SCHOOL OF POPULATION HEALTH THE UNIVERSITY OF QUEENSLAND Measuring diabetes mortality: Problems & Prospects Chalapati Rao April 2011

SCHOOL OF POPULATION HEALTHTHE UNIVERSITY OF QUEENSLAND

Measuring diabetes mortality:

Problems & Prospects

Chalapati Rao

April 2011


Outline Background

– Pathophysiology of diabetes mortality

– Conventions in coding and statistical presentation

– Current understanding of diabetes mortality

Multiple cause of death analyses– Diabetes – underlying cause vs ‘mentions’– Diabetes and IHD / Stroke– Diabetes and renal disease

Summary of issues

Prospects for consistent and comparable data


Background / context

Diabetes is an increasing public health problem

Measures of incidence / prevalence are difficult to attain for various reasons

– Sampling issues / response rates / measurement error

Population level measures of mortality appear straightforward, from national death registration systems


Medical certificate of cause of death


Example 1

Cardiac arrest

Congestive heart failure

Myocardial infarction

Metastatic lung cancer

Mode of dying

Ia

Ib

II

Underlying cause of death: Myocardial infarction

Death certification


Definition of underlying Cause of Death

When only one cause: select for tabulation

When several causes: “underlying” = the disease or injury, which initiated the train of morbid events leading directly to death, or the circumstance of the accident or violence which produced the fatal injury.


Rules for ICD Coding Recommend primary tabulations based on underlying causes

(data submitted to WHO Mortality Database)

Rules for selection of underlying causes

– General Principle

– Selection rules 1 – 3

– Modification rules A – F

– Specific rules for individual causes (decision tables ~ several hundred

pages) CURRENTLY, AUTOMATED CODING SYSTEMS IN SEVERAL COUNTRIES, INCLUDING

AUSTRALIA AND USA


Causes of diabetes deaths

DIABETES

Metabolic complications

Direct microvascular changes

Associated vascular changes

(Atherosclerosis)

Contributory cause

Ketoacidosis Hyperosmolar

coma Hypoglycaemia

Peripheral vascular / neural gangrene → sepsis

Diabetic renal disease

MI Stroke ? Diabetic

cardiomyopathy

Tuberculosis Cancer Pneumonia etc


Rules for diabetes

Underlying causes coded to three broad categories– Insulin dependent DM – E10– Non-insulin dependent DM – E 11– Unspecified DM – E14

Each with 9 sub categories with a fourth character extension .0 – coma .1 - ketoacidosis .2 – renal complications .3 – ophthalmic complications .4 – neurological complications .5 – peripheral circulatory complications .6 – other specified complications .7 – multiple complications .8 – unspecified complications .9 – without complications


Table 2: [C1] Distribution of numbers of fourth character diabetes UCOD coded deaths

Fourth character for UCOD E10 – E14

Australia USA

1999 2006 1999 2004

.0 – coma 34 29 688 464

.1 - ketoacidosis 38 65 1,894 1,810

.2 – renal complications 115 131 1,730 1,727

.3 – ophthalmic complications 7 5 49 31

.4 – neurological complications 11 12 392 394

.5 – peripheral circulatory complications 497 507 7,602 7,616

.6 – other specified complications 6 8 274 254

.7 – multiple complications 49 67 794 663

.8 – unspecified complications 3 - 177 55

.9 – without complications2,187

2,838

(78%)54,799

60,124 (82%)

Total deaths with diabetes as UCOD 2,947 3,662 68,399 73,138


Multiple causes of death analysis

To explore magnitude of diabetes mortality as a co-morbidity with other conditions in general; and with cardiovascular / renal disease in particular

To explore the phenomenon of mortality coded to ‘diabetes without complications’

To assess potential influence of certification / coding practices on statistics


Table 1. Trends in deaths from diabetes (underlying cause and total mentions on death certificate): numbers and percentage of total deaths for Australia (1999 and 2006) and USA (1999 and 2004).

