Cause‐specific mortality rates in patients with diabetes according to comorbid macro‐ and microvascular complications: BioBank Japan Cohort

Abstract Objective This study aimed to compare cause‐specific mortality rates in patients with type 2 diabetes with and without various vascular complications. Methods In Japanese hospitals, we followed up 30 834 patients with a mean age of 64.4 (standard deviation [SD]: 11.1) years. Patients were followed up from 2003 to 2007 for a median of 7.5 (interquartile range: 6.1‐9.7) years. We calculated cause‐specific mortality rates (number of deaths/1000 person‐years) and confounder‐adjusted hazard ratios in patients with macrovascular disease and in those with diabetic nephropathy, neuropathy and retinopathy, allowing for overlap of complications. Results All‐cause mortality rate was highest (51.4) in the nephropathy group, followed by the macrovascular disease group (45.2), the neuropathy group (39.5), the retinopathy group (38.7) and the nonvascular complication group (18.1). In the nephropathy group, morality rates of ischaemic heart, cerebrovascular, and infectious diseases and cancer were also highest among the groups. However, the cancer mortality rate was similar among the vascular complication groups. Relative to the nonvascular complication group, covariate‐adjusted hazard ratios for ischaemic heart and cerebrovascular disease mortality were triple to quadruple in the macro‐ and microvascular complication groups. All‐cause mortality rates rose exponentially according to age. Conclusion Highest risks of all‐cause, cancer, and ischaemic heart, infectious, and cerebrovascular disease mortality were determined in Japanese patients with diabetic nephropathy. Although cancer is the primary cause of death in Japanese patients with diabetes, cancer mortality rates are similar among those with and without vascular complications.


| INTRODUC TI ON
The number of patients with diabetes has been rapidly increasing worldwide. The International Diabetes Federation estimated that 4.2 million people aged between 20 and 79 years died from diabetes in 2019, and diabetes accounted for 10.7% of global all-cause mortality of people of this age. 1 The mortality rate of patients with diabetes is extremely high. Internationally, all-cause mortality rates are 29 and 23 per 1000 person-years in men and women with diabetes (mean age of 58 years), respectively, while they are 12 and 7 per 1000 person-years, respectively, in the general population (mean age of 55 years). 2 In Western countries and China, the primary cause of death in patients with type 2 diabetes is vascular diseases, followed by cancer. 2,3 In contrast, Taiwanese and Japanese patients with type 2 diabetes primarily die from cancer followed by vascular diseases. 4,5 Therefore, the ranking of causes of death in patients varies depending on the country. Since treatment of diabetes and self-management of patients improve with time, mortality rates of nonmacrovascular diseases will increase in patients with diabetes. However, evidence of cause-specific mortality rates in patients in Asian countries is insufficient.
The prevalence of vascular complications in patients with diabetes also varies, depending on the country and its data sources. Having In this study, we examined a hospital-based cohort that included patients with known prevalent diabetes to determine all-cause and cause-specific mortality rates of patients with diabetic complications. We also calculated the hazard ratios to compare the mortality risks among those with diabetic complications.

Between the fiscal years 2003 and 2007, the Tailor-Made Medical
Treatment Program with the BioBank Japan Project registered patients with type 2 diabetes at 66 hospitals. The study profile has been published elsewhere. 8 In brief, we followed participants with one of 47 targeted diseases in these hospitals from June 2003 to March 2013. We observed survival of the patients and new onset of diseases.
During the observation period, when participants without diabetes developed diabetes, we enrolled and followed them up as patients with diabetes. Medical coordinators in the participating hospitals collected information on survival or death of the patients using medical records. If the patients had not visited the hospitals for longer than 1 year, the medical coordinators requested residence registrations to municipalities to certify their life or death. The causes of deaths were investigated by requesting vital statistics by the Vital Statistics Act.
Consequently, the cohort achieved a 97% follow-up rate.

| Vascular complication categories
We collected information on diabetic complications using a standardized questionnaire from the medical records at enrolment. 8 We identi- The diabetic nephropathy group included patients with complicated diabetic nephropathy of stage 1 (prenephropathy), stage 2 (incipient nephropathy), stage 3 (overt nephropathy), stage 4 (kidney failure) and stage 5 (any status of continued dialysis therapy). 9 In this hospital setting, where only a subset of patients was followed up by diabetologists, most of the patients in this category had stage 2 or 3 nephropathy. (c) The diabetic neuropathy group included patients with diabetic polyneuropathy. (d) The diabetic retinopathy group included patients with simple, preproliferative and proliferative diabetic retinopathy. 10 This study permitted overlapping of vascular complications, which is usual in patients with diabetes. We aimed to calculate mortality risk according to each complication rather than the combination. If patients had two or more vascular complications, they were included in all of the relevant vascular complication groups.

| Measurements
We collected clinical information of the patients at baseline by interviewing the patients and reviewing medical records. 8 The informa-

| Statistical analysis
We described proportions of male sex, diabetic medication, cancer or macrovascular comorbidity, diabetes therapy and smoking, drinking, and physical activity habits, means of age, body mass index (BMI), systolic blood pressure, serum total cholesterol levels, HbA1c, disease and follow-up duration at baseline in each vascular complication group. The rates of all-cause, ischaemic heart, cerebrovascular, infectious and gastrointestinal diseases, aortic dissection, cancer, TA B L E 1 Baseline characteristics of patients with type 2 diabetes with macro-and microvascular complications    of the all-cause mortality rate was higher in male patients than in female patients. Figure 1 shows the Kaplan-Meier estimates for all-cause mortality.

