Cochrane Database of Systematic Reviews (Reviews) || Continuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus

Continuous subcutaneous insulin infusion (CSII) versus

multiple insulin injections for type 1 diabetes mellitus

(Review)

Misso ML, Egberts KJ, Page M, O’Connor D, Shaw J

This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library2010, Issue 1

http://www.thecochranelibrary.com

Continuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus (Review)

Copyright © 2010 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.

http://www.thecochranelibrary.com

T A B L E O F C O N T E N T S

1HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Figure 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Figure 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Figure 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Figure 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Figure 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Figure 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

18DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

88DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

90ADDITIONAL TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

93CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94DIFFERENCES BETWEEN PROTOCOL AND REVIEW . . . . . . . . . . . . . . . . . . . . .

94INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

iContinuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus (Review)


[Intervention Review]

Continuous subcutaneous insulin infusion (CSII) versusmultiple insulin injections for type 1 diabetes mellitus

Marie L Misso1 , Kristine J Egberts2, Matthew Page3, Denise O’Connor3 , Jonathan Shaw4

1Australasian Cochrane Centre, Clayton, Australia. 2Centre for Obesity Research and Education (CORE), Monash University, Mel-

bourne, Australia. 3Australasian Cochrane Centre, Monash Institute of Health Services Research, Monash University, Clayton, Aus-

tralia. 4Baker IDI Heart and Diabetes Institute, Victoria, Australia

Contact address: Marie L Misso, Australasian Cochrane Centre, Monash Institute of Health Services Research, Monash University, 43-

51 Kanooka Grove, Clayton, Victoria, 3168, Australia. [email protected].

Editorial group: Cochrane Metabolic and Endocrine Disorders Group.

Publication status and date: New, published in Issue 1, 2010.

Review content assessed as up-to-date: 19 July 2009.

Citation: Misso ML, Egberts KJ, Page M, O’Connor D, Shaw J. Continuous subcutaneous insulin infusion (CSII) versus multiple

insulin injections for type 1 diabetes mellitus. Cochrane Database of Systematic Reviews 2010, Issue 1. Art. No.: CD005103. DOI:

10.1002/14651858.CD005103.pub2.


A B S T R A C T

Background

Type 1 diabetes is a metabolic disorder resulting from a defect in insulin secretion. Onset of type 1 diabetes mellitus may occur at any

age and it is one of the most common chronic diseases of childhood and adolescence. Since there are no interventions known to prevent

onset, it is vital that effective treatment regimes are available. Glycaemic control is maintained by replacement of insulin and may be

in the form of ’conventional’ insulin therapy (multiple injections per day) or continuous subcutaneous insulin infusion (CSII).

Objectives

To assess the effects of CSII compared to multiple insulin injections (MI) in people with type 1 diabetes mellitus.

Search methods

Studies were obtained from electronic searches of The Cochrane Library, MEDLINE, EMBASE and CINAHL.

Selection criteria

Studies were included if they were randomised controlled trials comparing CSII with three or more insulin injections per day (MI) in

people with type 1 diabetes mellitus.

Data collection and analysis

Two authors independently assessed risk of bias and extracted characteristics of included studies. Authors contacted study investigators

to obtain missing information. Generic inverse variance meta-analyses using a random-effects model were performed.

Main results

Twenty three studies randomised 976 participants with type 1 diabetes to either intervention. There was a statistically significant

difference in glycosylated haemoglobin A1c (HbA1c) favouring CSII (weighted mean difference -0.3% (95% confidence interval -

0.1 to -0.4). There were no obvious differences between the interventions for non-severe hypoglycaemia, but severe hypoglycaemia

appeared to be reduced in those using CSII. Quality of life measures suggest that CSII is preferred over MI. No significant difference

was found for weight. Adverse events were not well reported, no information is available on mortality, morbidity and costs.

1Continuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus (Review)


mailto:[email protected]

Authors’ conclusions

There is some evidence to suggest that CSII may be better than MI for glycaemic control in people with type 1 diabetes. Non-

severe hypoglycaemic events do not appear to be reduced with CSII. There is insufficient evidence regarding adverse events, mortality,

morbidity and costs.

P L A I N L A N G U A G E S U M M A R Y

Continuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus

Type 1 diabetes results from a defect in insulin secretion, leading to elevated levels of plasma sugar or glucose and disturbances in

carbohydrate, fat and protein metabolism. Complications may effect the eyes, kidneys, nerves and the cardiovascular system. Type 1

diabetes may occur at any age and it is one of the most common chronic diseases of childhood and adolescence. Type 1 diabetes impacts

heavily on the lifestyle of the individual as well as their families. Since there is no cure or prevention for type 1 diabetes, life-long

insulin replacement and monitoring of blood glucose levels are required. It is vital that effective insulin therapy regimes are available

for optimal management and to minimise blood glucose fluctuation (known as too low or too high blood sugar levels - hypoglycaemia

or hyperglycaemia).

Insulin therapy may be in the form of ’conventional’ therapy of multiple (typically four) injections per day or continuous subcutaneous

insulin infusion. Continuous subcutaneous insulin infusion involves attachment (via catheter on the outside of the body) to an insulin

pump that is programmed to deliver insulin to match the individual’s needs, and doses are activated by the individual to cover meals

and correct blood glucose fluctuation. Here we explore whether continuous subcutaneous insulin infusion is better than three or more

insulin injections per day for good management of type 1 diabetes.

Twenty three studies randomised 976 participants with type 1 diabetes to either continuous subcutaneous insulin infusion or multiple

injections. Seven of the 23 studies were performed in participants under 18 years of age and the remainder were performed in adults.

Study duration ranged from six days to four years. The body of evidence suggests that continuous subcutaneous insulin infusion may be

better than multiple injections for glycaemic control in people with type 1 diabetes; continuous subcutaneous insulin infusion appears

to provide no benefit for reducing non-severe hypoglycaemic events. Future studies need to consider the short and long term adverse

effects, mortality, morbidity and costs of these interventions.

B A C K G R O U N D

Description of the condition

Type 1 diabetes results from a defect in insulin secretion. A con-

sequence of this is chronic hyperglycaemia (that is elevated lev-

els of plasma glucose) with disturbances of carbohydrate, fat and

protein metabolism. Long-term complications of diabetes melli-

tus include retinopathy, nephropathy and neuropathy. The risk

of cardiovascular disease is increased. For a detailed overview of

diabetes mellitus, please see under ’Additional information’ in the

information on the Metabolic and Endocrine Disorders Group

on the Cochrane Library (see ’About the Cochrane Collabora-

tion’, ’Collaborative Review Groups-CRGs’). For an explanation

of methodological terms, see the main Glossary on The CochraneLibrary.

The onset of type 1 diabetes mellitus may occur at any age and it is

one of the most common chronic diseases of childhood and ado-

lescence. In adults with type 1 diabetes, vascular complications are

a major cause of mortality (Couper 2002). Based on young onset

patients, the incidence of this disease worldwide varies from 0.1

to 37.4 per 100,000 (Karvonen 2000). The EURODIAB Study

Group predicts that between 2005 and 2020, new cases of type 1

diabetes in European children younger than five years will double

and that the prevalence of cases in those younger than 15 years will

increase by 70% (Dabelea 2009). Type 1 diabetes impacts heavily

on the lifestyle of the individual as well as their families. Life-long

insulin replacement and monitoring of blood glucose levels are

required and patients are prone to acute and late complications of

diabetes (for example cardiovascular disease) that are influenced

by diabetes control. Moreover, the economic burden is high for

the individual, the family, the health care sector and society as a



whole, where significant health care costs arise not only at diagno-

sis but also for subsequent admissions to hospital (Sutton 1990).

Description of the intervention

Since there are no interventions shown to delay or prevent the

onset of type 1 diabetes (DPT 2002; ENDIT 2003), it is vital that

effective treatment regimes are available for optimal management

of this disease. Glycaemic control is maintained by replacement of

insulin. Insulin is administered subcutaneously from where insulin

is absorbed into the bloodstream. Here, complexes are formed en-

abling binding to the insulin receptor for activation of glucose

uptake into the muscle. There are various regimes for insulin ad-

ministration. Ideally, an individual’s insulin replacement regimen

should provide appropriate basal insulin requirements across 24

hours, provide sufficient levels to cover food intake, have ade-

quate provision for correction when needed, minimise blood glu-

cose fluctuation and therefore risk of hypoglycaemia and hyper-

glycaemia and achieve short-term and long-term targets (APEG

2004). Insulin therapy may be executed in the form of ’conven-

tional’ insulin therapy of multiple injections (like 2 to 4 injec-

tions per day) or continuous subcutaneous insulin infusion (CSII).

The multiple injection form of therapy may consist of adminis-

tering rapid-acting, short-acting, intermediate-acting or long-act-

ing types of insulin (APEG 2004), which differ in relation both

to their speed of onset and duration of glucose-lowering action.

For example, rapid-acting insulins typically begin to lower glucose

within 5 to 10 minutes of injection, and last for 4 to 6 hours,

while short-acting insulins take at least 20 minutes to begin to

act, and can last for up to eight hours. In contrast, CSII typically

administers a rapid-acting insulin (analogue) via an insulin pump

programmed to deliver basal rates to match the individual’s needs

and bolus doses are activated by the individual to cover meals and

correct hyperglycaemia. Insulin pumps are worn by the individual

externally and insulin is delivered from the pump via insertion

of a catheter into the subcutaneous tissue of the abdominal wall

(Pickup 1980a). In this review, we define multiple insulin injec-

tions as three or more per day (MI).

How the intervention might work

While CSII and MI are both successful methods of insulin therapy,

it has been suggested that CSII may be a superior alternative since

it allows to more closely mimic normal physiology and eliminate

the need for repeat injections. CSII may provide a more efficient

mode of insulin delivery to the bloodstream, minimise the risk of

hypoglycaemic events and may be more beneficial to those who

find it difficult to adhere to a regimen of multiple injections.

Adverse effects of the intervention

Potential adverse events that may arise include pump malfunction,

catheter-site infection, irritation, diabetic ketoacidosis or discom-

fort (Guinn 1988).

Why it is important to do this review

Three systematic reviews comparing MI with CSII have been pub-

lished (HTA 2004; NICE 2008; Pickup 2002). The review by

Pickup et al provides an overview of the studies located, however

there are some limitations. Firstly, on-going study and abstract

searches have not been documented, therefore it is possible that

relevant studies may have been overlooked. Secondly, important

aspects such as the search strategy and methods of study randomi-

sation have not been documented. The Health Technology As-

sessment (HTA) report and the National Institute for Clinical Ex-

cellence (NICE) guidelines are methodologically rigorous, how-

ever the most up-to-date search strategies were conducted in 2007

and therefore do not include recent randomised controlled trials.

Given the limitations of these previous systematic reviews, it is

important that we have considered these factors and used specific

methodology and criteria outlined by the Cochrane Collaboration

in our aim to present a comprehensive systematic review to assess

the effects of CSII compared to MI. Where possible and feasible,

we have performed meta-analyses.

O B J E C T I V E S

To assess the effects of continuous subcutaneous insulin infusion

(CSII) compared to multiple insulin injection for type 1 diabetes

mellitus.

M E T H O D S

Criteria for considering studies for this review

Types of studies

All published and unpublished randomised and quasi-randomised

controlled trials (blind and open, parallel and cross-over) designed

to compare people with type 1 diabetes taking insulin in the form

of either continuous subcutaneous insulin infusion (CSII) or mul-

tiple insulin injections (MI) (three or more insulin injections per

day). There were no time or language restrictions.



Types of participants

Males and females with type 1 diabetes of any age taking insulin

treatment were considered in this review. The diagnostic criteria

for type 1 diabetes mellitus should have been described in the study.

To remain consistent with changes in classification and diagnostic

criteria through the years, diagnosis of type 1 diabetes status was

based upon the diagnostic criteria valid at the time of beginning

the study. Acceptable diagnostic criteria included those described

by the National Diabetes Data Group standards (NDDG 1979),

the World Health Organisation (WHO 1980), the American Di-

abetes Association (ADA 1997) or by the author of the study (in

these cases, attempts were made to contact study investigators to

determine specific criteria used to diagnose type 1 diabetes).

Types of interventions

Insulin preparation and dose of insulin varied depending on the

study, therefore all insulin preparations and doses were accepted.

Intervention

Continuous subcutaneous insulin infusion (CSII).

Control

Multiple insulin injections (MI) (three or more insulin injections

per day).

Types of outcome measures

Primary outcomes

• glycaemic control (glycosylated haemoglobin A1c (HbA1c),

daily mean blood glucose, fasting blood glucose or postprandial

blood glucose);

• number of overall, severe and non-severe hypoglycaemic

episodes;

• quality of life (ideally measured using a validated

instrument).

Secondary outcomes

• weight;

• insulin requirement to maintain glycaemic control;

• number and severity of adverse events (e.g. ketoacidosis,

local infections);

• diabetes late complications;

• mortality;

• costs.

Timing of outcome assessment

Short (less than or equal to one month), medium (more than one

month and less than six months) and longer term (equal to or

more than six months) outcome data were included in the review.

Search methods for identification of studies

Electronic searches

We searched the following sources for the identification of studies

up to 20 July 2009:

• The Cochrane Library;• MEDLINE;

• EMBASE;

• CINAHL.

For detailed search strategies please see under Appendix 1.

We also planned to search databases of ongoing studies: Cur-

rent Controlled Trials (www.controlled-trials.com - with links to

other databases of ongoing trials) and the National Research Reg-

ister (www.update-software.com/National/nrr-frame.html). This

will be performed in future updates of this review.

Searching other resources

We handsearched abstracts from international diabetes meetings

from 2002, 2003 and 2004 (including IDF, EASD, ADA).

Data collection and analysis

Selection of studies

To determine the studies to be assessed further, two independent

authors (MM, KE) scanned the titles, abstract sections and key-

words of every record retrieved, where available. Full articles were

retrieved for further assessment if the information provided was in-

sufficient to decide eligibility or if information provided suggested

that the study: 1. included people with type 1 diabetes mellitus; 2.

compared CSII and MI (three or more insulin injections per day);

3. assessed one or more relevant clinical outcome measure(s); 4.

was randomised. There were no disagreements about selection of

studies between authors.

Dealing with duplicate publications

In the case of duplicate publications and companion papers of

primary studies, we tried to maximise yield of information by

simultaneous evaluation of all available data. In cases of doubt,

the original publication (usually but not always the oldest version)

obtained priority.



Data extraction and management

For studies that met the selection criteria, two out of three possible

authors (MM, KE and MP) independently extracted relevant pop-

ulation and intervention characteristics using a standard data ex-

traction template (for details see Characteristics of included studies

and Table 1, Appendix 2, Appendix 8). The authors reached agree-

ment about the data extracted through discussion.

Assessment of risk of bias in included studies

Two authors assessed risk of bias of each study independently (for

details see Characteristics of included studies). The authors ex-

plored the influence of individual risk of bias criteria in a sensitiv-

ity analysis (see ’Sensitivity analysis’). The authors reached agree-

ment about risk of bias assessments through discussion. Potential

for bias was considered using the risk of bias table which addresses

the following domains:

• sequence generation;

• allocation concealment;

• blinding of participants, personnel and outcome assessors;

• incomplete outcome data;

• selective outcome reporting;

• other sources of bias.

Each question was answered ’Yes’, ’No’ or ’Unclear’, where ’Yes’

suggests a low risk of bias, ’No’ suggests a high risk of bias and

’Unclear’ suggests doubt about the level of bias in that domain.

Summary assessments for risk of bias were made for each outcome

across studies (see ’Effects of interventions’) using the approach

suggested in the Cochrane Handbook for Systematic Reviews of In-tervention (Higgins 2008).

Measures of treatment effect

Review Manager 5.0 software was used for data analysis. Results

are expressed as mean differences, calculated from end of treat-

ment values, with 95% confidence intervals (CI) for continuous

outcomes. There are no dichotomous outcomes in this review.

Pooled results were meta-analysed using the generic inverse vari-

ance method with a random-effects model. Statistical significance

was set at a P value less than 0.05 for primary and secondary out-

come measures.

Dealing with missing data

Relevant missing information about study design or results were

sought from the study investigators, where feasible. Evaluation

of important numerical data such as screened, eligible and ran-

domised participants as well as intention-to-treat and per-proto-

col population is presented in Table 1. Attrition rates like drop-

outs, losses to follow-up and withdrawn study participants were

investigated. Issues of last-observation-carried-forward were criti-

cally appraised and compared to specification of primary outcome

parameters and power calculation.

Six cross-over studies were not adequately analysed, and did not

report correlations between baseline and end of study data, thereby

ignoring within-person variation. We reanalysed the data from

these studies assuming different correlation coefficients (0.3, 0.5,

0.7, 0.9) and have discussed the results based on the most con-

servative approach, assuming a correlation coefficient of 0.3. The

data from these studies were also reanalysed assuming the largest

standard deviation of the differences, derived from the cross-over

studies that were adequately analysed. The results from these anal-

yses can be found in Appendix 3.

Assessment of heterogeneity

Results of clinically and statistically homogenous studies were

pooled to provide estimates of the efficacy of the interventions.

Clinical homogeneity was satisfied when participants, interven-

tions, outcome measures and timing of outcome measurement

were considered to be similar. For studies that were clinically het-

erogeneous or presented insufficient information for pooling, a

descriptive analysis was performed. Statistical homogeneity was

assessed using the I2 statistic where I2 values under 50% indicated

acceptable homogeneity (low heterogeneity) (Higgins 2003).

Assessment of reporting biases

Funnel plots were planned to be used in exploratory data analyses

to assess for the potential existence of small study bias. There are

a number of explanations for the asymmetry of a funnel plot,

including true heterogeneity of effect with respect to study size,

poor methodological design of small studies and publication bias

(Sterne 2001). Thus, the findings using this exploratory data tool

were interpreted with caution.

Data synthesis

Data were summarised statistically where available, sufficiently

similar, and of sufficient quality. Generic inverse variance meta-

analyses were performed using a random-effects model for the

included continuous outcomes. Effect estimates are expressed as

mean difference and 95% confidence interval (CI). Where data

could not be combined, narrative tables of results are presented.

Subgroup analysis and investigation of heterogeneity

Subgroup analyses were conducted according to age of study par-

ticipants and study duration, since these factors may cause varia-

tions in outcomes.

We also conducted post-hoc analyses where we explored studies

that included less than 20 participants in subgroup analyses for

each outcome (Appendix 4).



Sensitivity analysis

Sensitivity analyses were performed for every element on the risk

of bias table by excluding studies that had a high risk of bias. Other

elements that were employed in sensitivity analyses included:

• excluding very long or large studies to establish how much

they dominate the results;

• excluding studies using the following filters: diagnostic

criteria, date of publication, language of publication, source of

funding (industry versus other), country.

R E S U L T S

Description of studies

See: Characteristics of included studies; Characteristics of excluded

studies.

Results of the search

The search conducted up until 20 July 2009 for all databases, iden-

tified 4343 records, from these, 201 full text papers were retrieved

for further examination. After screening the full text of the selected

papers, 23 studies finally met the inclusion criteria. An adapted

PRISMA (Preferred Reporting Items for Systematic Reviews and

Meta-Analyses) flow-chart of study selection is attached (Liberati

2009), see Figure 1.

Figure 1. Adapted PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow-

chart of study selection



Assessment of inter-rater agreement

Two authors reviewed the studies, and were in agreement on those

to be fully assessed. From these, studies eligible for inclusion in

the review were identified. Both authors agreed on the final papers

chosen for assessment and on the risk of bias assessment of the

studies. There was complete agreement (100%) on the final papers

chosen for further assessment.

Included studies

Designs of included studies

Twenty-three randomised controlled trials (RCTs) meeting the

inclusion criteria of this review were identified. Details of these

studies are shown in the ’Characteristics of included studies’ ta-

ble. Eleven of the 23 included studies were of parallel design,

the other 12 had a cross-over design (Bak 1987 and Husted

1989; Bruttomesso 2008; Chiasson 1984 and 1985; Cohen 2003;

Hanaire-Broutin 2000; Hirsch 2005; Home 1982; Hoogma 2005;

Nathan 1982; Saurbrey 1988; Schmitz 1989; Weintrob 2004). Six

of the cross-over studies were analysed using adequate statistical

methodology to account for within-person variation. Six cross-

over studies were not adequately analysed, and ignored within-

person variation, so we reanalysed the data from these studies as

described above in ’Dealing with missing data’. The results from

these analyses can be found in Appendix 3.

Seventy-eight percent (18) of the studies had an initial run-in

phase lasting from two days to six months in order to achieve

stable metabolic conditions. Of the 12 studies with a cross-over

design, six reported a wash-out period before switching to the

other treatment.

The single centre design was the dominating setting (70%) but

multiple centre studies were also common (30%). Studies were

published from 1982 to 2008. Sixty-one per cent (14 studies)

could be clearly identified as partly or entirely industry sponsored

and in two studies, funding was not specified. Diagnostic criteria

for entry into the study was specified in eight of the studies.

Participants of included studies

In total, 976 people with type 1 diabetes took part in the 23

randomised controlled studies.

Seven of the 23 studies were performed in participants under 18

years of age (Cohen 2003; Doyle 2004; Meschi 1982; Nuboer

2008; Pozzilli 2003; Skogsberg 2008; Weintrob 2004).

All but one study investigated the effects in both genders. Botta

1986 only included pregnant females (n=10). It is important to

note that other studies included in this review have excluded preg-

nant females, suggesting clinical heterogeneity among the studies

in this review.

Interventions of included studies

All included studies compared continuous subcutaneous insulin

infusion (CSII) and multiple insulin injections (MI) (three or

more insulin injections per day). The duration of the intervention

ranged from six days to four years. It is important to note that

the study with an intervention period of six days (Meschi 1982)

should not be considered a real life setting and so was explored

in sensitivity analysis but the effect estimate was largely unaltered.

Most studies tried to achieve a comparable insulin dose regimen

throughout the investigation period, and treating physicians, to-

gether with the participants, tried to achieve optimisation of ther-

apy, usually by means of flexible insulin dosage in order to achieve

metabolic targets of heterogeneously defined adequate blood glu-

cose control.

Excluded studies

In total, 178 studies were excluded after careful evaluation of the

full publication. Reasons for exclusion of studies are given in the

’Characteristics of excluded studies’ table. The main reasons for

exclusion were inappropriate interventions, less than three injec-

tions per day in the MI group and non-randomised study design.

Risk of bias in included studies

For details of risk of bias in the included studies, see the

’Characteristics of included studies’ tables, Figure 2 and Figure 3.

For each outcome in ’Effects of interventions’, a summary assess-

ment for the risk of bias is provided.



Figure 2. Risk of bias summary: review authors’ judgements about risk of bias items for each included study.



Figure 3. Risk of bias graph: review authors’ judgements about each risk of bias item presented as

percentages across all included studies.

Allocation

All studies were described as randomised, however only eight

studies (Bruttomesso 2008; DeVries 2002; Doyle 2004; Hoogma

2005; Lepore 2003; Oslo Study 1985 to 1992; Pozzilli 2003; Tsui

2001) adequately described the method of randomisation and only

seven studies (Bruttomesso 2008; DeVries 2002; Hanaire-Broutin

2000; Hoogma 2005; Nosadini 1988; Nuboer 2008; Tsui 2001)

adequately described allocation concealment.

Blinding

Given the nature of the interventions, participant blinding was not

appropriate. Investigator blinding and outcome assessor blinding

was not described in any of the studies, however it is not expected

that this would introduce major bias since the majority of out-

comes are not subjective.

Incomplete outcome data

Three studies (Doyle 2004; Hoogma 2005; Tsui 2001) reported

the use of intention-to-treat analysis. Most studies described losses

to follow-up or drop-outs, if any occurred in the study.

For details about the number of screened and randomised partic-

ipants see Table 1.

Selective reporting

Selective reporting was unclear in all except one study, where this

was apparent (Hirsch 2005). For the remaining studies it was not

possible to track if all pre-specified outcomes were reported and if

they were reported in the pre-specified way.

Other potential sources of bias

The risk for other bias is unclear in all of the included studies.

Effects of interventions

Baseline characteristics

For details of baseline characteristics see Appendix 2.

Primary outcomes

Glycosylated haemoglobin A1c (HbA1c) (%)

Of the 23 studies included, 20 studies were included in this anal-

ysis. Fifteen of these (Bruttomesso 2008; Chiasson 1984 & 1985;

DeVries 2002; Doyle 2004; Home 1982; Lepore 2003; Meschi



1982; Nathan 1982; Nosadini 1988; Nuboer 2008; Oslo Study

1985 to 1992; Pozzilli 2003; Schmitz 1989; Skogsberg 2008;

Tsui 2001) were parallel or cross-over studies using appropriate

statistical methodology; and five (Cohen 2003; Hanaire-Broutin

2000; Hoogma 2005; Saurbrey 1988; Weintrob 2004) were cross-

over studies with inadequate statistical methodology (i.e. ignor-

ing within-person variation). Therefore, we reanalysed the data

from these studies assuming different correlation coefficients (0.3,

0.5, 0.7, 0.9). The data from these studies were also reanalysed

assuming the largest standard deviation of the differences derived

from the cross-over studies with adequate statistical methodology.

Three studies were excluded because they did not report HbA1c.

Assuming a conservative correlation of 0.3, and using generic in-

verse variance with a random-effects model, the mean difference

of HbA1c was estimated to be -0.3% (95% CI -0.4 to -0.1) in

favour of CSII compared with MI. The test of heterogeneity gave

an I2 statistic of 50% and the test for overall effect gave a P value

of 0.001 (Figure 4). The effect estimates, overall treatment effects

and heterogeneity assuming other correlations and the largest stan-

dard deviation of the differences are presented in Appendix 3.

Figure 4. Forest plot of comparison: CSII versus MI, outcome: HbA1c (%).

Risk of bias summary assessment

Most information is from studies at low or unclear risk of bias.

Detailed risk of bias assessments can be found in ’Characteristics

of included studies’ tables and Figure 2 and Figure 3.

Subgroup analyses

Subgroup analyses for all outcomes were performed assuming a

conservative correlation of 0.3, using generic inverse variance with

a random-effects model. The effect estimates, overall treatment ef-



fects and heterogeneity assuming other correlations and the largest

standard deviation of the differences are presented in Appendix 3.

HbA1c - age of study participants

In studies that included participants less than 18 years of age (

Cohen 2003; Doyle 2004; Meschi 1982; Nuboer 2008; Pozzilli

2003; Skogsberg 2008; Weintrob 2004) the mean difference of

HbA1c was estimated to be -0.2% (95% CI -0.4 to -0.03) in favour

of CSII compared with MI. The test of heterogeneity gave an I2

statistic of 20%. The test for overall effect gave a P value of 0.02

(Analysis 1.2). In studies that included participants more than 18

years of age (Bruttomesso 2008; DeVries 2002; Hanaire-Broutin

2000; Home 1982; Hoogma 2005; Lepore 2003; Nathan 1982;

Nosadini 1988; Oslo Study 1985 to 1992; Saurbrey 1988; Schmitz

1989; Tsui 2001) the mean difference of HbA1c was estimated to

be -0.3% (95% CI -0.5 to -0.1) in favour of CSII compared with

MI. The test of heterogeneity gave an I2 statistic of 62%. The test

for overall effect gave a P value of 0.01 (Analysis 1.2).

HbA1c subgroup analysis - study duration

There was only one short term study (up to one month), which is

insufficient for subgroup analysis (Meschi 1982).

In medium term studies (from one to six months) (Bruttomesso

2008; Chiasson 1984 & 1985; Cohen 2003; DeVries 2002;

Doyle 2004; Hanaire-Broutin 2000; Home 1982; Hoogma 2005;

Nathan 1982; Saurbrey 1988; Schmitz 1989; Weintrob 2004),

the mean difference of HbA1c was estimated to be -0.3% (95%

CI -0.5 to -0.1) in favour of CSII compared with MI. The test

of heterogeneity gave an I2 statistic of 53%. The test for overall

effect gave a P value of 0.002 (Analysis 1.3).

