Abstract
For diverse procedures, sizable geographic variation exists in rates and outcomes of surgery, including for degenerative lumbar spine conditions. Little is known about how surgeon training and experience are associated with surgeon-level variations in spine surgery practice and short-term outcomes. This retrospective observational analysis characterized variations in surgical operations for degenerative lumbar scoliosis or spondylolisthesis, two common age-related conditions. The study setting was two large spine surgery centers in one region during 2017–19. Using data (International Classification of Diseases-10th edition and current procedural terminology codes) extracted from electronic health record systems, we characterized surgeon-level variations in practice (use of instrumented fusion — a more extensive procedure that involves device-related risks) and short-term postoperative outcomes (major in-hospital complications and readmissions). Next, we tested for associations between surgeon training (specialty and spine fellowship) and experience (career stage and operative volume) and use of instrumented fusion as well as outcomes. Eighty-nine surgeons performed 2481 eligible operations. For the study diagnoses, spine surgeons exhibited substantial variation in operative volume, use of instrumented fusion, and postoperative outcomes. Among surgeons above the median operative volume, use of instrumented fusion ranged from 0% to >90% for scoliosis and 9% to 100% for spondylolisthesis, while rates of major in-hospital complications ranged from 0% to 25% for scoliosis and from 0% to 14% for spondylolisthesis. For scoliosis, orthopedic surgeons were more likely than neurosurgeons to perform instrumented fusion for scoliosis [49% vs. 33%, odds ratio (OR) = 2.3, 95% confidence interval (95% CI) 1.3–4.2, P-value = .006] as were fellowship-trained surgeons (49% vs. 25%, OR = 3.0, 95% CI 1.6–5.8; P = .001). Fellowship-trained surgeons had lower readmission rates. Surgeons with higher operative volumes used instrumented fusion more often (OR = 1.1, 95% CI 1.0–1.2, P < .05 for both diagnoses) and had lower rates of major in-hospital complications (OR = 0.91, 95% CI 0.85–0.97; P = .006). Surgical practice can vary greatly for degenerative spine conditions, even within the same region and among colleagues at the same institution. Surgical specialty and subspecialty, in addition to recent operative volume, can be linked to variations in spine surgeons’ practice patterns and outcomes. These findings reinforce the notion that residency and fellowship training may contribute to variation and present important opportunities to optimize surgical practice over the course of surgeons’ careers. Future efforts to reduce unexplained variation in surgical practice could test interventions focused on graduate medical education.
Graphical Abstract
Keywords: degenerative lumbar scoliosis, degenerative lumbar spondylolisthesis, practice variations, short-term outcomes
Introduction
For diverse operations, a large body of literature has documented substantial geographic variation in rates of surgery, types of procedures performed, surgical appropriateness, operative outcomes, and costs, including for degenerative spine conditions [1, 2]. Variations have been sizable across regions within the USA [3, 4], Europe [5–8], and Japan [3], for example. Even among surgeons within a region, large variations can exist in the performance and appropriateness of surgery [2]. Variations in rates of surgery have been more strongly associated with surgeons’ practice patterns than patients’ clinical characteristics [2, 4, 9, 10].
Training and operative experience have the potential to influence surgeons’ practice patterns by shaping their beliefs about the value of surgery in specific clinical situations [11]. For example, physicians who train in higher-cost geographic areas continue to deliver higher-cost care [12]. Surgeons who train at academic centers report greater preparedness for challenging cases [11, 13, 14]. Junior surgeons often prefer routine procedures, whereas experienced surgeons report comfort with complex cases [11]. After a major postoperative complication, surgeons often change their practices to prevent recurrences [15]. Across many studies, a surgeon’s operative volume has often been associated with quality of care as well as postoperative health outcomes [16].
Despite the relevance of surgeon training, few studies have directly examined the relationship between surgeon specialty or subspecialty and variations in surgical practice or outcomes. Additionally, while many studies have included recent operative volume, little is known about the role of a surgeon’s longitudinal professional experience. A unique opportunity to examine the role of surgeon training exists for degenerative lumbar spine conditions. The surgeons who routinely perform these operations include both orthopedic surgeons and neurosurgeons, and some surgeons pursue spine fellowship training, while others do not. Additionally, variations in practice are large for lumbar spine procedures [17]. Rates of elective lumbar spinal fusion (for any indication) vary from 9 to 128 operations per 100 000 Medicare beneficiaries across geographic regions. For degenerative lumbar spine conditions, center-level variation in outcomes is sizable [17–19], and rates of major surgical complications are high (5–15%) [20]. Moreover, use of complex lumbar spinal fusion procedures, instead of less complex ones, has been linked to higher rates of major operative complications and rehospitalizations [21].
This study sought to examine the role of surgeon training (specialty discipline and fellowship training) and experience (career stage and operative volume) in surgeon-level variations in practice patterns and outcomes for degenerative lumbar scoliosis and spondylolisthesis. The study was set at two large spine referral centers in the same region, limiting the role of geographic variation. First, we quantified surgeon-level variations in recent experience (operative volume and operative diagnosis mix), practice (use of instrumented fusion — a more extensive procedure that involves device-related risks) [21–23], and short-term clinical outcomes (major in-hospital complications and 30-day readmissions). Next, we examined whether surgeon training and experience were associated with use of instrumented fusion and short-term outcomes.
