Diagnosis of Type 2 Diabetes at an Older Age: Effect on mortality in men and women

The DARTS/MEMO (Diabetes Audit and Research in Tayside Scotland/Medicines Monitoring Unit) Collaboration (9) works on the record linkage of health care data to facilitate epidemiological and health services research in the population of Tayside, Scotland (estimated population of 389,800 in 1998). Record linkage is enabled by the widespread use of a unique health care identifier (community health index number) that is allocated to people when they register with a general practitioner in Scotland. This study was conducted in a defined population of general practitioner–registered people who were known to be residents of Tayside during the entire study period (unless they died).

DARTS is a validated population-based diabetes information system in Tayside that is compiled by record linking several independent data sources. It has 95% sensitivity at identifying people with diabetes (9). It was used to identify people aged >65 years who were diagnosed with type 2 diabetes between January 1993 and December 2002. For every diabetic individual, two nondiabetic comparators were identified from the general population, matched for age (within 1 year) and sex. The index date of a diabetic person was their date of diagnosis, as was the index dates of their matched comparators.

In a cohort study, the diabetic people and the nondiabetic comparators were followed for a maximum of 10 years for the primary outcomes of all-cause mortality and cardiovascular mortality. Mortality data were provided via death certification records, and the “underlying cause of death” was classified as a cardiovascular death according to ICD 9 codes 390–459 (10) and ICD 10 codes I00–99 (11). Kaplan-Meier survival curves were constructed. A Cox proportional hazards model with multiple covariates was fitted to the sample data. The covariates were population type (nondiabetic/diabetic), age at index date, sex, and Carstairs deprivation category (12) (an area-based measure of material deprivation).

There were 3,594 people with type 2 diabetes (48% male) and 7,188 matched comparators identified in the study. Subject survival is presented separately for men and women in Fig. 1. Over a mean follow-up period of 4.6 ± 2.9 years, 909 (25.3%) patients in the diabetic cohort and 1,651 (23.0%) in the nondiabetic cohort died. For men, the relative risk for mortality in the diabetic cohort compared with the nondiabetic cohort, adjusted for deprivation, was 1.06 (95% CI 0.94–1.19). There were no differences when patients were stratified by age at diagnosis. For women, the adjusted relative risk for mortality in the diabetic cohort compared with the nondiabetic cohort was 1.29 (1.15–1.45). The risks for those diagnosed with diabetes at 65–74 years, 75–84 years, and >85 years were 1.47 (1.21–1.78), 1.15 (0.97 -1.38), and 1.36 (1.06–1.73), respectively. Cardiovascular deaths accounted for 49.4% of the deaths in people with diabetes and 45.2% in those without (adjusted relative risk 1.01 [0.93–1.10]).

In this study, men diagnosed with type 2 diabetes after 65 years of age had no increased risk of mortality compared with nondiabetic men of the same age. This result contrasts sharply with the significant effect of diabetes on mortality in a population of older diabetic people that included newly diagnosed individuals and those with a longer duration of disease (5). Studies that have found small differences in mortality, even in men diagnosed at a later age, have not adjusted for the confounding effect of deprivation (possibly independently associated with diabetes prevalence [13] and mortality).

Our data came from a validated diabetes system and are known to be reliable (9). The mean follow-up of almost 5 years was similar to that of an Italian study (7). It may be that mortality of diabetic and nondiabetic men diverges in later years after diagnosis, although this did not occur for at least 7 years following diagnosis in our study. A weakness of the study was that we were unable to adjust for cardiovascular risk factors at baseline. However, we might expect a poorer profile in the diabetic cohort (14,15), with adjustment resulting in even less of a difference between the two cohorts. The proportion of cardiovascular deaths was relatively low; however, we used “underlying cause of death” to determine whether a death had a cardiovascular cause, and this may underestimate cardiovascular mortality. We also acknowledge that there may be a degree of under-ascertainment of diabetes in the nondiabetic cohort.

Our result may be due to a survivor effect, with men with more severe diabetes having presented and died before the age of 65 years. However, this does not detract from the important clinical point: that the effects of diabetes (in this case, mortality) in those diagnosed at an older age clearly differs markedly from those diagnosed when younger.

In contrast to men, women diagnosed with type 2 diabetes at an older age had an increased risk of mortality compared with older nondiabetic women. This was evident 3 years after diagnosis, particularly for women aged 65–74 years. Diabetes may be a stronger risk factor for women compared with men (as our results suggest), although whether this can be accounted for by a higher prevalence of classic cardiovascular risk factors in diabetic women compared with diabetic men is still controversial (16). We were unable to adjust for cardiovascular risk factors, but our study suggests that the effects of diabetes may vary according to sex, even in older individuals.

This study therefore highlights the importance of recognizing the heterogeneity of the older diabetic population.

The authors thank members of the DARTS/MEMO Collaboration and members of the DARTS Steering Group for their support.