Validation of an ICD-10-based algorithm for identifying inappropriate antibiotic prescribing for pediatric...

A diagnosis code-based algorithm classifies outpatient antibiotic prescribing as almost always (Tier 1), sometimes (Tier 2), and almost never (Tier 3) appropriate. This algorithm has been leveraged to quantify antibiotic appropriateness and direct stewardship initiatives for adults and children, primarily based on Tier 3 encounters receiving antibiotics, but has not been validated.

To assess the performance of the algorithm for assessing antibiotic appropriateness in pediatric outpatients.

Using a pediatric care network of primary care, urgent care, and emergency departments from January 1-December 31, 2024, encounters were classified into tiers using the ICD10 algorithm. Manual chart review was performed on a stratified random sample of 300 (100 per setting) Tier 3 encounters with antibiotics using a structured protocol derived from local and national guidelines to determine antibiotic appropriateness overall and by location. Survey weights were applied to obtain estimates for all Tier 3 encounters with antibiotic prescriptions. The positive predictive value (PPV) for Tier 3 in identifying inappropriate antibiotic prescribing was calculated.

Of 272,698 encounters, 115,508 (42%) had an antibiotic prescription. The algorithm classified 10,370 (9%) as Tier 3. In the stratified manual validation sample, 170/300 were classified as appropriate, corresponding to a survey-weighted estimate of 58% appropriate (95% CI, 52%-64%) and a PPV of 42% for Tier 3 as a marker of inappropriate antibiotic use.

The established algorithm did not reliably predict antibiotic appropriateness in this pediatric outpatient population, demonstrating a need for pediatric-focused algorithms for measuring outpatient antibiotic stewardship.