In long term studies (more than six months) (Lepore 2003;

Nosadini 1988; Nuboer 2008; Oslo Study 1985 to 1992; Pozzilli

2003; Skogsberg 2008; Tsui 2001), the mean difference of HbA1c

was estimated to be -0.2% (95% CI -0.4 to 0.1) in favour of CSII

compared with MI. The test of heterogeneity gave an I2 statistic

of 57%. The test for overall effect gave a P value of 0.21 (Analysis

1.3).

Small study bias

Assuming a conservative correlation of 0.3, using generic inverse

variance with a random-effects model for studies that included

less than 20 participants (Chiasson 1984 and 1985; Cohen 2003;

Home 1982; Meschi 1982; Nathan 1982; Schmitz 1989) the mean

difference of HbA1c was estimated to be -0.7% (95% CI -1.6 to

0.1) in favour of CSII compared with MI. The test of heterogeneity

gave an I2 statistic of 63%. The test for overall effect gave a P

value of 0.09 (Analysis 1.4). The effect estimates, overall treatment

effects and heterogeneity are presented in Appendix 3.

Daily mean blood glucose (mg/dl)

Of 23 studies included, 13 were included in this analysis. Nine of

these (Bak 1987 and Husted 1989; Bruttomesso 2008; DeVries

2002; Home 1982; Meschi 1982; Nathan 1982; Nosadini 1988;

Oslo Study 1985 to 1992; Schmitz 1989) were parallel or

cross-over studies using appropriate statistical methodology; and

four (Hanaire-Broutin 2000; Hoogma 2005; Saurbrey 1988;

Weintrob 2004) were cross-over studies where appropriate statis-

tical methodology was not used (i.e. ignoring within-person varia-

tion). Therefore, we reanalysed the data as described in the results

section for HbA1c. Ten studies were excluded because they did

not report daily mean blood glucose.


verse variance with a random-effects model, a pooled mean dif-

ference of daily mean blood glucose could not be reliably esti-

mated due to substantial heterogeneity with an I2 statistic of 94%.

The effect estimates, overall treatment effects and heterogeneity

assuming other correlations and the largest standard deviation of

the differences are presented in Appendix 3.





Subgroup analyses

Subgroup analyses were performed and presented as described in

the HbA1c subgroup analysis section.

Daily mean blood glucose - age of study participants

In studies that included participants less than 18 years of age

(Bak 1987 and Husted 1989; Meschi 1982; Weintrob 2004) the

mean difference of daily mean blood glucose was estimated to

be -4 mg/dl (95% CI -14 to 7) in favour of CSII compared

with MI. The test of heterogeneity gave an I2 statistic of 0%.

The test for overall effect gave a P value of 0.5 (Analysis 1.6).

In studies that included participants more than 18 years of age

(Bruttomesso 2008; DeVries 2002; Hanaire-Broutin 2000; Home

1982; Hoogma 2005; Nathan 1982; Nosadini 1988; Oslo Study

1985 to 1992; Saurbrey 1988; Schmitz 1989) the mean difference

of daily mean blood glucose was estimated to be -18 mg/dl (95%

CI -27 to -9) in favour of CSII compared with MI. The test of

heterogeneity gave an I2 statistic of 96%. Therefore, the pooled

effect estimate is not reliable (Analysis 1.6).

Daily mean blood glucose - study duration





In medium term studies (from one to six months) (Bak 1987 and

Husted 1989; Bruttomesso 2008; DeVries 2002; Hanaire-Broutin

2000; Home 1982; Hoogma 2005; Nathan 1982; Saurbrey 1988;

Schmitz 1989; Weintrob 2004), the mean difference of daily mean

blood glucose was estimated to be -14 mg/dl (95% CI -23 to -5)

in favour of CSII compared with MI. The test of heterogeneity

gave an I2 statistic of 95%. Therefore, the pooled effect estimate

is not reliable (Analysis 1.7).

In long term studies (more than six months) (Nosadini 1988; Oslo

Study 1985 to 1992), the mean difference of daily mean blood

glucose was estimated to be -21 mg/dl (95% CI -29 to -13) in


an I2 statistic of 0%. The test for overall effect gave a P value less

than 0.00001 (Analysis 1.7).

Small study bias


variance with a random effects model for studies that included less

than 20 participants (Home 1982; Meschi 1982; Nathan 1982;

Schmitz 1989) the mean difference of daily mean blood glucose

was estimated to be -29 mg/dl (95% CI -51 to -7) in favour of

CSII compared with MI. The test of heterogeneity gave an I2


(Analysis 1.8). The effect estimates, overall treatment effects and

heterogeneity are presented in Appendix 3.

Fasting blood glucose (mg/dl)

Of 23 studies included, 11 were included in this analysis. Eight of

these (Botta 1986; Bruttomesso 2008; Chiasson 1984 and 1985;

DeVries 2002; Doyle 2004; Home 1982; Lepore 2003; Oslo Study

1985 to 1992) were parallel or cross-over studies using appropriate

statistical methodology; and three (Hirsch 2005; Hoogma 2005;

Weintrob 2004) were cross-over studies where appropriate statis-

tical methodology was not used (i.e. ignoring within-person varia-

tion). Therefore, we reanalysed the data as described in the results

section for HbA1c. Twelve studies were excluded because they did

not report fasting blood glucose.



of fasting blood glucose was estimated to be -14 mg/dl (95% CI

-24 to -4) in favour of CSII compared with MI. The test of het-

erogeneity gave an I2 statistic of 71%. Therefore, the pooled ef-

fect estimate is not reliable (Figure 6). The effect estimates, overall

treatment effects and heterogeneity assuming other correlations

and the largest standard deviation of the differences are presented

in Appendix 3.


The proportion of information from studies at high risk of bias

is sufficient to affect the interpretation of results. Detailed risk

of bias assessments can be found in ’Characteristics of included

studies’ tables and Figure 2 and Figure 3.

Subgroup analyses



Fasting blood glucose - age of study participants


Chiasson 1984 and 1985; Doyle 2004; Weintrob 2004) the mean

difference of fasting blood glucose was estimated to be -8 mg/dl

(95% CI -24 to 9) in favour of CSII compared with MI. The test of

heterogeneity gave an I2 statistic of 0%. The test for overall effect

gave a P value of 0.35 (Analysis 1.10). In studies that included

participants more than 18 years of age (Botta 1986; Bruttomesso

2008; DeVries 2002; Hirsch 2005; Home 1982; Hoogma 2005;

Lepore 2003; Oslo Study 1985 to 1992) the mean difference of

fasting blood glucose was estimated to be -16 mg/dl (95% CI -29 to

-3) in favour of CSII compared with MI. The test of heterogeneity



Fasting blood glucose - study duration


insufficient for subgroup analysis (Hirsch 2005).

In medium term studies (from one to six months) (Botta 1986;

Bruttomesso 2008; Chiasson 1984 and 1985; DeVries 2002;

Doyle 2004; Home 1982; Hoogma 2005; Weintrob 2004), the

mean difference of fasting blood glucose was estimated to be -8

mg/dl (95% CI -19 to 2) in favour of CSII compared with MI.

The test of heterogeneity gave an I2 statistic of 63%. Therefore,

the pooled effect estimate is not reliable (Analysis 1.11).

In long term studies (more than six months) (Lepore 2003; Oslo

Study 1985 to 1992), the mean difference of fasting blood glucose

was estimated to be -37 mg/dl (95% CI -73 to -1) in favour of CSII

compared with MI. The test of heterogeneity gave an I2 statistic of

61%. Therefore, the pooled effect estimate is not reliable (Analysis

1.11).

Small study bias


variance with a random-effects model for studies that included less

than 20 participants (Botta 1986; Chiasson 1984 & 1985; Home

1982) the mean difference of fasting blood glucose was estimated



to be -14 mg/dl (95% CI -50 to 23) in favour of CSII compared


Therefore, the pooled effect estimate is not reliable (Analysis 1.12).


are presented in Appendix 3.

Post prandial blood glucose (mg/dl)

Of 23 studies included, five were included in this analysis. Three of

these (Botta 1986; Chiasson 1984 and 1985; DeVries 2002) were

parallel or cross-over studies using appropriate statistical method-

ology; and two (Hirsch 2005; Weintrob 2004) were cross-over

studies where appropriate statistical methodology was not used

(ie. ignoring within-person variation). Therefore, we reanalysed

the data as described in the results section for HbA1c. Eighteen

studies were excluded because they did not report post prandial

blood glucose.



of post prandial blood glucose was estimated to be -4 mg/dl (95%

CI -11 to 2) in favour of CSII compared with MI. The test of

heterogeneity gave an I2 statistic of 37%. The test for overall ef-

fect gave a P value of 0.18 (Figure 5). The effect estimates, overall



in Appendix 3.

Figure 5. Forest plot of comparison: CSII versus MI, outcome: Post prandial blood glucose (mg/dl).






Subgroup analyses



Post prandial blood glucose - age of study participants


Chiasson 1984 and 1985; Weintrob 2004) the mean difference of

post prandial blood glucose was estimated to be -3 mg/dl (95%

CI -12 to 6) in favour of CSII compared with MI. The test of


gave a P value of 0.49 (Analysis 1.14). In studies that included

participants more than 18 years of age (Botta 1986, DeVries 2002,

Hirsch 2005) the mean difference of post prandial blood glucose

was estimated to be -8 mg/dl (95% CI -23 to 8) in favour of CSII



1.14).

Post prandial blood glucose - study duration



In medium term studies (from one to six months) (Botta 1986;

Chiasson 1984 and 1985; DeVries 2002; Weintrob 2004), the

mean difference of post prandial blood glucose was estimated to

be -2 mg/dl (95% CI -4 to 0.3) in favour of CSII compared with


for overall effect gave a P value of 0.08 (Analysis 1.15).

There were no long term studies (more than six months) for sub-

group analysis (Analysis 1.15).



Small study bias


variance with a random-effects model for studies that included less

than 20 participants (Botta 1986; Chiasson 1984 & 1985) the

mean difference of post prandial blood glucose was estimated to

be 1 mg/dl (95% CI -15 to 16) in favour of CSII compared with


for overall effect gave a P value of 0.93 (Analysis 1.16). The effect

estimates, overall treatment effects and heterogeneity are presented

in Appendix 3.

Hypoglycaemic events

Non-severe hypoglycaemic events

Non-severe hypoglycaemia was often defined as a blood glucose

level lower than 70 mg/dl. Some studies did not provide a defini-

tion. The 17 studies (Bak 1987 and Husted 1989; Bruttomesso

2008; Chiasson 1984 and 1985; Cohen 2003; DeVries 2002;

Hanaire-Broutin 2000; Hirsch 2005; Home 1982; Hoogma 2005;

Meschi 1982; Nathan 1982; Nosadini 1988; Oslo Study 1985 to

1992; Pozzilli 2003; Schmitz 1989; Saurbrey 1988; Tsui 2001;

Weintrob 2004) that reported non-severe hypoglycaemia used dif-

ferent scales and it was not appropriate to conduct a meta-analysis,

however the data indicate that there is no relevant benefit of one

intervention over the other for reducing non-severe hypoglycaemic

events. A descriptive table of results can be found in Appendix 5.





Severe hypoglycaemic events

Severe hypoglycaemia was often defined as requiring assistance

from another person for recovery or resulting in coma or seizure.

Some studies did not provide a definition. The 15 studies

(Bruttomesso 2008; Cohen 2003; DeVries 2002; Doyle 2004;

Hanaire-Broutin 2000; Hirsch 2005; Hoogma 2005; Lepore

2003; Nosadini 1988; Nuboer 2008; Oslo Study 1985 to 1992;

Schmitz 1989; Skogsberg 2008; Tsui 2001; Weintrob 2004) that

reported severe hypoglycaemia used different scales and it was not

appropriate to conduct a meta-analysis, however the data indicate

that CSII may be better than MI for reducing the incidence of

severe hypoglycaemic events. A descriptive table of results can be

found in Appendix 6.





Quality of life

Quality of life was measured using different instruments by 15

studies (Bak 1987 and Husted 1989; Bruttomesso 2008; Chiasson

1984 and 1985; Cohen 2003; DeVries 2002; Doyle 2004;

Hanaire-Broutin 2000; Home 1982; Hoogma 2005; Nathan

1982; Nuboer 2008; Saurbrey 1988; Schmitz 1989; Skogsberg

2008; Tsui 2001), none of which have reported minimal clinically

important differences. Four studies (Bruttomesso 2008; Cohen

2003; DeVries 2002; Skogsberg 2008) used the validated dia-

betes treatment satisfaction questionnaire (DTSQ) of which two

of these included participants less than 18 years of age. In all four

studies the CSII groups had higher scores, representing better treat-

ment satisfaction, than the MI group. Of the two studies (Cohen

2003; Doyle 2004) that used the validated diabetes quality of life

for youth scale (DQOLY), one study in adults (Cohen 2003) re-

ported a higher score, representing better quality of life, in the CSII

group than the MI group. The other study (Doyle 2004), includ-

ing participants less than 18 years of age, did not report scores but

noted that a difference between the groups was not found. Two

studies (Hoogma 2005; Tsui 2001) used the validated diabetes

quality of life scale (DQoL) and found that the MI group scored

lower, representing better quality of life, than the CSII group. Use

of the SF-12 questionnaire (Hoogma 2005), SF36 general health

perceptions scale (DeVries 2002) and the paediatric quality of life

inventory (PedsQL) (Nuboer 2008) (all validated) revealed that

CSII was favoured over MI for perception of better mental health,

perception of better general health and better quality of life, re-

spectively. It is difficult to determine which treatment is better

when independent questionnaires (Bak 1987 and Husted 1989;

Chiasson 1984 and 1985; Hanaire-Broutin 2000; Home 1982;

Hoogma 2005; Nathan 1982; Saurbrey 1988; Schmitz 1989) were

used to measure satisfaction, acceptability, preferred treatment and

reaction to treatment. It was not appropriate to conduct a meta-

analysis, however the data suggest that the majority of participants

were more satisfied with CSII than MI. A descriptive table of re-

sults can be found in Appendix 7.





Secondary outcomes



Weight (kg)

Of 23 studies included, eight were included in this analysis. Six of

these (Bak 1987 and Husted 1989; DeVries 2002; Home 1982;

Lepore 2003; Oslo Study 1985 to 1992; Schmitz 1989) were par-

allel or cross-over studies using appropriate statistical methodol-

ogy; and two (Hanaire-Broutin 2000; Hoogma 2005) were cross-

over studies where appropriate statistical methodology was not

used (i.e. ignoring within-person variation). Therefore, we reanal-

ysed the data as described in the results section for HbA1c. Fifteen

studies were excluded because they did not report weight.



of weight was estimated to be -0.4 kg (95% CI -2 to 1.3). The

test of heterogeneity gave an I2 statistic of 0%. The test for overall

effect gave a P value of 0.65 (Figure 6). The effect estimates, overall



in Appendix 3.

Figure 6. Forest plot of comparison: CSII versus MI, outcome: Weight (kg).





Subgroup analyses



Weight - age of study participants

There were no studies that included participants less than 18 years

of age, therefore it was not possible to conduct subgroup analyses.

Weight - study duration

There were no short term studies (up to one month) for subgroup

analysis (Analysis 1.21).

In medium term studies (from one to six months) (Bak 1987 and

Husted 1989; DeVries 2002; Hanaire-Broutin 2000; Home 1982;

Hoogma 2005; Schmitz 1989), the mean difference of weight was

estimated to be -0.2 kg (95% CI -2 to 1.5) in favour of CSII

compared with MI. The test of heterogeneity gave an I2 statistic

of 0%. The test for overall effect gave a P value of 0.78 (Analysis

1.21).

In long term studies (more than six months) (Lepore 2003; Oslo

Study 1985 to 1992), the mean difference of weight was estimated

to be -1.6 kg (95% CI -7.7 to 4.5) in favour of CSII compared


The test for overall effect gave a P value of 0.61 (Analysis 1.21).

Small study bias


variance with a random-effects model in studies that included less

than 20 participants (Botta 1986; Chiasson 1984 and 1985; Home

1982), the mean difference of weight was estimated to be -13.7

kg (95% CI -50.4 to 23) in favour of CSII compared with MI.

The test of heterogeneity gave an I2 value of 79%. Therefore, the

pooled effect estimate is not reliable (Analysis 1.22). The effect


in Appendix 3.



Daily insulin requirement

Studies presented daily insulin requirement as either U or as U/

kg, therefore data were analysed separately according to the unit

used by the study investigators.

Daily insulin requirement (units - U)

Of 23 studies included, nine were included in this analysis. Seven

of these (Bak 1987 and Husted 1989; Botta 1986; Chiasson 1984

and 1985; DeVries 2002; Home 1982; Lepore 2003; Nathan

1982) were parallel or cross-over studies using appropriate statisti-

cal methodology; and two (Hanaire-Broutin 2000; Hirsch 2005)

were cross-over studies where appropriate statistical methodology

was not used (i.e. ignoring within-person variation). Therefore, we

reanalysed the data as described in the results section for HbA1c.

Fourteen studies were excluded because they did not report daily

insulin requirement (U).



of daily insulin requirement (U) was estimated to be -7 U (95%

CI -11 to -3) in favour of CSII compared with MI. The test of


gave a P value of 0.0003 (Figure 7). The effect estimates, overall



in Appendix 3.

Figure 7. Forest plot of comparison: CSII versus MI, outcome: Daily insulin requirement (U).






Subgroup analyses



Daily insulin requirement (U) - age of study participants

There was only one study that included participants less than 18

years of age, which is insufficient for subgroup analysis (Chiasson

1984 and 1985). In studies that included participants more than

18 years of age (Bak 1987 and Husted 1989; Botta 1986; DeVries

2002; Hanaire-Broutin 2000; Hirsch 2005; Home 1982; Lepore

2003; Nathan 1982) the mean difference of daily insulin require-

ment (U) was estimated to be -7 U (95% CI -11 to -3) in favour



(Analysis 1.24).

Daily insulin requirement (U) - study duration



In medium term studies (from one to six months) (Bak 1987

and Husted 1989; Botta 1986; Chiasson 1984 and 1985; DeVries

2002; Hanaire-Broutin 2000; Home 1982; Nathan 1982), the

mean difference of daily insulin requirement (U) was estimated to

be -9 U (95% CI -14 to -3) in favour of CSII compared with MI.



The test of heterogeneity gave an I2 statistic of 38%. The test for

overall effect gave a P value of 0.003 (Analysis 1.25).

There was only one long term study (more than six months), which

is insufficient for subgroup analysis (Lepore 2003).

Small study bias


variance with a random-effects model in studies that included less

than 20 participants (Bak 1987 and Husted 1989; Botta 1986;

Chiasson 1984 and 1985; Home 1982; Nathan 1982) the mean

difference of daily insulin requirement (U) was estimated to be

-6 U (95% CI -14 to 2) in favour of CSII compared with MI.

The test of heterogeneity gave an I2 statistic of 31%. The test for

overall effect gave a P value of 0.14 (Analysis 1.26). The effect


in Appendix 3.

Daily insulin requirement (units per kg - U/kg)

Of 23 studies included, 12 were included in this analysis. Ten of

these (Bruttomesso 2008; Cohen 2003; Doyle 2004; Meschi 1982;

Nuboer 2008; Oslo Study 1985 to 1992; Pozzilli 2003; Schmitz

1989; Skogsberg 2008; Tsui 2001) were parallel or cross-over stud-

ies using appropriate statistical methodology; and two (Hoogma

2005; Saurbrey 1988) were cross-over studies where appropriate

statistical methodology was not used (i.e. ignoring within-person

variation). Therefore, we reanalysed the data as described in the

results section for HbA1c. Ten studies were excluded because they

did not report daily insulin requirement (U/kg).


verse variance with a random-effects model, a pooled mean dif-

ference of daily mean blood glucose could not be reliably esti-

mated due to substantial heterogeneity with an I2 statistic of 94%.


assuming other correlations and the largest standard deviation of

the differences are presented in Appendix 3.





Subgroup analyses



Daily insulin requirement (U/kg) - age of study participants

In studies that included participants less than 18 years of age

(Cohen 2003; Doyle 2004; Meschi 1982; Nuboer 2008; Pozzilli

2003; Skogsberg 2008) the mean difference of daily insulin re-

quirement (U/kg) was estimated to be -0.16 U/kg (95% CI -0.31

to -0.01) in favour of CSII compared with MI. The test of hetero-

geneity gave an I2 statistic of 84%. Therefore, the pooled effect es-

timate is not reliable (Analysis 1.28). In studies that included par-

ticipants more than 18 years of age (Bruttomesso 2008; Hoogma

2005; Oslo Study 1985 to 1992; Saurbrey 1988; Schmitz 1989;

Tsui 2001) the mean difference of daily insulin requirement (U/

kg) was estimated to be -0.08 U/kg (95% CI -0.15 to -0.01) in


an I2 value of 94%. Therefore, the pooled effect estimate is not

reliable (Analysis 1.28).

Daily insulin requirement (U/kg) - study duration



In medium term studies (from one to six months) (Bruttomesso

2008; Cohen 2003; Doyle 2004; Hoogma 2005; Saurbrey 1988;

Schmitz 1989), the mean difference of daily insulin requirement

(U/kg) was estimated to be -0.12 U/kg (95% CI -0.21 to -0.03)

in favour of CSII compared with MI. The test of heterogeneity



In long term studies (more than six months) (Nuboer 2008; Oslo

Study 1985 to 1992; Pozzilli 2003; Skogsberg 2008; Tsui 2001),

the mean difference of daily insulin requirement (U/kg) was esti-

mated to be -0.14 U/kg (95% CI -0.32 to 0.04) in favour of CSII



1.29).

Small study bias


variance with a random-effects model in studies that included

less than 20 participants (Cohen 2003; Meschi 1982; Schmitz

1989) the mean difference of daily insulin requirement (U/kg)

was estimated to be -0.04 U/kg (95% CI -0.13 to 0.05) in favour



(Analysis 1.30). The effect estimates, overall treatment effects and

heterogeneity are presented in Appendix 3.

Adverse events

Information about adverse events was not well reported. Twenty-

one studies reported that there were no deaths, one study did not



address this outcome and one study (Nosadini 1988) reported a

death in the CSII group, further details were not provided. Among

eight studies that did not describe the nature of adverse events,

there were more than 19 events in the CSII group and more than 10

in the MI group. Ketoacidosis was reported in two studies (Cohen

2003; DeVries 2002), where there was one event of ketoacidosis

in the CSII group of each study and one event of ketoacidosis

in the MI group of DeVries et al. Hoogma et al (Hoogma 2005)

reported 21 injection and infusion site injuries reactions in the

CSII group and Weintrob et al (Weintrob 2004) reported 12 mi-

nor infusion site infections, 16 blockages and 42 dislodgments.

All but one of the studies (Hoogma 2005) either did not report se-

rious adverse events or reported that there were no serious adverse

events. Hoogma et al reported 33 serious adverse events in the

CSII group and 58 in the MI group. Three studies reported drop-

outs as a results of adverse events. One of these (DeVries 2002)

reported one femur fracture but did not report which group the

participant was in. A second study (Doyle 2004) reported one case

of dehydration and ketosis in the MI group and the third study

(Home 1982) reported one undescribed event. In total, five CSII

participants and six MI participants were hospitalised (Bak 1987

and Husted 1989; Doyle 2004; Nuboer 2008; Saurbrey 1988)

and one CSII participant received out-patient treatment (Home

1982). For details of adverse events see Appendix 8.

Diabetes late complications

None of the studies reported on morbidity.

Mortality

None of the studies reported on mortality.

Costs

Costs were not addressed in any of the included studies.

Heterogeneity

In the HbA1c analysis comparing CSII and MI, heterogeneity was

reflected by an I2 statistic of 50%. After elimination of Nathan

1982, the study with high risk of bias, heterogeneity was signifi-

cantly reduced to 34% without altering the direction of effect.

Considerable heterogeneity was found and investigated for daily

mean blood glucose, fasting blood glucose and daily insulin re-

quired (measured in U/kg), but could not be explained by sub-

group analyses.

Statistical heterogeneity was either not present or low in the meta-

analyses of the outcomes post-prandial blood glucose, daily insulin

requirement (measured in units) and weight.

Subgroup analyses

Subgroup analyses are described in detail for each outcome in

’Effects of interventions’.

Sensitivity analyses

Sensitivity analyses were performed for every element on the risk

of bias table by excluding studies that had a high risk of bias. Other

sensitivity analyses included:

• excluding very long or large studies to establish how much

they dominate the results;

• excluding studies using the following filters: diagnostic

criteria, date of publication, language of publication, source of

funding (industry versus other), country.

Values were noted (data not shown) and found to be similar to

overall effect estimates.

Assessment of reporting biases

Funnel plot symmetry was not found in any of the meta-analysis.

D I S C U S S I O N

Summary of main results

Twenty-three studies randomised 976 participants with type 1 dia-

betes to either continuous subcutaneous insulin infusion (CSII) or

multiple insulin injections (MI) (three or more insulin injections

per day). Study duration ranged from six days to four years. Upon

meta-analysis of 20 studies that reported glycosylated haemoglo-

bin A1c (HbA1c) (overall unclear risk of bias), there was a statisti-

cally significant difference in favour of CSII, the mean difference

was -0.3% (95% confidence interval (CI) -0.4 to -0.1, P = 0.001).

Without altering the direction of effect, heterogeneity was reduced

from an I2 statistic of 50% to 34% upon removal of Nathan 1982,

a study with high risk of bias. Subgroup analyses for age did not

significantly alter the effect estimate, however subgroup analyses

for study duration showed that medium term studies (12 studies,

one to six months) had a higher effect estimate (-0.3% (95% CI

-0.5 to -0.1)) and long term studies (seven studies, more than six

months) had a lower non-significant effect estimate (-0.2% (95%

CI -0.4 to 0.1)).

Thirteen studies were meta-analysed for daily mean blood glucose

(overall unclear risk of bias). Due to considerable heterogeneity (I2 statistic of 94%) a reliable pooled effect estimate could not be

established. Subgroup analysis for age demonstrated that the effect

estimate and heterogeneity were lower and non-significant among

the two studies with participants under 18 years of age (-4 mg/dl

(95% CI -14 to 7), I2 statistic of 0%). Subgroup analysis for study

duration showed that combination of the two long term studies



resulted in a higher effect estimate (-21 mg/dl (95% CI -29 to -

13), I2 statistic of 0%).

Eleven studies reported fasting blood glucose (overall high risk of

bias). Due to considerable heterogeneity (I2 statistic of 71%) a

reliable pooled effect estimate could not be established. Subgroup

analysis for age demonstrated that the effect estimate and hetero-

geneity were lower among the three studies with participants un-

der 18 years of age (-8 mg/dl (95% CI -24 to 9), I2 statistic of

0%).

Meta-analysis of the five studies that reported post-prandial blood

glucose (overall high risk of bias) demonstrated a non-significant

difference between the interventions with an effect estimate of -

4 mg/dl (95% CI -11 to 2, I2 statistic of 37%) in favour of CSII

compared with MI. Subgroup analyses for age and study duration

did not substantially alter the effect estimates or heterogeneity.

Non-severe hypoglycaemia, often defined in the studies as a blood

glucose level lower than 70 mg/dl, was reported in 17 studies (over-

all unclear risk of bias) using different scales. It was not appropriate

to conduct a meta-analysis, however the data suggested that there

is no relevant benefit of one intervention over the other for re-

ducing non-severe hypoglycaemic events. Severe hypoglycaemia,

often defined in the studies as requiring assistance from another

person for recovery or resulting in coma or seizure, was reported

in 15 studies (overall unclear risk of bias) using different scales.

Despite it not being appropriate to conduct a meta-analysis, the

data suggest that CSII may be better than MI for reducing the

incidence of severe hypoglycaemic events.

Quality of life was reported in 15 studies (overall unclear risk of

bias) using different scales. It was not appropriate to conduct a

meta-analysis, however many of the studies used validated ques-

tionnaires and found that CSII was preferred over MI for treat-

ment satisfaction, quality of life and perception of better general

and mental health. Often, participants randomised to CSII pre-

ferred to continue with CSII at study completion rather than re-

turn to MI.

