Prediction of Speech-Correcting Surgery in Patients With a Cleft Palate After Primary Palatoplasty: A Logistic Regression Model

Objective: To identify predictors of speech-correcting surgery after primary palatoplasty in patients with cleft palate with or without cleft lip (CP ± L) and to develop pre- and postoperative prediction models. Design: Retrospective cohort study. Setting: Wilhelmina Children's Hospital, Utrecht, the Netherlands. Patients, Participants: A total of 239 patients with CP ± L who underwent primary palatoplasty between 2008 and 2017 and completed standardized speech assessment at age 5. Interventions: Straight-line palatoplasty with intravelar veloplasty (Sommerlad) within the first year of life. Main Outcome Measure(s): Likelihood of speech-correcting surgery after primary palatoplasty. Potential predictors included cleft type, cleft width, age at palatoplasty, associated syndromes, and postoperative complications such as palatal dehiscence and oronasal fistula. Logistic regression models were developed using pre- and postoperative variables. Model performance was assessed by AUROC, calibration, Brier score, and R². Results: Of 239 patients, 49% required speech-correcting surgery. In the preoperative model, cleft width and presence of syndromes were significant predictors, showing moderate discrimination (AUROC: 0.694; 95% CI: 0.620–0.759) and good calibration. Adding oronasal fistula in the postoperative model minimally improved performance (AUROC: 0.697; 95% CI: 0.621–0.764). Conclusions: A clinically applicable model was developed to predict the likelihood of speech-correcting surgery following primary palatoplasty. Wide clefts, the presence of syndromes, and oronasal fistula were identified as key predictors.