A Bayesian Approach to Compare the Statewise Dengue
Death Counts in India

Atanu Bhattacharjee; Dibyojyoti Bhattacharjee

A Bayesian Approach to Compare the Statewise Dengue Death Counts in India

Atanu Bhattacharjee , Dibyojyoti Bhattacharjee

Atanu Bhattacharjee^1* and Dibyojyoti Bhattacharjee²

Department of Statistics, Gauhati University, Guwahati 781014, India
Department of Business Administration, Assam University, Silchar 788011, India

Related article at Pubmed, Scholar Google

Visit for more related articles at International Journal of Collaborative Research on Internal Medicine & Public Health

Abstract

Background: The Estimation of the true disease burden of dengue is a challenge, considering its varied epidemiology and dynamics of transmission. The true disease burden cannot be understood from ordinary count data on the number of deaths or number of infected cases.

Aim & Objectives: To compare the Indian states by the true disease burden due to dengue in the year 2010 by ranking with respect to their performance to deal with dengue death.

Methods: To compare the states by count data for observed deaths, the Bayesian approach is used. The source of data is the National Dengue Control Report published by National Vector Borne Disease Control Programme of 2009 and 2010 till the month of August. The statewise comparison of death due to dengue has been performed with all necessary computation done in the statistical software R.

Results: Certain states like Delhi, Gujarat and Andhra Pradesh had very high incidence of dengue cases. But the mortality per unit exposure due to dengue is less in Delhi in contrast to other state in the country.

Conclusion: This statewise dengue death comparison can be helpful to authorize the state specific dengue control programme, as several issues which are not clear from count data can surface from such analysis.

Keywords

Bayesian estimation, Dengue, Public health, Shrinkage Estimator

INTRODUCTION

Dengue Virus (DV) is an enveloped, singlestranded, positive RNA virus and a member of the family Flaviviridae, genus flavivirus. There are four antigenically related but distinct serologic subtypes; DV-1, DV-2, DV- 3 and DV-4[1]. Generally, infection with one serotype confers future protective immunity against that particular serotype but not against others [2] . Dengue infection (DI) is amongst the most emerging viral diseases transmitted by mosquitoes to humans, in terms of both illness and death [3]. Past few decades has turned this disease into a serious public health problem, especially in the tropical and subtropical countries [4]. An estimated 50-100 million cases of Dengue Fever (DF) and about 250,000–500,000 cases of Dengue Hemorrhagic Fever (DHF) occur every year [5].

As per the report of World Health organization [6], it is estimated that 50 million dengue infections occur every year with 500000 requiring hospitalization; 2.5% of those affected may die. These figures are likely to be underestimated as the numbers reported are influenced by different surveillance and reporting systems as well as varying interpretations of case definitions and the presence of missed and silent infections [7].

Dengue in India

Dengue virus was first isolated in India in 1945 [8]. All four virus types circulate and cause epidemics, but only occasional cases of dengue hemorrhagic fever/dengue shock syndrome (DHF/DSS) have been reported in India[9]. Delhi had its largest outbreak of DHF/DSS from August through November in 1996. During that year a total of 8,900 cases were reported, with a death rate of 4.2% [10]. As dengue is a mosquito borne viral fever with potential to turn out to be fatal so the disease has emerged to be a threat to the public health system in the country.

The status of public health across the country is being put to strain because of annual dengue infection and several other diseases those results from poor living conditions. Dengue is sweeping across Delhi, Haryana, Uttar Pradesh and Rajasthan and chikungunya through Karnataka, Maharashtra, Andhra Pradesh, Kerala, Madhya Pradesh and Gujarat and an outbreak of malaria in Mumbai in 2006. Poor sanitary conditions, water logging, accumulation of stagnant water bodies during the monsoons turn out to be perfect breeding grounds for mosquitoes. Mosquitoes breed in stagnant water pools and rotting garbage – unfortunately an all-too-familiar situation during and after the monsoons in towns and cities where civic services are over burdened. Several studies have shown that the out break/ incidence of dengue cases increases in India in the month of October i.e. at the fag end of the monsoon season [11, 12]. In the year 2003, India had experienced one of the wettest monsoons in 25 years, which led to a spate of mosquito growth creating an alarming situation of mosquito borne diseases in many states[12]. Both the public health authorities and the public themselves, are not aware about the special cleanliness drives that the surroundings are to be put to during or after the monsoon in order to restrict the breeding of mosquitoes. As per the estimates of the Ministry of Health and Family Welfare, in 2008, India recorded 12419 cases of Dengue, up from 5534 case in 2007. Delhi alone recorded 1307 cases of Dengue in 2008, up from 548 in 2007. Four Indian states, Punjab, Haryana, West Bengal, and Gujarat, have been the worst hit by Dengue, with Punjab reporting 4349 cases, and Haryana, West Bengal and Gujarat reporting 1137, 1050 and 1023 cases respectively.

