Central sensitization-related symptoms in patients undergoing lumbar spine surgery: development and internal validation of a clinical prediction rule for first-stage screening.

Abstract

Central sensitization (CS)-related symptoms are associated with poor outcomes after lumbar spine surgery, but universal screening may be inefficient when only a minority of patients are affected. This study aimed to develop and internally validate a clinical prediction rule (CPR) to screen for elevated CS-related symptom burden (Central Sensitization Inventory-9 [CSI-9] ≥ 20) using routinely collected clinical variables. In this cross-sectional diagnostic prediction model study, 674 consecutive patients scheduled for lumbar spine surgery were included; 97 (14.4%) had CSI-9 scores ≥ 20. Seven candidate predictors were evaluated using logistic regression with backward elimination. A 3-item CPR was derived by dichotomizing retained predictors at Youden-optimal cutpoints. Internal validation used 1,000 bootstrap resamples. Three predictors were retained: age ≤ 60 years, ≥ 5 pain sites, and sleep disturbance (Oswestry Disability Index sleep item ≥ 1). The CPR showed discrimination comparable to the full model (AUC, 0.730; DeLong P = 0.872). At the optimal cutoff (≥ 2 points), sensitivity was 73.2%, specificity 65.0%, and negative predictive value 93.5%. Low-risk patients (score 0-1; 59.5% of the cohort) had an event rate of 6.5%. Decision curve analysis confirmed positive net benefit across threshold probabilities of 5-40%. A simple 3-item CPR demonstrated moderate discrimination with high negative predictive value for elevated CS-related symptom burden before lumbar spine surgery. The rule may serve as a practical first-stage screen to guide selective CSI-9 administration.