Résumé : | Background
Worldwide, the incidence of wrist fracture is increasing. There are currently no externally validated prognostic models to inform early decision making for these patients.
Objectives
To develop and validate a prognostic model from a comprehensive range of candidate prognostic factors that can identify patients who are at risk of developing persistent pain following wrist or hand fracture.
Methods
We developed and validated a prognostic model using secondary data derived from a prospective cohort study (n = 715), with recruitment sites in 3 metropolitan hospitals in Sydney, Australia. The primary outcome was persistent pain 4 months following the injury. The current study used a backward stepwise regression analysis to develop the model in 2 hospitals (n = 408) and externally validate it in a third hospital (n = 307). To determine the accuracy of the model, we assessed calibration and discrimination in accordance with the PROGnosis RESearch Strategy framework.
Results
Complete data were available for 95% of the cohort. Of 14 candidate variables, the final model contained 2 prognostic factors: patient age and pain intensity reported at initial presentation. The area under the receiver operating characteristic curve was 0.63 (95% confidence interval: 0.56, 0.69) in the development sample and 0.61 (95% confidence interval: 0.51, 0.70) in the validation sample. The model systematically overestimated risk (intercept, −1.13; slope, 0.73).
Conclusion
We developed and externally validated a prognostic model to predict persistent pain 4 months after a wrist or hand fracture. Future studies are needed to assess whether the accuracy of this model can be improved by updating and validating it in local settings.
Level of Evidence Prognosis, level 1b |