Sustainable Cities and Society 37 (2018) 178–188

Contents lists available at ScienceDirect

Sustainable Cities and Society

journal homepage: www.elsevier.com/locate/scs

Urban adaptation to climate sensitive health effect: Evaluation of coping T

strategies for dengue in Delhi, India
Samraj Sahay
Department of Business Economics, University of Delhi, South Campus, New Delhi 110021, India

A R T I C L E I N F O A B S T R A C T

Keywords: In absence of adaptation initiative, residents resort to coping strategies to manage climate-related health risks.
Urban health adaptation Delhi, the capital city of India, is highly endemic to the climate sensitive vector-borne disease of dengue. The
Coping strategies weeks of a year with risk of dengue occurrence due to weather patterns are identified by using Poisson regression
Weather patterns and dengue which uses data for climate variables and reported dengue cases. The coping strategies adopted by household are
Empirical evaluation
then evaluated using survey data of 559 households. An economic evaluation of coping strategies to serve as an
Delhi city
important decision tool in developing adaptive capacity was carried out using household health production
function approach. The econometric tools of Negative Binomial Regression and probit regression have been used
for evaluation. The methodological framework provides a novel approach by linking climatic variability, health
impact and economic evaluation of adaptation strategies. The results reveal that the dengue risk is mainly
addressed by coping strategies at the household level and intervention measures are ineffective due to in-
appropriate timing and selective approach. The annual economic benefit of adopting coping strategies is esti-
mated as US$ 65 per household. The benefits of coping strategies in reducing dengue risk contribute to the
planning of adaptation strategies at city level.

1. Introduction aegypti mosquito, the principal urban vector (Reiter, 2001; Hales, de
Wet, Maindonald, & Woodward, 2002; Hopp & Foley, 2003; Morin,
As with most cities, Delhi, the capital city of India is vulnerable to Comrie, & Ernst, 2013; Ebi & Nealon, 2016). In Delhi, both Aedes aegypti
the health impacts of climate change. Among the climate sensitive in- and Aedes albopictus transmit dengue viruses with Aedes aegypti being
fectious diseases, the vector-borne disease (VBD) of dengue is most the most prevalent vector (Kumari et al., 2011, 2013; Sharma, Kumari,
prevalent, occurring regularly in Delhi. There has been a significant Srivastava, Barua, & Chauhan, 2014). Dengue is the only VBD having
increase in the number of confirmed cases in the last fifteen years with strong association with climate variables, both at the local as well as
outbreaks in 2003, 2006, 2010 and the worst in 2015 with a staggering global level (Smith et al., 2014). Incidence of dengue is exacerbated by
15867 dengue cases (Chakravarti & Kumaria, 2005; Nandi et al., 2009; increase in temperature, rainfall and humidity that alter the vector and
Kumari, Kumar, & Chauhan, 2011; NVBDCP, 2015; WHO, 2016). This pathogen life-cycle (Patz, Martens, Focks, & Jetten, 1998; Hales et al.,
makes interventions necessary to increase the adaptive capacity of the 2002; Hopp & Foley, 2001, 2003; Bhatt et al., 2013; Morin et al., 2013;
urban residents for future dengue risk. Appropriate adaptation strate- Naish et al., 2014; Estallo, Luduena-Almeida, Introini, Zaidenberg, &
gies would make the city resilient to future risk and contribute to Almiron, 2015; Ebi & Nealon, 2016). Increase in rainfall creates pools of
achieving the sustainable development goals (SDGs) of combating cli- water suitable for mosquito breeding. While increase in temperature
mate change impacts (SDG – Targets 13.1; 13.2 and 13.3), healthy lives shortens extrinsic incubation period (EIP) of the pathogens and in-
for all (SDG − Target 3.3), and resilient and sustainable cities (SDG – creases the biting and egg laying activities of the mosquitoes (Bradley,
Target 11.b) (SDSN, 2017). In India, the average economic burden of 1993; Costello et al., 2009), thereby increasing the vectoral capacity
dengue has been estimated to be US$ 29.3 million (Garg, Nagpal, (VC) of the mosquitoes (Reiter, 2001; Kovats, Campbell-Lendrum,
Khairnar, & Seneviratne, 2008; Conteh, Engels, & Molyneux, 2010). McMichael, Woodward, & Cox, 2001; Kramer & Ebel, 2003). High re-
Therefore, the effectiveness of interventions as adaptation strategies is lative humidity increases the life span of the mosquitoes, thereby en-
crucial in reducing the economic burden of the disease. abling them to infect more hosts (Patz et al., 2000).
Dengue is the most important mosquito-borne viral disease which is Recurrence of dengue in Delhi over the years shows that interven-
caused by four closely related viruses. It is mainly transmitted by Aedes tions have not been very effective and provides evidence of inadequate

E-mail address: [email protected].

https://doi.org/10.1016/j.scs.2017.11.017
Received 8 July 2017; Received in revised form 15 November 2017; Accepted 15 November 2017
Available online 21 November 2017
2210-6707/ © 2017 Elsevier Ltd. All rights reserved.
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Fig. 1. Interventions by the local government (MCD) and coping strategies by households.

planning. The present study makes an effort towards finding the reasons risk. Such evaluation identifies the effectiveness of coping in con-
for ineffectiveness of intervention measures and shortcomings in plan- tributing to urban resilience and determining the existing adaptation
ning. There is lack of evidence on the impact of weather patterns on deficiet (gaps in effectiveness of the intervention practices as a part of
dengue incidence and so far it has not been taken into account while public health policies and additional strategies which would be re-
planning interventions. Hence, no adaptation strategies have been quired specifically to minimize the risk). The paper aims at providing
planned for dengue risk due to weather patterns. For the first time, this the necessary policy inputs required for formulation of the adaptation
study analyses the effectiveness of the intervention measures for dengue strategies by local government. This is achieved by determining the
in Delhi by economic evaluation of household data for strategies impact of climate variables on dengue incidence and evaluating the
adopted by them to cope with dengue risk due to weather patterns. The coping strategies at the household level. The weeks of a year with high
methodology used in this paper links climatic variability, health impact probability (risk) of dengue occurrence due to weather patterns are
and coping strategies, thus, providing a novel framework for mon- identified by determining the relationship between climate variables
itoring and evaluation of adaptation strategies for climate related and dengue cases. This has been modelled using Poisson regression
health risk at city level. applying the generalized estimating equation (GEE) approach which
In most of the cities, in the absence of planned adaptation, residents takes care of overdispersion, serial correlation, non-linearity and allows
resort to coping strategies (private adaptation) which are autonomous, for hypothesis testing. Household coping strategies have been evaluated
reactive and with short-term benefits, practised at the individual, using the health production function (HPF) approach. A system of si-
household and community level but also with initiative by the private multaneous equations has been estimated using 2SRI (two-staged re-
sector (Revi et al., 2014; Ford et al., 2015; Araos et al., 2016). House- sidual inclusion) approach to deal with endogeneity. The econometric
holds in Delhi also practise coping strategies. However, repeated use of tools of Negative Binomial Regression (NBR) and probit regression are
coping strategies does not ensure resilience to future risk as it does not used for estimating the main and reduced form equations respectively.
contribute in building adaptive capacity, thereby making the residents
increasingly more vulnerable to the health hazards (Lavell et al., 2012;
Berman, Quinn, & Paavola, 2012; Revi et al., 2014). Coping strategies 2. Interventions and coping strategies
need to be supplemented by supportive measures by the local govern-
ment to build the adaptive capacity in the long-run (Revi et al., 2014). At the national level, a long term action plan was formulated by the
The existing coping strategies can be considered as building blocks for National Vector Borne Disease Control Programme (NVBDCP) in 2007
the adaptation strategies. This calls for evaluation of the coping stra- for control and prevention of dengue. It required the endemic states in
tegies adopted by households in Delhi. India including Delhi, to implement strategies for early case reporting
The economic evaluation of coping strategies serve as an important and management through establishment of sentinel surveillance sites,
decision tool for policy makers as it provide information on costs, vector management including anti-larval and anti-adult measures, and
benefits and effectiveness of coping strategies in reducing the dengue supporting interventions such as capacity building (NVBDCP, 2007).
Recently NVBDCP has also come up with operational guidelines specific

