Lumbar spinal stenosis (LSS) is a narrowing of the central spinal canal or lateral canal(s) caused by bone or soft tissue encroachment that causes mobility limitation because of pain and/or lower extremity paresthesias, i.e., neurogenic claudication. Lumbar stenosis impacts over 10% of older adults leading to millions of hospital visits annually (1–3). Decompressive laminectomy (DL) for LSS is the most common indication for spine surgery in older adults (1, 2). While DL provides symptom relief (i.e., reduction of pain and paresthesias) in some patients, nearly one-half of veterans undergoing DL will not achieve functional improvement at 12 months (4). Few studies have evaluated longer-term outcomes following DL, beyond 1 year, although some indicate that DL failure rates increase with extended follow-up beyond 1 year (5–7). As many as one in three DL for LSS patients undergo additional surgery within 5 years of the original procedure (8). While failed DL is often followed by more complex surgical procedures such as fusion, outcomes associated with these procedures are not superior (9).

A comprehensive understanding of factors associated with outcomes beyond 1 year is needed to optimize patient selection, surgical decision-making, and associated utilization of healthcare resources. Preliminary data raise the possibility that the more extensive the pathology (i.e., the greater the number of spinal levels decompressed), especially in the setting of altered spinal biomechanics, the greater the risk of DL failure (10). Others suggest that selecting patients with severe anatomical stenosis, no back pain, symptoms for less than 4 years, and no other conditions that impair walking will have superior DL outcomes (7). The burden of medical comorbidity also has been highlighted as a risk factor for worse long-term DL outcomes (11). Studies that examine a comprehensive set of predictors, including biopsychosocial factors, of long-term outcomes in patients undergoing DL for LSS are lacking (12). Such studies may inform patient selection and help identify patients likely to benefit from prehabilitation prior to DL for LSS.

We recently published the results of a prospective cohort study of Veterans with LSS who underwent DL. Among a comprehensive set of factors evaluated preoperatively, opioid use, apparent leg length inequality, and low self-efficacy were the strongest predictors of outcome 1 year following DL (3). We now examine predictors of 4-year DL success.

This prospectively collected cohort of United States Veterans included 193 participants with symptomatic spinal stenosis who underwent elective DL surgery with 1-year follow-up of whom 110 (focused herein) had 4-year follow-up. Table 1 summarizes the baseline participant characteristics. Overall, the average age was 65.1 ± 9.9, BMI 32.3 ± 5.3 with 70.9% being Caucasian. The majority (60.0%) of participants previously smoked. Only 18.2% reported smoking currently. All participants exhibited neurogenic claudication with the majority (55.5%) endorsing concurrent back and leg pain with imaging respectively confirming moderate (25.5%) or severe (60.9%) stenosis. Those included in the 4-year analysis had greater stenosis, symptom severity, comorbidity, poorer self-efficacy, poorer mood, and lower treatment credibility ratings (all p < 0.05; Supplementary Table S1).

Of the 68 participants demonstrating surgical success at 1 year, 11 did not sustain surgical success at 4 years (16%). Of the 69 participants demonstrating surgical success at 4 years, 12 did not demonstrate initial surgical success at 1 year (17%). Comparing surgical success rates at 1 and 4 years, there was no statistically significant difference (p = 0.8348; Table 2). In addition, 1- vs. 4-year success rates in BSS physical function (50.9 vs. 49.1%; p = 0.6831), BSS symptoms severity (65.5 vs. 67.3%; p = 0.6171), and BSS satisfaction (68.2 vs. 71.8%; p = 0.2850) were not statistically different (Supplementary Tables S2–S4).

Figure 2 highlights the BSS metric associations at 4 years. Table 3 summarizes the bivariate associations between surgical success and preoperative factors. A 0.2-point greater (worse) baseline BSS physical function score was associated with a 27% increase in odds of physical function improvement [OR 1.27 (1.07–1.51); p = 0.0063]. In addition, being on a non-CNS pain medication conferred over twice the odds of improving BSS physical function [OR 2.43 (1.13–5.26); p = 0.0237], while 5 points in the PTSD scale decreased the odds by 13% [OR 0.87 (0.78–0.98); p = 0.0186].

