Social and Housing Determinants of Dengue and Chikungunya in Indian Adults Aged 45 and Above: Analysis of a Nationally Representative Survey (2017-18)

Dengue and chikungunya (CHIKV) are the two major vector-borne diseases of serious public health concern in India. Studies on socio-economic and housing determinants of dengue and CHIKV at a pan-India level are lacking. Here, we took advantage of the recently carried out Longitudinal Ageing Study in India (LASI) carried across all the States and Union Territories of India to study the social determinants of dengue and CHIKV in India. LASI-1 (2017-2018) data on the self-reported period prevalence of dengue and CHIKV from 70,865 respondents aged ≥ 45 years were used for this analysis. The State-wise distribution of dengue and CHIKV was mapped. Prevalence was estimated for each study variable, and the difference was compared using the χ2 test. The adjusted odds ratios (AOR) of the socio-economic and housing variables for dengue and CHIKV were calculated using a multiple logistic regression model.


Background
The dengue virus primarily transmitted by Aedes aegypti, and to a lesser extent by Ae. albopictus consists of four serotypes (DENV1-4), which contribute to the distinct epidemiological spread [1]. In recent decades, global incidence of dengue has increased alarmingly, and about half the world's population are at risk [1]. Globally, there is an eightfold increase in dengue cases from 505,430 in 2000 to 5.2 million cases in 2019. Even though dengue risk is in 129 countries, the actual burden (70%) is in Asia, and India is one of the major contributors [2,3].
In India, Chikungunya (CHIKV) is the second major vector-borne viral disease transmitted by Ae. aegypti and Ae. albopictus. India has witnessed CHIKV outbreaks from 1963-74 [10]. This was followed by three decades of quiescence, and CHIKV re-emerged in 2005 with 1.39 million suspected cases in 2006, and after a gradual decline till 2014, the cases started rising in 2015, with 67,769 cases reported in 2017 [11].
An estimated 56.3-98% of the population in India are still susceptible to CHIKV, and this could explain the continuous transmission of CHIKV after re-emerging in 2005 [12].

Indian Council of Medical Research (ICMR) and Department of Health Research (DHR) have set up Viral
Research Diagnostic Laboratories throughout India to diagnose viral diseases [13]. The major vector control intervention for dengue and CHIKV in India are source reduction, larviciding the positive containers (Temephos 50% EC), indoor space spray (Pyrethrum, Cyphenothrin 5% EC) and outdoor fogging (Technical Malathion, Cyphenothrin 5% EC) [11].
In India, dengue and CHIKV pose a serious public health risk, and for effective control strategies, in addition to environmental risk factors, it is important to understand the socio-economic determinants of health (SDH) in uencing transmission. Income, education, employment-status, housing and access to affordable health care services are some of the important SDH that affect health equity [14]. Studies detailing the socio-economic and housing risk factors of viral vector-borne diseases are sparse in India, and are largely focused on selected districts [15][16][17][18]. A pan-India study on the socio-economic and housing determinants of vector borne diseases may provide important insights to their prevention and control. A nation-wide Longitudinal Ageing Study in India (LASI) wave 1 was carried out for the rst time in India (2017-18) to collect important information on health, health care, socio-economic status (SES) and self-reported prevalence of vector borne diseases including dengue and CHIKV among adults aged 45 and above [19]. Here, we have analyzed the LASI data, and detailed the SES and housing risk factors of dengue and CHIKV.

Data and participants
Data from the recent LASI wave 1 (2017-2018) carried out by the International Institute for Population Sciences (IIPS), Mumbai, India, was used for the analysis. The LASI wave 1 is a nationally representative study of adults aged ≥45 from all States and Union Territories (UT) except Sikkim. LASI has gathered important information on health, infectious diseases, socio-economic determinants, and consequences of population ageing in India. LASI utilized a multistage clustering sampling design to obtain the data from Indian residents aged ≥45 years and their spouses (regardless of age). Participants provided written consent to participate in the survey. There were total of 70,865 individuals, out of which 28,952 (40.7%) were between 45-54 years, 28,765 (40.4%) were in the age group of 55-69 years, and 13,417 (18.9%) were ≥70 years, and 58% were females. LASI data were obtained after written request to IIPS [19].