There was a statistically significant decrease in daily insulin re-

quirement measured in units (U) by nine studies (overall high risk

of bias). The mean difference was -7 U (95% CI -11 to -3, P =

0.0003, I2 statistic of 32%). There was an insufficient number of

studies for subgroup analyses for age and study duration. When

measured in units per kilogram (U/kg), the daily insulin require-

ment mean difference found among 12 studies (overall unclear

risk of bias) could not be reliably pooled due to considerable het-

erogeneity (I2 statistic of 94%).

No significant difference was found for weight (-0.4 kg (95% CI

-2 to 1.3), I2 statistic of 0%) among eight studies (overall unclear

risk of bias). Subgroup analysis for age did not significantly alter

the effect estimate, however subgroup analysis for study duration

showed that long term studies (seven studies) had a higher non-

significant effect estimate -1.6 kg (95% CI -7.7 to 4.5, I2 statistic

of 27%).

Information about adverse events other than non-severe and severe

hypoglycaemic events was not well reported. Most common other

adverse events included ketoacidosis, dehydration and injection or

infusion site complications. There were high numbers of injection

or infusion site complications, suggesting that training on the use

and care of these interventions is vital.

None of the studies reported on mortality, morbidity or costs.

Exploration of small study bias revealed that the effect estimates

for HbA1c, daily mean blood glucose, daily insulin requirement

(U/kg) and weight were higher than the effect estimates for these

outcomes when all studies were included, suggesting that small

study bias may be present for these outcomes.

Funnel plot asymmetry was found in all meta-analyses. It should

be noted that there are a number of explanations for the asymmetry

of a funnel plot, including true heterogeneity of effect with respect

to study size, poor methodological design of small studies and

publication bias (Sterne 2001).

Overall completeness and applicability ofevidence

There were many gaps in the information provided by the study

investigators about quality and data. Nineteen study investigators

were contacted for clarification of study design, data or both, and

53% (10 study investigators) replied to our questionnaire.

Six cross-over studies were not adequately analysed, and did not

report correlations between baseline and end-of-study data, there-

fore ignoring within-person variation. We reanalysed the data from

these studies assuming different correlation coefficients (0.3, 0.5,

0.7, 0.9) and have discussed the ’Effects of interventions’ based on

the most conservative approach, assuming a correlation coefficient

of 0.3. The data from these studies were also reanalysed assuming

the largest standard deviation of the differences derived from the

cross-over studies with adequate statistical methodology. The re-

sults from the analyses assuming different correlation coefficients

and the largest standard deviation of the differences (found in Ap-

pendix 3) do not vary substantially from that presented in ’Effects

of interventions’ for direction of effect and heterogeneity.

Participants in the included studies were both adults and children

with type 1 diabetes, suggesting that the results would be relevant

to a broad spectrum of age groups in similar contexts. None of the

studies were conducted in developing countries.

The mean reduction in HbA1c associated with CSII was relatively

small at 0.3 percentage points, and likely to be of only borderline

clinical relevance. However, the reduction frequency of severe hy-

poglycaemic events was of greater clinical relevance, with 8 out of

15 studies providing data on severe hypoglycaemia reporting a re-

duction of at least 50% in the frequency of severe hypoglycaemia.

Thus, CSII is likely to be a clinically useful alternative for those

individuals in whom there is concern about severe hypoglycaemia.



Quality of the evidence

Using the approach described in ’Assessment of risk of bias in

included studies’, the body of evidence was classified as high, un-

clear or low risk of bias for each outcome. Risk of bias was un-

clear for most of the evidence, but for studies reporting the out-

comes daily insulin requirement (U), fasting blood glucose and

post-prandial blood glucose, the body of evidence was at high risk

of bias.

Potential biases in the review process

This review consists of published data only. In future, updates will

attempt to identify non-published studies.

The selection criteria specified that MI required three or more

injections per day, therefore many studies were excluded because

their definition of MI was two or more injections per day. We

are uncertain whether these studies would have altered the effect

estimates seen in this review.

Many different scales and units were used to report measures of

non-severe and severe hypoglycaemia and quality of life. There

were insufficient studies to conduct meta-analyses for each of the

scales and units, and as a result, our interpretation of the overall

effects of the interventions on these outcomes is subjective and

open to bias.

Agreements and disagreements with otherstudies or reviews

A systematic review by Pickup et al (Pickup 2002), a health-tech-

nology assessment (HTA) report (HTA 2004) and a NICE report

(NICE 2008) were in agreement with the findings of this review.

The systematic review by Pickup et al found that glycosylated

haemoglobin A1c and mean blood glucose concentration were

lower in people receiving CSII compared with those receiving

MI. Blood glucose concentrations were less variable during CSII

and the insulin dose required to maintain glycaemic control was

reduced with CSII.

The HTA report found that the quality of included studies was

generally poor. When compared with at least three injections per

day (MI), CSII resulted in a modest but important improvement

in glycosylated haemoglobin A1c and a short term reduction in

insulin dose in adults with type 1 diabetes. Overall hypoglycaemic

events did not differ between CSII and MI. The authors of the

HTA review were unable to identify any economic evaluations

comparing these two interventions, however it is estimated that

the additional cost associated with using CSII over MI varies from

£1091 (1208 EURO, 11/2009 exchange rate) to £1680 (1860

EURO, 11/2009 exchange rate), per annum (including the cost

of the pump and consumables).

The NICE report focused on recommendations for when to

use CSII. The report included non-randomised and randomised

studies and found improvement in HbA1c and reduction in in-

sulin dose. No difference was found for hypoglycaemic events

and weight between the two interventions but patient preference

slightly favoured CSII.

A U T H O R S ’ C O N C L U S I O N S

Implications for practice

Both continuous subcutaneous insulin infusion (CSII) and mul-

tiple insulin injections (MI)CSII are currently used in practice.

There may be benefit in using CSII over MI for improving gly-

caemic control and improving health-related quality of life in peo-

ple with type 1 diabetes. Non-severe hypoglycaemic events do not

appear to be influenced differently by either intervention.

It is important to note that there is insufficient evidence regarding

adverse events, diabetes late complications, mortality and cost - all

essential considerations in deciding which treatment to administer

in practice. Until evidence is available for these outcomes we are

unable to make a recommendation about whether either interven-

tion is superior over the other in the management of glycaemic

control in people with type 1 diabetes.

Implications for research

Long term, large scale and methodologically rigorous studies are

required to determine the effect of CSII and MI on outcomes such

as hypoglycaemia, mortality, diabetes late complications and other

adverse effects, using validated scales. Cost effectiveness data are

also required to determine superiority between the two interven-

tions.

A C K N O W L E D G E M E N T S

The authors would like to acknowledge several colleagues for their

assistance in preparation of this review: Steve McDonald from

the Australasian Cochrane Centre and Karla Bergerhoff, the Tri-

als Search Coordinator of the Cochrane Metabolic and Endocrine

Disorders Group for their assistance in developing the search strat-

egy. Joanne McKenzie from the Australasian Cochrane Centre and

Damien Jolley from Monash Institute of Health Services Research

for biostatistical advice. Miranda Cumpston and Veronica Pitt

from the Australasian Cochrane Centre for methodological advice.

The authors would like to acknowledge the following study in-

vestigators for providing additional study information or data: JF

Bak, H Hanaire Broutin, RPLM Hoogma, JH DeVries, G Lepore,

N Weintrob and B Zinman (for Tsui 2001).



R E F E R E N C E S

References to studies included in this review

Bak 1987 and Husted 1989 {published and unpublished data}

Bak JF, Nielsen OH, Pedersen O, Beck-Nielsen H. Multiple

insulin injections using a pen injector versus insulin pump

treatment in young diabetic patients. Diabetes Research

1987;6(3):155–8.

Husted SE, Nielsen HK, Bak JF, Beck-Nielsen H.

Antithrombin III activity, von Willebrand factor antigen

and platelet function in young diabetic patients treated with

multiple insulin injections versus insulin pump treatment.

European Journal of Clinical Investigation 1989;19(1):90–4.

Botta 1986 {published data only}

Botta RM, Sinagra D, Angelico MC, Bompiani GD.

[Comparison of intensified traditional insulin therapy

and micropump therapy in pregnant women with type 1

diabetes mellitus]. Minerva Medica 1986;77(17):657–61.

Bruttomesso 2008 {published data only}

Bruttomesso D, Crazzolara D, Maran A, Costa S, Dal Pos

M, Girelli A, et al.In Type 1 diabetic patients with good

glycaemic control, blood glucose variability is lower during

continuous subcutaneous insulin infusion than during

multiple daily injections with insulin glargine. Diabetic

Medicine 2008;25(3):326–32.

Chiasson 1984 and 1985 {published data only}

Chiasson JL, Ducros F, Poliquin-Hamet M, Lopez D,

Lecavalier L, Hamet P. Continuous subcutaneous insulin

infusion (Mill-Hill Infuser) versus multiple injections

(Medi-Jector) in the treatment of insulin-dependent

diabetes mellitus and the effect of metabolic control on

microangiopathy. Diabetes Care 1984;7(4):331–7.

Cohen 2003 {published data only}

Cohen D, Weintrob N, Benzaquen H, Galatzer A, Fayman

G, Phillip M. Continuous subcutaneous insulin infusion

versus multiple daily injections in adolescents with type

I diabetes mellitus: a randomized open crossover trial.

Journal of pediatric endocrinology & metabolism 2003;16(7):

1047–50.

DeVries 2002 {published and unpublished data}

DeVries JH, Snoek FJ, Kostense PJ, Masurel N, Heine RJ.

A randomized trial of continuous subcutaneous insulin

infusion and intensive injection therapy in type 1 diabetes

for patients with long-standing poor glycemic control.

Diabetes Care 2002;25(11):2074–80.

Doyle 2004 {published data only}

Doyle EA, Weinzimer SA, Steffen AT, Ahern JA, Vincent

M, Tamborlane WV. A randomized, prospective trial

comparing the efficacy of continuous subcutaneous insulin

infusion with multiple daily injections using insulin

glargine. Diabetes Care 2004;27(7):1554–8.

Hanaire-Broutin 2000 {published and unpublished data}

Hanaire-Broutin H, Melki V, Bessieres-Lacombe S, Tauber

JP. Comparison of continuous subcutaneous insulin

infusion and multiple daily injection regimens using insulin

lispro in type 1 diabetic patients on intensified treatment: a

randomized study. The Study Group for the Development

of Pump Therapy in Diabetes. Diabetes Care 2000;23(9):

1232–5.

Hirsch 2005 {published data only}

Hirsch IB, Bode BW, Garg S, Lane WS, Sussman A, Hu

P, et al.Continuous subcutaneous insulin infusion (CSII)

of insulin aspart versus multiple daily injection of insulin

aspart/insulin glargine in type 1 diabetic patients previously

treated with CSII. Diabetes Care 2005;28(3):533–8.

Home 1982 {published data only}

Home PD, Capaldo B, Burrin JM, Worth R, Alberti

KG. A crossover comparison of continuous subcutaneous

insulin infusion (CSII) against multiple insulin injections in

insulin-dependent diabetic subjects: improved control with

CSII. Diabetes Care 1982;5(5):466–71.

Hoogma 2005 {published and unpublished data}

Hoogma RPLM, Hoekstra JB, Michels BP, Levi M.

Comparison between multiple daily insulin injection

therapy (MDI) and continuous subcutaneous insulin

infusion therapy (CSII), results of the five nations study.

Diabetes Research & Clinical Practice 2006;74 Suppl 2:

S144–7.

Lepore 2003 {published and unpublished data}

Lepore G, Dodesini AR, Nosari I, Trevisan R. Both

continuous subcutaneous insulin infusion and a multiple

daily insulin injection regimen with glargine as basal insulin

are equally better than traditional multiple daily insulin

injection treatment. Diabetes Care 2003;26(4):1321–2.

Meschi 1982 {published data only}

Meschi F, Beccaria L, Vanini R, Szulc M, Chiumello G.

Short-term subcutaneous insulin infusion in diabetic

children. Comparison with three daily insulin injections.

Acta Diabetologica Latina 1982;19(4):371–5.

Nathan 1982 {published data only}

Nathan Dm LPAJ. Intensive conventional and insulin

pump therapies in adult type I diabetes. A crossover study.

Annals of Internal Medicine 1982;97:31–6.

Nosadini 1988 {published data only}

Nosadini R. Frequency of hypoglycaemic and

hyperglycaemic-ketotic episodes during conventional and

subcutaneous continuous insulin infusion therapy in

IDDM. Diabetes Nutrition and Metabolism 1988;1:289–96.

Nuboer 2008 {published data only}

Nuboer R, Borsboom GJJM, Zoethout JA, Koot HM,

Bruining J. Effects of insulin pump vs. injection treatment

on quality of life and impact of disease in children with

type 1 diabetes mellitus in a randomized, prospective

comparison. Pediatric Diabetes 2008;9(4 Pt 1):291–6.

Oslo Study 1985 to 1992 {published data only}

Brinchmann-Hansen O, Dahl-Jorgensen K, Hanssen KF.

Effects of intensified insulin treatment on various lesions of



diabetic retinopathy. American Journal of Ophthalmology

1985;100(5):644–53.

Brinchmann-Hansen O, Dahl-Jorgensen K, Hanssen KF,

Sandvik L. Oscillatory potentials, macular recovery time,

and diabetic retinopathy through 3 years of intensified

insulin treatment.[see comment]. Ophthalmology 1988;95

(10):1358–66.

Brinchmann-Hansen O, Dahl-Jorgensen K, Hanssen KF,

Sandvik L. The response of diabetic retinopathy to 41

months of multiple insulin injections, insulin pumps, and

conventional insulin therapy. Archives of Ophthalmology

1988;106(9):1242–6.

Dahl Jorgensen K. Blood glucose control and progression of

diabetic neuropathy: eight years results from the Oslo study.

Diabetologia 1992;35:A15.

Dahl-Jorgensen K. Near-normoglycemia and late diabetic

complications. The Oslo Study. Acta Endocrinologica

Supplementum 1987;284:1–38.

Dahl-Jorgensen K, Bjoro T, Kierulf P, Sandvik L, Bangstad

HJ, Hanssen KF. Long-term glycemic control and kidney

function in insulin-dependent diabetes mellitus. Kidney

International 1992;41(4):920–3.

Dahl-Jorgensen K, Brinchmann-Hansen O, Hanssen KF.

Rapid tightening of blood glucose control leads to transient

deterioration of retinopathy in insulin dependent diabetes

mellitus: The Oslo study. British Medical Journal 1985;290

(6471):811–5.

Dahl-Jorgensen K, Brinchmann-Hansen O, Hanssen

KF, Ganes T, Kierulf P, Smeland E, et al.Effect of near

normoglycaemia for two years on progression of early

diabetic retinopathy, nephropathy, and neuropathy: the

Oslo study. British Medical Journal Clinical ResearchEducation 1986;293(6556):1195–9.

Dahl-Jorgensen K, Hanssen KF, Aagenaes O, Larsen S.

[New methods for subcutaneous insulin administration. A

year’s experience with the insulin pump and multiple insulin

injection therapy]. Tidsskrift for Den Norske Laegeforening

1984;104(13):856–61.

Dahl-Jorgensen K, Hanssen KF, Kierulf P, Bjoro T, Sandvik

L, Aagenaes O. Reduction of urinary albumin excretion

after 4 years of continuous subcutaneous insulin infusion in

insulin-dependent diabetes mellitus. The Oslo Study. ActaEndocrinologica 1988;117(1):19–25.

Dahl-Jorgensen K, Torjesen P, Hanssen KF, Sandvik

L, Aagenaes O. Increase in insulin antibodies during

continuous subcutaneous insulin infusion and multiple-

injection therapy in contrast to conventional treatment.

Diabetes 1987;36(1):1–5.

Hanssen KF, Dahl-Jorgensen K, Brinchmann-Hansen O.

The influence of strict control on diabetic complications.

Acta Endocrinologica Supplementum 1985;272:57–60.

Pozzilli 2003 {published data only}

Pozzilli P, Crino A, Schiaffini R, Manfrini S, Fioriti E,

Coppolino G, et al.A 2-year pilot trial of continuous

subcutaneous insulin infusion versus intensive insulin

therapy in patients with newly diagnosed type 1 diabetes

(IMDIAB 8). Diabetes Technology & Therapeutics 2003;5

(6):965–74.

Saurbrey 1988 {published data only}

Saurbrey N, Arnold-Larsen S, Moller-Jensen B, Kuhl C.

Comparison of continuous subcutaneous insulin infusion

with multiple insulin injections using the NovoPen.

Diabetic Medicine 1988;5(2):150–3.

Schmitz 1989 {published data only}

Schmitz A, Christiansen JS, Christensen CK, Hermansen

K, Mogensen CE. Effect of pump versus pen treatment

on glycaemic control and kidney function in long-term

uncomplicated insulin-dependent diabetes mellitus

(IDDM). Danish Medical Bulletin 1989;36(2):176–8.

Skogsberg 2008 {published data only}

Skogsberg L, Fors H, Hanas R, Chaplin JE, Lindman

E, Skogsberg J. Improved treatment satisfaction but no

difference in metabolic control when using continuous

subcutaneous insulin infusion vs. multiple daily injections

in children at onset of type 1 diabetes mellitus. Pediatric

Diabetes 2008;9(5):472–9.

Tsui 2001 {published and unpublished data}

Tsui E, Barnie A, Ross S, Parkes R, Zinman B. Intensive

insulin therapy with insulin lispro: a randomized trial of

continuous subcutaneous insulin infusion versus multiple

daily insulin injection.[see comment]. Diabetes Care 2001;

24(10):1722–7.

Weintrob 2004 {published and unpublished data}

Weintrob N, Schechter A, Benzaquen H, Shalitin S, Lilos P,

Galatzer A, et al.Glycemic patterns detected by continuous

subcutaneous glucose sensing in children and adolescents

with type 1 diabetes mellitus treated by multiple daily

injections vs continuous subcutaneous insulin infusion.

Archives of Pediatrics & Adolescent Medicine 2004;158(7):

677–84.

References to studies excluded from this review

Alemzadeh 2005 {published data only}

Alemzadeh R, Palma-Sisto P, Parton EA, Holzum MK.

Continuous subcutaneous insulin infusion and multiple

dose of insulin regimen display similar patterns of blood

glucose excursions in pediatric type 1 diabetes. Diabetes

Technology & Therapeutics 2005;7(4):587–96.

Amemiya 1983 {published data only}

Amemiya S, Kato K, Asayama K. The improved response

in endogenous insulin due to continuous subcutaneous

infusion of insulin therapy in juvenile diabetes. Tohoku

Journal of Experimental Medicine 1983;141 Suppl:713–7.

Arias 1985 {published data only}

Arias P, Kerner W, Zier H, Navascues I, Pfeiffer EF.

Incidence of hypoglycemic episodes in diabetic patients

under continuous subcutaneous insulin infusion and

intensified conventional insulin treatment: assessment

by means of semiambulatory 24-hour continuous blood

glucose monitoring. Diabetes Care 1985;8(2):134–40.

Bagdade 1991 {published data only}

Bagdade JD, Helve E, Taskinen MR. Effects of continuous

insulin infusion therapy on lipoprotein surface and core



lipid composition in insulin-dependent diabetes mellitus.

Metabolism: Clinical & Experimental 1991;40(5):445–9.

Bangstad 1994 {published data only}

Bangstad HJ, Osterby R, Dahl-Jorgensen K, Berg KJ,

Hartmann A, Hanssen KF. Improvement of blood glucose

control in IDDM patients retards the progression of

morphological changes in early diabetic nephropathy.

Diabetologia 1994;37(5):483–90.

Barbosa 1981 {published data only}

Barbosa J, Menth L, Eaton J, Sutherland D, Freier EF,

Najarian J. Long-term, ambulatory, subcutaneous insulin

infusion versus multiple daily injections in brittle diabetic

patients. Diabetes Care 1981;4(2):269–74.

Barrett 1995 {published data only}

Barrett DH, Davidson PC, Steed LJ, Abel GG, Loman KE,

Saudek CD. Evaluation of the psychosocial impact of the

MiniMed variable-rate implantable insulin pump. Southern

Medical Journal 1995;88(12):1226–30.

Bech 1996 {published data only}

Bech P, Gudex C, Staehr Johansen K. The WHO (Ten)

well-being index: Validation in diabetes. Psychotherapy and

Psychosomatics 1996;65(4):183–90.

Beck-Nielsen 1984 {published data only}

Beck-Nielsen H, Richelsen B, Hasling C, Nielsen OH,

Heding L, Sorensen NS. Improved in vivo insulin effect

during continuous subcutaneous insulin infusion in patients

with IDDM. Diabetes 1984;33(9):832–7.


Beck-Nielsen H, Richelsen B, Mogensen CE, Olsen T,

Ehlers N, Nielsen CB, et al.Effect of insulin pump treatment

for one year on renal function and retinal morphology in

patients with IDDM. Diabetes Care 1985;8(6):585–9.

Beck-Nielsen 1985a {published data only}

Beck-Nielsen H, Richelsen B, Sorensen NS, Nielsen OH.

Insulin pump treatment: Effect on glucose homeostasis,

metabolites, hormones, insulin antibodies and quality of

life. Diabetes Research 1985;2(1):37–43.


Beck-Nielsen H, Olesen T, Mogensen CE, Richelsen B,

Olsen HW, Ehlers N, et al.Effect of near normoglycemia

for 5 years on progression of early diabetic retinopathy and

renal involvement. Diabetes Research 1990;15(4):185–90.

Behme 1984 {published data only}

Behme MT, Dupre J. Insulin-binding to erythrocytes in

type I diabetes mellitus: effects of continuous subcutaneous

infusion of insulin. Clinical & Investigative Medicine -

Medecine Clinique et Experimentale 1984;7(2):109–14.

Bell 1985 {published data only}

Bell PM, Hayes JR, Hadden DR, Archer DB. The effect of

plasma glucose control by continuous subcutaneous insulin

infusion or conventional therapy on retinal morphology and

urinary albumin excretion. Diabete et Metabolisme 1985;11

(4):254–61.

Bell 1985a {published data only}

Bell PM, Sawhney B, Hayes JR, Hadden DR. Effect of

plasma glucose control by continuous subcutaneous insulin

infusion on nerve conduction. Irish Journal of MedicalScience 1985;154(10):378–84.

Bending 1986 {published data only}

Bending JJ, Viberti GC, Watkins PJ, Keen H. Intermittent

clinical proteinuria and renal function in diabetes: evolution

and the effect of glycaemic control. British Medical JournalClinical Research Ed 1986;292(6513):83–6.

Berg 1998 {published data only}

Berg TJ, Nourooz-Zadeh J, Wolff SP, Tritschler HJ,

Bangstad HJ, Hanssen KF. Hydroperoxides in plasma are

reduced by intensified insulin treatment. A randomized

controlled study of IDDM patients with microalbuminuria.

Diabetes Care 1998;21(8):1295–300.

Bergenstal 1985 {published data only}

Bergenstal RM, Dupre J, Lawson PM, Rizza RA, Rubenstein

AH. Observations on C-peptide and free insulin in the

blood during continuous subcutaneous insulin infusion and

conventional insulin therapy. Diabetes 1985;34 Suppl 3:

31–6.

Berger 1989 {published data only}

Berger M, Chantelau EA, Spraul M. [Treatment of type

I diabetes mellitus with continuous subcutaneous insulin

infusion: indications and quality control]. Verhandlungender Deutschen Gesellschaft fur Innere Medizin 1989;95:

135–40.

Bibergeil 1987 {published data only}

Bibergeil H, Huttl I, Felsing W, Felsing U, Seidlein I,

Herfurth S, et al.36 months continuous subcutaneous

insulin infusion (CSII) in insulin dependent diabetes

(IDDM)--influence on early stages of retinopathy,

nephropathy and neuropathy--psychological analysis.

Experimental & Clinical Endocrinology 1987;90(1):51–61.

Biesenbach 1988 {published data only}

Biesenbach G, Grafinger P, Kaiser W, Stuby U, Zazgornik

J. [Metabolic control in labile type I diabetes with

conventional insulin therapy, basal bolus insulin therapy

using pens and continuous subcutaneous insulin infusion

with the pump]. Medizinische Klinik 1988;83(12):

398–401.

Bode 1996 {published data only}

Bode BW, Steed RD, Davidson PC. Reduction in severe

hypoglycemia with long-term continuous subcutaneous

insulin infusion in type I diabetes. Diabetes Care 1996;19

(4):324–7.

Bode 2004 {published data only}

Bode B, Shelmet J, Gooch B, Hassman DR, Liang J,

Smedegaard JK, et al.Patient perception and use of an

insulin injector/glucose monitor combined device. Diabetes

Educator 2004;30(2):301–9.

Bombor 1984 {published data only}

Bombor H, Bruns W, Ratzmann KP, Jutzi E, Albrecht G,

Michaelis D, et al.Long-term improvement in metabolic

control of unstable type I diabetes by s.c. insulin injection



patterns based on the dose profiles required by bed-side

artificial beta-cell. Experimental & Clinical Endocrinology1984;83(2):143–51.

Borisova 1989 {published data only}

Borisova AM, Koev D, Kirilov G. [An improvement in

insulin efficacy in patients with type-1 diabetes mellitus

following treatment using an insulin infusion pump].

Vutreshni Bolesti 1989;28(1):74–7.

Brunetti 1984 {published data only}

Brunetti P, Bueti A, Antonella MA, Calabrese G, Fabietti

PG, Santeusanio F, et al.Metabolic effects of intensified

insulin therapy. Experimental & Clinical Endocrinology1984;83(2):130–5.

Buysschaert 1981 {published data only}

Buysschaert M, Lejeune M, Lambert AE. Continuous

subcutaneous insulin infusion in the treatment of juvenile-

onset diabetes. Diabete et Metabolisme 1981;7(3):189–93.

Buysschaert 1983 {published data only}

Buysschaert M, Marchand E, Ketelslegers JM, Lambert

AE. Comparison of plasma glucose and plasma free insulin

during CSII and intensified conventional insulin therapy.

Diabetes Care 1983;6(1):1–5.

Calabrese 1982 {published data only}

Calabrese G, Bueti A, Santeusanio F, Giombolini A, Zega

G, Angeletti G, et al.Continuous subcutaneous insulin

infusion treatment in insulin-dependent diabetic patients:

a comparison with conventional optimized treatment in a

long-term study. Diabetes Care 1982;5(5):457–65.

Canny 1985 {published data only}

Canny CL, Kohner EM, Trautman J, Puklin J, Morse P.

Comparison of stereofundus photographs in patients with

insulin-dependent diabetes during conventional insulin

treatment or continuous subcutaneous insulin infusion.

Diabetes 1985;34 Suppl 3:50–5.

Carta 1986 {published data only}

Carta Q, Meriggi E, Trossarelli GF, Catella G, Dal Molin V,

Menato G, et al.Continuous subcutaneous insulin infusion

versus intensive conventional insulin therapy in type I and

type II diabetic pregnancy. Diabete et Metabolisme 1986;12

(3):121–9.

Cavallo-Perin 1983 {published data only}

Cavallo-Perin P, Pagano G, Tagliaferro V, Jarre P, Ozzello

A, Lenti G. The treatment of unstable type I diabetes:

conventional versus portable pump insulin administration.

Acta Diabetologica Latina 1983;20(4):363–70.

Chisholm 1984 {published data only}

Chisholm DJ, Kraegen EW, Hewett MJ. Hypoglycemic

episodes during continuous subcutaneous insulin infusion:

decreased frequency but increased susceptibility. Australian& New Zealand Journal of Medicine 1984;14(3):255–9.

Christensen 1986 {published data only}

Christensen CK, Christiansen JS, Christensen T,

Hermansen K, Mogensen CE. The effect of six months

continuous subcutaneous insulin infusion on kidney

function and size in insulin-dependent diabetics. Diabetic

Medicine 1986;3(1):29–32.

Christensen 1987 {published data only}

Christensen CK, Christiansen JS, Schmitz A, Christensen

T, Hermansen K, Mogensen CE. Effect of continuous

subcutaneous insulin infusion on kidney function and size

in IDDM patients: a 2 year controlled study. Journal of

Diabetic Complications 1987;1(3):91–5.

