Tseng’s study (1) indicated that only 19.8% of deaths among Taiwanese diabetic patients had an underlying cause of death attributed to cardiovascular disease (CVD), which was relatively low compared with the U.S. (49.4%) and U.K. (49.1%). One possible explanation was that Taiwanese diabetic patients were less likely to experience CVD than their counterparts in western countries. An Asia Pacific Collaboration cohort study did not support this hypothesis: no discernible differences were found between the hazard ratios for CVD deaths in Asian and Australasian populations (2).
The underlying cause of death is determined by a combination of factors, including both the physician’s certification and the coder’s interpretation of coding rules. Therefore, another explanation was that coders in Taiwan were more likely to assign diabetes as the underlying cause of death than in other countries. An evaluation study revealed that the diabetes death rates calculated from manually coded death records did not show significant differences with those based on a widely used standard computerized coding system (3). Lu (3) proposed a third explanation, that Taiwanese physicians were more likely to certify diabetes as the underlying cause of death than physicians in other countries.
To test the first and the third hypotheses, we compared the diabetes-related multiple-cause-of-death mortality data of three countries as part of the International Collaborative Effort on Automating Mortality Statistics (4). The three countries used the same computerized coding system; therefore, there were no discrepancies in assigning the underlying cause of death among the three countries.
The proportion of death certificates with mention of diabetes was similar (9–10%) for Taiwan, Sweden, and the U.S. (Fig. 1). However, among those certificates with mention of diabetes, only 58% of certificates in Taiwan also mentioned CVD, considerably less than in Sweden (85%) and the U.S. (84%). Of those death certificates with mention of both diabetes and CVD in Taiwan, diabetes was selected as the underlying cause of death in 55% of cases, considerably higher than in Sweden (20%) and the U.S. (32%).
For CVD, the death rate calculated according to multiple-cause-of-death mortality data can be a proxy of the prevalence in the decedent population (5). We found a lesser coexistence of CVD among certificates with mention of diabetes in Taiwan than in Sweden and the U.S. Thus, the first explanation was partially supported. Our findings also confirmed the third hypothesis, that Taiwanese physicians were more likely to prefer diabetes over CVD as the underlying cause of death than their counterparts in Sweden and the U.S.
One limitation of using multiple-cause-of-death mortality data as a proxy for prevalence is that in some circumstances, the deceased may have had CVD but the certifying physician chose not to report it on the death certificate. However, we could still conclude that the interpretation of differences in cause-of-death statistics among countries should take into account the differences in cause-of-death certification behaviors among physicians of different countries.
Number of death certificates with mention of diabetes, with mention of both diabetes and CVD, and with diabetes selected as the underlying cause of death (UCD) in Taiwan, Sweden, and the U.S., 2001.
Number of death certificates with mention of diabetes, with mention of both diabetes and CVD, and with diabetes selected as the underlying cause of death (UCD) in Taiwan, Sweden, and the U.S., 2001.
