TOPLINE:

A clinical decision tool, comprising a validated decision tree, guided point-of-care testing of C-reactive protein, and safety-netting advice, led to reduced antibiotic prescriptions in children with acute illness compared with usual care. The tool maintained patient safety, with non-inferior outcomes found in terms of recovery time, additional testing, and follow-up visits.

METHODOLOGY:

• Researchers conducted a multicentre, cluster-randomised controlled trial in Belgium and involved 171 general practices to assess the impact of a clinical decision tool on antibiotic prescribing.
• Practices were randomly assigned in a 1:1 ratio to either the intervention (n = 82) or the usual care (n = 89) group; a total of 6750 children (aged 6 months to 12 years; 51% boys) with acute illness were included, with 2988 in the intervention group (median age, 3.7 years) and 3762 in the usual care group (median age, 3.9 years).
• The intervention group used a clinical decision tool with a decision tree and was trained to perform the point-of-care testing of C-reactive protein and provided with safety-netting advice. The usual care group managed the patients according to normal procedures.
• Physicians documented antibiotic considerations, performed point-of-care testing, and recorded readings for C-reactive protein and deviations from the tool's recommendations.
• Co-primary outcomes included antibiotic prescriptions at the index consultation, recovery time, additional testing, follow-up visits, and antibiotic prescriptions after the index consultation. Follow-up data, including adverse events and antibiotic prescriptions, were collected daily and recorded for up to 30 days post-consultation.

TAKEAWAY:

• The clinical decision tool led to reduced antibiotic prescriptions at the index consultation, with 22% of children in the usual care group vs 16% of those in the intervention group prescribed antibiotics (adjusted odds ratio, 0.72; 95% CI, 0.55-0.94).
• No significant differences in recovery time were found between the intervention and usual care groups, and additional testing, follow-up visits, and antibiotic prescribing after the index consultation were also non-inferior in the intervention group.
• The point-of-care testing of C-reactive protein was attempted in 25% of children in the intervention group: 87% of those with a "yes" response and 12% of those with a "no" response from the decision tree.
• The intervention group maintained safety, with adverse events and serious adverse events being less frequent (1%-2% in both the groups). Physicians reported high adherence to the clinical decision tool (for 95% of patients in the intervention group).

IN PRACTICE:

"This study highlights the effectiveness of a comprehensive approach to managing infection in children with acute illness in ambulatory care, combining a validated decision tree, guided POCT [point-of-care testing] of CRP [C-reactive protein], and safety-netting advice. By presenting a safe and evidence-based strategy for reducing unnecessary antibiotic prescribing, our findings contribute to the global fight against antimicrobial resistance," the authors wrote.

"Clinical guidelines could be adapted to incorporate the clinical decision tool and electronic health records could integrate it as part of a clinical-decision support system," they added.

SOURCE:

This study was led by Jan Yvan Verbakel, PhD, and Ruben Burvenich, PhD, Department of Public Health and Primary Care, KU Leuven, Leuven, Belgium. It was published online on September 25, 2025, in The Lancet.

LIMITATIONS:

Selection bias may have occurred due to the voluntary nature of participation. The absence of race and ethnicity data could have restricted the assessment of prescribing patterns across diverse groups. The 30-day follow-up period may not have captured long-term adverse effects or the sustainability of reduced prescribing rates.

DISCLOSURES:

This study was supported by the Belgian Health Care Knowledge Centre (KCE), as part of the KCE Trials programme. The authors declared having no relevant conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.