Having a 0.2-point greater baseline BSS symptom severity score (i.e., worse symptoms) was associated with 24% increased odds of improving BSS symptom severity post-DL [OR 1.24 (1.05–1.46); p = 0.0090]. A 10-point difference in treatment credibility was associated with a 24% increase in odds of symptom improvement [OR 1.24 (1.02–1.52); p = 0.0319]. In addition, having a college education was associated with an over threefold increase in odds of improvement [OR 3.36 (1.39–8.09); p = 0.0069]. Conversely, an additional year of symptom duration was associated with a 3% decrease in odds [OR 0.97 (0.94–1.00); p = 0.0409] of an improvement in BSS symptom severity post-DL.

Many variables demonstrated an association with BSS satisfaction. A 10-point greater CPSE physical function score (i.e., higher physical function disability) was associated with 20% greater odds [OR 1.20 (1.01–1.43); p = 0.0358] of BSS satisfaction. A 10-point greater CPSE coping score (i.e., better coping skills) was associated with a 23% increase in odds [OR 1.23 (1.04–1.47); p = 0.0165], a college education with an over twofold increase in odds [OR 2.81 (1.16–6.81); p = 0.0217], severe stenosis with an almost sevenfold increase in odds [OR 6.93 (1.55–31.1); p = 0.0114], and a 10-point increase in the preoperative expectations of postsurgical improvement score associated with a 23% increase in odds [OR 1.23 (1.00–1.51); p = 0.0456] of achieving BSS satisfaction. Conversely, each 1 kg/m² increase in BMI was associated with a 9% reduction in satisfaction [OR 0.91 (0.83–0.98); p = 0.0172], a 10-point greater BSS physical function score lowered the odds by 18% [OR 0.82 (0.68–0.99); p = 0.0431], and a 10-point greater BSS symptom severity resulted in a decrease of 29% [OR 0.71 (0.59–0.85); p = 0.0003]. Other factors associated with lower satisfaction included a 2 point greater coping strategies questionnaire (CSQ) score, which reduced the odds of achieving satisfaction by 14% [OR 0.86 (0.78–0.95); p = 0.0018], a year longer symptom duration with a decrease of 4% [OR 0.96 (0.93–1.00); p = 0.0309], 1 point in the widespread pain index with a 17% reduction [OR 0.83 (0.73–0.94); p = 0.0039]; and fibromyalgia, which decreased the odds of achieving satisfaction by 72% [OR 0.28 (0.12–0.70); p = 0.0064].

Factors demonstrating an association with improvements in at least two of three criteria, and thus overall surgical success at 4 years, include having some college education [OR 2.93 (1.27–6.77); p = 0.0120], moderate or severe stenosis on imaging (OR 7.16–7.39; p = 0.0195–0.0260), and having greater expectations for improvement and reduced limitations (OR for 10 points 1.24–1.35; p = 0.0048–0.0423). Prolonged symptom duration of an additional year [OR 0.96 (0.93–0.99); p = 0.0231] was associated with reduced odds of surgical success.

Multivariable models showed that a combination of several preoperative biopsychosocial factors could be used to predict the likelihood of 4-year surgical success (Table 4). CPSE pain management [adjusted OR or AOR for 10 points 0.81 (0.66–0.98); 0.0316], symptom duration [AOR for 12 months 0.96 (0.92–0.99); p = 0.0208], expectation of reduced limitations [AOR for 10 points 1.51 (1.15–1.98); p = 0.0031], and moderate or severe stenosis (AOR 14.4–17.4; p = 0.0045–0.0055) constituted a parsimonious subset of predictors of overall surgical success with an accuracy of 0.777 as indicated by the area under receiver operator characteristic curve (AUROC). Other criteria for improvement had similar levels of accuracy (AUROC 0.753–0.836).

