Background and Objectives Prognostic tools are widely used to guide early management of musculoskeletal (MSK) injuries. However, social identity factors such as age, gender, race, and socioeconomic status may influence symptom experience and reporting, potentially affecting the accuracy of risk classification. The purpose of this study is to identify those social identity variables associated with scores on common prognostic tools or their prognostic accuracy. Methods We analyzed data from a cohort of workers (n = 203) with acute musculoskeletal injuries. Participants completed the Numeric Pain Rating Scale (NPRS), Traumatic Injuries Distress Scale (TIDS), EuroQoL (EQ5D-5L), and a demographic survey including age, sex, race, income, education, and perceived discrimination. Differences in baseline scores were assessed between identity groups. Recovery was dichotomized (fully recovered/not fully recovered) using 3 different recovery indicators. Predictive accuracy for recovery was evaluated using area under the receiver operating characteristic curve, sample-wide and when stratified by social identity variables. Results Older participants reported significantly higher pain (5.5 vs. 4.3/10, p < 0.01) and those indicating more experiences of discrimination rated higher post-trauma distress (11.1 vs. 9.2/24, p < 0.05). 8-week recovery rates were 21.7% to 54.7%. Lower education predicted <full recovery in administrative data only. The TIDS and EQ5D-5L were significantly better at predicting recovery compared to the NPRS. When disaggregated by social identity, the TIDS functioned significantly better in females than males, while the EQ5D-5L functioned similarly across all social categories. Conclusion The results indicate that different risk/prognosis cut-scores, and even different tools, may be required for people with different intersectional identities. The results should be interpreted in light of some identities being reduced to broad categories. Most risk/prognosis tools in MSK recovery research use a single universal cut-score to distinguish low from high risk, but this study only partially supports that approach.
Bayattork et al. (2026) studied this question.