The objective of this evaluation was to estimate the risk for infection in persons responding to outbreaks by using data retrospectively obtained during December 2014–September 2017. We used several sources of data. We identified the domestic poultry detection events that were reported to USDA. Then, to identify the number of persons exposed during a response (i.e., responders), we used USDA reports from each incident and limited to persons who were deployed to the field. For the numerator, we used state reports to CDC of any ill responders and testing results. Specimens were tested for influenza viruses by reverse transcription PCR. We calculated the percent positive among exposed official responders and estimated 95% binomial CIs.

We found 264 detections of H5 or H7 viruses in poultry across 20 states; most (≈92%) were during the outbreak of infection with gs/GD H5 HPAI virus during 2014–2015, and 4,555 responders were potentially exposed to a virus (Table 2). Responders were from 3 main groups: USDA, USDA contractors, and state/local agriculture. All responders were recommended to receive seasonal influenza vaccine and were assumed to have properly worn adequate personal protective equipment (PPE) (4); no data were systematically collected on PPE breaches. CDC did not recommend antiviral chemoprophylaxis for persons using proper PPE. The APHIS Health and Safety and PPE Guidance for HPAI and APHIS-CDC monitoring guidance applied to all responders. Twenty-three persons became ill and were tested; no human infections with avian influenza viruses were detected. The risk for infection with avian influenza for responders was low, although our power to make this statement with confidence varied by year and virus because there was a wide range (74–3,962) of number of persons involved in each response. The data were most robust during the outbreaks of infection with gs/GD/96 H5N8 and reassortant HPAI H5N2 viruses in poultry during 2014–2015.

These results complement data previously published for outbreaks in the United States during 2014–2015, which found no avian influenza infections in 164 persons mostly exposed while not wearing PPE (5). Animal model data also support these epidemiologic data. The gs/GD/96 H5N8 and reassortant HPAI H5N2 viruses can replicate efficiently in the respiratory tract of ferrets, but illness was mild and did not transmit through direct contact between ferrets (6). North American lineage H7N8 virus also replicated in ferrets, and pathogenicity was greater for HPAI viruses than for LPAI viruses. Limited transmission in ferrets through direct contact was observed only for the LPAI virus (7). Similarly, North American lineage LPAI H7N9 viruses demonstrated limited transmissibility through direct contact in ferrets (8). The viruses from each of these events were evaluated by using the Influenza Risk Assessment Tool (9), which is used to assess the potential pandemic risk. The North American H7N8 and H7N9 viruses had low risk, and the gs/GD H5N8 and reassortant H5N2 and H5N1 viruses had low to moderate risk.

On the basis of these data, CDC and USDA revised the recommendations for monitoring illness in responders (https://www.cdc.gov/flu/avianflu/h5/infected-birds-exposure.htm). When responding to H5 or H7 viruses similar to those previously encountered in the United States to date, we recommend passive monitoring (i.e., self-report if ill) for persons wearing adequate PPE. For response personnel with inadequate or no PPE, or after experiencing a breach of PPE, we continue to recommend active monitoring. If personnel are responding to an avian influenza virus of unknown origin, we recommend active monitoring during exposure and for 10 days postexposure, regardless of PPE use. The change in procedure will substantially reduce the workload for safety officers and public health officials, and shift much of the reporting responsibility to the individual.

A limitation to this report is the lack of information regarding breaches in PPE. It is likely that breaches occurred, which, in turn, might have increased the risk for transmission. PPE noncompliance had been well-documented in the healthcare setting (10). The fact that we were unable to detect such a transmission event while actively monitoring, combined with what we know about the genetic and receptor binding characteristics of these viruses and the animal studies (7–9,11), might suggest that the viruses are not well adapted to humans. Another limitation was the challenge to follow-up responders after they returned to their home states. These data further support the rationale for revising the current monitoring recommendations. In contrast, if infection causes mild illness, it might have been missed. It is possible that in revising the monitoring recommendations we will risk missing a human infection. However, we might expect to detect an illness severe enough that the person seeks medical care and is tested for influenza.

The gs/GD/96 lineage clade 2.3.4.4 HPAI H5 viruses continue to cause outbreaks in poultry and wild birds in other parts of the world (12). One report from Canada monitored 50 household members or animal caretakers on affected farms potentially exposed to gs/GD/96 reassortant H5N2 or H5N1 viruses during 2014–2015, and no infections were identified (13). A recent study of 23 countries in Europe on outbreaks of infection with gs/GD/96 H5N8 virus in poultry during 2016–2017 reported no infections among 524 exposed persons who were monitored by a mixture of active and passive monitoring (14). Globally, many more persons have likely been exposed, but no human infections with gs/GD/96 H5N8 clade 2.3.4.4 viruses have been reported. Novel influenza virus infections in persons are reportable to the World Health Organization through the International Health Regulations (15). Given the nature of influenza viruses, we will continue to monitor the epidemiology and the viruses. Going forward, CDC and USDA should prospectively collect data on exposure and PPE use to better define the risk for responders exposed to avian influenza viruses.