Do psychological assessment tools predict spinal cord stimulation treatment outcomes?

Abstract

To determine whether psychological variables independently predict spinal cord stimulation success or instead moderate pain-response dynamics over time. This retrospective cohort included 217 patients undergoing SCS implantation, with available follow-up in 153 at 6 months and 155 at 12 months. Treatment success was defined as a ≥30% reduction in Numeric Rating Scale (NRS) pain from baseline. Baseline psychological measures included Beck Depression Inventory (BDI) categories (Minimal/Mild/Moderate/Severe), Pain Disability Index (PDI) categories (Low/High), and 10 binary Personality Assessment Inventory (PAI) traits. Descriptive analyses, Firth-penalized logistic regression, linear interaction models, partial Spearman correlations, and multivariable linear regression were performed using complete-case data. Available-case success rates were 38.6% at 6 months (59/153) and 30.3% at 12 months (47/155). At 6 months, baseline NRS was borderline associated with categorical success (OR 1.39, 95% CI 0.99 to 2.01; p=0.056); BDI, PDI, and PAI traits were not significant. In the 12-month model, most psychological variables were not independently associated with success, although one personality trait 'denial of depression' showed a trend toward lower odds of success (OR 0.34, 95% CI 0.09 to 1.11; p=0.074), and age showed a borderline association (p=0.056). In the linear models, the relationship between baseline and follow-up pain appeared strongest in the minimal BDI group and weaker in higher BDI categories. The overall model including these depression-by-pain terms was statistically significant (p=0.005, R²=0.145), though the individual interaction terms did not reach significance. Partial Spearman analysis showed an inverse association between BDI and ΔNRS at 6 months (ρ=-0.199; p=0.023; n=133), but not at 12 months (ρ=-0.060; p=0.493; n=137). Perfectionism and resistance-to-change were associated with smaller 12-month improvement in unadjusted analyses (p=0.0177 and p=0.0378) and remained significant in multivariable modeling (β = -1.195, p=0.010; β = -1.101, p=0.009; R²=0.077; p=0.0025). Psychological measures were not strong independent predictors of binary SCS success. Instead, depressive burden influenced early pain-response coupling, while cognitive-rigidity traits were associated with smaller long-term improvement.