Hostname: page-component-7b7567c8c-2rsb6 Total loading time: 0 Render date: 2026-11-12T09:37:51.835Z Has data issue: false hasContentIssue false
  • English
  • Français

Regional and seasonal variation in bacterial condition-related visit burden and antibiotic prescribing across Veterans Health Administration facilities

Published online by Cambridge University Press:  14 October 2026

Margaret CarrelOpen the ORCID record for Margaret Carrel[Opens in a new window],
Qianyi Shi,
Miah BoyleOpen the ORCID record for Miah Boyle[Opens in a new window] and
Michihiko Goto
Margaret Carrel*
Affiliation:
School of Earth, Environment, and Sustainability, The University of Iowa , Iowa City, IA, USA
Qianyi Shi
Affiliation:
Department of Internal Medicine, Carver College of Medicine, The University of Iowa, Iowa City, IA, USACenter for Access & Delivery Research and Evaluation (CADRE), Iowa City Veterans Affairs Health Care System, Iowa City, IA, USA
Miah Boyle
Affiliation:
School of Earth, Environment, and Sustainability, The University of Iowa , Iowa City, IA, USA
Michihiko Goto
Affiliation:
Department of Internal Medicine, Carver College of Medicine, The University of Iowa, Iowa City, IA, USACenter for Access & Delivery Research and Evaluation (CADRE), Iowa City Veterans Affairs Health Care System, Iowa City, IA, USA
*
Corresponding author: Margaret Carrel; Email: margaret-carrel@uiowa.edu

Abstract

Objective:

Estimate regional and seasonal variation in bacterial infection burden in Veterans Health Administration (VHA) facilities and its relationship to antibiotic prescribing .

Design:

Retrospective cross-sectional study.

Methods:

VHA records of ICD codes and prescription orders from October 2022 to September 2024 were aggregated to month and facility. The percent of patient visits for bacterial conditions (visit burden) as well as monthly total days supply and mean days supply of antibiotic prescriptions were calculated for each facility. Regional and seasonal variation in bacterial condition burdens in VHA facilities, and potential discordance between bacterial condition burden and antibiotic prescribing amounts were assessed.

Results:

Among the 135,694,718 care visits in 130 VHA facilities during the study period, 7,812,818 (5.8%) were for bacterial conditions. Bacterial condition visit burdens within VHA were highest in the US Northeast, with stable percentages of bacterial condition visits over time. The highest total antibiotic prescribing loads were observed in Southern facilities that did not have the highest bacterial condition percent. In models accounting for visit burden, region and season, total antibiotic days were significantly higher in non-Northeastern facilities, while mean prescribing days were not significantly associated with region but were significantly lower in seasons other than summer.

Conclusions:

While some VHA facilities in the US South have the highest ratios of bacterial conditions, regionally aggregated infection visit burdens are similar across the country. There is a disconnect in the relationship between bacterial condition visit burdens and total days of antibiotic prescribing in some Southern facilities, suggesting opportunities for stewardship efforts.

Information

Type
Original Article
Information
Antimicrobial Stewardship & Healthcare Epidemiology , Volume 6 , Issue 1 , 2026, e339
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2026. Published by Cambridge University Press on behalf of The Society for Healthcare Epidemiology of America

Introduction

In the US, antimicrobial-resistant infections continue to cause significant morbidity and mortality and influence clinician decision-making. Reference Sutton and Ashley1,Reference Hasegawa, Perencevich, Dukes and Goto2 Spatial and seasonal variation in rates of resistant infections suggests that behaviors and environments both within and outside of healthcare settings drive the likelihood of resistance in Staphylococcus aureus, Escherichia coli, and other commonly observed bacterial infections. Reference Carrel, Smith and Shi3–Reference Carrel, Shi, Hasegawa, Bricker, Boyle and Goto9 Inclusion of area-level factors and temporality has had limited predictive capacity for resistant S. aureus and E. coli within a nationwide cohort of Veterans Health Administration (VHA) patients, however, and coupled with individual patient-level data has had mixed influence on improvement in antibiogram predictive capabilities. Reference Tang, Shi, Hasegawa, Carrel, Oleson and Goto10,Reference Carrel, Shi and Clore11 Indeed, in many studies, the only significant predictor of an individual patient’s bacterial specimen exhibiting drug resistance is prior antibiotic exposure, at time scales varying from two weeks to ninety days. Reference Tang, Shi, Hasegawa, Carrel, Oleson and Goto10,Reference Kuster, Rudnick and Shigayeva12–Reference Goto, Bandyopadhyay and Shi14

