You are on page 1of 7

G Model

JIPH-1176; No. of Pages 7 ARTICLE IN PRESS


Journal of Infection and Public Health xxx (2019) xxx–xxx

Contents lists available at ScienceDirect

Journal of Infection and Public Health


journal homepage: http://www.elsevier.com/locate/jiph

Risk factors for dengue outbreaks in Odisha, India: A case-control


study
Subhashisa Swain a,c,∗ , Minakshi Bhatt a , Debasish Biswal b , Sanghamitra Pati d ,
Ricardo J. Soares Magalhaes e,f
a
Indian Institute of Public Health-Bhubaneswar, Public Health Foundation of India, Odisha, India
b
Department of Biotechnology, Ravenshaw University, Cuttack, Odisha, India
c
School of Medicine, University of Nottingham, Nottingham, United Kingdom
d
Regional Medical Research Center, Indian Council of Medical Research, Odisha, India
e
UQ Spatial Epidemiology Laboratory, School of Veterinary Science, The University of Queensland, Gatton 4343 QLD, Australia
f
UQ Child Health Research Centre, Children’s Health and Environment Program, The University of Queensland, South Brisbane 4101 QLD, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Background: Environmental and climatic risk factors of dengue outbreak has been studied in detail. How-
Received 21 May 2019 ever, the socio–epidemiological association with the disease is least explored. The study aims to identify
Received in revised form 13 August 2019 the social and ecological factors associated with emerging dengue in Odisha, India.
Accepted 25 August 2019
Methods: A population-based case-control study (age and sex matched at the ratio of 1:1) was conducted
in six districts of the state in 2017. A structured validated questionnaire was used to collect information
Keywords:
for each consenting participant. An ecological household survey was done using a checklist during the
Dengue
month of July–September. Along with the descriptive statistics, conditional logistic regression model was
Outbreak
Epidemiology
used to calculate the adjusted odds ratio using STATA.
Odisha Results: Of 380 cases, nearly 55% were male and the median age was 33 years. The adjusted odds of having
India dengue was nearly three times higher among the people having occupation which demands long travel,
presence of breeding sites (1.7; 95% CI 1.2–2.6), presence of swampy area near home (1.5; 95% CI 1.1–2.1)
and having travel history close to the index date (1.6; 95% CI 1.1–2.4). People staying in thatched houses
had three times higher risk of the disease, however, households keeping the swampy areas clean had
50% less risk for the disease (0.5; 95% CI 0.31–0.67). Nearly 22.2% of cases had a travel history during the
index date. Of them, 36% had diagnosis before the travel, whereas, 64% developed the disease after the
returning from the travel.
Conclusion: Household factors such as occupation and ecological condition of households play important
roles in dengue outbreaks in Odisha. However, our study suggests travel/commuting are also essential
factors to be considered during disease prevention planning.
© 2019 The Authors. Published by Elsevier Limited on behalf of King Saud Bin Abdulaziz University
for Health Sciences. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction 120 countries [1]. However, these estimates can be higher due
to large number of under-reporting and non-symptomatic cases.
Dengue, a neglected tropical vector born disease has become Most of the notified cases are concentrated in the South East Asia
challenge to the public health system in the world. The estimated (SEA) and Western Pacific Regions of the World Health Organiza-
annual dengue burden in the globe is nearly 390 million cases from tion (WHO) [2,3]. In the SEA, countries like Thailand, Indonesia,
India contribute in large extent to the burden [4].
Since the mid-1990s, dengue epidemic episodes in India have
grown rapidly and have become more frequent [5]. Currently India
∗ Corresponding author at: School of Medicine, University of Nottingham, United is endemic for both dengue fever (DF) and dengue hemorrhagic
Kingdom. fever (DHF). Initially the infection was geographically limited to
E-mail addresses: Subhashisa.Swain@nottingham.ac.uk (S. Swain), a few states, which later expanded to most of the states in the
bminakshi563@gmail.com (M. Bhatt), debasishbiswal1492@gmail.com (D. Biswal),
country [6]. Odisha, an Eastern state of the country historically
drsanghamitra12@gmail.com (S. Pati), r.magalhaes@uq.edu.au
(R.J. Soares Magalhaes). did not have any reported cases of the dengue till 2009, but now

https://doi.org/10.1016/j.jiph.2019.08.015
1876-0341/© 2019 The Authors. Published by Elsevier Limited on behalf of King Saud Bin Abdulaziz University for Health Sciences. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015
G Model
JIPH-1176; No. of Pages 7 ARTICLE IN PRESS
2 S. Swain et al. / Journal of Infection and Public Health xxx (2019) xxx–xxx

Table 1
Sociodemographic description [Univariate analysis].

