Transmission routes of highly pathogenic H5N1 between cows or to humans remain unclear due to limited data from affected dairy farms. We performed air, farm wastewater, and milk sampling on 14 H5N1 positive dairy farms across two different California regions. Infectious virus was detected in the air in milking parlors and in wastewater streams, while viral RNA was found in exhaled breath of cows. Sequence analysis of infectious H5N1 virus from air and wastewater samples on one farm revealed viral variants relevant for potential human susceptibility. Longitudinal analysis of milk from the individual quarters of cows revealed a high prevalence of subclinical H5N1 positive cows. Additionally, a heterogeneous distribution of infected quarters that maintained a consistent pattern over time was observed, inconsistent with shared milking equipment serving as the sole transmission mode. The presence of subclinically infected cows was further supported by detection of antibodies in the milk of animals that exhibited no clinical signs during the H5N1 outbreak on one farm. Our data highlight additional sources and potential modes of H5N1 transmission on dairy farms.

The outbreak of H5N1 in dairy cattle in the United States revealed challenges in identification and management of a novel disease. The virus showed an exceptional ability to spread between farms and among cows within a farm. The impact of the virus on dairy cattle varied from nonclinical to severe clinical signs and death. Many dairy producers did not report clinical signs in their cows or test for the virus. Cats and peridomestic birds on many affected dairies died from viral exposure. Dairy workers showed signs of conjunctivitis, which was confirmed to be due to H5N1. With the disease affecting multiple species and showing efficient cow-to-cow transmission, the situation only worsened. There was a negative impact on the relationships among dairy producers, dairy workers, poultry producers, and veterinarians, in which professional and personal relationships were severed and some experienced loss of employment. The regulatory response varied by geographic location, and in some states, animal health and human health authorities elevated producer fears of consequences of reporting. Authorities did quickly confirm that pasteurization inactivated the virus in dairy products and showed that it was very rarely detected in beef from affected cows. In this review, we will describe the relationships among the dairy industry and animal and human health and why the H5N1 outbreak requires a One Health perspective of all stakeholders in order to address it effectively.

Highly pathogenic avian influenza virus of the H5N1 subtype has been infecting U.S. dairy cattle and spreading among dairy operations since March 2024. H5N1 surveillance systems for dairies are needed, but information on whether environmental sampling can inform these systems is lacking. To guide a surveillance framework, we determined the environmental sites with Influenza A virus (IAV) detection on H5N1-affected dairies (n = 25) in four states (California, Colorado, Michigan, and Ohio) and explored sample characteristics that may have influenced detection. A total of 581 samples from dairy environmental sites were characterized for IAV RNA via rRT-PCR, and classified into six categories. A total of 94 samples (16.2%) had IAV detected, and the Ct values measured from these samples were typically higher than those measured in bulk tank milk from a subset of sampled herds. A majority of IAV detections were made from the following site categories: milking equipment/personal protective equipment, parlor surfaces, and wastewater/lagoons/manure. These results suggest that environmental sites most likely to be contaminated with IAV on dairies are those with exposure to milk. Meanwhile, mixed effect logistic modeling showed that days into an outbreak that samples were collected was associated with IAV detection. These results provide a framework within which to continue the assessment of environmental sampling as a surveillance tool for dairy H5N1. This framework can be strengthened by studies that perform further IAV viral characterization and collect samples from sites prior to, during, and after H5N1 outbreak periods.

