0888 Sleep Duration And Diabetes In The Us For 2007–2015: Moderation By Age, Sex, Race/ethnicity, And Body Mass Index

Abstract

Abstract Previous studies linked sleep and diabetes risk. Few studies, though, are sufficiently large to examine subgroups to determine who is at greatest risk. Data were obtained from the National Health Interview Survey conducted by the CDC from years 2007-2015 (N=272,338). Diabetes was assessed by self-report. Sleep duration was assessed as typical sleep within 24 hours and categorized as ≤4, 5, 6, 7 (reference), 8, 9, and ≥10 hours. Age was categorized as 18-34, 35-49, 50-64, and ≥65, sex and race/ethnicity were self-reported, and body mass index (BMI) was calculated from self-reported height and weight. Additional covariates included smoking, alcohol, employment, and survey year. Binary logistic regression using NHIS survey weights examined odds of diabetes predicted by sleep duration, adjusted for covariates. In a model adjusted for all covariates, increased diabetes risk was associated with ≤4 (OR=1.82,p<0.0001), 5 (OR=1.48,p<0.0001), 6 (OR=1.24,p<0.0001), 8 (OR=1.11,p<0.0001), 9 (OR=1.26,p<0.0001), and ≥10 (OR=1.81,p<0.0001) hours. In analyses stratified by age and sex, results were stronger among women for both short and long sleep. For age 18-34, short and long sleep duration were related to diabetes only among women. For both men and women, relationships with long and short sleep were strongest among those age 35-49 and 50-64. When results were stratified by race/ethnicity, both short and long sleep predicted diabetes for Non-Hispanic Whites, Blacks/African-Americans, Asians, and Native Americans, while only long sleep was associated with diabetes among Hispanics/Latinos and Multiracial/Others. Short and long sleep were associated with diabetes across all BMI categories, though the short sleep relationship was strongest in BMI<25 and the long sleep relationship was strongest in BMI≥40. Both short and long sleep are consistently related to diabetes. These relationships may disproportionately affect women and those in specific age and race/ethnicity groups. These data may be used to better identify at-risk groups. Support (If Any):