Deaths from symptomatically identifiable furious rabies in India: a nationally representative mortality survey

Abstract

Background: It is estimated that India has more deaths from rabies than any other country. However, existing estimates are indirect and rely on non-representative studies.

Methods and principal findings: We examined rabies deaths in the ongoing Million Death Study (MDS), a representative survey of over 122,000 deaths in India that uses enhanced types of verbal autopsy. We estimated the age-specific mortality rates of symptomatically identifiable furious rabies and its geographic and demographic distributions. A total of 140 deaths in our sample were caused by rabies, suggesting that in 2005 there were 12,700 (99% CI 10,000 to 15,500) symptomatically identifiable furious rabies deaths in India. Most rabies deaths were in males (62%), in rural areas (91%), and in children below the age of 15 years (50%). The overall rabies mortality rate was 1.1 deaths per 100,000 population (99%CI 0.9 to 1.4). One third of the national rabies deaths were found in Uttar Pradesh (4,300) and nearly three quarters (8,900) were in 7 central and south-eastern states: Chhattisgarh, Uttar Pradesh, Odisha, Andhra Pradesh, Bihar, Assam, and Madhya Pradesh.

Conclusions and significance: Rabies remains an avoidable cause of death in India. As verbal autopsy is not likely to identify atypical or paralytic forms of rabies, our figure of 12,700 deaths due to classic and clinically identifiable furious rabies underestimates the total number of deaths due to this virus. The concentrated geographic distribution of rabies in India suggests that a significant reduction in the number of deaths or potentially even elimination of rabies deaths is possible.

Figure 1. Regional variation of estimated rabies deaths and death rates: India, 2005.

State wise death rates are standardized to 2005 UN population estimates for India. Total estimated rabies deaths for India in the present study is 12,700, 99% CI (10,000, 15,500). Areas where no rabies deaths captured by this study represent 7% of the total India population. Figure S1 shows a comparison of state level rabies deaths reporting from present study and other mortality studies available. Abbreviations: Larger statesU: AP-Andhra Pradesh, AS-Assam, BR-Bihar, CG-Chhattisgarh, DL-Delhi, GJ-Gujarat, HR-Haryana, JK-Jammu & Kashmir, JH-Jharkhand, KA-Karnataka, KL-Kerala, MP-Madhya Pradesh, MH-Maharashtra, OR-Odisha, PB-Punjab, RJ- Rajasthan, TN-Tamil Nadu, UP-Uttar Pradesh, WB-West Bengal, Smaller statesU: AN-A & N Islands, AR-Arunachal Pradesh, CH-Chandigarh, DN-Dadra & Nagar Haveli, DD-Daman & Diu, GA-Goa, HP-Himachal Pradesh, LD-Lakshadweep, ML-Meghalaya, MN-Manipur, MZ-Mizoram, NL-Nagaland, PY-Puducherry, SK-Sikkim, TR-Tripura, UK-Uttarakhand.

Figure 2. Proportional rabies mortality reported from various sources in India 1991–2005.

We collected rabies deaths as reported from 6 different sources in India from 1991 to 2005 in order to compare our estimates with all other estimates available. (1) Medically Certified Causes of Death (MCCD) data for 1991–2004 are mostly urban deaths collected from selected urban hospitals mainly from the 10 biggest states. (2) Survey of Cause of Death (SCD) data are rural deaths between 1991–98 and were collected from about 1900 selected rural Primary Health Centers (PHC) in 23 states and 3 union territories. (3) Causes of Death Survey (Form 12) (RGI, 1998–2002) data are from the Registrar General of India. It shares the same sample framework (Sample Registration System) as our Million Death Study (MDS; 2001–2003) and includes both urban and rural deaths. (4) Sudarshan's study (January 2002 to March 2003) was a multi-centre community survey conducted by 23 university/research institutions in their territorial areas. (5) Government hospital data from the routine data collection of the Union Ministry of Health and Family Welfare . These data were significantly under reported and produced very inconsistent and sporadic results. Therefore, to calculate proportional deaths, our denominator was all causes of deaths from the states where at least one rabies death had been reported. (6) Knobel et. al. rabies deaths were calculated independently from a predictive probabilistic model based on hypothetical human-canine density, post-exposure treatment, and regional demographic features. The model does not consider any mortality statistics from India.