Title:
Neurological Recovery After Surgery for Traumatic Spinal Cord Injury: Associated Factors in an Ivorian Cohort

Authors:
TETI Faozo Stéphane Landry, DONGO Koffi Yves Soress, FIONKO Yao Bernard, YAO Konan Serges, KEKE Kouadio Jean-Baptiste, HAIDARA Aderehime, Cote d'Ivoire

Abstract:

Objective. To identify factors independently associated with neurological recovery, measured as ASIA grade improvement, after surgery for traumatic spinal cord injury (SCI) at an Ivorian referral centre.

Methods. Single-centre cohort (Bouaké University Hospital; retrospective phase 2017–2021, prospective phase January 2025–June 2026), reported according to STROBE. Of 345 operated patients, we analysed the 248 survivors with an initial deficit (ASIA A–D). The primary outcome was recovery, defined as an improvement of at least one ASIA grade between admission and the last documented assessment (median follow-up 45 days). Predictors were admission ASIA grade (complete vs incomplete injury), lesion level, AOSpine classification, time to surgery (counted from hospital admission) and age. We fitted a multivariable logistic regression model and assessed apparent discrimination (AUC) and calibration (Hosmer-Lemeshow test). Exploratory analyses tested alternative codings of surgical delay.

Results. Median age was 35 years [IQR 25–48]; 80.9% were men. A complete injury (ASIA A) was present in 41.7% and median time to surgery was 3 days [2–4]. Recovery occurred in 111 of 248 survivors with a deficit (44.8%), with a gradient by initial severity (ASIA A 30.9%; B 45.8%; C 56.6%; D 56.6%). In multivariable analysis (n = 247; 110 events), only incomplete injury (ASIA B–D vs A) was associated with recovery (adjusted OR 2.50; 95% CI 1.41–4.43; p = 0.002): 53.2% of incomplete injuries recovered versus 30.9% of complete injuries. We detected no association with time to surgery ≤ 3 days (adjusted OR 0.70; 95% CI 0.39–1.27), AOSpine type, lesion level or age. Apparent discrimination was modest (AUC 0.65), with no major calibration defect (Hosmer-Lemeshow p = 0.98). Alternative codings of delay (continuous, log-transformed, 24- and 48-hour thresholds, quartiles) did not change these findings.

Conclusion. In this West African surgical cohort, neurological recovery depended on initial severity. We detected no independent association with surgical timing or injury morphology, but estimates are imprecise and an important effect of timing cannot be excluded. The modest discrimination suggests unmeasured factors (MRI contusion extent, rehabilitation, secondary complications); prospective studies should test them.

DOI: http://dx.doi.org/10.51505/ijmshr.2026.10511