Journal of the American Academy of Orthopaedic Surgeons October 2008, Volume 16, Issue 10 Paul A. Anderson, MD, Paul C. McCormick, MD, MPH and Peter D. Angevine, MD, MPH
Randomized controlled trials are considered to provide the strongest data regarding the relative benefits of treatment alternatives for medical conditions. Uncertainty persists regarding the optimal treatment of patients with symptomatic lumbar disk herniation. Five randomized controlled trials were published between 1983 and 2007 that compared lumbar diskectomy with nonsurgical treatment. The studies enrolled more than 1,000 patients. Inclusion and exclusion criteria were generally similar, but there was substantial variation in the outcomes measurements used. In all studies, more than one third of patients assigned to nonsurgical care crossed over to have surgery. Crossover in the opposite direction ranged from 0% to almost 40%. As a result of the large number of crossovers, the estimated treatment effect size of diskectomy likely is underestimated. Valid inferences about the safety and effectiveness of continued nonsurgical care cannot be made. The use of frequentist statistical techniques threatens the validity of post hoc subgroup analysis. Large cohort studies and alternative statistical techniques may yield more accurate estimates of the effectiveness of lumbar diskectomy and aid in identifying patients who may benefit from early surgical intervention.