179
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

for control of VBDs in urban areas (NVBDCP, 2016). In Delhi, presently this study is based on the health production function approach as given
the intervention measures are being carried out by the local govern- in Deschenes and Greenstone (2011). The expression for willingness to
ment: Municipal Corporation of Delhi (MCD). The measures include pay/accept for the health impact of weather patterns can be expressed
early case management by the 34 designated sentinel surveillance in a measurable form using changes in the number of days of sickness
hospitals and 2 apex referral centres; vector management through in- (S) due to dengue and consumption of health-maintaining market goods
spection for potential breeding sites and presence of Ades larvae; use of XH having price pH. XH refers to coping strategies adopted to decrease
biolarvicide in major drains and large standing water bodies; and the probability of occurrence of dengue such as use of mosquito re-
thermal fogging in the affected areas. In addition to this different pellents, medicated nets, measures to prevent mosquito breeding, use of
media: print and electronic media; group interactions; personal contact; clothes that minimizes skin exposure and use of smoke to drive away
and public address system are used to spread awareness about dengue mosquitoes. The final equation for the marginal willingness to pay (See
(MCD, 2017). MCD also carries out cleaning of drains and garbage as Deschenes & Greenstone, 2011 for details) is given as
routine work which contribute to control of mosquito breeding.
dI* (C)/d C = −d S/d C (∂ U/∂S)/λ + pH (∂XH/∂C) (2)
The dengue control measures adopted by households, or the coping
strategies, mainly include strategies to prevent mosquito bites and the where λ is the Lagrange multiplier from the maximization problem or
breeding of mosquitoes. Bite prevention strategies include use of re- the marginal utility of income I. C refers to the weather patterns
pellents; both during day as well as at night, particularly in the dengue creating an environment conducive for exacerbation of the dengue and
affected areas of the city, as Aedes mosquito bites during the day time; is represented by the predicted probability of the dengue occurrence in
mosquito (medicated) net; wearing full body covering clothes; and Delhi. From Eq. (2) it can be deduced that willingness to pay/accept for
screens on windows and doors. Breeding prevention strategies include the health impact of weather patterns depends on the changes in S and
weekly emptying and drying of desert coolers; prevention of water XH. Depending on the exogenous factors, it is possible that there will be
accumulations around houses in empty containers; coconut shells, tyres a large change in the consumption of XH and little change in S.
and flower pots; cleaning and scrubbing of indoor flower vases, cov- In Eq. (2), the total derivatives of the S with respect to C, (dS/dC) is
ering of tanks for water storage; and cleaning of roof gutters and gar- independent of personal tastes and income and depends only on the
bage bins. Fig. 1 shows the intervention measures and coping strategies technical characteristics of the sickness (or dose-response) function and
specific to dengue being practised in Delhi. can be estimated without considering XH. (∂U/∂S)/λ is the dollar value
of the disutility of a change in S (time spent ill). The last term is the
3. Methododology partial derivative of XH with respect to C multiplied by the price of XH.
This gives us the estimate of changes in XH with changes in C. In this
3.1. Modeling framework study, full welfare effect cannot be estimated as it is very difficult to
measure disutility in monetary terms and cost of illness is independent
3.1.1. Impact of weather patterns on dengue incidence of XH. Hence, based on the objective of the study, only the benefit from
For this study, the impact of climate variables – maximum tem- change in XH with change in C has been measured.
perature (Tmax), minimum temperature (Tmin), rainfall and relative Thus, marginal willingness to pay (MWP) for health benefits from
humidity (RH) on dengue incidence have been modelled by Poisson reduction in number of days sick is the sum of observable reductions in
regression estimation using generalized estimating equation (GEE) ap- the cost of illness, cost of averting or coping activities and the monetary
proach (see Sahay, 2017 for details). Longitudinal or panel data of equivalent of disutility of illness. In this study, the MWP for a reduction
weekly dengue cases and climate variables are used as the objective is in time spent sick is calculated as the price of any averting or coping
to determine weeks of the year with probability of dengue occurrence activity divided by the marginal effect of the use of that −pH averting
(∂S / ∂XH)
due to the impact of climate variables. The number of confirmed or coping activity on time spent sick.
dengue cases is a count data. The GEE approach takes care of both the The estimation of MWP, (dI/dC) requires the estimation of health
issues of overdispersion and serial correlation associated with count production function (Eq. (3)) and the demand functions for XH (Eq. (4))
data (Liang & Zeger, 1986; Lu et al., 2009; Chen et al., 2010). The GEE, using household data. These are estimated as a system of simultaneous
also facilitates modelling of the non-linear relationship between dengue equations.
cases and climate variables (McCullagh & Nelder, 1989; Ballinger,
2004). The parameter estimated in GEE is essentially a multivariate
3.1.3. Econometric strategy
non-linear least square (Wooldridge, 2010). Following time lagged
The health production function equation is used to express the ac-
Poisson model is estimated to establish the relationship (Sahay, 2017)
tual time spent ill due to dengue. The equation measures the impact of
g {E (yij )} = '
x ij β , y ∼ Poisson weather patterns on the number of days ill and is given as
(1)
S = S (XH, C) (3)
for i = 1……., m and j = 1……., ni observation for each group or
cluster identifier i. g (.) is the log link function of the expectationE (yij ) Here, change in XH represent coping strategies with respect to the
where yij is the number of confirmed dengue cases with Poisson dis- change in C. XH is modelled as binary variable which takes a value of 1
tribution. xij used in the model are the weekly Tmax (°C), Tmin (°C), if coping action is taken and 0 otherwise. The basic model is
rainfall (mm) and RH (%). Factors other than the climate variables like
GDP per capita, population density or access to piped water may also XH = XH (I, C) (4)
influence dengue incidence (Colón-González, Fezzi, Lake, & Hunter, There are two major issues associated with the estimation of these
2013). In this study, the model is restricted to climate variables only as equations. Firstly, the discreteness and non-negativity of the dependent
there is almost no chance of any changes in these socio-economic de- variable, as number of days ill due to the dengue are count data and
terminants within weekly data (Sahay, 2017). The predicted probability hence Eq. (3) needs to be estimated using a count data model (Gerking
of dengue occurrence is obtained following estimation of Eq. (1). & Stanley, 1986; Gurmu, 1997; Mullahy, 1997; Deb & Trivedi, 1997;
Deb & Holmes, 2000; Manning & Mullahy, 2001; Shin & Moon, 2007;
3.1.2. Evaluation of the household coping strategies Gupta, 2008; Hidayat & Pokhrel, 2010; Richardson, Champ, & Loomis,
Model for evaluating the coping strategies is developed based on the 2012). Secondly, the variable for coping or averting measures in Eq. (3)
theory of utility maximizing consumer behaviour for estimating the is codetermined with the dependent variable leading to inconsistent
health-related welfare impact of weather patterns. The model used in and biased estimation due to endogeneity (Gerking & Stanley, 1986;

