Microvascular Disease and Risk of Cardiovascular Events Among Individuals With Type 2 Diabetes: a Population-Level Cohort Study

Brownrigg JRW, MRCS1, Hughes CO, MRCS2, Burleigh D, MSc3,

Karthikesalingam A, PhD1, Patterson BO, PhD1, Holt PJ, PhD1, Thompson

MM, MD1, de Lusignan S, MD3, Ray KK, MD, MPhil 4*, Hinchliffe RJ, MD1*

1 Division of Cardiovascular and Cell Sciences, St George’s University of London,

London, UK

2 Division of Surgery and Interventional Science, University College London,

London, UK

3 Department of Healthcare Management and Policy, University of Surrey,

Guildford, UK

4 Department of Primary Care and Public Health, Imperial College London,

London, UK

*KKR and RJH contributed equally to this study.

Word count:

4206

Corresponding author:

Jack RW Brownrigg

Division of Cardiovascular and Cell Sciences

St George’s University of London

Cranmer Terrace

London

SW17 OQT

jrwbrownrigg@gmail.com

1

ABSTRACT

BackgroundDiabetes confers a 2-fold excess risk of cardiovascular disease (CVD), yet predicting

individual risk remains challenging. The effect of total microvascular disease burden on CVD

risk among individuals with diabetes is unknown.

Methods A population-based cohort of patients with type 2 diabetes from the UK Clinical Practice

Research Datalink was studied (n=49 027). We used multivariable Cox models to estimate

hazard ratios for the primary outcome (cardiovascular death, non-fatal myocardial infarction or

non-fatal ischaemic stroke) associated with cumulative burden of retinopathy, nephropathy

and peripheral neuropathy among individuals with no history of cardiovascular disease at

baseline.

FindingsDuring a median follow-up of 5·5 years, 2822 (5·8%) individuals experienced a primary

outcome. Significant associations were observed for the primary outcome individually for

retinopathy, peripheral neuropathy, and nephropathy after adjustment for established risk

factors. The hazard ratios (with 95% confidence intervals) were 1·39 (1·09-1·76), 1·40 (1·19-

1·66), and 1·35 (1·15-1·58), respectively. For individuals with one, two or three microvascular

disease states versus none, the multivariable-adjusted hazard ratios for the primary outcome

were 1·32 (1·16-1·50), 1·62 (1·42-1·85) and 1.99 (1·70-2.34), respectively. Similar trends

were observed for cardiovascular death, all cause mortality and for hospitalisation for heart

failure. For the primary outcome, measures of risk discrimination showed significant

improvement when microvascular disease burden was added to models. In the overall cohort,

the net reclassification index for US and UK guideline risk strata were 3.6% (p<0.001) and

3.8% (p<0.001), respectively.

InterpretationThe cumulative burden of microvascular disease significantly impacts the risk of future

cardiovascular disease among individuals with type 2 diabetes. Given the prevalence of

diabetes globally, further work to understand the mechanisms behind this association and

strategies to mitigate this excess risk are warranted.

FundingCirculation Foundation

2

RESEARCH IN CONTEXT

Evidence before this studyWe searched Medline and EMBASE for studies published from Jan 1, 2000, to Nov 1, 2015,

with the terms “microvascular disease”, “cardiovascular disease”, “type 2 diabetes”, and

MeSH equivalents. The search period was selected to reflect contemporaneous data

immediately before, and following the introduction of routine screening for microvascular

disease in 2004 for the UK Quality and Outcomes Framework. We reviewed observational

studies and clinical trials examining the association between microvascular disease and

cardiovascular outcomes in ≥200 individuals. We identified 19 studies reporting positive

associations between retinopathy or nephropathy and cardiovascular disease, coronary

events, ischaemic stroke, and heart failure. More limited data also support a positive

association between neuropathy (cardiac autonomic neuropathy or peripheral neuropathy)

and cardiovascular disease outcomes. A single study in a Chinese cohort evaluated the

impact of disease in two microvascular beds, with reported hazard ratios of 1.69 (95% CI

0.99-2.89) for retinopathy alone and 2.25 (95% CI 1.40-3.63) with concomitant

microalbuminuria. Although some good quality studies were identified, all were limited in their

scope by either small sample size with individual studies reporting on fewer than 630 events

each, by the inclusion of individuals with pre-existing cardiovascular disease, selection bias,

or lack of adjustment for conventional risk factors and for the presence of disease in multiple

microvascular beds.

Added value of this studyBased on a detailed review of the literature, this study is the first to examine the effect of

disease in multiple microvascular beds in a large population cohort, with approximately 260

000 person years of exposure and 2822 first cardiovascular events. Our data reveal several

important findings. The presence of isolated retinopathy, peripheral neuropathy, or

nephropathy, independent of conventional risk factors, confer at least a similar risk of

cardiovascular events (cardiovascular death, non-fatal myocardial infarction or non-fatal

ischaemic stroke) as uncontrolled established risk factors including blood pressure (≥140/90

mmHg), HbA1c (≥7.0%) and low-density cholesterol (≥ 2.5 mmol/L). Individuals with disease

in multiple microvascular beds were, in a “dose dependent fashion”, at the greatest overall

risk, including for other endpoints such as hospitalisation for heart failure, cardiovascular

death and all-cause mortality.