| Crude survival curves and adjusted hazard ratios
The mortality in each vascular complication group was higher with statistical significance than that in the nonvascular complication

| D ISCUSS I ON
In our study, Japanese patients with type 2 diabetes who were complicated by diabetic nephropathy showed point estimates of the highest mortality rates from all causes, cancer, infectious disease, renal failure and ischaemic heart disease. Unlike Western patients with diabetes who mostly died from ischaemic heart disease, 2 Japanese patients mostly died from cancer. In Japanese patients, the second most common cause of death was infectious disease, and the third most common cause was ischaemic heart disease.
Western patients with diabetes and a history of macrovascular events have a high risk of mortality from recurrent macrovascular events. 13 However, in our Japanese cohort, the point estimates of mortality rates ( Table 2) and adjusted HRs (Figure 2) from ischaemic heart and cerebrovascular diseases in the nephropathy group were higher than those in the macrovascular disease group (  20 In contrast, a Japanese study reported that 47.1% of patients with diabetes had diabetic neuropathy-related symptoms, bilateral loss or a reduction in the Achilles tendon reflex, or bilateral reduction of sense vibration. 21 Other Japanese studies showed that 36.7% of patients with type 2 diabetes had some type of diabetic neuropathy 22 and 79.8% of patients with F I G U R E 1 Mortality of Japanese patients with diabetes and vascular complications diabetes in the capital sphere had peripheral or autonomic neuropathy. 23 The reason for the difference in the prevalence of diabetic neuropathy between previous Japanese studies 22,23 and our study (21.7%) may be that diagnosing diabetic neuropathy usually requires specific examinations by neurologists or diabetologists.
By contrast, the prevalence of diabetic retinopathy is similar between a previous Japanese survey and our data. The prevalence of retinopathy was 40.3% in American patients with diabetes aged 40 years or older, 24 while this prevalence was 22.9% in a report of Japanese patients with type 2 diabetes 22 and 18.9% in our study.
The reason for this similarity in Japanese data sets may be because diabetic retinopathy is relatively easily detected when older people consult ophthalmologists for problems with their eyes. However, the primary cause of death was cancer in our Japanese patients, as well as in Taiwanese patients with diabetes. 5 This finding is in contrast to ischaemic heart disease as the primary cause of death in patients with diabetes in the US 26 and in Italy. 27 4 This survey also showed that death from ischaemic heart disease in the general population had slightly decreased from 7.3% to 6.5%. Therefore, the high mortality rate of cancer in our study could be explained by the decrease in mortality rate of ischaemic heart disease in Japanese patients with diabetes, a high mortality rate of cancer in the Japanese general population, and the extended life span of patients with diabetes who are vulnerable to cancer.
In our study, the highest sixfold mortality rate of ischaemic heart disease in patients with diabetic nephropathy compared with those with no vascular complications ( Table 2) should be of concern to all healthcare professionals involved in diabetes. Surprisingly, patients  Our study showed little difference in cancer mortality rates among the vascular complication groups (Table 2 and Figure 2).
Previous studies reported a 20% (95% CI: 15%-26%) extra risk of the incidence of cancer 31 and a 25% (95% CI: 19%-31%) extra risk of cancer mortality in patients with diabetes compared with individuals without diabetes. 2 The reasons for these extra risks could be hyperinsulinemia 32 or hyperglycaemia, 33 and mortality of cancer increases in proportion to an elevation in fasting plasma glucose levels. 2 Although we also aimed to determine the difference in cancer mortality rates among patients with diabetes with various vascular complications, we did not find a large difference in an extra risk of cancer mortality in those with and without vascular complications. This lack of difference could be attributed to similar HbA1c levels among the vascular complication groups, according to previous studies. 2,34 There are a few limitations of this study. First, the patients were limited to those who were followed up in medium to large hospitals.
Because Japanese patients with mild diabetes are more likely to be followed up in clinics, the participating patients may have been bi- to be underdiagnosed. This situation also implies that the nonvascular complication group might have included patients who actually had retinopathy or neuropathy and that the actual HRs of the retinopathy and neuropathy groups might have been higher than those measured in this study. In this case, the death risks of the retinopathy and neuropathy groups should be more strongly emphasized in clinical practice. Third, a subset of data of causes of death were missing. This lack of data reduced the sum of the mortality rates of major causes.
There are several aspects of our study that strengthened our results. First, we could calculate various cause-specific mortality rates among patients with diabetes using vital statistics. 8 Use of clinical records, interviews, and administrative data enabled a high follow-up rate of 97.0% for the participants in the project. We consider that death of the patients was sufficiently recorded in this cohort. These results would be useful when clinicians prospectively quantify the prognostic mortality risks of their patients. Second, the number of participating patients with diabetes was large. This study may be unique in that the large sample size allowed estimation of cause-specific or sex-and age-specific mortality rates. Third, diagnoses of diabetes and its complications by medical doctors assured the internal validity and enabled comparisons among vascular complication groups.
In conclusion, data in Japanese hospitals show that among patients with diabetes, those with nephropathy have the highest mortality rates of all causes, cancer, and macrovascular and infectious diseases. The primary cause of death of these patients with diabetes with and without all vascular complications was cancer, and the cancer mortality rate was similar, regardless of whether vascular complications were present.

ACK N OWLED G M ENTS
We express our gratitude to all of the participants in the BioBank Japan Project. We thank all of the medical coordinators of the co- Knapp for editing a draft of this manuscript.

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest in this work.