Benchmarking AWaRe: estimating optimal levels of AWaRe antibiotic use in 186 countries, territories and areas based on clinical infection and resistance burden
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (PREreview)
Abstract
Background
Ensuring appropriate access to essential antibiotics is a critical global public health goal. The UN General Assembly agreed that 70% of global antibiotic use should be WHO Access group. A standard method to estimate optimal antibiotic use based on burden of disease, resistance and local context is needed to inform national policies.
Methods
Using data from multiple global datasets, we clustered 186 countries, territories and areas (CTAs) into ‘peer groups’ based on sociodemographic factors, infection and resistance incidence using a latent class model. Within each cluster, benchmark countries with low antibiotic use and infection mortality were identified. We estimated optimal Total DID given infection burden, Reserve DID based on relevant resistance burdens, Watch DID based on clinical infections requiring Watch antibiotics from the WHO AWaRe Book and Access DID as the residual volume.
Findings
Globally 43.0 billion DDDs (95%CI:35.4billion–57.7billion) of antibiotics in 2019 were needed in 186 CTAs, of which 76% would optimally be Access (95%CI:70%-82%). CTAs in lower-income clusters required more Watch and Reserve antibiotics than higher income CTAs. Among CTAs with actual use data available, 72% (48/67) of CTAs used more Total and 99% (66/67) used more Watch antibiotics than estimated optimal levels.
Implications
We present the first estimates of optimal AWaRe antibiotic levels for 186 CTAs. After accounting for country-specific needs, the UNGA 70% Access target is globally appropriate. AWaRe benchmarking enables CTAs to estimate under and overuse of AWaRe antibiotics to inform national policies.
Funding
The ADILA Project funded by the Wellcome Trust [222051/Z/20/Z].
R esearch in context
Evidence before this study
Member states agreed at the 2024 United Nations General Assembly (UNGA) AMR meeting that the WHO AWaRe (Access, Watch, Reserve) system should underpin global antibiotic surveillance and that 70% of global use should be from the AWaRe Access group, while accounting for national contexts but did not agree any method for derived country-level targets. The WHO GLASS antimicrobial use (GLASS-AMU) surveillance system and other studies have reported observed antibiotic use. GLASS reports actual national medicine-level antibiotic use from 2015-2022 for 60 CTAs. The GRAM study (Browne, et al., 2021) modelled estimated total antibiotic use from 2000-2018 for 204 CTAs but were only able to estimate AWaRe antibiotic use for 76 CTAs where data from the commercial IQVIA MIDAS® database were available. Recent estimates from Klein et al. (2024) reported changes in antibiotic use from 2016 – 2023 for 67 CTAs in IQVIA MIDAS®.
Despite these studies quantifying estimates of current national medicine-level antibiotic use, there are very few estimates of what levels of antibiotic use are “optimal” both overall and by AWaRe group. We searched PubMed for studies from January 2015 through October 2025 on national-level antibiotic targets and AWaRe antibiotic policy using Boolean search terms. We identified few studies that estimated optimal levels of antibiotic use, but none for all AWaRe groups. As part of the Lancet AMR series, Mendelson et al. (2024) estimated expected total antibiotic use based on infection burden using the WHO AWaRe book guidance. This study developed a framework for estimating required Watch antibiotic use but did not provide comprehensive estimates for Access and Reserve use. Summan et al. (2025) estimated antibiotic needs for chronic obstructive pulmonary disease (COPD) and pneumonia in 20 CTAs using disease burden and bacterial aetiology. These estimates focused only on penicillins and cephalosporins and did not provide population based standardised measures to allow for cross-national comparison among CTAs. Mishra et al. (2025) estimated the gap in Reserve antibiotic treatment courses for use in carbapenem resistant Gram-negative infections in 8 countries using GRAM estimates and IQVIA sales data. While these studies have estimated expected antibiotic use for specific antibiotics or infections, there is no reported method to estimate optimal national levels of total and AWaRe antibiotic use.