Number of deaths (% of all deaths)

Australia USA

1999 2006 1999 2004

Diabetes UCOD 2,947 (2.3%)

3,662 (2 .7%) 68,399 (2.9%)

73,138(3.1%)

Diabetes listed on, certificate 9,588

(7.5%)12,811

(9.6%)209,679

(8.8%)225,455 (9.4%)

Preliminary data assessment


Observations

Diabetes mentioned in Part II about twice as often as in part 1

Similar observations in data from Brazil, France, Sweden, and

several other European countries

This indicates its higher association with mortality than perceived

from underlying cause data

All this occurs in a backdrop of general under reporting of diabetes

on death certificates Diabetes listed as a single cause in only 1- 2% cases, so the deaths

coded to ‘DM without complications’ is a misrepresentation of reality


Diabetes and Myocardial Infarction (MI) / Stroke

SPECIAL RULE

If MI / stroke listed as a consequence of diabetes in Part 1, then underlying cause is diabetes

If MI / stroke listed in Part 1, with Diabetes in Part II, then underlying cause is MI / stroke


Example 1

Left ventricular failure


Diabetes

Ia

Ib1c

II

Underlying cause of death: Diabetes

Death certification


Example 2

Left ventricular failure


Diabetes

Ia

Ib1c

II

Underlying cause of death: Myocardial infarction

Death certification


Diabetes and Myocardial Infarction

Year Australia USA

UCOD Pt 2 UCOD % UCOD Pt 2 UCOD %

1999 1311 2312 36% 26168 52724 33%2000 1315 2345 36% 26578 52025 34%2001 1341 2400 36% 26870 51526 34%2002 1458 2452 37% 27474 50861 35%2003 1544 2291 40% 27544 50006 36%2004 1627 2235 42% 26916 47877 36%2006 1652 2129 44%

Deaths with MI in Part 1; and diabetes in Part 1 (UCOD) or Part 2


Diabetes and Stroke

Year Australia USA

UCOD Pt 2 UCOD % UCOD Pt 2 UCOD %

1999479 762 39% 7817 12769 38%

2000463 759 38% 7743 12462 38%

2001510 802 39% 7814 12455 39%

2002535 828 39% 7658 12623 38%

2003612 795 43% 7455 12039 38%

2004644 788 45% 7317 11793 38%

2006659 862 43%

Deaths with stroke in Part 1; and diabetes in Part 1(UCOD) or Part 2

Diabetes is being increasingly listed in Part 1, thereby augmenting its magnitude in underlying cause mortality statistics


Other evidence

Similar variations observed in other countries (Taiwan, Sweden)– South Africa is quite different with 98% of diabetes related deaths

having diabetes recorded in Part 1

Variation tested through clinical vignettes describing mortality from cardioavascular disease in diabetics

– Identified high degree of subjectivity in physcians, in the practice of listing Diabetes in part 1 vs Part 2, with concomitant MI / stroke


Clinical / pathological perspective ‘Diabetes predisposes to cardiovascular diseases’

– Harrison’s Principles of Internal Medicine; 17th Edition

‘Diabetes mellitus, whether type 1 or type 2, is a very strong risk factor

for the development of coronary artery disease (CHD) and stroke’

– Hurst: The Heart – 12th Edition

‘Diabetes mellitus induces hypercholesterolemia and a markedly

increased predisposition to atherosclerosis’

– Robbins Pathologic Basis of Disease, 6th edition The American Heart Association has designated DM as a major risk

factor for cardiovascular disease (same category as smoking, hypertension, and hyperlipidemia)


Inferences

Inconsistency in certification practices compounded by ambiguity in coding rules

As a result, Diabetes / CVD statistics based on underlying causes not readily comparable across countries, or over time


Diabetes with renal complications

Clear relationship from pathological perspective

– Microvascular and basement membrane changes in glomerulus

producing distinctive Kimmelstein Wilson lesions PLUS

atherosclerotic changes in renal arterial vessels

Designated code in ICD rules

– BUT – some interesting fine print…..


