Zhiqiang Wang, Huiying Zhang, Meina Liu

We aimed to assess if the gap in mortality between adults with and without diabetes has widened over time in US adults. This cohort study included 44,041 adults with diabetes from the US National Health Interview Survey between 1986 and 2009 linked to the National Mortality Index data up to 2011. Each participant with diabetes was matched to two participants without diabetes by age, sex, race, survey year, and region of residence (88,082 persons without diabetes). Mortality differences and hazard ratios were calculated for different time periods defined by three methods, according to 1) survey years with original follow-up durations, 2) follow-up calendar years, and 3) survey years with a fixed 3-year follow-up duration. Different methods of defining time periods produced substantially different mortality rates and changing patterns over time. The decline in mortality was higher when time periods were defined according to survey years with original follow-up durations than with the fixed 3-year duration. Different time periods had comparable baseline and attained ages only when the fixed duration was used. With this method, the gap between adults with and without diabetes progressively decreased from 224 (95% confidence interval 188-260) in 1992-1994 to 99 (65-132) per 10,000 person-years in 2007-2009. Hazard ratios declined significantly from 2.12 (1.88-2.38) in 1995-1997 to 1.70 (1.44-2.00) in 2007-2009. The decline in mortality over time was greater among adults with diabetes than those without diabetes. The gap in mortality between adults with diabetes and those without diabetes significantly narrowed in recent years, and was more than halved over the last 15 years.