Surgeons have always presumed that surgical outcomes vary by surgeon and medical center (stereotypically assuming they have better outcomes and fewer complications than their peers), though this has never been explicitly demonstrated by a high quality study looking at spine surgery outcomes. Some patients also clearly assume this is true given that some are willing to travel great distances and pay large sums in order to undergo surgery by well-known surgeons at respected medical centers. For probably the first time, the Spine Patient Outcomes Research Trial (SPORT) allows for analysis of validated surgical outcomes stratified by medical center. Dr. Desai and his colleagues from Dartmouth compared baseline patient and disease characteristics, surgical techniques, short-term and 4-year outcomes for patients undergoing surgery for spinal stenosis (SpS) or degenerative spondylolithesis (DS) among 13 spine centers across the United States. Not surprisingly, they discovered significant differences in all of these variables across the centers. The rate of listhesis varied significantly, with DS diagnosed in 33% of patients at one center and 63% at another. Not surprisingly, the proportion of patients undergoing decompression alone, uninstrumented fusion, and instrumented fusion also varied significantly. The patients themselves were also very different, with wide ranges observed for BMI, smoking status, education, and the rates of various medical comorbidities. On this background of variation in patient and disease characteristics, the authors reported significant differences in improvement on the SF-36 and ODI among the different centers, even after controlling for baseline differences.
The findings of this study come as no surprise if one considers the differences in patient populations, surgical techniques, and different patient and provider cultures that exist across the United States. While the authors did control for certain patient characteristics that varied across centers and were associated with outcomes, it is clearly not possible to control for all the baseline differences that exist. Some of the differences in outcomes (i.e. operative time, blood loss, hospital stay, etc.) are clearly due to differences in the proportion of DS patients vs. SpS patients and the fact that DS patients typically underwent fusion while the SpS patients did not. Given the differences in outcomes between these patient populations, performing separate analyses for DS and SpS or trying to control for the different proportion of DS patients across centers would have likely decreased some of the differences in outcomes.1 While controlling for more variables may have reduced some of the variation, this study does suggest that the center at which a patient undergoes surgery affects outcomes. Given this known variation, what should be done about it? Sending all patients to the top-performing centers is obviously not feasible. Instead, rather than continuing to work in the same way in our sometimes isolated centers, surgical teams would likely benefit from organized efforts to identify which local practices are associated with better outcomes and then share this knowledge with others. Much of the variation seen in outcome differences among centers is likely due to relatively small details in local practice patterns. For this reason, efforts like the High Value Healthcare Collaborative, a consortium of medical centers which strives to improve outcomes and decrease costs, will likely be fruitful. This study should alert us that it is time to get out of our local comfort zone and work with teams from other centers in order to identify practices which provide higher value healthcare. Until this is done, we are going to continue to see marked unwarranted geographic differences in practice patterns, costs, and outcomes.
Adam Pearson, MD, MS
Associate Web Editor
1. Pearson A, Blood E, Lurie J, et al. Degenerative spondylolisthesis versus spinal stenosis: does a slip matter? Comparison of baseline characteristics and outcomes (SPORT). Spine (Phila Pa 1976) 2010;35:298-305.