Antimicrobial resistance patterns amongst S. aureus and E. coli specimens within the VHA have shown distinct spatial patterning, with higher and increasing rates of resistance to antibiotic classes such as tetracyclines, trimethoprim-sulfamethoxazole, fluoroquinolones, and extended-spectrum beta-lactamases in the VHA facilities and patient residential counties of the US South. Reference Smith, Carrel and Goto4,Reference Boyle, Shi, Hasegawa, Carrel, Oleson and Goto15 Reasons for this have been speculated to be related to warmer and wetter climate, higher rates of poverty and structural drivers of infection risk. Reference See, Wesson and Gualandi16–Reference Herold, Immergluck and Maranan21 However, as previously stated, when these spatial, behavioral, and environmental factors are explicitly included in models, only prior individual-level antibiotic exposure is consistently significantly predictive of resistance profiles. Reference Tang, Shi, Hasegawa, Carrel, Oleson and Goto10,Reference Goto, Bandyopadhyay and Shi14 Antibiotic prescribing in veterans has been shown to vary according to factors such as patient rurality (longer antibiotic courses and higher use of broad-spectrum agents), facility complexity (lower complexity exhibits higher prescribing rates), and regionally (more prescribing for outpatient respiratory infections in the US South and Central regions). Reference Evans, Fitzpatrick and Poggensee22–Reference Jones, Sauer and Jones24

In non-VHA studies, information on prescribing also indicates higher rates of antibiotic use or inappropriate prescribing in the Southern US. Reference King, Bartoces, Fleming-Dutra, Roberts and Hicks25–Reference Roberts, Hicks and Bartoces30 These studies, however, utilize data from convenience samples of privately insured individuals, data from pharmacies without linkage to patient conditions or focus only on specific health conditions. The VHA represents an opportunity for nationwide examination of antibiotic prescribing amounts directly linked to facility-level need for prescribing to understand whether regional or seasonal variation in prescribing or inappropriate prescribing is driven by underlying variation in bacterial outcomes in facilities, that is is prescribing higher in the South because of generally higher bacterial condition burden.

We sought to explore whether there was variation in antibiotic usage, of any class for any bacterial condition, amongst VHA facilities. In particular, we examined whether there was a disconnect between bacterial condition burdens (the percent of visits for a bacterial condition or for which an antibiotic was prescribed) and antibiotic prescribing loads across regions, and whether this disconnect varied seasonally as bacterial conditions have been shown to vary seasonally. In doing so, we assessed whether some regional patient populations seen at VHA facilities exhibit higher rates of bacterial conditions as an overall burden of healthcare outcomes, and whether some regions have higher levels of antibiotic prescribing than their bacterial condition burden would indicate is warranted.

Data and methods

The VHA Corporate Data Warehouse (CDW) provides patient-level information on healthcare visits as well as prescription orders. The CDW was queried for the time period of October 1, 2022 to September 30, 2024, for all facilities located in the conterminous US (48+DC). These dates avoid the most intense period of the COVID-19 pandemic and capture the most recent patient data available within the CDW. Data from six VHA facilities were excluded from the analysis due to limited availability.

At the facility level, a monthly count of all in-person and telehealth visits with providers for patients >18 years of age was generated, excluding allied health/paramedical visits (eg, social work, nurse only, laboratory visits). To estimate the number of patients a facility handles, individual patients were only counted once per month, to avoid over estimation of facility use by a patient being seen in multiple clinics in a short time. To understand facility visit burdens, a more expansive definition using all visits by patients regardless of repeated timing during a month, was generated. Thus, we can distinguish between facilities with similar patient numbers but different numbers of provider visits, or facilities that serve a small number of veterans in each month.

For each facility, a monthly count of the visits with bacterial condition diagnoses or an antibiotic prescription was calculated. Bacterial condition diagnoses were those that fell within the Tier 1 and Tier 2 categories developed by Fleming-Dutra et al. (2016) as almost always having an antibiotic treatment indicated or sometimes having an antibiotic treatment indicated, and were captured via ICD-10 codes. Reference Fleming-Dutra, Hersh and Shapiro31 We used a composite measure because either component alone is biased: diagnosis codes omit appropriate antibiotic use without an infection code (eg, inappropriate prescribing, procedural or postexposure prophylaxis), and prescriptions omit indicated conditions that were not treated. The composite therefore captures visits at which an antibiotic could plausibly have been prescribed, appropriately or not. Because each visit contributes once to the numerator regardless of course length, the exposure is not mechanically linked to the prescribing-day outcomes. We included all bacterial infectious causes rather than focusing specifically on condition types (e.g, acute respiratory tract infections) or bacteria types, (eg, Gram negative or Gram positive or other bacterial classes), instead capturing overall infectious condition burden requiring antibiotic treatment within VHA facilities.