Highly pathogenic avian influenza virus H5N1 has been infecting dairy herds in the U.S. since its initial incursion into cows in early 2024. Although national strategies have aimed to detect affected herds, the best way to surveil herds for the H5N1 virus has not been formally studied and we also do not understand herd-level patterns of infection. To understand infection patterns of H5N1 in dairy herds over time, we conducted early surveillance of non-affected farms in California in the Fall of 2024 in an observational study. Daily bulk tank milk (BTM) samples were submitted from each herd and tested for influenza A (IAV) via rRt-PCR. In a subset of herds, IAV testing of multiple excretion types from cattle of different classes and pen-level daily milk was also completed soon after BTM detection. Daily detections of IAV occurred in BTM for a minimum of 33 days, with some herds continuing to have detection beyond a 75-day window. BTM Ct nadirs were seen between 1-3 weeks of detection. In herds that were tested, virus was detected in the milk from all pens of cattle within a very short time frame after BTM detection, or prior to the initiation of pen level sampling. A very low percentage (2.8%) of individual cow samples tested positive for IAV when collected soon after BTM detection, and although the virus was found in all excretion types, a majority of positive samples were from milk. This suggests that BTM may be the best early indicator of herd infection, and that movement of the virus to all lactating pens of cattle after herd incursion is relatively quick. These results also suggest that surveillance strategies with a long interval between BTM testing days may miss herds with short infection windows. Because most herds experienced test days where some submitted BTM samples had virus detected while others did not, and virus was detected in pen level milk samples when the BTM from the herd had become test negative, this work also highlights the necessity of studying the test sensitivity of IAV rRt-PCR detection in aggregate milk samples. Interpretive summary: This study characterized daily bulk tank and pen-level milk detections of H5N1 virus on affected dairies. Viral loads in bulk tank samples peaked within three weeks and declined by two months, though some herds showed prolonged detection. Individual cow testing revealed that bulk tank milk testing is more sensitive for H5N1 early outbreak detection, and also identified the likely presence of non-clinically affected cows shedding virus. Pen-level sampling revealed rapid herd-level spread, emphasizing the need for frequent and comprehensive milk sampling to improve detection sensitivity and support effective outbreak management for H5N1 on dairy operations.

Highly pathogenic avian influenza virus H5N1 has been infecting dairy herds in the US since its initial incursion into cows in early 2024. Although national strategies have aimed to detect affected herds during this outbreak, information regarding longitudinal herd-level viral detection has not been formally studied. To help understand viral shedding of H5N1 in milk from H5N1-affected dairy herds over time, an outbreak investigation was conducted on dairy farms in California in the Fall and Winter of 2024. The primary objective of this observational investigation was to identify initial H5N1 infection in these CA herds and then evaluate viral distribution via longitudinal bulk tank milk (BTM) testing. A secondary objective was to evaluate viral movement via longitudinal pen-level milk (PLM) sampling in a subset of affected herds. Daily BTM samples were submitted from 19 originally non-infected herds that were clients of a single veterinary clinic and volunteered to participate in this study. Samples were tested for influenza A (IAV) via rRT-PCR. In a subset of 5 herds, daily PLM IAV testing was completed soon after initial BTM IAV detection. Overall, a total of 4,749 BTM and 2,922 PLM samples were submitted for testing, and all herds eventually became infected with IAV. After initial detection, daily detections of IAV occurred in BTM for a minimum of 33 d, with some herds continuing to have detection at the end of the sampling period (>75 d). BTM Ct nadirs were seen 1-3 weeks after initial detection. In PLM-tested herds, virus was detected in samples from all pens of cattle soon after BTM detection, and in some cases was seen in all pens on the first day of PLM sampling. Overall, these results support BTM testing as an effective early detection tool for dairy H5N1 outbreaks but suggest that surveillance strategies that do not test aggregate milk from all lactating cows on an operation, or have a long interval between BTM testing days, may fail to identify all affected herds. Future investigations should study farm-level factors associated with long-term viral detection in some herds and assess detection sensitivities of the IAV rRT-PCR assay used on aggregate milk samples.

Since the initial detection in dairy herds in March 2024, Highly Pathogenic Avian Influenza H5N1 has spread extensively in the United States with over 1000 confirmed cases in dairy cattle across 17 states. Data on within-herd transmission of H5N1 are limited with fundamental knowledge gaps. We used a simplified disease transmission model and an Approximate Bayesian Computation (ABC) algorithm to estimate H5N1 within-herd transmission model parameters using disease morbidity, laboratory testing data from an outbreak herd, and experimental inoculation studies. The estimated adequate contact rate was used to simulate disease spread and predict the time to exceed various threshold fractions of cattle with clinical signs. The estimated adequate contact rate in the baseline model scenario was 0.78 (95% C.I.,0.64-0.97) per day and the estimated basic reproduction number was 8 (95% C.I.,7-11). Based on simulation model predictions for a known infected herd with 3,433 lactating cows, it took more than 2 weeks from disease introduction to attain a clinical signs prevalence threshold of 5 percent. The estimated parameters are essential for informing surveillance design, outbreak management approaches, risk analyses, and regional transmission models. Incorporating data from future experimental studies and outbreak herds may enhance model precision and help characterize the variability of transmission patterns.