Outcome variable
The outcomes of interest are dengue and CHIKV, and were based on the following questions: 1. In the past two years, have you had dengue? 2. Were you treated by health professionals for dengue? 3. In the past two years, have you had CHIKV? 4. Were you treated by health professionals for CHIKV? The options were 'Yes' and 'No'. Only those responses were considered 'Yes', wherein the respondent reported to have dengue/CHIKV and got treated by a health professional for the same. The untreated respondents were excluded from the analyses. The responses were coded as a binary variable ('0' for absence and '1' for presence-Dengue:620; CHIKV:1628).

Household variables
The household variables utilized are household size (1-5 or ≥6 members), type of house i.e. permanent (pucca) or temporary (kutcha), location of water source (own dwelling, yard/plot or outside dwelling), toilet-type (improved sanitation: ushed to piped sewer system/septic tank/pit latrine, pit latrine, twin pit, composting toilet; unimproved sanitation: open defecation), cooking fuel (clean fuel: LPG, biogas and electricity; unclean fuel: kerosene, charcoal, coal, crop residue, wood/shrub and dung cake) and having a damp wall or ceiling (yes/no).
Both the household and socio-economic variables were selected after prior literature search [20][21][22][23] Statistical analysis Frequency and percentage distribution tables were prepared for all the variables used in this study. Dengue and CHIKV prevalence with each of the housing conditions and SES variables were reported. The variables for multiple logistic regression analyses were chosen based on purposeful selection of variables [24]. Univariable analyses were conducted and predictors with <0.25 signi cance level were selected for the next-level analyses. From the model containing these variables, variables not signi cant at 0.10 level were removed one at a time, so that their effects on the odds ratios of other variables (15% change) could be assessed. Following this, variables eliminated (did not meet the criterion of 0.25 signi cant level) during the univariable analyses were added to the model and were retained if the predictors are signi cant at 0.1 level and their addition caused a change of 15% or more in the odds ratios for at least one of the categories of these variables. Hence, 'location of water source' (not signi cant in univariable analyses) remained in the analyses for both dengue and CHIKV as it changed the odds ratio of wealth quintiles by more than 15%. Collinearity was checked for the selected variables and vif (variance in ation factor) values were found to be <5. We have applied sampling weights computed by LASI during the data analysis to obtain accurate estimates. LASI individual and household datasets were merged to realize the study's objectives, and due to the merging of the datasets, few missing values were generated. However, missing values were less than 2%. The nal sample size considered for the analysis was 70,865. STATA MP statistical software version 16 was used for data analysis. The STATA Do le in txt format has been given as Additional le 1.

Spatial analysis
The State-wise prevalence (%) was used to visualize the spatial distribution of dengue and CHIKV using the ArcGIS software. The averaged prevalence of State and UT was analysed and grouped into four classes, two above and two below the national average. The choropleth technique was used for visualization where the darker hue was used to denote higher prevalence.

