H5N1 in US Dairy 2024 to 2026: A Slow-Motion Pandemic Stress Test

The B3.13 spillover: from one Texas herd to a national epizootic #

The 25 March 2024 USDA APHIS confirmation of H5N1 in a Texas dairy herd was, in retrospect, not the first infection. Phylogenetic analysis published in Nature in July 2024 (Caserta et al., 2024) traced the spillover to a single wild bird to cow transmission event in the Texas panhandle in late 2023, with subsequent cattle to cattle spread driven primarily by milking equipment, shared personnel, and lactating cow movements between herds. The viral genotype, designated B3.13, is a reassortant of Eurasian H5N1 clade 2.3.4.4b with North American low pathogenic avian influenza segments, first observed in wild birds in late 2023.

By April 2025 the USDA APHIS HPAI in livestock dashboard listed approximately 990 herd-level detections across 17 states, with California, Colorado, Michigan, Idaho, and Texas accounting for over three quarters of confirmed herds. California, which entered the outbreak in late August 2024, became the dominant focus in early 2025, recording over 760 herd detections by the end of Q1 2025 (CDFA HPAI dashboard). By April 2026 cumulative herd detections exceeded 1,070 nationally, with active genomic surveillance from USDA NVSL and academic groups (Cornell, St. Jude) continuing to track the B3.13 clade.

The economic damage at the herd level is mostly recoverable: affected cows show a 20 to 40 percent drop in milk yield for two to three weeks, with most herds returning to baseline within sixty days. The USDA Emergency Livestock Relief and the Dairy Herd Status Program have together disbursed roughly 1.7 billion dollars in indemnity, biosecurity, and lost-milk payments through Q1 2026 (USDA APHIS press releases, 2024 to 2026). The deeper damage is reputational and structural: shaken consumer confidence in milk, pressure on Mexico-bound dairy exports, and the slow erosion of the assumption that influenza A in cattle is a non-event.

Human cases: 70 confirmed, the Louisiana fatality, the Missouri mystery #

As of the CDC HPAI human case dashboard updated April 2026, the United States has recorded approximately 70 confirmed human H5N1 cases since the start of 2024, the largest mammalian-source human cluster in the country's history. Of these, 41 were associated with dairy cattle exposure (predominantly California and Colorado farmworkers), 23 with poultry culling operations (Colorado and Washington), and a small number with backyard flock contact or undetermined exposure. The dominant clinical presentation has been mild conjunctivitis and brief upper respiratory symptoms, treated with oseltamivir. Most cases recovered without hospitalization.

Two cases broke the mild pattern. In December 2024 Louisiana confirmed a severe H5N1 case in a patient over 65 with backyard poultry exposure; the patient developed acute respiratory distress and died on 6 January 2025, the first US H5N1 fatality (CDC HAN 521, Louisiana Department of Health). Genomic analysis identified the D1.1 genotype, distinct from the B3.13 dairy clade, with low-frequency hemagglutinin mutations (PB2 627K and HA 226L) suggestive of within-host adaptation but not transmissibility. A second severe case, a teenager in British Columbia in November 2024, required ECMO and survived; that case also carried D1.1 with similar adaptive markers (BC CDC, NEJM correspondence 2024).

The most unsettling 2024 event remains the Missouri spillover of September 2024: an adult hospitalized with chest pain and other symptoms tested positive for H5N1 with no documented animal, dairy, or poultry exposure. A household contact later seroconverted. CDC investigation concluded the source was undetermined. Whether this represents undetected environmental exposure, a missed common-source contact, or a brief and contained human-to-human chain remains unresolved (CDC MMWR, October 2024). For pandemic preparedness this is the single most important data point of the outbreak: it is the case that cannot be fully explained, in a virus that is one mutation set away from efficient airborne transmission.