Implications of the available evidenceThese data suggest that a continued broad assessment program for microvascular

complications of diabetes has prognostic value for routine clinical care globally as it further

risk stratifies people at higher cardiovascular risk than might be perceived, as well as

providing morbidity specific to individual microvascular disease states. The inclusion of

microvascular disease variables in cardiovascular risk algorithms resulted in a net correct

3

reclassification of 3.6% of our cohort into higher- or lower-risk strata based on incident events.

which is comparable if not slightly better than blood based biomarkers, but less than

improvement observed with coronary artery calcium scoring. If information on microvascular

disease were incorporated presently then 9·3% of individuals previously considered as

eligible for moderate intensity statins in US guidance (predicted risk <7.5%) would be

considered as candidates for high intensity statin therapy (observed risk 8·6%). Similarly,

microvascular disease would reclassify 9·0% of individuals in a higher risk group (predicted

risk ≥7.5%), currently considered eligible for high intensity statins, to a group who could be

offered moderate intensity therapy (observed risk 6·3%). In reference to UK NICE guidance,

of those currently considered ineligible for statin therapy (predicted risk <10%), 8·9% would

be reclassified into a higher risk group with an observed event rate of 11·6%. Of individuals

currently offered statin therapy (predicted risk ≥10%), 12·3% would be reclassified into a

lower risk group with an observed 10-year event rate of 8·1%. Based on the current known

prevalence of risk, in absolute terms this would represent a change in statin prescriptions for

10·6% of individuals with type 2 diabetes in the UK and 9·1% of those in the US, with

accurate reclassification in 59·5% and 65·7%, respectively.

As the assessment of microvascular disease should be part of routine clinical practice among

those with diabetes, our findings offer a simple, convenient and cheap method for improving

risk prediction as compared to more expensive blood based biomarkers or non-invasive

imaging modalities for better targeting preventive therapies. It might be possible to mitigate

against this excess risk, as we observed that among those with multiple microvascular

disease states, event rates were substantially lower when HbA1c, BP and LDL-C were better

controlled. High microvascular disease burden could be used as a criteria to enrich future

clinical outcome trials, identifying a very high risk cohort of patients who might derive greater

absolute benefit from more intensive risk factor control with conventional or novel therapies.

Finally, our observations should enthuse further research including a better understanding of

the impact of microvascular disease with different cardiovascular outcomes.

4

INTRODUCTION

Diabetes confers a 2-fold excess risk of cardiovascular disease1 and substantial premature

mortality from cardiovascular causes.2 However, individuals with diabetes are not

automatically considered as a coronary heart disease (CHD) risk equivalent and many

guidelines now recommend absolute risk assessment prior to considering lipid modification

therapy.3 Predicting individual risk remains challenging and external validation of available risk

algorithms in diabetic populations show moderate performance at best,4 highlighting the need

for cheap and routinely available measures that identify those with higher absolute risk over

and above established factors considered in contemporary risk algorithms.

Various microvascular disease states have been reported to be associated with risk of

vascular disease, including cardiac autonomic neuropathy (CAN),5,6 retinopathy,7,8

nephropathy,9,10 and peripheral neuropathy.11 Despite frequently co-existing, robust population

data evaluating the effect of cumulative microvascular disease burden on cardiovascular risk

in diabetes is absent. The aim of this study was to investigate whether microvascular disease

states alone or in unison are independently associated with cardiovascular disease (CVD),

and furthermore to compare any strength of association with conventional risk factors used in

current risk equations. To assess this relationship, we used routine healthcare data from a

large population-based cohort of individuals with type 2 diabetes free from CVD at baseline,

with approximately 259 686 person years of follow up and 2689 first cardiovascular events.

METHODSData sources and cohortThe Clinical Practice Research Datalink (CPRD) comprises data on individuals from over 600

practices in England, providing a representative UK primary care population.12 CPRD contains

information on anthropometric measurements, clinical diagnoses, laboratory tests and

prescription data, coded with the Read Clinical Coding system. Information on retinopathy,

nephropathy and peripheral neuropathy has been routinely collected in UK primary care

following the introduction of a pay for performance initiative, the Quality and Outcomes

Framework,13 in April 2004, which is linked to the National Institute for Health and Care

Excellence (NICE) guidance on standards of care for patients in the UK including appropriate

frequency of screening and risk factor control for those with chronic diseases.14

Individual patient data were linked across three datasets: the CPRD for demographic

characteristics and, Hospital Episode Statistics (HES) and the Office for National Statistics

(ONS) for the outcomes of interest. The HES are the English National Health Service

administrative dataset and contain information on every hospital admission including

diagnostic data, recorded as International Classification of Diseases, 10 th revision (ICD–10),

and procedural data based on the Office of Population, Census, and Surveys, version 4

5

(OPCS–4) codes. The ONS provide individual mortality records including cause of death

(ICD–10).

The study start date was 1 April 2008 to allow for 4 years of quality data on microvascular

disease status among participants. The data extract provided by CPRD included data on 49

027 individuals aged 18 years and over with type 2 diabetes and complete information on the

presence or absence of three microvascular diseases: retinopathy, nephropathy and

peripheral neuropathy. Individuals were screened for the presence of diabetes using

established criteria,15 and classified in accordance with methods described previously.16

Diabetes was defined by fasting plasma glucose ≥7.0 mmol per litre (126 mg per decilitre),

random plasma glucose ≥11.1 mmol per litre (200 mg per deciliter) or the use of glucose

lowering medications, based on recommendations from the American Diabetes

Association.15,17 In brief, classification of T2DM was performed according to the following

criteria: specific diagnostic code for T2DM (Read code C10F; ICD–10 code E11) with no

contradictory code; and patients with a diagnosis of diabetes at ≥35 years of age with no

insulin prescription within 1 year of diagnosis. Validation study of electronic health records

using this approach corrected miscoding of diabetes type in between 6–8% of cases.16 We

excluded individuals with a prior history of any cardiovascular disease.

Definition of baseline variablesAnthropometric measurements and numerical data, including systolic and diastolic blood

pressure, glycosylated haemoglobin, and cholesterol values were derived by taking the mean

of the three most recent values in the 12 months prior to the study start date. In cases where

three values were unavailable, the mean of two values was calculated. Values recorded more

than 12 months prior to the study start were not considered. Smoking status was stratified into

groups of never smoked, previously smoked and currently smoking at entry into the study.

Code lists used to define microvascular disease states were developed in accordance with

published guidance,18, 19 and are provided in the webappendix 1–3. Nephropathy was defined

as microalbuminuria (a moderate increase in albuminuria: 3-30 mg/mmol, 30-300 mg/g, 30-

300 mg/24h, or reagent strip urinalysis),20 and or eGFR <60ml/min per 1.73m2.