In United States Veterans with symptomatic LSS, 63% of participants demonstrated sustained DL success at 4 years, similar to the success rate at 1 year (3). Noteworthy preoperative characteristics associated with a higher likelihood of sustained, 4-year, surgical success included lower self-efficacy for pain management, shorter duration of symptoms, greater treatment credibility, and having moderate to severe anatomical stenosis.

Similar to the 1-year success rate, only about 50% of participants experienced significant functional improvement, a key outcome for older adults. Having worse baseline physical function and higher baseline self-efficacy for function were associated with a higher likelihood of significant functional improvement. Studies have highlighted the potential benefits of cognitive and physical prehabilitation prior to and immediately following spinal surgery (35–39), although heterogeneous methods prevent definitive conclusions from being drawn. Our findings suggest the potential value of future studies designed to evaluate the impact on DL outcomes of prehabilitation focused on enhancing self-efficacy.

Four-year improvements in symptom severity (e.g., pain and paresthesias) mirrored those at 1 year (3). Approximately 2/3 of participants had sustained a reduction in symptom severity. Significant predictors of 4-year symptom severity reduction were greater preoperative symptom severity, shorter symptom duration, and greater treatment credibility.

Satisfaction at 4 years post-DL also mirrors those reported at 1 year post-DL (3), with approximately 69% reporting satisfaction at 1 year and 72% reporting satisfaction at 4 years. Approximately 13% of participants at 1 year and 12% at 4 years post-DL reported being satisfied despite experiencing no significant improvement in pain or function. Predictors of significant satisfaction included lower BMI, lesser preoperative symptom severity, greater treatment credibility, and having moderate to severe anatomical stenosis. It has been highlighted in non-surgical patients with chronic pain that symptomatic improvement and satisfaction are not necessarily correlated and that the patient–provider relationship plays a prominent role in determining satisfaction with care (40). Future studies should include such contextual factors as a part of preoperative assessment.

Our findings suggest the importance of preoperative counseling that emphasizes personalized goals. If functional improvement is the patient's main goal, then preoperative efforts to optimize self-efficacy for function may be important. If reduction of symptoms is the patient's main goal, encouraging patients with more severe symptoms and anatomical pathology toward surgery may need to be considered. Future studies that include more participants are needed to validate our observations and evaluate the impact on patient outcomes of a personalized approach to care.

We note that approximately 20% of participants who demonstrated surgical success at 1 year did not demonstrate success at 4 years and vice versa. Despite most revision surgeries for simple DL being performed prior to 1 year (41), our data suggest that continued observation (vs. operative intervention) may be prudent for some participants with potentially delayed improvement. Factors that predict deterioration following initial success also require identification. While a more detailed analysis of participants with delayed improvement and delayed deterioration is beyond the scope of the present study, further investigation is warranted.

The main strength of our study was the rigorous assessment of a comprehensive set of biopsychosocial predictors, and additional preoperative factors, relevant to older adults. To our knowledge, we included the most comprehensive set of preoperative factors that have been studied in patients with LSS undergoing DL. Several study limitations also should be noted. The 4-year cohort represented approximately 57% of the original 1-year cohort, and there were some differences in baseline characteristics as noted earlier. However, the direction of the differences does not indicate the same informative censoring in typical longitudinal studies on aging, where the frailest participant dropout. In addition, our study participants were predominantly male veterans; thus, our findings may not generalize to the larger community of individuals undergoing DL for LSS. Although we observed acceptable predictive accuracy in multivariable models, there are very likely key extra-skeletal predictive factors that we did not collect such as muscle health, genetic profiles, and other biological parameters. Future studies should be conducted on a larger cohort with an even more comprehensive set of preoperative factors to optimize predictive accuracy. Estimates of predictive accuracy from an independent validation sample are more credible and necessary before definitive prediction rules can be considered.

Preoperative characteristics may predict long-term DL surgical success at 4 years. Further study is needed to not only establish predictive accuracy but also the impact of modifying presurgical risk factors on surgical outcomes.