Incident Diabetes and Mobility Limitations: Reducing Bias Through Risk-set Matching.

BACKGROUND: Increased prevalence of diabetes in the U.S. population could contribute substantially to increases in disability at older ages. Previous studies have examined the association between prevalent diabetes and various impairments and disabilities. Methods considering incident, rather than prevalent, diabetes as the exposure of interest can reduce bias in estimates of these associations.
METHODS: Risk-set matching, a type of propensity score matching meant to handle time-varying exposures, was used to estimate the relationship between incident diabetes and mobility limitations among adults in the Health and Retirement Study. This approach ensures that covariates precede diabetes onset rather than follow it.
RESULTS: Individuals who were diagnosed with diabetes during the study period accumulated more subsequent mobility limitations than were accumulated by matched controls. Among observationally similar pairs of individuals, those who developed diabetes reported an average of 24.9% more mobility limitations at study exit than those who did not.
CONCLUSIONS: The magnitude of the relationship between diabetes and limitations estimated in this article is smaller than that presented in previous studies, but the method presented here is likely to provide a less-biased estimate of the association between diabetes and accumulation of mobility limitations.