What factors predict the best outcomes for older patients operated on for grade I degenerative lumbar spondylolisthesis? A machine learning analysis from the Quality Outcomes Database.

OBJECTIVE: The prevalence of degenerative lumbar spondylolisthesis in older adults is increasing, yet the factors influencing surgical outcomes remain unclear. The authors' study used a machine learning approach to identify outcome clusters among older patients operated on for grade I lumbar spondylolisthesis. METHODS: The authors analyzed data from the prospective Quality Outcomes Database (QOD) registry for patients aged ≥ 65 years who underwent surgery for grade I degenerative lumbar spondylolisthesis. Principal components analysis was used to generate a composite outcome score from 5 patient-reported outcomes (PROs) collected 24 months postoperatively. Subsequently, K-means clustering was used to stratify patients by outcome, with the optimal number of clusters decided by silhouette scoring. Multivariable logistic regression and Boruta feature selection were used to assess variable importance. RESULTS: Of 608 total patients, 233 were at least 65 years of age and met 24-month follow-up. K-means clustering identified two distinct groups: cluster 1 corresponded to more optimal outcomes (less pain and disability, higher satisfaction and quality of life), and cluster 2 corresponded to suboptimal outcomes. Cluster 1 had significantly greater improvements across all PROs at 24 months (p < 0.001), with 87.1% achieving the minimal clinically important difference (MCID) across both EQ-5D and Oswestry Disability Index (ODI) scores versus 48.8% in cluster 2 (p < 0.001). Clusters were similar in age (mean 72.5 vs 73.4 years, p = 0.24), but cluster 1 had lower baseline back and leg pain scores. Notably, patients in the optimal outcomes cluster were significantly more likely to have received a fusion procedure (70.7% vs 51.2%, p = 0.003). In logistic regression predictor analysis, lower baseline Numerical Rating Scale (NRS)-back pain score (OR 0.70, 95% CI 0.50-0.98, p = 0.04) and addition of fusion (OR 1.70, 95% CI 1.24-2.33, p = 0.001) were significant independent predictors of optimal outcomes, which was subsequently confirmed in Boruta analysis. CONCLUSIONS: The authors' analysis of older (≥ 65 years old) patients undergoing surgery for grade I spondylolisthesis revealed two distinct clusters: one with optimal and the other with suboptimal outcomes. The addition of fusion to a procedure was associated with superior outcomes, as patients who received fusion had a 70% increase in the odds of reaching an optimal outcome. Age was not significantly associated with outcome, implying that there is no clear cutoff for surgical eligibility in older patients. These findings may inform decision-making between decompression alone versus decompression with fusion in older patients.

Duke Scholars

Author Christopher Ignatius Shaffrey Orthopaedic Surgery