Associations of modic changes, multifidus degeneration, and host factors with reoperation for recurrent lumbar disc...

Recurrent lumbar disc herniation (rLDH) remains a clinically important cause of reoperation after primary microdiscectomy, but its basis is likely multifactorial and extends beyond disc-centred structural features.

In this retrospective 1:1 propensity-matched case-control study, 146 registry patients were analysed: 73 who underwent reoperation for rLDH and 73 single-operation controls with at least 24 months of recurrence-free follow-up. Matching used age, sex, and operative level. Preoperative imaging variables included Modic changes, herniation location, disc morphology, disc height index, and multifidus fatty degeneration. Body mass index (BMI), active smoking, and occupational workload were analysed as host-related exposures. Independent associations with reoperation were assessed using multivariable logistic regression, with model estimates interpreted as within-sample associations given the matched case-control design. Model discrimination and internal validation were evaluated with receiver operating characteristic analysis and bootstrap resampling. Additional analyses included standardised mean difference diagnostics before and after matching, multicollinearity assessment, missing-data evaluation, and sensitivity analyses.

Compared with controls, reoperated patients had higher BMI, more active smoking, more Modic grade 2-3 changes, and more clinically significant multifidus degeneration. Independent factors associated with reoperation were Modic grade 2-3 changes (OR 18.02, 95% CI 5.89-55.15), active smoking (OR 7.31, 95% CI 2.25-23.81), heavy occupational workload (OR 5.44, 95% CI 1.50-19.78), multifidus degeneration grade ≥ 2 (OR 4.40, 95% CI 1.68-11.51), and higher BMI per 1 kg/m² (OR 1.36, 95% CI 1.16-1.60). These effect sizes should be interpreted as relative enrichment within the matched cohort rather than as population-level estimates. The model showed good apparent discrimination (AUC 0.88, 95% CI 0.81-0.94) and acceptable internal validation (optimism-corrected AUC 0.85, 95% CI 0.80-0.90) with acceptable calibration. Sensitivity analyses supported the robustness of the primary associations, and no meaningful multicollinearity or systematic missing-data bias was detected.

Reoperation for rLDH was associated with a combined structural-neuromuscular and host-related profile. These findings represent within-sample associations requiring external prospective validation and may support perioperative counselling and risk stratification discussions.