EVALUATING THE PERFORMANCE OF APACHE II AMONG CRITICALLY ILL PATIENTS ADMITTED AT MOI TEACHING AND REFERRAL HOSPITAL, ELDORET – KENYA

Abstract

Objective: This study evaluated the performance of the acute physiology and chronic health evaluation (APACHE) II model among critically ill patients admitted at Moi teaching and referral hospital (MTRH) intensive care unit (ICU) by assessing its calibration and discriminative ability.
Design: The study was a prospective analytical cross-sectional study
Setting: The study was carried out at MTRH, a public tertiary teaching hospital in Eldoret, Kenya. It has a 20-bed mixed use ICU admitting both surgical and medical patients. Participants: A total of 416 participants aged over 18 years were consecutively recruited. Main outcome measures: For the categorical variables, odds ratios were calculated at 95% confidence interval and p-values <0.05 considered statistically significant. Area under receiver operating characteristics (AUROC) curve was used to assess the discriminative ability of the model to distinguish between survivors and non-survivors. Hosmer-Lemeshow chi square statistic was used to assess the goodness of fit of the model.
Results: Of the 416 participants, there was a 57% (n=237) male majority. The ICU mortality rate was at 29% (n=120). Fever (AOR 1.2, 95% CI 0.1 – 13.7), low mean arterial pressure (AOR 1.8, 95% CI 0.5 – 6.1), tachycardia (AOR 1.9, 95% CI 0.9 – 4.1), hypoxaemia (AOR 1.1, 95% CI 0.5 – 2.5), acidosis (AOR 3.2, 95% CI 1.5 – 6.8) and hyponatremia (AOR 1.2, 95% CI 0.4 – 3.2) were associated with higher odds mortality. The AUROC curve yield for the model was 0.77 whilst the p-value for the Hosmer-Lemeshow statistic was 0.03, which showed a poor calibration.
Conclusion: The APACHE II has acceptable accuracy in discrimination of
survivors and non-survivors and can be used as a clinical adjunct in decision making in critically ill patients. Additional studies in resource limited settings required in order to improve its calibration