Organ support and early death in a Nigerian tertiary intensive care unit: A two-year retrospective cohort of 255 admissions

Abstract Introduction : Patient-level outcome data from Nigerian intensive care units (ICUs) are scarce, limiting service planning and quality improvement. Outcome depends on case mix and organisation, so local data are needed. Objectives : To describe case mix, organ-support use and resource use in our ICU; to identify factors independently associated with ICU death; and to establish a baseline for re-audit. Methods : Retrospective cohort of all 255 consecutive admissions to the ICU of Federal Medical Centre, Ebute Metta, between May 2022 and May 2024, with no exclusions. The primary outcome was ICU mortality. Factors associated with death were examined by multivariable logistic regression (age, sex, mechanical ventilation, inotropes, medical neurological diagnosis). Discussion : Median age was 51 years (IQR 37–65), and medical neurological conditions, mainly stroke, were the commonest diagnosis group (30.6%). Mechanical ventilation was used in 65.1% and inotropes in 33.7%. ICU mortality was 57.6% (95% CI 51.5–63.6), rising from 33.3% with neither intervention to 81.9% with both. Ventilation (adjusted OR 3.32, 95% CI 1.83–6.02), inotropes (3.02, 1.61–5.67) and a medical neurological diagnosis (2.47, 1.32–4.63) were independently associated with death; age and sex were not. Fifteen of 16 patients staying under 24 hours died, and 31.3% of deaths occurred within 48 hours of admission. Daily census never exceeded 7 in a ten-bed unit (mean occupancy 29%), so capacity was set by staffing and working equipment rather than beds. Organ support marks severity rather than causing death, and deaths this early implicate the pathway into critical care as much as care within the unit. Conclusion : Mortality was driven by severity of illness rather than age and was concentrated early after admission. Routine admission severity scoring and recording of referral-to- admission time, both achievable with existing resources, are needed before this mortality can be benchmarked.