180
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Joyce, Grossman, & Goldman, 1989; Alberini, Eskeland, Krupnick, & variable is expected to have a negative effect on the number of days ill.
McGranahan, 1996; Bresnahan, Dickie, & Gerking, 1997; Dasgupta, Inclusion of the error term in main equation control endogeneity and its
2004; Dickie, 2005; Hidayat & Pokhrel, 2010; Richardson et al., 2012). inclusion serve as a simple diagnostic tool to statistically test en-
The coping activities are chosen by the respondents and are not exo- dogeneity. A significant βu; the parameter associated with the error
genous.
The number of days ill, used as the dependent variable in Eq. (3) is a ˆ
term x u , confirms endogeneity of the coping strategies xe (Stuart et al.,
2009). e2SRI is the random error. As the dependent variable is count
count data and has Poisson distribution. As with most of the health data so the tests for overdispersion and excessive zero is carried out to
economics applications, the theoretical assumption of Poisson regres- decide on the count model to be used.
sion of equidisperion is often violated (Gurmu, 1997; Deb & Trivedi,
1997; Shin & Moon, 2007). Also, there can be many respondents for 3.2. Description of the study area – Delhi
whom the number of days ill is zero. The presence of excessive zeros
leads to a condition different from simple overdispersion due to in- The National Capital Territory (NCT) of Delhi with an area of
creasing variance. Both these issues have been addressed by using NBR, 1483 km2 and population density of 11297 persons/km2 in 2011
a simple extension of the Poisson model (Cameron, Trivedi, Milne, & (Census of India, 2011) is situated in the subtropical belt of the Indian
Pigott, 1988). The decision in favour of NBR is based on the results of sub-continent between the latitudes of 28° 22′–28° 54′ N and 76°
the Likelihood ratio test and Vuong test for the presence of over- 48′–77° 23′ E, at an altitude of 216 m above mean sea level. Delhi has a
dispersion and zero inflation in the data. typical tropical climate with maximum temperature reaching up to
To deal with endogeneity in the non-linear framework, a two-stage 45 °C during the summer months of April, May and June. While, the
residual inclusion (2SRI) model suggested by Terza, Basu, and Rathouz minimum temperature falls to 2 °C during December and January. Delhi
(2008) has been used. The 2SRI is generally consistent in the broad receives an average annual rainfall of 614 mm and three-fourth of this
class of models to deal with endogenous regressors (Terza et al., 2008) is received during the month of July, August and September (Sahay,
and has been suggested for the count data models by Wooldridge (1997, 2017).
2002). It has been widely used in health economics research to deal The city is divided into 11 districts with 3 sub-divisions in each
with the problem of endogeneity in non-linear framework (DeSimone, district and is controlled by three local bodies- MCD (94% of the total
2002; Baser, Bradley, Gardiner, & Given, 2004; Gibson et al., 2006; area), New Delhi Municipal Council (NDMC) (3%) and Delhi
Norton & Houtven, 2006; Lindrooth & Weisbrod, 2007; Shea, Terza, Cantonment (3%). The MCD has control over the major parts of Delhi.
Stuart, & Briesacher, 2007; Shin & Moon, 2007; Stuart, Doshi, & Terza, The area controlled by MCD includes 12 zones and 272 municipal
2009; Hidayat & Pokhrel, 2010; Richardson et al., 2012). To deal with wards. These municipal wards constitute of a total 2334 colonies which
endogeneity bias, the relationship between the endogenous regressors have been categorized into 8 categories from A to H. The categorization
and the unobservable confounder latent variables (omitted variables) is based on the recommendations of the third Municipal Valuation
that influence the outcome variable is formalized using instrumental Committee (MVC-III) and has been done mainly for the purpose of
variables (IV). The choice of instrument is mostly subjective and should determining the property tax. For categorization of colonies see
meet the criteria of strong instrument. In the first-stage of the 2SRI, an Appendix A in Supplementary file.
auxiliary or reduced form regression (Eq. (4)) for the endogenous
variable is estimated. The residual from the first-stage is then included
3.3. Data used
as an additional regressor in the second-stage outcome model or the
main equation (Eq. (3)) (Terza et al., 2008; Hidayat & Pokhrel, 2010;
3.3.1. Climate variables and dengue cases
Richardson et al., 2012).
The only data of confirmed dengue cases consistent with this study
Firstly, the following reduced form equation (Eq. (4)) is estimated
made available by the Health Department, (MCD) was from January
using a probit regression for the endogenous binary regressor for the
2008 to May 2013.1 Number of weekly confirmed cases of dengue from
coping strategies −
January 2008 to May 2013 have been used in this study (number of
Xit = Φ (xoδo + w+αs) + xu + e (5) weeks n = 278). The weekly averages of the daily data for the Tmax,
Tmin, rainfall and dew point for the same period recorded at Safdar-
Where, xo is set of exogenous regressors which are same as those used in jung, New Delhi have been used to establish the relationship. The data
the outcome equation, with associated parameters δo. The exogenous was extracted from the Global Surface Summary of the Day (GSOD)
variable, the probability of monthly occurrence of dengue C is expected database, archived by National Climatic Data Centre (NCDC). Daily
to have positive effect on the coping strategies. w+ represents the in- average temperature and dew point were used to determine RH by
strument variables with associated parameters αs. xu are the un- applying the formula based on August-Rosche-Magnus approximation
observable confounder or latent variables having influence on the (see Alduchov & Eskridge, 1996; Lawrence, 2005).
outcome variable y and are also correlated with the endogenous vari-
able.
3.3.2. Household data
As xu cannot be determined, so it is substituted by the error term
To evaluate the coping strategies, household data collected through
ˆ
defined as x u from the above reduced form Eq. (5) and used as one of
the regressor in the main equation as given below
survey in the colonies which were most affected during the last two
years has been used. Household surveys were carried out from the last
week of January to first week of March 2014 in the areas selected based
ˆ
yi = exp(x eβe + x oβo + x u βu) + e2SRI (6)

Where, yi are the number of days any family member stays ill due to the
on the disease prevalence and the socio-economic status. Purposive
random stratified sampling strategy was used with recall period of one
dengue. xo is the observable exogenous regressors (observable con-
founders) with associated parameters βo. Here also the exogenous 1
The data for confirmed dengue cases is collected under the Integrated Disease
variable C is expected to have a positive effect on the number of days ill. Surveillance Programme (IDSP). It is mandatory for all the medical institutions to report
The exogenous variables also include other variables that might affect confirmed cases to MCD every week, which then passes it on to the national nodal agency,
the Directorate of National Vector-borne Disease Control Programme (NVBDCP).
the number of days ill. These include variables representing socio-
Reporting is done only if the patient with symptoms of dengue, tests positive with dengue-
economic and demographic characteristics, and surrounding environ- specific IgM capture ELISA (MAC-ELISA) in a single serum specimen. MAC-ELISA man-
ment of the household. xe is the binary endogenous regressor re- ufactured by National Institute of Virology (NIV), Pune, India having 96% sensitivity is
presenting coping strategies, with associated parameter βe. This supplied to all hospitals for this purpose (Nandi et al., 2009).

