Sero‐epidemiologic study of influenza A(H7N9) infection among exposed populations, China 2013‐2014

Background The first human infections of novel avian influenza A(H7N9) virus were identified in China in March 2013. Sentinel surveillance systems and contact tracing may not identify mild and asymptomatic human infections of influenza A(H7N9) virus. Objectives We assessed the seroprevalence of antibodies to influenza A(H7N9) virus in three populations during the early stages of the epidemic. Patients/Methods From March 2013 to May 2014, we collected sera from the general population, poultry workers, and contacts of confirmed infections in nine Chinese provinces reporting human A(H7N9) infections and, for contacts, second sera 2‐3 weeks later. We screened for A(H7N9) antibodies by advanced hemagglutination inhibition (HI) assay and tested sera with HI titers ≥20 by modified microneutralization (MN) assay. MN titers ≥20 or fourfold increases in paired sera were considered seropositive. Results Among general population sera (n=1480), none were seropositive. Among poultry worker sera (n=1866), 28 had HI titers ≥20; two (0.11%, 95% CI: 0.02‐0.44) were positive by MN. Among 61 healthcare and 117 non‐healthcare contacts’ sera, five had HI titers ≥20, and all were negative by MN. There was no seroconversion among 131 paired sera. Conclusions There was no evidence of widespread transmission of influenza A(H7N9) virus during March 2013 to May 2014, although A(H7N9) may have caused rare, previously unrecognized infections among poultry workers. Although the findings suggest that there were few undetected cases of influenza A(H7N9) early in the epidemic, it is important to continue monitoring transmission as virus and epidemic evolve.


| BACKGROUND
The first human infections with influenza A(H7N9) virus were iden- Most influenza A(H7N9) virus infections in humans have been associated with direct or indirect exposure to poultry, including visiting live poultry markets (LPMs) 3-5 and farms 6,7 .
The large majority of persons identified with influenza A(H7N9) virus infection presented with severe disease 8 and approximately 40% died 2 .
Persons with mild, atypical, and asymptomatic infections are far less likely to be tested for influenza A(H7N9) and will usually not seek health care. The detection of mild infections through sentinel surveillance systems 9 and through the tracing of contacts of patients with confirmed infection 10 suggests that an unknown number of mild and subclinical infections went undetected. Serologic studies are needed to identify subclinical influenza A(H7N9) human infections and better describe the full spectrum of influenza A(H7N9) human infection. Although several serologic studies of influenza A(H7N9) have been published, they have been limited to single provinces [11][12][13][14] and have used either less sensitive serologic assays 11 or lacked confirmation by microneutralization testing 11,13,14 , which improves specificity of serologic testing results 15 .
In this study, we examined the seroprevalence of antibodies to influenza A(H7N9) in three populations: the general population, poultry workers, and close contacts of persons with influenza A(H7N9) virus infection in affected provinces using specific serologic methods.

| General population in two provinces
In the early stages following the identification of influenza A(H7N9) virus, we recruited members of the general population in two provinces, Jiangxi and Henan, where provincial governments expressed support for the study. Villages or counties with at least one reported human case of influenza A(H7N9) virus infection and the immediately adjacent neighboring villages and counties were eligible for enrollment. Village and county selection was based on willingness of the local governments to participate in the study. In Jiangxi Province, the first human influenza A(H7N9) virus infection was reported on April 24, 2013. We selected two villages with influenza A(H7N9) cases and seven neighboring villages, and collected data from May 21 to June 2, 2013. In Henan Province, the first human infection with influenza A(H7N9) virus was reported in one county on April 11, 2013. We selected two affected counties and 13 neighboring counties and conducted the study from April 21 to 27, 2013.
All residents in selected affected and adjacent villages were eligible to participate in this study. Residence was defined as the place where a person had spent the majority of nights in the past 3 months.
We interviewed all residents who were at home during our visit to the selected villages. Our target sample size was 1000 participants from both provinces.

| Poultry workers in six provinces
From April 2013 to May 2014, we collected data from poultry workers in six provinces: Jiangxi, Hunan, Anhui, Henan, Jilin, and Guangdong.
Villages or counties where at least 1 month had passed since the first confirmed human case of influenza A(H7N9) virus had been reported and the adjacent neighboring villages and counties were eligible for inclusion in the study. Selection was based on willingness of the local governments to participate. In selected sites, we focused on two different types of work sites: live poultry markets (LPMs) (including wholesale markets) and farms (commercial farms or households raising small-scale poultry).
All staff who earned at least 50% of their income from poultry work at selected markets and farms were eligible for this study.
Workers from live poultry and wholesale market locations included sellers, butchers, cleaners, and transporters. Workers from farms included those from commercial farms or small-scale farmers who raised poultry in their households for commercial sale. We enrolled all poultry industry workers from selected sites who consented to participate.

