To our knowledge, this is the first large-scale study to examine the association between SMT and spinal reoperation. The study findings support our hypothesis that adults with LSR receiving SMT at least one year after lumbar discectomy are less likely to undergo lumbar spine reoperation compared to matched controls not receiving SMT. A cumulative incidence graph demonstrated that surgeries tended to occur earlier in the two-year follow-up window in both cohorts.
In addition, it remains unclear whether mechanisms of SMT on pain or disability alone would account for our findings. It is possible that nonspecific therapeutic factors related to education and advice, the clinician-patient relationship, continuity of care, patient expectations, or avoidance of surgeon visits in the SMT cohort accounted for the observed reduction in lumbar spine reoperation [
46,
47]. With these considerations, follow-up studies might include other comparator cohorts such as physical therapy exercises, acupuncture, or psychological therapies to broadly explore the potential impact of conservative care on reoperation. In addition, a qualitative study could explore patients’ reasons for pursuing or avoiding lumbar spine reoperation.
Our findings are consistent with prior research suggesting that patients receiving SMT for a new diagnosis of LSR are less likely to undergo discectomy [
48]. However, our study population differed in that patients had undergone a discectomy previously. Nonetheless, the underlying rationale for this finding may be similar, as SMT has demonstrated efficacy in reducing pain and disability associated with LSR in two clinical trials [
49,
50]. We therefore suspect that SMT could reduce the likelihood of reoperation after discectomy by alleviating symptoms of LSR.
While our study did not directly measure costs, the observed decrease in lumbar spine reoperations in the SMT cohort may translate to potential cost savings. Lumbar spine reoperations incur substantial costs (2017: $11,161 per patient; approximately $13,750 in 2023, per the US Bureau of Labor Statistics) [
1]. In contrast, the costs of a single chiropractic visit are relatively lower (2023: $65 [
51]) yet would be additive over the duration of care. Future investigations using administrative claims data could be used to comprehensively compare total costs, including those incurred from adjunctive therapies, modalities, and medications over the duration of care, which would be missed in our current simplified model. This would provide an improved understanding of whether SMT offers cost-effective benefits in a population at heightened risk of substantial reoperation costs.
SMT is generally considered safe, with severe adverse events occurring at an incidence ranging from less than one to seven in 100,000 treatments [
52,
53]. However, there is a lack of studies examining the likelihood of SMT-related adverse events among patients with prior spine surgery [
12,
54]. Although our study did not focus on identifying adverse events, our findings of reduced lumbar spine reoperation in adults receiving SMT serve as a potential marker of safety of this treatment, suggesting that SMT may be unlikely to contribute to recurrent disc herniation requiring reoperation. However, our findings do not inform clinicians if or when it is safe to administer SMT in patients with previous discectomy, due to included patients being uniformly at least one-year post-operation, lack of a direct measure for adverse events, and lack of detailed patient-level data. Further study is needed to examine potential adverse events more rigorously among patients with prior spine surgery receiving SMT.
While our findings are promising, it is important to emphasize the necessity for well-designed prospective studies to further investigate this topic. Our study offers the advantage of a long-term follow-up and a relatively large population; however, due to the de-identified nature of the dataset, we were unable to assess patient-reported outcome measures. To gain insights into factors that mediate the risk of lumbar spine reoperation, it is necessary to conduct a longitudinal prospective study incorporating standardized assessments of pain, low back-related disability, quality of life, and pain medication utilization [
55,
56].
Strengths and limitations
Strengths of this study included having a large propensity matched population, multidisciplinary author team, and registered protocol. Our study findings provide markers of validity in alignment with prior research. For instance, the incidence of reoperation in our usual care cohort was approximately 14%, a value slightly higher than prior estimates (up to 12% within 4 years post-discectomy [
1‐
4]), which is justifiable given we only included patients with LSR. Furthermore, our cumulative incidence graph revealed an initial curvilinear increase in reoperation, suggesting that we succeeded in including patients with a heightened likelihood of reoperation [
6,
24]. Finally, our study also highlights patients’ adherence to care in the SMT cohort, as evidenced by most patients having multiple SMT visits.
This study also has limitations. As in all observational studies, we are unable to infer causality from our findings. Between-cohort differences in unmeasured confounding variables may have influenced results, such as socioeconomic variables (e.g., unemployment [
57]), patient-reported outcome measures of pain, disability or quality of life [
31], length of preoperative symptoms before the primary discectomy [
58], type of disc herniation (e.g., extrusion, sequestration; [
38]), presence of adjacent segment degeneration [
24], and lumbar spine range of motion [
38], which were largely unavailable in the TriNetX dataset. It is also possible that variation in geographic region, surgeon specialty, or surgeon experience would influence the likelihood of reoperation [
4]. Although we were unable to ascertain the exact type of discectomy per patient, one recent study found no difference in likelihood of reoperation among patients receiving open versus percutaneous lumbar discectomy [
2]. As our dataset did not provide detailed patient-level data, we were unable to determine the length of time since the primary discectomy, although it was at least one year for all patients per our selection criteria. A 20-year data range was necessary to maximize sample size and allow for a lengthy follow-up window yet may have affected our findings considering changes in lumbar spine surgery practices over this period. The natural language processing tool we used has not been validated for the specific purposes of our study methods and it’s performance in this context is uncertain.
As a real-world study, patients in both cohorts used several types of pharmacologic and/or non-pharmacologic interventions prior to the index date of meeting our selection criteria, yet we were unable to examine their response to these forms of care. While it remains possible that the type of response to previous interventions influenced patients’ likelihood of reoperation, available evidence regarding this is limited [
31,
57]. It is possible that cohorts had differential follow-up completeness, however, our required follow-up encounter, lengthy follow-up duration, and capacity for TriNetX to link patients’ encounters across included healthcare organizations minimized this potential variation.
Propensity score matching is an often-used method to adjust for confounding in observational studies, yet can paradoxically increase imbalance and bias especially when the model is misspecified [
59]. In addition, propensity score matching prunes potentially relevant patients, thereby reducing available sample size [
59]. While our propensity score density plot and post-matching SMD values for matched confounders suggested adequate matching [
41,
60,
61], alternative strategies such as direct matching may have offered advantages [
59].
This study focused on the endpoint of reoperation while other highly relevant clinical outcomes of changes in pain, disability, and quality of life could not be examined and may have differed between cohorts. Ideally, these measures would be controlled for and compared longitudinally from prior to the index date of inclusion, at the index date, and during care (SMT and usual care) for at least two years’ follow-up. Without these details, we cannot provide a comprehensive measure of effectiveness of SMT versus usual care in treatment of LSR following lumbar discectomy.
The results of this study may only generalize to adults attending US academic healthcare organizations as other regions may have different management strategies for LSR after lumbar discectomy.