181
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

year. The most affected zones in the previous two years were con- using DFBETA technique (Sahay, 2017). The robustness of the result is
sidered as the areas in Delhi where the disease is most prevalent. This confirmed as there was no change in the sign and significance of the
led to the identification of 6 zones as sampling areas. In each of these estimated parameters after dropping of the influential cases.
zones, the wards with the highest number of colonies falling in cate- Tmin at a lag of 9, rainfall and RH at a lag of 5 weeks had significant
gories A to H were identified, so as to have colonies with variation in and positive effect on the weekly dengue cases. The lagged time effect
their socio-economic status for the household survey. In each of these of the climate variables in this study matches with the finding of the
selected wards, one colony with the maximum cases of dengue in the studies carried out in different countries (Hurtado-Dı́az et al., 2007; Lu
last one year, belonging to each category was then identified by en- et al., 2009; Chen et al., 2010). The predicted probabilities of dengue
quiring from the respective Primary Urban Health Centre/Dispensaries occurrence were obtained following estimation of the model. Fig. 4
and the ASHA (Accredited Social Health Activist). Effort was made to shows the monthly number of dengue cases in the households surveyed
select households in a way so that all parts of the colony were re- and the predicted probability of dengue occurrence in these months due
presented. At least one household without a dengue case in the last one to the impact of weather patterns. The result shows that the probability
year and within a radius of 100 m of the households with cases was of occurrence of dengue extends beyond monsoon. It extends from April
surveyed. This was done to ensure that both types of households were to December. In the households surveyed cases were reported even in
covered. The households without cases within a radius of 100 m were month of April.
selected as adult Ae. aegypti, the most prevalent vector of dengue in
Delhi can typically travel up to ∼100 m, spreading the infection 4.2. Evaluation of the coping strategies
(Reiter, Amador, Anderson, & Clark, 1995; Harrington et al., 2005;
Hemme, Thomas, Chadee, & Severson, 2010). There were 35 wards in In the 2SRI approach used for the estimation of the system of si-
the affected area and hence 35 colonies were selected. It was decided to multaneous equations, first, a reduced form equation (Eq. (5)) was es-
survey 20 households in each colony. Thus, a total of 700 households timated using probit regression. This was followed by the estimation of
were supposed to be surveyed spread across different categories of the main equation (Eq. (6)) which is a count data model. The rejection
colonies selected for the study. of the null (alpha = 0) in the likelihood ratio test that there is no
Response in some of the colonies was very poor as households in overdispersion with chibar2 = 184.56 (Pr > =chibar2 = 0.0000),
these colonies were either unwilling to respond to all the questions or confirmed the presence of overdispersion and favoured the use of ne-
completely refused to respond. Only 10% of the households in category gative binomial regression (NBR). The results of Vuong test with AIC
A and B and 53 and 60% in category C and D responded to the survey. correction (–1.75; p < 0.04) and BIC correction (−17.5; p < 0.00),
In rest of the categories, 100% of the households selected were sur- confirmed the selection of standard negative binomial regression over a
veyed. Categories A, B and C represents the high income colonies with zero-inflated negative binomial regression (Desmarais & Harden, 2013)
A being the wealthiest. Categories D and E represents middle income, for estimation of the main equation (Eq. (6)). The data of 559 house-
while F, G and H represent lower income colonies with H being the holds used in the analysis was arranged in a panel of 12 months to
poorest (see Appendix A in Supplementary file). Total 559 households capture the monthly variation in the coping strategies and probability
(79.86% of the total 700 households) were surveyed spread across 6 of dengue occurrence. Further, the result of the Breusch-Pagan La-
zones and 8 categories of colonies. Of this, 309 households (55.28% of grange Multiplier test (Chi2 (1) = 1.49, p > Chi2 = 0.2341) con-
the total households surveyed) reported at least one dengue case in the firmed that there is no panel effect in the data. Hence, data has been
household in the last one year. Fig. 2 shows the map of Delhi depicting pooled for the analysis.
the intensity of dengue cases in each ward in the 6 selected zones, while The results obtained by estimating Eqs. (5) and (6) are presented in
Fig. 3 shows the total number of households selected, surveyed and Table 2. The variables having prominent effect on coping strategies and
households with cases in each category of the colonies in the 6 selected number of days ill with dengue have been discussed here. Probit re-
zones. Demographic, socio-economic and surrounding characteristics of gression was used to estimate the reduced form equation (Eq. (5)) for
the households surveyed are given in Table 1. the binary endogenous regressor timing_repellent representing the coping
strategies in the main equation (Eq. (6)). Aedes mosquito typically bites
4. Results during day time. Timing of the use of repellents (timing_ repellent) turns
out to be the most important coping strategy against dengue. The
4.1. Impact of weather patterns on dengue incidence variable takes a value of 1 if the household uses repellent both during
day and night and 0 if used only during night. The exogenous variables
Most of the dengue cases in Delhi occur from July to November with are the same as used in the main equation. The result of multi-
a peak observed in October. The result of cross-correlation analysis collinearity diagnostics revealed no correlation between the regressors
between confirmed cases and the climate variables was used to de- as the average VIF was found to be 2.00. The monthly probability of
termine the lags for the climate variables to be used in estimation of Eq. dengue occurrence (C) had a significant and positive effect on the
(1). Weekly dengue cases were modelled using a lag of 13 weeks for household’s decision to adopt coping strategies. This provides strong
Tmax, 9 for Tmin, 5 for rainfall and 5 for RH. Lags ranging from 1 to 20 evidence supporting the fact that the household’s decision to adopt
weeks have been reported in such association studies (Hurtado-Dı́az, coping strategies represents the household’s response to the effect of the
Riojas-Rodrı́guez, Rothenberg, Gomez-Dantés, & Cifuentes, 2007; Lu weather patterns on dengue incidence in Delhi. Per capita income (pci)
et al., 2009; Chen et al., 2010; Colón-González, Lake, & Bentham, of the household, as expected has a significant and positive effect on
2011). Two separate models were estimated by dropping the regressors household’s decision to adopt coping strategies as households with
to avoid multicollinearity. The results are presented in Appendix B in higher income are likely to spend more on the coping strategies.
Supplementary file. The model with the lowest QICu value is con- Type of the household (household _type) and ownership (own-
sidered to be the best fit model (Pan, 2001; Lu et al., 2009; Sahay, ership_type) had a significant and negative effect. The probability of
2017). Model 1 with all the regressors had the lowest QICu value. households adopting coping strategies decreased if the households had
However, the result of multicollinearity diagnostics for model 1 re- regular wage/salary earnings and if living in a rented house. This might
vealed correlation between Tmax and Tmin. The average variance in- be due to relatively low pci in these households. Out of the total 261
flationary factor (VIF) for model 1 was found to be 11.99. On dropping households with regular wage/salary earnings, 38% had pci in the
Tmax in model 2, the average VIF was 1.55, indicating no multi- range 1 (0–4999) as compared to only 28% of the total 298 self-em-
collinearity. Hence, it is considered to be the best fit model. The ade- ployed households in the same range. Of the total 184 rented house-
quacy of the model was assessed by identifying the presence of outliers holds, 38% had pci in range 1(0–4999) as compared to only 27% falling

182
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Fig. 2. Number of households with dengue cases per ward in the 6 survey zones selected for this study. The most affected wards of Delhi in the sample were 1–Narela zone with 33 out of
40 households surveyed in category G and H colonies in two separate wards (15 households in Alipur; 18 in Kanjhawala) reporting dengue cases; 2–Civil Lines with 28 out of 40
households in category G and H colonies; 3–Shahdara North with 28 out of 40 households in category F and H colonies reporting dengue cases.

in the same range out of the total 375 self owned households. Among significant effect on the household’s decision to adopt coping strategies.
the surrounding environment variables, source of drinking water Average age (hh_avg_age) and years of formal education (hh_avg_edu) of
(drinking_water) and presence of open drain near the household (drai- the family members above 15 years, used as instrument variables, had a
nage_type) had a significant and positive effect. These households are significant and positive effect. This is primarily because the adult
subjected to more mosquito bites. Hence, the probability of such members of the household are responsible for taking the decision re-
households adopting coping strategies increases. Both storage options garding adopting coping strategies and level of education enables them
(water_storage) and presence of standing water bodies within a radius of to react in more responsible manner. As these two variables are sig-
100 m from the house (dist_standing_water) had a significant and nega- nificant it supports the choice of these variables as instrument (Stuart
tive effect. This is mainly due to their inability to afford expenses in- et al., 2009).
curred on coping strategies. The pci was in the range 1 (0–4999) for The main or the illness equation (Eq. (6)) was estimated using NBR.
80.53% of the total 114 households which stored water in indoor The multicollinearity diagnostics revealed no correlation between the
containers and 68.70% of the total 131 households with standing water regressors as the average VIF was found to be 2.13. The result shows
bodies within a radius of 100 m. that the coefficient for Xu is highly significant and positive which
The awareness level about dengue (den_awrns) and inspection of the confirms the endogeneity of binary regressor ‘timing_ repellent’ and at
households for the presence of mosquito larvae (inspection) had a po- the same time controls for the endogeneity due to the latent variable
sitive and significant effect. Awareness increased the knowledge about (Shea et al., 2007; Terza et al., 2008; Stuart et al., 2009). It indicates
the disease and inspection compelled them to adopt coping measures to that unknown factors that influence decision of the household to adopt
avoid penalty. While fogging (fogging), which is typically a reactive coping strategies also has a positive influence on the expected count of
measure carried out only after the cases have been reported, had no number of days ill. This means that households that are more likely to

183
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Fig. 3. Category-wise number of households se-

lected, surveyed and with dengue cases. Number of
cases reported increase as we move from category A
to H colonies or as the household income decreases.
Total Households Surveyed – 559, Household with
dengue cases (in 2013) in our sample – 309; Survey
Period – January 2014–March 2014.

