Mortality and Causes of Death in a National Sample of Diabetic Patients in Taiwan

The study was approved and supported by the Department of Health, Executive Yuan, Republic of China. Because >96% of the total population of Taiwan is covered by the compulsory and universal NHI since March 1995, almost all diabetic patients have been using the NHI (12). Most diabetic patients receive their care in hospitals, and a diabetic patient will visit the outpatient clinic on average 35.8 times per year for any cause (13). Therefore, the hospitals’ databases for NHI claims are appropriate for deriving a national sample of diabetic patients. To assemble such a cohort, 66 hospitals distributed evenly across Taiwan were selected to participate in the recruitment. All of these hospitals provided the basal demographic data of all patients with a diagnosis of diabetes who submitted claims to the NHI for every month for 1 year. The diagnosis of diabetes was defined by the ninth revision of the International Classification of Diseases (ICD)-9 code 250 or the A-code (abridged code) of A181. Since the claim data included both inpatients and outpatients, it was believed that this recruitment procedure would not tend to select sicker patients. Although it is not easy to answer the question of sensitivity/specificity of this approach, both sensitivity and specificity were believed to be good for the following reasons: hospitals would tend to report all cases of diabetes in the claim data if the patients had diabetes in order to get better reimbursement and false report of diagnosis would face severe penalty from the Bureau of NHI.

Because the claim data were compiled every 1 month, there might be overlapping cases in the same hospitals in different months or the same case might appear in different hospitals. These cases were identified by using a unique identification number. The earliest date he or she appeared in the claim data was used as the date of entry into the study. This recruitment process began on 1 November 1995 and ended on 31 December 1998. All of the patients were then followed until the end of the year 2001.

The date and cause of death for a person was obtained by matching the computerized data file of the National Register of Deaths with an identification number. Because the contributing causes of death were not available in either the annual report or computer file, only the underlying cause of death was used for analyses in this study. The causes of death in the government computer files have been coded according to ICD-9 since 1981 (14).

Mortality rates were computed using a person-years denominator. The person-years of follow-up for each patient were calculated as the duration from the date of recruitment until the end of 2001 for those who were alive or to the date of death for those who died. Age- and sex-specific mortality rates were calculated, and the mortality rate ratios and their 95% CIs for men versus women in each stratum of age were estimated by Cox proportional hazard models.

Standardization was applied to compute the SMR using data for the Taiwan area (average of 1995–2001 midyear population) as the standard population. Byar’s approximation was used to calculate the 95% CIs for SMRs (15). Specific causes of death were classified into the following categories according to ICD-9 codes: diabetes (250), cancer (140–208), cardiopulmonary disease (401–429), stroke (430–438), disease of arteries, arterioles, and capillaries (440–448), nephropathy (580–589), infection (001–139, 320, 321, 326, 421, 460–466, 480–487, 510, 513, 551, 567, 590, 599, 680–686, 711, 730), digestive diseases (520–579, excluding 551), accidents (800–949), suicide (950–959), other causes (codes other than the above), and all causes.

At the end of recruitment, there were 256,036 diabetic individuals (118,855 men and 137,181 women). The mean age (±SD) of the patients was 61.2 ± 15.2 years.

With a total of 1,124,348.4 person-years of follow-up, 43,888 patients died and the crude mortality rate was 39.0/1,000 person-years. The age- and sex-specific mortality rates and SMRs and the mortality rate ratios for men versus women in each stratum of age are shown in Table 1. The mortality rates in the diabetic patients were age dependent, which increased from 14.1 to 86.3 per 1,000 person-years in patients aged <45 to ≥75 years, respectively, in the diabetic men and from 4.5 to 77.8 per 1,000 person-years in the diabetic women. Furthermore, the mortality rate ratios suggested a higher risk of mortality in the diabetic men. However, this sexual difference in mortality risk attenuated with increasing age. Excess mortality in the diabetic patients relative to the general population can be demonstrated by SMRs. A greater SMR in the diabetic men than in the diabetic women was observed in patients aged <45 years; and on the contrary, SMRs were greater in the diabetic women than the diabetic men in the older age-groups.

The observed numbers and percentages of specific causes of death and the SMRs for each specific cause and all causes of death by sex are shown in Table 2. Diabetes was listed as the underlying cause of death in about one-fourth and one-third of diabetic men and women, respectively. If diabetes was not considered as an underlying cause of death, then cancer was the most common cause of death in both sexes. However, if we considered deaths ascribed to cardiopulmonary disease, stroke, and disease of arteries, arterioles, and capillaries together as total cardiovascular deaths, then cardiovascular deaths would be ranked as the most common cause of death (19.8%) in all patients. For both sexes, the proportions ascribed to stroke were slightly greater than those ascribed to cardiopulmonary disease, and the proportions ascribed to disease of arteries, arterioles, and capillaries were actually small. Except for disease of arteries, arterioles, and capillaries in both sexes and accidents in the diabetic women, the diabetic patients had significantly higher risk of mortality than the general population from any of the specific causes of death listed in Table 2, as demonstrated by SMRs. Although the proportions ascribed to nephropathy were relatively small in both sexes, the excess mortality shown by the SMRs was greatest among all causes, except diabetes.

By using a cohort of diabetic patients from the hospitals’ claim data for the NHI and using the National Register of Deaths for record linkage, it was possible to study the mortality of a large number of diabetic patients for whom diabetes might not be listed as an underlying cause of death and who would therefore be concealed in vital statistics. Findings of the present study provide a more accurate look at many of the mortality issues related to diabetes in Taiwan.

