OBJECTIVES:To estimate trends in the prevalence and comorbidities of diabetes mellitus (DM) in U.S. nursing homes from 1995 to 2004. DESIGN: SAS callable SUDAAN was used to adjust for the complex sample design and assess changes in prevalence of DM and comorbidities during the study period in the National Nursing Home Surveys. Trends were assessed using weighted least squares linear regression. Multiple logistic regressions were used to calculate predictive margins. SETTING: A continuing series of two-stage, cross-sectional probability national sampling surveys. PARTICIPANTS: Residents aged 55 and older: 1995 (n 5 7,722), 1997 (n 5 7,717), 1999 (n 5 7,809), and 2004 (n 5 12,786). MEASUREMENTS: DM and its comorbidities identified using a standard set of diagnosis codes. RESULTS: The estimated crude prevalence of DM increased from 16.9% in 1995 to 26.4% in 2004 in male nursing home residents and from 16.1% to 22.2% in female residents (all Po.05). Male and female residents aged 85 and older and those with high functional impairment showed a significant increasing trend in DM (all Po.05). In people with DM, multivariate-adjusted prevalence of cardiovascular disease increased from 59.6% to 75.4% for men and from 68.1% to 78.7% for women (all Po.05). Prevalence of most other comorbidities did not increase significantly. CONCLUSION: The burden of DM in residents of U.S. nursing homes has increased since 1995. This could be due to increasing DM prevalence in the general population or to changes in the population that nursing homes serve. Nursing home care practices may need to change to meet residents' changing needs. J Am Geriatr Soc 58: 724-730, 2010.

Introduction