have cases are also more likely to adopt coping strategies. In other from reduction in number of days ill. The detailed calculation is pre-
words, a household is more likely to take decision in favour of adopting sented in Appendix C in Supplementary file. To determine the incre-
‘coping strategies’ if it is vulnerable due to presence of conditions which mental effect of coping strategies on the number of days ill due to
favour disease occurrence. As expected, the binary endogenous re- dengue, the marginal effect of the variable for coping strategies ‘timing_
gressor for coping strategies ‘timing_ repellent’ and ‘pci’ had significant repellent’ in Eq. (6) is calculated with all other variables kept at their
and negative effect on the number of days ill due to the dengue. means. The marginal value of reduction in cost of illness due to re-
Households with higher per capita income are capable of spending duction in days of illness as a result of coping strategies is determined
more on coping strategies which reduces their vulnerability towards by calculating the cost of illness for the reduced number of days of
disease occurrence. illness. The total cost associated with the reduction in illness days is
The ‘ownership_type’, ‘drinking_water’, ‘water_storage’ and ‘drainage_- estimated by adding the marginal values of cost of coping strategies and
type’ in close proximity of the households had positive and significant the cost of illness which comes out to be US$ 622 per year or US$ 5.20
effect on the expected number of days ill as they provide conducive per month per household.
environment for breeding of the vector. Also, these variables had a Both, the coping strategies and the probability of dengue occurrence
negative influence on household’s decision to adopt coping strategies as do not remain same throughout the year and vary in different months.
such households had low income. The other surrounding environment The monthly probability of dengue had a significant and positive effect
variables of dist_standing_water and flower_pots had no significant effect both on the household’s decision to adopt coping strategies and on the
on the number of days ill. The level of awareness about dengue (de- expected number of days ill due to dengue. Thus, the total cost asso-
n_awrns) had a significant and negative effect on the expected number ciated with the reduction in illness days due to coping strategies will
of days ill. As the awareness level increases, households adopt more also show monthly variations. The actual monthly cost of reduction in
dengue specific coping measure of using repellent both during day and illness days is calculated by multiplying the monthly probability of
night which reduces the risk of dengue. dengue occurrence and the average monthly cost of reduction in illness
The association between illness and the intervention measures of days. The average monthly cost of reduction in illness thus obtained can
inspection and fogging carried out by the local government was found be considered as the WTP by the households for reduction of one day of
to be insignificant. The inspection was mostly carried out after cases illness due to dengue and it comes out to be US$ 5.44 per month per
had been confirmed or notified to the authorities. While fogging was household. The benefit of the households adopting coping strategies
carried out in the worst affected areas and was done in a selective comes out to be US$ 65 per household per year. This is statistically
manner in areas which reported cases. According to respondents, fog- different (z = 2.08, p < 0.05) from the benefit obtained earlier
ging was done only on the main roads and broader lanes. Hence, no without considering the monthly variation in the probability of dengue
consistent and immediate causal impact is found in the econometric occurrence.
relationship. The coefficient associated with C is highly significant and
positive indicating that the expected number of days ill due to dengue
5. Discussion
increases in the month in which probability of dengue occurrence in-
creases. This establishes the relationship between weather patterns,
The study uses HPF approach for evaluating coping strategies. The
dengue incidence and coping strategies after considering the effects of
approach serves as an empirical decision tool for planning adaptation
exogenous variables representing demographic and socio-economic
strategies by the local government as it provides the vital information
characteristics, awareness, interventions, surrounding environment and
on economic benefit and effectiveness of coping strategies in reducing
location of the households.
dengue morbidity at household level. It also facilitates the decision
maker in identifying the gaps in the interventions which are being
4.3. Benefits of coping strategies practised as a part of public health initiative. The information gener-
ated through the evaluation contributes to the appropriate planning of
Coping strategies (timing_ repellent) have a negative and significant
effect on the expected number of days ill due to dengue. This is used to 2
Indian rupee (INR) converted to US$. In the survey year (2014) US$ 1 equals INR 61
calculate the household’s willingness to pay (WTP) for health benefits (The World Bank, 2017)

184
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Table 1
Demographic, Socio-economic and surrounding characteristic of the households.

Variable Coding Number of householdsa Meanb Min–Max

Demographic and Socio-economic characteristic

Size of the Household – hh_size 559 4.53 (1.08) 1–10
Age of respondent –age_resp 559 37.29 (10.23) 15–79
Education of respondent (years of formal education) – edu_resp 0 year (not literate)–01 20 (3.58) 12.28 (3.60) 0–17
8 years (middle school) –02 70 (12.52)
10–12 years (Secondary-Senior 204 (36.49)
Secondary)–03
15 years (graduate)–04 222 (39.71)
17 years (postgraduate and above) −05 43 (7.69)
Gender of respondent – gen_respndt Male – 1 89 (15.92) – –
Female – 0 470 (84.08)
Average age of adult members of households (age > 15 yrs) – hh_avg_age
c
559 33.66 (6.24) 15–65.5
Average years of education of adult members of households (age > 15 yrs)d – 559 12.60 (2.88) 0–17
hh_avg_edu
Age of the patient – age_patient 309 23.22 (13.45) 2–64
Gender of the patient – gender Female – 1 169 (54.69) – –
Male – 0 140 (45.3)
Household type – hh_type Regular wage/salary earnings – 1 261 (46.77) – –
Self-employed, Casual Labour – 0 298 (53.22)
Type of ownership – ownership_type Rented – 1 184 (32.92) – –
Self-owned – 0 375 (67.08)
Monthly per capita household income (in INRe) – pci 0–4999 – 01 181 (32.38) 1.77 (0.62) 1–4
5000–19999 – 02 330 (59.03)
20000–49999 – 03 43 (7.69)
50000–99999 – 04 5 (0.89)
100000 and above – 05 –

Surrounding characteristics of the households

Type of structure –structure Semi-pucca – 1 6 (1.07) – –
Pucca – 0 553 (98.93)
Drinking Water source – drinking_water Tube-well/hand pump/Tanker – 1 100 (17.89) – –
Tap water – 0 459 (82.11)
Water Storage – water_storage Indoor–With & without lid containers – 1 114(20.36) – –
Overhead Tanks – 0 446 (79.64)
Type of drainage – Drainage_type Open kutcha/pucca – 1 322(57.6) – –
Covered pucca – 0 237(42.4)
Distance of Garbage dump from household – dist_garbge_dump Less than 100 m – 1 259(46.34) – –
Beyond 100 m – 0 300 (53.67)
Distance of Standing water bodies from household – dist_standing_water Less than 100 m – 1 131(23.44) – –
Beyond 100 m – 0 428 (76.56)
Flower pots and vases – flower_pots Present Indoor/outdoor – 1 217 (38.82) – –
No flower pots – 0 342 (61.18)

a
Number of the households corresponding to the codes of the variables mentioned. Numbers in parenthesis represent percentage of the total 559 households.
b
Numbers in parenthesis represent standard deviations.
c
Total individuals in 559 households – 2047, of which 523 (20.52%) were under 15 years of age and the remaining 2024 (79.48%) were above 15 years. Average age of the members
above 15 is given.
d
Average of number of years of formal education of the family members above 15 is given.
e
INR – Indian Rupee. In the survey year (2014) US$ 1 equals INR 61 (The World Bank, 2017).

intervention measures and helps developing adaptation strategies

which would be required specifically to minimize future risk. This
would build resilience and increase the adaptive capacity of the re-
sidents. The planning of adaptation strategies by the local government
would supplement the coping strategies by residents, reducing de-
pendency on coping alone and serve to lower the economic burden of
coping for residents.
The findings suggest significant influence of weather patterns on
dengue incidence in Delhi. Normally, dengue cases occur during mon-
soon (July–October). It starts in July, peaks in October and then starts
declining in November (Katyal, Singh, & Kumar, 1996; Katyal, Kumar,
Gill, & Sharma, 2003; Nandi et al., 2009). The predicted probability of
dengue occurrence due to weather patterns shows an intra-annual ex-
pansion in disease occurrence beyond monsoon with cases being re-
Fig. 4. Month-wise dengue cases reported by the households and predicted probability of ported even in month of April in the sample. The intra-annual expan-
dengue occurrence in Delhi. Probability of dengue occurrence is observed from April to sion has further become evident with considerable increase in dengue
December − not restricted to monsoon season only.
cases recorded in April 2017 (The Times of India 11th April, 2017). This
provides valuable information on the risk period during which the in-
tervention measures needs to be planned. Empirical analysis of the
survey reveals that the coping strategies adopted by the households do

185
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Table 2
Results of Probit and Negative binomial regression using 2SRI approach.

Dependent variable: Coping strategy (timing_ repellent) Dependent variable: number of days ill due to
− Probit regression dengue − NBR

Major Factors Independent Variables Coef. Robust Std. Err. Coef. Robust Std. Err.

