Predictors of older adults' longitudinal dental care use. Ten-year results.

Our purpose was to identify factors predictive of reported dental care use by elders (65+) over a ten-year period in Massachusetts. The Massachusetts Health Care Panel Study began in 1974-75 (wave 1) as a statewide survey of 1625 noninstitutionalized elders. Wave 2 occurred in 1976, wave 3 in 1980, and wave 4 in 1985. The 540 persons who participated in all four waves are the subject of this report. The remainder either died, entered nursing home, or were lost to follow-up. This longitudinal design permits analysis of cohort, aging, and period effects. The outcome variables were self-reported dental care use within two years, or more than two years, as reported at waves 1, 3, and 4. Wave 2 was excluded because less than two years had elapsed since the previous wave. To identify factors predictive of reported use, we used a generalization of the logistic regression model that included a random effect term, which accounts for repeated measures being made on the same subjects. Covariates in the model were dentate status, education, income, cohort, sex, martial status, and time. The variable "time" served as a measure of aging/period effects. Persons were grouped into four birth cohorts. Before adjusting for other covariates, cohort was significantly associated with dental care use, but was not so in the full multivariate model. Dentate status, education, and income were significant predictors of use. The cohort effect was explained by dentate status, education, and income. Time was not significant, indicating no aging/period effects in this ten-year period, or that their net effect was zero.