Variables Category Control Case Total Crude OR


N = 380% [95% CI] N = 387% [95% CI] N = 767% [95% CI] [95% CI]

Sex Men 54.5 [49.4–59.4] 52.2 [47.2–57.1] 53.3 [49.8–56.8] Matched


Women 45.5 [40.6–50.6] 47.8 [42.9–52.8] 46.7 [43.2–50.2]

Age group <= 15 years 12.6 [9.6–16.4] 12.7 [9.7–16.4] 12.6 [10.5–15.2] Matched
16–25 years 20.0 [16.3–24.3] 23.3 [19.3–27.7] 21.6 [18.9–24.7]
26–35 years 23.9 [19.9–28.5] 22.2 [18.3–26.7] 23.1 [20.2–26.2]
36–45 years 21.3 [17.5–25.7] 18.6 [15.0–22.8] 19.9 [17.3–22.9]
>= 46 years 22.1 [18.2–26.6] 23.3 [19.3–27.7] 22.7 [19.9–25.8]

Education No schooling 19.5 [15.8–23.8] 14.7 [11.5–18.6] 17.1 [14.6–19.9] Reference


Primary 20.0 [16.3–24.3] 19.1 [15.5–23.4] 19.6 [16.9–22.5] 1.2 [0.7–1.9]
Secondary 47.6 [42.6–52.7] 42.4 [37.5–47.4] 45 [41.5–48.5] 1.2 [0.8–1.8]
High School and above 12.9 [9.9–16.7] 23.8 [19.8–28.3] 18.4 [15.8–21.3] 2.5 [1.5–4.2]*

Ethnicity Non SC/ST 22.6 [18.7–27.1] 15.5 [12.2–19.5] 19.1 [16.4–22.0] Reference
SC/ST 77.4 [72.9–81.3] 84.5 [80.5–87.8] 81.0 [78.0–83.6] 1.6 [1.1–2.3]*

Occupation Unemployed 22.4 [18.4–26.8] 9.8 [7.2–13.2] 16.1 [13.6–18.8] Reference


Agriculture/Daily worker 15.3 [11.9–19.2] 19.4 [15.7–23.6] 17.3 [14.8–20.2] 3.9 [2.1–7.5]*
Business 7.6 [5.3–10.8] 8.8 [6.3–12.1] 8.2 [6.5–10.4] 3.5 [1.7–7.4]*
Home maker/student 36.6 [31.9–41.6] 46.5[41.6–51.5] 41.6 [38.1–45.1] 4.6 [2.6–8.1]*
Industry/Office 19.1 [14.6–22.4] 15.5 [12.2–19.5] 16.8 [14.3–19.6] 2.7 [1.4–5.2]*

Housing type Asbestos roof 32.4 [27.8–37.2] 19.9 [16.2–24.2] 26.0 [23.1–29.3] Reference
Concrete roof 30.3 [25.8–35.1] 34.6 [30.1–39.5] 32.5 [29.2–35.9] 1.9 [1.3–2.9]*
Mixed roof 18.4 [14.8–22.6] 17.8 [14.3–21.9] 18.1 [15.5–21.0] 1.6 [1.1–2.6]*
Thatched roof 18.9 [15.3–23.2] 27.6 [23.4–32.3] 23.3 [20.5–26.5] 2.6 [1.7–4.1]*

SC- Schedule caste, ST- Schedule Tribe; OR- Odds Ratio; CI- Confidence Interval.
*
P value < 0.05.

there is wide reporting from all the districts. According to the burden and studying the serotypes [20,21]. Till date, there is scant
surveillance data, the state now contributes nearly 10–15% total of evidence on social epidemiology of the disease in the state of
dengue cases of the country [7]. These cases are scattered in distri- Odisha. We aimed to identify the socio–demographic and ecologi-
bution and uneven with circulation of four serotypes of the virus cal factors associated with dengue outbreaks in the state of Odisha,
[8]. India through population-based case-control study design.
The transmission of dengue is dependent on various macro and
micro level factors such as temperature, humidity, rainfall, the pop- Methods
ulation density, movement, immunity and virus load, urbanization,
environmental factors and socio-demographic and economic fac- Sample size calculation
tors [4,9,10]. These influence the spread of the disease through
increased Aedes aegypti mosquito population, transmission of the For estimating sample size and testing of the study tool, a pilot
vector, spreading of the disease and practices of protection mech- study was conducted among 30 subjects (10 cases and 20 controls)
anism. Vector dynamics [11], population movement [12] and their from non-study area. Findings from the pilot study was used for
association with the dengue outbreak is well searched. Climatic sample size calculation in the absence of relevant literature on risk
and environmental factors create the most favorable conditions of travel for dengue. To detect minimum 10% difference of travel
for dengue transmission and mostly studied. But investigating the exposure to endemic area among case (25%) and control (15%) of 1:1
socio-demographic and other individual factors is equally impor- ratio, with a power of 90% at a significant level of 0.05, the required
tant for identifying the people at risk. sample size was calculated as 354 in each group. Considering 10%
Especially, with current shift in the disease pattern it has become non-response rate, total required sample in each group was 393.
essential to study the mentioned factors in detail. Studies say, Proportionate sampling method was used to recruit patients from
dengue is no more a disease of women, children and urban prob- five districts, based on the disease burden in the year 2016. Within
lem [13–15]. Community and individual practices can prevent or each district, patients were selected randomly from the available
increase the chances of the disease depending on the nature of list of cases obtained from the health department. The study was
practices. Potential socio–demographic and ecological factors can conducted during the rainy seasons from July to September 2017
spread dengue to other regions due to increasing frequency of with additional purpose of inspecting the environmental condi-
transportation and travel [16]. Even though environmental factors tions (Table 4).
are crucial for the disease epidemic, a comprehensive prevention
mechanism would require information at individual level. Espe- Study setting
cially, in country like India, where diversity and differences are
predominant, studying these factors is essential. The eastern state of India, Odisha covers about 155,707 sq km of
The recent spike of cases in the state Odisha supports the the total area of the country and has nearly 48,903 km2 of forests
hypothesis of availability of favorable conditions for the disease spread over 31.41% of the state’s total area. The average rain-
spread [17]. Moreover, recurring dengue epidemics also cre- fall in Odisha is measured as 1482 mm, of which 78% is received
ates hyperendemic areas, typically large, densely populated areas between the months of June and September of a year. The summer
where several or all four serotypes dengue virus circulate in a sus- season in Odisha spans from March to June during which the max-
tained fashion. In the state, increased numbers of cases are reported imum temperature at most of the places goes well above 40 ◦ C.
among children and adults [8,18,19]. The available literature is lim- The state projects distinct yet homogeneous features of topogra-
ited to the epidemiological profile of the clinical cases, disease phy and divided into five morphological units such as mountainous

