A clinical decision tool including a decision tree, point-of-care testing of CRP, and safety-netting advice to guide antibiotic prescribing in acutely ill children in primary care in Belgium (ARON): a pragmatic, cluster-randomised, controlled trial

Abstract

Background

Antimicrobial resistance is a global health threat. Many children with acute illness in ambulatory care are unnecessarily prescribed antibiotics. We assessed the clinical effectiveness of a clinical decision tool for these children, including a validated decision tree, guided point-of-care C-reactive protein testing (POCT of CRP), and safety-netting advice.

Methods

ARON was a multicentre, unblinded, pragmatic, cluster-randomised, controlled trial conducted at eligibile Belgian general practitioner and community paediatrician practices able to recruit children with acute illness consecutively, and not already doing POCT of CRP. Practices were allocated (1:1) with equal size (n=4) block randomisation to the clinical decision tool or usual care, stratified by recruiting academic centre. Children with acute illness aged 6 months to 12 years were recruited and followed up for 30 days. The coprimary outcomes were antibiotic prescribing at the index consultation (tested for superiority), as well as recovery time, additional testing, follow-up visits, and antibiotic prescribing after index consultation (all tested for non-inferiority with margins of 1 day, 3%, 4%, and 2%, respectively). Coprimary outcomes were analysed with logistic regression, accounting for practice clustering, study arm, and age in the intention-to-treat population, except recovery time, which was analysed with Cox regression adjusting for the same covariates. Safety was assessed in the intention-to-treat population. This trial is registered with ClinicalTrials.gov (NCT04470518) and is completed.

Findings

Of 171 eligible practices, we randomly allocated 82 to the intervention group and 89 to the usual care group. Between Feb 24, 2021, and Dec 29, 2023, 7049 participants were screened, of whom 6760 were deemed eligible. Five patients in each study arm were excluded, so we analysed data from 6750 participants (2988 in the intervention group and 3762 in the control group; 3447 [51%] boys, 3302 [49%] girls, one [<1%] did not specify). The intervention significantly reduced antibiotic prescribing at the index consultation (466 [16%] vs 817 [22%], adjusted odds ratio 0·72 [95% CI 0·55–0·94]; p=0·017). Recovery time (adjusted mean difference –0·1 day [95% CI –0·5 to 0·3]), additional testing (adjusted absolute risk reduction [aARR] 2·0% [–1·7 to 5·0]), follow-up visits (aARR 2·8% [–0·9 to 6·1]), and antibiotic prescribing after index consultation (aARR 2·4% [0·2 to 4·2]) were all non-inferior in the intervention group versus the control group. 90 (88%) of 102 adverse events were serious (30 [1%] in the intervention group and 60 [2%] in the control group); none were deemed related to the study procedures. No child died throughout the trial.

Interpretation

The clinical decision tool reduced antibiotic prescribing in children without causing harm. Our results support its broader dissemination and implementation to improve the management of acutely ill children in ambulatory care.

Funding

Belgian Health Care Knowledge Centre.