A total of 2,402 participants were enrolled from 390 households in the 5 study sites (Appendix Table). The median number of participants per household was 5 (range 1–20). Various domestic mammals and poultry were raised by the households, most frequently chickens and ducks. Approximately one third of the households raised pigeons and geese, whereas ≈12% raised turkeys (Table 1). Table 2 shows the distribution of poultry-raising practices followed by participants. Approximately 17% of households reported raising poultry within the household, whereas ≈53% kept the poultry on the roof and ≈30% in a barn. Dead poultry was disposed of by burying, discarding in a closed trash bag, or burning by 44% of the respondents, whereas 56% either dumped the carcasses in the open trash or in small water canals. Approximately 38% said that they would consult a veterinarian when they noticed ill poultry, and 14% said that they would isolate the ill birds from the rest of the flock. Approximately 33% of respondents said they would get rid of the sick bird while it was alive, whereas ≈12% said they would slaughter and eat it. One third of the respondents reported using avian influenza vaccines for their poultry, but only 1% of those were able to determine that the vaccine was used against H5N1 influenza virus, whereas the rest did not know what the vaccine was used against. Vaccination was mostly performed by a veterinarian (77%) or by a hired aid or a family member.

Demographic and health data of the participants are summarized in Table 3. Respondents’ age ranged from 2 to 102 years. The mean age was 25 years, and the median age was 21 years. The distribution of enrolled participants by sex, age (adults versus children), and marital status (single versus other) was almost the same. Most participants were uneducated or received only a primary education (68%), and 12% had a job as a skilled laborer or professional. Approximately 10% had chronic health problems, and ≈2% had long-term respiratory problems. Less than 10% reported currently using tobacco. Only 4 participants reported ever receiving the influenza vaccine.

The median number of days per week that the participants had direct contact with poultry was 6 days. The participants spent a median of 10 minutes per day in direct contact each time they came in direct exposure with the poultry. Table 4 summarizes the poultry exposure practices of the participants. More than 70% reported cleaning poultry cages or feeding poultry, whereas 30% reported not having any direct poultry exposure. Among those who had direct poultry exposure (cleaning or feeding), only 10% reported using a dedicated garment. Approximately one third of the participants reported slaughtering poultry. Slaughtered poultry were mostly kept in a dedicated barrel to bleed (80%), but 20% of respondents left that to occur in the open. Most of the respondents cleaned the used utensils after slaughtering, mostly by using soap and water. Slaughter waste was disposed in closed bags and thrown into the trash (46% of respondents), thrown in open trash (25%), or dumped into small canals (29%).

Serologic findings at baseline and follow-up period 1 (2017–2018) are shown in Table 5. Serum samples were successfully collected from 2,397 persons at baseline and from 2,051 at follow-up period 1. Seroprevalence of H1N1 antibodies was ≈30% and of H3N2 antibodies was ≈50%. At baseline, 9 (0.4%) participants had H5N1 antibodies. Four participants had a titer of 1:80, and 5 had a titer of 1:160. Of the 9 participants, 7 came from 2 adjacent homes. At follow-up period 1, only 4 (0.2%) participants were seropositive for H5N1 antibodies, 3 with 1:80 titers and 1 with a 1:160 titer. None of the participants who were seropositive at baseline remained seropositive at follow-up period 1. At baseline, 266 (11%) participants had H9N2 antibodies. Of these, the 227 had 1:80 titers, 37 had 1:160 titers, and 2 had 1:320 titers. Household clusters of seropositive persons were observed for 223 participants, with cluster sizes ranging from 2 to 10 participants. At follow-up period 1, only 3 (0.1%) participants had H9N2 antibodies. All participants had a 1:80 titer and were from the same household, and only 1 of these participants was positive at baseline with the same titer. No participants had H5N8 antibodies at baseline or follow-up period 1.

During follow-up period 1, a total of 400 participants (16.7% of the cohort) were confirmed to have ILI symptoms. Of these, 113 were positive for influenza A by RT-PCR (28% of those with ILI and 4.7% of cohort overall). Four case-patients were subtyped as infected with H5N1 virus by RT-PCR and confirmed by sequencing. The incidence of H5N1 infection in this cohort of 2,402 persons was 17 cases/10,000 exposed persons. Phylogenetic analysis revealed that the viruses causing the infection were of clade 2.2.1.2, which are unique to and endemic in Egypt.