The patterns of changes of mortality rates with regard to age and sex as shown in Table 1 were similar to those of other studies. In a U.S. national cohort (1), mortality over a 22-year period for diabetic patients increased from 12.4 per 1,000 person-years for those aged 25–44 years at baseline to 89.7 per 1,000 person-years for those aged 65–74 years. Another study (5) carried out in Italy showed that the mortality rate increased from 11.9 to 49.3 per 1,000 person-years for diabetic men aged 45–54 to ≥75 years, and the rates for the diabetic women were 7.5 and 50.8 per 1,000 person-years, respectively. The diabetic men also showed a higher risk of death than their female counterparts in both the U.S. and the Italian studies (1,5). The excess mortality (indicated by SMRs) attenuating with age in the diabetic patients (Table 1) was also observed in several other population-based studies (1,7,16,17).

Heart disease is a more common cause of death than stroke in most western countries. For example, heart disease was listed as the underlying cause of death in 49.4% of the death certificates in diabetic patients in the U.S. (1) and in 49.1% in the U.K. (4). On the contrary, stroke is a more common cause of death than heart disease in Asian populations. A recent multinational study (18) comparing the effects of diabetes on the risks of major cardiovascular death in the Asia-Pacific region, including Taiwan, also confirmed stroke as a more common cause of death than coronary heart disease in Asian ethnicities than in Caucasians living in Australia and New Zealand. Stroke is also a more common cause of death than heart disease in the Japanese (18,19). The diabetic patients in the present study also showed more deaths ascribed to stroke than to cardiopulmonary disease (in which coronary heart disease was included) in both sexes, though the differences were not marked (Table 2). Genetic factors may be responsible for the discrepancy between Asian populations and ethnicities of the western world, but the contribution of other risk factors, such as socioeconomic status, obesity, glycemic control, hypertension, and hyperlipidemia, etc., cannot be excluded. In the present study, if deaths ascribed to cardiopulmonary disease, stroke, and disease of arteries, arterioles, and capillaries were taken together to represent total cardiovascular death, then the proportion was 19.8% (a figure close to the proportion of 18.5% for cancer) among all causes of deaths (Table 2). Diabetes was ascribed to 28.8% of the causes of death in the present study (Table 2). It was neither easy to discern how these patients died nor could we judge what proportion among these patients could be ascribed to a cardiovascular death. However, the patients in this category were less likely to die from cancer than a cardiovascular disease because cancer would be much easier to recognize and code as an underlying cause of death in death certificates if a patient with cancer died. If we considered all of these patients as dying from cardiovascular death, this would sum up to 48.6% of all deaths, a figure that would be close to the reported underlying cause of death in 49.4% in the U.S. (1) and in 49.1% in the U.K. (4). Thus, cardiovascular disease could be the most common cause of death in diabetic patients. However, whether a lesser proportion of our diabetic patients would die from cardiovascular disease when compared with that of the diabetic patients in the U.S. or U.K. is an issue requiring further investigation. The finding that only a small proportion (0.3%) of the causes of death was ascribed to disease of arteries, arterioles, and capillaries in diabetic patients was consistent with the findings that the prevalence of peripheral vascular disease (20) and diabetic foot problems (21) were not as commonly seen as in Caucasians.

It was interesting to observe that diabetes was listed as an underlying cause of death in a higher proportion of women than men (Table 2). This could possibly suggest that more diabetic men might have suffered from other causes of death that were easily recognized clinically and would be entered on the death certificates, such as cancer. However, other possibilities could not be excluded.

Underreporting of diabetes as an underlying cause of death on death certificates is very common. In the present study, ∼71.2% of diabetic patients in Taiwan will not have diabetes recorded as an underlying cause of death (Table 2). An early study (9) showed that only 12.5% of all death certificates in diabetic patients listed diabetes as an underlying cause and that diabetes was not mentioned on 51% of death certificates in diabetic patients. A recently published study (1) reported that diabetes was listed as the underlying cause of death for only 7.7% of diabetic men and 13.4% of diabetic women. Diabetes might not be recognized at the patient’s death, and thus it was not listed on death certificates. Another reason was that a diabetic patient could die of a cause unrelated to diabetes, such as cancer. Moreover, diabetic patients could die from other chronic diseases frequently associated with diabetes, such as ischemic heart disease, stroke, and renal disease.

The excess mortality ascribed to nephropathy was highest (if diabetes was not considered as an underlying cause of death) in our diabetic patients in both sexes (Table 2), suggesting that our diabetic patients are more prone to develop severe kidney diseases that lead to death. Excess mortality ascribed to other causes, such as cancer, infection, digestive diseases, accidents, and suicide, were all observed in our diabetic patients (Table 2). Further detailed analyses are required to explore the association between these causes and diabetes.

The strengths of this study include a large and national sample of diabetic patients and the use of national register data to allow for complete ascertainment of death cases. However, several limitations should be pointed out. First, it is not possible to make a distinction between type 1 and type 2 diabetes in this study. However, because more than one-half of the diabetes cases occurring in school-age children were of type 2 diabetes (22) and only a small proportion (<3%, data not published) of the diabetic patients had type 1 diabetes in Taiwan, it is believed that the results largely reflect the mortality of type 2 diabetic patients. Second, this study did not examine mortality in undiagnosed diabetic patients. Third, the potential misclassification of some causes of death on death certificates cannot be ruled out.

In conclusion, our data support that diabetic patients have an excess mortality compared with the general population and that diabetic men suffer from a higher mortality rate than women. These observations are especially marked in younger patients. Cardiovascular disease could be the most common cause of death in diabetic patients, followed by cancer. Whether our diabetic patients suffer from a lesser proportion of death than that ascribed to cardiovascular disease, as seen in the U.S. or the U.K., requires further investigation.

This study was supported in part by grants from the Department of Health (DOH89-TD-1035) and the National Science Council (NSC-90-2320-B-002-197, NSC-92-2320-B-002-156), Executive Yuan, Republic of China.