Risk Factors for Reoperation in Patients Treated Surgically for Degenerative Spondylolisthesis
A Subanalysis of the 8-Year Data From the SPORT Trial
Level of Evidence: 2
Degenerative spondylolisthesis (DS) is an acquired anterior displacement of one vertebrae over another most commonly occurring in patients older than 50 years.[1–3] This vertebral slip may produce various symptoms including back pain and neurogenic claudication.[1,2,4,5] Several well-designed, prospective studies support surgical intervention for patients with severe, unremitting symptoms after conservative treatment.[1,2,4–7] In the early 1990s, Herkowitz and Kurz showed significantly favorable outcomes in patients undergoing decompression with arthrodesis versus decompression alone. Surgical decompression with arthrodesis has now become the mainstay of treatment for patients with symptomatic DS.[1,2,4–6] Several high-quality studies have examined patient outcomes, complications, and costs following DS surgery, but few have identified consistent risk factors for reoperation.[1,2,4–6]
Although there are several known patient factors, including obesity, depression, and diabetes, which are predictors of poor outcome of treatment of DS, risk factors for reoperation are less well studied. The current literature shows a marked variance in reoperation rates for the treatment of DS, ranging from 5% to 35%.[1,6,9] Several factors contribute to the variability in reoperation rates: hospital and geographic variables; surgeon experience; surgical technique; and patient risk factors.[2,4,5,10,11]There are many potential risk factors, although few have been supported by a large prospective study.
The Spine Patient Outcomes Research Trial (SPORT) is a large, multicenter, prospective study with strict inclusion criteria across three arms: intervertebral disc herniation, DS, and spinal stenosis. The purpose of this study was to perform a subanalysis of the 8-year SPORT data to determine baseline risk factors for and outcomes of reoperation in patients treated surgically for DS.
This study was a subgroup analysis of data from the SPORT trial. Enrollment began in March of 2000 and ended in February 2005.
The SPORT trial is a multicenter study carried out among 13 institutions in 11 states across the United States. Patients diagnosed with DS were enrolled in the observational or randomized groups and underwent either surgical or nonsurgical treatment. Data were collected at 6 weeks, 3, 6, and 12 months, and then yearly to 8 years post enrollment.
Inclusion criteria were symptoms of spinal stenosis for at least 12 weeks plus a confirmatory imaging study (CT or MRI) showing spinal stenosis on cross-sectional imaging, and DS on standing lateral radiographs of L3/4 or L4/5. Exclusion criteria were previous lumbar spine surgery; isthmic spondylolisthesis; cauda equina syndrome; scoliosis >15°; vertebral fractures; infection or tumor; inflammatory spondyloarthropathy; pregnancy; or medical contraindications for surgery.
The protocol surgery was a standard posterior decompressive laminectomy, with or without bilateral single-level fusion with or without posterior pedicle-screw instrumentation. Nonsurgical protocol included at least active physical therapy, education or counseling, home exercises, and non-steroidal anti-inflammatory drugs (NSAIDs) if tolerated with additional treatments individualized to patient need. “Reoperation” included any surgical treatment performed in the lumbar spine either at or adjacent to the level of the index procedure.
Primary endpoints were the Medical Outcomes Study 36-Item Short-Form General Health Survey bodily pain and physical function scores (SF 36 BP and PF), as well as Oswestry Disability Index, American Academy of Orthopaedic Surgeons/Modems version (ODI). Secondary outcomes included patient-reported self-improvement, Stenosis Bothersomeness Index (SBI), satisfaction with current symptoms and care, and work status.
Patient characteristics included age, sex, work lift demand, race, education, marital status, work status, compensation, body mass index (BMI), and smoking status. Assessed comorbidities included hypertension, diabetes, osteoporosis, depression, heart problems, stomach problems, bowel or intestinal problems, depression, joint problems, and “other.” Additional candidate variables included duration of symptoms, patient’s self-assessment of health, patient satisfaction with symptoms, expectation of being pain-free with surgery or nonoperative treatment, opioid use, injections, physical therapy, antidepressant use, NSAIDs, missed work, and low back pain severity.