The total bacterial condition patient and visit burden was calculated monthly by dividing the count of patients and visits associated with either a diagnosis or prescription indicating infection by the count of unique patients and total visits to facilities. The monthly percent of visits to facilities for bacterial conditions was charted to explore any variation by region or season. Region was assigned based on the four US Census regions. 32 Seasons were defined as: Winter (December, January, February), Spring (March, April, May), Summer (June, July, August), Fall (September, October, November).

Also at the facility level, monthly measures of antibiotic prescribing were calculated. All classes of systemic antimicrobials were included, as exposure to one class can sometimes confer resistance to another class and potentially drive area or facility level resistance. Reference Szybalski and Bryson33,Reference Sakenova, Cacace and Orakov34 For inpatients, antibiotics delivered as solution, additives, oral or injection were included, while for outpatients, antibiotics delivered as oral or injection were included. From each prescription, the number of prescribing days (days supply) was determined. For each facility and month, both the total number of prescribing days across all prescriptions as well as a mean days of prescribing was calculated. Reference Kim, Strominger, Zivin, Van and Maust35 Total prescribing days estimates overall durations of antibiotic treatments being prescribed, while mean days of prescribing across all prescriptions smoothes across potentially high variability in the total number and duration of prescriptions. We use these terms rather than days of therapy as both outpatient and inpatient treatments are being considered.

Facilities were stratified by the four US Census regions and monthly percent bacterial condition visits versus total and mean prescribing days in that month were calculated and charted to understand whether there exist regional mismatches, eg, whether some facilities have higher prescribing duration than their bacterial condition visit burden would suggest is necessary. The total prescribing days per 1,000 bacterial condition visits in regions and seasons was also calculated. Significance in variation between bacterial condition visit burden, total antibiotic days prescribed, total prescribing days per 1,000 bacterial condition visits and mean antibiotic days prescribed by region and season were assessed via Kruskal-Wallis tests. Smoothed loess lines stratified by region for data observed in each of the four seasons were also used to understand whether regional relationships between total and mean prescribing and bacterial condition visit percentages varied seasonally. Generalized estimating equation models accounting for correlation within facilities and across time further estimated the relationship between prescribing and bacterial condition visit percentageswhile simultaneously accounting for region and season. Models were clustered on facility with repeated measures defined as month/year and an autoregressive correlation structure with robust standard errors using the geepack package in R Studio (2025.09.2). Reference Højsgaard, Halekoh and Yan36

This cross-sectional study was approved by the institutional review board at the University of Iowa and the Research and Development Committee at the Iowa City Veterans Affairs Health Care System with a waiver of informed consent because the study was a retrospective analysis of health records with no direct contact with patients. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline.

Results

From October 1, 2022 to September 30, 2024, there were 61,141,210 unique patients seen for 135,694,718 care visits at 130 VHA facilities in the conterminous US. Among the 130 facilities, all provide outpatient care; ∼120 provide acute medical and surgical inpatient care, and ∼10 provide only mental health or nursing home care. Of these, 6,190,552 patients had 7,812,818 visits meeting the bacterial condition burden definition, defined as visits with either a Tier 1 or Tier 2 diagnosis code or an associated antibiotic prescription (10.1% of all patients, 5.8% of all visits). VHA facilities are not uniformly distributed across the US Census regions, the highest number are in the South (n = 49) and Midwest (n = 32), the Northeast region has n = 26 facilities and the West has n = 23 facilities. The burden of bacterial condition visits amongst VHA facilities varied significantly by region and season (Table 1). Bacterial condition visit percentages were highest in facilities in the Northeast and in the winter. This pattern varies by facility, however, with some facilities in the South exhibiting the highest bacterial condition percents among total visits, though fit lines indicate relative stability in rates over time (Figure 1).

Figure 1.

The percentage of VHA visits for a tier 1 or tier 2 condition or with an antibiotic prescription by month and region, October 2022–September 2024. Winter months are indicated by solid lines, summer months by dashed lines.

A line graph showing bacterial condition percent by month and region from October 2022 to September 2024.

Table 1.

Summary statistics by region, season, and region: season for total patients, visits, bacterial conditions patients and visits, total prescribing days, total prescribing days per 1,000 bacterial condition visits, and mean prescribing days

A table summarizing statistics by region, season, and region-season for patients, visits, bacterial conditions, and antibiotic prescribing.
Table 1. Long description

Significance: *** P < .001 ** P < .01 * P < .05, Kruskal-Wallis rank sum test.

Total antibiotic prescribing days and mean prescribing days significantly vary by region, and mean prescribing days significantly vary by season (Table 1). Average total prescribing days are highest in the South, lowest in the Northeast, while the average of mean prescribing days is highest in the Northeast. Seasonally, the highest average monthly mean prescribing days is in the summer.