Methods
This retrospective, observational study is reported in accordance with the Reporting of studies Conducted using Observational Routinely-collected Data Statement [24].
Setting
The sites included two spine referral centers within large regional healthcare systems in Los Angeles. Site 1 is a large private healthcare system that receives tertiary and quaternary referrals from southern CA, NV, and AZ; surgeons are paid on a fee-for-service basis. Site 2 is an integrated healthcare system with 4.7 million capitated members; patients with spinal conditions are referred (almost at random) to spine surgeons within the system; surgeons are salaried. Each system performs 2000–2200 spine procedures per year, including >500 related to degenerative lumbar scoliosis and spondylolisthesis. Sites use similar electronic health record systems.
Participants/study subjects
Surgeon eligibility
We included orthopedic surgeons and neurosurgeons who performed eligible procedures during the study period, excluding those with missing data on training or career stage.
Patient eligibility
Each site identified patients over age 40 years who had International Classification of Diseases-10th edition (ICD-10 CM) codes for degenerative lumbar scoliosis or spondylolisthesis (Supplementary Appendix Table 1) and underwent an eligible operation (see later). We excluded patients with trauma, scoliosis, or spondylolisthesis arising before age 18 years, spinal tumors, prior spine surgery, and concurrent scoliosis and spondylolisthesis (to reduce this as a contributor to variation).
Surgical episode eligibility
Eligible procedures occurred between 1 January 2017 and 31 December 2019, involved current procedural terminology (CPT) codes reflecting decompression, fusion, and deformity correction procedures (Supplementary Appendix Table 1), and represented the first lumbar procedure performed on the patient. Surgical episodes of care included the operation, associated index hospitalization, and 30-day readmissions.
Data sources
Surgeon characteristics
Study sites provided characteristics based on surgeons’ publicly available information.
Patient and episode characteristics
Administrative data from study sites provided ICD-10 CM, CPT codes, and demographic information.
Measures
Surgeon characteristics
These were specialty (orthopedics and neurosurgery), fellowship training in spine (yes and no), and career stage [early (<15 years since medical school graduation), middle (15–30 years), and late (>30 years)].
From administrative datasets, we derived each surgeon’s relevant experience including eligible operative volume (at study site, during study period) and operative diagnosis mix (percentage of cases involving spondylolisthesis vs. scoliosis). We also derived each surgeon’s use of instrumented fusion (percentage of eligible procedures).
Patient characteristics
Sociodemographic variables included gender, age, and self-reported race/ethnicity (Black, Hispanic, White, Asian, and others). Comorbidity measures included osteoporosis (present/absent) and the Elixhauser Comorbidity Index (based on ICD-10 CM codes from up to 3 years prior) [25].
Spine-related variables included central spinal stenosis, foraminal stenosis, leg weakness, neurogenic claudication, and cauda equina syndrome (based on ICD-10 CM codes within the three years prior to surgery).
Episode characteristics
A major study measure was whether surgical procedures involved instrumented fusion. We also characterized other aspects of the procedures (based on CPT codes; Supplementary Appendix Table 2).
Major in-hospital complications included hemorrhage, mechanical complications, dural tears, surgical site infections, anesthesia complications, or mortality during the index hospitalization (based on ICD-10 CM codes; Supplementary Appendix Table 3). We used administrative variables to identify 30-day same-site inpatient readmissions.
Analysis
Surgeon characteristics
We performed descriptive analyses of training, specialty and fellowship training, and career stage.
For eligible operative volume and operative diagnosis mix, we calculated medians, interquartile ranges (IQRs), and ranges, stratified by study site and diagnosis.
Patient and episode characteristics
We performed descriptive analyses of primary diagnoses, sociodemographic characteristics, clinical presentations, surgical approaches, and surgical outcomes.
Surgeon-level variation in practice
For use of instrumented fusion, we calculated medians, IQRs, and ranges, stratified by study site and diagnosis. To describe relationships between each surgeon’s eligible operative volume and his/her use of instrumented fusion, we graphed funnel plots by specialty along with 90% and 95% confidence intervals (CIs) around the mean [26].
To examine associations between surgeon characteristics (specialty, fellowship training, career stage, and operative volume) and use of instrumented fusion, we applied a generalized estimating equations [27] logistic model taking into account correlated within-surgeon observations and assuming an exchangeable working correlation structure. Robust standard errors were computed for the estimates of the regression coefficients with surgeons as clusters. Generalized estimating equation logistic models were stratified by diagnosis (scoliosis or spondylolisthesis). In an alternative modeling approach, we used patient-level random-effects logistic regression with surgeon characteristics including volume as covariates. We performed both sets of analyses with and without adjustment for patient age, gender, race/ethnicity, and Elixhauser Comorbidity Index.
Surgeon-level variation in outcomes
We determined medians, IQRs, and ranges for rates of major in-hospital complications and 30-day readmissions and created funnel plots as described earlier.
To examine associations between surgeon characteristics and episode outcomes, we used the same modeling approaches as earlier except that we did not stratify by diagnosis. Rates of missing data were very low (<1%) so we excluded observations with key missing variables [28].