Operative outcomes included procedure (decompression only, noninstrumented fusion, instrumented fusion), multilevel fusion, levels decompressed, operative time, blood loss, blood replacement, length of hospital stay, and complications. Intraoperative complications included dural tear or spinal fluid leak, nerve root injury, or other. Postoperative complications were measured up to 8 weeks postoperatively and included nerve root injury, wound hematoma, or other postoperative complications. Reasons for reoperation were defined by categories including progressive spondylolisthesis, recurrent stenosis, complications, or other. The operative outcomes were analyzed for the reoperation versus no reoperation group, as well as for the noninstrumented fusion versus instrumented fusion groups.
We performed a subgroup analysis of surgically treated patients, including both randomized and observational cohorts, stratified into two groups: reoperation or no reoperation. Baseline characteristics, operative factors, complications, and medical events were compared between the two groups using χ 2 tests for categorical variables and t tests for continuous variables. A Cox proportional hazards model was used to analyze the risk factors for reoperation. Variables that were significant at the P < 0.10 level were candidates for inclusion in the final multivariable regression model. Final selection for the model was done using the stepwise method, removing variables that did not maintain P < 0.05. Age and sex were forced into the model. Primary outcomes analyses compared the reoperation and no reoperation groups using changes from baseline at each follow-up, with a mixed-effects longitudinal regression model including a random individual effect to account for correlation between repeated measurements within individuals. The analyses were adjusted for age, sex, race, work status, BMI, neuroforaminal involvement, depression, joint problem, stomach problem, reflex deficit, number of moderately or severely stenotic levels, other comorbidities, baseline scores (for SF-36, ODI, SBI), and center. Across the 8 years of follow-up, overall comparisons of “area under the curve” between groups were performed using a Wald test. Computations were performed using SAS procedure PROC MIXED for continuous data and PROC GENMOD for binary outcome (SAS version 9.2; SAS Institute Inc, Cary, NC). Statistical significance was defined as P < 0.05 based on a two-sided hypothesis test with no adjustments made for multiple comparisons.
Overall, 607 of 892 eligible participants were enrolled into the study: 304 in the randomized cohort and 303 in the observational cohort (Figure 1). Of these, 601 completed at least one follow-up visit. A total of 406 surgery patients were available at 8 years post enrollment, constituting 69% of the randomized and 57% of the observational participants.
Flow diagram of the Spine Patient Outcomes Research Trial degenerative spondylolisthesis (DS) study: randomized and observational cohorts. Numbers reported at follow-up time points are cumulative and percentages are calculated from the number of patients in each respective arm.
Univariate analysis revealed patients who had undergone a reoperation by 8 years were younger than those without a reoperation (62.2 vs. 65.3 years, P= 0.008). Patients who had a reoperation were less likely to have neurogenic claudication at baseline (73% vs. 89%, P < 0.002). There were no significant differences in other demographic variables or specific comorbidities (Table 1). Patients who underwent a reoperation had worse SF-36 BP (28.1 vs. 32.6, P= 0.044), ODI (47.0 vs. 43.1, P = 0.048), and Stenosis Frequency Index scores (15.8 vs. 14.3, P = 0.021) at baseline.
Univariate analysis revealed patients in the reoperation group had a higher incidence of total postoperative complications following the index surgery (39% vs. 27%, P = 0.036). However, there were no significant differences in wound hematoma rate, infection rate, neurological injury, or dural tear between the reoperation and nonreoperation groups. There were no significant differences in other operative outcomes measures including type of operation, operative time, blood loss, length of hospital stay, or incidence of intraoperative complications (Table 2).
When comparing all patients included in the study, the noninstrumented fusion group had a higher rate of decompression at L2-L3 (20% vs. 10%, P = 0.014) and L3-L4 (61% vs. 46%, P = 0.022) than the instrumented fusion group. The noninstrumented fusion group had a significantly shorter operative time (157.6 vs. 226.6 minutes, P < 0.001), less blood loss (506.5 vs.632.2, P = 0.031), and a lower rate of intraoperative blood replacement (27% vs. 38%, P = 0.062) than the instrumented fusion group (Table 3). There was no significant difference in the rate of wound hematoma, neurological injury, or infection between instrumented and noninstrumented cases.
