Associations of paraspinal muscle quantity and quality with patient-reported outcomes in patients aged ≥85 years undergoing lumbar spine surgery: an exploratory retrospective study.

Background: Population aging has driven an increase in lumbar spine surgery among the super-elderly. The associations of preoperative paraspinal muscle quantity and quality with patient-reported outcomes in patients aged ≥85 years undergoing lumbar spine surgery remain unclear. We therefore examined these associations at baseline and 1 year after surgery. Methods: This is a retrospective observational study. Forty-three patients aged ≥85 years undergoing surgery for degenerative lumbar disorders were studied. Preoperative L3 muscle quantity [paraspinal muscle index (PMI)] and quality [computed tomography (CT) attenuation (Hounsfield units, HU); magnetic resonance imaging (MRI)-based Goutallier grading] were assessed. Patient-reported outcome measures (PROMs) were collected preoperatively and at 1 year. Associations at baseline, 1 year, and change from baseline (Δ) were examined using Spearman correlations. Multivariable linear and baseline-adjusted Firth logistic regression assessed the preoperative Japanese Orthopaedic Association Back Pain Evaluation Questionnaire (JOABPEQ) low back pain score and treatment effectiveness, respectively. An exploratory receiver operating characteristic (ROC) analysis evaluated HU to identify low preoperative JOABPEQ low back pain scores (≤29). Results: PMI showed no statistically detectable association with any PROM in this sample. In separate exploratory multivariable models, HU and Goutallier grade showed nominal associations with the preoperative JOABPEQ low back pain score, whereas no statistically detectable association was observed for PMI; no muscle measure showed a statistically detectable association with treatment effectiveness after baseline adjustment. HU correlated positively with preoperative JOABPEQ low back pain (ρ=0.379, P=0.02), whereas Goutallier grade correlated inversely with the same PROM (ρ=-0.374, P=0.02) and with preoperative 8-Item Short Form Health Survey physical component summary (SF-8 PCS) (ρ=-0.307, P=0.048). In exploratory ROC analysis, HU showed fair discrimination for low JOABPEQ low back pain scores [area under the curve (AUC) =0.740, 95% confidence interval (CI): 0.586-0.893]; the data-derived cutoff of 33.7 HU requires external validation and is not ready for clinical application. Conclusions: In this small, single-center exploratory study, HU and Goutallier grade showed nominal associations with baseline pain-related disability, whereas PMI showed no statistically detectable association. No muscle measure showed a statistically detectable association with treatment effectiveness after baseline adjustment. Given limited power, no multiplicity adjustment, and potential selection bias, these findings require validation and should not guide preoperative decision-making.