| Close contacts in four provinces and one city
We selected all provinces with at least one influenza A(H7N9)-confirmed case reported before March 2014 where the local CDCs agreed to participate in the study. We collected data from contacts of the following All healthcare contacts and non-healthcare contacts of confirmed influenza A(H7N9) cases who were willing to participate were eligible.
A healthcare contact was defined as one who provided direct medical care to an influenza A(H7N9) case, either before or after confirmation of diagnosis, and who did not use standardized personal protective equipment (PPE) protection as defined by national guidelines 16 . A nonhealthcare contact was defined as a family member who lived with or cared for the patient, as well as other persons who had close contact with the patient during the illness but prior to isolation and without PPE.

| Data collection
Each study participant ≥18 years of age and guardians of participants <18 years of age completed one questionnaire, designed to collect information on demographic characteristics, health habits, and information on exposure to birds in the household, LPMs and other places (such as live poultry farms, wetland, and parks) in the preceding 3 months, respiratory symptoms in the preceding 3 weeks, and contact with persons with fever and respiratory illness in the preceding month.

| Specimen collection
We collected a blood sample from each participant using venous vacuum blood collection tubes (Becton Dickinson Medical Devices (BD) Vacutainer SST). When possible, we collected paired blood samples from close contacts. We collected the first blood sample 7 days after their exposure to the confirmed case, and the second blood sample 2-3 weeks later. The sera were temporarily stored at 4°C; after separation, sera were stored at −20°C in local public health laboratories.
After the field investigation was completed, sera were sent to the National Influenza Center of China CDC.

| Laboratory testing
We used a modified HI assay with horse red blood cells (RBCs) to screen for antibody response to influenza A(H7N9) virus according to the standard protocol endorsed by the World Health Organization (WHO) 17 . The influenza A(H7N9) antigen used in the HI assay was A/ Anhui/1/2013, propagated in specific pathogen-free (SPF) embryonated chicken eggs and inactivated with 1‰ β-propiolactone (BPL). were included in each assay. We modified the order of treatment by conducting hemadsorption before applying receptor-destroying enzyme (RDE) treatment on serum samples. Samples with an HI titer ≥20 were tested using the same virus (A/Anhui/1/2013) by neutralization antibody detection by a modified microneutralization (MN) assay for confirmation 15 . Seropositivity was defined as a MN titer ≥20 or a fourfold titer increase in paired samples 15 .

| Data analysis
Epidata3.0 was used for parallel data entry and checking. spss18.0 (SPSS Inc., Chicago, IL, USA) was used to conduct frequency analyses.

| General population
We enrolled and collected blood from a total of 1480 people from the general population; 1054 subjects were from Jiangxi Province and 426 subjects from Henan Province (Figure 1). The median age of these participants was 49 years, ranging from 6 months to 93 years; 42% (627) were male. Participants from every age group were enrolled in this study (Table 1).
Among all samples, none had an HI titer to influenza A(H7N9) ≥20, which allowed us to exclude with 95% confidence a true seropositivity in this population of more than 0.32%.

| Poultry workers
We enrolled and collected blood from 1866 poultry workers from six provinces (Figure 1). The median age of these participants was  Table 3. Neither market had been tested for influenza A(H7N9) avian influenza virus. However, the wholesale market provided poultry to another market with positive influenza A(H7N9) environmental samples 18 .  (Table 4). There was no seroconversion within the 131 paired blood samples. We could exclude a true seropositivity of more than 1.53% in this population (upper limit of the 95% CI).

| DISCUSSION
This study used hemagglutination and neutralization assays to exam-  Although we identified two seropositive cases in poultry workers, the overall seropositivity in this study was much lower than that reported in previous serologic studies conducted in a similar time period 11,13,14 . Compared with studies that did not perform MN testing, the proportion of elevated HI titers in our study (0.1% poultry workers with HI titer ≥20) was also much lower. For example, one study among poultry workers in Shenzhen found 7.2%-14.9% with HI titers ≥160 13 .
Another in Guangzhou found 1.6% poultry workers with HI titers ≥40 14 , and a study in Zhejiang Province reported 6.3% poultry workers with HI titers ≥80 7 . Although our study found a lower seropositive rate among poultry workers in Guangdong than prior studies conducted in this province [12][13][14] , our study showed a higher seroprevalence among  In summary, our study suggests that there was minimal transmis- involved in the field investigation, specimen collection and labora-