About

Joseph Lewnard is an Associate Professor at the School of Public Health within the Division of Epidemiology. His research interests include infectious diseases, antimicrobial resistance, public health surveillance, mathematical modeling, and Bayesian inference. He studies the transmission dynamics of infectious disease agents and evaluates the effectiveness of interventions such as vaccination. His work integrates biostatistics, clinical data analysis, ecological modeling, evolutionary biology and phylogenetics, as well as network and systems biology. Lewnard's research aims to inform public health strategies and improve understanding of infectious disease spread and control.

Research topics

• Medicine
• Environmental health
• Internal medicine
• Demography
• Sociology
• Computer Science
• Business
• Geography
• Virology
• Demographic economics
• Gerontology
• Emergency medicine
• Economics
• Pediatrics
• Intensive care medicine
• Immunology
• Telecommunications
• Pathology

Selected publications

• Frequent seasonal reassortment between high and low path viruses drives the diversification of influenza A/H5N1

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-04-18

  articleOpen access

  Abstract Since 2021, highly pathogenic (HPAI) H5N1 viruses have spread across the Americas, diversifying via reassortment into new genotypes that have spilled into humans and livestock, raising fears of a new influenza pandemic. Pandemic lineages are typically associated with reassortment, but we currently have limited understanding of where and when reassortment is expected to occur, which limits our ability to assess pandemic risks. Using a dataset of 9,052 full-genome sequences, we show that reassortment and novel genotype formation are associated with seasonal variation in low pathogenicity avian influenza (LPAI) cases and with the spatial and host distributions of viral transmission. We pinpoint ducks, geese, and the Central flyway as frequent sources of new genotypes, and show that reassortment rates vary seasonally, driven by mixing between high- and low-pathogenicity viruses. Cattle spillover genotypes (B3.13 and D1.1) evolved during periods of high reassortment, implicating reassortment as a common occurrence in lineages evolving during particular time periods. Together, these findings reframe reassortment as a predictable ecological process, with direct implications for how surveillance and pandemic risk assessment should be designed.

  PublisherDOI

• Bacterial sexually transmitted infections and related antibiotic use among individuals eligible for doxycycline post-exposure prophylaxis in the United States

  Nature Communications · 2025-10-16 · 5 citations

  articleOpen accessCorresponding

  While doxycycline postexposure prophylaxis (doxyPEP) can prevent bacterial sexually transmitted infections (STIs), concern surrounds the volume of antibiotic use needed to realize this benefit. We estimated incidence rates of gonorrhea, chlamydia, and syphilis diagnoses and related antibiotic prescribing among US males and transgender individuals using Merative MarketScan® Research Databases during 2017-2019. Follow-up encompassed 38,543 person-years among recipients of HIV pre-exposure prophylaxis (PrEP), 29,228 person-years among people living with HIV (PLWH), and 19,918 person-years among people with prior-year STI diagnoses. Incidence rates of STI diagnoses among PLWH and PrEP recipients with ≥1 prior-year STI diagnosis totaled 33.3-35.5 per 100 person-years. Direct effects of doxyPEP could prevent 7.4-9.6 gonorrhea diagnoses, 7.3-8.1 chlamydia diagnoses, and 3.1-5.9 syphilis diagnoses per 100 person-years of use. However, expected increases in tetracycline consumption resulting from doxyPEP implementation totaled 271.9-312.9 additional 7-day doxycycline treatment courses per 100 person-years of use. These increases corresponded to 37.0-38.7, 36.5-37.0, and 46.1-100.2 additional 7-day doxycycline treatment courses for each prevented gonorrhea, chlamydia, and syphilis diagnosis, respectively. Increases in doxycycline use exceeded anticipated reductions in STI-related prescribing of cephalosporins, macrolides, and penicillins by 16-69-fold margins. Anticipated changes in antibiotic use as well as STI incidence should inform priority-setting for doxyPEP.