Fighting Dengue Out: Some Indian Issues

Considering the huge population of India, the proportion of people infected by dengue would not be significant and the fatalities due to dengue negligible. But a relatively small number of fatalities and low incidence of dengue infection should not be a cause for contentment. Such viruses and diseases should not surface at all and even if they do, public health authorities should be in a position to avert the loss of lives. To do that it is important to understand the burden of dengue to which the different states of India are subjected. However, estimating the true disease burden of dengue in the country like India is a challenge task considering its varied epidemiology and dynamics of transmission. India has a highly complex and colorful social mosaic. The topographical difference, hot and humid living conditions at places, difference in the share of average annual rainfall, disparity in the share of resources, unplanned urbanization and several other factors may be held responsible for the unsuccessful attempts in the uprooting diseases like malaria, dengue etc. from the country. It is necessary that the issues of public health be studied at the different regional level, going down to each individual states and if possible to the level of districts, at which the implementation of several government policies including those related to public health initiates. However, to understand the extent and pace to which the different public health policies are to be implemented at the different states it is necessary to estimate the actual burden of the corresponding health problem, with dengue being no exception.

Considering the continual return of dengue in the country and the need of estimating the true burden of disease at the different states of the country provides the backdrop of such a study. This works intent to quantify, the burden of dengue, in the different states of the country. The study can help the corresponding public health authorities of the states to adjust their programs, leading to eradication of dengue from India.

Review of Literature

The dengue disease and viruses is a great problem in the Indian subcontinent for at least the past 50 years [13, 14, 15]. Since last twenty years the epidemiology of dengue has dramatically changed. It has been reported many times in Indian subcontinent [16, 17, 18]. Thus, other sources of information are needed to help determine the probable underlying causes and to detect proper dengue cases. Delhi, a city in North India, has experienced seven outbreaks of dengue virus infection since 1967 with the last reported in 2003 [19, 20, 21].

Ekta et al [23] have compared the serological and virological profiles of the confirmed dengue cases reported to All India Institute of Medical Sciences (AIIMS) in these three years i.e.2003, 2004, and 2005. However, Jha et al.[11,21] have concluded that the major deaths in rural India take place at home, without prior attention by any qualified healthcare worker, so most causes are not medically certified.

Material and Methods

The data required for the study is based on the National Dengue Control Report published by National Vector Borne Disease Control Programme ( NVBDCP) of 2009 and 2010 till the month of August. The report provides state wise total blood slides examinations, dengue cases and death in each year. The state wise comparison of death due to dengue is performed by comparing the mortality rate per unit exposure rate. The estimate of which for each of the states is obtained through a Bayesian analysis. It is practical to assume the true rates are similar in size that generates the dependency between the parameter. To deal with such problem it is good practice to call a hyper-parameter to reduce the dependency between parameters. The whole process produces the system of a hierarchical prior guiding us to use hierarchical Bayesian approach. All the relevant calculations are performed in the statistical software R¹.

The main goal of this work is the comparison by the mortality rates due to dengue for 23 Indian states. Each state has a true mortality rate λi, and the objective is to compare estimate of 23 rates λ1, ..., λ23. It is logical to assume a priori that the true rates are similar in size. However, it can be affected due the dependence between parameters. The particular information about one state’s true mortality rate can influences the idea about other states. Suppose we are interested in simultaneously estimating the true mortality rates {λi} for all states. The simple procedure is the compare the individual mortality rates by

But this pooled estimate is based on the assumption that the true mortality rate is the same across the states. This estimate shrinks or moves the individual estimate yi/ei toward the pooled estimate Σyj/Σej where the parameter 0 < λ < 1 determines the size of the shrinkage. The shrinkage estimate is a natural byproduct of the application of an exchangeable prior model on the true mortality rates [26].

