Oral antibiotic prescribing patterns and antibiotic-associated diarrhea risk: public health implications in North Macedonia (2011–2022)
Amoxicillin/clavulanate, cefixime, and clindamycin showed the highest combined indicators of antibiotic-associated diarrhea (AAD) risk, while Saccharomyces boulardii and Lactobacillus rhamnosus GG held the strongest external evidence for AAD prevention, with inconsistent local product availability.
| Population | General population of the Republic of North Macedonia with access to reimbursed oral antibiotics (ATC J01), 2011–2022 |
|---|---|
| Exposure | Pharmacoepidemiological analysis of oral antibiotic consumption + AAD risk estimation using pharmacovigilance indicators (EudraVigilance, FAERS, SmPC) + evidence synthesis on probiotics |
| Comparator | WHO AWaRe classification (Access/Watch/Reserve) as normative reference; probiotics assessed versus no probiotic (data from external systematic reviews) |
| Outcome | Oral antibiotic consumption pattern (DDD/1000 inhabitants/day); AAD risk per antibiotic (combined SmPC + pharmacovigilance indicator); S. boulardii efficacy for AAD prevention (evidence from external meta-analyses); L. rhamnosus GG efficacy for AAD prevention (evidence from external meta-analyses); Adherence to WHO AWaRe classification in prescribing profile; Market availability of evidence-supported probiotics for AAD; Pharmacovigilance signals (gastrointestinal ADR reports EudraVigilance/FAERS) |
Summary of findings
| Outcome | Effect | 95% CI | Certainty | Clinical relevance | Notes |
|---|---|---|---|---|---|
| Oral antibiotic consumption pattern (DDD/1000 inhabitants/day) | descriptive aggregate data; in the inferential statistic calculated | — | Low | — | 1 studies |
| AAD risk per antibiotic (combined SmPC + pharmacovigilance indicator) | normalized composite score; in the RR/OR/CI reported | — | Low | — | 1 studies |
| S. boulardii efficacy for AAD prevention (evidence from external meta-analyses) | RR ~0.47, 95% CI 0.35-0.63 (sourced from external meta-analyses cited by study) | — | High | — | |
| L. rhamnosus GG efficacy for AAD prevention (evidence from external meta-analyses) | RR ~0.49, 95% CI 0.29-0.83 (sourced from external meta-analyses cited by study) | — | High | — | |
| Adherence to WHO AWaRe classification in prescribing profile | descriptive; predominance of Watch-category agents identified; in the ratio with CI | — | Low | — | 1 studies |
| Market availability of evidence-supported probiotics for AAD | descriptive survey; considerable composition variability reported; in the quantitative CI | — | Low | — | 1 studies |
| Pharmacovigilance signals (gastrointestinal ADR reports EudraVigilance/FAERS) | normalized disproportionality indicators; in the RR/OR/CI calculated | — | Low | — | 1 studies |
Context
Irrational antibiotic use drives antimicrobial resistance and preventable adverse drug reactions, including AAD, which increases morbidity, costs, and risk of Clostridioides difficile infection. The WHO AWaRe classification guides rational prescribing but its application in middle-income countries remains limited. Mapping local consumption and AAD risk by class is a prerequisite for targeted pharmacotherapeutic and public health interventions.
What the study showed
Penicillins and cephalosporins dominated oral antibiotic consumption throughout the period. Amoxicillin/clavulanate, cefixime, and clindamycin scored highest on composite AAD risk indicators, combining high utilization frequency and strong pharmacovigilance signals. S. boulardii and L. rhamnosus GG were the only probiotics with consistent evidence from external meta-analyses; absolute effect sizes were not calculated by this study itself. Considerable variability was identified in declared composition and information quality of probiotic products available in local online pharmacies.
How it was done
Observational pharmacoepidemiological study without a prospective control group, integrating four data sources: (1) reimbursement data from the National Health Insurance Fund (ATC J01, 2011–2022); (2) gastrointestinal adverse drug reaction reports from EudraVigilance and FAERS; (3) synthesis of systematic reviews and meta-analyses on probiotics for AAD; (4) online pharmacy availability survey. Sample size is not reported as individual patients — data are population-level aggregates.
Effect magnitude
The study does not calculate its own RR, OR, SMD, or 95% CI for AAD; risk is estimated via SmPC frequency categories and normalized pharmacovigilance indicators, without original inferential statistics. Probiotic effect sizes are imported from external meta-analyses, not generated by this study.
Risk of bias
Ecological/pharmacoepidemiological observational design with aggregate data precludes causal inference and per-patient analysis (ecological fallacy). Pharmacovigilance data (EudraVigilance, FAERS) are subject to differential underreporting and notoriety bias, potentially overestimating risk for more widely prescribed antibiotics. No formal risk-of-bias tool (RoB 2, ROBINS-I) was applied to the probiotic evidence synthesis, and probiotic clinical effects were not re-analyzed with local data.
What this study does NOT prove
This study does not prove causality between prescribing patterns and AAD incidence, nor probiotic efficacy in the Macedonian population — probiotic effects are extrapolated from meta-analyses conducted in other populations. Findings are not directly generalizable to other countries without local consumption profiling.
In clinical practice
Clinicians in North Macedonia and similar settings should prioritize WHO AWaRe Access-category antibiotics and treat amoxicillin/clavulanate, cefixime, and clindamycin as high AAD-risk agents when making prescribing decisions. When probiotic co-prescription is indicated, S. boulardii CNCM I-745 and L. rhamnosus GG carry the most consolidated evidence; verifying actual product availability and identity in the local market is required before recommending.
Limitations
Ecological/pharmacoepidemiological observational design with aggregate data precludes causal inference and per-patient analysis (ecological fallacy). Pharmacovigilance data (EudraVigilance, FAERS) are subject to differential underreporting and notoriety bias, potentially overestimating risk for more widely prescribed antibiotics. No formal risk-of-bias tool (RoB 2, ROBINS-I) was applied to the probiotic evidence synthesis, and probiotic clinical effects were not re-analyzed with local data.
What is still missing
Prospective studies with individual patient data in the region are needed to quantify actual AAD incidence by antibiotic class and validate the clinical impact of probiotics in the local context.
Technical appendix
Version history
- 1.0 · 2026-09-14 — Auto-generated under Evidence Standard v1.0