Coping Strategy timing_ repellent −3.9152* 1.6763

Demographic-Socio-economic age_patient −0.0685* 0.0027 0.1320* 0.0407
age_patient_sqre 0.0009* 5.54E-05 −0.0014* 0.0005
pci 0.3643* 0.0401 −0.1812* 0.2486
household _type −0.0441 0.0369 0.1522 0.1753
ownership_type −0.2348* 0.0400 0.4037 0.2586
Awarenessa and Interventionsb den_awrns 0.5423* 0.0408 −0.7554* 0.2578
inspection 0.6443* 0.1104 1.4609 0.9980
fogging −0.1275 0.0840 0.2436 0.3308
Surrounding Environment drinking_water 0.1397** 0.0644 0.6165** 0.2828
water_storage −0.2622* 0.0619 0.8408* 0.2301
drainage_type 0.1307** 0.0584 0.6816* 0.2130
dist_standing_water −0.2061* 0.0507 0.2557 0.2494
flower_pots 0.0824** 0.0414 −0.1428 0.2099
Instrument hh_avg_age 0.0159* 0.0030
hh_avg_edu 0.0593* 0.0202
Location − Zone and Category Zone: Reference zone − South
Narela −0.3822* 0.0929 −1.2817* 0.3309
Rohini 0.1803* 0.0588 0.7358* 0.2988
Shahdara North 0.1964* 0.0710 0.0106 0.3162
Shahdara South 0.2122* 0.0695 0.5811*** 0.3380
Civil Lines 0.2526* 0.0580 1.1339*** 0.3555
Category (Reference category – F)
A 3.7392 159.14 −16.2323* 0.7322
B −0.3173 0.2456 −17.2365* 0.6522
C −0.2688* 0.0790 −3.2110* 0.3855
D −0.3647* 0.0769 −3.1744* 0.4041
E −0.2136* 0.0576 −2.1827* 0.2734
G −0.0780 0.0682 −1.6939* 0.2705
H −0.0292 0.0547 −1.3585* 0.2675
Weather Change C (Probability of dengue) 0.2015* 0.0468 6.9488* 0.4143
Latent variable Xu 3.3384** 1.6749
Constant −2.2233* 0.1892 −9.8075*
Dispersion parameters Ln α 3.2826
alpha 26.6457
Likelihood-ratio test of α = 0: 7018.79*
chibar2 01 c
Model Features N observations 6708 6708
Wald chi2 1601.11* 983.92*

‘*’, ‘**’ and ‘***’ indicates the significance at 1%, 5% and 10% level of significance respectively.
a
The awareness score ranged from 0 to 32 with following codes: 1 (0–8), 2 (10–18), 3 (20–26) and 4 (28–38).
b
Thermal fogging in the affected areas and household inspection for the detection of potential breeding sites and presence of mosquito larvae were the only intervention measures
carried out by the local government – MCD, specifically for control and prevention of dengue in 2013–2014, the year of survey and a year before that.
c
Rejection of the null that α = 0, indicates overdispersion and supports Negative Binomial model.

have a significant impact on reducing dengue incidence, intervention in the range 1 (score ranged from 0 to 8) and 2 (score ranged from 10 to
measures by the local government failed to achieve the desired goals, 18). Less than 30% of the total households surveyed in categories E, F,
while income levels acts as a major factor influencing household’s de- G and H confirmed fogging. The low awareness level along with lack of
cision to adopt coping strategies. Increase in awareness level of fogging and inspections in these low income colonies directly indicates
households prompted them to adopt dengue specific coping strategies the negligence of the local government in carrying out interventions in
which reduced the chance of dengue occurrence. The intervention these colonies. Further, low income prevented households from
measures of inspection and fogging by the local government were in- adopting coping strategies. The indoor water storage facilities and
effective in reducing dengue incidence. This is mainly due to in- standing water bodies, typically found in low income colonies nega-
appropriate timing and selective manner of carrying out the interven- tively influenced their decision as they could not afford coping strate-
tions. Fogging was done only after the cases have been reported in a gies. These findings highlight that the low income colonies are the most
colony and was restricted only to broader lanes. vulnerable section of the city which deserve major attention by the
The per capita income ‘pci’ of the households turns out to be one of policy makers while framing the adaptation measures.
the most significant limiting factors for coping strategies. The high per The outcome of the study provides significant input for planning
capita income makes the coping measures affordable leading to re- adaptation strategies for the dengue. Inappropriate timing of the in-
duction in number of days ill due to dengue. Households in the low terventions is one of the major reasons for ineffectiveness of these
income colonies were most affected by dengue. Of the total 120 measures. The study finds that the predicted probability of dengue
households surveyed each in categories G and H, 70% households in occurrence due to weather patterns extends from April to December.
category G and 72.5% households in category H reported cases, while This is very crucial for planning the exact timing of the intervention
66 and 60% households in categories G and H respectively, had per measures. Considering the lag effects of Tmin, rainfall and RH ranging
capita household income in the range 1 (0–4999). These two categories between 5 and 9 weeks, inspection should be initiated at least five
also had the highest number of households with lowest awareness score weeks prior to the start of weeks with probability of dengue occurrence.

186
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Accordingly it should be carried out from the month of March to References

