Biological Age Affecting Attrition and Tooth Loss in a Follow-up Study

Abstract

In population-based longitudinal studies, bias caused by nonresponse among eligible participants and attrition during follow-up thwarts conclusions. As this issue is not commonly addressed in dental studies, it is the aim of this study to examine the consequences of attrition with respect to tooth loss and mortality in a 10-y follow-up study. From the Study of Health in Pomerania (SHIP-0), a biological age (BA) score was constructed from 10 systemic biomarkers and related to one’s actual chronological age (CA). The 3,417 dentate participants were stratified according to their BA-CA scores into tertiles: individuals with younger BA than their CA, those with concurrent BA and CA, and those with older BA than their CA. Baseline characteristics and propensity of leaving or remaining in the study were compared across these tertiles. We compared the characteristics within BA strata in the remainers of SHIP-2 (10-y follow-up) and their impact on tooth loss. Besides dropout by those who died, the attrition propensity of baseline study participants was dose dependent as related to BA-CA scores and socioeconomic factors. BA younger participants were underrepresented in dropouts but overrepresented in remaining follow-up participants. BA younger participants had a more favorable risk profile, better oral health, and a lower mortality rate than BA older participants. For the BA older participants, the opposite was observed. Remainers attaining the follow-up SHIP-2 were healthier and more health conscious. After 10 y, their tooth retention was still directed by BA constructed at baseline. The results support the assumption that individual risk profiles aggregated in BA constitute characteristic susceptibility patterns affecting perseverance or attrition in long-term follow-up studies. Attrition, which is common to follow-up studies, changes the study composition of participants depending on their BA and hence the transferability of results to the baseline population. The baseline BA gradient persists even after a long time.