We reviewed available data from investigations of all of the human Nipah infections recognized in Bangladesh from 2001 through 2007. Information from the separate investigations was combined into a single database. Not all variables of interest were collected from the earliest outbreaks, but because many of the same investigators were involved across the outbreaks, data were collected in similar formats.

Persons were classified as being infected with Nipah virus if they had fever with new onset of altered mental status, seizures, or severe shortness of breath and either had specific antibodies against Nipah virus or were part of a cluster of similar case-patients in the same region, with at least 1 of the case-patients being Nipah-antibody positive. In addition, if a person had fever and immunoglobulin (Ig) M antibody to Nipah, that person was classified as being infected with Nipah virus.

We classified Nipah cases as part of a cluster if at least 1 other Nipah case was identified in the same community within 3 weeks of onset of illness. If no other cases appeared in the same community within 3 weeks, the Nipah case was classified as an isolated case.

We counted distinct introductions of Nipah virus into the human population. Each cluster of Nipah case-patients and each sporadic case was counted as a separate Nipah introduction.

We classified persons as primary case-patients if illness developed without known contact with any other Nipah case-patients, as secondary case-patients if Nipah disease developed 5–15 days after close contact with other Nipah case-patients, and as Nipah spreaders if at least 1 person with whom that person had close contact had Nipah illness develop 5–15 days after that contact. We collected geographic coordinates by using global positioning systems from the home of each case-patient.

A total of 122 Nipah case-patients were identified; 74 (61%) were male, and the mean of all case-patients age was 27 years (range 2–75). Eighty-seven (71%) of 122 died. Fifty-nine (48%) infections were serologically confirmed. Of the 63 case-patients that were not laboratory confirmed, 59 (94%) died before any blood was collected; 4 (6%) had an initial serum specimen without detectable antibodies but died before a follow-up specimen could be collected.

Figure 1

Figure 1. Locations of human Nipah virus introductions (red dots), Bangladesh, 2001–2007.

We identified 23 introductions of Nipah virus into human populations in central and northwestern Bangladesh from 2001 through 2007 (Figure 1). Ten introductions involved clusters with a median of 10 affected persons (range 2–35). Thirteen Nipah introductions resulted in isolated human infections.

Figure 2

Figure 2. Human Nipah virus infections in Bangladesh, by month of illness onset, 2001–2007.

Figure 3

Figure 3. Human Nipah virus infections in Bangladesh, by year of illness onset, 2001–2007.

The onset of illness for the index cases for the 23 human Nipah introductions occurred from December 31 through April 20 (Figure 2). The probability that all 23 index cases would randomly occur in 5 contiguous months is (5 / 12)²³ × 12 = 0.00000002. The number of Nipah case-patients varied by year; most cases were recognized in 2004, and no cases were recognized in 2002 and 2006 (Figure 3).

In 62 (51%) Nipah case-patients, illness developed from apparent person-to-person transmission. None of these cases were professional healthcare workers exposed to Nipah case-patients on the job. The case-fatality ratio was similar for primary and secondary cases (75% vs. 66%, respectively; p = 0.28).

Nine case-patients (7%) were Nipah spreaders who transmitted Nipah infection to other persons. All Nipah spreaders died; 4 of these were classified as secondary cases. Nipah case-patients who had difficulty breathing during their illness were more likely to be Nipah spreaders (12% vs. 0%, p = 0.03). Similarly, Nipah case-patients who had cough were more likely to be Nipah spreaders (12% vs. 2%, p = 0.080). Patients classified as secondary cases were just as likely to have difficulty breathing as those classified as primary cases (65% vs. 66%, p = 0.94).

Nipah case-patients who died were more likely to be Nipah spreaders than were Nipah case-patients who survived (10% vs. 0%; Fisher exact test, p = 0.057). Persons who spread Nipah were no more likely to be primary Nipah case-patients than were Nipah patients who did not spread Nipah (56% vs. 49%, odds ratio 1.3, 95% confidence interval 0.34–5.1, p = 0.74). Nipah spreaders transmitted Nipah to a mean of 7 persons (range 1–22).

Among the 10 clusters, 5 involved person-to-person transmission ranging from 2 to 5 generations of transmission. The 60 primary Nipah case-patients infected 29 subsequent persons. The estimated basic reproductive number was 0.48.