Milk safety regime and the federal testing order #

The FDA's bulk tank surveillance program, launched April 2024 and expanded under the USDA Federal Order of 28 April 2024, sampled retail milk products across 17 states in the first phase and detected viral RNA fragments by qPCR in roughly 20 percent of pasteurized retail samples. Critically, follow-up egg inoculation and cell culture studies at FDA, USDA ARS, and university partners found no infectious virus in any pasteurized retail product. The HTST pasteurization standard (72 degrees Celsius for 15 seconds) inactivates H5N1 below detection in spiked-milk experiments at the FDA Laurel and Cornell laboratories (Spackman et al., 2024; Schaefer et al., 2024). Raw milk products, by contrast, contain infectious virus when sourced from affected herds, and the FDA has formally advised against raw milk consumption from any herd with H5N1 detections.

On 13 December 2024 the USDA issued a National Milk Testing Strategy federal order requiring nationwide bulk tank surveillance, mandatory reporting of positive results, and herd-level testing of any dairy with confirmed virus. By April 2026 over 40 states had entered Stages 1 to 4 of the strategy, with Stage 4 (free-status surveillance) achieved in 8 states. The federal order resolved the 2024 paradox of voluntary testing, where many producers declined surveillance for fear of market or labor disruption, by making participation a condition of interstate movement under 9 CFR 86.

The remaining gap is on the labor side. There is no federal PPE mandate for dairy workers handling lactating cows in HPAI-positive herds, and uptake of voluntary PPE (face shields, N95s, gloves) was below 30 percent in the CDC and NIOSH 2024 farmworker assessments in Colorado and Michigan. A substantial share of the dairy workforce is undocumented and reluctant to seek testing or report symptoms; CDC and state health departments have used Spanish-language outreach and 75 dollar incentive payments to lift testing rates, with mixed results. The Salus health platform tracks this surveillance gap as the highest-leverage and lowest-cost lever in the US response.

Egg price echoes: 166 million birds, an 8 dollar dozen, and a political headache #

The dairy outbreak has unfolded in parallel with the broader 2022 to 2026 H5N1 panzootic in commercial poultry, which is the larger immediate consumer-price story. Per USDA APHIS commercial poultry data, cumulative bird losses from HPAI control measures (depopulation of infected and exposed flocks) reached approximately 166 million birds by April 2026, including 130 million layers and pullets. Q1 2025 saw the worst layer flock losses on record, with over 30 million laying hens depopulated in January and February 2025 alone (USDA APHIS Stamping Out reports).

The price response was sharp. BLS Producer Price Index data for eggs (PPI series WPS017106) rose 156 percent year on year in February 2025. USDA AMS daily wholesale large eggs (Midwest, grade A, loose) peaked at 8.17 dollars per dozen on 25 February 2025, more than triple the 2.50 dollar baseline. Retail prices, tracked in the BLS CPI series APU0000708111, reached 6.23 dollars per dozen in March 2025, the highest nominal value on record. By April 2026 wholesale had retreated to roughly 2.90 dollars per dozen as flocks were rebuilt, but cage-free supplies remained tight and California Proposition 12 compliant retail eggs continued to trade at a 35 to 50 percent premium.

The political and trade consequences extended beyond grocery aisles. The Trump administration moved to expand egg imports from Turkey, Brazil, and South Korea in March 2025, and Mexico maintained partial restrictions on US dairy and poultry from affected states. The Mexico-bound dairy export risk is non-trivial: Mexico is the largest single buyer of US dairy products (4.4 billion dollars in 2024 per US Dairy Export Council), and a full SPS suspension would dwarf the herd-level losses.

The vaccine and antiviral stockpile #

The federal preparedness response, coordinated through ASPR and BARDA, has assembled a layered H5N1 vaccine and antiviral stockpile. BARDA's existing pre-pandemic agreements with CSL Seqirus, GSK, and Sanofi cover egg-based and cell-based H5N1 vaccines, several of which use the AS03 or MF59 adjuvant systems and are indexed against clade 2.3.4.4b strains close to B3.13. As of mid-2025 HHS reported approximately 10 million prefilled doses available within weeks of activation, with surge capacity to roughly 100 million doses over four to six months given the existing fill-finish footprint.

