Background: Sex estimation represents a pivotal element of forensic anthropological investigation, conventionally dependent on highly dimorphic skeletal components such as the pelvis and skull. The purpose of the current study was to systematically evaluate the diagnostic accuracy of sternal measurements for sex estimation and to identify methodological- or population-based moderators that influence classification performance. Methods: A systematic review and meta-analysis were conducted in accordance with PRISMA 2020 guidelines. R programming software was used to perform statistical meta-analysis. Pooled sensitivity, specificity, diagnostic odds ratio (DOR), likelihood ratios (LR±), and overall accuracy were calculated using random-effects meta-analysis. Subgroup analyses and meta-regression were performed based on population origin, study design, statistical approach, and measurement protocol. Results: Forty-one studies comprising 293 predictive models were included. The overall pooled sensitivity and specificity were 80.9% (95% CI: 79.7–82.1) and 74.0% (95% CI: 72.4–75.5), respectively, with a mean accuracy of 77.3%. Subgroup analysis revealed that studies involving African populations and imaging-based methods achieved the highest accuracy. Machine learning- and ROC-based methods outperformed traditional discriminant analysis. Combined sternal measurements (manubrium and body) yielded the most robust diagnostic performance (accuracy: 87.3%). Significant heterogeneity (I2 > 85%) was observed. Conclusions: Sternal morphometry exhibits a moderate to high degree of diagnostic accuracy in sex estimation and possesses significant forensic importance, especially in situations where more sexually dimorphic features are inaccessible. Nonetheless, variations across populations, the absence of standardized protocols, and methodological heterogeneity constrain its universal applicability.
Triantafyllou et al. (Wed,) studied this question.