We used a line list of high-risk exposure contacts of the initial 18 laboratory-confirmed NiV patients, including 235 HCWs and 191 household and community contacts. We needed to survey 300 contacts (150 HCW and 150 household and community contacts) because our assumption was that 5% of contacts would have asymptomatic infection de-velop, and absolute precision of the estimate would be 2.5% for a 95% confidence level. The Institutional Human Ethics Committee of the ICMR–National Institute of Epidemiology, Chennai (approval no. NIE/IHEC/201806-01) and the Government Medical College, Kozhikode (approval no. GMCKKD/RP2018/IEC/97) approved the study protocol.

We approached the contacts at their residences or workplaces and interviewed them using a structured questionnaire. We collected sociodemographic information, data on the type and frequency of contact with >1 NiV patient, and history of febrile illness after contact with NiV patients. For each person who consented to participate, we collected a 3-mL blood sample, separated serum, and transported samples to the National Institute of Virology (Pune, India), where they were tested for human IgM and IgG against NiV.

We collected 279 blood samples from 155 HCWs and 124 household and community contacts. The median age for HCWs was 37 years (interquartile range 29–48 years) and for household and community contacts was 39 years (interquartile range 30–51 years). Thirty-two HCWs and 36 household contacts reported exposure to body fluids of NiV patients; 123 (79.4%) HCWs and 88 (71.0%) household contacts reported physical contact with ≥1 NiV patient (Table 1).

We performed ELISA on samples with reagents provided by the US Centers for Disease Control and Prevention (Atlanta, GA, USA) and tested serum at 4 dilutions: 1:100, 1:400, 1:1,600, and 1:6,400. For IgM assays, we considered samples positive when the sum of the optical density for all 4 dilutions was >0.45 (10). For IgG assays, we considered samples positive when the sum of the optical density for all dilutions was >0.95.

Of the 279 serum samples tested, 2 had IgM and IgG and 1 had only IgM against NiV. We calculated the overall seroprevalence of NiV as 1.08% (95% CI 0.37–3.11). None of the seropositive persons reported having a febrile illness after their last contact with an NiV patient, indicating subclinical infections. Two seropositive persons were family members of a laboratory-confirmed patient, and the third was a HCW in the emergency medicine department. All 3 had a history of exposure to body fluids of >1 NiV patient (Table 2).

The risk for subclinical infection was higher among the contacts who had exposure to body fluids (3/68, 4.4% [95% CI 1.5%–12.2%]) than for those who only had physical contact with ≥1 NiV patient (0/211, 0% [95% CI 0%–1.8%]; p = 0.007). The epidemiologic association between exposure and seropositivity suggests our results are accurate. Applying the proportion of asymptomatic infection found in our sample of 279 to all 426 persons exposed to laboratory-confirmed NiV infection yields an expected total of 23 NiV infections among contacts, including 5 (21.7%) asymptomatic cases.

Although NiV is known to cause subclinical infections, the extent of these infections among close contacts varies during outbreaks. For instance, no subclinical infections have been reported from outbreaks in Bangladesh (11), but 1%–15% of infections were subclinical during outbreaks in Malaysia (12–15). Parashar et al. reported clinically undetected NiV infection in 6% of 166 community-farm controls and in 11% of 178 case-farm controls (12). Another study of household contacts of hospitalized NiV patients indicated that 8% had subclinical infections (13). In an outbreak in Singapore, infections were reported in 2 (4.6%) of 43 asymptomatic abattoir workers (14). Another study conducted in Singapore among 1,460 HCWs having contact with NiV patients identified antibodies specific for NiV in 22 (1.5%), 10 of whom were asymptomatic (15). These studies suggest that infection with the Malaysian strain of NiV causes less severe illness, a lower case-fatality rate, and higher prevalence of asymptomatic infections compared with outbreaks involving the Bangladesh strain. Studies in African green monkeys also suggest the Bangladesh strain of NiV is more pathogenic than the Malaysian strain (1). The NiV strain responsible for the Kerala outbreak was closer to the Bangladesh strain and was more pathogenic (8). Although previous studies did not show any subclinical infections during NiV outbreaks with the Bangladesh strain, our study suggested that NiV strain of Kerala outbreak generated asymptomatic infections. Our study also found that IgM could be detected ≤2 months after NiV infection and the immunoglobulin class switch to IgG could occur beyond 2 months.

Our study had 1 limitation. Although we approached all line listed contacts, we collected samples from only 124 of 191 household and community members. The remaining contacts were either unavailable (17%) or declined to give a blood sample (18%). However, this limitation is unlikely to affect overall seroprevalence because nonparticipation in the survey was not based on exposure status.

Our findings indicate that subclinical infections occurred among close contacts of patients during an NiV outbreak in Kerala, India, but were infrequent. In addition, we found the risk for subclinical infections was higher among persons with a history of exposure to body fluids of NiV patients than for those with only physical contact.