The first case was in a 5-year-old boy with exposure to poultry at the household and at a live bird market. Only the swabs obtained on day 1 of the illness were positive for H5N1. The serum sample titer obtained on day 1 was <1:10, whereas a titer of 1:40 was detected on day 14. The boy’s symptoms included fever, cough, sore throat, myalgia, malaise, headache, runny nose, and diarrhea. Fever persisted for 4 days; cough and sore throat continued to occur throughout the 14-day follow-up period. No household contacts showed symptoms, virus shedding, or seroconversion.

Case 2 was in an 11-year-old girl with direct contact with chickens and ducks. Only the swabs obtained on day 1 of the illness were positive for H5N1. The girl’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, malaise, headache, and runny nose. Fever and sore throat persisted for 2 days; all other symptoms cleared by the fourth day. No household contacts showed symptoms, virus shedding, or seroconversion.

Case 3 was in a 5-year-old boy with direct contact with chickens and ducks. Swabs continued to be positive up to day 9 of sampling. The boy’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, malaise, headache, diarrhea, myalgia, and runny nose. Fever and cough persisted until day 9; all other symptoms were clear by the fourth day. No household contacts showed symptoms, virus shedding, or seroconversion.

Case 4 was in a 27-year-old woman who had direct contact with chickens and ducks. Swabs continued to be positive up to day 9 of sampling. The woman’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, malaise, headache, myalgia, and runny nose. Sore throat and cough persisted until day 12; all other symptoms were clear by the fourth day. No household contacts showed symptoms, virus shedding, or seroconversion.

During follow-up period 2, a total of 2,189 participants remained in the study, compared with 2,402 who participated in follow-up period 1. Of these, 740 (33.8%) participants were confirmed to have ILI symptoms, of whom 158 were positive for influenza A by RT-PCR (21% of those with ILI and 7.2% of the cohort overall). Four case-patients were infected with H9N2 and 1 with H5N1 virus according to RT-PCR results, which were confirmed by sequencing. The incidence of H9N2 infection was 18 cases/10,000 exposed persons. Incidence of H5N1 was 5 cases/10,000 exposed persons. Phylogenetic analysis revealed that the H9N2 viruses causing the infection were G1-like viruses (Appendix Figure 1), similar to viruses circulating in poultry in Egypt, whereas the H5N1 viruses were of clade 2.2.1.2 (Appendix Figure 2).

The H5N1 case was in a 10-year-old girl with direct contact with chickens and ducks. Swabs were positive on day 1 only. The girl’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, and myalgia. Sore throat and cough persisted until day 12; all other symptoms were clear by the fourth day. Two household contacts showed symptoms but were negative for H5, virus shedding, or seroconversion.

The first H9N2 case was in a 64-year-old woman with direct contact with chickens and ducks. Swabs were positive up to day 5 of sampling. The woman’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, diarrhea, runny nose, and myalgia, all of which persisted up to 8 days. No household contacts showed symptoms, virus shedding, or seroconversion.

The second H9N2 case was in an 8-year-old girl with direct contact with chickens and ducks. Swabs were positive up to day 5 of sampling. The girl’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, runny nose, and myalgia, all of which persisted up to 5 days except for coughing, which continued for 3 more days. No household contacts showed symptoms, virus shedding, or seroconversion.

The third H9N2 case was in a 15-year-old boy with direct contact with chickens and ducks. Swabs were positive up to day 5 of sampling. The boy’s serum sample titers were <1:10. Symptoms included fever, cough, sore throat, runny nose, and myalgia, all of which persisted up to 10 days. No household contacts showed symptoms, virus shedding, or seroconversion.

The fourth H9N2 case was in a 13-year-old girl. This case was detected 3 days after follow-up of another H3N2-positive case-patient in her household. Of the 5 members in her household, 4 had low-grade fever. H9N2 was detected in nasal and oral swab specimens from the case-patient, and the remaining household contacts were either negative or had H3N2 infection. The case-patient had low-grade fever, sore throat, runny nose, malaise, and breathing difficulty up to 6 days after detection.