In a priori power calculations, we assumed a 15.3% rate of major in-hospital complications [20], a 31.5% relative difference between two groups of surgeons [29], and an intraclass correlation coefficient of 0.005. Under these assumptions, studying 2936 episodes of surgical care would have provided 80% power at 5% significance level.
Statistical programming and statistical analyses were conducted in SAS version 9.4 (on Linux), pandas version 1.4.4 (for Python version 3.9.13), R version 4.3.2 for desktop [30], and Stata MP version 17.0. All hypotheses were two-tailed with a 5% significance level.
Institutional review boards at each study site approved this analysis and waived informed consent because all analyses were based on de-identified, coded data.
Results
Surgeon characteristics
The study included 89 eligible spine surgeons (Table 1); surgeon characteristics were missing for three (3%, ∼1% of surgical episodes). Respectively, at Sites 1 and 2, 38% and 74% trained in neurosurgery, while 93% and 57% completed spine fellowships. About half of surgeons were in the midcareer stage.
Table 1.
Characteristics of study spine surgeons, by site.
| Site 1 (N = 42) |
Site 2 (N = 47) |
|
|---|---|---|
| Training | ||
| Specialty, N (%) | ||
| Neurosurgery | 16 (38) | 35 (74) |
| Orthopedics | 26 (62) | 12 (26) |
| Spine fellowship, N (%) | 39 (93) | 27 (57) |
| Duration of experience | ||
| Career stage, N (%) | ||
| Early (<15 years) | 11 (26) | 18 (38) |
| Middle (15–29 years) | 22 (52) | 24 (51) |
| Late (30+ years) | 9 (21) | 5 (11) |
| Recent experience | ||
| Eligible operative volume over 3 years, median (IQR, range) | 14 (5–29, 1–115) | 9 (4–62, 1–119) |
| Eligible volume <15, N (%) | 21 (50) | 26 (55) |
| Operative diagnosis mix, median (IQR, range)a | 58% (29%–73%, 0%–100%) | 76% (56%–88%, 0%–100%) |
| Practice patterns | ||
| Use of instrumented fusion, median (IQR, range) | ||
| Scoliosis | 35% (19%–55%, 0%–100%) | 50% (18%–74%, 0%–100%) |
| Spondylolisthesis | 68% (50%–93%, 0%–100%) | 72% (63–84%, 0%–100%) |
| Operative outcomes | ||
| Rate of major in-hospital complications, median (IQR, range) | 7% (4%–10%, 0%–17%) | 5% (3%–7%, 0%–13%) |
| Rate of 30-day same-hospital readmission, median (IQR, range) | 4% (0%–6%, 0%–12%) | 4% (3%–8%, 0%–17%) |
Proportion of eligible operations for spondylolisthesis vs. scoliosis.
Patient and episode characteristics
Across both sites, we studied 2481 surgical episodes. More eligible operations were for spondylolisthesis than scoliosis. Study populations and surgical episodes were similar between sites except that more patients had scoliosis at Site 1 than at Site 2 (Table 2).
Table 2.
Characteristics of study patients and surgical episodes of care.
| Site 1 | Site 2 | |||
|---|---|---|---|---|
| Scoliosis (N = 463) | Spondylolisthesis (N = 433) | Scoliosis (N = 355) | Spondylolisthesis (N = 1230) | |
| Patient characteristics | ||||
| Demographics | ||||
| Age, mean | 69.2 | 66.0 | 67.4 | 66.4 |
| Female, N (%) | 232 (50) | 226 (52) | 201 (57) | 702 (57) |
| White, non-Hispanic, N (%) | 360 (78) | 316 (73) | 232 (65) | 699 (57) |
| Hispanic ethnicity, N (%) | 42 (9) | 62 (14) | 77 (22) | 338 (27) |
| Asian, N (%) | 24 (5) | 13 (3) | 20 (6) | 84 (7) |
| Black, non-Hispanic, N (%) | 16 (3) | 25 (6) | 22 (6) | 88 (7) |
| Clinical presentation | ||||
| Central or foraminal stenosis, N (%) | 407 (88) | 382 (88) | 225 (63) | 902 (73) |
| Leg weakness, N (%) | 11 (2) | 3 (1) | 3 (1) | 6 (<1) |
| Neurogenic claudication, N (%) | 174 (38) | 176 (41) | 83 (23) | 390 (32) |
| Cauda equina syndrome, N (%) | 4 (1) | 1 (<1) | 5 (1) | 13 (1) |
| Comorbidities | ||||
| Elixhauser mortality risk index, mean | −1.8 | −2.4 | −3.0 | −4.0 |
| Osteoporosis, N (%) | 69 (15) | 65 (15) | 57 (16) | 167 (14) |
| Episode characteristics | ||||
| Surgical approach | ||||
| Instrumented fusion, N (%) | 176 (38) | 296 (68) | 199 (56) | 948 (77) |
| Anterior lumbar interbody fusion, N (%) | 86 (19) | 150 (35) | 39 (11) | 140 (11) |
| Posterior lumbar interbody fusion, N (%) | 118 (25) | 178 (41) | 66 (19) | 390 (32) |
| Anterior and posterior approach, N (%) | 43 (9) | 41 (9) | 33 (9) | 112 (9) |
| Decompression, N (%) | 326 (70) | 229 (53) | 226 (64) | 793 (64) |
| Discectomy, N (%) | 390 (84) | 410 (95) | 180 (51) | 657 (53) |
| Posterior approach, N (%) | 237 (51) | 211 (49) | 271 (76) | 1119 (91) |
| Surgical outcomes | ||||
| Major in-hospital complication (including death), N (%) | 29 (6) | 35 (8) | 24 (7) | 62 (5) |
| Dural laceration, N (%) | 23 (5) | 24 (6) | 20 (6) | 50 (4) |
| Mechanical complication, N (%) | 5 (1) | 9 (2) | 1 (<1) | 4 (<1) |
| Length of stay, mean | 4.0 | 3.5 | 4.3 | 3.8 |
| Readmission within 30 days, N (%) | 19 (4) | 19 (4) | 22 (6) | 66 (5) |
Note: For age, length of stay, and Elixhauser index, the results represent subpopulation means; for other variables, results represent the raw counts as well as the percentages of each subpopulation. In-hospital complications that are not itemized in the table (hemorrhage, surgical site infection, or mortality during the index hospitalization) had incidence rates <0.5%.