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015
G Model
JIPH-1176; No. of Pages 7 ARTICLE IN PRESS
S. Swain et al. / Journal of Infection and Public Health xxx (2019) xxx–xxx 3

Fig. 1. Mapping of case and control samples.


Red dots represent cases and green dot represents controls. These are only spot maps, does not represent the numbers exactly.

and highlands region, coastal plains, western rolling uplands, cen- sociodemographic, travel history and exposure, environmental
tral plateaus and flood plains. According to the 2011 census of India, condition and cleanliness practice, disease status of the family.
the total population of Odisha was 41,947,358, of which 21,201,678 Socio–demographic and household profile section had questions on
(50.54%) are male and 20,745,680 (49.46%) are female with popu- age, sex, ethnicity (defined as ‘schedule caste/tribe’ or not according
lation density of 269 per km2 . to government classification), education (based on the schooling
The study was carried out in 5 of the 9 high dengue burden completed), occupation (current and during the disease) and hous-
districts of the state, which contributes to more than 85% of the ing conditions. The housing type was categorized according to the
total cases. Details of the distribution of dengue has been provided type of roof, like asbestos, concrete, thatched or mixed. Travel his-
elsewhere [17]. These districts represent nearly, 40% of the total tory section asked questions on travel history during the disease
population. Within each district, high burden blocks were selected, period, duration of travel, place of travel, onset of disease in relation
based on reported number of cases. This purposive selection was to travel time and purpose of travel. Environmental and sanita-
done after consulting with the state nodal officers. tion section constituted questions on the presence of mosquito
breeding sites, current and past cleaning practices (during the dis-
Study participants ease period). Reported environmental conditions were validated
with the visit made by investigators with a check list. Participant’s
The list of confirmed dengue patients within last one year (2016) houses and gardens, including those of the neighbors (living within
recorded at public & private hospitals of Odisha was used for selec- 10 ms radius of the house of the study participants) were inspected
tion of cases. The place of residence at the time of getting disease to check the sources for the stagnant water, one of the favorable
was chosen as the study areas. This information was obtained breeding sites. The potential dengue breeding sites such as, water
from National Vector Borne Disease Control Program (NVBDCP) storage containers or tanks (whether the containers were tightly
cell, Odisha, India. We excluded cases, who had disease history covered or not), gutters (collect rainwater), empty receptacles (bot-
of more than one year to avoid the recall bias. For the participant tles, cans, tires, plastic containers, coconut shells etc.), presence of
aged less than 18 years, parents were interviewed, preferably those any cattle shed was noted by the investigators. The heads of fami-
who spend more time with them (for e.g. mother for single parent lies (cases, controls and the respective neighbors) who were asked
working and grandparents or elder siblings in case of both parents to participate in the study permitted inspection of their houses,
working). courtyards and gardens and voluntarily gave all information
Cases were confirmed dengue fever patients tested through IgM requested.
method and hospitalized to any public or private health facilities Final approved tool was created in epi collect+ software and
of the state. Controls were from same living area of cases matched installed in two tablets which was used during data collection.
with sex and age (±2 years) with no history of dengue, chikungunya We used digital data collection methods to obtain accurate geo-
or acute febrile illness in last one year. People having any history coordinates of the houses, easy data management and real time
of dengue in last one year but were not living in the study area in monitoring. Ten percent of the randomly picked data from total
during that period were excluded from the study. The index date was cross checked by the principal investigator. Members of the
of the matched control was same that of (diagnosis date) cases. All research team had access to collected data through unique pass-
the questions referred to the index date only. The distribution of word.
the cases and controls are given in Fig. 1.
Ethical considerations
Data collection procedures
The study was approved by state research and ethics committee,
For the data collection, cases and controls were interviewed Department of Health and Family Welfare, Government of Odisha,
with the help of a questionnaire designed for the study in vernacu- India (letter number 141/SHRMU). Participants were explained
lar language (Odia). The tool was tested for validity and reliability about the nature and the purpose of the study. Anonymity and
and reported Cronbach’s alpha (0.76) for all the items. Based confidentiality of information provided were guaranteed. Before
on the responses from pilot study, necessary modification was the commencement of the interview, their right to refuse or with-
done. The survey questionnaire comprised four sections namely; draw from the interview at any time during the process was clearly

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015
G Model
JIPH-1176; No. of Pages 7 ARTICLE IN PRESS
4 S. Swain et al. / Journal of Infection and Public Health xxx (2019) xxx–xxx

Table 2
Environmental practices [univariate analysis].