  PublisherOA PDFDOI

• Pediatric antibiotic use associated with respiratory syncytial virus and influenza in the United States, 2008-2018

  medRxiv · 2025-03-26

  preprintOpen accessSenior author

  Background: Understanding of the contributions of respiratory syncytial virus (RSV) and influenza infections to pediatric antibiotic use is limited. We aimed to estimate the proportions and incidence of outpatient antibiotic prescriptions associated with RSV and influenza infections in a sample of commercially-insured US children. Methods: DataMart from 2008-2018. We used negative binomial time-series models regressing weekly antibiotic prescriptions against RSV and influenza circulation measures to estimate counterfactual rates of antibiotic prescriptions in the presence and absence of RSV and influenza circulation overall, by age group, census division, and antibiotic class. We considered both syndromic (medical claims) and laboratory (National Respiratory and Enteric Virus Surveillance System) RSV and influenza measures and controlled for age, division, 13-valent pneumococcal conjugate vaccine introduction, and seasonal and secular trends. Results: An estimated 6.3% (95% confidence interval 5.2-7.3%) and 3.4% (3.1-3.8%) of antibiotic prescriptions were associated with RSV and influenza, respectively. These estimates translate to 72.6 (59.7-85.9) RSV-associated and 40.0 (35.1-45.1) influenza-associated antibiotic prescriptions per 1000 children annually. RSV-associated antibiotic prescription incidence was highest among children aged ≤5 years while influenza-associated antibiotic prescriptions were highest among children >5 years. Macrolides were the antibiotic class for which RSV and influenza accounted for the greatest share of prescribing. Conclusions: RSV and influenza account for meaningful proportions of pediatric antibiotic prescriptions. Measures to prevent RSV and influenza infections in children, including immunization, may reduce antibiotic use and aid in mitigating antibiotic resistance.

  PublisherOA PDFDOI

• Acute respiratory infections due to antibiotic-nonsusceptible <i>Streptococcus pneumoniae</i> in United States adults

  medRxiv · 2025-04-07

  preprintOpen accessSenior author

  Background: (pneumococcus) in US adults (≥18 years). Methods: We estimated antibiotic-nonsusceptible pneumococcal sinusitis and non-bacteremic pneumonia incidence as products of non-bacteremic pneumococcal pneumonia and sinusitis incidence rates, serotype distribution, and serotype-specific antimicrobial nonsusceptibility prevalences by antibiotic class and guideline-recommended agents from 2016-2019. We derived pneumonia and sinusitis incidence rates from national healthcare utilization surveys and administrative datasets; pneumococcal-attributable attributable percents and serotype distributions from published data; and serotype-specific nonsusceptibility estimates from Active Bacterial Core surveillance data. We evaluated nonsusceptibility for all serotypes and those targeted by 15-, 20- and 21-valent pneumococcal conjugate vaccines (PCV15/20/21). Results: An estimated 16.4% (95% confidence interval 12.8-21.4%) of non-bacteremic pneumococcal pneumonia and 19.0% (14.8-24.9%) of sinusitis cases were nonsusceptible to ≥3 antibiotic classes, translating to 243,521 (179,673-333,675) and 1,844,726 (1,070,763-2,904,089) outpatient visits for pneumonia and sinusitis, respectively, and 10,155 (7,542-13,803) pneumonia hospitalizations annually. An estimated 31.2% (26.6-36.3%) of non-bacteremic pneumococcal pneumonia and 10.5% (9.4-12.0%) of pneumococcal sinusitis cases were nonsusceptible to ≥1 outpatient first-line antibiotic agent. Cases attributable to serotypes targeted by PCV15, PCV20, and PCV21 that were nonsusceptible to ≥3 antibiotic classes accounted for 7.4% (4.7-11.1%), 8.5% (5.8-12.1%), and 12.6% (9.2-17.5%) of all non-bacteremic pneumococcal pneumonia cases, and 8.4% (5.3-12.5%), 9.4% (6.2-13.4), and 14.4% (10.4-20.0%) of all pneumococcal sinusitis cases. Conclusions: We demonstrated high proportions of antibiotic nonsusceptibility in non-bacteremic pneumococcal pneumonia and sinusitis in US adults. PCVs and antibiotic stewardship may mitigate antibiotic nonsusceptibility in pneumococcal disease.

