Performance of a modified respiratory index of severity in children (RISC) score to predict poor outcomes in children admitted with lower respiratory tract infections at Kenyatta National Hospital

Abstract

Background: Respiratory tract infections including pneumonia are a leading cause of mortality and morbidity in children under five years of age. A large burden of these cases is in Africa. Clinical prediction scores such as respiratory index of severity in children (RISC) score have been developed to predict outcomes in children with respiratory tract infections. RISC score is a composite score that uses the oxygen saturation, ability to feed, presence of chest indrawing and nutritional status. The score ranges from zero to six with six being the poorest score. Such a score would make it easier for clinicians to predict outcomes and prioritize care based on available resources to reduce morbidity and mortality Objectives: To determine the RISC scores and evaluate its validity to predict poor outcomes in children aged 2-59 months who are hospitalized with acute lower respiratory tract infections at Kenyatta National Hospital. Methods: This study recruited children aged 2-59 months presenting with symptoms of cough or difficulty in breathing for a duration of less than two weeks that was severe enough to require hospitalization. Demographic, clinical information and the child’s RISC score was also captured in a pretested questionnaire on admission. Children were followed up in the ward for outcomes of interest that included death within seven days of admission, prolonged hospital stay of more than one week and mechanical ventilation. We analyzed the rates of poor outcomes and relationship between the observed and expected outcomes based on the modified score Results: A total of 146 children were recruited. The male to female ratio was 1:0.8 with a median age of 10 months. The modified RISC scores ranged between zero and six with a median RISC score of two. Thirty nine percent of the study participants had a poor outcome with a mortality rate of 0.7%. The score predicted mortality perfectly for higher risk groups based on the modified score (≥ 4) but performed poorly for values representing low risk (0 to 1) or moderate risk (2 to 3) of poor outcomes. The Hosmer-Lemeshow test for goodness-of-fit provided evidence of differences between the observed and estimated frequencies (chi square = 4.45; DF=2; p value = 0.35).