Surgeon-level variation in practice
Surgeons varied in eligible operative volume and diagnosis mix (Table 1). Per surgeon, the median operative volume over 3 years was 9 (range 1–119) and 14 (range 1–115) at the two sites, respectively. At least half of surgeons had volumes <15.
Surgeons varied greatly in their use of instrumented fusion across as seen in Table 1 and Fig. 1. Among surgeons with ≥15 eligible cases, use of instrumented fusion ranged from 0% to 92% for scoliosis and 9% to 100% for spondylolisthesis and reflecting 11-fold to >90-fold variation. Variation was similar across sites.
Figure 1.
(A) Individual surgeons’ volumes of eligible scoliosis operations over 3 years and the proportion of those operations involving instrumented fusion. (B) Individual surgeons’ volumes of eligible spondylolisthesis operations over 3 years and the proportion of those operations involving instrumented fusion.
Surgeon characteristics and use of instrumented fusion
In multivariable analyses, orthopedic surgeons performed instrumented fusions at higher rates than neurosurgeons for scoliosis (49% vs. 33%) and spondylolisthesis (74% vs. 49%); these differences were statistically significant for scoliosis but not spondylolisthesis (Table 3). Using the alternative modeling approach, the differences were significant for both diagnoses. Results were similar without adjustment (Supplementary Appendix Table 4A–C).
Table 3.
Results for main analyses: adjusted differences in surgeon-level practices and short-term outcomes by surgeon training and experience.
| Practices | Outcomes | ||
|---|---|---|---|
| Instrumented fusion | Major in-hospital complication | 30-day readmission | |
| Training | |||
| Specialty, orthopedics vs. neurosurgery: mean rates, OR (95% CI; P-value) | – | 6.4% vs. 5.5%, 1.2 (0.74–1.9; .49) |
4.3% vs. 6.2%, 0.71 (0.45–1.1; .14) |
| Scoliosis |
49% vs. 33%, 2.3
(1.3–4.2; .006) |
– | – |
| Spondylolisthesis | 74% vs. 49%, 1.4 (0.87–2.2; .17) |
– | – |
| Spine fellowship vs. no fellowship: mean rates, OR (95% CI; P-value) | – | 6.2% vs. 5.1%, 1.3 (0.71–2.2; .43) |
4.5% vs. 7.4%, 0.55 (0.32–0.94; .027) |
| Scoliosis |
49% vs. 25%, 3.0 (1.6–5.8; .001) |
– | – |
| Spondylolisthesis | 76% vs. 71%, 1.3 (0.90–2.0; .16) |
– | – |
| Experience | |||
| Career stage: mean rates, OR (95% CI; P-value) | |||
| Early vs. late | — | 6.2% vs. 8.6%, 1.1 (0.68–1.9; .61) |
5.4% vs. 4.6%, 1.1 (0.70–1.7; .67) |
| Scoliosis | 42% vs. 42%, 0.76 (0.39–1.5; .45) |
– | – |
| Spondylolisthesis | 76% vs. 69%, 1.1 (0.62–1.9; .79) |
– | – |
| Eligible operative volume: OR for 10-patient increase (95% CI; P-value) | — |
0.91
(0.85–0.97; .006) |
1.0 (0.43–1.1; .91) |
| Scoliosis |
1.1
(1.0–1.2; .036) |
– | – |
| Spondylolisthesis |
1.1
(1.0–1.2; .005) |
– | – |
Note: Statistically significant results (P<0.05) are indicated in bold.
For scoliosis, surgeons with fellowship training performed instrumented fusion significantly more often than surgeons without fellowship training (49% vs. 25%), but there was no statistically significant difference for spondylolisthesis (76% vs. 71%). Rates of instrumented fusion did not differ by surgeon career stage for scoliosis or spondylolisthesis. Results were similar using the alternative approach and in unadjusted analyses.
A surgeon’s operative volume was associated with use of instrumented fusion. Among the three highest volume surgeons, rates of instrumented fusion varied from 8% to 63%. For both diagnoses, operating on 10 more patients over 3 years had an OR of 1.1 for the use of instrumented fusion (Table 3). Using the alternative modeling approach, the association was only statistically significant for spondylolisthesis. Unadjusted results were similar.
