Adult Age Threshold Estimation Using Radiographic Evaluation of Wrist–Hand Skeletal Maturation: A Systematic Review and Meta-Analysis

Background: Wrist-hand radiographic skeletal maturation methods are widely used for forensic age estimation of living individuals, remaining among the most widely accepted imaging modalities in forensic practice despite limited evidence supporting their use at the legally relevant 18-year threshold. This systematic review aims to provide a quantitative synthesis of the widely recognized but previously unsynthesized body of evidence regarding their diagnostic accuracy (sensitivity, specificity, AUC-curve, mean absolute error-MAE) for forensic adult age estimation. Methods: Six databases (Scopus, PubMed, Google Scholar, Embase, Cochrane, Clarivate) between 1980 and 2026 were searched. Studies evaluating wrist-hand skeletal maturation for forensic age estimation in individuals ≥16 years with verified chronological age were included. Two reviewers independently screened studies and extracted data. Quality was assessed using QUADAS-2. DerSimonian-Laird random-effects model estimated the sensitivity and specificity of the wrist-hand bones maturation at the age of 18 years, and pooled MAE. GRADE assessed the certainty of evidence. Results: From 747 records, 23 studies (11,425 participants, 15 countries, 2003–2025) were included. Pooled MAE was 0. 537 years (95% CI: 0.387–0.686; 95% PI: 0.33–0.79 years), but obtained from 3 Italian studies conducted on clinical populations, and all characterized by high risk of bias. At the 18-year threshold, pooled sensitivity was 69.5% (95% CI: 61.6–77.3%; 95% PI: 35.3–100%) and specificity 85.56% (95% CI: 83.53–87.60%; 95% PI: 77.2–93.9%). Nevertheless, the heterogeneity was extreme (I2 > 98%) for all diagnostic measures, and calculated prediction intervals confirm that individual study results are expected to vary widely. QUADAS-2 assessment revealed 82.6% of studies with high/unclear patient selection bias. GRADE-certainty was VERY LOW for both outcomes due to serious risk of bias, very serious inconsistency, and serious indirectness. Conclusions: Wrist-hand skeletal maturation shows low accuracy and high rates of misclassification for age estimation at the 18-year threshold. The wide prediction intervals (sensitivity 35–100%, specificity 77–94%, MAE 0.33–0.79 years) indicate that performance in a new study population may differ substantially from the pooled estimates. Very low certainty evidence, extreme heterogeneity, and substantial methodological limitations preclude confident application. Results should be interpreted with caution in forensic contexts, preferably combining multiple age indicators, in particular dental evidence.