Outcome ascertainmentThe follow-up period extended to the study end: either December 2014, the date of patient

transfer from an included practice, or death. The primary outcome was the time to first major

cardiovascular event (an a priori composite of cardiovascular death, non-fatal myocardial

infarction or non-fatal ischaemic stroke). Ischaemic stroke events were defined by ICD-10

codes (I63) in accordance with published guidance.21 We combined ischaemic strokes with

unclassified strokes (I64) because previous studies have shown that 87% of unclassified

strokes were ischaemic.22 Information about cause-specific mortality and date of death was

obtained through the established record linkage with ONS. Fatal myocardial infarction (MI)

6

and ischaemic stroke were defined by primary cause of death (ICD–10 codes I21–I22 and I64

respectively). Patients were censored on the date of first primary outcome event. The pre-

specified secondary endpoints were cardiovascular death (fatal MI or fatal ischaemic stroke),

hospitalisation for heart failure and all-cause mortality. Study approval was granted by the

Independent Scientific Advisory Committee of the Medicines and Healthcare products

Regulatory Agency.

Statistical analysesWe defined clinical characteristics and outcome data both overall and according to risk

groups (absence of microvascular disease at baseline, or stratified by the number of prevalent

microvascular disease states). All reported p values are two-sided. Adjusted hazard ratios

and corresponding 95% confidence intervals were estimated with Cox proportional-hazards

models. Adjustment in all models was performed for age, gender, on treatment systolic and

diastolic blood pressure, high- and low-density cholesterol, HbA1c, body-mass index, duration

of diabetes, smoking history (defined by either ex-smoker or current smoker status),

antiplatelet therapy, lipid-lowering therapy, use of angiotensin converting enzyme inhibitor/

angiotensin receptor blocker, any treatment for blood pressure, ethnicity and index of multiple

deprivation. The group free of microvascular disease at baseline were used as the reference

category. Missing data for ethnicity and index of multiple deprivation were imputed using

multiple imputation by chained equations in the “mice” algorithm in R, and these imputed data

were used in the primary analysis.

We assessed differences in predictive accuracy of a model including established risk factors

from the Framingham risk function for a first primary outcome event (model A),23 and the

same model incorporating microvascular disease variables (model B). Model discrimination

was assessed with the use of the C-statistic.24 To evaluate the overall improvement in risk

stratification with the addition of microvascular disease to fully adjusted models, we calculated

net reclassification improvement (NRI) statistic and the integrated-discrimination-improvement

(IDI) statistic.25 Discrimination indices are reported across risk strata defined in both the

American College of Cardiology (ACC)/ American Heart Association (AHA) treatment

guidelines (lower risk <7.5%, higher risk ≥7·5%3 for 10-year CVD risk and the UK National

Institute for Health and Care Excellence (NICE) guidelines which consider higher risk

individuals as those with ≥10% 10-year risk of CVD.26 Statistical analyses were performed

with the use of R software version 15·2.

Role of the funding source

The sponsors had no role in the original protocol design, data collection, data analysis, data

interpretation, writing of the report, or the decision to submit the report for publication. The

corresponding author had full access to all the data in the study and had final responsibility for

the decision to submit for publication.

7

RESULTS

Patient CharacteristicsWe identified a cohort of 49 027 individuals with type 2 diabetes, of whom just less than half

were women. Baseline characteristics of the study population, both overall and according to

microvascular disease burden, are shown in Table 1. Individuals with microvascular disease

were more likely to have an adverse cardiovascular risk profile with significantly greater levels

of HbA1c, systolic blood pressure and smoking history. Age and duration of diabetes

significantly increased in a linear fashion with increasing burden of microvascular disease.

Exceptions included a trend for more favourable low-density lipoprotein cholesterol with

increasing burden of microvascular disease, likely related to the greater use of lipid-lowering

therapy. A comparison of the demographic characteristics of individuals with a single

manifestation of microvascular disease versus those without is provided in the webappendix

4.

Primary and Secondary Outcome MeasuresEvent rates for the primary outcome per 1000 person years in those without microvascular

disease were 5·00 compared with 8.22, 10.12 and 10.04 among individuals with isolated

retinopathy, nephropathy and peripheral neuropathy, respectively. Each microvascular

disease state studied was significantly associated with the primary outcome, and remained so

following adjustment for established risk factors and after excluding individuals with multiple

manifestations of microvascular disease (Table 5 webappendix). Single manifestations of

microvascular disease appear to confer at least as much risk as the failure to control

conventional risk factor goals in adjusted analyses (webappendix 6–8). Microalbuminuria in

the absence of low eGFR (<60ml/min per 1.73m2) was independently associated with the

primary outcome (webappendix 9). Further adjustment for the number of antihypertensive

treatments resulted in no qualitative difference in the hazard ratios for the primary outcome.

Figure 1 shows the linear relationship between increasing burden of microvascular disease

and the primary outcome (Panel A), cardiovascular mortality (Panel B), and hospitalisation for

heart failure (Panel C), P for linear trend <0·001 for all. Analyses for all-cause mortality were

qualitatively similar (webappendix 10); we found a 4·7-fold excess risk of death from any

cause among individuals with three manifestations of microvascular disease compared with

none (webappendix 11). Unadjusted event rates for the primary outcome among individuals

free of microvascular disease at baseline and among those with one, two, or three

microvascular disease states were 5·0, 9·8, 15·7 and 22·1 per 1000 person years,

respectively. After adjustment for potential confounders, the hazard ratios for the primary

outcome, cardiovascular death and hospitalisation for heart failure remained significant but

were attenuated across all three groups, suggesting that conventional risk factors account, in

part, for the excess risk observed with cumulative burden of microvascular disease (Table 2).