Environmental condition Control Case Total Crude OR


during the index date N = 380% [95% CI] N = 387% [95% CI] N = 767% [95% CI] [95% CI]

Presence of swampy area 71.0 [66.3–75.4] 78.6 [74.2–82.4] 74.8 [71.6–77.8] 1.5 [1.1–2.0]*
surrounding home
Participants clean the 51.1 [46–56.1] 43.7 [38.8–48.7] 47.3 [43.8–50.9] 0.7 [0.5–0.9]
swampy Area
Presence of breeding sites 44.5 [39.5–49.5] 55.5 [50.5–60.4] 50.1 [46.5–53.6] 1.6 [1.2–2.1]*
surrounding home
Presence of container 18.4 [14.8–22.7] 29.5 [25.1–34.2] 24.0 [21.1–27.1] 1.8 [1.3–2.6]*
breeding surrounding
home
Participants clean the 20.0 [16.3–24.3] 22.7 [18.8–27.2] 21.4 [18.6–24.4] 1.2 [0.8–1.8]
breeding sites
Presence of the 18.4 [14.8–22.7] 21.4 [17.6–25.8] 19.9 [17.3–22.9] 1.2 [0.8–1.7]
construction sites

Reference group for all the variables was ‘No’; OR- Odds Ratio; CI- Confidence Interval.
*
P value < 0.05.

Table 3
Travel history during the index date.

Variables Categories Control Case Total Crude OR


N = 380 n[95% CI] N = 387 N = 767 [95% CI]
n[95% CI] n[95% CI]

Travelled during the Yes 16.3 [12.9–20.4] 22.2 [18.3–26.7] 19.3 [16.7–22.3] 1.5 [1.1–2.1]*
index datea
Purposes of the travel Never 83.7 [79.6–87.1] 77.8 [73.3–81.7] 80.7 [77.7–83.3] Reference
Business 4.5 [2.8–7.1] 2.1 [1.0–4.1] 3.3 [2.2–4.8] 0.5 [0.2–1.2]
Work 4.7 [3–7.4] 12.4 [9.5–16.1] 8.6 [6.8–10.8] 3.3 [1.8–6.1]*
Other 7.1 [4.9–10.2] 7.8 [5.5–10.9] 7.4 [5.8–9.5] 1.2 [0.7–2.1]

Place visited during the Did not travel 83.7 [79.6–87.1] 77.8 [73.3–81.7] 80.7 [77.7–83.3] Reference
travel
Nearby town 7.4 [5.1–10.5] 11.4 [8.6–14.9] 9.4 [7.5–11.7] 1.7 [1.1–2.7]*
Cities within the state 4.7 [3.0–7.4] 2.1 [1.0–4.1] 3.4 [2.3–4.9] 0.5 [0.2–1.1]
Out of the state 4.2 [2.6–6.8] 8.8 [6.3–12.1] 6.5 [5.0–8.5] 2.4 [1.3–4.7]*

Period of travel with Did not travel 83.7 [79.6–87.1] 77.8 [73.3–81.7] 80.7 [77.7–83.3] Reference
reference to index date
Within 7 days of 8.7 [6.2–12] 14.2 [11.1–18.1] 11.5 [9.4–13.9] 1.8 [1.1–2.8]*
getting the disease
7-30 days 1.3 [0.5–3.1] 3.1 [1.8–5.4] 2.2 [1.4–3.5] 3.2 [1.02–10.1]*
1month to 6 months 2.9 [1.6–5.2] 1.8 [0.9–3.8] 2.3 [1.5–3.7] 0.80 [0.29–2.1]
More than one year 3.4 [2.0–5.8] 3.1 [1.8–5.4] 3.3 [2.2–4.8] 1.1 [0.41–2.4]

OR- Odds Ratio; CI- Confidence Interval.


