Use of Survival Analysis to Predict Attrition Among Women Participating in Longitudinal Community-Based Nutrition Research

Abstract

ObjectiveTo identify participant characteristics and study methodology that influenced the completion of a 15-month community-based longitudinal study evaluating the impact of the Supplemental Nutrition Assistance Program Education and Expanded Food and Nutrition Education Program. DesignObservational longitudinal 15-month study across 12 data collection timepoints. Sociodemographic characteristics were collected with a paper-based survey at baseline. SettingFive counties across central and southern Illinois. ParticipantsWomen, aged 18 to 65 years (n = 297), recruited at sites likely to serve families eligible for the Supplemental Nutrition Assistance Program Education and Expanded Food and Nutrition Education Program (housing departments, child care centers, etc). Main Outcome MeasuresPredictors of participant attrition during the study duration. AnalysisCox proportional hazard models. ResultsNinety-seven participants were retained across the full study. In unadjusted models, greater income and education levels were significantly related to lower attrition; however, this relationship did not persist in a multivariate model. When adjusted for other characteristics, larger household size was the only measured variable significantly related to greater odds of attrition (odds ratio = 1.09; 95% confidence interval = 1.02, 1.17). Conclusions and ImplicationsSeveral characteristics predicting attrition in other settings were not significant in this study. Future attrition analyses that evaluate social support, transportation capacity, and type of phone in longitudinal nutrition education studies are warranted.