The total number of exposed person due to dengue has been denoted by e. The estimate of mortality rate per unit of exposure rate assumed by λ. It has been assumed that the death count Y follow Poisson distribution with mean eλ. The standard estimate of λ is, λˆ = y / e .

The comparison due to death rate among the states in 2010, has been performed by the prior information of mortality rate generated by reports of 2009 death counts. In the annual state report many state’s death count is zero or nearer to zero and many states are having higher number of death counts. The prior information about death has been obtained from 10 randomly selected states as a representative of dengue endemic and epidemic area.

Unfortunately, these individual rates can be poor estimates, especially for the states with small exposures. We saw that some of these states did not experience any deaths and the individual death rate yi/ei = 0 would likely underestimate the states’ true risk of mortality. Also it is found that the rates for the states with small exposures have high variability.

Since the individual death rates are not reliable estimates of the actual situation, so it seems desirable to combine the individual estimates in some way to obtain improved estimates.

with small dengue death the shrinkage size is close to 50%.

Discussion

In India, maximum dengue cases are recorded in Delhi (Table 1). The national capital, Delhi is one of the highest exposed dengue states but with no fatal cases. It indicates that the health service availability in Delhi is excellent but the drainage system, sanitation and hygiene awareness of the citizens needs improvement. Similarly, in the other states inhabited by ethnic tribes mainly in the forest ecosystems, meso to hyper-endemic conditions of dengue exist with the preponderance of dengue to the extent of 90% or even more. During August 2009 to July, 2010 of resurgence of dengue, certain states in India like Kerala, Karnataka, Gujrat and Maharastha are found to have high incidence of dengue infection. Kerala, Haryana and Maharastha performed worst due to high amount of dengue deaths. Among all the states Bihar, Nagaland and Jammu and Kashmir performed best recording zero number of death and lowest number of dengue cases. However, when the true burden of dengue is considered via the posterior expectation of λ the ranking of the states showed several changes as evident from Table 1.

It is interesting to note that Haryana, one of the richest states in India, and with moderate number of dengue cases has a high mortality rate due to dengue and is worst compared to the other states. This may be attributed to some chance causes occurring in a particular year or may be due to some hidden reasons. But this definitely calls for serious concern in subsequent years.

However, the data available with NVBDCP are mostly based on hospital records and as reported by different surveillance programs. Sometimes cases of dengue fever may not be correctly detected. Shekatkar et al. [23] while writing about Leptospirosis comments that- "leptospirosis is easily mistaken for other febrile illnesses including influenza, dengue fever, meningitis, or hepatitis. Therefore, rapid and appropriate laboratory diagnostic tests are needed to aid clinical case identification and to facilitate the implementation of rapid outbreak investigations for optimal treatment and patient management." Thus, testing, diagnosing, relieving symptoms, and expecting a cure treatment are important for the contemporary health care service [24].

Poor economic condition and deplorable conditions of living of the people in the country is a hindrance in the control of Dengue in spite of several efforts from the government and NGOs. Misconceptions and wrong beliefs are common, which increases the gap between knowledge and practice in the public [25], ultimately leading to diseases that can be otherwise controlled by public awareness.

Conclusion

The statewise ranking based on dengue status can be computed in each year based on its status report presented by NVBDCP. This statewise dengue death comparison can be helpful in providing necessary guidelines for planning the course of action for the state specific dengue control programme. The public health facility to prevent the dengue and the health service facility to stop the dengue fatal are played important role to rank the states. In future these two factors are needed to control to deal with dengue cases and deaths.

Conflict of interest

None declared.

