Do expectations of recovery improve risk assessment for people with whiplash-associated disorders? Secondary analysis of a prospective cohort study

Background WhipPredict, which includes prognostic factors of pain-related disability, age and hyperarousal symptoms, was developed and validated for prediction of outcome in people with whiplash associated disorders (WAD). Patient expectations of recovery was not an included factor, though is known to mediate outcomes. The aim of this study was to determine whether the addition of expectations of recovery could improve the accuracy of WhipPredict. Methods Two hundred twenty-eight participants with acute WAD completed questionnaires (WhipPredict and expectations of recovery) at baseline. Health outcomes (neck disability index (NDI) and Global Perceived Recovery (GPR)) were assessed at 6- and 12-months post injury. Cut-off points for expectations of recovery predictive of both full recovery (NDI <= 10 % , GPR >= 4) and poor outcome (NDI >= 30 % , GPR <= - 3) were determined, and multivariate logistic regression analyses were used to compare models with and without this variable. Results Expectations of recovery improved or maintained the accuracy of predictions of poor outcome (6-months: sensitivity 78 to 83%, specificity maintained at 79.5%; 12-months: sensitivity maintained at 80%, specificity 69 to 73%). The sensitivity of predictions of full recovery improved (6-months: 68 to 76%; 12-months: 57 to 81%), though specificity did not change appreciably at 6 months (80 to 81%) and declined at 12 (83 to 76%). ROC curves indicated a larger and more consistent improvement in model performance when expectations of recovery were added to the pathway predictive of full recovery. Conclusions The addition of expectations of recovery may improve the accuracy of WhipPredict, though further validation is required.