Christiansen 1987 {published data only}

Christiansen JS, Ingerslev J, Bernvil SS, Christensen CK,

Hermansen K, Schmitz A. Near normoglycemia for 1 year

has no effect on platelet reactivity, factor VIII, and von

Willebrand factor in insulin-dependent diabetes mellitus: a

controlled trial. Journal of Diabetic Complications 1987;1

(3):100–6.

Ciavarella 1985 {published data only}

Ciavarella A, Vannini P, Flammini M, Bacci L, Forlani G,

Borgnino LC. Effect of long-term near-normoglycemia

on the progression of diabetic nephropathy. Diabete et

Metabolisme 1985;11(1):3–8.

Colette 1986 {published data only}

Colette C, Monnier L, Arnal J, Selam JL, Mirouze J. Effects

of different insulin administration modalities on vitamin

D metabolism of insulin-dependent diabetic patients.

Advances in Experimental Medicine & Biology 1986;208:

501–8.

Colette 1989 {published data only}

Colette C, Pares-Herbute N, Monnier L, Selam JL, Thomas

N, Mirouze J. Effect of different insulin administration

modalities on vitamin D metabolism of insulin-dependent

diabetic patients. Hormone & Metabolic Research 1989;21

(1):37–41.

Connis 1989 {published data only}

Connis RT, Taylor TR, Gordon MJ, Mecklenburg RS,

Liljenquist JE, Stephens JW, et al.Changes in cognitive and

social functioning of diabetic patients following initiation of

insulin infusion therapy. Experimental Aging Research 1989;

15(1-2):51–60.

Coustan 1986 {published data only}

Coustan DR, Reece EA, Sherwin RS, Rudolf MC, Bates

SE, Sockin SM, et al.A randomized clinical trial of the

insulin pump vs intensive conventional therapy in diabetic

pregnancies. JAMA 1986;255(5):631–6.

Davies 1984 {published data only}

Davies AG, Price DA, Houlton CA, Burn JL, Fielding BA,

Postlethwaite RJ. Continuous subcutaneous insulin infusion

in diabetes mellitus. A year’s prospective trial. Archives of

Disease in Childhood 1984;59(11):1027–33.

de Beaufort 1985 {published data only}

de Beaufort CE, Bruining GJ, Aarsen RS, den Boer NC,

Grose WF. Does continuous subcutaneous insulin infusion

(CSII) prolong the remission phase of insulin-dependent

diabetic children? Preliminary findings of a randomized

prospective study. Netherlands Journal of Medicine 1985;28

Suppl 1:53–4.



de Beaufort 1989 {published data only}

de Beaufort CE, Houtzagers CM, Bruining GJ, Aarsen RS,

den Boer NC, Grose WF, et al.Continuous subcutaneous

insulin infusion (CSII) versus conventional injection

therapy in newly diagnosed diabetic children: two-year

follow-up of a randomized, prospective trial. DiabeticMedicine 1989;6(9):766–71.

Dicker 1987 {published data only}

Dicker D, Feldberg D, Karp M, Yeshaya A, Samuel N,

Goldman JA. Preconceptional diabetes control in insulin-

dependent diabetes mellitus patients with continuous

subcutaneous insulin infusion therapy. Journal of PerinatalMedicine 1987;15(2):161–7.

DiMeglio 2004 {published data only}

DiMeglio LA, Pottorff TM, Boyd SR, France L, Fineberg

N, Eugster EA. A randomized, controlled study of insulin

pump therapy in diabetic preschoolers. Journal of Pediatrics

2004;145(3):380–4.

Edelmann 1987 {published data only}

Edelmann E, Walter H, Biermann E, Schleicher E,

Bachmann W, Mehnert H. Sustained normoglycemia and

remission phase in newly diagnosed type I diabetic subjects.

Comparison between continuous subcutaneous insulin

infusion and conventional therapy during a one year follow-

up. Hormone & Metabolic Research 1987;19(9):419–21.

Feldt-Rasmussen 1986 {published data only}

Feldt-Rasmussen B, Mathiesen ER, Deckert T. Effect of two

years of strict metabolic control on progression of incipient

nephropathy in insulin-dependent diabetes. Lancet 1986;

328(8519):1300–4.

Feldt-Rasmussen 1986a {published data only}

Feldt-Rasmussen B, Mathiesen ER, Hegedus L, Deckert

T. Kidney function during 12 months of strict metabolic

control in insulin-dependent diabetic patients with incipient

nephropathy. New England Journal of Medicine 1986;314

(11):665–70.

Feldt-Rasmussen 1991 {published data only}

Feldt-Rasmussen B, Mathiesen ER, Jensen T, Lauritzen T,

Deckert T. Effect of improved metabolic control on loss

of kidney function in type 1 (insulin-dependent) diabetic

patients: an update of the Steno studies. Diabetologia 1991;

34(3):164–70.

Flores 1984 {published data only}

Flores d’Arcais A, Morandi F, Beccaria L, Meschi F,

Chiumello G. Metabolic control in newly diagnosed type

1 diabetic children. Effect of continuous subcutaneous

infusion. Hormone Research 1984;19(2):65–9.

Garg 2004 {published data only}

Garg SK, Walker AJ, Hoff HK, D’Souza AO, Gottlieb

PA, Chase HP. Glycemic parameters with multiple daily

injections using insulin glargine versus insulin pump.[see

comment]. Diabetes Technology & Therapeutics 2004;6(1):

9–15.

Garmo 2004 {published data only}

Garmo A, Pettersson-Frank B, Ehrenberg A. Treatment

effects and satisfaction in diabetic patients changing from

multiple daily insulin injections to CSII. Practical Diabetes

International 2004;21(1):7–12.

Giocolea 1988 {published data only}

Giocolea I, Hernandez I, Fombellida J, Vazquez JA. Renal

function of insulin-dependent diabetic patients, treated

with continuous subcutaneous insulin infusion pump. A

comparison with conventional insulin therapy during a

follow-up period of one year. Anales de Medicina Interna1988;5(4):169–72.

Goicolea 1986 {published data only}

Goicolea Opacua I, Hernandez Colau I, Vazquez Garcia JA.

[Comparative study between the subcutaneous continuous

insulin infusion pump and optimized conventional

treatment. Effects at 6 months]. Revista Clinica Espanola1986;179(1):3–7.


Goicolea Opacua I, Hernandez Colau I, Cortazar Galarza

A, Vazquez Garcia JA. [Comparison of metabolic control

between the continuous subcutaneous insulin infusion

pump and augmented conventional treatment. Effects after

12 months]. Medicina Clinica 1987;88(16):617–20.


Goicolea Opacua I, Martinez Castillo A, Santamaria Pelarda

E, Sola Sarabia C, Vazquez Garcia JA. [Metabolic control

using a subcutaneous insulin infusion pump: effects 2 years

after treatment]. Revista Clinica Espanola 1988;182(4):

200–2.


Goicolea I, Santamaria E, Pardo C, Vazquez JA, Martin A,

Castresana I. [Course of diabetic retinopathy and metabolic

control with subcutaneous insulin infusion pump: 18-

month study]. Anales de Medicina Interna 1989;6(2):71–3.

Greene 1983 {published data only}

Greene SA, Smith MA, Baum JD. Clinical application of

insulin pumps in the management of insulin dependent

diabetes. Archives of Disease in Childhood 1983;58(8):

578–81.

Grimm 1987 {published data only}

Grimm JJ, Haardt MJ, Thibult N, Goicolea I, Tchobroutsky

G, Slama G. Lifestyle, metabolic control and social

implications of pump therapy in 54 routine type I diabetic

patients. Diabete et Metabolisme 1987;13(1):3–11.

Guerci 1998 {published data only}

Guerci B, Meyer L, Delbachian I, Kolopp M, Ziegler O,

Drouin P. Blood glucose control on Sunday in IDDM

patients: intensified conventional insulin therapy versus

continuous subcutaneous insulin infusion. Diabetes Research& Clinical Practice 1998;40(3):175–80.

Gulan 1988 {published data only}

Gulan M, Perlman K, Sole M, Albisser AM, Zinman B.

Counterregulatory hormone responses preserved after long-

term intravenous insulin infusion compared to continuous

subcutaneous insulin infusion. Diabetes 1988;37(5):

526–31.



Haakens 1989 {published data only}

Haakens K, Hanssen KF, Dahl-Jorgensen K, Vaaler S,

Torjesen P, Try K. Early morning glycaemia and the

metabolic consequences of delaying breakfast/morning

insulin. A comparison of continuous subcutaneous insulin

infusion and multiple injection therapy with human

isophane or human ultralente insulin at bedtime in insulin-

dependent diabetics. Scandinavian Journal of Clinical &Laboratory Investigation 1989;49(7):653–9.

Haakens 1990 {published data only}

Haakens K, Hanssen KF, Dahl-Jorgensen K, Vaaler S,

Aagenaes O, Mosand R. Continuous subcutaneous insulin

infusion (CSII), multiple injection (MI) and conventional

insulin therapy (CT) in self-selecting insulin-dependent

diabetic patients. A comparison of metabolic control, acute

complications and patient preferences. Journal of Internal

Medicine 1990;228(5):457–64.

Hall 1989 {published data only}

Hall K, Johansson BL, Povoa G, Thalme B. Serum levels

of insulin-like growth factor (IGF) I, II and IGF binding

protein in diabetic adolescents treated with continuous

subcutaneous insulin infusion. Journal of Internal Medicine1989;225(4):273–8.

Haug 1987 {published data only}

Haug PJ, Kelly TM, Cannon RB, Edwards CQ. A self-

controlled study of the effect of continuous subcutaneous

insulin infusion on diabetic neuropathy. Journal of theNeurological Sciences 1987;82(1-3):123–32.

Heber 1977 {published data only}

Heber D, Molitch ME, Sperling MA. Low-dose continuous

insulin therapy for diabetic ketoacidosis. Prospective

comparison with “conventional” insulin therapy. Archives ofInternal Medicine 1977;137(10):1377–80.

Helve 1987 {published data only}

Helve E. High density lipoprotein subfractions during

continuous insulin infusion therapy. Atherosclerosis 1987;64

(2-3):173–80.

Helve 1987a {published data only}

Helve E, Koivisto VA, Lehtonen A. A crossover comparison

of continuous insulin infusion and conventional injection

treatment of type I diabetes. Acta Medica Scandinavica

1987;221(4):385–93.

Helve 1987b {published data only}

Helve E, Laatikainen L, Merenmies L, Koivisto VA.

Continuous insulin infusion therapy and retinopathy in

patients with type I diabetes. Acta Endocrinologica 1987;

115(3):313–9.

Hermansen 1987 {published data only}

Hermansen K, Moller A, Christensen CK, Christiansen

JS, Schmitz O, Orskov H, et al.Diurnal plasma profiles

of metabolite and hormone concentration in insulin-

dependent diabetic patients during conventional insulin

treatment and continuous subcutaneous insulin infusion. A

controlled study. Acta Endocrinologica 1987;114(3):433–9.

Hermansen 1988 {published data only}

Hermansen K, Schmitz O, Boye N, Christensen CK,

Christiansen JS, Alberti KG, et al.Glucagon responses to

intravenous arginine and oral glucose in insulin-dependent

diabetic patients during six months conventional or

continuous subcutaneous insulin infusion. Metabolism:Clinical & Experimental 1988;37(7):640–4.

Hoogma 2004 {published data only}

Hoogma RP, Spijker AJ, van Doorn-Scheele M, van Doorn

TT, Michels RP, van Doorn RG, et al.Quality of life and

metabolic control in patients with diabetes mellitus type

1 treated by continuous subcutaneous insulin infusion or

multiple daily insulin injections.[see comment]. Netherlands

Journal of Medicine 2004;62(10):383–7.

Hung 1984 {published data only}

Hung J, Menth L, Thompson MJ, Saner B, Rao KV,

Barbosa J. The Minnesota Diabetes Complications Clinical

Trial cognitive functions under long-term maximized and

standard metabolic controls. Diabete et Metabolisme 1984;

10(1):48–51.

Jakobsen 1988 {published data only}

Jakobsen J, Christiansen JS, Kristoffersen I, Christensen

CK, Hermansen K, Schmitz A, et al.Autonomic and

somatosensory nerve function after 2 years of continuous

subcutaneous insulin infusion in type I diabetes. Diabetes1988;37(4):452–5.

Jensen 1986 {published data only}

Jensen BM, Kuhl C, Molsted-Pedersen L. Preconceptional

treatment with insulin infusion pumps in insulin-dependent

diabetic women with particular reference to prevention of

congenital malformations. Acta Endocrinologica, Supplement

1986;112(277):81–5.

Kamoi 2004 {published data only}

Kamoi K, Miyakoshi M, Maruyama R. A quality-of-life

assessment of intensive insulin therapy using insulin lispro

switched from short-acting insulin and measured by an ITR-

QOL questionnaire: a prospective comparison of multiple

daily insulin injections and continuous subcutaneous

insulin infusion. Diabetes Research & Clinical Practice 2004;

64(1):19–25.

Kelly 1984 {published data only}

Kelly TM, Sanborn GE, Haug PJ, Edwards CQ. Effect of

insulin infusion pump use on diabetic retinopathy. Archivesof Ophthalmology 1984;102(8):1156–9.

Knight 1986 {published data only}

Knight G, Boulton AJM, Drury J, Ward JD. Long term

glycaemic control by alternative regimens in a feasibility

study of continuous subcutaneous insulin infusion. DiabetesResearch 1986;3(7):355–8.

Kobayashi 1987 {published data only}

Kobayashi T, Sawano S, Itoh T, Kosaka K. Plasma

immunoreactive somatostatin response to arginine after

glycemic control with continuous subcutaneous insulin

infusion in type I diabetics. Diabetes Care 1987;10(3):

286–92.



Koivisto 1983 {published data only}

Koivisto VA, Tronier B. Postprandial blood glucose response

to exercise in type I diabetes: comparison between pump

and injection therapy. Diabetes Care 1983;6(5):436–40.

Kordonouri 2006 {published data only}

Kordonouri O, Hartmann R, Lauterborn R, Barnekow C,

Hoeffe J, Deiss D. Age-specific advantages of continuous

subcutaneous insulin infusion as compared with multiple

daily injections in pediatric patients: one-year follow-up

comparison by matched-pair analysis. Diabetes Care 2006;

29(1):133–4.

Kritz 1983 {published data only}

Kritz H, Hagmuller G, Lovett R, Irsigler K. Implanted

constant basal rate insulin infusion devices for Type 1

(insulin-dependent) diabetic patients. Diabetologia 1983;25

(2):78–81.

Kroc 1984 {published data only}

Kroc. Blood glucose control and the evolution of diabetic

retinopathy and albuminuria. A preliminary multicenter

trial. The Kroc Collaborative Study Group. The NewEngland journal of medicine 1984;311:365–72.

Kroc 1988 {published data only}

Kroc. Diabetic retinopathy after two years of intensified

insulin treatment. Follow-up of the Kroc Collaborative

Study. The Kroc Collaborative Study Group. JAMA :the journal of the American Medical Association 1988;260:

37–41.

Kuno 1996 {published data only}

Kuno T, Tasaki H, Miyazaki S. Continuous subcutaneous

insulin injection for self-care of young patients with insulin-

dependent diabetes mellitus. Acta Paediatrica Japonica1996;38(5):464–9.

Laatikainen 1987 {published data only}

Laatikainen L, Teramo K, Hieta-Heikurainen H, Koivisto

V, Pelkonen R. A controlled study of the influence of

continuous subcutaneous insulin infusion treatment on

diabetic retinopathy during pregnancy. Acta Medica

Scandinavica 1987;221(4):367–76.

Lager 1983 {published data only}

Lager I, Lonnroth P, von Schenck H, Smith U. Reversal of

insulin resistance in type I diabetes after treatment with

continuous subcutaneous insulin infusion. British MedicalJournal Clinical Research Ed 1983;287(6406):1661–4.

Lager 1986 {published data only}

Lager I, Attvall S, Blohme G, Smith U. Altered recognition

of hypoglycaemic symptoms in type I diabetes during

intensified control with continuous subcutaneous insulin

infusion. Diabetic Medicine 1986;3(4):322–5.

Lapolla 2003 {published data only}

Lapolla A, Dalfra MG, Masin M, Bruttomesso D, Piva I,

Crepaldi C, et al.Analysis of outcome of pregnancy in type 1

diabetics treated with insulin pump or conventional insulin

therapy. Acta Diabetologica 2003;40(3):143–9.

Lauritzen 1985 {published data only}

Lauritzen T, Frost-Larsen K, Larsen HW, Deckert T. Two-

year experience with continuous subcutaneous insulin

infusion in relation to retinopathy and neuropathy. Diabetes

1985;34 Suppl 3:74–9.

Lawson 1984 {published data only}

Lawson P, Trayner I, Rosenstock J, Kohner E. The effect of

continuous subcutaneous insulin infusion on serum lipids.

Diabete et Metabolisme 1984;10(4):239–44.

Lawson 1985 {published data only}

Lawson P, Home PD, Bergenstal R. Observations on blood

lipid and intermediary metabolite concentrations during

conventional insulin treatment or CSII. Diabetes 1985;34

Suppl 3:27–30.

Leblanc 1986 {published data only}

Leblanc H, Obadia G, Passa P. [Reduction of hypoglycemia

in 10 diabetic patients treated with continuous subcutaneous

infusion of insulin]. Presse Medicale 1986;15(10):463–6.

Lecavalier 1987 {published data only}

Lecavalier L, Havrankova J, Hamet P, Chiasson J-L. Effects

of continuous subcutaneous insulin infusion versus multiple

injections on insulin receptors in insulin-dependent

diabetics. Diabetes Care 1987;10(3):300–5.

Lepore 2004 {published data only}

Lepore G, Dodesini AR, Nosari I, Trevisan R. Effect of

continuous subcutaneous insulin infusion vs multiple daily

insulin injection with glargine as basal insulin: an open

parallel long-term study. Diabetes, Nutrition & Metabolism -Clinical & Experimental 2004;17(2):84–9.

Levy 1984 {published data only}

Levy I, Bergua M, Esmatjes E, Halperin I, Figuerola D.

[Unstable type I diabetes mellitus: comparative study

between intensive conventional treatment and continuous

subcutaneous infusion of insulin]. Medicina Clinica 1984;

83(13):525–8.

Levy-Marchal 1983 {published data only}

Levy-Marchal C, Albisser AM, Zinman B. Overnight

metabolic control with pulsed intermittent versus

continuous subcutaneous insulin infusion. Diabetes Care1983;6(4):356–60.

Levy-Marchal 1988 {published data only}

Levy-Marchal C, Czernichow P. Feasibility of continuous

subcutaneous insulin infusion in young diabetic patients.


Litton 2002 {published data only}

Litton J, Rice A, Friedman N, Oden J, Lee MM, Freemark

M. Insulin pump therapy in toddlers and preschool children

with type 1 diabetes mellitus. Journal of Pediatrics 2002;141

(4):490–5.

Ludvigsson 2003 {published data only}

Ludvigsson J, Hanas R. Continuous subcutaneous glucose

monitoring improved metabolic control in pediatric patients

with type 1 diabetes: a controlled crossover study. Pediatrics2003;111(5 Pt 1):933–8.

Lunetta 1996 {published data only}

Lunetta M, De Mauro M, Le Moli R, Nicoletti F. Effect

of octreotide on blood glucose and counterregulatory

hormones in insulin-dependent diabetic patients: the role



of dose and route of administration. European Journal of

Clinical Pharmacology 1996;51(2):139–44.

Mancuso 1986 {published data only}

Mancuso S, Caruso A, Lanzone A. Continuous

subcutaneous insulin infusion (CSII) in pregnant diabetic

women. Acta Endocrinologica, Supplement 1986;112(277):

112–6.

Marshall 1987 {published data only}

Marshall SM, Home PD, Taylor R, Alberti KG. Continuous

subcutaneous insulin infusion versus injection therapy:

a randomized cross-over trial under usual diabetic clinic

conditions. Diabetic Medicine 1987;4(6):521–5.

Marshall 1988 {published data only}

Marshall SM, Taylor R, Home PD, Alberti KG.

Intermediary metabolism, insulin sensitivity and insulin

receptor status under comparable long-term therapy with

insulin injections and continuous subcutaneous insulin

infusion. Acta Endocrinologica 1988;117(4):417–27.

McDonald 1982 {published data only}

McDonald JW, Dupre J, Rodger NW, Champion MC,

Webb CD, Ali M. Comparison of platelet thromboxane

synthesis in diabetic patients on conventional insulin

therapy and continuous insulin infusions. ThrombosisResearch 1982;28(6):705–12.

Mecklenburg 1982 {published data only}

Mecklenburg RS, Benson JW Jr, Becker NM, Brazel PL,

Fredlund PN, Metz RJ, et al.Clinical use of the insulin

infusion pump in 100 patients with type I diabetes. NewEngland Journal of Medicine 1982;307(9):513–8.

Minkina-Pedras 2005 {published data only}

Minkina-Pedras M, Jarosz-Chobot P, Malecka-Tendera

E, Deja G. [Assessment of metabolic control and

safety of continuous subcutaneous insulin infusion in

prepubertal children with type 1 diabetes mellitus].

Endokrynologia, Diabetologia i Choroby Przemiany MateriiWieku Rozwojowego 2005;11(3):171–6.

Moberg 1994 {published data only}

Moberg EA, Lins PE, Adamson UK. Variability of blood

glucose levels in patients with type 1 diabetes mellitus on

intensified insulin regimens. Diabete et Metabolisme 1994;

20(6):546–52.

Moller 1986 {published data only}

Moller A, Rasmussen L, Ledet T, Christiansen JS,

Christensen CK, Mogensen CE, et al.Lipoprotein changes

during continuous subcutaneous insulin infusion in insulin-

dependent diabetic patients. Scandinavian Journal ofClinical & Laboratory Investigation 1986;46(5):471–5.

Monnier 1987 {published data only}

Monnier LH, Rodier M, Gancel A. Plasma lipid fatty acids

and platelet function during continuous subcutaneous

insulin infusion in type I diabetes. Diabete et Metabolisme1987;13(3):210–6.

Muller 1999 {published data only}

Muller UA, Femerling M, Reinauer KM, Risse A, Voss

M, Jorgens V, et al.Intensified treatment and education of

type 1 diabetes as clinical routine. A nationwide quality-

circle experience in Germany. ASD (the Working Group

on Structured Diabetes Therapy of the German Diabetes

Association). Diabetes Care 1999;22 Suppl 2:B29–34.

Nathan 1996 {published data only}

Nathan DM, Dunn FL, Bruch J, McKitrick C, Larkin

M, Haggan C, et al.Postprandial insulin profiles with

implantable pump therapy may explain decreased

frequency of severe hypoglycemia, compared with intensive

subcutaneous regimens, in insulin-dependent diabetes

mellitus patients. American Journal of Medicine 1996;100

(4):412–7.

Ng 1986 {published data only}

Ng Tang Fui S, Pickup JC, Bending JJ, Collins AC, Keen H,

Dalton N. Hypoglycemia and counterregulation in insulin-

dependent diabetic patients: a comparison of continuous

subcutaneous insulin infusion and conventional insulin

injection therapy. Diabetes Care 1986;9(3):221–7.

Nijs 1991 {published data only}

Nijs HG, Radder JK, Poorthuis BJ, Krans HM. Glucose

disposal and intermediary metabolism during one year of

continuous subcutaneous insulin infusion (CSII). DiabetesResearch & Clinical Practice 1991;12(1):1–9.

Olsen 1985 {published data only}

Olsen T, Ehlers N, Nielsen CB, Beck-Nielsen H. Diabetic

retinopathy after one year of improved metabolic control

obtained by continuous subcutaneous insulin infusion

(CSII). Acta Ophthalmologica 1985;63(3):315–9.

Olsen 1987 {published data only}

Olsen T, Richelsen B, Ehlers N, Beck-Nielsen H. Diabetic

retinopathy after 3 years’ treatment with continuous

subcutaneous insulin infusion (CSII). Acta Ophthalmologica1987;65(2):185–9.

Pickup 1979 {published data only}

Pickup JC, Keen H, Parsons JA, Alberti KG. Continuous

subcutaneous insulin infusion:good blood glucose control

for up to 4 days. Diabetologia 1979;16(6):385–9.

Pickup 1979a {published data only}

Pickup JC, Keen H, Parsons JA, Alberti KG, Rowe AS.

Continuous subcutaneous insulin infusion: improved

blood-glucose and intermediary-metabolite control in

diabetics. Lancet 1979;1(8129):1255–7.


Pickup JC, Collins AC, Keen H. Urinary insulin excretion

rate: an index of free insulinaemia in insulin-dependent

diabetics. Diabetic Medicine 1984;1(4):291–4.


Pickup JC, Sherwin RS, Tamborlane WV, Rizza RA, Service

FJ. Conference on insulin pump therapy in diabetes.

Multicenter study of effect on microvascular disease. The

pump life. Patient responses and clinical and technological

problems. Diabetes 1985;34 Suppl 3:37–41.


Pickup JC, Collins AC, Walker JD, Viberti GC, Pasic J.

Patterns of hyperinsulinaemia in type 1 diabetic patients



with and without nephropathy. Diabetic Medicine 1989;6

(8):685–91.

Pietri 1980 {published data only}

Pietri A, Dunn FL, Raskin P. The effect of improved diabetic

control on plasma lipid and lipoprotein levels: a comparison

of conventional therapy and continuous subcutaneous

insulin infusion. Diabetes 1980;29(12):1001–5.

Pietri 1983 {published data only}

Pietri AO, Dunn FL, Grundy SM, Raskin P. The effect

of continuous subcutaneous insulin infusion on very-

low-density lipoprotein triglyceride metabolism in type I

diabetes mellitus. Diabetes 1983;32(1):75–81.

Reeves 1982 {published data only}

Reeves ML, Seigler DE, Ryan EA, Skyler JS. Glycemic

control in insulin-dependent diabetes mellitus. Comparison

of outpatient intensified conventional therapy with

continuous subcutaneous insulin infusion. American

Journal of Medicine 1982;72(4):673–80.

Regal 1988 {published data only}

Regal H, Koivisto VA, Maury CP, Irsigler K. No sign of

amyloidosis in pump-treated diabetics. Diabetes Research1988;8(4):195–9.

Rizza 1980 {published data only}

Rizza RA, Gerich JE, Haymond MW, Westland RE, Hall

LD, Clemens AH, et al.Control of blood sugar in insulin-

dependent diabetes: comparison of an artificial endocrine

pancreas, continuous subcutaneous insulin infusion, and

intensified conventional insulin therapy. New EnglandJournal of Medicine 1980;303(23):1313–8.

Rodger 1985 {published data only}

Rodger NW, Dupre J, Canny CLB. Continuous

subcutaneous insulin infusion in adults: Glycemic

advantage is predicted by venous plasma C-peptide

concentrations. Diabetes Care 1985;8(5):447–55.

Rodger 1988 {published data only}

Rodger NW, Behme MT, Dupre J, Chamberlain MJ.

Metabolic effects of continuous subcutaneous insulin

infusion: evidence that a rise and fall of portal vein

insulin concentration with each major meal facilitates

post-absorptive glycemic control. Clinical & InvestigativeMedicine - Medecine Clinique et Experimentale 1988;11(3):

167–86.

Saibene 1981 {published data only}

Saibene V, Melandri M, Brembilla L, Spotti D, Pozza

G. Comparison between multi-injection and continuous

subcutaneous insulin therapy in insulin-dependent diabetic

inpatients. Acta Diabetologica Latina 1981;18(1):45–50.

Schaepelynck-Belicar 2003 {published data only}

Schaepelynck-Belicar P, Vague P, Simonin G, Lassmann-

Vague V. Improved metabolic control in diabetic adolescents

using the continuous glucose monitoring system (CGMS).

Diabetes & Metabolism 2003;29(6):608–12.

Schiffrin 1982 {published data only}

Schiffrin A, Belmonte M. Multiple daily self-glucose

monitoring: its essential role in long-term glucose control

in insulin-dependent diabetic patients treated with pump

and multiple subcutaneous injections. Diabetes Care 1982;

5(5):479–84.

