Use of clinical prediction rules after spinal cord injury to guide early locomotor training

Context Determining walking prognosis during early recovery phases after spinal cord injury (SCI) is challenging.Objective We retrospectively examined the use of clinical prediction rules (CPRs) for a cohort of patients with SCI undergoing locomotor training during inpatient rehabilitation.Methods Patients participating in a randomized controlled trial received overground robotic exoskeleton (ORE) or usual care (UC) locomotor training during inpatient rehabilitation. Patient data extracted from the medical record (Admission ISNCSCI scores, age) were utilized to populate each of three CPRs. Admission and discharge Walking Index for spinal cord injury-II (WISCI-II) and CARE Tool (walking) items characterized walking ability.Results Eighty-nine inpatients with SCI [AIS B (20%), C (28%), and D (52%)] were included [males (79%), median age 56 years (IQR: 33, 66), tetraplegic (61%), median time since injury 23 days (17, 42) and median length of stay 36 days (28, 54)]. Accuracy of walking prediction by the CPRs was 71.9%, with 55.4% specificity and 100% sensitivity. Fifty-eight patients were predicted to walk independently 1-year post-injury, with the 33 (56.9%) walking independently by inpatient rehabilitation discharge [ORE (AIS C = 4, AIS D = 19) vs. UC (AIS D = 10)] being younger [median age = 46 vs. 62, P = 0.042], shorter time from injury to first locomotor training [median days = 29 vs. 40, p = 0.001], AIS D [88% vs. 54%, p = 0.009], and higher admission WISCI-II and CARE Tool scores [p < 0.05].Conclusions Integration of CPRs into clinical practice appears to differentiate those with higher versus lower walking potential and may guide therapists to better prioritize locomotor interventions earlier in recovery.