A CLASSIFICATION TREE ANALYSIS OF SELECTION FOR DISCRETIONARY TREATMENT

OBJECTIVES. TO study treatment bias in observational outcomes research , the authors present a nonlinear classification tree model of clinica l and psychosocial factors influencing selection for interventional ma nagement (lower extremity bypass surgery or angioplasty) for patients with intermittent claudication. METHODS. The study sample includes 532 patients with mild to moderate lower extremity vascular disease, with out prior peripheral revascularization procedures or symptoms of disea se progression. All patients were enrolled in a prospective outcomes s tudy at the time of an initial referral visit for claudication to one of the 16 Chicago-area vascular surgery offices or clinics in 1993-95. The influence of baseline sociodemographic, clinical, and patient sel f-reported health status data on subsequent treatment is analyzed. Stu dy variables were derived from lower extremity blood flow records and patient questionnaires. Follow-up home health visits were used to asce rtain the frequency of lower extremity revasculariztion procedures wit hin 6 months of study enrollment. Hierarchically optimal classificatio n tree analysis (CTA) was used to obtain a nonlinear model of treatmen t selection. The model retains attributes with the highest sensitivity at each node based on cutpoints that maximize classification accuracy . Experimentwise Type I error is ensured at P < 0.05 by the Bonferroni method and jackknife validity analysis is used to assess model stabil ity. RESULTS. Seventy-one of 532 patients (13.3%) underwent interventi onal procedures within 6 months; Ten patient attributes were used in t he CTA model, which had an overall classification accuracy of 89.5% (6 7.6% sensitive and 92.9% specific), achieving 57.7% of the theoretical possible improvement in classification accuracy beyond chance. Eleven model prediction endpoints reflected a 33-fold difference in odds of undergoing lower extremity revasculariztion. CONCLUSIONS. Initial ankl e-brachial index (100%), leg symptom status over the previous six mont hs (89%), self-reported community walking distance (74%) and prior wil lingness to undergo a lower extremity hospital procedure (39%) were us ed to classify most patients in the sample. These attributes are criti cal control variables for a valid observational study of treatment eff ectiveness.