Schiffrin 1982a {published data only}

Schiffrin A, Belmonte MM. Comparison between

continuous subcutaneous insulin infusion and multiple

injections of insulin. A one-year prospective study. Diabetes

1982;31(3):255–64.


Schiffrin A, Desrosiers M, Moffatt M, Belmonte MM.

Feasibility of strict diabetes control in insulin-dependent

diabetic adolescents. Journal of Pediatrics 1983;103(4):

522–7.


Schiffrin A, Parikh S, Marliss EB, Desrosiers MM.

Metabolic response to fasting exercise in adolescent insulin-

dependent diabetic subjects treated with continuous

subcutaneous insulin infusion and intensive conventional

therapy. Diabetes Care 1984;7(3):255–60.

Schiffrin 1984a {published data only}

Schiffrin AD, Desrosiers M, Aleyassine H, Belmonte MM.

Intensified insulin therapy in the type I diabetic adolescent:

a controlled trial. Diabetes Care 1984;7(2):107–13.


Schiffrin A, Parikh S. Accommodating planned exercise in

type I diabetic patients on intensive treatment. Diabetes

Care 1985;8(4):337–42.


Schiffrin A, Suissa S. Predicting nocturnal hypoglycemia

in patients with type I diabetes treated with continuous

subcutaneous insulin infusion. American Journal of Medicine1987;82(6):1127–32.

Schmitz 1987 {published data only}

Schmitz O, Sorensen SS, Alberti KG, Orskov H, Hansen

HE. Metabolic control in newly kidney transplanted

insulin-dependent diabetics: improvement by insulin pump

treatment (CSII). Journal of Diabetic Complications 1987;1

(3):81–6.

Schottenfeld-Naor 1985 {published data only}

Schottenfeld-Naor Y, Galatzer A, Karp M, Josefsberg Z,

Laron Z. Comparison of metabolic and psychological

parameters during continuous subcutaneous insulin

infusion and intensified conventional insulin treatment in

type I diabetic patients. Israel Journal of Medical Sciences1985;21(10):822–8.

Scuffham 2003 {published data only}

Scuffham P, Carr L. The cost-effectiveness of continuous

subcutaneous insulin infusion compared with multiple daily

injections for the management of diabetes.[see comment].

Diabetic Medicine 2003;20(7):586–93.

Service 1985 {published data only}

Service FJ, Rizza RA, Daube JR, O’Brien PC, Dyck PJ.

Near normoglycaemia improved nerve conduction and

vibration sensation in diabetic neuropathy. Diabetologia

1985;28(10):722–7.



Siebenhofer 2004 {published data only}

Siebenhofer A, Plank J, Berghold A, Horvath K, Sawicki

PT, Beck P, et al.Meta-analysis of short-acting insulin

analogues in adult patients with type 1 diabetes: continuous

subcutaneous insulin infusion versus injection therapy.

Diabetologia 2004;47(11):1895–905.

Simonson 1985 {published data only}

Simonson DC, Tamborlane WV, Sherwin RS, Smith JD,

DeFronzo RA. Improved insulin sensitivity in patients with

type I diabetes mellitus after CSII. Diabetes 1985;34 Suppl

3:80–6.

Skare 1986 {published data only}

Skare S, Dahl-Jorgensen K, Kriz V, Hanssen KF. Plasma

somatostatin and plasma glucagon in long-term IDDM

without residual B-cell function. No effect of different long-

term metabolic control. Scandinavian Journal of Clinical &

Laboratory Investigation 1986;46(7):635–8.

Skyler 1982 {published data only}

Skyler JS, Seigler DE, Reeves ML. A comparison of insulin

regimens in insulin-dependent diabetes mellitus. DiabetesCare 1982;5 Suppl 1:11–8.

Sleightholm 1986 {published data only}

Sleightholm MA, Gallimore R, Tennent GA. Continuous

subcutaneous insulin infusion does not provoke significant

acute-phase response. Diabetes Care 1986;9(1):50–2.

Slijper 1990 {published data only}

Slijper FME, De Beaufort CE, Bruining GJ, De Visser JJ,

Aarsen RSR, Kicken DAM, et al.Psychological impact of

continuous subcutaneous insulin infusion pump therapy in

non-selected newly diagnosed insulin dependent (type 1)

diabetic children: Evaluation after two years of therapy.


Steno 1982 {published data only}

Steno. Effect of 6 months of strict metabolic control on eye

and kidney function in insulin-dependent diabetics with

background retinopathy. Steno study group. Lancet 1982;1

(8264):121–4.

Sulli 2003 {published data only}

Sulli N, Shashaj B. Continuous subcutaneous insulin

infusion in children and adolescents with diabetes mellitus:

decreased HbA1c with low risk of hypoglycemia. Journal of

Pediatric Endocrinology 2003;16(3):393–9.

Tamborlane 1979 {published data only}

Tamborlane WV, Sherwin RS, Genel M, Felig P. Reduction

to normal of plasma glucose in juvenile diabetes by

subcutaneous administration of insulin with a portable

infusion pump. New England Journal of Medicine 1979;300

(11):573–8.

Tamborlane 1979a {published data only}

Tamborlane WV, Sherwin RS, Genel M, Felig P. Restoration

of normal lipid and aminoacid metabolism in diabetic

patients treated with a portable insulin-infusion pump.

Lancet 1979;1(8129):1258–61.


Tamborlane WV, Hintz RL, Bergman M, Genel M, Felig P,

Sherwin RS. Insulin-infusion-pump treatment of diabetes:

influence of improved metabolic control on plasma

somatomedin levels. New England Journal of Medicine1981;305(6):303–7.


Tamborlane WV, Puklin JE, Bergman M, Verdonk C,

Rudolf MC, Felig P, et al.Long-term improvement of

metabolic control with the insulin pump does not reverse

diabetic microangiopathy. Diabetes Care 1982;5 Suppl 1:

58–64.

Testa 1985 {published data only}

Testa MA, Puklin JE, Sherwin RS, Simonson DC. Clinical

predictors of retinopathy and its progression in patients

with type I diabetes during CSII or conventional insulin

treatment. Diabetes 1985;34 Suppl 3:61–8.

Thuesen 1986 {published data only}

Thuesen L, Christiansen JS, Sorensen KE, Falstie-Jensen N,

Christensen CK, Hermansen K, et al.Exercise capacity and

cardiac function in type 1 diabetic patients treated with

continuous subcutaneous insulin infusion. A controlled

study. Scandinavian Journal of Clinical & Laboratory

Investigation 1986;46(8):779–84.

Tilvis 1986 {published data only}

Tilvis RS, Helve E, Miettinen TA. Improvement of diabetic

control by continuous subcutaneous insulin infusion

therapy changes fatty acid composition of serum lipids

and erythrocytes in type 1 (insulin-dependent) diabetes.

Diabetologia 1986;29(10):690–4.

Toni 2004 {published data only}

Toni S, Reali MF, Fasulo A, Festini P, Medici A,

Martinucci ME. The use of insulin pumps improves the

metabolic control in children and adolescents with type 1

diabetes.[comment][erratum appears in Arch Dis Child.

2004 Oct;89(10):985 Note: Festini, F [corrected to Festini,

P]]. Archives of Disease in Childhood 2004;89(8):796–7.

Turk 1996 {published data only}

Turk Z, Flego I, Kerum G. Platelet aggregation in type 1

diabetes without microvascular disease during continuous

subcutaneous insulin infusion. Hormone & MetabolicResearch 1996;28(2):95–100.

van Ballegooie 1985 {published data only}

van Ballegooie E, de Jong PE, Donker AJ, Sluiter WJ.

The effect of continuous subcutaneous insulin infusion

on renal function in type I diabetic patients with and

without proteinuria. Proceedings of the European Dialysis &Transplant Association - European Renal Association 1985;21:

722–4.

Viberti 1981 {published data only}

Viberti G, Pickup JC, Bilous RW, Keen H, Mackintosh

D. Correction of exercise-induced microalbuminuria in

insulin-dependent diabetics after 3 weeks of subcutaneous

insulin infusion. Diabetes 1981;30(10):818–23.

Viberti 1984 {published data only}

Viberti GC, Home PD, Bilous RW, Alberti KG, Dalton

N, Keen H, et al.Metabolic effects of physical exercise

in insulin-dependent diabetics controlled by continuous



subcutaneous insulin infusion or conventional injection

therapy. Acta Endocrinologica 1984;105(4):515–20.

Warmolts 1987 {published data only}

Warmolts JR, Mendell JR, O’Dorisio TM, Cataland S.

Comparison of the effects of continuous subcutaneous

infusion and split-mixed injection of insulin on nerve

function in type I diabetes mellitus. Journal of the

Neurological Sciences 1987;82(1-3):161–9.

Weinzimer 2004 {published data only}

Weinzimer SA, Ahern JH, Doyle EA, Vincent MR, Dziura

J, Steffen AT, et al.Persistence of benefits of continuous

subcutaneous insulin infusion in very young children with

type 1 diabetes: a follow-up report. Pediatrics 2004;114(6):

1601–5.

Willi 2003 {published data only}

Willi SM, Planton J, Egede L, Schwarz S. Benefits of

continuous subcutaneous insulin infusion in children with

type 1 diabetes. Journal of Pediatrics 2003;143(6):796–801.

Wilson 2005 {published data only}

Wilson DM, Buckingham BA, Kunselman EL, Sullivan

MM, Paguntalan HU, Gitelman SE. A two-center

randomized controlled feasibility trial of insulin pump

therapy in young children with diabetes. Diabetes Care

2005;28(1):15–9.

Wredling 1997 {published data only}

Wredling R, Hannerz L, Johansson U. Variability of

blood glucose levels in patients treated with continuous

subcutaneous insulin infusion: a pilot study. Practical

Diabetes International 1997;14(1):5–8.

Yki-Jarvinen 1984 {published data only}

Yki-Jarvinen H, Koivisto VA. Continuous subcutaneous

insulin infusion therapy decreases insulin resistance in type

1 diabetes. Journal of Clinical Endocrinology & Metabolism1984;58(4):659–66.

Ziegler 1988 {published data only}

Ziegler D, Wiefels K, Dannehl K, Gries FA. Effects of one

year of near-normoglycemia on peripheral nerve function

in type 1 (insulin-dependent) diabetic patients. KlinischeWochenschrift 1988;66(9):388–96.

Additional references

ADA 1997

American Diabetes Association. Report of the Expert

Committee on the Diagnosis and Classification of Diabetes

Mellitus. Diabetes Care 1997;20(7):1183–97.

APEG 2004

Australian Paediatric Endocrine Group. Clinical Practise

Guidelines: Type 1 Diabetes in Children and Adolescents.

www.chw.edu.au/prof/services/endocrinology/apeg

www.chw.edu.au/prof/services/endocrinology/apeg http://

www.chw.edu.au/prof/services/endocrinology/apeg/ 2004.

Couper 2002

Couper JJ. Children with type 1 diabetes: where are we at?.

The Medical Journal of Australia 2002;177(5):228–9.

Dabelea 2009

Dabelea D. The accelerating epidemic of childhood

diabetes. The Lancet 2009;373(9680):1999–2000.

DPT 2002

Diabetes Prevention Trial-Type 1 Diabetes Study Group.

Effects of insulin in relatives of patients with type 1 diabetes

mellitus. New England Journal of Medicine 2002;346(22):

1685–91.

ENDIT 2003

The European Nicotinamide Diabetes Intervention Trial

(ENDIT) Group. Intervening before the onset of Type 1

diabetes: baseline data from the European Nicotinamide

Diabetes Intervention Trial (ENDIT). Diabetologia 2003;

46(3):339–46.

Guinn 1988

Guinn TS, Bailey GJ, Mecklenberg RS. Factors related

to discontinuation of continuous subcutaneous insulin-

infusion therapy. Diabetes Care 1988;11(1):46–51.

Higgins 2003

Higgins JP, Thompson SG, Deeks JJ, Altman DG.

Measuring inconsistency in meta-analyses. British Medical

Journal 2003;327(7414):557–60.

Higgins 2008

Higgins JPT, Green S (editors). Cochrane Handbook for

Systematic Reviews of Interventions Version 5.0.1 [updated

September 2008]. The Cochrane Collaboration. Available

from www.cochrane-handbook.org. WileyBlackwell, 2008.

HTA 2004

Colquitt JL, Green C, Sidhu MK, Hartwell D, Waugh N.

Clinical and cost-effectiveness of continuous subcutaneous

insulin infusion for diabetes. Health Technology Assessment

2004; Vol. 8, issue number 43.

Karvonen 2000

Karvonen M, Viik-Kajander M, Moltchanova E, Libman

I, LaPorte R, Tuomilehto J. Incidence of childhood type 1

diabetes worldwide. Diabetes Mondiale (DiaMond) Project

Group. Diabetes Care 2000;23(10):1516–26.

Liberati 2009

Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC,

Ioannidis JPA, et al.The PRISMA Statement for Reporting

Systematic and Meta-Analyses of Studies That Evaluate

Interventions: Explanation and Elaboration. PLoS Med

1999;6(7):1–28. [DOI: 10.1371/journal.pmed.1000100]

NDDG 1979

National Diabetes Data Group. Classification and diagnosis

of diabetes mellitus and other categories of glucose

intolerance. Diabetes 1979;28(12):1039–57.

NICE 2008

National Institute for Health and Clinical Excellence

(NICE). Continuous subcutaneous insulin infusion for the

treatment of diabetes mellitus. London (UK): National

Institute for Health and Clinical Excellence (NICE);

Technology appraisal guidance no. 151 2008 July.



Pickup 1980a

Pickup JC. A new approach to improved metabolic control

in diabetics: Continuous subcutaneous insulin infusion.

American Heart Journal 1980;100(4):417–20.

Pickup 2002

Pickup J, Mattock M, Kerry S. Glycaemic control with

continuous subcutaneous insulin infusion compared with

intensive injections in patients with type 1 diabetes: meta-

analysis of randomised controlled trials. British MedicalJournal 2002;324:1–6.

Sterne 2001

Sterne JAC, Egger M, Davey Smith G. Investigating and

dealing with publication and other biases.. British Medical

Journal 2001;323:101–105.

Sutton 1990

Sutton L, Plant AJ, Lyle DM. Services and cost of

hospitalization for children and adolescents with insulin-

dependent diabetes mellitus in New South Wales. The

Medical Journal of Australia 1990;152(3):130–6.

WHO 1980

World Health Organisation. WHO Expert Committee

on Diabetes Mellitus: second report. World Health

Organization Technical Report Series 1980;646:1–80.∗ Indicates the major publication for the study



C H A R A C T E R I S T I C S O F S T U D I E S

Characteristics of included studies [ordered by study ID]

Bak 1987 and Husted 1989

Methods RANDOMISED CONTROLLED CLINICAL TRIAL (RCT): Yes

RANDOMISATION RATIO: Not reported

NON-INFERIORITY DESIGN: Not reported

EQUIVALENCE DESIGN: Not reported

PARALLEL / CROSSOVER / FACTORIAL RCT: Cross-over

CONTROLLED CLINICAL TRIAL (CCT): Yes

Participants WHO PARTICIPATED: 20 participants were randomised in this study. Half of the

participants were randomised to CSII and the other half were randomised to MI. After

six months they were crossed over, serving as their own control. In total, 16 participants

completed CSII and 20 completed MI

SEX (female% / male%): 1 female (6%), 15 males (94%)

AGE (mean years (SD)): 24 (2) years

ETHNIC GROUPS (%): Not reported

DURATION OF DISEASE (mean (SD)): 70 (45) months

INCLUSION CRITERIA: IDDM of more than 2.5 years and diagnosed before age

30, no sign of nephropathy, hypertension or advanced retinopathy (ie. normal value of

serum creatinine), absence of albuminuria (ie. Albustic negative urine), diastolic blood

pressure below 95 mmHg, absence of cataract, retinal proliferations, haemorrhages or

exudates and less than 5 microaneurysms in the retina

EXCLUSION CRITERIA: Not reported

DIAGNOSTIC CRITERIA: WHO 1980

CO-MORBIDITIES: Not reported

CO-MEDICATIONS: Not reported

Interventions NUMBER OF STUDY CENTRES: 1

COUNTRY/ LOCATION: Denmark

SETTING: Out-patient, however transition phases to CSII or MI were made during

hospitalisation

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY):100 U/ml regular insulin (Ac-

trapid HM) was delivered by a Graseby MS 36 insulin infuser with a 25 G butterfly

needle placed in the anterior abdominal wall. About half of the total daily dose was

delivered as a basal infusion and the other half as bolus injections, 30 minutes before

meals

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): About 30% of the total daily dose

was delivered as intermediate acting NPH insulin (40 U/ml insulin protaphan HM)

in the thigh late at night. Short-acting insulin was delivered by a pen injector with a

disposable cartridge of 1.5 ml of short acting insulin (100 U/ml insulin actrapid HM).

Bolus injections were delivered in the abdominal wall 30 minutes before breakfast, lunch

and dinner

TREATMENT BEFORE STUDY: All of the participants were treated (as out-patients)

with intermediate-acting insulin alone or in combination with short-acting insulin once

or twice a day

TITRATION PERIOD: Not reported



Bak 1987 and Husted 1989 (Continued)

Outcomes PRIMARY OUTCOME(S) (as stated in the publication): Not reported

SECONDARY OUTCOMES (as stated in the publication): Not reported

ADDITIONAL OUTCOMES:

Bak 1987 - metabolic regulation (mean diurnal blood glucose, serum fructosamine,

fasting plasma ketone bodies, FFA, HGH and glucagon), weight, estimation of blood

glucose instability, patient’s evaluation (questionnaire)

Husted 1998 - metabolic regulation (mean diurnal blood glucose, serum fructosamine),

weight, coagulation and platelet function (platelet aggregation, serum and plasma throm-

boxane B2, plasma antithrombin III activity, plasma von Willebrand factor antigen)

Study details DURATION OF INTERVENTION: 6 months

DURATION OF FOLLOW-UP: 12 months

RUN-IN PERIOD: Duration not reported. All of the participants were treated (as out-

patients) with intermediate-acting insulin alone or in combination with short-acting

insulin once or twice a day

Publication details LANGUAGE OF PUBLICATION: English

COMMERCIAL FUNDING: No

NON-COMMERCIAL FUNDING: Yes

PUBLICATION STATUS (PEER REVIEW JOURNAL): Yes

PUBLICATION STATUS (JOURNAL SUPPLEMENT): No

PUBLICATION STATUS (ABSTRACT): No

Stated aim of study Bak 1987 - “We therefore made a cross-over study in order to compare 6 months of

CSII therapy with 6 months of MII by evaluating metabolic control, insulin needs, and

patient acceptability.”

Husted 1998 - “The aim of the present study with cross-over design was to compare 6

months of CSII therapy with 6 months of MII by evaluating metabolic control, platelet

aggregability, thromboxane B2 levels in serum and plasma as well as ATIII activity and

von Willebrand factor antigen.”

Notes Standard deviations of the difference were extracted from paired t-test data

Risk of bias

Bias Authors’ judgement Support for judgement

Adequate sequence generation? Unclear risk Details not provided.

Allocation concealment? Unclear risk Details not provided.

Blinding?

All outcomes

High risk Not applicable for patient blinding, but

outcome assessor blinding not reported

Incomplete outcome data addressed?

All outcomes

Low risk Withdrawals, drop outs, and losses to fol-

low up were described. 4 drop-outs (20%)

Free of selective reporting? Unclear risk Details not provided.



Bak 1987 and Husted 1989 (Continued)

Free of other bias? Unclear risk Compliance was not reported.

Botta 1986





PARALLEL / CROSSOVER / FACTORIAL RCT: Parallel


Participants WHO PARTICIPATED: 10 participants were included in this study, all were pregnant

females. 5 were randomised to, and completed CSII and another 5 females were ran-

domised to, and completed MI

SEX (female% / male%): 100% female

AGE (mean years (SD)): The mean age of the participants receiving CSII was 27.8 (6.

18) years and the mean age of the participants receiving MI was 25.4 (5.36) years


DURATION OF DISEASE (mean years (SD)): The mean disease duration for the

participants receiving CSII was 8.2 (8.04) years and in those receiving MI, the mean

disease duration was 9.4 (9.4) years

INCLUSION CRITERIA: Not reported


DIAGNOSTIC CRITERIA: Not reported




COUNTRY/ LOCATION: Italy

SETTING: Out-patient

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The Micropump CPI 9100

Lilly was used to attempt to maintain the glycaemic values between 90 mg/dl and 130

mg/dl during the course of the day

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Glycaemic values were main-

tained between 90 mg/dl and 130 mg/dl during the course of the day. Rapid MC insulin

was delivered before breakfast, lunch and dinner, with the addition of intermediate in-

sulin MC with lunch and dinner. Participants were instructed to modify the dosage of

rapid insulin by two units if on the preceding day glycaemia (two hours before the meal)

was higher than 150 mg/dl or lower than 90 mg/dl

TREATMENT BEFORE STUDY: Not reported




ADDITIONAL OUTCOMES: fetal weight index, fasting and post-prandial glycemia,

ketonuria, hypoglycemia and pregnancy related outcomes



Botta 1986 (Continued)

Study details DURATION OF INTERVENTION: 24.68 weeks for CSII and 28.80 weeks for MI

DURATION OF FOLLOW-UP: 24.68 weeks for CSII and 28.80 weeks for MI

RUN-IN PERIOD: Not reported

Publication details LANGUAGE OF PUBLICATION: Italian

COMMERCIAL FUNDING: Not reported

NON-COMMERCIAL FUNDING: Not reported




Stated aim of study “Considerando il particolare atteggiamento emotivo della donna diabetica nei confronti

della sua condizione patologica durante la gravidanza abbiamo tentato di valutare se

l’applicazione del microinfusore nelle donne gravide con diabete mellito di tipo I possa

determinare una maggiore compliance alla terapia ed all’autocontrollo rispetto alla ter-

apia insulinica tradizionale e ci possa consentire di ottenere dei risultati migliori nel

trattamento della gravidanza diabetica.”

Translated: Considering the detail emotional attitude of the diabetic woman regarding its

pathological condition during the pregnancy we have tried to estimate if application of

the microinfusore in the pregnant women with mellito diabetes can determine a greater

compliance to the therapy and self-control regarding the traditional insulinica therapy

and can concur to obtain to us of the better result in the treatment of the diabetic

pregnancy

Notes

Risk of bias




Blinding?

All outcomes


outcome assesor blinding not reported


All outcomes

High risk Number and reasons of drop-outs, with-

drawals, and losses to follow up were not

reported





Bruttomesso 2008







Participants WHO PARTICIPATED: 42 participants were included in this study. All participants

were already on CSII treatment. 24 participants were randomised to continue with CSII

and 18 participants were randomised to start MI for four months, after which the groups

crossed over, serving as their own control

SEX (female% / male%): The final sample comprised 20 female (51%), 19 male (49%)

AGE (mean years (SD)): 38.1 (9.3) years


DURATION OF DISEASE (mean years (SD)): 16.6 (8.2) years

INCLUSION CRITERIA: Willingness to measure blood glucose frequently and profi-

ciency in adjusting insulin administration according to blood glucose values and carbo-

hydrate intake

EXCLUSION CRITERIA: Individuals with insulin allergy, untreated retinopathy, car-

diovascular disease, hepatic or renal insufficiency, drug abuse or life-threatening diseases,

women who were pregnant or lactating or intending to become pregnant were excluded







CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro was infused and

adjusted according to the following targets: 5.0 mmol/L in the morning and before

meals, 6.0 to 8.5 mmol/L 90 minutes after meals and 6.1 to 7.5 mmol/L at bed time

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro was delivered at

meals and adjusted according to the following targets: 5.0 mmol/L in the morning and

before meals, 6.0 to 8.5 mmol/L 90 minutes after meals and 6.1 to 7.5 mmol/L at bed

time. Participants with less than optimal glycemic control with one glargine injection

were instructed to take twice daily glargine or extra injections of insulin lispro

TREATMENT BEFORE STUDY: All participants were treated by CSII with lispro or

aspart

TITRATION PERIOD: Four weeks - all participants using lispro continued, but two

participants changed from aspart to lispro

Outcomes PRIMARY OUTCOME(S) (as stated in the publication): blood glucose variability

SECONDARY OUTCOMES (as stated in the publication): blood glucose concentra-

tions, number and severity of hypo/hyperglycemic episodes, number of episodes of ke-

toacidosis, insulin requirement, body weight, HbA1c, lipid profile, serum FFA and hGH,

and treatment satisfaction

ADDITIONAL OUTCOMES: NA



Bruttomesso 2008 (Continued)



RUN-IN PERIOD: For four weeks basal insulin infusion rate and meal-related boluses

were optimised as much as possible, while avoiding hypoglycaemia and hyperglycemia.

The target blood glucose for the trial was 5.0 mmol/L in the morning and before meals,

6.0 to 8.5 mmol/L 90 minutes after meals and 6.1 to 7.5 mmol/L at bed time


COMMERCIAL FUNDING: Yes - Disetronic Medical System AG (one of the study

investigators was a full time employee and another was funded to attend conferences)

NON-COMMERCIAL FUNDING: No




Stated aim of study “The superiority of continuous subcutaneous insulin infusion (CSII) over multiple daily

injections (MDI) with glargine is uncertain. In this randomized cross-over study, we

compared CSII and MDI with glargine in patients with Type 1 diabetes well controlled

with CSII. The primary endpoint was glucose variability.”


Risk of bias


Adequate sequence generation? Low risk Participants were randomised by Dis-

etronic Medical Systems AG (Burgdorf,

Switzerland) and treatment assignment was

determined with random lists for each cen-

tre

Allocation concealment? Low risk Participants and investigators were blinded

to the treatment sequence up to randomi-

sation

Blinding?

All outcomes

High risk Not applicable to patient blinding, but out-

come assessor blinding is not reported


All outcomes

Low risk Drop outs described.


Free of other bias? Unclear risk Compliance not reported.



Chiasson 1984 and 1985







Participants WHO PARTCIPATED: 12 participants were included in this study. Half of the partic-

ipants were randomised to start with CSII and the other half were randomised to start

with MI. After three months they were crossed over

SEX (female% / male%): 7 female (58%), 5 male (42%)










COUNTRY/ LOCATION: Canada


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Using the Mill-HIll Infuser

(Muirhead Medical Products Ltd, London, England) and on the basis of previous study

results, initial basal rate was set at 50% of the previous insulin dose and was delivered

as a continuous infusion of regular insulin. The boluses were delivered 20 minutes

before meals as regular insulin according to the following initial schedules: 2.7 U/20g

carbohydrates before breakfast and 1.8 U/20g carbohydrates before other meals and

snacks containing more than 20g carbohydrates. The target blood glucose was 70 to 120

mg/dl before meals and <150 mg/dl 1hr after meals. All blood glucose <50 mg/dl and

>180 mg/dl were undesirable. Basal rate was increased or decreased by 1 U/24h if fasting

blood glucose was >120 mg/dl or <70 mg/dl. The boluses were increased or decreased by

0.2 U/20g carbohydrates if one hour after meal blood glucose was >150 mg/dl or below

the before meal blood glucose. Participants on CSII were also instructed to add 2 U of

regular insulin to boluses for every 100 mg/dl above blood glucose target

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Using the Medi-Jector (Derata

Corporation, Minneapolis, Minnesota) and on basis of previous study results, initial

basal rate was set at 50% of the previous insulin dose and was delivered as ultralente at 22

hours. The boluses were delivered 20 minutes before meals as regular insulin according

to the schedules described above





ADDITIONAL OUTCOMES: metabolic regulation (capillary blood glucose, glyco-

suria, glycated haemoglobin), nerve conductivity, proteinuria & required insulin dose to

maintain blood glucose



Chiasson 1984 and 1985 (Continued)



RUN-IN PERIOD: All of the participants were admitted to the unit at the hospital to

initiate either treatment and learn how to operate the (CSII) pump or (MI) medijector,

how to calculate carbohydrates and how to modify insulin dose according to prespecified

algorithms/schedules

Publication details LANGUAGE OF PUBLICATION: English & French


NON-COMMERCIAL FUNDING: Yes - medical research councils, NIH etc




Stated aim of study “The present study was designed to compare continuous subcutaneous insulin infusion

(CSII) using the Mill-Hill Infuser (Muirhead Medical Products Ltd, London, England)

with multiple injections using the Medi-Jector (Derata Corporation, Minneapolis, Min-

nesota) in the treatment of insulin dependent diabetes mellitus (IDDM) and to assess

the effect of glucose control on diabetes late complications.”


