Predictive value of Acute Physiology and Chronic Health Evaluation II (APACHE II) scoring in patients admitted to intensive care units

BackgroundRisk stratification tools are essential for guiding care and allocating limited resources in intensive care units (ICUs). Evidence on the performance of the Acute Physiology and Chronic Health Evaluation II (APACHE II) score in Nepal is limited. We evaluated the association of APACHE II measured within 24 hours of ICU admission with mortality, ICU length of stay, and discharge disposition in a tertiary hospital in Nepal. Methods and findingsWe conducted a prospective observational cohort study of consecutive adult patients admitted to three multidisciplinary ICUs at a tertiary hospital in Kathmandu, Nepal. Patients with ICU stay <24 hours and those admitted to coronary or neonatal ICUs were excluded. APACHE II was calculated from the worst values in the first 24 hours. The primary outcome was ICU mortality; secondary outcomes were ICU length of stay and discharge disposition (home, ward, high-care/step-down, or in-hospital death). Discrimination for mortality was assessed using receiver operating characteristic (ROC) analysis. Associations with length of stay were examined using linear regression; discharge disposition was evaluated using multinomial logistic regression. Among 200 patients (54% male; mean age 54.7 {+/-} 21.0 years), ICU mortality was 23.0% (46/200). Non-survivors were older than survivors (mean difference 13.31 years; 95% CI 6.60-20.02; p < 0.001). The mean APACHE II score was 13.19 {+/-} 7.89 overall, higher in non-survivors vs survivors (19.85 {+/-} 7.14 vs 11.21 {+/-} 6.98; p < 0.001). APACHE II discriminated ICU mortality well (AUC 0.806); a cutoff [&ge;]14.5 yielded 76.1% sensitivity and 71.4% specificity. Mortality increased across APACHE II strata (trend p < 0.001). Glasgow Coma Scale scores were lower in non-survivors (10.85 {+/-} 4.43) than survivors (13.29 {+/-} 3.11; p < 0.001). Higher APACHE II scores were associated with in-hospital death versus ward discharge, but not with high-care versus ward (OR 1.063; 95% CI 0.994-1.136; p = 0.074). Limitations include the single-center design, modest sample size, and short study duration, which may limit generalizability. ConclusionsAPACHE II measured within 24 hours of ICU admission demonstrated good discrimination for ICU mortality and a modest association with ICU length of stay in a Nepalese tertiary setting, while being less informative for intermediate discharge dispositions. Together with Glasgow Coma Scale, APACHE II offers a pragmatic, cost-effective approach to early prognostication and resource planning in resource-limited ICUs. Multicenter studies are warranted to validate these findings across Nepal.