Surgeon equipoise as an inclusion criterion for the evaluation of nonoperative versus operative treatment of thoracolumbar spinal injuries

Abstract Background context Prospective studies have failed to demonstrate the superiority of either operative or nonoperative treatment of thoracolumbar fractures. Similar to other surgical fields, research has been limited by the variability in surgical interventions, difficult recruitment, infrequent pathology, and the urgency of interventions. Purpose To outline factors precluding randomized controlled trials in spinal fractures research, and describe a novel methodology that seeks to improve on the design of observational studies. Study design/setting A preliminary report describing an observational study design with clinical equipoise as an inclusion criterion. The proposed methodology is a cohort study with head-to-head comparison of operative and nonoperative treatment regimens in an expertise-based trial fashion. Patients are selected retrospectively by an expert panel and clinical outcomes are assessed to compare competing treatment regimens. Surgeon equipoise served as an inclusion criterion. Patient sample Patients with closed or open thoracolumbar spinal fracture with or without neurological impairment, presenting to one of two different trauma centers between 1991 and 2005 (N=760). Outcome measures Homogeneity of baseline clinical and demographic data and distribution of prognostic risk factors between the operative and the nonoperative cohort. Methods Patients treated for spine fractures at two University hospitals practicing opposing methods of fracture intervention were identified by medical diagnosis code searches (n=760). A panel of spine treatment experts, blinded to the treatment received clinically has assessed each case retrospectively. Patients were included in the study when there was disagreement on the preferred treatment, that is, operative or nonoperative treatment of the injury. Baseline and initial data of a study evaluating nonoperative versus operative spinal fracture treatment are presented. Results One hundred and ninety patients were included in the study accounting for a panel discordance rate of 29%. The distribution of baseline characteristics and demographics of the study populations were equal across the parallel cohorts enrolled in the study, that is, no differences in prognostic factors were observed. Conclusions The use of clinical equipoise as an inclusion criterion in comparative studies may be used to avoid selection bias. Using multivariate analysis of retrospectively assembled parallel cohorts, a valid compa