*
P value < 0.05.
a
Reference group —‘did not travel’.

explained. Written consent was obtained from each adult partic- hold had more mosquito breeding sites (55.5%) and container
ipants and parental consent was obtained for study participants breeding empty receptacles (29.5%), which was statistically signif-
aged less than 18 years before the interview. icant (Table 2).
Travel history collected during the survey is provided in Table 3.
Nearly 22.2% of cases had travel history during the index date. Of
Results them, 36% had diagnosis before the travelling, whereas, 64% were
diagnosed with dengue after returning from the travel. Describ-
Dataset for analysis ing the purposes of travel, 12.4% of the cases travelled for work,
whereas, 7.8% travelled for other purposes (including visit to rela-
In total, 767 individuals (387 cases and 380 controls) from 5 dis- tive home). Of the cases, who travelled, 80% travelled within one
tricts participated in the study. Of 387 cases, 380 had age (±2 years) month of the disease and 53% travelled to the nearby town (Fig. 2).
and sex matched controls. Proportion of males (53.3%) were higher
than females but not statistically significant. Participants age var-
ied from 3 to 91 year (median age 33 year, IQR 8). The response Risk factors for dengue outbreaks in Odisha
rate during the study was 97.54% (780 subjects were contacted but
only 767 responded positively). Other sociodemographic charac- Univariate regression models show ethnicity, occupation,
teristics is described in Table 1. Significant association was seen higher education, housing types, environmental factors like pres-
with occupation, education and type of housing. ence of swampy area around home, travel history, travelling from
Compared to controls, cases (78.6%) had more swampy areas outside of the state and having disease before the travelling had
(including gutter) surrounding their home. Nearly, 51% of the con- statistically significant unadjusted odds ratio (OR) at p value 0.05.
trol population clean their environment compared to 43.7% cases. Multivariate conditional logistic regression model explains the
Whereas, equal proportion of people in both the groups, were association with occupation (OR ranged from 3–5), housing (OR
involved in cleaning of wastes surrounding their home. Case house- ranged from 2–3), travelling during the disease [1.6, 95% CI: 1.1–2.4]

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015
G Model
JIPH-1176; No. of Pages 7 ARTICLE IN PRESS
S. Swain et al. / Journal of Infection and Public Health xxx (2019) xxx–xxx 5

Fig. 2. Distribution of the travel details among cases. Percentage in each cell has been calculated with respect to the total people travelled during the disease (86).

and presence of the breeding sites [1.7, 95% CI: 1.2–2.6] were sta- Housing structure is proved to be linked with dengue outbreak.
tistically significant after adjusting for other variables. Studies reported that, staying in sheds/old flats creates higher
chances of dengue [16]. Especially, densely and nearly located
houses increase the dengue spread chances because of the crowd-
Discussion ing and environmental conditions [24]. I Peri–urban areas of India
are crowded, and the houses are closely located. Thus, the outbreak
Researchers have studied the risk factors of dengue transmis- is thought to have affected all the houses irrespective of the struc-
sion, especially the environmental and climatic associations. More ture. But gradient association with the housing structure provides
advanced researches use the population dynamics of both human a proxy indication towards the socio–economic condition of the
and vector to understand the outbreak. However, most of these people. People living with thatched houses, may represent lower
population-based studies are based on simulative mathematical socio–economic strata compared to others. So, increased dengue
models [11]. Few in India have researched the epidemiological from these houses could be because of the low protective mea-
determinants for dengue. This is the first ever study done to identify sures in practice. Also, the housing characteristics such as windows,
the risk factors in the Eastern India. We found the leading associ- curtains, gardens and its relationship with dengue breeding sites
ated factors are (i) occupation demanding travel has nearly three cannot be ruled out. This specific aspect needs further investigation
times higher risk; (ii) presence of breeding sites increases the risk in an Indian context.
by nearly two times and (iii) travel during the disease increases the Environmental factors as the risk factor of dengue is well estab-
risk by two times as well. The case-control study is matched for lished in different countries [16,25–27]. Interestingly, we found
age, gender and small extent of area of living; thus, the identified the presence of swampy area and breeding sites increases the
individual and housing level factors making the estimates real and dengue outbreak by two times but, the presence of container was
comparable. not significant. Similar findings were reported other studies [28].
We found significant association of dengue with occupation and The non-association of the container breeding sites could have
housing type. Which demonstrates a clear difference in the occupa- two possible explanations. One, there could be lack of knowl-
tion and housing characteristics of cases compared to controls. The edge in understanding the ‘container breeding’ sites. Even though
positive association with businesspersons and agriculture/daily the researchers explained the participants with examples, still the
worker indicates the jobs demanding travel to other places or under-reporting could have been because of the knowledge gap
nearby towns predisposes them more at risk towards the disease. [29,30]. Another factor could be unorganized and unstructured
There is inconsistent evidence on dengue association with occu- urbanization as most of the participants were from the peri-urban
pation [16,22,23]. The reported age group of cases in our study area, the environmental sanitation conditions are often neglected.
is quite same as reported in other studies from the country [6]. Thus, the presence of container breeding sites in surrounding areas
However, the four times high association with students and home- instead of the house of the participants which could have been over-
makers needs further investigation. Also, the reporting of the cases looked and under reported. We did find similar responses among
within same families need to be studied in detail for better expla- school children in our previous study [31]. Presence of breeding
nation. As there is no difference across the groups observed, our sites and their association necessitates a comprehensive preven-
findings indicate the possible source of the disease could be the tion strategy for stopping the dengue spread, which is supported
workplace. by the findings from our study. Dengue is being new to the state,

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015
G Model
JIPH-1176; No. of Pages 7 ARTICLE IN PRESS
6 S. Swain et al. / Journal of Infection and Public Health xxx (2019) xxx–xxx