Prevalence of dengue and CHIKV in adults ≥45 years
The distribution of all the study variables is shown in Table 1. The period prevalence of dengue and CHIKV is 0.87% (95% CI 0.81-0.95%), and 2.3% (95% CI 2.19-2.41%) respectively. Figures 1-2 show the distribution of dengue and CHIKV across India's States and UT. Dengue is highly prevalent in the northern states of India, and the highest prevalence is observed in Delhi (5.6%), followed by, Chandigarh (3.1%), and Dadra & Nagar Haveli (3.1%), while Delhi (14%) and Haryana (7%) show the highest prevalence of CHIKV cases.
The prevalence of dengue and CHIKV associated with SES and housing factors in adults ≥45 years is shown in Table 2. Dengue and CHIKV are less in males (0.74% and 2.05%, respectively) compared to females (0.97% and 2.48%, respectively). Dengue and CHIKV are higher in the urban (1.12% and 2.92%, respectively) than in the rural areas (0.76% and 2.02%, respectively). ST had the lowest prevalence of dengue (0.55%) and CHIKV (1.18%), while the highest is seen in SC (1.12%) and the forward caste (2.95%) respectively. In the income category, a slightly higher prevalence is seen in the richest for both the diseases. College-educated respondents have the lowest prevalence of dengue (0.48%). The prevalence of both diseases is higher in pucca/semi-pucca houses. The prevalence of CHIKV is slightly high in households with no toilet facility, while it is the opposite in dengue. For both dengue and CHIKV, prevalence marginally increased in households using clean fuel. Damp walls or ceilings in households decreased the prevalence of CHIKV (2.07% vs 3.10%).

Relationship between the study variables (SES and housing) and dengue/CHIKV in adults ≥ 45 years
The odds ratios of dengue and CHIKV for the SES and housing variables are shown in Table 3. Residence in an urban area, adults in the age group of 45-54 years, wealth, education less than six school years, not working individuals increase the odds for dengue. Urban residents have 1.6 times (AOR :1.57; 95% CI: 1.17-2.10) higher odds for dengue than rural. The odds for dengue are highest in richest (AOR: 2.10; 95% CI: 1.33-3.31). Compared to illiterates (0 school years), the lowest risk (AOR: 0.24; 95% CI: 0.12-0.49) for dengue is seen in the college-educated. Households with water-source in the yard/plot have lesser odds (AOR: 0.55; 95% CI: 0.37-0.80, p value=0.02) for dengue.
Residence in an urban area, increasing MPCE quintiles, education with less than six school years, SC and forward castes, people in the pucca/semi pucca house, no water source within a dwelling, increase the odds for CHIKV. Urban residents have 1.6 times more odds (AOR: 1.56; 95% CI: 1.20-2.02) for CHIKV than rural residents. CHIKV odds are higher by 1.