8

In fully adjusted models, a single manifestation of microvascular disease appears to as

strongly associated with the primary outcome as blood pressure, low-density cholesterol,

glycosylated haemoglobin and smoking history in the present analysis (Figure 2), although

this may in part be due to the greater variability around the measurement of conventional risk

factors when compared to a diagnosis of microvascular disease. A similar relationship was

observed for cardiovascular death, hospitalisation for heart failure (Figure 2), and death from

any cause (webappendix 12). This association remained when established risk factors were

dichotomised to reflect recommended risk factor goals (webappendix 13). When assessed

across strata of risk factor control for HbA1c (<7·0%, and ≥7·0%), low-density cholesterol

(<2·5, and ≥2·5 mmol per litre) and blood pressure (<140/90, and ≥140/90 mm Hg), a

consistent linear trend of greater risk of the primary outcome with cumulative burden of

microvascular disease and uncontrolled risk factors was observed (Figure 3).

In comparison to a Cox model based on established risk factors included in the Framingham

model (model A), the addition of information on microvascular disease (model B) yielded

improvements in the C-statistic from 0·679 to 0·689 respectively and an improvement in the

integrated discrimination index (0·003, 95% CI, 0·003–0·004, P<0·001). Across the two

ACC/AHA categories of cardiovascular risk (<7.5% and ≥7.5% 10–year risk of CVD),

microvascular disease reclassified 9·1% of the cohort into higher or lower risk groups as

defined by US guidelines, and did so with 65·7% correct reclassification (net reclassification

index 0.036, 95% CI 0.017-0.055, p<0.001). Of those individuals with a predicted <7.5% 10

year risk of CVD (32.9% of the overall cohort), 9·3% were reclassified into a higher risk group

(≥7·5% 10–year risk of CVD), with an observed 10–year event rate of 8·6%. Similarly,

microvascular disease reclassified 9·0% of individuals considered at higher risk (67.1% of

overall cohort) to a lower risk group, with an observed 10–year CVD risk of 6·3%. According

to risk categories quoted in UK guidance, microvascular disease reclassified 10·6% of

individuals to a higher or lower risk group, of whom 59·5% were reclassified accurately (net

reclassification index 3.8%, 95% CI 0.013-0.060, p<0.001). Among those considered at lower

risk (<10% 10–year risk CVD, 48.6% of cohort), 8·9% were reclassified to a higher risk group

with a 10 year observed event rate of 11·6%. Of individuals considered at higher risk (≥10%

10–year risk CVD, 51.4% of overall cohort) by conventional models, 12·3% were reclassified

into a lower risk group with an observed 10-year event rate of 8·1%.

In separate analyses the sequential addition of data on duration of diabetes and, in turn,

microvascular disease to a model based on the Framingham risk function yielded C-statistics

of 0.679 and 0.682, with respective NRIs of 0.011 (95% CI 0·001-0·022, p=0·050) and 0.024

(95% CI 0.007–0.042, p=0.007). The addition of information on duration of diabetes

corresponded with a small improvement in IDI (0.0005 (95% CI 0.0002–0.0009, p=0.006), with

further improvement after the addition of microvascular disease data (IDI 0.003, 95% 0.018–

0.041, p<0.001).

9

DISCUSSION

In a population cohort of individuals with type 2 diabetes, our findings show that burden of

microvascular disease is a determinant of future cardiovascular risk. The risk of a first

cardiovascular event increased linearly with the number of manifestations of microvascular

disease present. Furthermore, the presence of isolated retinopathy, peripheral neuropathy, or

nephropathy confer at least a similar risk of cardiovascular events as factors contained in

contemporary risk equations such as blood pressure, low-density lipoprotein cholesterol and

haemoglobin A1c. Despite significant differences in baseline values of glycosylated

haemoglobin, low-density cholesterol and blood pressure among individuals with increasing

burden of microvascular disease, these factors did not abolish the associations between

microvascular disease and cardiovascular outcomes. We noted no deviations from linearity in

subgroups stratified by varying degrees of risk factor control.

Consistent with our findings, previous reports have documented an increase in cardiovascular

risk with individual microvascular disease states.5–11 However, the true impact of

microvascular disease may have been overestimated because risk ratios provided in the

literature are subject to confounding by a lack of adjustment for the presence of disease in

multiple microvascular beds. An important advance of this study was our ability to examine

the effect of both cumulative burden, and isolated microvascular disease states on first

presentation of cardiovascular disease. This approach was enabled by the routine collection

of microvascular disease data in the UK, and the availability of electronic health record

linkage.

Individual participant data from 97 prospective studies suggests the presence of diabetes is

associated with a 1·8 times increased risk of death from any cause.2 However, a study of

individuals with type 2 diabetes in the Swedish National Diabetes Register suggests that

excess mortality risk has declined in recent years, driven in part, by substantial reductions in

CVD mortality.27 The reported hazard ratios for all-cause and cardiovascular mortality, based

on follow-up to 2011 in that study were 1·15 (95% CI 1·14-1·16) and 1·14 (95% CI 1·13-1·15),

respectively. Although event rates in type 2 diabetes are falling and do not imply a CHD risk

equivalent as previously described,28,29 lifetime risk of cardiovascular disease remains high

emphasizing the need to identify early markers of risk.30 At diagnosis of type 2 diabetes, the

UK Prospective Diabetes Study identified retinopathy alone in 36% of participants.31 Currently,

data recorded on the presence or absence of retinopathy, nephropathy and peripheral

neuropathy are used in the UK to inform risk of developing blindness, renal failure, and

amputation, respectively. Our findings suggest these data may offer a simple tool to identify

very high-risk individuals with type 2 diabetes who are currently perceived to be at lower

absolute risk using contemporary risk models.