When standardized by bacterial condition visits, total antibiotic days per 1,000 visits are highest in the Midwest and then the South, and in the Midwest across all seasons (Table 1). However, when examining facility level variation by region, there are other differences in the total days of antibiotics prescribed across bacterial condition visit percentages (Figure 2A). Facilities in the US South and Midwest have higher total days at bacterial condition percentages of 4%–8% than Western and Northeastern facilities with similar percentages of patient visits for bacterial conditions. When the mean prescribing days of antibiotics rather than total days are considered, regional differences attenuate (Figure 2B). Western and Midwest facilities comprise the majority of the upper section of the point cloud at the lower ranges of bacterial condition visit percentages, while Southern facilities with the highest burdens of patient visits for bacterial conditions have comparable mean prescribing days to other regions.

Figure 2.

Monthly bacterial condition percent in VHA facilities versus total prescribing days (A) and mean prescribing days (B), stratified by region.

Scatter plots depict antibiotic days versus bacterial condition percent by region.

No regions show expected globally linear positive relationships between total days of prescribing and bacterial condition visit percentages when stratified by season (Figure 3). Instead, total days of prescribing peaks in facilities in all regions with ranges of bacterial condition visit percentages that are not at upper or lower extremes, and is highest in all seasons in facilities in the South. In contrast, when mean days of prescribing is considered by season and region (Figure 4), facilities in the Midwest and West with the lowest bacterial condition percent of visits have the highest prescribing durations.

Figure 3.

Monthly bacterial condition percent in VHA facilities versus total prescribing days with smoothed fit lines stratified by region for spring (A), summer (B), fall (C), and winter (D).

Four scatter plots showing total antibiotic days versus bacterial condition percent by region and season.

Figure 4.

Monthly bacterial condition percent in VHA facilities versus mean prescribing days with smoothed fit lines for stratified by region for spring (A), summer (B), fall (C), and winter (D).

Four scatter plots showing the relationship between bacterial condition percent and mean antibiotic days across different regions and seasons.
Figure 4. Long description

Panel A: Spring. A scatter plot showing the relationship between bacterial condition percent (x-axis) and mean antibiotic days (y-axis) for the spring season. Data points are color-coded by region: Midwest (blue), Northeast (purple), South (orange), and West (green). Smoothed fit lines are included for each region. Panel B: Summer. A scatter plot showing the relationship between bacterial condition percent (x-axis) and mean antibiotic days (y-axis) for the summer season. Data points are color-coded by region: Midwest (blue), Northeast (purple), South (orange), and West (green). Smoothed fit lines are included for each region. Panel C: Fall. A scatter plot showing the relationship between bacterial condition percent (x-axis) and mean antibiotic days (y-axis) for the fall season. Data points are color-coded by region: Midwest (blue), Northeast (purple), South (orange), and West (green). Smoothed fit lines are included for each region. Panel D: Winter. A scatter plot showing the relationship between bacterial condition percent (x-axis) and mean antibiotic days (y-axis) for the winter season. Data points are color-coded by region: Midwest (blue), Northeast (purple), South (orange), and West (green). Smoothed fit lines are included for each region.

Models indicate that monthly bacterial condition visits percentages are positively and significantly associated with total antibiotic prescribing days but negatively and significantly associated with mean prescribing days (Table 2). Compared with facilities in the Northeast, facilities in other US regions are significantly associated with higher total antibiotic prescribing days but not with any significant differences in mean prescribing days. Season was significantly associated with total and mean prescribing days; Fall and Winter were negatively associated with total prescribing days compared to Summer while Spring was positively associated with total prescribing days, and all seasons were negatively associated with mean prescribing days compared to Summer.

Table 2.

GEE model results for total and mean prescribing days in facilities by month. Clustering within facilities based on facility ID and time waves by month/year was accounted for, an autoregressive correlation structure was assumed

A table comparing total and mean antibiotic prescribing days across regions and seasons.
Table 2. Long description

A table with 11 rows and 6 columns comparing total and mean antibiotic prescribing days across different regions and seasons. The columns are labeled ‘Monthly Visit Bacterial Condition Percent’, ‘Region’, ‘Season’, ‘Total Rx days Estimate’, ‘Total Rx days Std. Error’, ‘Mean Rx days Estimate’, and ‘Mean Rx days Std. Error’. The table includes data for regions Northeast, Midwest, South, and West, and seasons Summer, Fall, Winter, and Spring. Notable values include a significant positive association of monthly visit bacterial condition percent with total Rx days and a significant negative association with mean Rx days. Compared to the Northeast, the Midwest, South, and West show higher total Rx days. Seasons Fall and Winter show negative associations with total Rx days compared to Summer, while Spring shows a positive association. All seasons show negative associations with mean Rx days compared to Summer.

Significance: *** P < .001 ** P < .01 * P < .05.