Of the 406 surgery patients, 72% underwent instrumented fusion, 21% noninstrumented fusion, and 7% underwent decompression alone; 97 underwent a multilevel fusion (24%). At 8-year follow-up, 91 of 406 (22%) had a reoperation. Of these patients, 25 occurred within the first year (28%), 49 within 2 years (54%), 64 within 4 years (70%), and 78 within 6 years (86%) (Table 1 ; Figure 1). The reasons for reoperation included recurrent stenosis or progressive spondylolisthesis (45%), complications such as hematoma, dehiscence, or infection (complete list in Table 3) (36%), and the development of a new condition (14%).
Patient-reported outcomes at 8 years were compared between the reoperation and no reoperation groups. For both primary and secondary outcomes (including SF-36 BP, SF-36 PF, ODI, SBI, and satisfaction with symptoms), patients in the reoperation group had significantly less improvement, represented by mean change in scores from baseline to 8-year follow up, (Table 4; Figure 2).
Primary outcomes over time with area under curve P value that compares average 8-year mean change from baseline between reoperation and no reoperation groups.
Risk Factors for Reoperation
Multivariate regression analysis using all of the baseline demographic variables in Table 1 showed that age (P = 0.018, hazard ratio [HR] 0.97); patients with 2/3 moderate or severe stenotic levels (P = 0.02, HR 1.71); predominant back pain (P = 0.023, HR 2.09); no physical therapy (P = 0.017, HR 1.74); absence of neurogenic claudication (P = 0.019, HR 1.89); and greater leg pain score at baseline (P = 0.003, HR 1.39) predicted higher reoperation rates (Table 5). Factors that did not affect the risk for reoperation included: smoking, diabetes, obesity, longer duration of symptoms, or workman’s compensation.
Of the patients who underwent reoperation, 80 of 91 patients had complete information available on reoperation levels. Eighteen of 80 (22.5%) patients had a reoperation at the same level as the index operation, 40 of 80 (50%) had a reoperation of at least one different level than the index operation, and 37 of 80 (46%) had levels unspecified. Of these 80 patients, 13 patients had two reoperations and one had three reoperations (data not shown).
Multivariate analysis was used to further investigate the demographic risk factors for reoperation for patients who received noninstrumented fusion or instrumented fusion. Using all of the demographic factors in Table 1, in the noninstrumented fusion group, patients without neurogenic claudication were more likely to undergo reoperation (P = 0.0034, HR 7.12), as were patients of older age (P = 0.009, HR 0.92). In the instrumented fusion group, patients without neurogenic claudication (P = 0.0098, HR 2.16) and those with a higher baseline SBI score (P = 0.0161, HR 1.06) were more likely to undergo reoperation (data not shown).
This study performed a subanalysis of the 8-year SPORT data to determine baseline risk factors for reoperation in patients treated surgically for DS, and compare outcomes with patients that did not undergo reoperation. Several similar studies have been limited by small demographic groups, homogenous cohorts, or by retrospective analysis.[7,10–13] This is the first subanalysis of its kind using SPORT data.
The incidence of reoperation for patients treated surgically for DS was 22% at 8 years, with 54% of these occurring in the first 2 years. The most common indication for reoperation was progressive spondylolisthesis or recurrent stenosis (45%). These findings are consistent with several reports in the literature with similar patient demographics.[1–3,5,7,10,11,13]
Previously reported risk factors for adjacent segment disease and radiographic degeneration after lumbar fusion include age,[10,11,13] number of levels fused,[9,10,12] preexisting disc degeneration,[14,15] smoking, BMI,and preexisting facet degeneration. The strongest baseline positive predictors for reoperation in this study were lack of neurogenic claudication and back pain predominant symptoms (HR = 1.89, HR = 2.09, respectively). Interestingly, demographic factors including smoking, obesity, diabetes, duration of symptoms, and workman’s compensation were not associated with an increased risk of reoperation. Similarly, surgical factors such as the type of surgery performed, blood loss, time of operation, intraoperative complications, level of fusion, and multilevel laminectomy were not associated with an increased risk for reoperation.
Compared to patients with neurogenic claudication, patients without neurogenic claudication who underwent fusion showed a higher risk for reoperation. They were seven times as likely to undergo reoperation with noninstrumented fusion and twice as likely to undergo reoperation after instrumented fusion. Further studies are needed to address this question, but the findings suggest that fusion with instrumentation may result in fewer reoperations in patients without neurogenic claudication.