Added value of this study
We developed a standard method for deriving optimal ranges of total and AWaRe antibiotic use in defined daily doses (DDD) per 1000 inhabitants per day (DID) accounting for national level infection and antibiotic resistance burden, population socio-demographics, national income, health system infrastructure and healthcare access using a benchmarking approach. Our study provides the first comprehensive estimates of optimal levels of total antibiotic use and disaggregated by AWaRe group in DID for a 2019 baseline for 186 CTAs and compares actual levels of AWaRe use to expected optimal use for 67 CTAs using the IQVIA MIDAS® dataset. We used publicly available data from the Global Burden of Disease (GBD), Global Research on Antimicrobial Resistance (GRAM) and the World Bank to provide an open-access, standardised AWaRe-clinical framework that could inform national and global evidence-based antibiotic policy development and implementation.
Implications of all the available evidence
Although some recent studies have attempted to derive estimates for optimal levels of antibiotic use, they have been focused on specific infections or antibiotics. The UNGA AMR commitments provide a clear direction for global target setting for antibiotic use. Our study provides the first method to estimate optimal antibiotic levels at a country level for both total and AWaRe, based on local infection burden and antibiotic resistance for 186 CTAs. Estimating optimal antibiotic levels with an agreed standard methodology across CTAs would allow for benchmarking and peer-comparison, assisting with target setting and shared learning across National Action Plans from different policy initiatives and outcomes.
Article activity feed
-
This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/19586511.
This study presents a timely and policy-relevant effort to estimate "optimal" antibiotic use across 186 countries using the AWaRe framework. Its central contribution—moving beyond descriptions of current antibiotic consumption toward defining need-based targets grounded in infection burden and resistance patterns—addresses an important gap in the literature and aligns well with global policy priorities such as the UNGA 70% Access target. At the same time, the concept of "optimal" use, while useful, relies heavily on modeling choices and assumptions that would benefit from clearer qualification throughout the manuscript.
The abstract succeeds in highlighting key findings, including the …
This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/19586511.
This study presents a timely and policy-relevant effort to estimate "optimal" antibiotic use across 186 countries using the AWaRe framework. Its central contribution—moving beyond descriptions of current antibiotic consumption toward defining need-based targets grounded in infection burden and resistance patterns—addresses an important gap in the literature and aligns well with global policy priorities such as the UNGA 70% Access target. At the same time, the concept of "optimal" use, while useful, relies heavily on modeling choices and assumptions that would benefit from clearer qualification throughout the manuscript.
The abstract succeeds in highlighting key findings, including the estimate of 43 billion DDDs and 76% Access use, and conveys the potential policy relevance of the work. However, it tends to oversimplify the underlying methodology, providing limited insight into how these estimates are derived or how sensitive they are to assumptions. This is particularly important given that one of the initial analytical approaches—the ecological regression model—did not yield meaningful results and was not used in the final analysis. Explicitly acknowledging this and clarifying the reliance on the benchmarking approach would improve transparency.
The study is methodologically innovative, especially in its use of a benchmarking framework combined with clustering of "peer" countries. This approach offers a creative way to account for heterogeneity across settings and provides a potentially useful tool for policy comparison and target setting. However, it also depends on strong assumptions about comparability between countries and on modeled infection and resistance data, particularly in low-income settings where empirical data are sparse. These constraints do not undermine the value of the approach but do suggest that results should be interpreted with appropriate caution.
The figures reflect a clear effort to communicate complex results in a policy-relevant way. For example, Figure 4 effectively highlights discrepancies between observed and estimated optimal use, particularly the overuse of Watch antibiotics. In contrast, other figures (e.g., Figures 1–3) are visually dense and sometimes difficult to interpret, and in some cases include analytical components that were not ultimately used. Simplifying these visuals and more clearly representing uncertainty would enhance their accessibility and impact, especially for non-specialist audiences.
The manuscript demonstrates technical rigor and comprehensive data integration, drawing on multiple large-scale datasets to produce global estimates. At the same time, the extensive use of technical nomenclature and abbreviations (e.g., DDD, DID, CTAs, ABU, ABR) can reduce readability and make it harder to follow the main arguments. Streamlining terminology and improving figure labeling would help make the findings more accessible without compromising rigor.
Overall, the study provides important and potentially impactful insights into global patterns of antibiotic use, including evidence of both overuse and underuse across settings. Its benchmarking framework is a valuable step toward more context-sensitive antibiotic policy. However, greater clarity around key assumptions, more explicit communication of limitations, and a more cautious interpretation of "optimal" targets would strengthen the credibility and usability of the findings.
Competing interests
The author declares that they have no competing interests.
Use of Artificial Intelligence (AI)
The author declares that they did not use generative AI to come up with new ideas for their review.
-
