Recovery trajectory phenotypes and their baseline predictors following conservative treatment for cervical radiculopathy: A cohort study using latent class growth analysis

BACKGROUND

Recovery after conservative treatment for cervical radiculopathy (CR) is heterogeneous, but VAS-based pain recovery patterns and their baseline predictors remain unclear.

OBJECTIVE

To identify VAS-based pain recovery trajectory phenotypes after conservative treatment for CR, compare clinical outcomes across phenotypes, and explore baseline predictors of unfavorable trajectories.

METHODS

In this single-center retrospective cohort study, consecutive CR patients receiving standardized conservative treatment were enrolled. Latent class growth analysis (LCGA) was applied to Visual Analogue Scale (VAS) scores measured at five time points from baseline to 12 months. Neck Disability Index (NDI) trajectories were modeled in parallel to assess robustness. Multinomial logistic regression was used to identify predictors of unfavorable trajectories, with multiple imputation and bootstrap resampling for internal validation.

RESULTS

Among 195 patients, LCGA identified four VAS-based pain recovery trajectories: Rapid Recovery (31.8%), Gradual Improvement (37.9%), Slow Incomplete Recovery (19.5%), and Persistent Poor Outcome (10.8%). Composite treatment-failure rates at 12 months were progressively higher across these groups (4.8%, 21.6%, 52.6%, and 85.7%, respectively; P < 0.001), and conversion-to-surgery rates ranged from 1.6% to 42.9%. Longer symptom duration, higher baseline NDI, multi-segment involvement, and severe foraminal stenosis were independently associated with the Persistent Poor Outcome trajectory. The bias-corrected C-statistic was 0.76 (95% CI: 0.70-0.82).

CONCLUSION

Four VAS-based pain recovery trajectories were identified after conservative treatment for CR. Baseline functional and imaging features may offer preliminary prognostic information but capture only part of multidimensional recovery. Prospective validation incorporating psychosocial factors is required.