  PublisherOA PDFDOI

• Strength and durability of indirect protection against SARS-CoV-2 infection through vaccine and infection-acquired immunity

  Nature Communications · 2025-01-29 · 7 citations

  articleOpen access

  Early investigation revealed a reduced risk of SARS-CoV-2 infection among social contacts of COVID-19 vaccinated individuals, referred to as indirect protection. However, indirect protection from SARS-CoV-2 infection-acquired immunity and its comparative strength and durability to vaccine-derived indirect protection in the current epidemiologic context of high levels of vaccination, prior infection, and novel variants are not well characterized. Here, we show that both vaccine-derived and infection-acquired immunity independently yield indirect protection to close social contacts with key differences in their strength and waning. Analyzing anonymized SARS-CoV-2 surveillance data from 9,625 residents in California state prisons from December 2021 to December 2022, we find that vaccine-derived indirect protection against Omicron SARS-CoV-2 infection is strongest within three months of COVID-19 vaccination [30% (95% confidence interval: 20-38%)] with subsequent modest protection. Infection-acquired immunity provides 38% (24-50%) indirect protection for 6 months after SARS-CoV-2 infection, with moderate indirect protection persisting for over one year. Variant-targeted vaccines (bivalent formulation including Omicron subvariants BA.4/BA.5) confer strong indirect protection for at least three months [40% (3-63%)]. These results demonstrate that both vaccine-derived and infection-acquired immunity can reduce SARS-CoV-2 transmission which is important for understanding long-term transmission dynamics and can guide public health intervention, especially in high-risk environments such as prisons.

  PublisherOA PDFDOI

• Clinical progression parameters associated with SARS-CoV-2, influenza, and respiratory syncytial virus infections

  medRxiv · 2025-05-29

  preprintOpen accessSenior authorCorresponding

  ABSTRACT Mathematical and computational models are often used to forecast respiratory infectious disease burden, including to inform healthcare capacity requirements. We aimed to characterize pathways of clinical progression associated with SARS-CoV-2, influenza, and respiratory syncytial virus (RSV) infections using data from patients in an integrated healthcare system, whose encounters were monitored across all levels of acuity spanning virtual, ambulatory, and inpatient care settings. Using parametric survival models, we estimated probabilities of progression and distributions of time to progression from each care setting to all higher-acuity settings on a cascade encompassing the following classes of events or healthcare encounters: symptoms onset; diagnostic testing; telehealth or other virtual care appointment; outpatient physician office visit; urgent care presentation; emergency department presentation; hospital admission; mechanical ventilation; and death. Our analyses included data from 59,668, 22,705, and 1,668 episodes associated with positive SARS-CoV-2, influenza, and RSV tests, respectively, between 1 April 2023 and 31 March 2024. First clinical encounters occurred in inpatient settings for only 4.7%, 3.4%, and 18.7% of SARS-CoV-2, influenza, and RSV episodes, respectively, with median times (interquartile range) of 6.8 (3.6-13.2), 6.6 (3.5-12.1), and 6.4 (3.8-10.6) days from symptoms onset to admission. Overall, 7.9% of SARS-CoV-2 episodes, 5.8% of influenza episodes, and 33.8% of RSV episodes resulted in inpatient admission, ventilation, or death. Between 40.4-62.1%, 71.6-87.3%, and 47.9-58.7% of SARS-CoV-2, influenza, and RSV infections, respectively, had encounters in lower-acuity virtual care, outpatient, or urgent care settings. For all three viruses, the proportions of cases receiving care at each level of acuity increased with older age and greater numbers of comorbid conditions. Median durations of hospital stay were 4.2 (2.6-7.3), 4.0 (2.3-6.8), and 4.3 (2.5-7.4) days for SARS-CoV-2, influenza, and RSV episodes resulting in admission. These estimates provide a basis for modeling real-world clinical care requirements and the progression of respiratory viral infections. AUTHOR SUMMARY Models of respiratory infections such as SARS-CoV-2, influenza, and RSV are used to forecast disease burden and plan the allocation of healthcare resources. Early healthcare system encounters—for instance, for receipt of care in virtual or outpatient settings—may provide a basis for anticipating higher-acuity care requirements as patients progress to more severe disease. However, limited data are available addressing patterns of healthcare utilization among patients with these infections. Using electronic healthcare records from an integrated healthcare system, we estimated probabilities and rates of progression from lower-acuity states, such as virtual or outpatient visits, to increasingly higher-acuity states including inpatient admission, ventilation, and death. We quantified associations of demographic and clinical risk factors with progression probabilities for each infection. We provide a databank containing fitted distributions for progression to inform infectious disease modeling.