Discussion
Dengue and CHIKV infections are often mild, and may be undiagnosed or misdiagnosed. Hence, we have only considered those who self-reported that they were treated for dengue or CHIKV. Dengue is the dominant vector-borne viral disease in India; population level serosurvey carried out in 2017-2018 (5-45 years) showed 48.7% seropositivity for dengue [7] vs. 18.1% for CHIKV [12]. Analysis of LASI data shows dengue and CHIKV prevalence to be 0.87% and 2.3%, respectively. Dengue is endemic in most States of India [5], and population level serosurvey carried out in 2017-2018 in the age group of 5-45 years has reported a seropositivity of 60.3%, 5%, 18.3%, 62.3% and 76.9% in the Northern, North-Eastern, Eastern, Western and Southern regions respectively, with an overall 48.7% seropositivity for India [7]. The low selfreported prevalence could be due to the high seropositivity across India, except for the North-Eastern and Eastern regions. The North Indian States of Delhi, Uttar Pradesh, Punjab and Haryana are the only ones to report ≥2% prevalence. Delhi is highly endemic for dengue, and multiple serotypes co-circulate [6].
Secondary infections resulting in severe dengue illness are known to occur due to the circulation of numerous serotypes [25], and may explain the highest self-reported prevalence (5.6%) in Delhi.
The high prevalence of CHIKV could be explained by the study period of the LASI survey. Even though, the LASI survey was carried out in 2017-18, the respondents were asked to self-report if they had the disease in the preceding two years. In 2016, there was a massive outbreak of CHIKV in North India [26,27]. The highest prevalence (>4%) of the self-reported CHIKV cases were in the northern States of Delhi, Uttar Pradesh, Haryana and Rajasthan. Even though, the population level serosurvey shows South India to have the highest seropositivity (43.1%) [12], the self-reported cases in the LASI survey are lower. The Southern States were the most affected in the CHIKV outbreak in 2005-06 [28 -30]. A multicentric hospital-based study carried out in 2008-2009 to detect CHIKV cases by RT-PCR and/ or IgM-ELISA reported highest positive cases in South India (49.36%), followed by West (16.28%), and the lowest was in North (0.56%) [31]. Prior exposure to CHIKV could explain the low self-reported prevalence rates in the South when compared to North India. The eastern States of Odisha and West Bengal, and the adjacent States of Bihar, Jharkhand and Chhattisgarh have <1% prevalence, and this overlaps well with the 4.4% seropositivity in the East [12]. Similarly, the prevalence was 0% in the North-East, and is in line with the 0.3% seropositivity in this region [12]. In line with the population level serosurvey data [12], LASI survey shows the Eastern and the North-Eastern region of India to have low prevalence of CHIKV, and are susceptible to future outbreaks.
Analysis of LASI data indicates urban residence, wealth, education and location of water-source to be the common risk factors for dengue and CHIKV in India. In addition, adults (45-54 years) are also at more risk for dengue, while for CHIKV, caste (SC and forward), pucca/semi-pucca house type are additional risk factors. Among the various factors of dengue transmission, urbanization, globalization and lack of effective vector control are considered to be the three major drivers [32]. Ae. aegypti, the primary driver of dengue and CHIKV lives in urban and peri-urban human habitation. In urban tropics, large swathes of human and Ae. aegypti population live in intimate association, and provide the perfect setting for the maintenance and generation of epidemic strains of vector-borne viruses [32,33]. In this analysis, urban residence increases the odds for both dengue and CHIKV. Positive association has been reported with dengue and CHIKV prevalence, and population density [34][35][36][37][38]. Even though dengue is present both in rural and urban India, incidence in urban areas are much higher; a nation-wide dengue serosurvey has recorded 70.9% (64·3-76·6) seropositivity in urban compared to 42.3% (36·0-48·9) in rural districts [7]. The urban incidence of CHIKV is even higher; 40·2% (31·7-49·3) in urban vs. 11·5% (8·8-15·0) in rural [12]. Ae. aegypti's breeding preferences coupled with population density makes urban areas a signi cant risk factor for vector-borne viral diseases in India. Among all the States and UT of India, the National Capital Territory of Delhi and the Union Territory of Chandigarh are most urbanized with 97.5% and 97.25% urban population respectively, followed by Daman and Diu at 75.2% [39]. Delhi shares borders with Haryana and Uttar Pradesh, and the urban expansion has accelerated in the border regions of these States [40]. Thus, this region has emerged as hot spot of dengue and CHIKV prevalence in the country. Even though Himachal Pradesh is bordering this hot spotregion, the level of urbanization in Himachal Pradesh is least (10%) in the country, and this could explain the low period prevalence of dengue and CHIKV. Overall, urbanization appears to be a major driver of dengue and CHIKV.
