Retrospective analysis underestimates neurological deficits in complex spinal deformity surgery: a Scoli-RISK-1 Study

Abstract

OBJECTIVEThe authors conducted a study to compare neurological deficit rates associated with complex adult spinal deformity (ASD) surgery when recorded in retrospective and prospective studies. Retrospective studies may underreport neurological deficits due to selection, detection, and recall biases. Prospective studies are expensive and more difficult to perform, but they likely provide more accurate estimates of new neurological deficit rates.METHODSNew neurological deficits were recorded in a prospective study of complex ASD surgeries (pSR1) with a defined outcomes measure (decrement in American Spinal Injury Association lower-extremity motor score) for neurological deficits. Using identical inclusion criteria and a subset of participating surgeons, a retrospective study was created (rSR1) and neurological deficit rates were collected. Continuous variables were compared with the Student t-test, with correction for multiple comparisons. Neurological deficit rates were compared using the Mantel-Haenszel method for standardized risks. Statistical significance for the primary outcome measure was p < 0.05.RESULTSOverall, 272 patients were enrolled in pSR1 and 207 patients were enrolled in rSR1. Inclusion criteria, defining complex spinal deformities, and exclusion criteria were identical. Sagittal Cobb measurements were higher in pSR1, although sagittal alignment was similar. Preoperative neurological deficit rates were similar in the groups. Three-column osteotomies were more common in pSR1, particularly vertebral column resection. New neurological deficits were more common in pSR1 (pSR1 17.3% [95% CI 12.6–22.2] and rSR1 9.0% [95% CI 5.0–13.0]; p = 0.01). The majority of deficits in both studies were at the nerve root level, and the distribution of level of injury was similar.CONCLUSIONSNew neurological deficit rates were nearly twice as high in the prospective study than the retrospective study with identical inclusion criteria. These findings validate concerns regarding retrospective cohort studies and confirm the need for and value of carefully designed prospective, observational cohort studies in ASD.