  PublisherOA PDFDOI

• Pneumococcal Serotype Distribution and Coverage of Existing and Pipeline Pneumococcal Vaccines

  The Journal of Infectious Diseases · 2025-07-21 · 7 citations

  articleOpen accessSenior author

  BACKGROUND: Next-generation pneumococcal conjugate vaccines (PCVs) target an expanding array of serotype antigens. We assessed the proportions of invasive pneumococcal disease (IPD) and pneumococcal acute respiratory infections (ARIs) caused by serotypes targeted by existing and pipeline PCVs, and the annual United States pneumococcal disease burdens potentially preventable by these products. METHODS: We estimated serotype distribution and proportions of pneumococcal ARIs (acute otitis media [AOM; children only], sinusitis, nonbacteremic pneumonia) and IPD attributable to serotypes targeted by each PCV using Markov chain Monte Carlo approaches incorporating data from epidemiological studies and Active Bacterial Core surveillance. We then estimated annual numbers of outpatient-managed ARIs, nonbacteremic pneumonia hospitalizations, and IPD cases potentially preventable by PCVs by multiplying disease incidence rates by PCV-targeted disease proportions and vaccine effectiveness estimates. RESULTS: In children, PCV15, PCV20, PCV24, PCV25, and PCV31 serotypes account for 16% (95% confidence interval, 15%-17%), 31% (30%-32%), 34% (32%-35%), 43% (42%-44%), and 68% (67%-69%) of pneumococcal AOM, respectively. In adults, PCV15, PCV20, PCV21, PCV24, PCV25, and PCV31 serotypes account for 43% (38%-47%), 52% (47%-57%), 69% (64%-73%), 65% (61%-70%), 62% (57%-67%), and 87% (83%-90%) of pneumococcal nonbacteremic pneumonia. For IPD, 42%-85% of pediatric and 42%-94% of adult cases were due to PCV-targeted serotypes. PCV-preventable burdens encompassed 270 000-3 300 000 outpatient-managed ARIs, 2000-17 000 pneumonia hospitalizations, and 3000-14 000 IPD cases annually. CONCLUSIONS: Across pneumococcal conditions, coverage and preventable burdens were lowest for PCV15 and highest for PCV31, with PCV21 also targeting sizeable burdens of adult disease. Comparative estimates of preventable disease burden may inform future policy.

  PublisherOA PDFDOI

• Pediatric Antibiotic Use Associated With Respiratory Syncytial Virus and Influenza in the United States, 2008–2018

  The Journal of Infectious Diseases · 2025-06-05 · 3 citations

  articleOpen accessSenior author

  BACKGROUND: Understanding the contributions of respiratory syncytial virus (RSV) and influenza infections to pediatric antibiotic use is limited. We aimed to estimate the proportions and incidence of outpatient antibiotic prescriptions associated with RSV and influenza infections in a sample of commercially insured US children. METHODS: We conducted a retrospective study of outpatient antibiotic prescriptions dispensed to children in the Optum Clinformatics DataMart from 2008 to 2018. We used negative binomial time series models regressing weekly antibiotic prescriptions against RSV and influenza circulation measures to estimate counterfactual rates of antibiotic prescriptions in the presence and absence of RSV and influenza circulation overall and by age group, census division, and antibiotic class. We considered syndromic RSV and influenza measures (medical claims) and laboratory measures (National Respiratory and Enteric Virus Surveillance System) and controlled for age, division, 13-valent pneumococcal conjugate vaccine introduction, and seasonal and secular trends. RESULTS: An estimated 6.3% (95% CI, 5.2%-7.3%) and 3.4% (3.1%-3.8%) of antibiotic prescriptions were associated with RSV and influenza, respectively. These estimates translate to 72.6 (59.7-85.9) RSV-associated and 40.0 (35.1-45.1) influenza-associated antibiotic prescriptions per 1000 children annually. RSV-associated antibiotic prescription incidence was highest among children aged ≤5 years while influenza-associated antibiotic prescriptions were highest among children >5 years. Macrolides were the antibiotic class for which RSV and influenza accounted for the greatest share of prescribing. CONCLUSIONS: RSV and influenza are associated with meaningful proportions of pediatric antibiotic prescriptions. Measures to prevent RSV and influenza infections in children, including immunization, may reduce antibiotic use and aid in mitigating antibiotic resistance.