Risk of bias




Blinding?

All outcomes

High risk Not applicable to patient blinding, but out-

come assessor blinding is not reported


All outcomes

High risk Number and reasons of drop-outs, with-

drawals, and losses to follow up were not

reported


Free of other bias? Unclear risk Compliance was measured and checked by

going over the participant’s notebook at

each visit



Cohen 2003







Participants WHO PARTICIPATED: 16 participants were included in this study. The participants

were divided into two groups, one group was randomised to CSII and the other group

was randomised to MI. After six months they were crossed over, serving as their own

control. 12 participants completed the study


AGE (mean years (SD)): Mean was not reported. The median age of all participants was

14.2 years (range 14.5-17.9)


DURATION OF DISEASE (mean years (SD)): Mean was not reported. At least two

years

INCLUSION CRITERIA: Type 1 diabetes mellitus of at least two years, low c-peptide

secretion (<0.6 ng/ml), and no other chronic disease that could interfere with treatment

EXCLUSION CRITERIA: Participants unable to detect hypoglycemia or who had

evidence of microvascular complications or other clinically significant disorders





COUNTRY/ LOCATION: Israel

SETTING: Out-patient (ambulatory)

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Tayco, Disetronic (Burgdorf,

Switzerland) pump with short acting analog insulin (Humalog, Eli Lily)

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Four daily injections of a com-

bination of NPH and regular insulin before breakfast, regular insulin only before lunch

and dinner, and NPH only at bed time





ADDITIONAL OUTCOMES: Body mass index, body mass index standard deviation

score for age, HbA1c, frucosamine, daily insulin dose, hypo- and hyperglycemic events,

diabetic ketoacidosis, Diabetes Treatment Satisfaction Questionnaire (DTSQ) and the

Diabetes Quality of Life Questionnaire for Youth (DQOLY)





COMMERCIAL FUNDING: Yes - Tayco Diagnostica



Cohen 2003 (Continued)





Stated aim of study “We designed an open randomized crossover trial to compare CSII with MDI for

glycemic control, dose requirements, weight change, incidence of adverse events, quality

of life and satisfaction in adolescents. ”

Notes This cross-over study was not adequately analysed, and did not report correlations be-

tween baseline and end of study data, ignoring within-person variation. Therefore, stan-

dard deviations of the difference were derived assuming different correlation coefficients

(0.3, 0.5, 0.7, 0.9) and the largest standard deviation of the differences was derived from

cross-over studies with adequate statistical methodology

Risk of bias




Blinding?

All outcomes

High risk Not applicable in patient blinding, but out-

come assessor blinding not reported


All outcomes

Low risk Withdrawals, drop outs, and losses to fol-

low up were described - three were satisfied

with pump and so didn’t want to start MI,

one who started with MI refused to connect

to pump. There was 0.25% withdrawals/

dropouts/losses to follow up





DeVries 2002






It is important to note that at the design phase, the study investigators anticipated the possibilitythat they would have to analyse the study as a parallel RCT (rather than a cross-over) becauseof the demanding nature of the protocol and an expected high drop out rate. However, atthe point of cross-over, there had been no drop outs so the study investigators continued witha cross-over design and stopped inclusion at 79 participants (48 participants per arm werecalcluated to be required to have an 80% chance of detecting a 0.75% difference in HbA1cwith an assumed SD of 1.3%). Finally, the study investigators decided that due to the dropout rate after cross-over, they would analyse the data as a parallel study using only pre cross-over data.CONTROLLED CLINICAL TRIAL (CCT): Yes

Participants WHO PARTICIPATED: 79 participants were included in this study. 39 participants

were randomised to start with CSII and 40 participants were randomised to start with

MI. 32 participants completed CSII in the first stage and all 40 participants completed

MI (data from these participants is analysed)


AGE (mean years (SD)): The mean age of participants in the group starting with CSII

was 36.2 (10.3) and the mean age of the participants starting with MI was 37.3 (10.6)


DURATION OF DISEASE (mean years (SD)): The mean disease duration in the group

staring with CSII was 17.6 (9.8) years and the mean disease duration in the group staring

with MI was 18.0 (9.4) years

INCLUSION CRITERIA: Type 1 diabetes, age between 18 and 70 years, persistent

poor control while on three or more insulin injections a day (defined as mean of all

HbA1c values measured ≥ 8.5% in the last six months before the study)

EXCLUSION CRITERIA: Severe active retinopathy (requiring laser therapy), impaired

hepatic function (alkaline phosphatase or alanine aminotransferase at least two times the

upper limit or normal), nephropathy (defined as serum creatinine > 150 µmol/L), insulin

resistance (defined as the use of > 1.5 units insulin/kg body weight), substance abuse,

cardiac disease (decompensated heart failure NYHA III and IV, unstable angina pectoris,

or a myocardial infarction within the last 12 months), uncontrolled hypertension (blood

pressure ≥ 180/110 mmHg), insulin allergy, and past or current psychiatric treatment

for schizophrenia, mental disorder, or bipolar disorder. Women were excluded if they

were (or intending to become) pregnant or breastfeeding

DIAGNOSTIC CRITERIA: Defined as diabetes diagnosed at or before age 30 years

with a C-peptide level ≤ 0.20 nmol/L at a concomitant glucose level ≥ 7.0 mmol/L, or

diagnosed at or before age 40 with a C-peptide level ≤ 0.05 nmol/L at a concomitant

glucose level ≥ 7.0 mmol/L

CO-MORBIDITIES: 36 participants had retinopathy, 19 had microalbuminuria, 8 had

proteinuria and 10 had neuropathy



COUNTRY/ LOCATION: Netherlands




DeVries 2002 (Continued)

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The Disetronic H-TRONplus

insulin pump was used with insulin aspart. The starting dose was 90% of the previously

used total insulin dose per day or 80% when hypoglycaemia was a problem to the

participants. It was advised to start with one basal rate, or two at the most. The nighttime

rate then being 0.2 units/hour lower than the day time rate. Initially, 50% of the total

daily insulin dose was delivered as basal insulin, the rest equally divided before meals. The

meal time bolus insulin aspart was adjusted according to the postprandial measurements

and the basal rate was adjusted according to the preprandial measurements, targeting for

the same levels as the MI group

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY):An algorithm was given to adjust

insulin doses. 80% of the previously given before meal human insulin dose was delivered

as insulin aspart before meals and participants were advised to add the preprandial dose

increment to their night time NPH insulin dose, leaving the total daily insulin dose

unchanged. When the interval between two day time injections was greater than five

hours, additional NPH doses were added: 40% of the previously given premeal human

insulin was delivered as NHP insulin and 60% as insulin aspart for each interval exceeding

five hours. Insulin aspart was adjusted according to the postprandial measurement and

NPH was adjusted according to the preprandial measurements



Outcomes PRIMARY OUTCOME(S) (as stated in the publication): Change in HbA1c in both

randomisation groups from baseline to 12 and 16 weeks (the mean of both)

SECONDARY OUTCOMES (as stated in the publication): Number of hypoglycemias,

means of blood glucose values at the nine points in the blood glucose profiles, SD of all

measurements in these profiles, and changes in dimensions of the quality-of-life measures

ADDITIONAL OUTCOMES: Insulin requirement, body weight.

Study details DURATION OF INTERVENTION: 16 weeks

DURATION OF FOLLOW-UP: 32 weeks

RUN-IN PERIOD: Qualification phase consisted of a short re-education and intensi-

fication of the frequency of out-patient clinic visits. At week two (of 14 weeks) a 30

minute re-education session was scheduled with the diabetes nurse specialist, which in-

cluded checking for possible deficits in insulin injection technique, self monitored blood

glucose, self-regulation of insulin dose in anticipation of physical activity and meals, and

in case of holidays and illness, actions to be taken in case of hypo- or hyperglycaemia,

storage or insulin and self monitoring strips, the advice to keep bedtime insulin dosage

constant, solely to be adjusted on consistent out-of-range glucose values at awakening,

and finally the advice to take additional carbohydrate in case of a glucose value <8 mmol/

L at bed time. There were five outpatient clinic visits during this phase. A minimum of

two self monitored blood glucose measurements per day were required and a minimum

of 70% of these had to be made to qualify for randomisation


COMMERCIAL FUNDING: Yes, but the study investigators stated that they were the

sole originators of the study protocol






DeVries 2002 (Continued)


Stated aim of study “The primary objective was to compare the efficacy in improving glycemic control of

continuous subcutaneous insulin infusion and insulin injection therapy in diabetic pa-

tients in persistent poor control. The secondary objective was to investigate possible

different effects of these modes of treatment on health-related quality of life.”

Notes

Risk of bias


Adequate sequence generation? Low risk The randomisation list was computer gen-

erated and permuted blocks of six were used

for each centre

Allocation concealment? Low risk Subjects were randomised with scratch la-

bels to each intervention, sequentially as-

signed to participants by the coordinating

center

Blinding?

All outcomes

High risk Not applicable for patient blinding but out-

come assessor blinding was not reported


All outcomes

Low risk Due to high drop out rate after cross-over,

only the first half of the study (pre cross-

over) was analysed as a parallel study


Free of other bias? Unclear risk Compliance was measured.

Doyle 2004







Participants WHO PARTCIPATED: 32 participants were included in this study. 16 were randomised

to and completed CSII. 16 were randomised to MI and 15 completed MI


AGE (mean years (SD)): The mean age of participants randomised to CSII was 12.5 (3.

2) years and the mean age of participants randomised to MI was 13 (2.8) years

ETHNIC GROUPS (%): There were 11 white (69%), 3 hispanic (19%) and 2 black



Doyle 2004 (Continued)

(12%) participants in the CSII group and in the MI group there were 13 white (81%),

2 (13%) hispanic and 1 (6%) black participants

DURATION OF DISEASE (mean years (SD)): The mean disease duration in partic-

ipants randomised to CSII was 6.8 (3.8) years and in those randomised to MI, mean


INCLUSION CRITERIA: Eight to 21 years of age, otherwise healthy except for treated

thyroid or celiac disease, treated with insulin for at least six months, naïve to CSII and

glargine, willing to do at least four blood tests per day and had a screening HbA1c

between 6.5% and 11%






COUNTRY/ LOCATION: Connecticut, USA


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The Medtronic MiniMed 508

or Paradigm 511 was used to deliver insulin aspart. The initital base dose was 50% of the

total daily insulin dose. Initial carbohydrate-to-insulin ratios and correction doses were

based on their prerandomisation insulin doses or on age and pubertal stage. Participants

were advised to use the prescribed carbohydrate-to-insulin ratio for all meals and snacks

that were ≥15 g carbohydrates. Participants were contacted daily in the first two weeks

for dosage adjustments. Blood glucose targets were 70 to120 mg/dl before meals and

90 to 150 mg/dl at bed time. Participants were instructed to treat two consecutive high

blood glucose levels as potential catheter occulsions, and were advised to change the site

and take a correction dose of aspart by injection

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The initial dose of glargine was

calculated as 80% of their total daily insulin dose of NPH or lentre according to usual

practice guidelines with glargine. Initial carbohydrate-to-insulin ratios and correction

doses were based on their pre-randomisation insulin doses or on age and pubertal stage.

Particpants were advised to use the prescribed carbohydrate-to-insulin ratio for all meals

and snacks that were ≥15 g carbohydrates. Participants were contacted daily in the first

two weeks for dosage adjustments. Blood glucose targets were 70 to 120 mg/dl before

meals and 90 to 150 mg/dl at bed time. Glargine was given in the morning or at bed

time





ADDITIONAL OUTCOMES: HbA1c, blood glucose, insulin dose, adverse events,

quality of life



RUN-IN PERIOD: For a period of 1 to 2 weeks, participants were asked to do four finger

stick blood glucose tests per day, and instructed to keep written records of the blood

glucose levels. These were evaluated by the investigator, along with the participant’s ability



Doyle 2004 (Continued)

to count carbohydrates. If further teaching was necessary, the participant met with the

dietician before the next visit. Participants were instructed on how to treat hypoglycemic

episodes with glucose tablets and were sent home with instructional aids such as videos

and written literature about CSII and MI therapy to review before randomisation


COMMERCIAL FUNDING: Yes - pharmaceutical and industry.

NON-COMMERCIAL FUNDING: Yes - government and foundations.




Stated aim of study “The present randomized clinical trial was undertaken with a principal aim of comparing

the efficacy of CSII and MDIs with glargine lin lowering HbA1c levels in youth with

type 1 diabetes in this short-tern study”

Notes

Risk of bias


Adequate sequence generation? Low risk Participants stratified according to sex and

age, within each stratum, a randomisa-

tion scheme was generated using a random

number table with a block size of four. Con-

ducted by the centre’s Investigational Phar-

macy


Blinding?

All outcomes


outcome assessor blinding was not reported


All outcomes

Low risk Intention-to-treat analysis described. The

study investiagtors reported 3.10% drop-

outs/withdrawals/losses to follow up, rea-

sons for these were described


Free of other bias? Unclear risk Compliance was measured only in the first

1-2 week run in phase. Selection criteria

not reported



Hanaire-Broutin 2000








were randomised to and completed CSII, and 21 participants were randomised to MI

but only 20 completed MI

SEX (female% / male%): 20 female, 21 male



DURATION OF DISEASE (mean years (SD)): 20 (11.3) years

INCLUSION CRITERIA: HbA1c < 10%, negative C-peptide, and experience of inten-

sified insulin therapy. None had untreated retinopathy, impaired renal function, gastric

neuropathy, BMI > 30 kg/m2, daily insulin dose > 2 U/kg, history of hypoglycaemia

unawareness, or any sever disease that could interfere with study

EXCLUSION CRITERIA: As above

DIAGNOSTIC CRITERIA: History of type 1 diabetes and C-peptide < 0.3 ng/ml or

negative Glucagon test




COUNTRY/ LOCATION: Toulouse, France


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The MiniMed 506 or 507 and

Htron D or V was used to deliver lispro Humalog U-100 according to glycemic targets:

70 to 120 mg/dl two hours before meal or during fasting and 120 to 180 mg/dl two

hours after meal

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Three injections of lispro before

meals and two injections of NHP before breakfast and at bedtime. Lispro Humalog U-

100 was delivered according to glycemic targets: 70 to 120 mg/dl two hours before meal

or during fasting and 120 to 180 mg/dl two hours after meal. Participants were advised

to add a third injection of NPH at noon if before dinner blood glucose was not within

the target range

TREATMENT BEFORE STUDY: At enrolment, 32 participants were treated by CSII

with regular insulin and nine participants were treated by MI with regular insulin or

insulin lispro




ADDITIONAL OUTCOMES: Insulin dose, weight, HbA1c, glycemia, hypoglycaemia,

satisfaction



RUN-IN PERIOD: Regardless of previous treatment all participants were submitted to



Hanaire-Broutin 2000 (Continued)

six week run in period of CSII therapy with regular insulin


COMMERCIAL FUNDING: Yes





Stated aim of study “Therefore, the aim of the present work was to compare the efficacy on glycemic control

and hypoglygemia frequency of 2 new intensified insulin regimens, CSII and MDI, with

insulin lispro in a randomized study.”






Risk of bias


Adequate sequence generation? Unclear risk The study investigators used two tables of

randomisation, according to the previous

treatment of the patient used before the

study (injection or external pump)

Allocation concealment? Low risk Envelopes containing treatment allocation

were prepared before the beginning of the

study and were opened at the randomisa-

tion visit (#2) by the participant

Blinding?

All outcomes




All outcomes

Low risk 2.4% drop-outs/withdrawals/losses to fol-

low up. One participant dropped out in

first period of MI, he was orginially on CSII

and found MI too difficult





Hirsch 2005






CONTROLLED CLINICAL TRIAL (CCT): No


were randomised to start with CSII and 45 completed CSII and were subsequently

crossed over to MI. 50 participants were randomised to start with MI and 46 completed

MI and were subsequently crossed over to CSII


AGE (mean years (SD)): The mean age of participants starting with CSII was 41.7 (11.

1) years and the mean age of participants starting with MI was 44.2 (11) years

ETHNIC GROUPS (%): In the group starting with CSII, 48 participants were caucasian

(96%) and two were hispanic (4%) and in the group starting with MI, 48 participants

were caucasian (96%), one was hispanic (2%) and one was asian (2%)

DURATION OF DISEASE (mean years (SD)): The mean disease duration of partici-

pants starting with CSII was 19.7 (11.3) years and in those starting with MI, the mean


INCLUSION CRITERIA: People with type 1 diabetes who were ≥ 18 years, had BMI

≤ 40 kg/m2, and had HbA1c ≤ 9%.

EXCLUSION CRITERIA: Impaired hepatic or renal function, impaired cardiac func-

tion, hypoglycaemia awareness or recurrent major hypoglycemia, women of child bear-

ing age if they were pregnant, breast feeding or not practising contraception





COUNTRY/ LOCATION: USA


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Participants used own pumps

with insulin aspart

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Unit for unit basis single basal

bed time injection of insulin glargine and meal time boluses of insulin aspart delivered

by the Induo device (LifeScan, Milpitas, CA)

TREATMENT BEFORE STUDY: All participants were previously treated with CSII

for at least three months before the screening visit

TITRATION PERIOD: After randomisation, participants had one week to settle and

then four weeks study. At cross over, participants had one week to settle and four week

study



ADDITIONAL OUTCOMES: Glycemic control (fructosamine, blood glucose,

HbA1c), insulin dose, hypoglycemic events, hematology and blood chemistry



Hirsch 2005 (Continued)



RUN-IN PERIOD: Basal insulin coverage could be modified through the scheduled

phone communications with the participants during a one week dose adjustment period

using insulin aspart with the Induo device (LifeScan, Milpitas, CA) with the goal of

fasting and before dinner glycemia in the range of 90 to 126 mg/dl (5 to 7 mmol/L)







Stated aim of study “...we compared insulin aspart as the only approved rapid-acting analog for CSII with

insulin aspart/glargine in MDI therapy. Because the principles applied to mealtime

coverage were identical during both treatments, the study tests whether the theoretically

smooth profile of glargine will provide safety and efficacy comparable to the ”customized“

basal insulin delivery system that can be programmed by an infusion pump during CSII.

”






Risk of bias




Blinding?

All outcomes




All outcomes

Unclear risk 9% drop-outs/withdrawals/losses to follow

up, however, reasons for these are not re-

ported

Free of selective reporting? High risk Complete data for CSII vs MI at the end

of treatment or end of study not provided

for HbA1c and mean daily blood glucose

Free of other bias? Unclear risk Compliance was measured and reported.



Home 1982






CONTROLLED CLINICAL TRIAL (CCT): No

Participants WHO PARTCIPATED: 10 participants were included in this study. It is not reported

how many started with each intervention






EXCLUSION CRITERIA: Women who were pregnant, those taking drugs known to

alter carbohydrate metabolism (ie. oral contraceptives), renal disease (plasma creatinine

concentration > 2.26 mg/100ml), proliferative retinopathy, prior CVD event (ischemic

heart disease, myocardial infarct)





COUNTRY/ LOCATION: England

SETTING: General practice

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): 60% daily basal infusion and

40% preprandial boluses delivered by a portable infuser via injection cannula into the

abdominal wall using highly purified porcine insulin (Actrapid) and 0.15 mol/L saline

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Actrapid and Ultratardwas used.

Highly purified porcine neutral soluble insulin (Actrapid) was delivered by injection

before each main meal and highly purified bovine ultralente (Ultratard) was delivered

before main evening meal

TREATMENT BEFORE STUDY: Twice daily injection therapy.




ADDITIONAL OUTCOMES: Insulin dose, HbA1c, mean blood glucose, urinary glu-

cose, weight, hypoglycaemia, cholesterol, overall control, patient reaction



RUN-IN PERIOD: 15 months for initial intensive education program and blood glucose

monitoring study







Home 1982 (Continued)



Stated aim of study “In an attempt to answer these questions we have compared CSII with a thrice-daily

injection regimen in a crossover study, using patients whose control has previously been

optimised on twice-daily insulin injections.”


Risk of bias


Adequate sequence generation? Unclear risk Only reports that each patient was studied

on either intervention in random order


Blinding?

All outcomes




All outcomes

Unclear risk 9.1% drop-outs/withdrawals/losses to fol-

low up - one (of eleven) patient suffered

a myocardial infarction before complet-

ing the study, however the authors have

reported throughout the paper that there

were only ten participants from the begin-

ning of the study and it is not reported at

which stage the participant was at when

they withdrew



Hoogma 2005








were randomised to start with CSII and 129 participants were randomised to start with

MI. After six months the participants were crossed over, serving as their own controls.

223 participants completed the study



Hoogma 2005 (Continued)

SEX (female% / male%): 135 (53%) female, 121 (47%) male

AGE (mean years (SD)): The mean age of participants starting with CSII was 35.3 (9.

8) years and the mean age of participants starting with MI was 37 (10.6) years


DURATION OF DISEASE (mean years (SD)): The mean disease duration of those

starting with CSII was 14.4 (8.9) years and in those staring with MI, the mean disease

duration was 15.4 (10.1) years

INCLUSION CRITERIA: Type 1 diabetes and absent C-peptide secretion, aged 18 to

65 years, on multiple injection therapy for at least six months

EXCLUSION CRITERIA: Hypoglycaemia awareness, progressive retinopathy, renal

insufficiency (creatinine ≥ 250 µmol/L), acute coronary syndrome or cerebrovascular

accident within the last six months, uncontrolled hypertension, autonomic neuropathy,

planned or existing pregnancy, or other clinically significant concomitant disorders





COUNTRY/ LOCATION: Europe


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro via a multi pro-

grammable insulin infusion pump (H-TRON V100 or H-TRONplus V100, Disetronic

Medical Systems AG, Burgdorf, Switzerland). Adjusted according to glycaemic targets

i.e.. blood glucose 4.0 to 7.0 mmol/L fasting and before meals, blood glucose at bed

time 6.0 to 10.0 mmol/L and blood glucose 8.0 to 10.0 mmol/L one hour after meals.

Participants were advised to decrease or increase the dose of NPH insulin if fasting glu-

cose was repeatedly < 4.0 or > 7.0 mmol/L and to decrease or increase the dose of rapid-

acting insulin at meals if the one hour post prandial blood glucose was repeatedly < 8.0

or > 10.0 mmol/L. In case of high before dinner blood glucose values it was decided to

supplement NPH insulin at lunch time and at breakfast and dinner

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): At least three injections of insulin

lispro before meals and at least one injection of insulin NPH to achieve optimal control.

Adjusted according to glycaemic targets, as above

TREATMENT BEFORE STUDY: Multiple injection therapy for at least six months


Outcomes PRIMARY OUTCOME(S) (as stated in the publication): Mean HbA1c, daily blood

glucose fluctuation (defined as the average standard deviation of the mean daily blood

glucose, using measurements taken during the 14 days prior to the final visit in each

treatment period

SECONDARY OUTCOMES (as stated in the publication): mean daily blood glucose

values (from standardised 24 hour blood glucose profiles), severity of hypoglycaemic

episodes, quality of life (using the diabetes quality of life (DQoL) questionnaire, the

SF-12 questionnaire and an additional questionnaire related to lifestyle and therapy

manageability

ADDITIONAL OUTCOMES: Insulin requirements, body weight, injection and infu-

sion site problems, adverse events






RUN-IN PERIOD: Each treatment phase started with a two month run in period,

during which patients received education to optimise intensive insulin therapy







Stated aim of study “The aim of this study was to compare CSII with MDI with respect to metabolic

parameters - including HbA1c, blood glucose levels and fluctuation, and incidence of

hypoglycaemia, quality of life and adverse events, in an adequately powered RCT with

a crossover design.”






Risk of bias


Adequate sequence generation? Low risk Randomized 1:1 with a block size of four.

Allocation concealment? Low risk The study investigators used numbered,

sealed envelopes containing the group

number handed to the investigators. The

envelope with the lowest available number

had to be used first

Blinding?

All outcomes




All outcomes

Low risk Efficacy ITT cross-over cohort (229 partic-

ipants), safety cohort (256 participants)

272 were enrolled and randomised, 223

completed the study. Analysis was based

on the intention-to-treat cross-over cohort

(229 participants)

31 (15 in MDI-CSII, 16 in CSII-MDI)

withdrew during the first treatment phase,

12 completed phase one (two in MDI-

CSII, ten in CSII-MDI) but did not cross




over to the second phase. Six withdrew at

phase two (three in MDI-CSII, three in

CSII-MDI)

Main reason for withdrawal was refusal to

start CSII from group MDI-CSII, or to

leave CSII for group CSII-MDI


Free of other bias? Unclear risk Compliance was measured by review of pa-

tients diary regarding number of recorded

SMBG measurements

Lepore 2003








were randomised to and completed CSII and the other 16 participants were randomised

to and completed MI


AGE (mean years (SD)): In the CSII group, the mean age of the participants was 37.7

(11.2) years and in the MI group, the mean age of the participants was 42.9 (15.6) years


DURATION OF DISEASE (mean years (SD)): In the CSII group, the mean disease

duration was 19.6 (9.2) years and in the MI group, the mean disease duration was 14.7

(11.1) years

INCLUSION CRITERIA: Type 1 diabetes, age 18 - 65 years, poor metabolic control

(defined as HbA1c constantly above 8% in the previous year), treatment with multiple

daily insulin injections (regular or lispro insulin before each meal plus NPH as basal

insulin) for at least 1 year before the study, patients consenting to begin treatment with

CSII

EXCLUSION CRITERIA: Evidence of severe late complication of diabetes (i.e.. overt

nephropathy and/or proliferative retinopathy), significant cardiovascular and hepatic

disease, patients unwilling to perform at least four blood glucose measures per day

DIAGNOSTIC CRITERIA: ADA 1997






CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro at multiple basal

infusion rates plus boluses at meals



Lepore 2003 (Continued)

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro at each meal com-

bined with glargine injected at dinner or at bed time

TREATMENT BEFORE STUDY: Treatment with multiple daily insulin injections

(regular or lispro insulin before each meal plus NPH as basal insulin) for at least one

year before the study




ADDITIONAL OUTCOMES: Insulin requirements, weight, HbA1c, blood glucose,

hypoglycaemia, total cholesterol, HDL-cholesterol, triglycerides, uric acid

Study details DURATION OF INTERVENTION: 1 year

DURATION OF FOLLOW-UP: 1 year

RUN-IN PERIOD: During the run-in period all patients received a structured education

programme (eight hours) concerning intensive insulin therapy (blood glucose self mon-

itoring, insulin therapy, carbohydrate counting, self-management strategies). Patients

treated with CSII received another two sessions (eight hours) about CSII treatment




PUBLICATION STATUS (PEER REVIEW JOURNAL): Yes - Letter



Stated aim of study “For this purpose we evaluated, in an open parallel group trial of 1-year duration involving

32 type 1 diabetic patients that had been treated with MDIs (regular or lispro insulin

before each meal plus NPH as basal insulin) for at least 1 year, the efficacy of two regimens

of intensive insulin treatment: CSII versus MDIs with lispro at each meal plus glargine

as basal insulin.”

Notes

Risk of bias


Adequate sequence generation? Low risk Computer generated sequence.

Allocation concealment? Unclear risk A progressive identification number (from

1 to 32) was given to every participant

who was randomly allocated to interven-

tion groups using a randomisation software

on an Internet service provider

Blinding?

All outcomes

High risk Not applicable for participant blinding but




Lepore 2003 (Continued)


All outcomes

High risk


Free of other bias? Unclear risk Compliance - the study investigators evalu-

ated the number of blood glucose self mea-

sures recorded in the memory of glucose

meter in both groups, the data of insulin in-

jections recorded in the diary of the patient

in MI group, the data of bolus and basal

insulin infusion recorded in the memory of

pump in CSII group

Meschi 1982







Participants WHO PARTICIPATED: 16 participants were included in this study. Eight participants

were randomised to and completed CSII and the other eight participants were ran-

domised to and completed MI


AGE (mean years (SD)): The mean age of participants in the CSII group was 13.3 (4)

years and the mean age of participants in the MI group was 12.9 (2.4) years


DURATION OF DISEASE (mean years (SD)): Not reported








SETTING: In-patient

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Regular insulin was infused

through a 27 gauge short butterfly catheter implanted subcutaneously in the abdominal

wall. The portable battery driven Mill Hill Infuser pump Model 1001 was used to deliver

insulin. Preprandial insulin delivery was manually delivered 30 minutes before main

meals as single boluses. On the first day, the 24 hour infusion dose was calculated as 1.

