Seroprevalence and Transmission of Human Influenza A(H5N1) Virus before and after Virus Reassortment, Cambodia, 2006–2014

Thirty-five human influenza A(H5N1) cases were reported in Cambodia during 2013–2014 after emergence of a clade 1.1.2 reassortant virus. We tested 881 villagers and found 2 cases of pauci- or asymptomatic infection. Seroprevalence after emergence of the reassortant strain (0.2%) was lower than the aggregate seroprevalence of 1.3% reported in earlier studies.


The Study
In the first week of February 2014, a suspected case and 2 laboratory-confirmed cases of the new influenza A(H5N1) We selected these 2 villages to conduct seroprevalence studies within a month of case occurrence to determine point-seroprevalence in the general population ( Figure). Epidemiology teams from IPC and the Cambodia Ministry of Health interviewed and obtained serum samples from all village residents who provided informed consent. Sampling was repeated >2 weeks later to test for a possible increase in H5N1 virus antibody levels. We obtained National Ethics Committee for Human Research approval for all serostudies conducted as part of pandemic risk assessments.
We used hemagglutination inhibition (HI) and microneutralization assays as described previously (2) to test paired serum samples for H5N1 virus antibodies. Patients whose first serum sample was antibody-negative and whose second blood sample showed seroconversion (HI titer >80 and a microneutralization titer >40) were considered H5N1 virus positive (3). Patients who had detectable antibodies in the first serum sample and a >4-fold rise in antibody titer (minimum HI titer of 40 and microneutralization titer of 20) for the second sample were also considered H5N1 positive (3).
We used EpiData (EpiData Association, Odense, Denmark) to double-enter questionnaires. We excluded index cases from the analysis to assess risk of infection due to coexposure or secondary transmission and calculated prevalence by using Poisson confidence intervals. We used the Fisher exact test to compare our data to those from community seroprevalence studies conducted before emergence of this H5N1 reassortant virus.
Paired samples were obtained from 238 (34.2%) of the 695 children and adults in village 1 and from 643 (69.8%) of 921 persons in village 2 ( Figure). All persons in direct contact with the index case-patient in village 1 were screened, but 2 healthy family contacts of the second case-patient in village 1 refused participation. By laboratory testing, we found 1 additional case of pauci-or asymptomatic human  (4). We therefore compared these 2014 results to the 10-year findings of 7 community seroprevalence studies (Table). These comparisons were undertaken by the same teams of IPC epidemiologists and virologists around confirmed cases in H5N1 epidemic foci in Cambodia with the cooperation and support of the Cambodian Ministry of Health.

Seroprevalence and Transmission of Human Influenza A(H5N1) Virus before and after
Studies we conducted prior to the emergence of the reassortant strain, including 2 unpublished serosurveys, tested 2,713 persons, of whom 35 (1.3%) were positive for H5N1 antibodies (Table). Our surveys of the population surrounding cases of infection by the reassortant H5N1 virus in 2014 shows a seroprevalence of 0.2%, lower than the aggregate seroprevalence of 1.3% for 2005-2010. This difference is statistically significant (p = 0.004).
Even if the 2 healthy but untested contacts in village 1 had been found positive, the seroprevalence of H5N1 virus in humans would still be significantly lower following the emergence of the clade 1.1.2 reassortant virus when compared with historical IPC data (2,5,6). Despite the 2013-2014 surge in cases, there is no evidence of increased H5N1 virus transmission to humans or altered clinical and epidemiologic characteristics associated with the emergence of this reassortant virus. A review of literature showed that with the exception of 3 studies in China (7-9), all community seroprevalence studies conducted around H5N1 cases outside Cambodia found 0% prevalence (references not listed).

Conclusions
Routine biocontrol measures in Cambodia are poor, and poultry surveillance is lacking. Farmers either delay reporting and detection, or sell sick poultry, due in part, to lack of compensation from the government. This enables poultry outbreaks to spread. These factors could explain why prevalence in populations around confirmed cases was 10-fold higher in Cambodia than elsewhere until 2012.
The drop in the number of positive cases in the 2014 community serosurveys raises the possibility that villagers' knowledge of avian influenza and prevention may have improved after 2012, probably as a result of Cambodian authorities' commitment to and international funding for community education campaigns. Recent studies, however, have shown a >2.5-fold increase in the circulation of avian H5N1 virus in live bird markets in Cambodia during 2013 compared with 2011 (10,11). The peak in confirmed human infections with H5N1 reassortant virus during 2013-2014 may therefore be linked to this increased circulation of H5N1 reassortant virus among poultry, but it may also reflect improved surveillance in the public health sector because our 2014 seroprevalence study does not suggest increased transmissibility of the virus to humans.
Our study has limitations. Although there were 9 human cases of infection with H5N1 virus reported in 2014, all followed by direct-contact tracing, only 2 spatiotemporal seroprevalence studies were conducted around H5N1 clusters and foci. There was also some minor loss to follow-up. Transmission risk estimates in villages with clusters might be influenced by the role of host genetics in susceptibility to infection (12,13). Our findings may therefore apply to genetically vulnerable persons and may have been overestimated, but they remain comparable to or lower than historical data in Cambodia. It is difficult to compare seroprevalence results between studies due to differences in methods and the lack of consensus on neutralizing antibody levels associated with pauci-and asymptomatic influenza H5N1 infections.
In conclusion, our findings provide no evidence for increased human-to-human H5N1 virus transmission in Cambodia after emergence of the clade 1.1.2 reassortant virus. Direct-contact tracing is essential, but regular community-wide seroprevalence studies may not be required. However, each new epidemic phase or emergence of a new reassortant virus must be carefully investigated to address specific public health and research questions that can be answered only in an outbreak setting.