Multistage analysis of quality in pit-and-fissure sealant programs for first permanent molars: a real-world study

Abstract

This study examined the associations of operator routine procedural exposure and children’s residential setting (urban vs. rural) with quality indicators across three stages—diagnostic attribution, operational execution, and long-term retention—in a pit-and-fissure sealant program. A single-center, retrospective observational study was conducted using electronic medical records and follow-up data from September 2024 to March 2026. Children aged 6–13 years were included. Operators were classified as having higher or lower routine procedural exposure based on their routine exposure to pediatric preventive bonding procedures. Stage I assessed indication burden and recorded reasons for non-sealing using child-level and clustering-aware models, as appropriate. Stage II evaluated sealant application among indicated teeth, sealant coverage among examined teeth, and mean sealed teeth per child using binomial GEE and HC3-robust factorial models, as appropriate. Operator-level clustering and one-tooth-per-child analyses were performed as sensitivity analyses. Stage III used a nested follow-up cohort (≥ 2 visits) to estimate sealant survival using Kaplan–Meier methods. Associations of residential setting and operator routine procedural exposure with sealant failure were evaluated using a Cox model stratified by age and sex and adjusted for tooth position, with robust standard errors clustered at the child level. The proportional hazards assumption was assessed using scaled Schoenfeld residuals. The cohort included 7,692 children. A total of 20,922 first permanent molars were sealed (2.72 ± 1.48 per child). Rural children had a slightly greater mean number of indicated teeth per child, whereas the residential difference in tooth-level indication status was borderline after accounting for within-child clustering ( P = 0.053). Non-sealing reason distributions varied according to residential setting and operator procedural exposure in clustering-aware analyses, although category-specific associations were not uniformly robust. For sealant application, operator exposure was associated with the outcome in the primary child-clustered model ( P = 0.002), whereas the residence-by-exposure interaction was not significant ( P = 0.282). Sealant coverage was associated with residence ( P = 0.002), with no evidence of interaction ( P = 0.833). Operator-related associations were less consistent in operator-clustered sensitivity analyses. In the retention cohort (489 children; 1,249 teeth), median sealant survival was 249 days (95% CI 237–261). The initial Cox model showed non-proportional hazards for age and sex. In the final Cox model stratified by age and sex, with residential setting, routine procedural exposure, and tooth position included as covariates, neither higher routine procedural exposure (aHR = 0.96, 95% CI 0.75–1.21; P = 0.706) nor urban residence (aHR = 0.82, 95% CI 0.42–1.62; P = 0.576) was associated with sealant failure. Program quality exhibited stage-specific patterns. Diagnostic attribution varied across residential and operator-exposure groups, whereas execution-stage analyses showed comparatively robust residential differences in sealant coverage but less consistent operator-related associations across clustering specifications. In the retention stage, neither operator routine procedural exposure nor residential setting showed evidence of an association with sealant failure after stratification by age and sex and adjustment for tooth position. Because age and sex were handled as stratification variables, their associations could not be represented by single time-constant hazard ratios. These findings support stage-specific quality assessment while underscoring the need for cautious interpretation of operator-related associations.

The authors' abstract, as published at the source. BMC Oral Health, 2026 · DOI ↗