Discussion

In the VHA, patient visits for bacterial conditions or resulting in an antibiotic prescription represented 5.8% of all 135 million patient visits from October 2022 through September 2024, and were observed for 10.1% of all 61 million unique patients. While facilities in the US South saw far more patients in far more visits than those in other regions, percentages of patients and visits for bacterial conditions or with antibiotic prescriptions were highest regionally in the Northeast. Variation within regions exists, however, with highest monthly bacterial condition visit percentages consistently observed in individual facilities in the south.

Prior research has indicated that the US South has higher rates of antimicrobial-resistant infections in VHA patients and facilities, potentially due to climate or social variables. Reference Carrel, Smith and Shi3,Reference Livorsi, Goto and Carrel5,Reference Carrel, Shi, Hasegawa, Bricker, Boyle and Goto9 Minor, though significant, seasonal variation in bacterial condition visit burdens are observed across VHA facilities, but the inclusion of all species of bacteria likely diminishes the appearance of seasonal effects typically observed in classes such as Gram-negatives or Staphylococci. Reference Leekha, Diekema and Perencevich6–Reference Perencevich, McGregor and Shardell8

Higher antibiotic use in the South has been well documented in outpatient claims and pharmacy data, and our findings are consistent with that literature. What prior studies have generally lacked, however, is a facility-level denominator of clinical need against which prescribing can be compared. By linking prescribing to the burden of conditions for which antibiotics are indicated, within a single integrated system with uniform formulary, coding, and record-keeping practices, this study shows that regional differences in total prescribing persist after accounting for the frequency of antibiotic-indicated visits, and that they are attributable to volume rather than to longer courses per prescription. This suggests regional variation is not explained by payer mix, coding practice, or differences in underlying condition burden alone.

While intraregional and seasonal variation in bacterial condition burden was anticipated, the high variability in total prescribing and the non-linearity of the observed relationship with infection burdens were not. Facilities in the South had far higher levels of total prescribing days than other regions, followed by the Midwest which had the highest total prescribing days when standardized by bacterial condition visits, and the highest levels of total prescribing were not observed at the highest levels of bacterial infection burden. This disconnect between patient need/provider perception of need and prescribing was not observed, however, when the outcome under consideration was mean prescribing days; in this case a negative relationship between monthly bacterial infection burden and prescribing duration was observed.

This difference, coupled with a peak in total prescribing at the middle of observed levels of bacterial condition burden, offers opportunities for VHA antimicrobial stewardship programs to focus their attention on facilities in the South and Midwest that perhaps have lower frequency of patients requiring antibiotics but prescribing practices that result in longer durations of antibiotic regimens. Alternatively, these facilities could serve different types of patients, such as those from rural locations, who are prescribed longer treatment courses in advance to prevent burdensome returns to the treatment facility. Reference Appaneal, Caffrey, Lopes, Dosa and LaPlante23

Compared to the summer months, total and mean prescribing days were lower in other seasons, except for the spring when total prescribing days is modeled. Longer durations could reflect clinician habit, a condition mix weighted toward infections requiring prolonged therapy (eg, skin and soft tissue, bone and joint), or patient factors such as rurality and comorbidity, none of which would increase the number of antibiotic-indicated visits. Total visit volume also varied modestly by season, with the lowest volume in winter (approximately 8% below spring), reflecting reduced routine and elective visits around the holiday period. Because bacterial condition burden is expressed as a share of all visits, part of the modest winter peak in visit percentage may reflect a smaller denominator of non-urgent visits rather than a true increase in infection-related care.

Limitations of the study include those related to patient composition and data management. Our bacterial condition burden measure was a composite union of visits with Tiers 1 and 2 diagnosis codes or associated antibiotic prescriptions. Because antibiotic prescribing was one component of this numerator, the measure is not fully independent of prescribing behavior; future work should decompose this composite into diagnosis-only, diagnosis-plus-prescription, and prescription-only components. We chose to include counts of patient visits resulting in an antibiotic prescription without an associated ICD code in Tiers 1 or 2 to capture more fully the prescribing behaviors within VHA, which could include either prescriptions for codes such as pneumonia which may be viral or bacterial in nature or for antibiotic prescribing done prophylactically for events such as tick bites.