Population based national serosurveys show that the incidence of dengue and CHIKV increases with age; compared to 5-8 (Dengue: 28.3%; CHIKV: 9.2%) and 9-17 (Dengue: 41.0%; CHIKV: 14%), seropositivity is high in the 18-45 age group (Dengue: 56.2%; CHIKV: 21.6%) [7,12]. The age group more susceptible to clinical dengue infection varies among different geographical regions, and is in uenced by host immunity and the circulating viral genotypes. Epidemiology of the 2017 dengue outbreak in Sri Lanka show adults ≥ 50 years are least affected [41]. In Taiwan, dengue prevalence from 2010-2015 show signi cantly higher prevalence rates in adults ≥ 60 years [42]. Cyclical pattern of dengue epidemics driven by DENV-1 and DENV-2 serotypes have been observed in Singapore from 2004-2016; in DENV-2 predominant years (2007-12 and 2016), the incidence rate of dengue in 55+ age group is almost equal to the 15-24 years age group, while in DENV-1 predominant years (2004-2006 and 2013-2015), the incidence rate in 55+ years is about half [43]. In the 2007 epidemic in Brazil, there was a shift in the age pattern, with dengue hemorrhagic fever affecting predominantly children<15 (>53%), compared to 22.6% in 2001 [44]. For pan-India, reliable estimates of age-strati ed dengue caseloads are not available in the public domain. A nine year (2007-2015) dengue trend in Mumbai, western India shows dengue morbidity to be highest in young adults aged 21-40 years [45]. Analysis of the LASI data among the three age groups (45-54, 55-69 and ≥70) shows adults in the 45-54 years age group to have higher odds for dengue. One possible reason for the higher likelihood in this group could be their active life style related to employment, which would also make them travel frequently. A case-control study in Odisha, India shows the odds of dengue are three times higher in individuals whose work requires long travel [17].
Location of water source outside the house was found to increase the odds of both dengue and CHIKV. An individual-level cohort study carried out in Vietnam shows households that do not have access to tap water close to their dwelling have increased risk of dengue fever [46]. Lack of access to piped water supply will lead to households resorting to using containers for water-storage; these storage containers will provide the ideal breeding sites for mosquitoes resulting in increased dengue risk for the household [46]. A retrospective study carried out in Delhi has identi ed lack of access to tap water to be a key factor in dengue IgG seropositivity [15]. Lack of proper toilet facility in the household also increases the likelihood of CHIKV. Ae. aegypti's peak biting periods are early in the Morning, and in the period before dusk [47]; the need to use outside toilet facilities increases the likelihood of mosquito bites and vectorborne diseases.
Individuals with less than 6 years of schooling have higher odds of dengue and CHIKV. Several studies have shown the association between low education levels and dengue [20,48,49]. Education helps in understanding the etiology of the disease, mode of transmission, symptoms, treatment, prevention and control measures [23]. Wealthy households have higher odds of dengue and CHIKV. Also, residents in pucca houses have higher likelihood of getting infected with CHIKV. Possible reasons include: 1) wealth is likely to be positively associated with urban residence; both dengue and CHIKV have higher prevalence in densely populated urban settings in India, and 2) health seeking behaviour may be better in wealthy households. In Delhi, dengue burden was higher in wealthier districts despite lower mosquito load [15]. In contrast, low SES is shown to be a key risk factor of dengue in Brazil [21,22,48,50] and Cuba [51]. Unlike dengue hemorrhagic fever and dengue shock syndrome, dengue fever is self-limiting characterized by fever, myalgia, headache and constitutional systems [52]. The well-educated individuals from wealthy urban background are more likely to get diagnosed promptly compared to the lower socio-economic class, and this may have increased the odds of dengue and CHIKV in the former. Future studies in different SES settings of India should be carried out to better understand the association between SES and dengue/CHIKV incidence.

Conclusions
The major limitation of the study is that the data analyzed to understand the socio-economic and housing determinants are only from adults ≥ 45, therefore, it may not be appropriate to generalize these ndings to all age groups. As the disease is self-reported, only respondents with symptomatic infection who got diagnosed may have reported, and this would affect the accuracy of the prevalence estimates. Furthermore, as LASI is a cross-sectional survey, the association of socio-economic and household variables with dengue or CHIKV in this study does not imply causation. This is a secondary data analysis of LASI wave 1, conducted by IIPS. Hence, ethical approval is not applicable.

Consent for publication
Not applicable Availability of data and materials The datasets analyzed during the study are available after submitting a data request form to IIPS, https://www.iipsindia.ac.in/content/lasi-publications.

Competing Interests
The authors declare that they have no competing interests.  Legend not included with this version Legend not included with this version

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. AdditionalFile1.txt