Increased Mortality Rate Associated with Chikungunya Epidemic, Ahmedabad, India

A total of 3,056 excess deaths epidemiologically linked to chikungunya occurred in 2006.

C hikungunya virus, an alphavirus of the family Togaviridae, is native to tropical Africa and Asia. This virus is transmitted to humans by mosquitoes. Aedes aegypti and Ae. albopictus are the 2 main vectors that transmit this disease (1). The fi rst reported chikungunya outbreak occurred in Tanganyika (now Tanzania) in 1952-1953 (2). The word chikungunya is derived from the Makonde language in southeastern Tanzania and means "bent down or become contorted," which indicates the classic posture the patient adopts because of severe joint pain. Symptoms of chikungunya include sudden onset of fever, severe arthralgia, and maculopapular rash. A specifi c symptom is severe incapacitating arthralgia, often persistent, which can result in long-lasting disability (3).
A major epidemic of this disease was reported in [2005][2006] in Réunion Island; ≈266,000 residents (34.3% of the population) of this Indian Ocean island were affected by chikungunya fever as of February 19, 2007. This epidemic also spread to France through imported cases from Réunion Island (4). Historically, chikungunya was considered selflimiting and nonfatal. However, 254 deaths on Réunion (case-fatality rate 1/1,000) that were attributed directly or indirectly to chikungunya during the epidemic changed this perspective (1,4).
India reported a massive chikungunya epidemic in 2006. Chikungunya has reemerged in India since 1973, when the attack rate was 37.5%. However, in the 2006 epidemic, the attack rate increased to 45% in some places (4). More than 1.39 million cases across 151 districts and 10 states were reported during this period (5). However, unlike the epidemic on Réunion Island, no deaths directly attributable to this disease were reported (6). The dominant vectors are Ae. albopictus on Réunion Island and Ae. aegypti in India (4). However, Ae. albopictus was also implicated in Kerala State, India (7).
Studies have indicated that the recent outbreak in the Indian Ocean islands was initiated by a strain related to East African isolates, from which viral variants have evolved with a traceable history of microevolution. This history could provide information for understanding the unusual magnitude and virulence of this chikungunya epidemic (8).
The purpose of this study was to analyze the association between the chikungunya epidemic in India and the mortality rate in the city of Ahmedabad. Such fi ndings could show correlations between reported genomic mutations in chikungunya virus and its increased virulence. Such information is valuable for public health systems in Increased Mortality Rate Associated with Chikungunya Epidemic, Ahmedabad, India Dileep Mavalankar,* Priya Shastri,* Tathagata Bandyopadhyay,* Jeram Parmar,* and Karaikurichi V. Ramani* developing countries that frequently underreport or misreport epidemics.

Collection of Death Data
The registrar of births and deaths (RBD) of Ahmedabad, who is a subordinate offi cer to the medical offi cer of health, registers all births and deaths within the city limits under the Registration of Births and Deaths Act. Deaths are registered in 2 ways. Deaths that occur in a hospital are reported by hospital authorities, who provide a medical certifi cate of death that is sent to the RBD offi cer in the city ward in which the hospital is located. Deaths that occur at home are reported by the family to the local RBD offi cer of the ward in which their home is located.
Deaths are compiled and sent from the RBD ward offi ce to the RBD central offi ce and subsequently communicated to the state level registrar of birth and death. Death data used in this study were provided by the medical offi cer of health of the city. Data include monthly total deaths registered in Ahmedabad during 2002-2006.

Collection of Chikungunya Case Data
During the chikungunya epidemic, the city health department collected, compiled, and reported data on suspected cases of chikungunya from municipal hospitals and health centers. Data include monthly reported cases of chikungunya, blood samples sent for testing, and samples positive for chikungunya virus infection in Ahmedabad starting in April 2006. Few data were reported by private hospitals, dispensaries, and private practitioners in the city, who treat many patients.

Statistical Analysis
Average mortality rate for each month during 2002-2005 (years before the epidemic) was calculated by dividing the average number of deaths for each month by the average population. Average mortality rate for each month in 2006 was calculated by dividing the number of registered deaths for each month by the monthly population. The expected number of monthly deaths for each month in 2006 was calculated by multiplying the average mortality rate for each month (2002)(2003)(2004)(2005) by the monthly population in 2006. Because there were 12 estimates of expected deaths (1 for each month), we used the more conservative simultaneous confi dence interval (CI) and the Bonferroni method (9) instead of a simple CI for each month separately. Excess deaths for each month in 2006 were the difference between actual observed number of deaths and expected number of deaths. Average monthly mortality rates for 2002-2005 were then compared with the mortality rate for 2006 (epidemic year).

Results
The medical offi cer of health in Ahmabadad reported 60,777 suspected chikungunya cases in 2006. The peak of the epidemic occurred in August and September 2006 when 55,593 (91.5%) of the cases were reported. A total of 84 (54.5%) of 154 blood samples tested were positive for chikungunya virus. Of these 84 confi rmed chikungunya cases, 10 were fatal (case-fatality rate 11.9%).
The temporal relationship between chikungunya cases and expected mortality rates and actual mortality rates in 2006 is shown in the Figure. The peak in chikungunya cases in August-September coincides with the peak in actual deaths in 2006.

