Predicting Physical Therapy Visits Needed to Achieve Minimal Functional Goals After Arthroscopic Knee Surgery

Abstract
Retrospective, cross-sectional regression modeling. To predict physical therapy visits following arthroscopic knee surgery. The number of physical therapy visits required to achieve a set of specific minimal-level goals (full knee extension, straight leg raise, normalized gait pattern, bicycle pedaling, and independent home exercises) that are related to decreased complication rates has not previously been modeled. A multiple regression model to predict postoperative physical therapy visits was developed using subject demographics and 2 simple clinical measures, degree of straight leg raise lag and total range of motion. All data were collected from 148 patient charts. Model validity was examined by the predicted residual sum of squares technique and a second independent sample of 157 charts. Diagnosis group, surgery group, and range of motion were the significant variables predicting visits in the final model (R2 = 0.384). Results of model validation analyses using predicted residual sum of squares technique (R2 = 0.346) and the second set of data (R2 = 0.282) were satisfactory. Analysis of residuals (difference of observed and predicted visits) showed prediction of the number of physical therapy visits within 3 visits for approximately 75% of the cases in both sets of data. Using the model to predict physical therapy visits following arthroscopic knee surgery was more accurate than using diagnosis alone, except for lateral retinacular release. This study demonstrates how regression models could be used to explain variance in physical therapy visits for a given set of minimal functional goals.

This publication has 47 references indexed in Scilit: