Statistical Analysis

The PSI was dichotomized, with responses 1–3 considered “improved” and response 4 considered “nonimproved.” Responses 1–3 identify patients who are able to subjectively appreciate clinical improvement after surgical intervention. Response 4 alone was sufficient for defining “nonimprovement” given that it explicitly captured patients who felt they were the “same or worse” compared to before surgery.

The PSI dichotomization was treated as the anchor in a receiver operating characteristic (ROC) analysis in order to define a MCID in each of 4 PROMs: EQ-5D index, ODI, leg pain VAS, and low back pain VAS. For each of the 4 PROMs, a ROC curve was plotted, an area under the curve was calculated, and an optimal MCID was computed from both the top-left-corner criteria and the Youden index.13 All computations and statistical tests were performed in R (R Core Team, Vienna, Austria). The pROC package in R was used for all components of the ROC analysis.14