Decision analysis as an aid to determining the management of early low rectal cancer for the individual patient

Purpose: Because there are no data available from randomized controlled tri als (RCT), a decision analysis was performed to aid in the decision of whic h option, a local excision with or without radiotherapy or an abdominal per ineal resection (APR), should be offered to medically fit patients with ear ly (suspected T1/T2) low (< 5 cm) rectal cancer. Methods: All clinically relevant outcomes, including complications of surge ry and radiotherapy, cure, salvageability after local recurrence, distant d isease, and death, were modeled for both options, The probabilities of comp lications and outcomes after radiotherapy and/or local excision were derive d from weighted averages of results from studies conducted between 1969 and 1997, The probabilities for the APR option were extracted from relevant RC Ts, Long- and short-term patient-centered utilities for each complication a nd outcome were extracted from the literature and from expert opinion, Results: The expected utility of local excision (EU = 0.81) for the base ca re was higher than the expected utility for APR (EU = 0.78). Although the r esult was sensitive to all variables, local excision was always favored ove r APR within the plausible ranger of the variables taken one, two, or three at a time. The model illustrated the tension between the patient's percept ion of a colostomy and the higher recurrence rates with local excision. Conclusion: The results of this decision analysis suggest that local therap y for early low rectal cancer is the preferred method of treatment. However , there must be careful preoperative assessment, patient selection, and con sideration of patient concerns. In addition, decision analysis may be usefu l in providing patient information and assisting in decision making. J Clin Oncol 17:312-318. (C) 1999 by American Society of Clinical Oncology.