10

Cardiovascular risk estimation in diabetes has important implications for primary prevention

strategies. The 2013 ACC/ AHA guidelines on the control of blood cholesterol advocate

moderate-intensity statin therapy in persons with diabetes who are 40–75 years of age; while

high-intensity therapy is restricted to individuals with a ≥7·5% estimated 10–year risk of

cardiovascular disease.3 Our findings suggest that individuals with more than one

manifestation of microvascular disease would be eligible for high–intensity statin treatment

based on the recorded event rates. The 10–year risk of the primary outcome in the present

study was 9·8% in participants with a single manifestation of microvascular disease, 15·7%

with two, and 22·1% with three microvascular beds affected. Overall, microvascular disease

reclassified 9·1% of the cohort into higher or lower risk groups as defined by US guidelines,

and did so with 65·7% correct reclassification. When extrapolated to the 27.9 million

individuals with type 2 diabetes in the US,32 this would represent a change in the intensity of

statin therapy for over 2.5 million people. If information on microvascular disease were

incorporated presently then 9·3% of individuals previously considered as eligible for moderate

intensity statins (predicted risk <7.5%) could now be considered as candidates for high

intensity statin therapy (observed risk 8·6%). Similarly, microvascular disease would

reclassify 9·0% of individuals currently considered eligible for high intensity statins (predicted

risk ≥7.5%), to a group who could be offered moderate intensity therapy (observed risk

6·3%). Improvements in reclassification as suggested above not only offer potentially the

correct intensity of therapy, but also offer the best net benefit avoiding potential exposure of

lower CVD risk patients to potentially unnecessary dose dependent side effects on higher

intensity statins, which may impact on compliance and patient engagement.

In the UK the potential relevance of the present findings may be more profound. NICE

guidance recommends initiating atorvastatin 20 mg or a statin of equivalent potency for

primary prevention in people with type 2 diabetes and ≥10% 10–year risk of developing CVD

with no recommendations for statins below this predicted risk threshold.26 The use of

microvascular disease would reclassify 10·6% of individuals to a higher or lower risk group, of

whom 59·5% would be reclassified accurately. This figure corresponds to 370 000 of the 3.2

million people living with type 2 diabetes in the UK presently.33 Among those currently

considered ineligible for statin therapy (predicted risk <10%), 8·9% would be reclassified into

a higher risk group with an observed event rate of 11·6%, reflecting potentially 135 000 new

statin prescriptions in the UK. Of individuals currently offered statin therapy per NICE

guidance (predicted risk ≥10%), 12·3% would be reclassified into a lower risk group (200 000

when extrapolated to UK population) with an observed 10-year event rate of 8·1%. The

inclusion of microvascular disease would potentially offer cost benefits from the opportunity to

prevent more events as higher risk patients would be targeted, despite resulting in a net

reduction of statin prescriptions in the UK and therefore cost reductions or neutrality.

11

Real world data suggests that acceptance of preventive therapies and implementation, for

instance of statin guidelines, has been problematic in younger patients.34 A potential practical

application of these data might be to highlight individuals who, despite their age, are at higher

risk due to multiple manifestations of microvascular disease, and may help to overcome

patient and physician reluctance to initiate statins. Furthermore these data might serve as the

basis for identifying patient groups with high absolute risk who might, under current EMEA

and FDA licences, benefit most from further lipid lowering with novel (more expensive)

therapies, or could be used to enrich patients with higher event rates for future trials, thus

reducing sample size, duration and cost of conducting large outcome studies.

Among individuals with three manifestations of microvascular disease, our data indicate that

good control of risk factors (HbA1c <7·0%, low-density cholesterol <2·5 mmol per litre, and

blood pressure <140/90) is associated with a 43% lower risk of future cardiovascular events

compared to when these factors are not at goal (17·1 versus 29·8 events per 1000 person

years). However, these data are observational in nature and, although they support a positive

association between poor risk factor control and cardiovascular events among individuals with

prevalent microvascular disease, they cannot prove the benefit of treatments to modify

HbA1c, low-density cholesterol or blood pressure to target in this population. Insights from the

Steno-2 study support this observation that aggressive management of multiple risk factors

might mitigate some of the excess risk associated with microvascular disease.35 It randomised

patients with type 2 diabetes and persistent microalbuminuria to receive either intensive or

conventional therapy for a number of modifiable risk factors including glucose control, blood

pressure, total cholesterol and triglyceride levels. Intensive therapy was associated with a

lower risk of both fatal and non-fatal cardiovascular events at a median follow-up of 13 years.

An important caveat however is that baseline cardiovascular risk factors were significantly

more adverse in Steno–2 compared to the present cohort.

We also assessed the associations of microvascular disease burden with hospitalisation for

heart failure and report event rates around half those observed in the recent Reduction of

Atherothrombosis for Continued Health (REACH) registry.36 Among participants with

established atherothrombosis and a prior ischaemic event enrolled in REACH, 6·5% of

patients were hospitalised for heart failure corresponding to a rate of 16 per 1000 person

years. This compared to an overall rate of 6 per 1000 person years in this study of individuals

free of cardiovascular disease at baseline. Those with disease in three microvascular beds

were at significantly greater risk, with event rates of 15 per 1000 person years, similar to

those with a history of MI or stroke in REACH. In comparison with diabetic patients free from

microvascular disease, the adjusted hazards for heart failure with the presence of one, two, or

three microvascular disease states were 1·63, 2·24, and 2·90, respectively. The mechanisms

behind this association are unclear but plausible contributors include CAN, which frequently

co-exists with other microvascular disease states,37 and may be the diabetes-specific process

12

that explains part of the excess risk of heart failure not accounted for by increased burden of

atherothrombosis.38,39

While the present data derive from a validated and nationally representative sample of

England, results should not be extrapolated to dissimilar populations. Important limitations of

the study include our reliance on comprehensive code lists for any given baseline or outcome

variable. This is a limitation common to all studies using routinely recorded data and was

mitigated through the use of a validated approach for defining baseline and outcome

parameters.16,19 Individuals were screened for diabetes using established criteria that may not

reflect population samples identified through other methods and may imply lower overall

cardiovascular risk compared to cohorts with type 2 diabetes diagnosed through case finding

or clinical symptoms. Limitations exist in the amount of clinical detail presently recorded in

national administrative datasets such as CPRD, which offer the benefit of large cohorts at the

expense of granularity that is common to bespoke epidemiological studies. In this regard,

quantitative data on albuminuria or albumin-to-creatinine ratio was not consistently available

and would have been preferable, given these measures have been previously shown by the

CKD Prognosis Consortium to improve the discrimination of cardiovascular outcomes beyond

traditional risk factors among individuals with diabetes.10 Furthermore, greater detail on the

classification of retinopathy into non-proliferative and proliferative types was not available in

sufficient numbers to permit meaningful analyses across these categories. Analyses were

restricted to individuals in whom complete information was available on prevalent

microvascular disease and may be subject to selection bias. Examination of the association

between microvascular disease and CVD among individuals with data missing on all three

diseases showed no qualitative difference with the complete cohort (webappendix 14).