While the VHA is the single nationwide provider of healthcare, enabling comparison across space and time in outcomes, the demographics of Veterans do not mirror those of the US more generally as Veterans tend to be older in age, predominantly male, and White. Due to data completeness, six facilities in Oregon, Illinois Washington and Ohio were excluded from the analysis; their spread across multiple Census regions likely limits the impact this has on the overall conclusions, however. Finally, decisions such as how to assign prescription refills, what types of patient visits were included in the total patient visit count (eg, not social work or lab only), calculation of bacterial condition visits from a list of candidate health record codes and seasonal delineations could impact findings. Reference Fleming-Dutra, Hersh and Shapiro31,Reference Degnan, Cluzet and David37

Higher likelihood of drug-resistant infections is linked to prior antibiotic exposure, and higher and increasing rates of drug resistance in the VHA have previously been observed in the US South. Though the US South has the largest number of VHA facilities and serves the largest number of patients, the share of their visits for bacterial conditions or resulting in an antibiotic prescription are similar to other regions. The total days of antibiotics some Southern patients are being prescribed, however, is not consistent with the burden of bacterial infections in their provider’s facilities, though average days of prescribing are. Outlier prescribing behavior, then, of longer duration than average, perhaps driven by clinician behavior, types of infections (eg, joint or valve), or patient characteristics (eg, rural residence), may be partially responsible for the higher rates of resistant infections observed regionally in facilities, and maybe be why area level investigations of social or environmental drivers of resistance are not detecting significant associations. This suggests that stewardship efforts should perhaps be directed towards shortening antibiotic courses in facilities that may not appear to be top candidates based upon levels of bacterial conditions that are not at the highest end of the observed range but for whom the mixture of infection types, provider behaviors or patient characteristics, such as rural patients, results in longer days supply of antibiotics.

Data availability statement

Data is not publicly available, and may only be accessed via permission from the Veterans Health Administration.

Acknowledgments

None.

Financial support

M.C. and Q.S. received support from the Centers for Disease Control & Prevention Epicenter program award CK000613 (total award amount $500,000). M.G. received support from the Agency for Healthcare Research and Quality under award number K08HS027472 (total award amount $741,346). M.B. reports no relevant funding sources. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States Government.

Competing interests

All authors report no conflicts of interest relevant to this article.

