The Spine Blog

Friday, January 17, 2014

The Challenge of Performing High Impact Research

The Evidence-Based Medicine for Spine Surgery group published their latest biannual article in the January 1 issue, accompanied by an editorial discussing the challenges of creating high impact literature that answers clinically important questions in spine surgery. The group reviewed six recent potentially “high impact” papers on a wide range of topics, from the timing of surgery for thoracolumbar fractures to the cost-effectiveness of surgery for cervical spondylotic myelopathy. The first article is a systematic review comparing early (< 72 hours) and late surgery for thoracolumbar fractures. Not surprisingly, the early group had shorter ICU and hospital stays and lower morbidity, though the quality of the included papers was low, limiting the strength of the conclusions. The 2nd article compared retrograde ejaculation (RE) rates between patients undergoing ALIF with rhBMP-2 and those undergoing ALIF with autograft or total disk replacement by retrospectively reviewing 5 trials. While the authors of this industry sponsored study reported an RE rate twice as high in the BMP patients (3.4% vs. 1.7%), this difference failed to reach statistical significance. Given the low rate of RE, this study was likely underpowered to answer the question at hand, and patients were also not specifically queried about this complication, so many cases were likely missed. The 3rd article was a case series of spine trauma patients, and the authors evaluated risk factors for infection. While this study included patients from 13 centers, only three months’ worth of cases were included, totaling 169 patients, of whom 6 had surgical site infections. While the authors attempted to look at risk factors for infection, with numbers this low no definitive conclusions could be made. The 4th study used a Korean administrative database to determine the rate of re-operation following surgery for disk herniation. Patients undergoing open discectomy, endoscopic discectomy, laminectomy, fusion, and nucleolysis were included. Unfortunately, the information in the database did not include the level or indication for subsequent surgery. As such, few conclusions could be drawn. The 5th study was a cost-utility analysis based on the AOSpine North America cervical spondylotic myelopathy study. Unfortunately, costs and utilities were not compared among treatment groups (i.e. surgical vs. non-operative, anterior vs. posterior, etc.), precluding calculation of an incremental cost effectiveness ratio (ICER), the most relevant outcome of such a study. The final study was a secondary analysis of the SPORT disk herniation data attempting to determine the effect of lumbar epidural steroid injection (LESI) on outcomes. Those treated with LESI had increased crossover from surgery to non-operative treatment, though LESI did not improve short or long-term outcomes. Unfortunately, the purely observational nature of this study with no specific protocol for LESI and the high level of confounding make it hard to draw strong conclusions about the effect of LESI for disk herniation patients.


The evidence-based medicine editorial board makes a strong effort to find high impact articles for their reviews, and all of the studies included had the potential to be high impact. Unfortunately, methodological issues limited the strength of the conclusions that could be made based on the reported data. Insufficient power, lack of randomization, inappropriate inclusion and exclusion criteria, inappropriate outcome measures, and lack of a comparison group are all common problems which plagued the studies included in this review. While these studies were performed by experienced, well-meaning investigators, the nature of the questions they were attempting to answer made designing a study that avoids these issues challenging. Randomizing patients to early vs. late surgery for thoracolumbar fractures is unethical. It is difficult to enroll a sufficient of patients to do an analysis on relatively rare complications such as infection or retrograde ejaculation. Administrative database studies are inherently limited by the data they include. Studying interventions such as LESI is challenging as patients are free to crossover to other treatments if the injection does not work. The only solution to most of these problems is to design large, prospective, multicenter trials with appropriate outcome measures. While some questions are suited to randomized studies, others—such as myelopathy and timing of fracture surgery—can only be ethically studied using observational designs. Unfortunately, this type of research is expensive, requires high levels of cooperation among investigators at different institutions, and is extremely work intensive. Hopefully the spine surgery community will continue to put forth the effort to perform high quality research such that “high-impact” literature will be a reality.


Please read the latest installment of “Evidence-based recommendations for spine surgery” in the January 1 issue and the accompanying editorial. What do you think about the challenges inherent in studying important questions in spine surgery? Let us know by leaving a comment on The Spine Blog.

Adam Pearson, MD, MS

Associate Web Editor