Surgeon-level variation in outcomes
Rates of major in-hospital complications and 30-day same-hospital readmissions varied among surgeons (Table 1 and Fig. 2) across the two sites. Rates of complications and the degree of variation were similar when limited to instrumented fusion procedures (Site 1 median 7%, IQR 1%–12%, range 0%–33%; Site 2 median 5%, IQR 2%–8%, range 0%–16%). Among surgeons with >15 eligible cases, complication rates ranged from 0% to 25% for scoliosis and from 0% to 14% for spondylolisthesis, reflecting >14-fold to >25-fold variation.
Figure 2.
(A) Individual surgeons’ total volumes of eligible operations over 3 years and the proportion of operations involving major in-hospital complications. (B) Individual surgeons’ total volumes of eligible operations over 3 years and the proportion of operations involving 30-day same-hospital readmissions.
The median rate of readmission was 4% at both sites; rates and variation were similar when limited to instrumented fusion procedures (Site 1 median 4%, IQR 0%–6%, range 0%–19%; Site 2 median 5%, IQR 3%–8%, range 0%–21%).
Surgeon characteristics and outcomes
In multivariable analyses (Table 3), surgeons with orthopedic vs. neurosurgical specialty training did not differ in rates of complications (6.4% vs. 5.5%) or readmissions (4.3% vs. 6.2%). Spine fellowship training was not associated with complication rates (6.2% vs. 5.1%), but it was statistically significantly associated with lower readmission rates (4.5% vs. 7.4%). Early-career, midcareer, and late-career surgeons did not differ in rates of complications or readmissions.
Surgeons with higher eligible operative volumes had statistically significantly lower complication rates: operating on 10 more patients over 3 years had an OR of 0.91 for any complication. Volume was not associated with the surgeon’s readmission rate.
Results were similar using the alternative modeling approach and in unadjusted analyses (Supplementary Appendix Table 4A–C).
Discussion
Statement of principal findings
This work has two key findings. First, for degenerative lumbar scoliosis and spondylolisthesis, we observed large variations in surgical practice and short-term operative outcomes among individual surgeons at two spine referral centers in one region. Among surgeons above the median for volume, use of instrumented fusion varied >90-fold for scoliosis and >11-fold for spondylolisthesis, while rates of major in-hospital complications varied >14-fold and >25-fold, respectively. Second, we detected associations between the surgeons’ training and experience and both practices and outcomes. For scoliosis, orthopedic surgeons were more likely than neurosurgeons to perform instrumented fusions as were spine fellowship-trained surgeons. Surgeons with higher operative volumes were more likely to use instrumentation for both diagnoses, but use did not vary with the surgeon’s career stage. Spine fellowship training was associated with lower readmission rates. Most notably, having higher eligible operative volumes was statistically significantly associated with lower rates of major in-hospital complications.
Strengths and limitations
The study size—almost 90 surgeons and 2500 patients—enhanced our ability to detect variations in surgical practice. We included two sites with distinct organizational cultures and financial incentives, since these have the potential to influence practice patterns, and findings were generally consistent across study sites. At one study site, surgeons only practice at that institution and patients are assigned to surgeons almost at random, which diminishes the likelihood that selection bias in patient referrals could explain our findings. We included multiple measures of training (specialty and spine fellowship) and experience (career stage and operative volume). A key measure of surgical practice, the use of instrumented fusion, is likely to be coded accurately because it affects billing. We employed widely accepted clinical outcome measures.
Nonetheless, the work also has limitations. The ICD-10 CM codes for scoliosis and spondylolisthesis can be subject to variable documentation practices. The operative volume measure did not capture surgeons’ procedures for similar conditions or at outside institutions. Data sources did not capture readmissions to outside hospitals; however, study sites report that these were rare. Statistical power was only sufficient to detect relatively large (>31%) differences between groups. Statistical models could not fully control for selection bias because the study only included people who underwent surgery, and because administrative data lack important clinical details. However, results did not change with vs. without adjustment for patient demographic characteristics and comorbidity. Furthermore, adjusting for the clinical variables that surgeons weigh in decisions to operate (e.g. symptoms, physical exam signs, and imaging test results) would remove a key mechanism by which practice patterns and outcomes vary [31, 32].
Interpretation within the context of the wider literature
This study adds important nuance to prior literature examining variations in spine surgery practice. Several prior studies have documented large variations across geographic regions in rates of spine surgery as well as the procedure types selected [4, 8, 33]. One study documented large variations in the performance and appropriateness of surgery within regions [2]. By focusing on two large referral centers in one region, our analysis shifts the emphasis toward documenting variation among individual surgeons. Furthermore, few prior studies have examined the roles of surgeon training and longitudinal operative experience [34]. Our results provide evidence that surgical specialty and fellowship training, as well as recent operative experience, can be associated with large variations in surgical practice, but longitudinal experience did not appear to be associated with large variations.
Despite linking surgeon training and experience to practice patterns, our data do not indicate that specific training is advantageous for these diagnoses because we found that major complication rates were generally similar. Our rates of complications and readmissions, as well as surgeon-level variation in these measures, are similar to prior studies of lumbar spine surgery [17, 21, 35]. In future research, we hope to assess whether surgeon training is associated with longer-term patient-reported outcomes, including physical function, pain interference, and depression.
