Geospatial analysis of risk factors contributing to loss to follow-up in cleft lip/palate care

Abstract

Background: Multidisciplinary cleft care depends on follow-up at specified time points to monitor and address functional or aesthetic concerns that may arise during a child's development. However, loss to follow-up (LTFU) is common and can lead to missed opportunities for therapeutic and surgical intervention. This study explores clinical, demographic, and geographic determinants of LTFU in cleft care. Methods: Medical records were retrospectively evaluated for 558 pediatric patients of a single mid-volume cleft team. The primary outcome was LTFU. Spatial dependency was evaluated using variograms. The probability of LTFU was assessed using a generalized linear geostatistical model within a Bayesian framework. Risk maps were plotted to identify vulnerable communities within our state at higher risk of LTFU. Results: Younger age at last encounter was a strong predictor of LTFU (P < 0.0001), even when ignoring spatial dependency among observations. When accounting for spatial dependency, lower socioeconomic status [OR = 0.98; 95% CI = (0.97 0.99)] and cleft phenotype [OR = 0.55; 95% CI = (0.36, 0.81)] were significant predictors of LTFU. Distance from the cleft team and rural/urban designation were not statistically significant predictors. Cartographic representation of predicted probability of LTFU revealed vulnerable communities across our state, including in the immediate vicinity of our cleft center. Conclusions: Geostatistical methods are able to identify risk factors missed by traditional statistical analysis. Knowledge of vulnerable populations allow a cleft team to allocate more resources toward high-risk areas to rectify or prevent deficiencies in care.