November at regular intervals. Fogging should be carried out from last
week of March to December at regular intervals to prevent adult Aedes Alberini, A., Eskeland, G. S., Krupnick, A., & McGranahan, G. (1996). Determinants of
from transmitting the disease. These interventions should include all diarrheal disease in Jakarta. Water Resources Research, 32, 2259–2269.
Alduchov, O. A., & Eskridge, R. E. (1996). Improved Magnus form approximation of sa-
categories of the colonies covering all the households irrespective of the turation vapour pressure. Journal of Applied Meteorology and Climatology, 35,
ease of accessibility. The planning should focus more on low income 601–609.
colonies as they were found to be worst affected due to poor inter- Araos, M., Berrang-Ford, L., Ford, J. D., Austin, S., Biesbroek, R., & Lesnikowski, A.
(2016). Climate change adaptation planning in large cities: A systematic global as-
vention measures, unaffordability of coping strategies and low aware- sessment. Environmental Science and Policy. http://dx.doi.org/10.1016/j.envsci.2016.
ness level. Also, to avoid storing water for longer period which serves as 06.009.
the breeding sites, it should be ensured that there is regular water Ballinger, G. A. (2004). Using generalized estimating equations for longitudinal data
analysis. Organizational Research Methods, 7, 127.
supply, the network of piped water service is extended and water tan- Baser, O., Bradley, C. J., Gardiner, J. C., & Given, C. (2004). Testing and correcting for
kers reach in colonies without piped water supply on a regular basis. non-random selection bias due to censoring: An application to medical costs. Health
Services and Outcomes Research Methodology, 4, 93–107.
Berman, R., Quinn, C., & Paavola, J. (2012). The role of institutions in the transformation
6. Conclusion
of coping capacity to sustainable adaptive capacity. Environmental Development, 2,
86–100.
The evaluation of the coping strategies for dengue at household Bhatt, S., Gething, P. W., Brady, O. J., Messina, J. P., Farlow, A. W., Moyes, C. L., et al.
level confirm the impact of weather patterns on dengue incidence and (2013). The global distribution and burden of dengue. Nature, 496(7446), 504–507.
http://dx.doi.org/10.1038/nature12060.
at the same time demonstrate the benefits of coping strategies. It also Bradley, D. J. (1993). Human tropical diseases in a changing environment. Environmental
provides strong evidence that the intervention measures by the local change and human health, Ciba foundation symposium, Vol. 175, Chichester, UK:
government did not achieve the desired result. The finding suggests that Wiley146–170.
Bresnahan, B. W., Dickie, M., & Gerking, S. (1997). Averting behavior and urban air
the health impact of weather patterns in Delhi is being addressed pollution. Land Economics, 73(3), 340–357.
mainly by the coping strategies at the household level. The ineffec- Cameron, A. C., Trivedi, P. K., Milne, F., & Pigott, J. A. (1988). Microeconometric model
tiveness of the intervention measures in reducing dengue incidence as of the demand for health care and health insurance in Australia. Review of Economic
Studies, 55, 85–106.
revealed by the empirical exercise suggests poor preparedness or lack of Census of India (2011). Office of the Registrar General, Government of India, New Delhi,
planned adaptation. The study identifies the most vulnerable areas of India. http://censusindia.gov.in/2011-prov-results/prov_data_products_delhi.html.
the city and provides valuable information on timing of the year when Chakravarti, A., & Kumaria, R. (2005). Eco-epidemiological analysis of dengue infection
during an outbreak of dengue fever, India. Virology Journal, 2, 32. http://dx.doi.org/
interventions should be carried out. It precisely highlights the measures 10.1186/1743-422X-2-32.
which would eliminate breeding sites in and around households. Chen, S. C., Liao, C. M., Chio, C. P., Chou, H. H., You, S. H., & Cheng, Y. H. (2010). Lagged
Avoiding breeding sites would ensure minimizing the risk of dengue. temperature effect with mosquito transmission potential explains dengue variability
in southern Taiwan: Insights from a statistical analysis. Science of the Total
The present study is an effort to fill the gap in adaptation research at
Environment, 408, 4069–4075.
the city level. It provides a framework for determining the impact of Colón-González, F. J., Lake, I. R., & Bentham, G. (2011). Climate variability and dengue
weather patterns on dengue incidence which is essential for framing fever in warm and humid Mexico. American Journal of Tropical Medicine and Hygiene,
adaptation strategies. The assessment of the intervention measures 84(5), 757–763. http://dx.doi.org/10.4269/ajtmh.2011.0-0609.
Colón-González, F. J., Fezzi, C., Lake, I. R., & Hunter, P. R. (2013). The effects of weather
carried out by the local government and the coping strategies practised and climate change on dengue. PLOS Neglected Tropical Diseases, 7(11), e2503. http://
at the household level for dengue provides insights and policy inputs to dx.doi.org/10.1371/journal.pntd.0002503.
the public health agencies entrusted with reducing the risks and in- Conteh, L., Engels, T., & Molyneux, D. H. (2010). Socioeconomic aspects of neglected
tropical diseases. The Lancet, 375, 239–247.
cidence of the disease. It thereby helps to achieve community and Costello, A., Abbas, M., Allen, A., Ball, S., Bell, S., Bellamy, R., et al. (2009). Managing the
household level adaptation to such climate related vulnerabilities. The health effects of climate change. The Lancet, 373, 1693–1733.
framework used in this study provides a novel approach for measuring Dasgupta, P. (2004). Valuing health damages from water pollution in urban Delhi, India:
A health production function approach. Environment and Development Economics, 9,
the effectiveness of adaptation strategies and planning of alternative 83–106.
strategies for climate related health effects at city level. DeSimone, J. (2002). Illegal drug use and employment. Journal of Labor Economics, 20,
952–977.
Deb, P., & Holmes, A. M. (2000). Estimates of use and costs of behavioural health care: A
Funding
comparison of standard and finite mixture models. Health Economics, 9(6), 475–489.
Deb, P., & Trivedi, P. K. (1997). Demand for medical care by the elderly: A finite mixture
This research did not receive any specific grant from funding approach. Journal of Applied Econometrics, 12(3), 313–336.
Deschenes, O., & Greenstone, M. (2011). Climate change, mortality, and adaptation:
agencies in the public, commercial, or not-for-profit sectors.
Evidence from annual fluctuations in weather in the U.S. American Economics Journal:
Applied Economics, 3(4), 152–185.
Conflicts of interest Desmarais, B. A., & Harden, J. J. (2013). Testing for zero inflation in count models: Bias
correction for the Vuong test. The Stata Journal, 13(4), 810–835.
Dickie, M. (2005). Parental behavior and the value of children’s health: A health pro-
None. duction approach. Southern Economic Journal, 71, 855–872.
Ebi, K. L., & Nealon, J. (2016). Dengue in a changing climate. Environmental Research, 15,
Acknowledgements 115–123.
Estallo, E. L., Luduena-Almeida, F. F., Introini, M. V., Zaidenberg, M., & Almiron, W. R.
(2015). Weather variability associated with Aedes (stegomyia) aegypti (dengue vector)
The author wishes to acknowledge and thank Dr. Yamini Gupt, oviposition dynamics in northwestern Argentina. PLoS One, 10(5), e0127820.
University of Delhi, New Delhi and Prof. Purnamita Dasgupta, Institute Ford, J., Berrang-Ford, L., Biesbroek, R., Araos, M., Austin, S., & Lesnikowski, A. (2015).
Adaptation tracking for a post-2015 climate agreement. Nature Climate Change, 5,
of Economic Growth, Delhi, India for their guidance, comments and 967–969.
suggestions. The author appreciates and thanks officials of the Health Garg, P., Nagpal, J., Khairnar, P., & Seneviratne, S. L. (2008). Economic burden of dengue
Department of the South Delhi Municipal Corporation for providing the infections in India. Transactions of the Royal Society of Tropical Medicine and Hygiene,
102, 570–577.
data. The author is also thankful to the unknown reviewers of this paper
Gerking, S., & Stanley, L. R. (1986). An economic analysis of air pollution and health: The
for their extremely valuable and constructive suggestions and com- case of St. Louis’. The Review of Economics and Statistics, 68, 115–121.
ments. Gibson, T. B., Mark, T. L., Axelsen, K., Baser, O., Rublee, D. A., & McGuigan, K. A. (2006).
Impact of statin co-payments on adherence and medical care utilization and ex-
penditures. American Journal of Managed Care, 12, SP11–SP19.
Appendix A. Supplementary data Gupta, U. (2008). Valuation of urban air pollution: A case study of Kanpur City in India.
Environmental and Resource Economics, 41, 315–326.
Supplementary data associated with this article can be found, in the Gurmu, S. (1997). Semi-parametric estimation of hurdle regression models with an ap-
plication to medicaid utilization. Journal of Applied Econometrics Special Issue:
online version, at http://dx.doi.org/10.1016/j.scs.2017.11.017.

187
S. Sahay Sustainable Cities and Society 37 (2018) 178–188

Econometric Models of Event Counts, 12(3), 225–234. Family Welfare http://www.nvbdcp.gov.in/Doc/OPERATIONAL-GUIDELINES-FOR-