References

Sutton, KF, Ashley, LW. Antimicrobial resistance in the United States: origins and future directions. Epidemiol Infect 2024;152:e33.10.1017/S0950268824000244CrossRefGoogle ScholarPubMed
Hasegawa, S, Perencevich, EN, Dukes, K, Goto, M. 964. Physicians acceptable treatment failure rates and interpretation of antibiogram for Gram-negative infections: A pilot survey study of infectious diseases specialists. US: Oxford University Press; 2022. ofac492. 807.Google Scholar
Carrel, M, Smith, M, Shi, Q, et al. Antimicrobial resistance patterns of outpatient Staphylococcus aureus isolates. JAMA Netw Open 2024;7:e2417199–e2417199. doi:10.1001/jamanetworkopen.2024.17199.CrossRefGoogle ScholarPubMed
Smith, MW, Carrel, M, Goto, M. County-level spatiotemporal distribution of fluoroquinolone-resistant Enterobacteriaceae in outpatient settings of the Veterans’ Health Administration, 2000–2017. Infect Control Hosp Epidemiol 2023;44:1497–1499 .10.1017/ice.2022.291CrossRefGoogle ScholarPubMed
Livorsi, DJ, Goto, M, Carrel, M, et al. Regional variations in fluoroquinolone non-susceptibility among Escherichia coli bloodstream infections within the Veterans Healthcare Administration. Antimicrob Resist Infect Control 2016;5:38.10.1186/s13756-016-0135-2CrossRefGoogle ScholarPubMed
Leekha, S, Diekema, D, Perencevich, E. Seasonality of staphylococcal infections. Clin Microbiol Infect 2012;18:927–933.10.1111/j.1469-0691.2012.03955.xCrossRefGoogle ScholarPubMed
Eber, MR, Shardell, M, Schweizer, ML, Laxminarayan, R, Perencevich, EN. Seasonal and temperature-associated increases in gram-negative bacterial bloodstream infections among hospitalized patients. PLoS One 2011;6:e25298.10.1371/journal.pone.0025298CrossRefGoogle ScholarPubMed
Perencevich, EN, McGregor, JC, Shardell, M, et al. Summer peaks in the incidences of gram-negative bacterial infection among hospitalized patients. Infect Control Hosp Epidemiol 2008;29:1124–1131.10.1086/592698CrossRefGoogle ScholarPubMed
Carrel, M, Shi, Q, Hasegawa, S, Bricker, C, Boyle, M, Goto, M. Variation in facility-level likelihood of drug-resistant Staphylococcus aureus in outpatients remains after patient-level risk adjustment. Antimicrob Steward Healthc Epidemiol 2025;5:e205. doi:10.1017/ash.2025.10127.CrossRefGoogle ScholarPubMed
Tang, Z, Shi, Q, Hasegawa, S, Carrel, M, Oleson, J, Goto, M. Spatial pattern and risk factors of resistance to important antibiotics among E. coli from veterans in seven U.S. Midwest states. Antimicrob Steward Healthc Epidemiol 2026;6:e35. doi:10.1017/ash.2025.10292.CrossRefGoogle Scholar
Carrel, M, Shi, Q, Clore, GS, et al. Assessing the potential for improved predictive capacity of antimicrobial resistance in outpatient Staphylococcus aureus isolates using seasonal and spatial antibiograms. Antimicrob Resist Infect Control 2024;13:1–6.10.1186/s13756-024-01388-3CrossRefGoogle ScholarPubMed
Kuster, SP, Rudnick, W, Shigayeva, A, et al. Previous antibiotic exposure and antimicrobial resistance in invasive pneumococcal disease: results from prospective surveillance. Clin Infect Dis 2014;59:944–952.10.1093/cid/ciu497CrossRefGoogle ScholarPubMed
Opatowski, M, Brun-Buisson, C, Touat, M, et al. Antibiotic prescriptions and risk factors for antimicrobial resistance in patients hospitalized with urinary tract infection: a matched case-control study using the French health insurance database (SNDS). BMC Infect Dis 2021;21:571.10.1186/s12879-021-06287-1CrossRefGoogle ScholarPubMed
Goto, M, Bandyopadhyay, A, Shi, Q, et al. Personalized antibiogram: a novel multitask machine learning framework for simultaneous prediction of antimicrobial resistance profile with enhanced detection of carbapenem resistance in enterobacteriaceae. Clin Infect Dis 2026;83:e1–e9. doi: 10.1093/cid/ciag027.CrossRefGoogle ScholarPubMed
Boyle, M, Shi, Q, Hasegawa, S, Carrel, M, Oleson, J, Goto, M. Spatiotemporal trends in tetracycline- and trimethoprim-sulfamethoxazole-resistant S. aureus among veteran outpatients in the eastern United States. Epidemiol Infect 2026;154:e31. doi: 10.1017/S0950268826101216.CrossRefGoogle ScholarPubMed
See, I, Wesson, P, Gualandi, N, et al. Socioeconomic factors explain racial disparities in invasive community-associated methicillin-resistant Staphylococcus aureus disease rates. Clin Infect Dis 2017;64:597–604.10.1093/cid/ciw808CrossRefGoogle ScholarPubMed
Immergluck, LC, Leong, T, Malhotra, K, et al. Geographic surveillance of community associated MRSA infections in children using electronic health record data. BMC Infect Dis 2019;19:1–12.Google ScholarPubMed
Anthony, CA, Peterson, RA, Polgreen, LA, Sewell, DK, Polgreen, PM. The seasonal variability in surgical site infections and the association with warmer weather: a population-based investigation. Infect Control Hosp Epidemiol 2017;38:809–816.10.1017/ice.2017.84CrossRefGoogle Scholar
Portela, GT, Leong, T, Webster, A, et al. Risk factors for non-invasive (skin and soft tissue) and invasive Staphylococcus aureus infections among children and adults living in southeastern USA: a retrospective cohort study. BMJ Open 2022;12:e059327.10.1136/bmjopen-2021-059327CrossRefGoogle Scholar
Ali, F, Immergluck, LC, Leong, T, Waller, L, Malhotra, K, Jerris, RC, Edelson, M, Rust, GS. Aspatial analysis of health disparities associated with antibiotic resistant infections in children living in Atlanta (2002–2010). eGEMs 2019;7:50. doi: 10.5334/egems.308 CrossRefGoogle ScholarPubMed
