Evaluating the influence of patient traits versus surgical intervention on outcomes across spine surgeries: a latent state-trait analysis.

Abstract

Patient-reported outcomes have been used to assess surgical success from a physical and mental health standpoint. PROMIS-10 surveys offer a standardized approach to assess outcomes. The baseline variances inherent to patients (trait) and fluctuations based on external forces (state) have not been established. The purpose of this study was to identify the contributions of a patient's state (surgery) and trait (patient characteristics) across different spine surgeries on a patient's outcome. Patients who underwent spine surgery between January 2007 and December 2024 with PROMIS-10 mental and physical health scores at 4 different time points 2-5 years and 0-6 months preoperatively and 0-6 months and 6-24 months postoperatively were included. Structural equation modeling with the lavaan R code package was used for latent state-trait analysis (LSTA). Standardized root-mean square residual was used to characterize model fit. Loadings of ≥ 0.5 were used to determine latent trait and state dependence. In total, 5304 surgeries across 4248 patients (51.3% female, mean age ± SD 64.54 ± 11.65 years) were included. All surgery types, except multilevel thoracic kyphoplasty, had a state variance loading of ≥ 0.5 across all PROMIS-10 sections. The mean percentage of variance explained by trait and state factors was 56.6% ± 15.3% and 43.4% ± 15.3% for physical health and 44.7% ± 7.38% and 55.3% ± 7.38% for mental health, respectively. LSTA of the PROMIS-10 instrument provides context to highlight which surgical types may be most impactful in affecting patient-reported outcomes. For those with high state dependence, the surgery or surgical experience influenced reported outcomes, while those with high trait dependence were primarily driven by innate patient characteristics.