Ethnicity data were missing in just under three quarters of patients and social deprivation was

missing in a third; these confounding variables were imputed and included in the primary

analysis. Results may have been affected by unmeasured variables such as diet, which was

not considered in our analyses because these data are unreliably recorded. Finally, the data

presented here are observational in nature and although attempts have been made to reduce

confounding by statistical adjustment we cannot exclude the possibility of residual

confounding as part of the explanation for our findings.

In this linked primary and secondary care study of diabetic adults, microvascular disease was

found to confer a risk equivalent to conventional factors including smoking, hypertension and

dyslipidaemia. Cardiovascular risk and mortality increased with the total number of

microvascular beds affected, suggesting a continued broad assessment program for

retinopathy, nephropathy and peripheral neuropathy can provide reliable information on

cardiovascular risk, in addition to morbidity linked to individual microvascular disease states.

Such prognostic data has implications for cardiovascular risk stratification and prevention

strategies.

13

AcknowledgementsThe study was supported by a grant from the Circulation Foundation.

ContributorsJB, RH, and KR designed the study protocol. JB, CH, and DB did the statistical analyses.

SdeL, BP, and AK provided support in the statistical analyses and interpretation of results.

PH, MT provided critical appraisal of initial drafts. All authors took part in the writing of this

report.

Declaration of InterestsKKR reports to having received honoraria for serving on the steering committee, clinical

endpoint adjudication committee, advisory boards or lectures from Agerion, Abbvie, Pfizer,

AZ, Sanofi, Regeneron, Amgen, MSD, Roche, Kowa, Algorithm, Novartis, Novo Nordisk, Lily,

Resverlogix, ISIS Pharma, Cipla, Takeda, Boehringer Ingelheim. RJH is supported by a

career salary award from The Higher Education Funding Council for England. MMT has

received research grants from Medtronic, Cook Endovascular, and Endologix. PH is a

Clinician Scientist financially supported by the National Institute for Health Research (NIHR-

CS-011–008). All other authors report no declarations of interests.

14

REFERENCES

1. The Emerging Risk Factors Collaboration. Diabetes Mellitus, fasting blood glucose

concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective

studies. Lancet 2010;375:2215–2222.

2. Seshasai SRK, Kaptoge S, Thompson A, et al. Diabetes mellitus, fasting glucose, and

risk of cause-specific death. N Engl J Med 2011;364:829–841.

3. Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA Guideline on the

Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults: A

Report of the American College of Cardiology/American Heart Association Task Force on

Practice Guidelines. Circulation 2014;129:S46–48.

4. van Dieren S, Beulens JWJ, Kenge AP, et al. Prediction models for the risk of

cardiovascular disease in patients with type 2 diabetes: a systematic review. Heart

2012;98:360–369.

5. Valensi P, Sachs RN, Harfouche B, et al. Predictive value of cardiac autonomic

neuropathy in diabetic patients with or without silent myocardial ischemia. Diab Care

2001;24:339–43.

6. Astrup AS, Tarnow L, Rossing P, et al. Cardiac autonomic neuropathy predicts

cardiovascular morbidity and mortality in type I diabetic patients with diabetic nephropathy.

Diab Care 2006;29:334–9.

7. van Hecke MV, Dekker JM, Stehouwer CDA, et al. Diabetic retinopathy is associated

with mortality and cardiovascular disease Incidence. The EURODIAB Prospective Diabetes

Study. Diab Care 2005;28:1383–1389.

8. Kramer CK, Gross JL, Rodrigues TC, et al. Diabetic retinopathy predicts all-cause

mortality and cardiovascular events in both type 1 and 2 diabetes. Diab Care 2011;34:1238–

1244.

9. Gerstein HC, Mann JFE, Yi Q, et al. Albuminuria and risk of cardiovascular events,

death, and heart failure in diabetic and nondiabetic individuals. JAMA 2001;286:421–426.

10. Matsushita K, Coresh J, Sang Y, et al. Estimated glomerular filtration rate and

almbuminuria for prediction of cardiovascular outcomes: a collaborative meta-analysis of

individual participant data. Lancet Diabetes Endocrinol 2015;3:514–25.

11. Brownrigg JR, de Lusignan S, McGovern A, et al. Peripheral neuropathy and the risk

of cardiovascular events in type 2 diabetes mellitus. Heart 2014;100:1837–1843.

12. Walley T, Mantgani A. The UK General Practice Research Database. Lancet

1997;350:1097–1099.

13. Roland M. Linking physician pay to quality of care: a major experiment in the UK. N

Engl J Med 2004;351:1448–54.

14. National Institute for Health and Care Excellence (2009). Type 2 diabetes: the

management of type 2 diabetes. NICE guideline (CG87)

15. American Diabetes Association. Diagnosis and classification of diabetes mellitus.

Diab Care 2014;37:S81–90.

15

16. de Lusignan S, Khunti K, Belsey J, et al. A method of identifying and correcting

miscoding, misclassification and misdiagnosis in diabetes: a pilot and validation study of

routinely collected data. Diabet Med 2010;27:203–207.

17. Report of the Expert Committee on the Diagnosis and Classification of Diabetes

Mellitus. Diab Care 1997;20:1183–97.

18. Dave S, Petersen I. Creating medical and drug code lists to identify cases in primary

care databases. Pharmacoepidemiol Drug Saf 2009;18:704–7.