Table 4 questionnaire and limited the recall period to 12 weeks. Interviews


Conditional logistic regression for association of dengue.
were always performed by the same investigators to minimize the
Variable Categories Adjusted OR [95% inter-observer bias. We reduced the non-response bias by active
CI] case recruitment through the health workers. These health work-
Ethnicity SC/ST Reference ers live in the study areas and have most of the information about
Non SC/ST 1.4 [0.9–2.1] the patient. Because of financial constraints and logistic reasons, it
Education Illiterate Reference was not possible to perform laboratory test for the control group.
Primary 0.8 [0.47–1.5] Some controls accidentally may have been asymptomatic cases,
Secondary 0.9 [0.54–1.5] and misclassification bias cannot be excluded. We reduced this bias
High School and 1.4 [0.76–2.8]
by excluding rigorously subjects from the control group who pre-
above
sented with fever or any of the symptoms associated with dengue
Occupation Unemployed Reference fever. For the control group, who was not sure about the questions,
Agriculture/Daily 3.8 [1.8–7.9]*
help of other family members were taken. Our findings are based
worker
Business 4.6 [1.9–10.6]* on cases reported in 2016 only. Further, expanding the analysis,
Home 4.4 [2.3–8.4]* we found nearly 2/3 of cases developed the dengue after returning
maker/student from the travel. However, we did not have enough sample size to
Industry/Office 3.0 [1.4–6.5]* run any further regression model to estimate the risk. Thus, gener-
Housing Asbestos Reference alization of the risk factor needs caution because of the change in
Concrete 1.9 [1.2–3.1]* dengue virus circulation in the state.
Mixed 2.0 [1.1–3.5]*
Thatched 2.9 [1.7–5.1]*
Conclusions
Presence of swampy area No Reference
surrounding home
Yes 1.5 [1.1–2.1]* Our study is the first in the Eastern India to explore the risk
factors at individual and environmental level. Our findings corrob-
Participants clean the swampy No Reference
area
orate the available evidence on dengue transmission and suggest
Yes 0.5 [0.31–0.67] a dynamic pathway. It can hypothesize that, the initial infection
of the cases happened at other places than home. The travel his-
Presence of container breeding No Reference
surrounding the home tory might help the infected person to carry the virus to home and
Yes 1.5 [0.95–2.3] the favorable environmental conditions and mosquitos could have
spread the disease. At present, we do not have enough evidence to
Presence of any breeding site No Reference
Yes 1.7 [1.2–2.6]* support our hypothesis, but the future research should explore this
aspect.
Travelled during the index date No Reference
Yes 1.6 [1.1–2.4]*
Authors’ contributions
Variables with significant association in univariate analysis were included in mul-
tivariate analysis; OR- Odds Ratio; CI- Confidence Interval; SC- Schedule Caste; ST-
Schedule Tribe. SS and RSM conceptualised the study. MB and DB did data col-
*P value < 0.05. lection. SS and MB did data analysis. SS and SP drafted the full paper.
RSM and SP revised edited the manuscript. All authors contributed
to final development of the article.
the lower knowledge about the disease and risk factors cannot be
excluded.
Funding
The travel history among cases during the index date was asso-
ciated with more risk. This explains the link between the travel
This study was financially supported by the PHRI-Research
and dengue infection, which to some extent gives information on
Grant awarded by Public Health Foundation of India with the
the source of the infection. There is a clear evidence on the travel
financial support from Department of Science and Technology, Gov-
history and dengue infections both nationally and internationally
ernment of India (No. IN-DL220960833674480).
[32]. Studies from other cities of India also reported similar findings
[11]. We found majority of the people got the disease within 7 days
Competing interests
of the travel and of them nearly 70% travelled to the nearby cities.
Even though we did not enquire about the reason of the travel, the
None declared.
association of occupation and dengue it is an attempt for it. The
purposes of the travel could be for business, occupation and visit to
other places/houses. A study mapped the epidemic dynamics of the Ethical considerations
vector borne diseases in villages and city with commuter network
found positive association with travel to the cities and acquiring the The study was approved by state research and ethics committee,
infection [33]. Studying the age structure in detail we find, nearly Department of Health and Family Welfare, Government of Odisha,
30% of the people are below 25 years and 45% of the cases were India with letter number 141/SHRMU.
women. So, the higher chances of the travel could be to the schools,
workplace or to relative’s houses. Stoddard et al. reported the posi- Consent for publication
tive association with house–house movement [34]. Similar findings
are reported in Brazil [35]. Verma et al. too mentioned the travel Not applicable.
history as an essential criterion for suspicion of dengue in India
[36]. A mathematical model used similar concept to estimate the Availability of data and materials
outbreak in the western India [37].
Our study has some inherent limitations of case-control study The data that support the findings of this study are avail-
design. To minimize recall bias, we used a pre-tested standardized able from National Vector Borne Disease Control Cell, Department