Discussion
Analysis of our data shows that the mortality rate in Ahmedabad increased substantially in 2006 when compared with rates for the previous 4 years. A total of 3,056 excess deaths occurred in 2006 (the epidemic year) when compared with the expected number of deaths for that year. A substantial increase in deaths reported was observed from August through November 2006 (2,944 excess deaths in these months). The number of reported chikungunya cases also showed a peak in August and September 2006, which coincided temporally with the peak in number of deaths in Ahmedabad (Figure).

RESEARCH
The main issues of contention are whether these excess deaths were caused by chikungunya and whether such excess deaths will occur in future years without chikungunya epidemics. No major adverse event or other epidemic occurred in Ahmebabad in August-November 2006 other than the chikungunya epidemic. Our epidemiologic evidence shows that the epidemic is the most plausible explanation for the large increase in deaths in Ahmedabad in August-November 2006. However, other unidentifi ed causes cannot be ruled out. Similar data from other cities and areas affected by the chikungunya epidemic may help establish the link between chikungunya and excess deaths.
There are 2 major problems with reporting of deaths in Ahmedabad. The cause of death is poorly reported, and the RBD does not separate death data for residents and nonresidents. Inclusion of patients from surrounding rural areas who died in city hospitals could have resulted in excess deaths being reported during the epidemic. However, this was a problem in years before the epidemic (2002)(2003)(2004)(2005) as well. A review of deaths registered in rural areas outside the city limits of Ahmedabad showed no major decrease during the epidemic months of 2006 over previous years. Thus, the increase in number of deaths caused by migration of sick patients cannot explain this major increase in deaths in 2006, although this factor may have contributed to it.
An excess in total deaths was also reported for the chikungunya epidemic on Réunion Island during February-April 2006 (10). A total of 260 excess deaths were reported on Réunion Island during the epidemic, of which 254 were directly or indirectly attributed to chikungunya (mortality rate attributed to chikungunya 1/1,000) (4,10). Most of the excess deaths on Réunion Island were persons >75 years of age. Of 10 confi rmed deaths in Ahmedabad caused by chikungunya, 2 were persons >80 years of age, 4 were persons 60-70 years of age, and 3 were persons <60 years of age.
The genomic sequences of chikungunya virus isolates from India were similar to that of a recent isolate from Réunion Island (11). Because of this fi nding, the mortality rate on Réunion Island can be applied to the epidemic in India to estimate the probable number of deaths that may have occurred. With limited case data reported from India and a mortality rate on Réunion Island of 1 per 1,000 cases, it was previously estimated that India would have ≈1,200-19,000 deaths caused by the chikungunya epidemic (12). The excess number of deaths observed during the epidemic in Ahmedabad suggests that estimates of deaths caused by chikungunya in India need to be revised.
Despite the increase in deaths in Ahmedabad and reports of suspected deaths caused by chikungunya in Kerala State, India (13), no systematic and comprehensive investigation of deaths in relation to this epidemic has been conducted by government authorities at the national or state level in India. The government of India has declared repeatedly in the parliament that "there are no deaths directly attributable to Chikungunya" in India (6). Although 10 deaths caused by chikungunya were reported by the medical offi cer of health in Ahmebadad, the website of the government of India continues to report "zero deaths" (5). Further investigations on the cause of excess deaths are urgently needed to conclusively establish that chikungunya was the cause of excess deaths in Ahmedabad. Given the possible association of deaths with the chikungunya fever epidemic in Ahmedabad, public health authorities should investigate such epidemics in other countries. These investigations will help determine whether the virus has increased in virulence, which may also pose a greater threat outside the Indian Ocean region. Such studies would help detect and control similar epidemics and help governments to provide adequate warnings to travelers to chikungunya-endemic countries.
We report an increase in mortality rates in Ahmedabad during August-November 2006 (when a chikungunya epidemic occurred in this city) compared with previous months in 2006 and the same months in the past 4 years. The highest number of chikungunya cases was also reported in  August and September. The city had ≈2,944 additional deaths during August-November 2006. Epidemiologic evidence shows that the increase in deaths in Ahmedabad was largely attributable to the chikungunya epidemic. Given poor reporting of deaths, an unexplained cause of death cannot be ruled out. Mortality rate data for Ahmedabad are consistent with observations of other researchers that the virus may have mutated and become more dangerous than reported (8). Public health authorities must investigate recent epidemics. Otherwise, developing countries may not be able to detect and combat severe future epidemics of other reemerging diseases such as avian infl uenza and severe acute respiratory syndrome. If our fi ndings are validated by studies in other regions of India and elsewhere, it would assist the international health community to be better prepared in dealing with future epidemics of emerging infectious diseases and reduce associated deaths.