19. de Lusignan S, Liaw ST, Michalakidis G, et al. Defining datasets and creating data

dictionaries for quality improvement and research in chronic disease using routinely collected

data: an ontology-driven approach. Inform Prim Care 2011;19:127–34.

20. Levin A, Stevens PE. Summary of KDIGO 2012 CKD guideline: behind the scenes,

need for guidance, and a framework for moving forward. Kidney International 2014;85:49-61.

21. Sacco RL, Kasner SE, Broderick SE, et al. An updated definition of stroke for the 21st

century a statement for healthcare professionals from the American Heart Association/

American Stroke Association. Stroke 2013;44:2064–89.

22. Go AS, Mozaffarian D Roger VL, et al. Heart disease and stroke statistics-2013

update: a report from the American Heart Association. Circulation 2013;127:e1-240.

23. D’Agostino RB, Vasan RS, Pencina MJ, et al. General Cardiovascular risk profile for

use in primary care. The Framingham Heart Study. Circulation 2008;117:743–753

24. Harrell FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing

models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med

1996;15:361–387.

25. Pencina MJ, D'Agostino RB, D'Agostino RB Jr, Vasan RS. Evaluating the added

predictive ability of a new marker: from area under the ROC curve to reclassification and

beyond. Stat Med 2008;27:157–172

26. National Institute for Health and Care Excellence (2014). Cardiovascular disease: risk

assessment and reduction, including lipid modification. NICE guideline (CG181)

27. Tancredi M, Rosengren A, Svennson AM, et al. Excess mortality among persons with

type 2 diabetes. N Engl J Med 2015;373:1720–1732.

28. Gregg EW, Li Y, Wang J, et al. Changes in diabetes-related complications in the

United States, 1990-2010. N Engl J Med 2014;370:1514–1523.

29. Haffner SM, Lehto S, Ronnemaa T, et al. Mortality from coronary heart disease in

subjects with type 2 diabetes and in non-diabetic subjects with and without prior myocardial

infarction. N Engl J Med 1998;339(4):229–234.

30. Lloyd-Jones DM, Leip EP, Larson MG, et al. Prediction of lifetime risk for

cardiovascular disease by risk factor burden at 50 years of age. Circulation 2006;113:791–

798.

31. UK Prospective Diabetes Study Group. Intensive blood-glucose control with

sulphonylureas or insulin compared with conventional treatment and risk of complications in

patients with type 2 diabetes (UKPDS 33). Lancet 1998;352:837–853.

16

32. American Diabetes Association. National Diabetes Statistics Report 2015: estimates

of diabetes and its burden in the epidemiologic estimation methods. Natl Diabetes Stat Rep

2014:2009–12.

33. Quality and Outcomes Framework, Health and Social Care Information Centre,

Information Services Division Scotland. Statistics for Wales, Department of Health, Social

Services and Public Safety, 2014-2015.

34. Gaede P, Lund-Andersen H, Parving HH, Pederson O. Effect of multifactorial

intervention on mortality in type 2 diabetics. N Engl J Med 2008;358:580–91.

35. Berthold HK, Gouni-Berthold I, Bohm M, Kron W, Bestehorn KP. Patterns and

predictors of statin prescription in patients with type 2 diabetes. Cardiovasc Diabetol

2009;8:25.

36. Cavender MA, Steg PG, Smith SC, et al. Impact of diabetes mellitus on

hospitalization for heart failure, cardiovascular events, and death outcomes at 4 years from

the Reduction of Atherothrombosis for Continued Health (REACH) Registry. Circulation

2015;132:923–931.

37. Vinik AI, Ziegler D. Diabetic cardiovascular autonomic neuropathy. Circulation

2007;115:387–397.

38. Udell JA, Cavender MA, Bhatt DL, et al. Glucose-lowering drugs or strategies and

cardiovascular outcomes in patients with or at risk of type 2 diabetes: a meta-analysis of

randomised controlled trials. Lancet Diabetes Endocrinol 2015;3:356–366.

39. Doehner W, Frenneaux M, Ander SD. Metabolic impairment in heart failure: the

myocardial and systemic perspective. J Am Coll Cardiol 2014;64:1388–1400.

17

Table 1. Baseline Characteristics

Number microvascular disease states† All n=49 027

p value

0n=12 385

1n=18 631

2n=13 886

3n=4125

Age, years 62·6 (11·3) 69·0 (11·4) 73·7 (10.5) 74·5 (10·5) 69·2 (11·8) <0·001

Women 5405 (43·6) 8754 (47·0) 6731 (48·5) 1943 (47·1) 22833 (46·6) <0·001

White ethnicity 3104 (89·7) 4730 (90·8) 3418 (92·5) 981 (91·2) 12233 (91·0) 0·12

BMI, kg/m2 31.0 (6·2) 30·6 (6·3) 30·3 (6·2) 30·4 (6·2) 30·6 (6·2) <0·001

HbA1c, % 7·23 (1·25) 7·23 (1·28) 7·32 (1·34) 7·64 (1·45) 7·29 (1·31) <0·001Duration diabetes, years 5·7 (4·5) 7·2 (5·5) 9·5 (6·7) 13·8 (8·0) 8·0 (6·3) <0·001

Systolic blood pressure, mmHg 135·5 (12·6) 136·9 (13·3) 138·0 (14·4) 139·2 (14·8) 137·1 (13·6) <0·001

Diastolic blood pressure, mmHg 78·1 (7·6) 76·3 (8·0) 74·5 (8·3) 74·5 (8·3) 76·0 (8·2) <0·001

Total cholesterol, mmol/L 4·36 (0·89) 4·29 (0·89) 4·23 (0·93) 4·14 (0·91) 4·28 (0·90) <0·001

HDL cholesterol, mmol/L 1·27 (0·37) 1·27 (0·37) 1·26 (0·38) 1·24 (0·40) 1·27 (0·38) <0·001