5 U/kg of body weight divided into 0.75 U/kg as the basal rate and 0.75 U/kg in three

preprandial boluses (i.e.. 20% 30 minutes before breakfast, 40% before lunch and 40%

before dinner) adjusted to optimise metabolic control



Meschi 1982 (Continued)

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): To achieve the same daily insulin

distribution, the total 24 hour insulin dose (1.5 U/kg body weight) was divided into 0.

75 U/kg as lente insulin (two injections) and 0.75 U/kg as regular insulin injections 30

minutes before main meals (three injections)





ADDITIONAL OUTCOMES: Blood glucose, HbA1c, insulin dose, hypoglycaemia

Study details DURATION OF INTERVENTION: 6 days

DURATION OF FOLLOW-UP: 6 days

RUN-IN PERIOD: Four days in hospital with conventional insulin therapy of two daily

injections with a mixture of regular and lente insulin before breakfast and dinner. The

dose was increased until slight hypoglycaemic symptoms were noted



NON-COMMERCIAL FUNDING: Yes - Consiglio Nazionale delle Ricerche



PUBLICATION STATUS (ABSTRACT):No

Stated aim of study “We have therefore compared CSII with an intensified conventional insulin treatment

(3 daily insulin injections) (3II).”

Notes

Risk of bias




Blinding?

All outcomes

High risk Not applicable for participant blinding, but



All outcomes

Unclear risk Details not provided.





Nathan 1982







Participants WHO PARTICIPATED: Five participants were included in this study


AGE (mean years (SD)): The mean age of participants was 31 (5.7) years


DURATION OF DISEASE (mean years (SD)): 7 (1.8) years




CO-MORBIDITIES: None of the participants had significant neuropathy, nephropathy,

retinopathy, or vascular disease at the outset of the study



COUNTRY/ LOCATION: USA

SETTING: In-patient and out-patient

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): An auto syringe AS6C pump

was used with a 27-gauge pump catheter. Catheters were changed every two to seven days

depending on participant preference. About half of the total insulin dose was delivered

as basal infusion and the remainder was delivered as before meal boluses as per MI

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The timing and amount of doses

of intermediate, long-acting and rapid acting insulins were adjusted according to gly-

caemic targets. Subcutaneous NPH insulin (intermediate-acting) was delivered 30 to 60

minutes before breakfast and either 30 minutes before dinner or at 8pm. Regular insulin

was delivered from 15 to 60 minutes before each meal depending on the response of

participant

Glycemic target - to obtain fasting plasma glucose concentrations less than 90 mg/dL

and postprandial concentrations less than 150 mg/dL without hypoglycaemic reactions

TREATMENT BEFORE STUDY: Before the study, participants were using various

insulin regimens




ADDITIONAL OUTCOMES: HbA1c, mean plasma glucose, hypoglycaemic reactions,

daily insulin dose, fluorophotometry

Study details DURATION OF INTERVENTION: 8 to 12 weeks


RUN-IN PERIOD: Subjects were admitted to the General Clinical Research Center

of the Massachusetts General Hospital for an evaluation and adjustment of therapies

according to glucose profiles done at this time. Once randomised, participants were

monitored for two to three days as in-patients while the regimen was established



Nathan 1982 (Continued)







Stated aim of study “A crossover trial was conducted in five adult patients with type I diabetes to examine the

relative efficacy of intensive conventional and insulin pump therapies in a more realistic

outpatient setting.”

Notes One of the participants was crossed over three times, initially treated with MI for eight

weeks, them CSII for 12 weeks, MI for another four weeks and finally with CSII for 12

weeks

Another of the participants participated in a double blind cross-over fashion, so that

both the participant and the investigator were blind to the intervention. The participant

received both interventions simultaneously but when on the CSII phase, saline was

injected and when on the MI phase, saline was infused through the pump. This was

controlled by the participant’s wife, who was a registered nurse, so that the participant

was unaware of the cloudy insulin solution in either device

Standard deviations of the difference were extracted from paired t-test data

Risk of bias




Blinding?

All outcomes

Unclear risk One of the participants participated in a

double blind cross-over fashion, so that

both the participant and the investigator

were blind to the intervention. The partic-

ipant received both interventions simulta-

neously but when on the CSII phase, saline

was injected and when on the MI phase,

saline was infused through the pump. This

was controlled by the participant’s wife,

who was a registered nurse, so that the par-

ticipant was unaware of the cloudy insulin

solution in either device


All outcomes

Unclear risk Details not provided.




Nathan 1982 (Continued)

Free of other bias? Unclear risk Compliance was measured by blood glu-

cose measurements.

Nosadini 1988







Participants WHO PARTICIPATED: 64 participants were included in this study (25 female, 39

male), however the following groups are relevant to this review: MI (ICIT), CSII

(CSII-fixed basal overnight insulin infusion rate (FBR) and CSII-higher programmable

overnight insulin infusion rate (HOR)) there were 44 participants (16 female, 28 male)

CSII-FBR - 19 participants randomised to and completed.

CSII-HOR - 10 participants randomised to and completed.

MI-ICIT - 15 participants randomised to and completed.

SEX (female% / male%):

CSII-FBR - 8 female (42%), 11 male (58%)

CSII-HOR - 4 female (40%), 6 male(60%)

MI-ICIT - 4 female (27%), 11 male (73%)

AGE (mean years (SD)):

CSII-FBR - 36 (6) years

CSII-HOR - 34 (3) years

MI-ICIT - 32 (9) years


DURATION OF DISEASE (mean years (SD)):

CSII-FBR - 8 (3) years

CSII-HOR - 7 (3) years

MI-ICIT - 7 (4) years

INCLUSION CRITERIA: Diabetes according to NDDG, absent C-peptide response

EXCLUSION CRITERIA: Refusal of assignment to group, present C-peptide response

DIAGNOSTIC CRITERIA: NDDG






CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The Microjet MC20 was used

to deliver insulin on the CSII-FBR regimen and the Betatron II was used to deliver insulin

on the CSII-HOR regimen. The insulin dose was adjusted according to the glycaemic

target of a blood glucose value after and overnight fast between 90 to 110 mg/dL and

of after meal value between 130 to 160 mg/dL. The dose of the before meal insulin was

decided based on the before meal blood glucose value. The dose of intermediate insulin

in the evening was decided on the basis of the overnight fasting blood glucose value.



Nosadini 1988 (Continued)

Each participant was seen every three weeks to modify the regimen to achieve better

blood glucose control

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Three daily soluble insulin injec-

tions were delivered during the day with a long-acting insulin in the evening. Participants

were allowed to modify each insulin injection (regular and long acting) within a 4 U

range. Further changes had to be by physician by phone or during examination. The

insulin dose was adjusted according to glycaemic targets of a blood glucose value after

an overnight fast between 90 to 110 mg/dL and of an after meal value between 130 to

160 mg/dL. The dose of before meal insulin was decided based on before meal blood

glucose value. The dose of intermediate insulin in the evening was decided on the basis

of the overnight fasting blood glucose value. Each participant was seen every three weeks

to modify the regimen to achieve better blood glucose control

TREATMENT BEFORE STUDY: Two to three injections per day.




ADDITIONAL OUTCOMES: plasma glucose, mean daily blood glucose, HbA1c, basal

and glucagon stimulated C-peptide, urinary albumin, hyper- and hypoglycaemic events

Study details DURATION OF INTERVENTION: 1 year

DURATION OF FOLLOW-UP: 1 year

RUN-IN PERIOD: Metabolic data for each participant were recorded for three months

while on existing treatment (two to three injections per day) and glucagon challenge was

performed. Each participant spent at least a week in hospital doing group teaching from

a team of a physician, diabetes nurse and dietician. Training in self monitoring and self

regulation of therapy was also conducted



NON-COMMERCIAL FUNDING: Yes - CNR Italy’s National Research Council




Stated aim of study “The purpose of the present investigation was to assess whether intensified conventional

insulin therapy and continuous subcutaneous insulin infusion are characterized by any

difference in the incidence of hypoglycaemic and ketoacidotic episodes in comparison

with more conventional insulin regimes.”

Notes

Risk of bias





Nosadini 1988 (Continued)

Allocation concealment? Low risk The type of treatment was enclosed in an

envelope and opened by participant

Blinding?

All outcomes




All outcomes

High risk



Nuboer 2008








(13 female, 7 male) were randomised to CSII and 19 completed (12 female, 7 male),

and the other 19 participants (9 female, 10 male) were randomised to and completed

MI


AGE (mean years (SD)): The mean age of participants in the CSII group was 10 (3)

years (excluding one drop out) and the mean age of participants in the MI group was 10

(3.7) years


DURATION OF DISEASE (mean years (SD)): Disease duration in the CSII group was

5.6 (3.3) years (excluding one drop out) and in the MI group was 4.7 (2.9) years

INCLUSION CRITERIA: Type 1 diabetes, diagnosed by the presence of islet antigen-

2, glutamic acid decarboxylase-65 or islet cell cytoplasmic autoantibodies, daily insulin

administration for 1 yr or longer, random C-peptide < 200 pmol, HbA1c > 8.0%, a

history of repeated symptomatic hypoglycaemia, age four to 16 years, and attendance of

a regular school

EXCLUSION CRITERIA: Clinically manifest chronic complications, pregnancy, co-

morbidity, mental retardation, psychiatric treatment or symptoms in a child or a parent,

insufficient

Dutch language capabilities and absence of a telephone at home






Nuboer 2008 (Continued)


COUNTRY/ LOCATION: The Netherlands


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): For CSII, only insulin aspart was

used. H-tron Disetronic insulin pumps (Roche, Basel, Switzerland) were used, changing

the infusion catheter every two days. CSII insulin dosage was initiated as 75% of the total

daily MI dose, mostly 50% as basal insulin day and night and 50% premeal, including

snacks

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Three short-acting insulin doses

before breakfast, lunch and supper. During MI, 26 children used insulin aspart and 12

used regular insulin before meals. At bed time, longer acting insulins were given, 23

using intermediate-acting insulin neutral protamine Hagedorn (NPH) and 15 insulin

glargine

TREATMENT BEFORE STUDY: Participants had a 3.5 month run-in period


Outcomes PRIMARY OUTCOME(S) (as stated in the publication): Not specified

SECONDARY OUTCOMES (as stated in the publication): Not specified

ADDITIONAL OUTCOMES: Disease impact using the Diabetes Quality of Life Ques-

tionnaire, Pediatric Quality of Life Inventory as rated by parents and children, HbA1c,

daily insulin dose (U/kg), and adverse events

Study details DURATION OF INTERVENTION: 10.5 months for CSII and 3.5 months for MDI

DURATION OF FOLLOW-UP: 10.5 months

RUN-IN PERIOD: 3.5-month run-in phase on MDII for all children, consisting of

injecting three short-acting insulin doses before breakfast, lunch and supper



NON-COMMERCIAL FUNDING: Yes - Ministry of Health and the trust Funds of

National Health Insurance CIEs and of University Hospitals (VAZ/VWS). Additional

funding was obtained from the Child Health and Wellbeing Fund, the Volkskracht Fund

and the van Leeuwen Lignac Fund, all in Rotterdam, the Netherlands




Stated aim of study “The aim of the present study was to investigate the changes in quality of life and of

impact of disease by either CSII or MDII prospectively in a randomized study preceded

by a 3.5-month run-in phase of MDII, following medical effects simultaneously.”

Notes

Risk of bias





Nuboer 2008 (Continued)

Allocation concealment? Low risk After the 3.5 months of run-in phase, each

child was randomly assigned to either (on-

going) MDII or CSII using the ’closed

envelop’ method, supplied by the Public

Health Department

Blinding?

All outcomes

Unclear risk Not applicable for participant blinding, but



All outcomes

Low risk Of the 39 children enrolled, 38 completed

the study. One 11 year old girl consistently

refrained from the pump phase of CSII af-

ter three weeks because of its visibility on

the beach


Free of other bias? Low risk Compliance was checked - Injection sites

were checked at all visits and rotation ad-

vised accordingly with intensive out-pa-

tient guidance, including the 24 hour tele-

phone service

Oslo Study 1985 to 1992







Participants WHO PARTICIPATED: 15 participants randomised to CSII and 15 randomised to

MI. It is unclear how many participants completed the study


AGE (mean years (SD)): The mean age of participants in the CSII group was 26 years

(range 18-38) and the mean age of participants in the MI group was 26 years (range 19-

42)


DURATION OF DISEASE (mean years (SD)): Disease duration in the CSII group was

153 months (range 77-280) and in the MI group was 154 months, (range 81-250)

INCLUSION CRITERIA: Age between 18 to 45 years, diagnosis of diabetes before

30 years of age, duration of disease more than seven but less than 30 years, no clinical

signs of nephropathy or systemic hypertension, no history of neuropathy, no proliferative

retinopathy, not pregnant, negative for C-peptide, using no medication other than insulin

(except contraceptives)





Oslo Study 1985 to 1992 (Continued)



Interventions NUMBER OF STUDY CENTRES: Multi-centre

COUNTRY/ LOCATION: Norway


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Autosyringe AS6C pump was

used in 12 participants and the Nordisk infuser was used in 3 participants during the

first two years. After the second year the following pumps were used: Nordisk Infuser,

Autosyringe AS6C, Eugly, Graseby MS26, Nipro. Subcutaneous abdominal site of in-

fusion. Approximately half of the daily insulin dose was given continuously and rest as

boluses 15 minutes pre meals. Basal and meal doses were adjusted individually in each

participant, according to glycaemic targets. Thirteen participants used four pre meal bo-

luses per day, one used three and one used six. The Nordisk infuser, Graseby and Eugly

pumps delivered regular porcine insulin at a concentration of 100 U/ml, the other pumps

delivered 40 U/ml. In the auto syringe AS6C pumps, daily insulin dose was diluted to 2.

8 ml by the corresponding insulin buffer. Fourteen participants received Velosulin and

one participant received Actrapid

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Regular porcine insulin was in-

jected 15 to 30 minutes pre meal and NPH insulin was given at bed time. Fourteen

participants were injected five times per day and one participant was injected four times

per day. Doses were adjusted according to glycaemic targets. Patients were allowed to

inject via butterfly needle for the first two years. After the second year Novopen was used

for injection of regular insulin





ADDITIONAL OUTCOMES: Insulin requirements, weight, HbA1c, blood glucose

(mean and fasting), hypoglycaemia, peripheral nerve function, retinopathy

Study details DURATION OF INTERVENTION: 4 years

DURATION OF FOLLOW-UP: 4 years. Mean follow up was 40 months (range 24 to

46), all follow up was completed by the same two clinicians in the outpatient clinic after

one and three weeks, monthly during the first year, then bi-monthly

RUN-IN PERIOD: Two month pre-period with home blood glucose monitoring. All

participants injected NPH and regular porcine insulin pre breakfast and pre dinner.

Appropriate dose adjustments were made aiming to maintain blood glucose between 3

and 7 mmol/L pre meals without hypoglycaemia

Publication details LANGUAGE OF PUBLICATION: English and Norweigan

COMMERCIAL FUNDING: Yes - Nordisk Gentofte, Boehringer Mannheim

NON-COMMERCIAL FUNDING: Yes - Norwegian Council for Science and the

Humanities, Norwegian Diabetes Association, Norwegian Council on Cardiovascular

Disease, University of Oslo, Jahres Medical Foundation






Oslo Study 1985 to 1992 (Continued)

Stated aim of study “1. To study the influence of long term near-normoglycaemia one early stages of mi-

croangiopathy and neuropathy in young insulin dependent diabetes patients

2. To compare blood glucose control in three different insulin regimens:

-Continuous subcutaneous insulin infusion with portable pumps (CSII)

-Multiple insulin injections (MI)

-Optimized conventional treatment with two daily insulin injections

all regimens combined with home blood glucose monitoring.

3. To assess the frequency of acute complications of treatment

4. To establish a routine laboratory method for the determination of stable glycosylated

haemoglobin (HbA1c) to accurately assess blood glucose control.”

Notes This study was published in multiple journals at different time points. For this review,

only data reported at the four year time point have been used, since after four years, the

participants were no longer randomised

Risk of bias


Adequate sequence generation? Low risk Block randomisation by computer pro-

gram.

Allocation concealment? Unclear risk Statistician without prior knowledge of

participants or mode of treatment

Blinding?

All outcomes

High risk Not applicable.


All outcomes

High risk



Pozzilli 2003







Participants WHO PARTICIPATED: 23 participants were included in this study. Eleven participants

were randomised to CSII and seven completed. Twelve participants were randomised to,

and completed MI




Pozzilli 2003 (Continued)

AGE (mean years (SD)): 18.4 (9) years. In the completed CSII group, the mean age of

participants was 17.9 (9) years and the mean age of participants in the MI group was

18.9 (8.5) years


DURATION OF DISEASE (mean years (SD)): ≥ 1 week

INCLUSION CRITERIA: Diagnosis according to ADA with age presentation between

12 to 35 years, including positivity for islet cell autoantibodies; duration of clinical disease

(since beginning of insulin therapy) less than four weeks; no medical contraindications

(including pregnancy) or any other major chronic disease; and willingness and capability

to participate in regular follow up


DIAGNOSTIC CRITERIA: ADA



Interventions NUMBER OF STUDY CENTRES: Multi-centre



CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The 507C Medtronic MiniMed

Pump was used to deliver Humalog adjusted according to glycaemic targets. Participants

calculated before meal boluses by means of carbohydrate counting and individualised

carbohydrate/insulin ratios. In participants with no history of severe hypoglycaemia, the

before meal target blood glucose range was 3.9 to 8.3 mmol/L. In participants with

reduced hypoglycaemic awareness, the before meal target blood glucose range was 4.4

to 8.9 mmol/L. If hypoglycaemia was recurrent, a higher range was used

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Three injections of short acting

insulin at meal times plus NPH at bed time. The dose was adjusted according to glycaemic

targets. If before meal blood glucose was below 6.5 mmol/L, the dose was decreased by

10%, if blood glucose levels were consistently below 4.5 mmol/L for more than three

days, the dose was decreased by 20%. Therapy was discontinued if two hours after meal

blood glucose was consistently below 7.5 mmol/L. Participants with blood glucose above

10 mmol/L received a 10% increase in their dose or had their regimen modified





ADDITIONAL OUTCOMES: HbA1c, drug toxicity, basal C-peptide concentration,

insulin requirement, body mass index

Study details DURATION OF INTERVENTION: 2 years

DURATION OF FOLLOW-UP: 2 years




NON-COMMERCIAL FUNDING: Yes - International Centre for Diabetes Research,

Italy




Pozzilli 2003 (Continued)



Stated aim of study “The present study was designed to find out whether more intense insulin therapy such

as continuous subcutaneous insulin infusion (CSII) together with NA administration

for up to 2 years after Type 1 diabetes diagnosis would be acceptable to patients with

recent-on-set Type 1 diabetes and improve the metabolic control compared with patients

receiving the intensive subcutaneous insulin therapy (ISIT) mode of intervention.”

Notes

Risk of bias


Adequate sequence generation? Low risk Permutated block design.


Blinding?

All outcomes




All outcomes

Low risk Four (17%) dropped out (first four ran-

domised to CSII decided to drop out be-

fore starting treatment)


Free of other bias? Unclear risk Compliance not reported.

Saurbrey 1988







Participants WHO PARTICIPATED: 21 participants were included in this study. Of the 21 partici-

pants randomised, 19 completed the study (13 in CSII and 6 in MI) (8 female, 11 male)


AGE (mean years (SD)): 32 (2.1) years



INCLUSION CRITERIA: Unclear (with IDDM and able to comply with intensive

insulin regimen, were assumed to be C-peptide negative)




Saurbrey 1988 (Continued)

DIAGNOSTIC CRITERIA: Unclear

CO-MORBIDITIES: Twelve participants had background retinopathy, four partici-

pants had symptoms of incipient sensory neuropathy and three had intermittent pro-

teinuria


Interventions NUMBER OF STUDY CENTRES: Single centre


SETTING: In-patient for the first five days. Week three to six were out-patient. Week ten

was in-patient for change of treatment. Weeks 13, 16, 20 were out-patient. Participants

had access to a physician 24/7

CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The Auto-Syringe pump (n=

6) or the Medix pump (n=13) were used to deliver insulin. Insulin dose was adjusted

according to glycaemic targets and participants were advised to inject an additional 2 to

4 U of soluble insulin if blood glucose went above 12 mmol/L

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Three or more before meal in-

jections of short acting insulin (Actrapid) with a novopen plus one injection of inter-

mediated-acting insulin delivered at bedtime with conventional syringe. Insulin dose

was adjusted according to glycaemic targets and participants were advised to inject an

additional 2 to 4 U of soluble insulin if blood glucose went above 12 mmol/L

TREATMENT BEFORE STUDY: All participants were treated with twice daily injec-

tions of short-acting plus intermediate-acting or long-acting




ADDITIONAL OUTCOMES: Blood glucose, HbA1c, number of hypoglycaemic

episodes, participant’s attitudes about treatments



RUN-IN PERIOD: Over three weeks, participants were taught self monitoring of blood

glucose using BM-glycemie strips (Boehringer, Mannheim) and a Hypocount B meter

(Hypoguard, Melton, England)







Stated aim of study “The aim of the present cross-over study was to investigate whether CSII treatment and

ICT using NovoPen for the pre-meal injections of short-acting insulin, were equally

acceptable in daily use and also equally efficient in achieving good metabolic control in

patients with IDDM.”



Saurbrey 1988 (Continued)






Risk of bias




Blinding?

All outcomes




All outcomes


low up - two participants found wearing

the pump unacceptable


Free of other bias? Unclear risk Participants were asked to report blood glu-

cose etc on a daily basis (as a possible mea-

sure of compliance)

Schmitz 1989







Participants WHO PARTICIPATED: Ten participants were included in this study. The participants

were randomised to either CSII or MI for six months, after which they were crossed

over for another six months serving as their own control. All participants completed the

study





INCLUSION CRITERIA: Diabetes duration ≥ 20 years, absence of severe diabetic late

complications i.e.. urinary albumin excretion rate (UAE) ≤ 20µg/min, normal blood

pressure and serum creatinine and no proliferative retinopathy. Furthermore, the patients



Schmitz 1989 (Continued)

should have no need for medication other than insulin




CO-MEDICATIONS: Part of the inclusion criteria was that the participants should

have no need for medication other than insulin




CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin was delivered by the

Nordisk Infuser using Pharmaplast 25-gauge catheters and a 5.7 ml carpule of highly

purified porcine regular insulin (Velosulin, 100 IU/ml). Participants received three to six

boluses during the day

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Participants delivered three to

four bolus injections per day using the Insuject (Nordisk) containing carpules of 2 ml

Velosulin, 100 IU/ml during day time and an injection of NPH insulin at bed time

(Insulatard, 100 IU/ml)





ADDITIONAL OUTCOMES: Weight, insulin dose, HbA1c, blood glucose, kidney

function (glomerular filtration rate, renal plasma flow, urinary albumin excretion)





COMMERCIAL FUNDING: Not reported

NON-COMMERCIAL FUNDING: Not reported




Stated aim of study “The aims of the present investigation were in this type of patient, 1) to evaluate glycaemic

control on CSII versus MIT treatment as well as the preferential treatment modality of

the patients and 2) to study the effect of optimised glycaemic control on kidney function

in very long-term uncomplicated IDDM.”

Notes The measure of variance was not specified (i.e.. whether SD or SE), the data has been

treated as SD

There were ten participants in this study, however information about the number of

participants in each group was not provided. For data analysis, it has been assumed that

each group comprised five participants

Standard deviations of the difference were extracted from paired t-test data



Schmitz 1989 (Continued)

Risk of bias




Blinding?

All outcomes




All outcomes

Low risk There were no drop-outs/withdrawals/

losses to follow up.



Skogsberg 2008







Participants WHO PARTICIPATED: 72 participants were randomised and 67 participants com-

pleted the study


AGE (mean years (SD)): The mean age of the participants in the CSII group was 11.8

(4.9) years and the mean age of the participants in the MI group was 12.3 (4.5) years.

Standard deviations are not provided, instead 95% confidence intervals are provided


DURATION OF DISEASE (mean years (SD)): The mean disease duration in the CSII

group was 12.2 (2.0) years and the mean disease duration in the MI group was 10.4 (1.

7) years



DIAGNOSTIC CRITERIA: National guidelines for children with diabetes in Sweden

- National Guidelines for Children with Diabetes in Sweden. Lund: Studentlitteratur,

1996





Skogsberg 2008 (Continued)

Interventions NUMBER OF STUDY CENTRES: Multicentre (n=9)

COUNTRY/ LOCATION: Sweden


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The CSII group used insulin

aspart (NovoRapid; Novo Nordisk). The insulin pumps that were used in this study were

H-Tron (Roche, Burgdorf, Switzerland)

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): The MI group received morning

and bed time intermediate-acting insulin NPH (Insulatard; Novo Nordisk, Bagsvaerd,

Denmark) and rapid-acting insulin, aspart (NovoRapid; Novo Nordisk, Bagsvaerd, Den-

mark) three to four times a day as mealtime insulin. The NPH injections were given as

separate injections, that is, not mixed with the rapid-acting insulin



Outcomes PRIMARY OUTCOME(S) (as stated in the publication): Outcomes not specified as

primary or secondary

SECONDARY OUTCOMES (as stated in the publication): Outcomes not specified as

primary or secondary

ADDITIONAL OUTCOMES: Hemoglobin A1c (HbA1c), insulin dosage, number of

contacts with the hospital for ketoacidosis (defined as pH < 7.3), severe hypoglycemic

episodes and/or technical problems, and the Diabetes Treatment Satisfaction Question-

naire (DTSQ)


DURATION OF FOLLOW-UP: 24.68 weeks for CSII and 28.80 weeks for MI




NON-COMMERCIAL FUNDING: No - This research was partly supported by Re-

search and Development Center, county of Gavleborg’, Sweden, The Swedish Children’s

Diabetes Foundation’, Novo Nordisk Scandinavia, and Roche Diagnostics, Sweden




Stated aim of study ”We therefore designed this study to compare CSII with MDI in newly diagnosed diabetic

children and adolescents. Our outcome measurements were metabolic control, safety,

and treatment satisfaction

Notes

Risk of bias


Adequate sequence generation? Unclear risk Details not reported.



Skogsberg 2008 (Continued)

Allocation concealment? Unclear risk Details not reported.

Blinding?

All outcomes

Unclear risk Not applicable for participant blinding, but



All outcomes

Low risk Of 72 participants, 67 completed the whole

study. Two in the MDI group changed to

insulin pump between 18 and 21 months

at their own/parents’ request. Two, also

in the MDI group, were switched to the

new long-acting insulin analogue, glargine,

after 18 months (initiated by the dia-

betes team). One additional participant in

the MDI group dropped out around 12

months. These participants did not differ

from the others in the MDI group regard-

ing metabolic control or treatment satisfac-

tion when they dropped out of the study

Free of selective reporting? Unclear risk Details not reported.


