Decision trees for indication of total hip replacement on patients with osteoarthritis

Objective. To develop a decision tree based on health-related quality of life outcomes rather than expert consensus for determining the appropriateness of total hip replacement (THR) among patients with hip OA. Methods. This is a prospective observational study of two independent cohorts. The derivation cohort included 590 patients recruited from seven hospitals between March 1999 and March 2000. The validation cohort included 339 patients recruited from six hospitals between September 2003 and September 2004. Socio-demographic and clinical data were collected for the participants, all of whom completed the WOMAC before hip replacement and 6 months later. Univariate and Regression Trees, by classification and regression trees (CART), analyses were performed in the derivation cohort. The decision trees derived in the derivation cohort were validated in the validation cohort. Results. Main variables that predicted change in the WOMAC pain and functional limitation domains were pre-intervention pain or functional limitation and the application of non-surgical treatments. CART analysis showed that when pre-intervention pain was classified as minor, or WOMAC pain or functional limitation scores were <= 40, there was an odds ratio of 0.076 (95% Cl 0.031, 0.185) of having an expected gain after THR in the WOMAC pain domain of >30 or >25 in the WOMAC functional limitation domain. Conclusions. A simple decision tree based on WOMAC outcomes can help to determine the appropriate application of THR. It could also be used to evaluate clinical practice or for quality control.