  PublisherOA PDFDOI

• P-604. Preliminary real-world Abrysvo vaccine effectiveness (VE) against Respiratory Syncytial Virus (RSV)-related lower respiratory tract disease (LRTD) hospitalizations and emergency department (ED) visits—Kaiser Permanente of Southern California (KPSC), November 2023-April 2024

  Open Forum Infectious Diseases · 2025-01-29

  articleOpen access

  Abstract Background While the pivotal randomized clinical trial documented the efficacy of the RSVpreF vaccine (Abrysvo) against RSV-related LRTD, the limited number of hospitalizations and ED visits hampered VE assessment for these clinical outcomes. We evaluated Abrysvo’s effectiveness against 1st occurrence of RSV-related inpatient or ED LRTD in KPSC, a large US healthcare system.Table 1.Characteristics of study population, KPSC Methods In this test-negative case control study, we assessed VE among adults ≥60 years of age by analyzing KPSC members’ electronic health records and enhancing specimen testing from 11/24/2023-4/9/2024. Events from inpatient or ED settings meeting an ICD-10 code-based LRTD case definition with a respiratory specimen tested on GenMark RP2 expanded multiplex PCR assay were included. Cases were defined as RSV+ episodes. Analyses were conducted using two sets of pre-specified controls: 1) ‘Strict’: RSV- and positive for a non-vaccine preventable disease (VPD) etiology, and 2) ‘Broad’: all RSV-. The strict approach accounted for lower sensitivity of RSV testing among adults and potential bias associated with VPD controls. Exposure was defined as Abrysvo receipt ≥21 days before encounter. Adjusted odds ratios (OR) and 95% confidence intervals (CI) were estimated from a multivariable logistic regression model with adjustment for patient demographic and clinical characteristics. VE was calculated as 1−OR multiplied by 100%.Table 1.Characteristics of study population, KPSC, cont. Results The overall study population included 4,469 KPSC members: &amp;gt;50% were ≥75 years, 93% had ≥1 comorbidities, and 15% immunocompromised. In the strict analysis, there were 425 cases and 522 controls who received Abrysvo a median of 59 days before illness onset; VE compared to no vaccine receipt was 84% (95% CI: 29–97). In the broad analysis, there were 425 cases and 4,044 controls who received Abrysvo a median of 53 days before illness onset; adjusted VE was 82% (95% CI: 28–96).Table 2.Abrysvo Vaccine Effectiveness against RSV LRTD hospitalizations and ED visits, KPSC Conclusion This study demonstrated real world VE of Abrysvo during its first 5 months of use against severe LRTD in the hospital and ED settings among US adults ≥60 years of age. These results expand on efficacy results from clinical trial populations by including a substantial number of persons age 75+ years and with comorbidities and immunocompromising conditions. Disclosures Sara Y. Tartof, PhD MPH, GSK: Grant/Research Support|Pfizer Inc: Grant/Research Support Negar Aliabadi, MD, MS, Pfizer Inc: employment|Pfizer Inc: Stocks/Bonds (Public Company) Jeff Slezak, MS, Dynavax Technologies: Grant/Research Support|Pfizer: Grant/Research Support Vennis Hong, MPH, Pfizer: Grant/Research Support Bradley Ackerson, MD, Dynavax: Grant/Research Support|GlaxoSmithKline: Grant/Research Support|Moderna: Grant/Research Support|Pfizer: Grant/Research Support Qing Liu, M.S., Pfizer Inc.: Stocks/Bonds (Public Company) Sally Shaw, DrPH, MPH, Pfizer: Grant/Research Support Sabrina Welsh, MPH, Pfizer Inc: Stocks/Bonds (Public Company) Julie Stern, MPH, GlaxoSmithKline: Grant/Research Support|Pfizer: Grant/Research Support|Sanofi: Grant/Research Support Michael Dutro, PharmD, Pfizer: Stocks/Bonds (Public Company) Erica Chilson, PharmD, Pfizer Inc: Employee|Pfizer Inc: Stocks/Bonds (Public Company) Elisa Gonzalez, MS, Pfizer: Stocks/Bonds (Private Company) Robin Hubler, MS, Pfizer, Inc.: Employee of Pfizer Inc.|Pfizer, Inc.: Stocks/Bonds (Private Company) Luis Jodar, PhD, Pfizer Inc: Employment|Pfizer Inc: Stocks/Bonds (Public Company) Bradford D. Gessner, M.D., M.P.H., Pfizer: Employee|Pfizer: Stocks/Bonds (Public Company) Elizabeth Begier, MD, M.P.H., Pfizer Vaccines: Employee|Pfizer Vaccines: Stocks/Bonds (Private Company)