Tsui 2001








(five female, eight male) were randomised to and 12 completed CSII. 14 participants

were randomised to and completed MI (four female, ten male)


AGE (mean years (SD)): The mean age of the participants in the CSII group was 36

(12) years and the mean age of the participants in the MI group was 36 (10) years


DURATION OF DISEASE (mean years (SD)): The mean disease duration in the CSII

group was 17 (10) years and the mean disease duration in the MI group was 15 (9) years

INCLUSION CRITERIA: Adults between 18 to 60 years with endocrine diagnosis

of type 1 diabetes for more than two years, with onset before 40 years, and able to

comply with treatment regimen. Participants were required to be treated with one to

two injections per day and be motivated to use CSII

EXCLUSION CRITERIA: History of more than two severe hypoglycemic events (coma,

seizure, loss of conciousness) in the last year, hemoglobinopathy, insulin resistance, ex-

treme obesity (BMI > 35 kg/m2), severe late complications of diabetes, evidence of sig-



Tsui 2001 (Continued)

nificant cardiovascular disease, hepatic disease, cancer or cerebrovascular or severe pe-

ripheral vascular disease, alcohol or drug abuse and/or participation in another clinical

trial in the past four weeks. Females who were or intending to be pregnant were also

excluded

DIAGNOSTIC CRITERIA: ADA criteria




COUNTRY/ LOCATION: Canada


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro (Humalog, Eli

Lilly, Indianapolis, IN) was delivered by a MiniMed (Sylmar, CA) 507 insulin infusion

pump

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Insulin lispro (Humalog) was

delivered before meals and with NPH insulin at bed time (no later than 11.30pm)

TREATMENT BEFORE STUDY: Part of the inclusion criteria was that participants

were required to be treated with one to two injections per day




ADDITIONAL OUTCOMES: Insulin requirements, HbA1c, blood glucose profile,

hypoglycaemia, quality of life



RUN-IN PERIOD: Unclear (there was a two week screening period)







Stated aim of study “To evaluate glycaemic control, hypoglycaemic events and quality of life in patients

treated with continuous subcutaneous insulin infusion (CSII) and multiple daily insulin

injection (MDI), with insulin lispro as the principle insulin.”

Notes

Risk of bias


Adequate sequence generation? Low risk Computer-generated randomisation

schedule used.



Tsui 2001 (Continued)

Allocation concealment? Low risk Sealed envelopes that had been indepen-

dently prepared.

Blinding?

All outcomes

High risk Not applicable for participant blinding but



All outcomes

Low risk One drop out from CSII, 3.7% drop-outs/

withdrawals/losses to follow up. ITT con-

ducted



Weintrob 2004







Participants WHO PARTICIPATED: 23 participants were included in this study. Eleven participants

were randomised to start with CSII-MI and ten completed CSII-MI. Twelve participants

were randomised to start with MI-CSII and ten completed MI-CSII


AGE (mean years (SD)): The age range of the participants was 9.25 to 13.75 years,

median 11.9 years


DURATION OF DISEASE (mean years (SD)): The disease duration was 2.5 to 11

years, median 6 years

INCLUSION CRITERIA: Age 8 to 14 years, treated with insulin for at least two years,

deficient C-peptide secretion (fasting level, < 0.6 ng/mL), absence of other health prob-

lems except for treated hypothyroidism or antibody-negative celiac disease controlled

with gluten-free diet

EXCLUSION CRITERIA: Participants with microvascular complications, mental re-

tardation or a psychiatric disorder were excluded from this study

DIAGNOSTIC CRITERIA: ADA 1997




COUNTRY/ LOCATION: Israel


CSII (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Participants used their pre study

insulin and doses were adjusted according to glycemic targets and amount of carbohy-

drates per meal according to 1U of insulin per 10 to 20 grams of carbohydrates. The



Weintrob 2004 (Continued)

target range for glycemia was 80 to 150 mg/dL (4.4 to 8.3 mmol/L) before meals and at

midnight and 120 to 180 (6.7 to 10.2 hours) after meals. Regular insulin was delivered

20 to 30 minutes before meals and lispro was delivered immediately before meals and

snacks. Participants were asked to change the infusion site every three days and under

the following conditions: when blood glucose > 300 mg/dL (16.7 mmol/L) with positive

urine ketones or when blood glucose was > 300 mg/dL (16.7 mmol/L) and failed to

respond to a corrective insulin dose, and at any sign of insertion site infection

MI (ROUTE, TOTAL DOSE/DAY, FREQUENCY): Participants used their pre study

insulin and doses were adjusted according to glycaemic targets and amount of carbohy-

drates per meal. Participants were instructed to use a fixed caloric amount of carbohy-

drates for each meal and then adjust insulin according to postprandial glucose values.

Insulin consisted of combined neutral protamine Hagedorn (NPH) and regular insulin

before breakfast, regular insulin before lunch and supper and NHP at bed time

TREATMENT BEFORE STUDY: Three or more injections of insulin in the form

of Insulatard and Actrapid (Novo Nordisk, Bagsvaerd, Denmark) or Humalin N and

Humalin R (Eli Lilly, Inianapolis, Ind)




ADDITIONAL OUTCOMES: Sensor performance, insulin requirements, body

weight, HbA1c, blood glucose, hypo- and hyperglycaemia, quality of life, body mass

index

Study details DURATION OF INTERVENTION: 3.5 months


RUN-IN PERIOD: All children were treated with three or more injections of insulin

(insulatard and actrapid or humulin N and Humulin R) for two weeks







Stated aim of study “In the present randomized crossover study, we compared the rate, degree, and duration of

hypoglycaemia and hyperglycaemia and the variability in glucose levels between children

and adolescents treated by CSII or MDI. We also sought to determine whether the

CGMS (Continuous Glucose Monitoring System) can improve glycaemic control and

glucose patterns.”


Risk of bias





Weintrob 2004 (Continued)


Blinding?

All outcomes




All outcomes


low up. One out of 23 dropped out because

they were afraid to insert the sensor



Characteristics of excluded studies [ordered by study ID]

Study Reason for exclusion

Alemzadeh 2005 Not a randomised controlled trial.

Amemiya 1983 Not a randomised controlled trial.

Arias 1985 Not a randomised controlled trial and the number of multiple insulin injections per day is less than

three

Bagdade 1991 The number of multiple insulin injections per day is less than three

Bangstad 1994 The number of multiple insulin injections per day is less than three

Barbosa 1981 The number of multiple insulin injections per day is less than three

Barrett 1995 Not a randomised controlled trial.

Bech 1996 Following all reasonable attempts to contact the authors, we were unable to confirm the number of

multiple injections per day

Beck-Nielsen 1984 The number of multiple insulin injections per day is less than three


Beck-Nielsen 1985a The number of multiple insulin injections per day is less than three


Behme 1984 Not a randomised controlled trial and the number of multiple insulin injections per day is less than

three



(Continued)

Bell 1985 Not a randomised controlled trial and the number of multiple insulin injections per day is less than

three

Bell 1985a Not a randomised controlled trial and the number of multiple insulin injections per day is less than

three

Bending 1986 The number of multiple insulin injections per day is less than three

Berg 1998 The number of multiple insulin injections per day is less than three

Bergenstal 1985 The number of multiple insulin injections per day is less than three

Berger 1989 Not a randomised controlled trial.

Bibergeil 1987 The number of multiple insulin injections per day is less than three

Biesenbach 1988 Not a randomised controlled trial.

Bode 1996 Not a randomised controlled trial.

Bode 2004 The number of multiple insulin injections per day is less than three

Bombor 1984 Not a randomised controlled trial.

Borisova 1989 Not a randomised controlled trial.

Brunetti 1984 Not a randomised controlled trial.

Buysschaert 1981 Not a randomised controlled trial.

Buysschaert 1983 Not a randomised controlled trial.

Calabrese 1982 Not a randomised controlled trial.

Canny 1985 The number of multiple insulin injections per day is less than three

Carta 1986 Following all reasonable attempts to contact the authors, we were unable to obtain extractable data

Cavallo-Perin 1983 Not a randomised controlled trial.

Chisholm 1984 Not a randomised controlled trial.

Christensen 1986 The number of multiple insulin injections per day is less than three

Christensen 1987 The number of multiple insulin injections per day is less than three

Christiansen 1987 The number of multiple insulin injections per day is less than three



(Continued)

Ciavarella 1985 The number of multiple insulin injections per day is less than three

Colette 1986 Not a randomised controlled trial.

Colette 1989 Not a randomised controlled trial.

Connis 1989 Not a randomised controlled trial.

Coustan 1986 The number of multiple insulin injections per day is less than three

Davies 1984 The number of multiple insulin injections per day is less than three

de Beaufort 1985 The number of multiple insulin injections per day is less than three

de Beaufort 1989 The number of multiple insulin injections per day is less than three

Dicker 1987 Following all reasonable attempts to contact the authors, we were unable to determine whether the trial

was randomised.

DiMeglio 2004 The number of multiple insulin injections per day is less than three

Edelmann 1987 The number of multiple insulin injections per day is less than three

Feldt-Rasmussen 1986 The number of multiple insulin injections per day is less than three

Feldt-Rasmussen 1986a The number of multiple insulin injections per day is less than three

Feldt-Rasmussen 1991 The number of multiple insulin injections per day is less than three

Flores 1984 Following all reasonable attempts to contact the authors, we were unable to obtain extractable data

Garg 2004 Not a randomised controlled trial.

Garmo 2004 Not a randomised controlled trial.

Giocolea 1988 The number of multiple insulin injections per day is less than three

Goicolea 1986 The number of multiple insulin injections per day is less than three




Greene 1983 Not a randomised controlled trial.



(Continued)

Grimm 1987 The number of multiple insulin injections per day is less than three

Guerci 1998 Not a randomised controlled trial.

Gulan 1988 Inappropriate interventions.

Haakens 1989 Not a randomised controlled trial.

Haakens 1990 Not a randomised controlled trial.

Hall 1989 The number of multiple insulin injections per day is less than three

Haug 1987 Not a randomised controlled trial.

Heber 1977 Inappropriate interventions.

Helve 1987 The number of multiple insulin injections per day is less than three

Helve 1987a The number of multiple insulin injections per day is less than three

Helve 1987b The number of multiple insulin injections per day is less than three

Hermansen 1987 The number of multiple insulin injections per day is less than three

Hermansen 1988 The number of multiple insulin injections per day is less than three

Hoogma 2004 Not a randomised controlled trial.

Hung 1984 The number of multiple insulin injections per day is less than three

Jakobsen 1988 The number of multiple insulin injections per day is less than three

Jensen 1986 Not a randomised controlled trial.

Kamoi 2004 Not a randomised controlled trial.

Kelly 1984 Not a randomised controlled trial.

Knight 1986 Not a randomised controlled trial.

Kobayashi 1987 Following all reasonable attempts to contact the authors, we were unable to confirm the number of


Koivisto 1983 Not a randomised controlled trial.

Kordonouri 2006 Not a randomised controlled trial.



(Continued)

Kritz 1983 Not a randomised controlled trial.

Kroc 1984 The number of multiple insulin injections per day is less than three

Kroc 1988 Not a randomised controlled trial.

Kuno 1996 Not a randomised controlled trial.

Laatikainen 1987 The number of multiple insulin injections per day is less than three

Lager 1983 Not a randomised controlled trial.

Lager 1986 The number of multiple insulin injections per day is less than three

Lapolla 2003 Not a randomised controlled trial.

Lauritzen 1985 The number of multiple insulin injections per day is less than three

Lawson 1984 The number of multiple insulin injections per day is less than three

Lawson 1985 The number of multiple insulin injections per day is less than three

Leblanc 1986 The number of multiple insulin injections per day is less than three

Lecavalier 1987 Inappropriate interventions.

Lepore 2004 Not a randomised controlled trial.

Levy 1984 Not a randomised controlled trial.

Levy-Marchal 1983 Not a randomised controlled trial.

Levy-Marchal 1988 Not a randomised controlled trial.

Litton 2002 Not a randomised controlled trial.

Ludvigsson 2003 Not a randomised controlled trial.

Lunetta 1996 Following all reasonable attempts to contact the authors, we were unable to obtain extractable data

Mancuso 1986 Following all reasonable attempts to contact the authors, we were unable to determine whether the trial

was randomised.

Marshall 1987 The number of multiple insulin injections per day is less than three

Marshall 1988 The number of multiple insulin injections per day is less than three



(Continued)

McDonald 1982 Not a randomised controlled trial.

Mecklenburg 1982 Not a randomised controlled trial.

Minkina-Pedras 2005 Following all reasonable attempts to contact the authors, we were unable to confirm whether the trial

was randomised or the number of multiple injections per day

Moberg 1994 Not a randomised controlled trial.

Moller 1986 The number of multiple insulin injections per day is less than three

Monnier 1987 The number of multiple insulin injections per day is less than three

Muller 1999 Not a randomised controlled trial.

Nathan 1996 Inappropriate interventions.

Ng 1986 The number of multiple insulin injections per day is less than three

Nijs 1991 Not a randomised controlled trial.

Olsen 1985 The number of multiple insulin injections per day is less than three

Olsen 1987 The number of multiple insulin injections per day is less than three

Pickup 1979 Not a randomised controlled trial.

Pickup 1979a Not a randomised controlled trial.

Pickup 1984 Not a randomised controlled trial.

Pickup 1985 The number of multiple insulin injections per day is less than three

Pickup 1989 The number of multiple insulin injections per day is less than three

Pietri 1980 Not a randomised controlled trial.

Pietri 1983 Not a randomised controlled trial.

Reeves 1982 Following all reasonable attempts to contact the authors, we were unable to confirm whether the trial

was randomised

Regal 1988 Not a randomised controlled trial.

Rizza 1980 Not a randomised controlled trial.

Rodger 1985 The number of multiple insulin injections per day is less than three



(Continued)

Rodger 1988 The number of multiple insulin injections per day is less than three

Saibene 1981 Not a randomised controlled trial.

Schaepelynck-Belicar 2003 Not a randomised controlled trial.

Schiffrin 1982 Following all reasonable attempts to contact the authors, we were unable to confirm the number of


Schiffrin 1982a Following all reasonable attempts to contact the authors, we were unable to confirm the number of






Schiffrin 1984a Following all reasonable attempts to contact the authors, we were unable to confirm the number of






Schmitz 1987 Inappropriate interventions.

Schottenfeld-Naor 1985 Not a randomised controlled trial.

Scuffham 2003 Not a randomised controlled trial.

Service 1985 The number of multiple insulin injections per day is less than three

Siebenhofer 2004 Inappropriate interventions.

Simonson 1985 Not a randomised controlled trial.

Skare 1986 Following all reasonable attempts to contact the authors, we were unable to obtain extractable data

Skyler 1982 Following all reasonable attempts to contact the authors, we were unable to confirm whether the trial

was randomised

Sleightholm 1986 The number of multiple insulin injections per day is less than three

Slijper 1990 The number of multiple insulin injections per day is less than three



(Continued)

Steno 1982 The number of multiple insulin injections per day is less than three

Sulli 2003 Not a randomised controlled trial.

Tamborlane 1979 Not a randomised controlled trial.

Tamborlane 1979a Not a randomised controlled trial.



Testa 1985 The number of multiple insulin injections per day is less than three

Thuesen 1986 The number of multiple insulin injections per day is less than three

Tilvis 1986 The number of multiple insulin injections per day is less than three

Toni 2004 Not a randomised controlled trial.

Turk 1996 The number of multiple insulin injections per day is less than three

van Ballegooie 1985 Not a randomised controlled trial.

Viberti 1981 The number of multiple insulin injections per day is less than three

Viberti 1984 The number of multiple insulin injections per day is less than three

Warmolts 1987 Not a randomised controlled trial.

Weinzimer 2004 Not a randomised controlled trial.

Willi 2003 Not a randomised controlled trial.

Wilson 2005 The number of multiple insulin injections per day is less than three

Wredling 1997 Not a randomised controlled trial.

Yki-Jarvinen 1984 Not a randomised controlled trial.

Ziegler 1988 Following all reasonable attempts to contact the authors, we were unable to obtain extractable data



D A T A A N D A N A L Y S E S

Comparison 1. CSII versus MI

Outcome or subgroup titleNo. of

studies

No. of

participants Statistical method Effect size

1 HbA1c 20 Mean Difference (Random, 95% CI) -0.25 [-0.40, -0.10]

1.1 Cross-over trials 10 Mean Difference (Random, 95% CI) -0.25 [-0.45, -0.05]

1.2 Parallel trials 10 Mean Difference (Random, 95% CI) -0.26 [-0.52, -0.00]

2 HbA1c subgroup analysis - age 20 Mean Difference (Random, 95% CI) -0.25 [-0.40, -0.10]

2.1 < 18 years 8 Mean Difference (Random, 95% CI) -0.22 [-0.41, -0.03]

2.2 > 18 years 12 Mean Difference (Random, 95% CI) -0.29 [-0.52, -0.06]

3 HbA1c subgroup analysis - study

duration

20 Mean Difference (Random, 95% CI) -0.25 [-0.40, -0.10]

3.1 Short term (< 1 month) 1 Mean Difference (Random, 95% CI) 0.2 [-1.94, 2.34]

3.2 Medium term (> 1 month

≤ 6 months)


3.3 Long term (> 6 months) 7 Mean Difference (Random, 95% CI) -0.17 [-0.44, 0.10]

4 HbA1c small study bias 6 Mean Difference (Random, 95% CI) -0.72 [-1.55, 0.10]

5 Daily mean blood glucose 13 Mean Difference (Random, 95% CI) Subtotals only

6 Daily mean blood glucose

subgroup analysis - age

13 Mean Difference (Random, 95% CI) -15.26 [-23.37, -7.

14]

6.1 < 18 years 3 Mean Difference (Random, 95% CI) -3.70 [-14.38, 6.98]

6.2 > 18 years 10 Mean Difference (Random, 95% CI) -17.60 [-26.71, -8.

48]

7 Daily mean blood glucose

subgroup analysis - study

duration


14]

7.1 Short term (< 1 month) 1 Mean Difference (Random, 95% CI) -3.0 [-80.47, 74.47]

7.2 Medium term (> 1 month

≤ 6 months)


13]

7.3 Long term (> 6 months) 2 Mean Difference (Random, 95% CI) -21.41 [-29.40, -13.

43]

8 Daily mean blood glucose small

study bias


31]

9 Fasting blood glucose 11 Mean Difference (Random, 95% CI) -14.02 [-24.42, -3.

61]

10 Fasting blood glucose subgroup

analysis - age


61]

10.1 < 18 3 Mean Difference (Random, 95% CI) -7.67 [-23.88, 8.53]

10.2 >18 8 Mean Difference (Random, 95% CI) -16.25 [-29.07, -3.

42]

11 Fasting blood glucose subgroup

analysis - study duration


61]

11.1 Short term (< 1 month) 1 Mean Difference (Random, 95% CI) -20.0 [-35.48, -4.52]

11.2 Medium term (> 1

month ≤ 6 months)

8 Mean Difference (Random, 95% CI) -8.48 [-19.35, 2.39]



11.3 Long term (> 6 months) 2 Mean Difference (Random, 95% CI) -37.04 [-73.46, -0.

62]

12 Fasting blood glucose small

study bias

3 Mean Difference (Random, 95% CI) -13.71 [-50.37, 22.

96]

13 Post prandial blood glucose 5 Mean Difference (Random, 95% CI) -4.44 [-10.91, 2.03]

14 Post prandial blood glucose




14.2 >18 3 Mean Difference (Random, 95% CI) -7.51 [-23.03, 8.00]


subgroup analysis - study

duration




month ≤ 6 months)


15.3 Long term (> 6 months) 0 Mean Difference (Random, 95% CI) 0.0 [0.0, 0.0]


small study bias

2 Mean Difference (Random, 95% CI) 0.67 [-14.52, 15.86]

17 Non-severe hypoglycaemic

events

Other data No numeric data

18 Severe hypoglycaemic events Other data No numeric data

19 Quality of life Other data No numeric data

20 Weight 8 Mean Difference (Random, 95% CI) -0.37 [-1.98, 1.25]

21 Weight subgroup analysis -

study duration


21.1 Short term (< 1 month) 0 Mean Difference (Random, 95% CI) 0.0 [0.0, 0.0]


month ≤ 6 months)



22 Weight small study bias 3 Mean Difference (Random, 95% CI) 1.74 [-4.24, 7.72]

23 Daily insulin requirement (U) 9 Mean Difference (Random, 95% CI) -6.98 [-10.77, -3.19]

24 Daily insulin requirement (U)




24.2 > 18 8 Mean Difference (Random, 95% CI) -6.59 [-10.66, -2.52]

25 Daily insulin requirement

(U) subgroup analysis - study

duration




month ≤ 6 months)



26 Daily insulin requirement (U)

small study bias


27 Daily insulin requirement (U/

kg)

13 Mean Difference (Random, 95% CI) Subtotals only


kg) subgroup analysis - age


28.1 < 18 6 Mean Difference (Random, 95% CI) -0.16 [-0.31, -0.01]

28.2 > 18 6 Mean Difference (Random, 95% CI) -0.08 [-0.15, -0.01]




kg) subgroup analysis - study

duration


29.1 Short term (< 1 month) 1 Mean Difference (Random, 95% CI) 0.1 [-0.12, 0.32]


month ≤ 6 months)




kg) small study bias


A D D I T I O N A L T A B L E S

Table 1. Overview of study populations

study ID intervention [n] screened [n]

randomised

[n] safety [n] ITT [n] finishing

study

comments

Bak 1987 &

Husted 1989

CSII (I1)

MI (I2)

For de-

tails about the

inter-

ventions, see ’

Characteris-

tics

of included

studies’.

Not reported Total: 20 Not reported Not reported Total: 16 Four par-

ticipants failed

to com-

plete the study

due to incom-

patibility with

CSII

Botta 1986 Not reported I1: 5

I2: 5

Total: 10

Not reported Not reported I1: 5

I2: 5

Total: 10

Drop outs not

reported

Bruttomesso

2008

I1: 24

I2: 18

Total:42

I1: 24

I2: 18

Total: 42


I2: 15

Total: 39

Chiasson

1984 & 1985

Not reported I1: 6

I2: 6

Total: 12


I2: 6

Total: 12

Drop outs not

reported

Cohen 2003 Not reported Total: 16 Not reported Not reported Total: 12 Three partici-

pants were sat-

isfied with

pump and so

didn’t want to

start MI, one

who started

with MI re-

fused to con-

nect to pump

DeVries 2002 To-

tal: 150 (asked

to participate)

I1: 39

I2: 40

Total: 79

Not reported I1: 39

I2: 40

I1: 34

I2: 21

It is impor-

tant to note



Table 1. Overview of study populations (Continued)

To-

tal: 89 (quali-

fication phase)

Total: 79 Total: 55 that at the

design phase,

the study

investigators

anticipated

the possibility

that they

would have

to analyse the

study as a

parallel RCT

(rather than

a cross-over)

because of the

demanding

nature of the

protocol and

an expected

high drop out

rate. However,

at the point

of cross over,

there had been

no drop outs

so the study

investigators

continued

with a cross-

over design

and stopped

inclusion at

79 partic-

ipants (48

participants

per arm were

calculated to

be required

to have an

80% chance

of detecting

a 0.75%

difference in

HbA1c with

an assumed

SD of 1.3%)




. Finally, the

study investi-

gators decided

that due to

the drop out

rate after cross

over, they

would analyse

the data as a

parallel study

using only pre

cross over data

Doyle 2004 Total: 32 I1: 16

I2: 16

Total: 32

Not reported I1: 16

I2: 16

Total: 32

I1: 16

I2: 15

Total: 31

Hanaire-

Broutin 2000

Not reported I1: 20

I2: 21

Total: 41


I2: 20

Total: 40

Hirsch 2005 Not reported I1: 50

I2: 50

Total: 100


I2: 46

Total: 91

Home 1982 Total: 12 Total: 11 Not reported Not reported Total: 10 The study in-

vesti-

gators have not

provided clear

infor-

mation about

what stage the

drop out oc-

curred

Hoogma 2005 Not reported Total: 272 Not reported Total: 229 Total: 223

Lepore 2003 Not reported I1: 16

I2: 16

Total: 32


I2: 16

Total: 32

Meschi 1982 Not reported I1: 8

I2: 8

Total: 16


I2: 8

Total: 16

Nathan 1982 Not reported Total: 5 Not reported Not reported Total: 5

Nosadini

1988

Not reported I1: 19 FBR

10 HOR

I2: 15

Not reported Not reported I1: 19 FBR

10 HOR




Total: 44 I2: 15

Total: 44

Nuboer 2008 Total: 39 I1: 19

I2: 19

Total: 38


I2: 19

Total: 38

Oslo Study

1985-1992

Total: 180 I1: 15

I2: 15

Total: 30


I2: 15

Total: 30

Pozzilli 2003 I1: 11

I2: 12

Total: 23

I1:11

I2:12

Total: 23


I2: 12

Total: 19

Four dropped

out of CSII be-

fore treatment

started.

Saurbrey 1988 Not reported I1:15

I2: 6

Total: 21

Not reported Not reported I1:13

I2: 6

Total: 19

Schmitz 1989 Not reported Total: 10 Not reported Not reported Total: 10

Skogsberg

2008

Not reported I1: 34

I2: 38

Total: 72


I2: 33

Total: 67

Tsui 2001 Total: 28 I1: 13

I2: 14

Total: 27


I2: 14

Total: 26

Weintrob

2004

Total: 24 I1: 11

I2: 12

Total: 23


I2: 12

Total: 22

Total 976 867

C O N T R I B U T I O N S O F A U T H O R S

MARIE MISSO (MM): co-ordinated each stage of the review and was responsible for: design of the review (in collaboration with

DOC); developing the protocol (in collaboration with DOC, JS and KE); developing the search strategy; undertaking the searches for

studies (independently of, but in addition to KE); screening the search results (independently of, but in addition to KE); organising

retrieval of papers; screening retrieved papers against inclusion/exclusion criteria (independently of, but in addition to KE); appraising

the quality of papers (independently of, but in addition to KE and MP); extracting data from papers (independently of, but in addition

to KE and MP); writing to study investigators for additional information; providing additional data about papers; summarising the risk

of bias of the studies (rated independently by MM, KE and MP); compiling the summary of comparisons, tables of included, excluded

awaiting and ongoing studies; entering data into RevMan 5; performing analysis of data; interpreting the findings; writing the review

(with contribution from DOC, JS, KE and MP); final approval of the version to be published.



DENISE O’CONNOR (DOC): was involved in the following stages of the review: the design of the review (in collaboration with

MM); developing the protocol (in collaboration with MM, JS & KE); contributing to the writing of the review (in collaboration with

MM, JS, KE and MP).

KRISTINE EGBERTS (KE): was involved in the following stages of the review: contributing to the drafting of the protocol (in

collaboration with MM, JS & DOC); screening the search results (independently of, but in addition to MM); screening retrieved

papers against inclusion/exclusion criteria (independently of, but in addition to MM); appraising the quality of papers (independently

of, but in addition to MM and MP); extracting data from papers (independently of, but in addition to MM and MP); contributing to

the writing of the review (in collaboration with MM, DOC, JS and MP).

MATTHEW PAGE (MP): was involved in the following stages of the review: appraising the quality of papers (independently of, but

in addition to MM and KE); extracting data from papers (independently of, but in addition to MM and KE); assisting MM with

compiling the table of included studies; contributing to the writing of the review (in collaboration with MM, DOC, JS and MP).

JONATHAN SHAW (JS): was involved with the following stages of the review: developing the protocol (in collaboration with MM,

DOC & KE) and contributing to the writing of the review (in collaboration with MM, DOC, KE and MP).

D E C L A R A T I O N S O F I N T E R E S T

JONATHAN SHAW: Received honoraria for lectures and for advisory boards; his department has received funding for research and

for organizing educational activities from NovoNordisk, Eli Lilly, Sanofi Aventis and Medtronic.

D I F F E R E N C E S B E T W E E N P R O T O C O L A N D R E V I E W

Matthew Page has been added as an author. The background has been updated. Study quality assessment methodology has been replaced

with risk of bias methodology as recommended for adapting the review to Review Manager 5 and the Cochrane Handbook for SystematicReviews of Interventions Version 5.0.0 (Higgins 2008).

Subgroup analysis regarding sex and type of insulin was not performed. We also conducted post-hoc analyses where we explored studies

that included less than 20 participants in subgroup analyses for each outcome.

I N D E X T E R M SMedical Subject Headings (MeSH)

Diabetes Mellitus, Type 1 [∗drug therapy]; Hypoglycemic Agents [∗administration & dosage]; Infusions, Subcutaneous; Insulin

[∗administration & dosage]; Randomized Controlled Trials as Topic

MeSH check words

Adolescent; Adult; Child; Humans



Documents

Cochrane Database of Systematic Reviews (Reviews) || Continuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus