Social predictors of less frequent dental attendance over time among older people: population-averaged and person-specific estimates.

Abstract

Longitudinal studies considering social disparities in the utilization of dental services are scarce. Repeated measures should be accounted for by the use of appropriate statistical methods. The purpose of this study was first to describe the patterns of less frequent dental attendance (less than once a year) over time from the age of 65-70 in Norwegian and Swedish 1942 cohorts. Second, this study estimated the influence of predisposing, enabling and need-related social predictors using marginal model with robust variance estimators and random intercept model, RIM, to account for the clustered structure of the repeated observations. Third, the study aimed to compare the estimates of associations between social predictors and less frequent dental attendance derived from marginal and random intercept models. In 2007 and 2012, all residents born in 1942 in selected counties of Norway and Sweden were invited to participate in a questionnaire survey. In Norway, the response rate was 58.0% (n = 4211) in 2007 and 54.5% (n = 3733) in 2012 with a follow-up rate of 70%. The corresponding figures in Sweden were 73.1% (n = 6078) and 72.2% (n = 5697), with a follow-up rate of 80%. Marginal and random intercept models were fitted for population-averaged and person-specific estimates. Design effects were calculated by comparing the results from ordinary logistic regression analyses and the marginal model with robust variance estimators. The proportion of the total variation due to differences between persons was reported using the intraclass correlation coefficient (ICC). Less frequent dental attendance declined from 14.5% to 12.2% in Norway and from 13.6% to 12.9% in Sweden. According to marginal and random intercept models, time-invariant (gender, country of birth, education) and time-variant social predictors (working status, social network, marital status, smoking and perceived health) contributed to less frequent dental attendance. A likelihood ratio test confirmed that adjustment for clustered observations was appropriate. The ICC was 0.90 in Norway and 0.85 in Sweden. The prevalence of less frequent dental attendance was low and dropped by increasing age from 65 to 70 years. Both at population and at person-specific levels, being advantaged on social aspects protects against less frequent dental attendance after 65 years of age in the Norwegian and Swedish cohorts investigated.