Herold, BC, Immergluck, LC, Maranan, MC, et al. Community-acquired methicillin-resistant Staphylococcus aureus in children with no identified predisposing risk. JAMA 1998;279:593–598.10.1001/jama.279.8.593CrossRefGoogle ScholarPubMed
Evans, CT, Fitzpatrick, MA, Poggensee, L, et al. Outpatient prescribing of antibiotics and opioids by Veterans Health Administration providers, 2015–2017. Am J Prev Med 2021;61:e235–e244.10.1016/j.amepre.2021.05.009CrossRefGoogle ScholarPubMed
Appaneal, HJ, Caffrey, AR, Lopes, V, Dosa, D, LaPlante, KL. Antibiotic prescribing in outpatient settings: rural patients are more likely to receive fluoroquinolones and longer antibiotic courses. Antibiotics 2023;12:224.10.3390/antibiotics12020224CrossRefGoogle ScholarPubMed
Jones, BE, Sauer, B, Jones, MM, et al. Variation in outpatient antibiotic prescribing for acute respiratory infections in the veteran population: a cross-sectional study. Ann Intern Med 2015;163:73–80.10.7326/M14-1933CrossRefGoogle ScholarPubMed
King, LM, Bartoces, M, Fleming-Dutra, KE, Roberts, RM, Hicks, LA. Changes in US outpatient antibiotic prescriptions from 2011–2016. Clin Infect Dis 2020;70:370–377.10.1093/cid/ciz225CrossRefGoogle ScholarPubMed
King, LM, Tsay, SV, Hicks, LA, Bizune, D, Hersh, AL, Fleming-Dutra, K. Changes in outpatient antibiotic prescribing for acute respiratory illnesses, 2011 to 2018. Antimicrob Steward Healthc Epidemiol 2021;1:e66.10.1017/ash.2021.230CrossRefGoogle ScholarPubMed
Hersh, AL, Shapiro, DJ, Pavia, AT, Fleming-Dutra, KE, Hicks, LA. Geographic variability in diagnosis and antibiotic prescribing for acute respiratory tract infections. Infect Dis Ther 2018;7:171–174.10.1007/s40121-017-0181-yCrossRefGoogle ScholarPubMed
Hicks, LA, Bartoces, MG, Roberts, RM, et al. US outpatient antibiotic prescribing variation according to geography, patient population, and provider specialty in 2011. Clin Infect Dis 2015;60:1308–1316. doi:10.1093/cid/civ076.CrossRefGoogle ScholarPubMed
Hicks, LA, Taylor, TH Jr., Hunkler, RJ. U.S. outpatient antibiotic prescribing, 2010. N Engl J Med 2013;368:1461–1462. doi: 10.1056/NEJMc1212055.CrossRefGoogle ScholarPubMed
Roberts, RM, Hicks, LA, Bartoces, M. Variation in US outpatient antibiotic prescribing quality measures according to health plan and geography, 2008–2012. Am J Manag Care 2016;22:519.Google Scholar
Fleming-Dutra, KE, Hersh, AL, Shapiro, DJ, et al. Prevalence of inappropriate antibiotic prescriptions among US ambulatory care visits, 2010-2011. JAMA 2016;315:1864–1873.10.1001/jama.2016.4151CrossRefGoogle ScholarPubMed
Bureau USC. Geographic terms and concepts - census divisions and census regions. 2010.Google Scholar
Szybalski, W, Bryson, V. Genetic studies on microbial cross resistance to toxic agents I. Cross resistance of Escherichia coli to fifteen antibiotics. J Bacteriol 1952;64:489–499.10.1128/jb.64.4.489-499.1952CrossRefGoogle ScholarPubMed
Sakenova, N, Cacace, E, Orakov, A, et al. Systematic mapping of antibiotic cross-resistance and collateral sensitivity with chemical genetics. Nat Microbiol 2025;10:202–216.10.1038/s41564-024-01857-wCrossRefGoogle ScholarPubMed
Kim, HM, Strominger, J, Zivin, K, Van, T, Maust, DT. Relationship between facility number of clinicians and prescribing intensity of psychotropic medications, opioids, and antibiotics in ambulatory practice. BMC Health Serv Res 2024; 24:217. doi:10.1186/s12913-024-10613-z.CrossRefGoogle ScholarPubMed
Højsgaard, S, Halekoh, U, Yan, J. The R package geepack for generalized estimating equations. J Stat Softw 2006;15:1–11.Google Scholar
Degnan, KO, Cluzet, V, David, MZ, et al. Development and validation of antibiotic stewardship metrics for outpatient respiratory tract diagnoses and association of provider characteristics with inappropriate prescribing. Infect Control Hosp Epidemiol 2022;43:56–63.10.1017/ice.2021.44CrossRefGoogle Scholar
Figure 0

Figure 1.The percentage of VHA visits for a tier 1 or tier 2 condition or with an antibiotic prescription by month and region, October 2022–September 2024. Winter months are indicated by solid lines, summer months by dashed lines.

Figure 1

Table 1.Summary statistics by region, season, and region: season for total patients, visits, bacterial conditions patients and visits, total prescribing days, total prescribing days per 1,000 bacterial condition visits, and mean prescribing daysTable 1 long description.

Figure 2

Figure 2.Monthly bacterial condition percent in VHA facilities versus total prescribing days (A) and mean prescribing days (B), stratified by region.

Figure 3

Figure 3.Monthly bacterial condition percent in VHA facilities versus total prescribing days with smoothed fit lines stratified by region for spring (A), summer (B), fall (C), and winter (D).

Figure 4

Figure 4.Figure 4 long description.Monthly bacterial condition percent in VHA facilities versus mean prescribing days with smoothed fit lines for stratified by region for spring (A), summer (B), fall (C), and winter (D).

Figure 5

Table 2.GEE model results for total and mean prescribing days in facilities by month. Clustering within facilities based on facility ID and time waves by month/year was accounted for, an autoregressive correlation structure was assumedTable 2 long description.