On 4 October 2024 BARDA announced a 176 million dollar contract with Moderna to advance an mRNA-based H5N1 vaccine candidate (mRNA-1018) into Phase 3 trials, with a 590 million dollar option for additional pandemic influenza mRNA work signed in early 2025. Phase 1 and Phase 2 readouts published in early 2025 showed robust HAI titers against H5 antigens at 50 microgram and 100 microgram doses with a tolerable reactogenicity profile (Moderna Q1 2025 earnings call, NIAID DMID 23-0001 trial results). Pfizer separately holds a smaller BARDA contract for mRNA-based H5N1 development. The mRNA platforms remain the most credible path to surge capacity in a pandemic, with theoretical scale of several hundred million doses per quarter.

On the antiviral side, oseltamivir is the front line and the Strategic National Stockpile holds tens of millions of treatment courses. Baloxavir marboxil (Xofluza) is in the SNS but at lower volume. CDC monitoring through the IRAT framework continues to score B3.13 as moderate emergence and high impact risk. The combined pharmaceutical posture is materially better than 2009, but the binding constraint in a pandemic would remain fill-finish capacity, distribution, and uptake, not bulk antigen.

Reassortment risk and a 2026 to 2027 stress test #

The pandemic risk is not the current B3.13 virus, which has poor airborne transmissibility in mammals and binds preferentially to avian-type sialic acid receptors. The risk is reassortment. If H5N1 co-infects a host (a pig, a human, or potentially a cow) carrying seasonal H1N1 or H3N2, internal gene swaps could produce a virus retaining the H5 antigen (against which the population is largely naive) while gaining mammalian transmissibility. The 2009 H1N1 pandemic emerged from such triple reassortment in swine. WHO's 2025 global preparedness assessment and the CDC IRAT framework both flag this scenario explicitly.

Salus's stress-test model for a 2026 to 2027 H5N1 pandemic assumes a moderately transmissible reassortant emerging in late 2026, a 9 to 14 month vaccine ramp, and a clinical attack rate of 25 percent in the first wave. Direct healthcare costs run 35 to 60 billion dollars over 18 months. Lost output and labor force absences add another 60 to 160 billion dollars depending on the case fatality rate (modeled at 0.4 to 1.2 percent across scenarios), school and childcare closures, and the duration of voluntary distancing. Total economic loss in the central scenario sits at 95 to 220 billion dollars, an order of magnitude smaller than COVID-19 but large enough to compress GDP growth by 80 to 180 basis points in the first pandemic year.

The policy ledger is short and concrete. Close the dairy-worker surveillance gap with mandatory PPE and protected testing for undocumented workers. Fund the National Milk Testing Strategy through Stage 4 in all 48 contiguous states. Expand BARDA mRNA contracts to lock in 2027 fill-finish capacity. Maintain genomic surveillance funding (NVSL, USDA ARS, academic) at current levels rather than the proposed FY26 cuts. None of these are expensive. All of them are cheaper than the next pandemic.

Sources #

• USDA APHIS, HPAI in livestock dashboard, updated April 2026
• USDA APHIS, Federal Order on testing and reporting of HPAI in livestock, 24 April 2024
• USDA AMHS, National Milk Testing Strategy Federal Order, 13 December 2024
• FDA, Updates on bulk tank milk and retail milk surveillance, 2024 to 2025
• CDC, H5 bird flu human case dashboard, updated April 2026
• CDC HAN 521, Severe H5N1 case in Louisiana, 18 December 2024
• Caserta et al., Spillover of H5N1 to dairy cattle, Nature, July 2024
• Spackman et al., Pasteurization inactivates H5N1 in milk, NEJM, 2024
• BARDA, Moderna mRNA-1018 H5N1 vaccine award, 4 October 2024
• BLS, Producer Price Index, eggs for fresh use (WPS017106) and CPI eggs (APU0000708111)
• USDA AMS, Daily wholesale shell egg report, February 2025 peak
• WHO, Global influenza preparedness assessment, 2025