Patients who had a reoperation were more likely to have a history of antidepressant use (11% vs. 4%, P < 0.014). However, depression itself was not a significant risk factor. Although other studies have identified the correlation between inferior outcomes and depression for patients undergoing lumbar spine surgery, they have not reported on outcomes in patients taking antidepressants without underlying depression.[1,2,8,16,17] This may indicate that patients with DS on antidepressants for treatment of chronic pain are predisposed to worse clinical outcomes.
Contrary to previous studies, the overall use of instrumentation showed no significant difference in reoperation rates at 8-year follow-up (21% vs. 24%, P= 0.51)[1,2,5,7,11,17] Also, rates of late complications requiring reoperation, (progressive spondylolisthesis, recurrent stenosis, pseudoarthrosis, wound exploration, or other complications) were not significantly different between these two groups (P = 0.60). This is contrary to several reports that suggest instrumented fusion may facilitate the development of complications such as adjacent segment disease that may require a reoperation.[9,13,17–20]
In this analysis, there was a surprisingly low pseudoarthrosis reoperation rate (1%–2%, Table 2) among the entire SPORT DS cohort who underwent fusion. However, patients in the SPORT trial were not routinely imaged during follow-up; thus, pseudoarthrosis rates, particularly those that were asymptomatic, may be underreported. Kornblum et al noted that patients with a solid fusion have significantly better outcomes than patients who develop a pseudoarthosis. Subsequent articles have established no difference in outcome because of fusion technique or use of instrumentation.[17,19] The discrepancy between the present study pseudoarthrosis rate and the historical pseudoarthrosis rate may lie in the fact that this represents the symptomatic pseudoarthrosis rate, which may be different from the radiographic pseudoarthrosis rate, which was evaluated in previous studies. Similar to other studies, number of levels fused was not associated with increased risk of reoperation.
The SF-36 BP and PF, ODI, SBI, and satisfaction with symptoms showed significantly greater improvement in the no reoperation group. These findings are consistent with other reports showing worse outcomes in patients requiring a reoperation.[7,10,11,13] Although both groups of patients improved from baseline, our data indicate that reoperation may not be able to “rescue” the symptoms of an unsuccessful initial surgery.
Martin et al studied reoperation rates in a large retrospective study (n = 24,882) and found patients with spondylolisthesis had a lower reoperation rate after fusion surgery than after decompression alone (17.1% vs. 28%, P = 0.002).[7,13] In a similar study, Lad et al found higher complication rates in patients undergoing arthrodesis compared with decompression alone (8.3% vs. 4.8%; P < 0.0001). These findings were similar to our study, although both of these studies included heterogenous patient groups.
The drop-out rate with any long-term study can also confound the study findings. Patients with worse outcomes may seek treatment elsewhere or not at all, and patients doing well following surgical or nonsurgical intervention may also be less likely to follow-up. Reasons for reoperation are unknown other than broad categories such as recurrent stenosis, progressive spondylolisthesis, infection, or pseudoarthrosis. We do not have information on the specific symptoms or findings that prompted reoperation. Another possible limitation of this study is limited sample size for subgroup analyses, which could limit the ability to identify a significant factor associated with reoperation. However, this study comprises one of the largest series in the literature on a pure population with DS with good follow-up. Finally, radiographic parameters such as sagittal alignment and pseudarthrosis may influence the rate of adjacent and same segment disease,[21,22] yet the limited available radiographic data precluded our ability to include these radiographic analyses.
Overall, this study demonstrated a modest reoperation rate of 22% at the 8-year follow-up for patients treated surgically in the DS arm of the SPORT trial. Reoperations were most commonly because of recurrent stenosis or progressive spondylolisthesis. The strongest predictors for reoperation were patients with predominant back pain symptoms and patients without neurogenic claudication at enrollment. As patients are guided through the shared decision-making process, these risk factors and the understanding of patient-specific outcomes can facilitate making the best decisions between physicians and patients. Future studies should involve larger prospective studies of patient subgroups with DS to improve the individualization of treatment in the clinical setting.