Similar to many prior studies, including in spine surgery [1, 16, 36], we detected a significant association between higher surgeon-level operative volumes and improved health outcomes. Most prior research has focused on patient-reported outcomes [1, 16, 36], whereas ours is innovative in focusing on highly morbid and costly short-term outcomes. Prior research has indicated that an individual surgeon’s volume can be important when the operation requires specific intra-operative processes and skills as is the case for degenerative lumbar scoliosis and spondylolisthesis, whereas the referral center/hospital’s volume can be important when specific hospital-based services or a long in-hospital stay are involved [1, 16].
Implications
Our findings reinforce the notion that residency and fellowship training contribute to surgical variation and present important opportunities to optimize surgical practice over the course of surgeons’ careers. Surgical training programs already routinely teach surgeons how to select patients for surgery and operations for surgical candidates, but marked variations in care practices have persisted over the long term. Future efforts to reduce unexplained and unwarranted variation in surgical practice could rigorously test diverse quality improvement interventions focused on graduate medical education, such as didactics, surgical indications case conferences, use of quality metrics, and project-based learning, as well as novel interventions [37–40]. For surgeons who have completed training, similar strategies may be helpful to assure the appropriateness of surgery and minimize complications. Any future interventions should consider the fact that variation in surgical practice does not equate to unnecessary or inappropriate care [10], and that variation can occur with beneficial advances in operative care.
In addition, future work should include formally assessing spine surgeons’ adherence to rigorously developed appropriateness criteria [41–43]. Designed to lessen unwarranted and variation in surgical practice, appropriateness criteria provide recommendations about the risks and benefits of a specific operation for individual patients. The criteria are a type of quality measure, distinct from practice guidelines or shared decision-making aids [44].
Conclusion
Surgical practice can vary greatly for degenerative spine conditions, even within the same region and among colleagues at the same institution. Surgical specialty and subspecialty training, in addition to recent operative volume, can be linked to variations in spine surgeons’ practice patterns and outcomes. This suggests that residency and fellowship training may contribute to variation and present important opportunities to optimize surgical practice over the course of surgeons’ careers. Future efforts to reduce unexplained variation in surgical practice could test interventions focused on graduate medical education.
Supplementary Material
Acknowledgements
We thank Dr Babak Khandehroo (Cedars-Sinai Medical Center) and Kim Holmquist and Joanie Chung (Kaiser Permanente Southern CA) for assistance with extracting data from their respective sites.
Contributor Information
Kanaka D Shetty, RAND Health Care, RAND Corporation, 1776 Main Street, Santa Monica, CA 90401, USA.
Peggy G Chen, RAND Health Care, RAND Corporation, 1776 Main Street, Santa Monica, CA 90401, USA.
Harsimran S Brara, Kaiser Permanente, Los Angeles Medical Center, 4867 W Sunset Blvd, Los Angeles, CA 90027, USA.
Neel Anand, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, CA 90048, USA.
David L Skaggs, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, CA 90048, USA.
Vinicius F Calsavara, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, CA 90048, USA.
Nabeel S Qureshi, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, CA 90048, USA.
Rebecca Weir, RAND Health Care, RAND Corporation, 1776 Main Street, Santa Monica, CA 90401, USA.
Karma McKelvey, Rocky Vista University, Montana College of Osteopathic Medicine, 4130 Rocky Vista Way, Billings, Montana 59106, USA.
Teryl K Nuckols, RAND Health Care, RAND Corporation, 1776 Main Street, Santa Monica, CA 90401, USA; Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, CA 90048, USA.
Author contributions
As per the International Committe of Medical Journal Editors (ICMJE), the criteria for authorship were met for all authors, with specific contributions as follows.
Substantial contributions to the (a) conception or design of the work or (b) the acquisition, (c) analysis, or (d) interpretation of data for the work;
(a) Drafting the work or (b) revising it critically for important intellectual content;
Final approval of the version to be published; and
Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
| Criterion | ||||||||
|---|---|---|---|---|---|---|---|---|
| Author last name | 1a | 1b | 1c | 1d | 2a | 2b | 3 | 4 |
| Shetty | X | X | X | X | X | X | X | X |
| Chen | X | X | X | X | X | X | ||
| Brara | X | X | X | X | X | X | ||
| Anand | X | X | X | X | X | X | ||
| Skaggs | X | X | X | X | X | |||
| Weir | X | X | X | X | X | X | X | X |
| McKelvey | X | X | X | X | X | |||
| Nuckols | X | X | X | X | X | X | X | X |
Supplementary data
Supplementary data is available at IJQHC online.
Conflict of Interest
None declared.
Funding
This work was supported by the National Institute on Aging (1R21AG059214-01A1).
Data availability
Due to human subjects’ protections, datasets are not available for use outside the study team.
Ethics and other permissions
Institutional review boards at the RAND Corporation and each study site approved this analysis, including a waiver of informed consent for both spine surgeons and study patients. All analyses were based on de-identified, coded data. No copyrighted material was used.
