Background context: There is significant variability in minimal clinically important difference (MCID) criteria for lumbar spine surgery that suggests population and primary pathology specific thresholds may be required to help determine surgical success when using patient reported outcome measures (PROMs).

Purpose: The purpose of this study was to estimate MCID thresholds for three commonly used PROMs after surgical intervention for each of four common lumbar spine pathologies.

Study design/setting: Observational longitudinal study of patients from {BLINDED} national registry.

Patient sample: Patients undergoing surgery from 2015-2018 for lumbar spinal stenosis (LSS; n=856), degenerative spondylolisthesis (DS; n=591), disc herniation (DH; n=520) or degenerative disc disease (DDD n=185) were included.

Outcome measures: PROMs were collected pre-surgery and one-year post-surgery: the Oswestry Disability Index (ODI), and back and leg Numeric Pain Rating Scales (NPRS). At 1-year, patients reported whether they were ‘Much better’/’Better’/’Same’/’Worse’/’Much worse’ compared to before their surgery. Responses to this item were used as the anchor in analyses to determine surgical MCIDs for benefit (‘Much better’/’Better’) and substantial benefit (‘Much better’).

Methods: MCIDs for absolute and percentage change for each of the three PROMs were estimated using a receiving operating curve (ROC) approach, with maximization of Youden’s index as primary criterion. Area under the curve (AUC) estimates, sensitivity, specificity and correct classification rates were determined. All analyses were conducted separately by pathology group.

Results: MCIDs for ODI change ranged from -10.0 (DDD) to -16.9 (DH) for benefit, and -13.8 (LSS) to -22.0 (DS,DH) for substantial benefit. MCID for back and leg NPRS change were -2 to -3 for each group for benefit and -4.0 for substantial benefit for all groups on back NPRS. MCID estimates for percentage change varied by PROM and pathology group, ranging from -11.1% (ODI for DDD) to -50.0% (leg NPRS for DH) for benefit and from -40.0% (ODI for DDD) to -66.6% (leg NPRS for DH) for substantial benefit. Correct classification rates for all MCID thresholds ranged from 71%-89% and were relatively lower for absolute vs. percent change for those with high or low pre-surgical scores.

Conclusions: Our findings suggest that the use of generic MCID thresholds across pathologies in lumbar spine surgery is not recommended. For patients with relatively low or high pre-surgery PROM scores, MCIDs based on percentage change, rather than absolute change, appear generally preferable. These findings have applicability in clinical and research settings, and are important for future surgical prognostic work.