Table 2: [C1] Distribution of numbers of fourth character diabetes UCOD coded deaths

Fourth character for UCOD E10 – E14

Australia USA

1999 2006 1999 2004

.0 – coma 34 29 688 464

.1 - ketoacidosis 38 65 1,894 1,810

.2 – renal complications115

(3.9%) 131

(3.6%) 1,730

(2.5%) 1,727

(2.4%)

.3 – ophthalmic complications 7 5 49 31

.4 – neurological complications 11 12 392 394

.5 – peripheral circulatory complications 497 507 7,602 7,616

.6 – other specified complications 6 8 274 254

.7 – multiple complications 49 67 794 663

.8 – unspecified complications 3 - 177 55

.9 – without complications2,187

2,838

(78%)54,799

60,124 (82%)

Total deaths with diabetes as UCOD 2,947 3,662 68,399 73,138


Coding rules for ‘DM with renal complications’

Coded as underlying cause IF any of the following

terms listed in part 1 of the death certificate

– Diabetic nephropathy

– Kimmelstein Wilson Disease

– Intracapillary Glomerulosclerosis


Specific additional rule Coded to DM with renal complications IF

A) the death certificate lists DM in part 1; AND

B) with any of the following listed as a consequence, in Part 1– Unspecified disorder of the kidney or ureter (N28.9)– chronic nephritic syndrome (N03)– nephrotic syndrome (N04)– unspecified nephritic syndrome (N05) or – unspecified contracted kidney (N26)

RENAL FAILURE (N17-N19) listed as a consequence of Diabetes is not considered as Diabetes with renal complications


Research hypothesis

Certifying physicians are more likely to write ‘RENAL FAILURE’ on death certificates, rather than the more complex terms listed in the rules

‘Diabetes with renal failure’ can be considered as ‘diabetic renal disease’

Question: What do the multiple causes of death data tell us about mortality from diabetic renal disease?


Multiple cause of death data

1. Extract all deaths coded to DM as underlying cause

2. Create subset of deaths coded to ‘DM with renal complications’

3. THEN Screen all remaining DM UCOD deaths, for renal failure listed as a consequence of DM in Pt 1, to create additional deaths subsets

4. Additional deaths A: Deaths with UCOD -DM without complications; but with Renal failure listed in Pt 1

5. Additional deaths B: Deaths coded to other DM UCOD categories, but with renal failure listed in Pt 1


Deaths from diabetic renal disease

Deaths

Australia USA

1999 2006 1999 2004

Diabetes with renal complications 115 131 1,730 1,727

Additional deaths ‘A’ 301* 577 11,543 13,739

Additional deaths ‘B’ 118# 169 2,166 2,241

Total diabetes UCOD 2,947 3,662 68,399 73,138

* out of 2187 deaths coded to DM without complications

# out of 645 deaths coded to diabetes with other categories of complications except renal or *


Comparison of underlying vs mulitple cause rates

Age-standardised death

rates (per 100,000)*

USA Australia

1999 2004 % change 1999 2006 % change

Underlying cause rate

(UCR)0.68 0.62 -7.7 0.67 0.64 -5.3

Multiple cause rate 1

(MCR1)5.2 5.6 7.3 2.5 3.4 37.7

Rate ratio†

(95% CI)

7.7

(7.3 – 8.1)

8.9

(8.5 – 9.4)−

3.7

(3.0 – 4.6)

5.4

(4.5 – 6.5)−

Multiple cause rate 2

(MCR2)6.1 6.4 5.6 3.2 4.3 34.2

Rate ratio†

(95% CI)

9

(8.5 – 9.4)

10.3

(9.8 – 10.8)−

4.7

(3.9 – 5.8)

6.7

(5.6 – 8.0)−

*Standardised to Australian population in 2006. † To the UCR


Time trends


Summary

Conventional mortality statistics based on underlying causes mask true patterns of mortality from diabetes and its complications

Issues in certification, coding and statistical presentation result in this situation

These need to be addressed, to improve prospects of more realistic understanding of diabetes mortality, given the impending epidemic in the Asia Pacific region


Global availability of cause of death data : 2001

The boundaries and names shown and the designations used on this map do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement.

© WHO 2003. All rights reserved

goodincompletepoor/non existant

Vital Registration


References / acknowledgements

1. Rao C, Adair T, Bain C, Doi SA. Mortality from diabetic renal disease: a hidden epidemic. Eur J Public Health. 2011 Jan 18. [Epub ahead of print]

2. Adair T, Rao C. Changes in certification of diabetes with cardiovascular diseases increased reported diabetes mortality in Australia and the United States. J Clin Epidemiol. 2010 Feb;63(2):199-204.

Documents

SCHOOL OF POPULATION HEALTH THE UNIVERSITY OF QUEENSLAND Measuring diabetes mortality: Problems & Prospects Chalapati Rao April 2011