References
- 1. Birkmeyer JD, Stukel TA, Siewers AE et al. Surgeon volume and operative mortality in the United States. New Engl J Med 2003;349:2117–27. 10.1056/NEJMsa035205 [DOI] [PubMed] [Google Scholar]
- 2. Song Z, Kannan S, Gambrel RJ et al. Physician practice pattern variations in common clinical scenarios within 5 US metropolitan areas. JAMA Health Forum 2022;3:e214698. 10.1001/jamahealthforum.2021.4698 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Hurley MP, Schoemaker L, Morton JM et al. Geographic variation in surgical outcomes and cost between the United States and Japan. Am J Manag Care 2016;22:600–7. [PubMed] [Google Scholar]
- 4. Martin BI, Tosteson AN, Lurie JD et al. Variation in the care of surgical conditions: spinal stenosis. A Dartmouth Atlas of Health Care Series. Dartmouth Institute; 2014. [PubMed] [Google Scholar]
- 5. Heijink R, Engelfriet P, Rehnberg C et al. A window on geographic variation in health care: insights from EuroHOPE. Health Econ 2015;24:164–77. 10.1002/hec.3287 [DOI] [PubMed] [Google Scholar]
- 6. Walsh ME, Boland F, O’Byrne JM et al. Geographical variation in musculoskeletal surgical care in public hospitals in Ireland: a repeated cross-sectional study. BMJ Open 2019;9:e028037. 10.1136/bmjopen-2018-028037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Weeks WB, Paraponaris A, Ventelou B. Geographic variation in rates of common surgical procedures in France in 2008–2010, and comparison to the US and Britain. Health Policy 2014;118:215–21. 10.1016/j.healthpol.2014.08.015 [DOI] [PubMed] [Google Scholar]
- 8. Mäntymäki H, Ponkilainen VT, Huttunen TT et al. Regional variations in lumbar spine surgery in Finland. Arch Orthop Trauma Surg. 2023;143:1451–8. 10.1007/s00402-021-04313-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Rosenberg BL, Kellar JA, Labno A et al. Quantifying geographic variation in health care outcomes in the United States before and after risk-adjustment. PLoS One 2016;11:e0166762. 10.1371/journal.pone.0166762 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Birkmeyer JD, Reames BN, McCulloch P et al. Understanding of regional variation in the use of surgery. Lancet 2013;382:1121–9. 10.1016/S0140-6736(13)61215-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Gunaratnam C, Bernstein M. Factors affecting surgical decision-making—a qualitative study. Rambam Maimonides Med J 2018;9:e0003 10.5041/RMMJ.10324 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Phillips RL Jr, Petterson SM, Bazemore AW et al. The effects of training institution practice costs, quality, and other characteristics on future practice. Ann Fam Med 2017;15:140–8. 10.1370/afm.2044 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Phillips RL, Holmboe ES, Bazemore AW et al. Purposeful imprinting in graduate medical education: opportunities for partnership. Fam Med 2021;53:574–7. 10.22454/FamMed.2021.264013 [DOI] [PubMed] [Google Scholar]
- 14. Chen C, Petterson S, Phillips R et al. Spending patterns in region of residency training and subsequent expenditures for care provided by practicing physicians for medicare beneficiaries. JAMA Intern Med 2014;312:2385–93. 10.1001/jama.2014.15973 [DOI] [PubMed] [Google Scholar]
- 15. Simianu VV, Basu A, Alfonso-Cristancho R et al. Assessing surgeon behavior change after anastomotic leak in colorectal surgery. J Surg Res 2016;205:378–83. 10.1016/j.jss.2016.06.075 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Morche J, Mathes T, Pieper D. Relationship between surgeon volume and outcomes: a systematic review of systematic reviews. Syst Rev 2016;5:204. 10.1186/s13643-016-0376-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Martin BI, Mirza SK, Franklin GM et al. Hospital and surgeon variation in complications and repeat surgery following incident lumbar fusion for common degenerative diagnoses. Health Serv Res 2013;48:1–25. 10.1111/j.1475-6773.2012.01434.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Desai A, Bekelis K, Ball PA et al. Variation in outcomes across centers after surgery for lumbar stenosis and degenerative spondylolisthesis in the spine patient outcomes research trial. Spine (Phila Pa 1976) 2013;38:678–91. 10.1097/BRS.0b013e318278e571 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Chotai S, Sivaganesan A, Sielatycki JA et al. Surgeon-level variability in outcomes, cost, and comorbidity adjusted-cost for elective lumbar decompression and fusion. Neurosurgery 2018;82:506–15. 10.1093/neuros/nyx243 [DOI] [PubMed] [Google Scholar]
- 20. Mannion AF, Fekete TF, Porchet F et al. The influence of comorbidity on the risks and benefits of spine surgery for degenerative lumbar disorders. Eur Spine J 2014;23:S66–71. 10.1007/s00586-014-3189-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Deyo RA, Mirza SK, Martin BI et al. Trends, major medical complications, and charges associated with surgery for lumbar spinal stenosis in older adults. JAMA Intern Med 2010;303:1259–65. 10.1001/jama.2010.338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Hirase T, Ling JF, Haghshenas V et al. Instrumented versus noninstrumented spinal fusion for degenerative lumbar spondylolisthesis: a systematic review. Clin Spine Surg 2022;35:213–21. 10.1097/BSD.0000000000001266 [DOI] [PubMed] [Google Scholar]
