For the better part of history, scientific literature was produced with a singular intent: communication of the results of research experiments for the purposes of edification, replication, and debate. The scientific method (formulation of hypotheses, experimental and measurement-based testing, objective assessment, etc.) represented the cornerstone of any research effort, with the randomized controlled experiment considered the sine qua non. The adequate performance of a randomized controlled trial (RCT) necessitates adherence to several important principles rooted in the scientific method, including a priori declaration of hypotheses, primary outcomes, and time points for evaluation; calculation of the number of patients necessary to detect meaningful differences; and an analytic plan to address protocol nonadherence (when patients assigned to a particular treatment fail to receive the intervention, or choose to be treated with a different intervention). Not only are these guideposts critical to the researchers performing an RCT, they should also be equally vital to the clinician-reader, who is left to apply the findings in clinical practice.
In the last 2 decades, several large RCTs funded by the U.S. National Institutes of Health (NIH) have appeared in the spine surgical literature, including the Spine Patient Outcomes Research Trial (SPORT) studies1,2, and the work of Kallmes et al. regarding the use of vertebroplasty for insufficiency fractures3. To this compendium, we can now add the Adult Symptomatic Lumbar Scoliosis (ASLS)-1 trial, the results of which are presented in this issue of The Journal. This NIH-supported study was conducted at 9 North American centers, with enrollment over the course of 4 years (April 2010 to July 2014), and involved parallel enrollment of adults with lumbar scoliosis into randomized and observational study arms. The authors intended to compare functional outcomes following operative and nonoperative treatment for lumbar scoliosis at 2 years following the intervention, using an intention-to-treat analysis, and estimated a need for 82 patients in the randomized cohort to adequately power the study. However, the authors were unable to achieve the level of enrollment deemed necessary in their power calculation and also experienced an extensive degree of nonadherence, particularly within the group of patients assigned to nonoperative management; the authors reported that, in the first 2 years of the investigation, 21 (64%) of 33 individuals assigned to nonoperative treatment nonetheless elected to undergo surgery. The authors responded to this by emphasizing the results of the as-treated and observational cohort analyses, couching their findings in the context of what they term “interpretation of this study as pragmatic in approach….” They also note that an interim power analysis, performed using 1-year outcomes available for 26 patients, revealed that only 36 total randomized patients would be required to detect meaningful differences. After appropriately recognizing these shortcomings, the authors proposed that, if a patient with lumbar scoliosis is dissatisfied with current spinal health and function, surgical intervention is the preferred approach.
While RCTs are lionized in a number of respects, they remain experiments and are prone to many of the same pitfalls as other types of prospective research, including bias introduced through study design, patient nonadherence, and the analytic plan2,4. The main advantage of RCTs lies in the random allocation of treatment, which mitigates the influence of selection and indication bias on outcomes. The greater the extent of nonadherence, especially when 1-sided, as is always the case in surgical research (i.e., a patient who is initially treated nonoperatively can always choose surgery at a later date, but a patient who receives an operation can never go back to reverse the surgical event and be considered a nonoperative patient), the more the advantages conferred by randomization are lost2. The authors are certainly within their right to orient our attention to the pragmatic stance they prefer when contextualizing their findings. I believe that the most pragmatic approach to applying these results to clinical practice, however, requires falling back on the tenets of the scientific method. To do otherwise, whether using a pragmatic or other interpretive rubric, seems like creating “moving goal posts” and engendering a slippery slope when it comes to applying research we like, or intuitively agree with, in a clinical setting.
In the end, the actionable message from this trial stems from the as-treated analysis and the observational arm of the study and not the randomized component. Thus, the findings are prone to confounding by selection and indication on the part of the surgeon-researchers as well as the patients. This is why the level of evidence applied to this study is II as opposed to I. Furthermore, the authors’ stance seems to keep concerns regarding adverse events on the periphery, including a 10% rate of neurological deficits and a revision rate of 14%, not to mention the personal and societal costs associated with these procedures.
The authors must be applauded for bringing an extensive and logistically challenging investigation to completion after what amounts to >8 years of ongoing effort. I also agree that the information presented can be viewed as best-available evidence regarding the natural history of nonoperative and operative management in the setting of adult lumbar scoliosis and the impact that these treatments can have on functional status, pain levels, and the performance of normal daily activities. The importance of this report is further enhanced by the exceedingly low rates of patients lost to follow-up across both arms. I believe that the most pragmatic application of these findings consists of their inclusion in pretreatment discussions regarding a strategy that aligns best with patients’ goals, desire for functional improvement, and willingness to accept the risks of deformity-correction surgery. Given the limitations associated with this study and its reliance on nonrandomized observational data, however, the claim that surgical intervention should be recognized as the superior treatment option for adult lumbar scoliosis due to results of the as-treated analysis alone cannot be completely substantiated.
1. Weinstein JN, Tosteson TD, Lurie JD, Tosteson AN, Hanscom B, Skinner JS, Abdu WA, Hilibrand AS, Boden SD, Deyo RA. Surgical vs nonoperative treatment for lumbar disk herniation: the Spine Patient Outcomes Research Trial (SPORT): a randomized trial. JAMA. 2006 Nov 22;296(20):2441-50.
2. Anderson PA, McCormick PC, Angevine PD. Randomized controlled trials of the treatment of lumbar disk herniation: 1983-2007. J Am Acad Orthop Surg. 2008 Oct;16(10):566-73.
3. Kallmes DF, Comstock BA, Heagerty PJ, Turner JA, Wilson DJ, Diamond TH, Edwards R, Gray LA, Stout L, Owen S, Hollingworth W, Ghdoke B, Annesley-Williams DJ, Ralston SH, Jarvik JG. A randomized trial of vertebroplasty for osteoporotic spinal fractures. N Engl J Med. 2009 Aug 6;361(6):569-79.
4. Bono CM. Commentary: “Why oh why is kyphoplasty use on the rise?” said the little vertebroplasty needle. Spine J. 2011 Aug;11(8):745-6.