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015
G Model
JIPH-1176; No. of Pages 7 ARTICLE IN PRESS
S. Swain et al. / Journal of Infection and Public Health xxx (2019) xxx–xxx 7

of Health and Family Welfare, Government of Odisha, India but [16] Chen B, Yang J, Luo L, Yang Z, Liu Q. Who is vulnerable to dengue fever? A
restrictions apply to the availability of these data, which were used community survey of the 2014 outbreak in Guangzhou, China. Int J Environ Res
Public Health 2016;13:712.
under license for the current study, and so are not publicly avail- [17] Swain S, Bhatt M, Pati S, Soares Magalhaes RJ. Distribution of and associated
able. Data are however available from the authors upon reasonable factors for dengue burden in the state of Odisha, India during 2010–2016. Infect
request and with permission of Department of Health and Family Dis Poverty 2019 [cited 2019 May 22];8. Available from: https://www.ncbi.nlm.
nih.gov/pmc/articles/PMC6501402/.
Welfare, Government of Odisha, India. [18] Padhi S, Dash M, Panda P, Parida B, Mohanty I, Sahu S, et al. A three year retro-
spective study on the increasing trend in seroprevalence of dengue infection
CRediT authorship contribution statement from southern Odisha, India. Indian J Med Res 2014;140:660–4.
[19] Mishra S, Ramanathan R, Agarwalla SK. Clinical profile of dengue fever in
children: a study from Southern Odisha, India. Scientifica 2016;2016 [cited
Subhashisa Swain: Conceptualization, Formal analysis, Funding 2017 Dec 25]. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/
acquisition, Investigation, Methodology, Project administration, PMC4860230/.
[20] Rao MRK, Padhy RN, Das MK. Episodes of the epidemiological factors correlated
Resources, Software, Writing - original draft, Writing - review &
with prevailing viral infections with dengue virus and molecular characteriza-
editing. Minakshi Bhatt: Data curation, Formal analysis, Software. tion of serotype-specific dengue virus circulation in eastern India. Infect Genet
Debasish Biswal: Data curation. Sanghamitra Pati: Investigation, Evol 2018;58:40–9.
[21] Sahu SK, Pasupalak S, Mohanty I, Narasimham MV. Recent
Methodology, Resources, Validation, Writing - review & editing.
trends of seroprevalence of dengue in a tertiary care hospital in
Ricardo J. Soares Magalhaes: Conceptualization, Methodology, Southern Odisha. J Clin Diagn Res 2018;12(01) [cited 2018 Feb
Supervision, Writing - review & editing. 21]; Available from: http://jcdr.net/article fulltext.asp?issn=0973-
709x&year=2018&volume=12&issue=1&page=DC05&issn=0973-
709x&id=11138.
Acknowledgments [22] Harapan H, Rajamoorthy Y, Anwar S, Bustamam A, Radiansyah A,
Angraini P, et al. Knowledge, attitude, and practice regarding dengue
We are thankful to Dr Madan Mohan Pradhan, Joint Director of virus infection among inhabitants of Aceh, Indonesia: a cross-sectional
study. BMC Infect Dis 2018;18 [cited 2019 Mar 11]. Available from:
National Vector Borne Disease Control Cell, Odisha, India for his https://bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-018-3006-
kind support. The authors acknowledge the support provided by z.
the Directorate of Health Services and National Vector Borne Dis- [23] Koyadun S, Butraporn P, Kittayapong P. Ecologic and sociodemographic risk
determinants for dengue transmission in urban areas in Thailand. Interdiscip
ease Control Cell, Government of Odisha for providing the data. We Perspect Infect Dis 2012;2012:1–12.
thank to all the health workers, medical officers, field investigators [24] Seidahmed OME, Lu D, Chong CS, Ng LC, Eltahir EAB. Patterns of urban housing
and participants for their valuable support. shape dengue distribution in Singapore at neighborhood and country scales:
dengue, housing, and drainage in Singapore. GeoHealth 2018;2:54–67.
[25] Schmidt W-P, Suzuki M, Thiem VD, White RG, Tsuzuki A, Yoshida L-M, et al.
References Population density, water supply, and the risk of dengue fever in Vietnam:
cohort study and spatial analysis. PLoS Med 2011;8:e1001082.
[1] WHO | Epidemiology [Internet]. [Cited 2019 Mar 13]. Available from: https:// [26] Kajeguka DC, Msonga M, Schiøler KL, Meyrowitsch DW, Syrianou P, Tenu F,
www.who.int/denguecontrol/epidemiology/en/. et al. Individual and environmental risk factors for dengue and chikungunya
[2] Stanaway JD, Shepard DS, Undurraga EA, Halasa YA, Coffeng LE, Brady OJ, et al. seropositivity in North-Eastern Tanzania. Infect Dis Health 2017;22:65–76.
The global burden of dengue: an analysis from the Global Burden of Disease [27] Sharmin S, Viennet E, Glass K, Harley D. The emergence of dengue in