  PublisherOA PDFDOI

• Can We Compare Attributable Risk of Adverse Events with the Self-Controlled Case Series Design in Vaccine Safety Studies? A Use Case of Guillain-Barré Syndrome

  medRxiv · 2025-05-08 · 1 citations

  preprintOpen access

  Abstract Background Estimating attributable risk (AR) through self-controlled case series (SCCS) analyses alone may limit generalizability because SCCS only incorporate vaccinated patients with the outcome (e.g., Guillain-Barré syndrome [GBS]) who may differ from the overall population recommended for vaccination. Objective We aimed to demonstrate background event incidence rates’ impact on vaccine-specific GBS ARs and to standardize ARs across different vaccine studies by applying a generalizable, population-based GBS background rate for improved comparability and to better estimate the expected population-level ARs. Methods We identified post-licensure SCCS vaccine studies and GBS background rates using US Medicare data via targeted literature review. GBS control period rates from SCCS vaccine studies were extracted or calculated. Population-level ARs were calculated for each vaccine using published background GBS rates and the original SCCS-generated incidence rate ratios (IRRs). Results Published vaccine-specific GBS IRRs ranged from 2.02 (95%CI: 0.93–4.40) for RSVPreF to 4.96 (95%CI: 1.43–17.27) for recombinant zoster vaccine (RZV). Study-specific ARs per 100,000 doses ranged from 0.28 (H1N1) to 0.90 (RSVPreF). Vaccines with lower control period GBS rates had a lower attributable risk for a given IRR. After standardization using published background GBS rates, population-level ARs were higher for vaccines with higher IRRs. For example, when using the H1N1 vaccine control period rate as the GBS background rate for AR calculation, RSVPreF had the lowest AR per 100,000 doses (0.21) and RZV the highest (2.37). Conclusion AR calculation is dependent on the control period rate within a given SCCS analysis. ARs derived exclusively from SCCS analyses may lead to incorrect conclusions regarding an adverse event’s absolute risk if included in product labels due to a lack of external validity. Using a representative background rate from the recommended vaccinee population, along with SCCS-derived IRR, to calculate the expected population-level AR may offer more accurate vaccine risk-benefit assessments.

  PublisherOA PDFDOI

Frequent coauthors

• Nicholas P. Jewell

  London School of Hygiene & Tropical Medicine

  73 shared

• Kristin L. Andrejko

  Centers for Disease Control and Prevention

  65 shared

• Ana M. Mora

  University of California, Berkeley

  50 shared

• Katherine Kogut

  Center for Environmental Health

  49 shared

• Brenda Eskenazi

  Center for Environmental Health

  47 shared

• Nina Holland

  45 shared

• Stephen Rauch

  University of California, Berkeley

  43 shared

• Maximiliano Cuevas

  40 shared