- 23. Shillingford JN, Laratta JL, Sarpong NO et al. Instrumentation complication rates following spine surgery: a report from the Scoliosis Research Society (SRS) morbidity and mortality database. J Spine Surg 2018;5:110–5. 10.21037/jss.2018.12.09 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Benchimol EI, Smeeth L, Guttmann A et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med 2015;12:e1001885. 10.1371/journal.pmed.1001885 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Menendez M, Neuhaus V, van Dijk C et al. The Elixhauser comorbidity method outperforms the charlson index in predicting inpatient death after orthopaedic surgery. Clin Orthop Relat R 2014;472:2878–86. 10.1007/s11999-014-3686-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Spiegelhalter DJ. Funnel plots for comparing institutional performance. Stat Med 2005;24:1185–202. 10.1002/sim.1970 [DOI] [PubMed] [Google Scholar]
- 27. Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13–22. 10.1093/biomet/73.1.13 [DOI] [Google Scholar]
- 28. French B, Farjah F, Flum DR et al. A general framework for estimating volume-outcome associations from longitudinal data. Stat Med 2012;31:366–82. 10.1002/sim.4410 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Bederman SS, Kreder HJ, Weller I et al. The who, what and when of surgery for the degenerative lumbar spine: a population-based study of surgeon factors, surgical procedures, recent trends and reoperation rates. Can J Surg J Canadien de Chirurgie 2009;52:283–90. [PMC free article] [PubMed] [Google Scholar]
- 30. R Core Team . R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2023. [Google Scholar]
- 31. Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology 2009;20:488–95. 10.1097/EDE.0b013e3181a819a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Steiner PM, Cook TD, Shadish WR et al. The importance of covariate selection in controlling for selection bias in observational studies. Psychol Methods 2010;15:250–67. 10.1037/a0018719 [DOI] [PubMed] [Google Scholar]
- 33. Bae HW, Rajaee SS, Kanim LE. Nationwide trends in the surgical management of lumbar spinal stenosis. Spine (Phila Pa 1976) 2013;38:916–26. 10.1097/BRS.0b013e3182833e7c [DOI] [PubMed] [Google Scholar]
- 34. McCulloch P, Nagendran M, Campbell WB et al. Strategies to reduce variation in the use of surgery. Lancet 2013;382:1130–9. 10.1016/S0140-6736(13)61216-7 [DOI] [PubMed] [Google Scholar]
- 35. Bernatz JT, Anderson PA. Thirty-day readmission rates in spine surgery: systematic review and meta-analysis. Neurosurg Focus 2015;39:E7. 10.3171/2015.7.FOCUS1534 [DOI] [PubMed] [Google Scholar]
- 36. Farjoodi P, Skolasky RL, Riley LH. The effects of hospital and surgeon volume on postoperative complications after LumbarSpine surgery. Spine (Phila Pa 1976) 2011;36:2069–75. 10.1097/BRS.0b013e318202ac56 [DOI] [PubMed] [Google Scholar]
- 37. Anderson CI, Basson MD, Ali M et al. Comprehensive multicenter graduate surgical education initiative incorporating entrustable professional activities, continuous quality improvement cycles, and a web-based platform to enhance teaching and learning. J Am Coll Surg 2018;227:64–76. 10.1016/j.jamcollsurg.2018.02.014 [DOI] [PubMed] [Google Scholar]
- 38. Kobritz M, Xie Z, Antonacci A et al. Surgical Quality at the Intersection Between Education and Accountability—Overview for Trainees and Surgeons in Practice. https://www.facs.org/for-medical-professionals/news-publications/journals/rise/articles/surgical-quality-at-the-intersection-between-education-and-accountability/ (26 November 2023, date last accessed).
- 39. Sethi R, Buchlak QD, Yanamadala V et al. A systematic multidisciplinary initiative for reducing the risk of complications in adult scoliosis surgery. J Neurosurg Spine 2017;26:744–50. 10.3171/2016.11.SPINE16537 [DOI] [PubMed] [Google Scholar]
- 40. Buchlak QD, Yanamadala V, Leveque JC et al. Complication avoidance with pre-operative screening: insights from the Seattle spine team. Curr Rev Musculoskelet Med 2016;9:316–26. 10.1007/s12178-016-9351-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Chen PGC, Daubs MD, Berven S et al. Surgery for degenerative lumbar scoliosis: the development of appropriateness criteria. Spine 2016;41:910–8. 10.1097/BRS.0000000000001392 [DOI] [PubMed] [Google Scholar]
- 42. Mannion AF, Pittet V, Steiger F et al. Development of appropriateness criteria for the surgical treatment of symptomatic lumbar degenerative spondylolisthesis (LDS). Eur Spine J 2014;23:1903–17. 10.1007/s00586-014-3284-0 [DOI] [PubMed] [Google Scholar]
- 43. Reitman CA, Cho CH, Bono CM et al. Management of degenerative spondylolisthesis: development of appropriate use criteria. Spine J 2021;21:1256–67. 10.1016/j.spinee.2021.03.005 [DOI] [PubMed] [Google Scholar]
- 44. Fitch K, Bernstein SJ, Aguilar MD et al. The RAND/UCLA appropriateness method user’s manual. Santa Monica, CA: RAND Corporation; 2001. Report No.: MR-1269-DG-XII/RE. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Due to human subjects’ protections, datasets are not available for use outside the study team.