Fourteen secondary Nipah case-patients had contact with a primary Nipah case-patient for <2 days. The median incubation period of these secondary cases was 9 days (range 6–11 days).

These introductions of Nipah virus from its presumed reservoir in Pteropus bats to humans in Bangladesh were clustered, both in time of year as well as in specific years. Occurring during winter and spring in certain years, this clustering in Bangladesh suggests that the specific conditions necessary for Nipah virus transmission from bats to humans occurs only periodically. Perhaps shedding of Nipah virus by Pteropus bats is seasonal, and because of population dynamics and the accumulation of susceptible juvenile bats over time, transmission is quite low in some years compared with other years when widespread shedding and transmission occur.

The presumed wildlife reservoir of Nipah virus, bats of the genus Pteropus, is widely distributed across Bangladesh, the rest of the Indian subcontinent, and Southeast Asia (18). Wherever Pteropus bats have been collected and tested, they commonly have had antibodies against Nipah or a related virus (8,9,13,23,24). Human Nipah infections in Bangladesh, however, have been recognized in a confined geographic area. The Institute of Epidemiology Disease Control and Research in Bangladesh has national surveillance for disease outbreaks. Outbreaks involving the deaths of several previously healthy persons, which is characteristic of Nipah outbreaks, typically generate substantial local concern, media attention, and notification of local and central health authorities. Despite investigations of outbreaks in all regions of the country, all confirmed Nipah outbreaks have occurred in the same central and northwestern regions (Figure 1). Notably, the only 2 outbreaks that have been reported from India have been in regions within 50 kilometers of the border with Bangladesh and immediately contiguous with the affected areas in Bangladesh (25,26). One hypothesis that would explain this geographic concentration of human cases is that the bats are attracted to specific foods available in these areas during the winter and spring; people living in these communities are occasionally exposed to foods contaminated with bat urine or saliva that contains Nipah virus. One outbreak of Nipah virus was associated with consumption of raw date palm sap, which is harvested from December through March in this region (14). In 21 of the 23 recognized index case-patients, Nipah virus illness developed during this harvest season.

Person-to-person transmission is a major pathway for human Nipah virus infection in Bangladesh, accounting for 51% of recognized cases in this review. By contrast, in Malaysia and Singapore, person-to-person transmission was not confirmed, although 1 nurse who cared for Nipah patients in an intensive care unit in Malaysia and reported no clinical illness had serologic evidence of Nipah virus infection and brain magnetic resonance imaging consistent with Nipah virus infection (27). Even if occasional unrecognized person-to-person transmission of Nipah virus occurred in the outbreak in Malaysia, person-to-person transmission is much more apparent and common in Bangladesh. Moreover, the outbreak in Siliguri, India, in 2001 was also characterized by widespread person-to-person transmission (25).

Three factors likely contributed to the higher frequency of person-to-person transmission of Nipah virus in Bangladesh than was observed in Malaysia. First, respiratory disease associated with Nipah infection was more common and more severe in Bangladesh compared with that in Malaysia and Singapore (28–30). Earlier studies demonstrated that Nipah virus was present in respiratory secretions of some Nipah virus–infected patients (11,31). We found that Nipah case-patients who had difficulty breathing were much more likely to be Nipah spreaders. Together, these findings suggest that when a Nipah virus–infected patient has a symptomatic respiratory tract infection associated with Nipah virus, the patient can shed an infectious inoculum of Nipah virus in his respiratory secretions. In the largest recognized Nipah outbreak in Bangladesh, touching a patient who had respiratory difficulties was a risk factor for developing Nipah infection (16). The personal care typically provided to ill and dying relatives in Bangladesh is likely a second important contributor to person-to-person transmission. This care is characterized by close physical interaction, frequent contact with a patient’s saliva, and a lack of basic infection control practices because the paradigm of communicable disease is inconsistent with the prevailing cultural interpretation of illness (32). Third, all Nipah virus strains from human cases in Malaysia were genetically similar, in contrast to the marked diversity of the strains in Bangladesh (11). Some strains possibly possess genetic characteristics that facilitate person-to-person transmission.

Our 0.48 estimate of the basic reproductive number of Nipah virus in rural Bangladesh, a resource-poor setting with extremely limited infection control practices, suggests that Nipah virus is unlikely to cause a sustained human pandemic from person-to-person transmission. This conclusion is further supported by our observation of 23 separate introductions of the virus into the human population; none of these introductions resulted in sustained person-to-person transmission. Indeed, only 1 of the introductions exceeded 2 generations of transmission (16). However, we could study only 9 patients who transmitted the Nipah virus. Thus, our understanding of the variability in transmission of different strains of the virus in different contexts is limited.

