The Effects of Including Quality of Life Responses in Models to Predict Walking Performance of Patients with Intermittent Claudication

ObjectivesTo examine the accuracy of previously developed prediction models of treadmill walking performance in patients with intermittent claudication (IC) due to peripheral arterial disease (PAD); and to examine the accuracy of new prediction models.DesignAnalysis of data collected in a previous randomised clinical trial.MaterialsNinety-three assessments of 28 patients diagnosed with IC due to PAD.MethodsPatients undertook routine clinical assessments, quality of life (QOL) questionnaires and treadmill walking tests. Walking performance and estimates based on prediction models were compared via paired t-tests or Wilcoxon Rank Sum tests. Stepwise linear regression generated models to predict walking performance from clinical measures and QOL responses. Accuracy was determined as the absolute error between model estimates and patient results.ResultsWalking performance was significantly underestimated (35–71% error) by previously developed prediction models. Models developed in the current study identified QOL responses as the most significant predictors of current walking performance but these models still resulted in substantial errors (19–84%).ConclusionsPreviously published predictors of walking performance significantly underestimated patient’s ability in practise. Predictions based upon clinical measurements and QOL responses were developed however, their accuracy was also limited. Further research is needed regarding walking performance prediction to assist clinicians with assessment of PAD severity and treatment effectiveness.