Prospective decision analysis for peripheral vascular disease predicts future quality of life.

OBJECTIVE: Decision making for peripheral vascular disease can be quite complex as a result of pre-existing compromise of patient functional status, anatomic considerations, uncertainty of favorable outcome, medical comorbidities, and limitations in life expectancy. The ability of prospective decision-analysis models to predict individual quality of life in patients with lower extremity arterial occlusive disease was tested.
METHODS: This was a prospective cohort study. The settings were university and Veterans Administration vascular surgery practices. All 214 patients referred with symptomatic lower extremity arterial disease of any severity over a 2-year period were screened, and 206 were enrolled. A Markov model was compared with standard clinical decision-making. Utility assessment and generalized (Short Form-36; SF-36) and disease-specific (Walking Impairment Questionnaire; WIQ) quality of life were derived before treatment. Estimates of treatment outcome probabilities and intended and actual treatment plans were provided by attending vascular surgeons. The main outcome measures were generalized (SF-36) and disease-specific (WIQ) variables at study entry and at 4 and 12 months.
RESULTS: Primary intervention consisted of amputation for 9, bypass for 42, angioplasty for 8, and medical treatment for 147 patients. Considering all patients, no improvement in mean overall patient quality of life measured by the SF-36 Physical Component Score (27 +/- 8 vs 28 +/- 8; P = .87) or WIQ (39 +/- 22 vs 39 +/- 23; P = .13) was noted 12 months after counseling and treatment by the vascular surgeons. Individually considered SF-36 categories were improved only for Bodily Pain (40 +/- 23 vs 49 +/- 25; P = .03), with the most significant improvement observed among patients with the most severe pain (68 +/- 25 vs 37 +/- 23; P = .02) and among those undergoing bypass (60 +/- 29 vs 31 +/- 22; P = .02). It is noteworthy that when the treatment chosen was incongruent with the Markov model, patients were more likely to report a poorer quality of life at 1 year (Physical Component Score, 25 +/- 8 vs 29 +/- 8; P < .001). The quality of life predicted at baseline by the Markov model correlated positively with the Physical Component Score (r = 0.23), Bodily Pain (r = 0.33), and Fatigue (r = 0.44) and negatively with WIQ (r = -0.08) observed 1 year later.
CONCLUSIONS: Prospective application of an individualized decision Markov model in patients with vascular disease was predictive of patient quality of life at 1 year. The patient's outcome was worse when the treatment received did not follow the model's recommendation. This decision analysis model may be useful to identify patients at risk for poor outcomes with standard clinical decision making.