Study 2013. Lancet Infect Dis 2016;16:712–23. Bangladesh: epidemiology, challenges and future disease risk. Trans R Soc Trop
[3] Bhatt S, Gething PW, Brady OJ, Messina JP, Farlow AW, Moyes CL, et al. The Med Hyg 2015;109:619–27.
global distribution and burden of dengue. Nature 2013;496:504–7. [28] Nagpal BN, Gupta SK, Shamim A, Vikram K, Srivastava A, Tuli NR, et al. Control
[4] Bhatia R, Dash A, Sunyoto T. Changing epidemiology of dengue in South-East of Aedes aegypti breeding: a novel intervention for prevention and control of
Asia. WHO South-East Asia J Public Health 2013;2:23. dengue in an endemic zone of Delhi, India. PLoS One 2016;11:e0166768.
[5] Chakravarti A, Arora R, Luxemburger C. Fifty years of dengue in India. Trans R [29] Kohli C, Kumar R, Meena G, Singh M, Ingle G. A study on knowledge and pre-
Soc Trop Med Hyg 2012;106:273–82. ventive practices about mosquito borne diseases in Delhi. MAMC J Med Sci
[6] Ganeshkumar P, Murhekar MV, Poornima V, Saravanakumar V, Sukumaran K, 2015;1:16.
Anandaselvasankar A, et al. Dengue infection in India: a systematic review and [30] Kumar V, Rathi A, Lal P, Goel S. Malaria and dengue: knowledge, attitude, prac-
meta-analysis. PLoS Negl Trop Dis 2018;12:e0006618. tice, and effect of sensitization workshop among school teachers as health
[7] NVBDCP | National Vector Borne Disease Control Programme [Internet]. [Cited educators. J Fam Med Prim Care 2018;7:1368.
2017 Dec 25]. Available from: http://nvbdcp.gov.in/den-cd.html. [31] Swain S, Pati S, Pati S. ‘Health Promoting School’ model in prevention of
[8] Mishra B, Turuk J, Sahu SJ, Khajuria A, Kumar S, Dey A, et al. Co-circulation of all vector-borne diseases in Odisha: a pilot intervention. J Trop Pediatr Jan. 21
four dengue virus serotypes: first report from Odisha. Indian J Med Microbiol 2019; [cited 2019 Mar 11]; Available from: https://academic.oup.com/tropej/
2017;35:293–5. advance-article/doi/10.1093/tropej/fmy077/5298566.
[9] Akter R, Hu W, Naish S, Banu S, Tong S. Joint effects of climate variability [32] Wichmann O, Jelinek T. Dengue in travelers: a review. J Travel Med
and socioecological factors on dengue transmission: epidemiological evidence. 2006;11:161–70.
Trop Med Int Health 2017;22:656–69. [33] Mpolya EA, Yashima K, Ohtsuki H, Sasaki A. Epidemic dynamics of a vector-
[10] Mutheneni SR, Morse AP, Caminade C, Upadhyayula SM. Dengue burden in borne disease on a villages-and-city star network with commuters. J Theor Biol
India: recent trends and importance of climatic parameters. Emerg Microbes 2014;343:120–6.
Infect 2017;6:e70. [34] Stoddard ST, Forshey BM, Morrison AC, Paz-Soldan VA, Vazquez-Prokopec GM,
[11] Enduri MK, Jolad S. Dynamics of dengue disease with human and vector mobil- Astete H, et al. House-to-house human movement drives dengue virus trans-
ity. Spatial Spatio-Temporal Epidemiol 2018;25:57–66. mission. Proc Natl Acad Sci 2013;110:994–9.
[12] Sirisena P, Noordeen F, Kurukulasuriya H, Romesh TA, Fernando L. Effect [35] da Silva-Nunes M, de Souza VAF, Pannuti CS, Sperança MA, Terzian
of climatic factors and population density on the distribution of dengue ACB, Nogueira ML, et al. Risk factors for dengue virus infection in rural
in Sri Lanka: a GIS based evaluation for prediction of outbreaks. PLoS One Amazonia: population-based cross-sectional surveys. Am J Trop Med Hyg
2017;12:e0166806. 2008;79:485–94.
[13] Arunachalam N, Murty US, Kabilan L, Balasubramanian A, Thenmozhi V, Nara- [36] Verma S, Kanga A, Singh D, Verma GK, Mokta K, Ganju SA, et al. Emergence
hari D, et al. Studies on dengue in rural areas of Kurnool District, Andhra of travel: associated dengue fever in a non-endemic, hilly state. Adv Biomed
Pradesh, India. J Am Mosq Control Assoc 2004;20:87–90. Res 2014;3 [cited 2018 Feb 21]. Available from: https://www.ncbi.nlm.nih.gov/
[14] Biswas D, Bhunia R, Basu M. Dengue fever in a rural area of West Bengal, pmc/articles/PMC4260275/.
India, 2012: an outbreak investigation. WHO South-East Asia J Public Health [37] Enduri MK, Jolad S. Estimation of reproduction number and non stationary
2014;3:46. spectral analysis of dengue epidemic. Math Biosci 2016;288:140–8.
[15] Singru S, Bhosale S, Debnath D, Fernandez K, Pandve H. Study of knowledge,
attitude and practices regarding dengue in the urban and rural field practice
area of a tertiary care teaching hospital in Pune, India. Med J Dr Patil Univ
2013;6:374.

Please cite this article in press as: Swain S, et al. Risk factors for dengue outbreaks in Odisha, India: A case-control study. J Infect Public
Health (2019), https://doi.org/10.1016/j.jiph.2019.08.015

You might also like