LDL cholesterol, mmol/L 2·38 (0·85) 2·30 (0·84) 2·25 (0·82) 2·20 (0·84) 2·23 (0·84) <0·001

eGFR, mL/min/1.73m2 81·4 (16.9) 70·0 (22·5) 69·5 (21·4) 52·7 (19·1) 68·5 (22·6) <0·001

Smoking history 8999 (72·9) 13967 (75·2) 10694 (77·2) 3265 (79·2) 36925 (75·5) <0·001Deprivation index ≤ 5th decile 4550 (54·0) 6658 (52·4) 5141 (54·0) 1527 (52·9) 17876 (53·3) 0·05

Statin use 8631 (69·7) 13558 (72·8) 10333 (74·4) 3078 (74·6) 35600 (72·6) <0·001

ACEi/ARB 7479 (60·4) 13903 (74·6) 11717 (84·4) 3766 (91·3) 36865 (75·2) <0·001Blood pressure treatment 9216 (74.4) 16146 (86.7) 13001 (93.6) 4006 (97.1) 42369 (86.4) <0.001

Antiplatelet 6790 (54·8) 11963 (64·2) 9904 (71·3) 3136 (76·0) 31793 (64·8) <0·001

† Microvascular diseases considered include retinopathy, microalbuminuria and peripheral neuropathy. Data are mean (SD) or number (%). BMI indicates body mass index; HbA1c, glycosylated haemoglobin; HDL, high-density lipoprotein; LDL, low density lipoprotein; eGFR, estimated glomerular filtration rate; ACEi/ARB, angiotensin converting enzyme inhibitor/ angiotensin receptor blocker. P values from Chi square test or ANOVA are provided for the overall trend with increasing number of microvascular disease states. Missing values: The following variables had missing values: Ethnicity (n=35590, 72.6%), BMI (n=94, 0.2%), HbA1c (n=94, 0·2%), Systolic BP (n=6, 0·01%), Diastolic BP (n=6, 0·01%), Total cholesterol (n=25, 0·05%), HDL cholesterol (n=3778, 7·7%), LDL cholesterol (n=8347, 17·0%), eGFR (n=523, 1.1%), Smoking status (n=108, 0·2%), Deprivation index (n=.15495, 31·6%)

18

Table 2. Adjusted Hazard Ratios of Clinical Outcomes by Burden of Microvascular Disease*

Number microvascular disease states0

n=12 3851

n=18 6312

n=13 8863

n=4125Primary outcome

N 351 (2·8%) 975 (5·2%) 1072 (7·7%) 424 (10·3%)

Event rate per 1000 person years 5·00 9·82 15·69 22·10

Unadjusted hazard ratio 1·00 1·97 (1·74–2·22) 3·15 (2·80–3·56) 4·45 (3·87–5·13)

Adjusted hazard ratio (95% CI)* 1·00 1·32 (1·16–1·50) 1·62 (1·42–1·85) 1.99 (1·70–2·34)

Hospitalisation for heart failure

N 114 (0·9%) 449 (2·4%) 611 (4·4%) 270 (6·5%)

Event rate per 1000 person years 2·76 4·85 9·53 14·88

Unadjusted hazard ratio 1·00 2·77 (2·25–3·40) 5·45 (4·46–6.66) 8·53 (6·86–10·62)

Adjusted hazard ratio (95% CI)* 1·00 1·63 (1·31–2·03) 2.24 (1·80–2·80) 2.90 (2·27–3·71)

Cardiovascular mortality

n 92 (0·7%) 314 (1·7%) 384 (2·8%) 177 (4.3%)

Event rate per 1000 person years 1·55 3·67 6·41 10·36

Unadjusted hazard ratio 1·00 2·38 (1·88–3·00) 4·16 (3·31–5·22) 6.73 (5·23–8·66)

Adjusted hazard ratio (95% CI)* 1·00 1·43 (1·12–1·83) 1·83 (1·42–2·34) 2·53 (1·91–3·36)* Adjusted for age, gender, systolic BP, diastolic BP, LDL-C, HDL-C, HbA1c, BMI, duration of diabetes, smoking status, antiplatelet therapy, lipid-lowering treatment, RAS blockade, other blood pressure treatment, ethnicity, index of multiple deprivation. Numerical data were entered into models as continuous data.

19

A

20

B

21

C

Figure 1. Unadjusted freedom from the primary outcome (A), hospitalisation for heart failure (B), and cardiovascular mortality (C) by cumulative burden of microvascular disease. The primary outcome measure was cardiovascular mortality, non-fatal myocardial infarction or non-fatal ischaemic stroke. Log-rank test for the linear association between cumulative burden of microvascular disease for the primary outcome p<0.001; hospitalisation for heart failure p<0.001; and for all-cause mortality p<0.001.

22

A

B

23

C

Figure 2. Adjusted hazard ratio for the primary outcome (A), hospitalisation for heart failure (B), and cardiovascular mortality (C) by cumulative burden of microvascular disease and per 1 SD difference in values for established risk factors*The primary outcome measure was cardiovascular mortality, non-fatal myocardial infarction or non-fatal ischaemic stroke. 1 SD of each established risk factor is: BP 13.5/8.4 mmHg; LDL 0.9 mmol/L; BMI 6.3 kg/m2; HbA1c 1.3%.Adjusted for age, gender, systolic BP, diastolic BP, LDL-C, HDL-C, HbA1c, BMI, duration of diabetes, smoking status, antiplatelet therapy, lipid-lowering treatment, RAS blockade, other blood pressure treatment, ethnicity, index of multiple deprivation

24

Figure 3. Adjusted event rates for the primary outcome by cumulative burden of microvascular disease and established risk factor goals*The primary outcome measure was cardiovascular mortality, non-fatal myocardial infarction or non-fatal ischaemic stroke.* Adjusted for age, gender, systolic BP, diastolic BP, LDL-C, HDL-C, HbA1c, BMI, duration of diabetes, smoking status, antiplatelet therapy, lipid-lowering treatment, RAS blockade, any blood pressure treatment, ethnicity, index of multiple deprivation

25