¹The Software can be freely downloaded from https://cran.r-project.org/

References

Erum K , Mehreen K, Nabil K, Amna N, Salma A, Rumina H. Demographic and Clinical Features of Dengue Fever in Pakistan from 2003–2007: A Retrospective Cross-Sectional Study. PLoS ONE.2010;5(9): 1-7.
G N Malavige1, P K, V G N S, Fernando S, D H, Aaskov J and Seneviratne S L Patterns of disease in Sri Lankan dengue patients. Arch Dis Child. 2006;91: 396- 400.
Gubler DJ. Dengue and dengue hemorrhagic fever. Clinical Microbiology Reviews, 1998; 11(3): 480-496.
Teixeira MDG, Costa MCN, Guerra Z, Barreto ML: Dengue in Brazil: Situation- 2001 and trends. Dengue Bull 2002;26: 70-76.
Guzman MG, Kouri G . Dengue and dengue hemorrhagic fever in the Americas: lessons and challenges. J Clin Virol.2003; 27: 1-13.
World Health Organization.Dengue and Dengue Haemorrhagic Fever.2008 WHO Fact Sheet. WHO, Geneva.
Guha-Sapir D, Schimmer, B. Dengue fever: new paradigms for a changing epidemiology. Emerging Themes in Epidemiology, 2005; 2;2(1):1.
Sabin AB. Research on dengue during World War II. Am J Trop Med Hyg 1952;1: 30-50.
Rao CVRM. Dengue fever in India. Indian J Pediatr 1987; 54:11-4.
Sharma PL, Sood OP, editors. Round table conference series—dengue outbreak in Delhi: Gurgaon, India: Ranbaxy Science Foundation; 1996.
Jha P, G Vendhan , G C Prakash, R Kumar, M Prem, D Neeraj ,P Richard , and RGI-CGHR Prospective Study Collaborators. Prospective study of one million deaths in India: rationale, design, and validation results. PLoS Medicine 2006; 3(2): e18.
Chakravarti A and Kumaria R. Ecoepidemiological analysis of dengue infection during an outbreak of dengue fever, India. Virology Journal, 2005.2:32.
Carey DE, Myers RM, Reuben R. Dengue types 1 and 4 viruses in wild-caught mosquitoes in south India. Science. 1964.143: 131-132.
Myers RM, Varkey MJ, Reuben R, Jesudass ES. Dengue outbreak in Vellore, southern India, in 1968, with isolation of four dengue types from man and mosquitoes. Indian J Med Res. 1970;58: 24-30.
Vitarana UT, Jayasekera N, Withane N, Gubler DJ. Finding the cause of dengue hemorrhagic fever outbreaks in Sri Lanka. Arbovirus Research in Australia. 1993;6: 125-129.
Chan YC, Salahuddin NI, Khan J, Tan HC, Seah CL, Li J, et al. Dengue haemorrhagic fever outbreak in Karachi, Pakistan, 1994. Trans R Soc Trop Med Hyg. 1995;89: : 619-620.
Bharaj P, Chahar HS, Pandey A, Diddi K, Dar L, Guleria R, et al. Concurrent infections by all four dengue virus serotypes during an outbreak of dengue in 2006 in Delhi, India. Virol J. 2008;5:1. 5- 1.
Dar L, Broor S, Sengupta S, Xess I, Seth P. The first major outbreak of dengue hemorrhagic fever in Delhi, India. Emerg Infect Dis. 1999;5:. 589-590.
Broor S, Dar L, Sengupta S, Chakaraborty M, Wali JP, Biswas A, Kabra SK, Jain Y, Seth P: Recent dengue epidemic in Delhi, India. In Factors in the emergence of arbovirus diseases Edited by: Saluzzo JE, Dodet B. Paris: Elsevier; 1997: 123-127.
Gupta E, Dar L, Narang P, Srivastava VK, Broor S: Serodiagnosis of dengue during an outbreak at a tertiary care hospital in Delhi. Indian J Med Res 2005, 121: 36- 38.
Jha P. Reliable mortality data: a powerful tool for public health. Natl Med J India 2001;14( 3): 129-131.
Ekta Gupta, Lalit Dar, Geetanjali Kapoor and Shobha Broor The changing epidemiology of dengue in Delhi, India. Virology Journal. Virology Journal 2006, 3:92. 1-5.
Shekatkar S, Harish BN, Parija SC. IgM Dot-ELISA Assay using prevalent Leptospira strain for diagnosis of leptospirosis. International Journal of Collaborative Research on Internal Medicine & Public Health. 2010; 2(10): 338-346.
Smego RA. Welcoming the Next Generation of Diseases. International Journal of Collaborative Research on Internal Medicine & Public Health. 2011,3(1): 2-6.
Janahi E, Awadh M, Awadh S. Public knowledge, risk perception, attitudes and practices in relation to the swine flu pandemic: A cross sectional questionnaire-based survey in Bahrain. International Journal of Collaborative Research on Internal Medicine & Public Health. 2011; 3(6):451-464.
Jim Albert. Bayesian Computation with R. New York. Spinger Verlag. 2009.