Antibiotic effectiveness and risk of illness progression for children with lower respiratory infections presenting to primary care: prospective cohort study and trial

Problem

Antibiotic resistance is a global public health threat. Antibiotics are very commonly prescribed for children presenting with uncomplicated lower respiratory tract infections (LRTI) but there is little randomised evidence for children, and randomised trials commonly recruit selected populations. We estimated the effectiveness of antibiotics using both trial and observational data. We assessed the external validity of the STARWAVe prediction rule, and developed a new model to predict illness progression.

Approach

Children aged one to twelve presenting to UK general practices with an acute LRTI were randomised to receive Amoxicillin 50mg/kg/day in divided doses for 7 days, or placebo. Children not randomised (either ineligible or clinician/parent choice) participated in an observational study collecting the same data. Propensity scores controlled for confounding by indication in the observational data. The primary outcome was the duration of symptoms rated moderately bad or worse (measured using a validated diary). Secondary outcomes were reconsultations for new or worsening symptoms; side effects, and illness progression requiring attendance at, or admission to, hospital.

Findings

432 children entered the trial and 326 children the observational study. The estimate of benefit of antibiotics for the primary outcome were similar for trial alone (trial Hazard Ratio (HR) 1.13, 95% CIs 0.90,1.43) and when also including the observational data HR 1.16 (0.95, 1.41). The STARWAVe model had moderate performance in predicting illness progression (AUROC 0.66;0.50, 0.77). A new, internally validated, prognostic model for illness progression (consisting of 7 variables: baseline severity/respiratory rate (difference from normal for age)/duration of prior illness/oxygen-saturation/sputum-rattly chest/passing urine less often/diarrhoea) had good discrimination (bootstrapped AUROC 0.83) and calibration. A 3 item model (respiratory rate ; oxygen saturation; sputum-rattly chest) also performed well (AUROC 0.81), as did a score (ranging from 19 to 102) derived from coefficients of the model (AUROC was 0.78; 0.68, 0.88): a score of less than 70 classified 89% (600/674) of children having a low risk (<5%) of progression of illness.

Consequences

Amoxicillin for uncomplicated chest infections in children is unlikely to be clinically effective. A simple prognostic score could be useful as a tool to help guide clinical management.