We conducted a large prospective household cohort study of AIV infections among persons exposed to backyard poultry in Egypt. Our study design solved several problems that were noted by other similar studies (19,20). The larger sample size provided enough statistical power to detect the rare event of detecting active infection with AIVs. Following the households closely enabled us to detect those cases and document case-patients’ shedding, symptoms, and seroconversion. This approach also enabled us to verify whether human-to-human transmission was occurring.

Our epidemiologic findings confirm that backyard poultry raising practices have low to no biosecurity measures; these practices included keeping poultry within the household, disposing dead poultry in the open trash or water streams, and letting ill poultry loose. Growers did not frequently use personal protective equipment while exposed to poultry. Few of them reported using dedicated garments while tending to poultry, even though this measure was a main recommendation of previous educational campaigns. Similarly, only 10% of participants who reported slaughtering poultry (the riskiest behavior because of the aerosols generated [21]) used any personal protective equipment. Reviewing, revising, and updating health education and awareness campaigns based on scientific evidence might assist in decreasing the incidence of infection with AIVs in Egypt (21).

Very low seroprevalence of H5N1 antibodies was detected, whereas up to 11% of the participants had H9 antibodies, possibly because during the study period H5N1 infection was rare in poultry but H9N2 infection was common (2). Another explanation for the low level of seroprevalence noted, especially against H5N1, is that some sampling was conducted over summer months, when avian influenza activity is low. Further explanation might include waning antibody titers or lack of seroconversion.

None of the participants had H5N8 antibodies, even though these viruses are not uncommon in poultry in Egypt. In the United States, humans who were exposed to birds infected with clade 2.3.4.4 H5 viruses did not show any acute respiratory symptoms (22). Viruses of this clade might not be transmitting efficiently from birds to humans because they have limited capacity for replication and transmission in mammals and mammalian cell lines (23).

None of the exposure, health status, or demographic variables we collected were associated with being seropositive. This finding might indicate that being seropositive is related to factors not collected in this study, such as host genetic factors or other behaviors.

Our study had enough power to detect active infection with AIVs. Human infections with H5N1 and H9N2 but not H5N8 were detected. Egypt had previously reported human cases of H5N1 and H9N2 infection; human infection with H5N8 has not been reported. Incidence of infection with any avian influenza virus was 5–18 cases/10,000 poultry-exposed persons, meaning that the number of reported cases is much lower than the number of infections that actually occur. This occurrence can be explained by the fact that none of the case-patients we detected died or required hospitalization, which would make them easily missed by hospital- or clinic-based surveillance. Although the number of reported cases might be underestimated, the reported case-fatality rates are overestimated because of the mild and asymptomatic cases that were missed.

The reported symptoms were similar to symptoms of infection with seasonal influenza viruses. All case-patients had fever, cough, sore throat, and myalgia. Malaise and diarrhea occurred in a few cases. Most case-patients used over-the-counter antipyretic or anti-inflammatory drugs to treat their symptoms. Shedding duration ranged from 1 to 9 days, but no human-to-human transmission was detected, meaning that exposure to viruses that infected the poultry remains the source of human infection. Because of delays in confirming avian influenza in study participants, we were not able to obtain samples from poultry at the time when the participant was ill. Sampling poultry at the same time of sampling humans would have provided better information on how poultry infections correlate with human infections. The exact routes of transmission from poultry to humans, whether direct contact or aerosol, remain to be determined. Observing or documenting behaviors through surveys does not pinpoint transmission routes; hence, future studies should consider measuring individual exposure by using virologic methods, such as determining presence of virus on surfaces or in the air to which humans are exposed, especially in household settings, similar to what has been done in live bird market settings (24–28).

Serologic and virologic findings were not correlated, especially for H9N2 infections. In follow-up period 1, no cases of H9N2 were detected, yet 11% of the participants had antibodies. In follow-up period 2, cases were detected, but seroprevalence was almost negligible. These findings indicate that relying on serologic tests alone to estimate disease incidence might be misleading, given that many case-patients do not seroconvert, as we found in this study.

In conclusion, backyard poultry growers in Egypt continue to be infected with AIVs that are enzootic in their poultry. To eliminate human cases, poultry infections should be controlled and growers’ awareness increased to decrease the adverse effects of substandard poultry-raising practices.