This analysis has limitations. First, we almost certainly did not identify all human Nipah virus infections that occurred in Bangladesh from 2001 through 2007. Many persons in Bangladesh, especially the impoverished, do not access the formal healthcare system, even when seriously ill (33,34). The available data on human Nipah virus infection in Bangladesh are biased toward infections acquired in outbreaks recognized by public health authorities. Because meningoencephalitis is a common cause of hospitalization in Bangladesh, sporadic cases that are unrecognized as Nipah virus infection may be the more common presentation of Nipah virus infection in the country. Indeed, during 2004 and 2007, years when multiple outbreaks were identified, many patients who sought treatment for symptoms of encephalitis in hospitals located near identified outbreak areas were tested for Nipah virus. This testing identified several additional patients infected with the virus who lived quite a distance from the outbreak villages and who had no apparent connection to other cases. Thus, surveillance almost certainly underestimates the public health impact of Nipah virus infection in Bangladesh and may underestimate its geographic and seasonal range. Because outbreaks are more likely to lead to recognition of Nipah virus infection than sporadic cases and person-to-person transmission can occur only in clusters, the overall proportion of Nipah virus infections in Bangledesh that are transmitted person to person is probably <50%.

A second limitation is that the calculation of the basic reproductive rate assumed that all persons infected by a primary case-patient were identified. The investigation team could have failed to recognize all cases, especially in persons with milder or asymptomatic infection. In Malaysia, among 178 persons without symptoms who lived on farms where at least 1 person with confirmed Nipah encephalitis was identified, 20 (11%) had antibodies against Nipah virus (3). This possibility of asymptomatic infection suggests that our estimate of the basic reproductive rate is a minimal estimate. However, in outbreaks, when mild or asymptomatic persons were screened in Bangladesh, few additional Nipah cases were identified (J. Hossain, pers. comm.). Moreover, only patients who died transmitted the infection, so the possible infection of persons in whom mild illness developed is unlikely to contribute to the risk for pandemic transmission.

A third limitation is that we identified only 9 persons who transmitted Nipah virus and so have limited statistical power to assess their characteristics. Indeed, the association of cough with Nipah virus transmission and death with Nipah transmission are above the traditional guideline for statistical significance. However, the weight of the evidence, including an association with respiratory illness that meets the traditional criterion for statistical significance and the isolation of Nipah virus in respiratory secretions, suggests that person-to-person transmission occurs occasionally from virus-infected patients who are efficient respiratory transmitters of the virus.

A fourth limitation is that this analysis assumes that persons in whom Nipah illness developed 5–15 days after contact with a Nipah patient were considered infected by the contact. If the subsequent case-patient had a similar environmental exposure to Nipah virus as the initial case-patient, this approach may overestimate the proportion of cases that result from person-to-person transmission. However, Nipah virus is readily recovered from the saliva of infected persons (11,31), and epidemiologic studies of individual outbreaks implicate contact with Nipah-infected patients as a risk factor for transmission (16). Moreover, the observed 6–11-day window between exposure and disease is consistent with incubation periods for other human viral infections. Thus, person-to-person transmission is the most likely route of transmission in these scenarios.

The recurrent introduction of Nipah virus from its fruit bat reservoir to humans in Bangladesh causes outbreaks with high fatality rates and substantial neurologic sequelae among survivors (35). Fruit bats have a critical niche in the ecosystem, contributing to both plant pollination and distribution of seeds (36,37). Preventing human Nipah virus infections in Bangladesh is difficult because these infections are relatively rare compared with many other serious and more common health threats faced by the large population of this low-income country. Improved understanding of the mechanism of zoonotic transmission from bats to humans may help identify feasible approaches to prevent future introductions of the virus into the human population. Interventions to reduce the risk for pathogen exposure by minimizing saliva contact or improving hand hygiene among persons who care for seriously ill patients in severely resource-constrained settings could reduce person-to-person transmission of Nipah as well as transmission of other dangerous pathogens. In confirmed or highly suspected cases of Nipah virus infection, respiratory droplet precautions may also be warranted.