Classifying Patients Operated for Degenerative Lumbar Spondylolisthesis: A Machine-Learning Clustering Analysis to Identify Patterns of Clinical Presentation. (16th November 2020)

• Record Type: Journal Article
• Authors: Chan, Andrew K; Wozny, Thomas; Bisson, Erica F; Bydon, Mohamad; Glassman, Steven D; Foley, Kevin T; Shaffrey, Christopher I; Potts, Eric A; Shaffrey, Mark E; Coric, Domagoj; Knightly, John J; Park, Paul; Wang, Michael Y; Fu, Kai-Ming G; Slotkin, Jonathan; Asher, Anthony L; Virk, Michael S; Kerezoudis, Panagiotis; Alvi, Mohammed A; Guan, Jian; Haid, Regis W; Mummaneni, Praveen V
• Is Part Of: Neurosurgery. Volume 67(2010)Supplement 1
• Publication Date: 2020-11-16
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1093/neuros/nyaa447_687
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract: INTRODUCTION: Comparisons of controlled trials in patients with lumbar spondylolisthesis are difficult due to heterogeneity in the populations studied. METHODS: This is a study of the prospective QOD Spondylolisthesis Module, which includes patients who underwent single-segment surgery for Meyerding grade 1 degenerative lumbar spondylolisthesis. 24-month patient-reported outcomes (PRO) including ODI, NRS back pain (NRS-BP), NRS leg pain (NRS-LP), EQ-5D, and NASS Satisfaction were collected. Age, BMI, ASA grade, baseline NRS-BP, NRS-LP, and baseline responses to each sub-item of the ODI and EQ-5D were introduced into a k-means clustering analysis to identify patterns of clinical presentation. RESULTS: Overall, 507 patients were included (mean age: 62.5 ± 12.2 years; 57.0% women). Clustering revealed 2 distinct cohorts. Cluster 1 (high disease burden) was defined by younger patients, higher BMI and ASA grades, and globally worse baseline PRO responses. Cluster 2 (intermediate disease burden) was defined by older patients, lower BMI and ASA grades, and globally intermediate baseline PRO responses. Baseline radiographic parameters-mm of listhesis and proportion of dynamic listhesis-did not differ between clusters (p > 0.05). The high disease burden cluster had higher blood loss (188.5 ± 197.5 vs. 152.3 ± 166.5ml, P = .03), longer hospital stays (3.2 ± 1.9 vs. 2.3 ± 1.6days, P < .001), and fewer home discharges (88.0% vs. 95.3%, P = .003). Both clusters improved at 24-months for all outcomes ( P < .001). In both univariate and multivariable analyses, absolute mean 24-month ODI, NRS-BP, NRS-LP, and EQ-5D were markedly worse for the high disease burden cluster (adjusted P < .001). However, the high disease burden cluster demonstrated greater 24-month improvements for ODI, NRS-BP, and EQ-5D (adjusted P < .05) and also had a higher proportion reaching ODI minimal clinically important difference (14.3 improvement)(78.6% vs. 60.9%, adjusted P = .001). However, the high disease burden cluster had lower satisfaction [55.1% vs. 67.4% most satisfied(NASS 1), adjusted OR = 0.6;95%CI(0.4-0.9); P = .02]. NRS-LP improvement was similar (adjusted p = 0.30). CONCLUSION: Here, we define two distinct groups of patients-those with high versus intermediate disease burden-who ultimately present for single-segment surgery for lumbar spondylolisthesis. The clusters are associated with significantly different outcomes and it may be helpful for future investigations to consider these clinical phenotypes as distinct entities.