Psychometric performance of the perinatal bereavement care confidence scale among maternity healthcare workers in Lagos, Nigeria: A cross‐sectional evaluation

OBJECTIVE: This study evaluates the internal structure, internal consistency, and score distributions of the 43-item Perinatal Bereavement Care Confidence Scale (PBCCS) among nurses and midwives working in public maternity services in Lagos, Nigeria. METHODS: This secondary psychometric analysis used cross-sectional BEACON data from 215 participants. Six negatively phrased items were reverse-scored. Internal consistency, ordinal reliability, item performance, floor/ceiling effects, and exploratory construct validity were examined. Structural validity was assessed primarily by polychoric weighted least squares mean and variance adjusted (WLSMV) confirmatory factor analysis (CFA) of the published nine-subdomain model, with one-factor, four-factor, and bifactor alternatives; scale-specific exploratory factor analysis (EFAs) were secondary diagnostics. RESULTS: Seven subdomains had Cronbach alpha values of 0.774-0.938; Functional Skills was lower, at 0.672 and Environmental/Workload lowest, at 0.591. The published nine-factor CFA fitted below conventional benchmarks (robust comparative fit index [CFI] = 0.869, Tucker-Lewis index = 0.856, root mean square error of approximation [RMSEA] = 0.097, 90% confidence interval [CI] 0.092-0.101, standardized root mean squared residual = 0.118) but better than one-factor (CFI = 0.564) and four-factor (CFI = 0.763) alternatives. A bifactor model yielded hierarchical omega = 0.78 and explained common variance = 0.53 but fitted worse than the nine-factor model. Exact ceiling effects were 26.0% for Emotional Knowledge and 37.7% for Continuous Education. Environmental/Workload (mean 2.71) and Institutional Support (2.93) were the lowest-scoring subdomains. CONCLUSION: Most PBCCS subdomains showed acceptable or stronger reliability after corrected scoring, but the published structure was only partially supported. The nine-subdomain model preserved meaningful distinctions better than simpler alternatives, while wording effects and Environmental/Workload heterogeneity contributed to misfit. A total score might summarize some common variance, but domain profiles remain necessary for interpretation and service planning.