Diabetes-Attributable Mortality in the United States from 2003-2016 Using a Multiple-Cause-of-Death Approach.

AIMS Deaths attributable to diabetes may be underestimated using an underlying cause of death (COD) approach in U.S. death records. This study sought to characterize the burden of diabetes deaths using a multiple-cause of death approach and to identify temporal changes in co-reported causes of death among those with diabetes listed anywhere on their death records. METHODS COD were identified using data from the National Center for Health Statistics from 2003-2016. We calculated age-adjusted mortality rates for diabetes as the underlying or contributing COD by race/ethnicity. We used ICD-10 codes to identify leading causes of death among those with and without diabetes on their death records. We compared temporal changes in deaths due to cardiovascular disease, cerebrovascular disease, cancer, and other causes. RESULTS The study population included 34,313,964 decedents aged ≥25 from 2003-2016. Diabetes was listed as an underlying COD in approximately 3.0% (n=1,031,000) and 6.7% (n=2,295,510) of the death records, respectively. Decedents with diabetes listed as an underlying COD experienced a 16% decline in mortality, and the race/ethnicity-specific average annual percentage changes (AAPC) showed significant declining trends for most groups (AAPC ranged from 0.18 to -2.83%). Cardiovascular disease remained the leading underlying COD among diabetes-attributable deaths, although its proportion of deaths fell from 31 to 27% over time. Co-reported COD diversified, and were more likely to include hypertension and hypertensive renal disease among those with diabetes on their death records. CONCLUSIONS Our findings underscore the importance of using a multiple-cause-of-death approach for more completely characterizing diabetes' contribution to mortality