Hales, S., de Wet, N., Maindonald, J., & Woodward, A. (2002). Potential effect of popu- URBAN-VBD-SCHEME-2016.pdf. [Accessed on 19 September 2017].
lation and climate changes on global distribution of dengue fever: An empirical Naish, S., Dale, P., Mackenzie, J. S., McBride, J., Mengersen, K., & Tong, S. (2014).
model. Lancet, 360(9336), 830–834. Climate change and dengue: A critical and systematic review of quantitative mod-
Harrington, L. C., Scott, T. W., Lerdthusnee, K., Coleman, R. C., Costero, A., Clark, G. G., elling approaches. BMC Infectious Diseases, 14, 167. http://www.biomedcentral.com/
et al. (2005). Dispersal of the dengue vector Aedes aegypti within and between rural 1471-2334/14/167.
communities. American Journal of Tropical Medicine and Hygiene, 72, 209–220. Nandi, J., Sharma, R. S., Dasgupta, R. K., Katyal, P., Dutta, P. K., & Dhillon, G. P. S.
Hemme, R. R., Thomas, C. L., Chadee, D. D., & Severson, D. W. (2010). Influence of urban (2009). Epidemiological analysis of hospitalized cases of dengue fever/dengue hae-
landscapes on population dynamics in a short-distance migrant mosquito: Evidence morrhagic fever and extent of breeding of Aedes aegypti in major hospitals in the
for the dengue vector Aedes aegypti. PLOS Neglected Tropical Diseases, 4(3), e634. National Capital Territory of Delhi (NCT Delhi), 2005–2009. Dengue Bull, 33,
http://dx.doi.org/10.1371/journal.pntd.0000634. 130–139.
Hidayat, B., & Pokhrel, S. (2010). The selection of an appropriate count data model for Norton, E. C., & Houtven, C. H. (2006). Intervivos transfers and exchange. Southern
modelling health insurance and health care demand: Case of Indonesia. International Economic Journal, 73, 157–172.
Journal of Environmental Research and Public Health, 7, 9–27. http://dx.doi.org/10. Pan, W. (2001). Akaike's information criterion in generalized estimating equations.
3390/ijerph7010009. Biometrics, 57, 120–125.
Hopp, M. J., & Foley, J. A. (2001). Global-scale relationships between climate and the Patz, J. A., Martens, W. J. M., Focks, D. A., & Jetten, T. H. (1998). Dengue fever epidemic
dengue fever vector, Aedes aegypti. Climatic Change, 48, 441–463. potential as projected by general circulation models of global climate change. Environ
Hopp, M. J., & Foley, J. A. (2003). Worldwide fluctuations in dengue fever cases related Health Perspective, 106, 147–153.
to climate variability. Climate Research, 25, 85–94. Patz, P. A., McGeehin, M. A., Bernard, S. M., Ebi, K. L., Epstein, P. R., Grambsch, A., et al.
Hurtado-Dı́az, M., Riojas-Rodrı́guez, H., Rothenberg, S. J., Gomez-Dantés, H., & Cifuentes, (2000). The potential health impacts of climate variability and change for the United
E. (2007). Impact of climate variability on the incidence of dengue in Mexico. States: executive summary of the report of the health sector of the U.S. National
Tropical Medicine and International Health, 12(11), 1327–1337. Assessment. Environment and Health Perspective, 108, 367–376.
Joyce, T., Grossman, M., & Goldman, F. (1989). An assessment of the benefits of air Reiter, P., Amador, M. A., Anderson, R. A., & Clark, G. G. (1995). Short report: Dispersal
pollution control: The case of infant health. Journal of Urban Economics, 25, 32–51. of Aedes aegypti in an urban area after blood feeding as demonstrated by rubidium-
Katyal, R., Singh, K., & Kumar, K. (1996). Seasonal variations in Aedes aegypti population marked eggs. American Journal of Tropical Medicine and Hygiene, 52, 177–179.
in Delhi, India. Dengue Bulletin, 20, 78–81. Reiter, P. (2001). Climate change and mosquito-borne disease. Environ Health Perspective,
Katyal, R., Kumar, K., Gill, K. S., & Sharma, R. S. (2003). Impact of intervention measures 109, 141–161.
on DF/DHF cases and aedes aegypti indices in delhi, India: An update, 2001. Dengue Revi, A., Satterthwaite, D. E., Aragón-Durand, F., Corfee-Morlot, J., Kiunsi, R. B. R.,
Bulletin, 27, 163–167. Pelling, M., et al. (2014). Urban areas. In C. B. Field, W. R. Barros, D. J. Dokken, K. J.
Kovats, R. S., Campbell-Lendrum, D., McMichael, A., Woodward, A., & Cox, J. (2001). Mach, M. D. Mastrandrea, T. E. Bilir, M. Chatterjee, K. L. Ebi, Y. O. Estrada, R. C.
Early effects of climate change: Do they include changes in vector-borne disease? Genova, B. Girma, E. S. Kissel, A. N. Levy, S. MacCracken, P. R. Mastrandrea, & L.
Philosophical Transaction of the Royal Society of London B, 356, 1057–1068. White (Eds.). Climate change 2014: Impacts, adaptation, and vulnerability. part A: Global
Kramer, L. D., & Ebel, G. D. (2003). Dynamics of flavivirus infection in mosquitoes. and sectoral aspects. contribution of working group II to the fifth assessment report of the
Advances in Virus Research, 60, 187–232. intergovernmental panel of climate change (pp. 535–612). Cambridge, United Kingdom
Kumari, R., Kumar, R., & Chauhan, L. S. (2011). First dengue virus detection in Aedes and New York, NY, USA: Cambridge University Press.
albopictus from Delhi, India: Its breeding ecology and role in dengue transmission. Richardson, L. A., Champ, P. A., & Loomis, J. B. (2012). The hidden cost of wildfires:
Tropical Medicine and International Health, 16(8), 949–954. http://dx.doi.org/10. Economic valuation of health effects of wildfire smoke exposure in Southern. Journal
1111/j.1365-3156.2011.02789.x. of Forest Economics, 18, 14–35.
Kumari, R., Sharma, R. S., Kumar, K., Singh, P., Krishnan, S., Dash, A. P., et al. (2013). Sustainable Development Solutions Network (2017). Targets –Indicators and a monitoring
Mapping of dengue vectors and dengue virus activity in Delhi during 2011–2012. framework http://indicators.report/targets/. [Accessed on 23rd May, 2017].
Dengue Bulletin, 37, 87–100. Sahay, S. (2017). Climatic variability and dengue risk in urban environment of Delhi
Lavell, A., Oppenheimer, M., Diop, C., Hess, J., Lempert, R., Li, J., et al. (2012). Climate (India). Urban Climate. http://dx.doi.org/10.1016/j.uclim.2017.10.008.
change: new dimensions in disaster risk, exposure, vulnerability, and resilience. In C. Sharma, R. S., Kumari, R., Srivastava, P. K., Barua, K., & Chauhan, L. S. (2014).
B. Field, V. Barros, T. F. Stocker, D. Qin, D. J. Dokken, K. L. Ebi, M. D. Mastrandrea, K. Emergence of dengue problem in India –A public health challenge. Journal of
J. Mach, G. K. Plattner, S. K. Allen, M. Tignor, & P. M. Midgley (Eds.). Managing the Communicable Diseases, 46(2), 17–45.
risks of extreme events and disasters to advance climate change adaptation. a special report Shea, D., Terza, J., Stuart, B., & Briesacher, B. (2007). Estimating the effects of pre-
of working groups I and II of the intergovernmental panel on climate change (IPCC) (pp. scription drug: Coverage for medicare beneficiaries. Health Services Research, 43,
25–64). Cambridge, UK, and New York, NY, USA: Cambridge University Press. 933–949.
Lawrence, M. G. (2005). The relationship between relative humidity and the dewpoint Shin, J., & Moon, S. (2007). Do HMO plans reduce health care expenditure in the private
temperature in moist air a simple conversion and applications. American sector. Economic Inquiry, 45, 82–99.
Meteorological Society. http://dx.doi.org/10.1175/BAMS-86-2-225. Smith, K. R., Woodward, A., Campbell-Lendrum, D., Chadee, D. D., Honda, Y., Liu, Q.,
Liang, K. Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear et al. (2014). Human health: impacts, adaptation, and co-benefits. In C. B. Field, V. R.
models. Biometrika, 73, 13–22. Barros, D. J. Dokken, K. J. Mach, M. D. Mastrandrea, T. E. Bilir, M. Chatterjee, K. L.
Lindrooth, R. C., & Weisbrod, B. A. (2007). Do religious nonprofit and for-profit orga- Ebi, Y. O. Estrada, R. C. Genova, B. Girma, E. S. Kissel, A. N. Levy, S. MacCracken, P.
nizations respond differently to financial incentives? The hospice industry. Journal of R. Mastrandrea, & L. L. White (Eds.). Climate change 2014: Impacts, adaptation, and
Health Economics, 26, 342–357. vulnerability. part A: Global and sectoral aspects. contribution of working group II to the
Lu, L., Lin, H., Tian, L., Yang, W., Sun, J., & Liu, Q. (2009). Time series analysis of dengue fifth assessment report of the intergovernmental panel on climate change (pp. 709–754).
fever and weather in Guangzhou, China. BMC Public Health, 9, 395. Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press.
Municipal Corporation of Delhi (MCD) (2017). http://mcdonline.gov.in/tri/sdmc_ Stuart, B. C., Doshi, J. A., & Terza, J. V. (2009). Assessing the impact of drug use on
mcdportal/healthindex.php. [Accessed on 19 September 2017]. hospital costs. Health Research and Educational Trust. http://dx.doi.org/10.1111/j.
Manning, W. G., & Mullahy, J. (2001). Estimating log models: To transform or not to 1475-6773.2008.00897.x.
transform? Journal of Health Economics, 20, 461–494. Terza, J., Basu, A., & Rathouz, P. (2008). Two-stage residual inclusion estimation: ad-
McCullagh, P., & Nelder, J. A. (1989). Generalized linear models (2nd ed.). London: dressing endogeneity in health econometric modeling. Journal of Health Economics,
Chapman and Hall. 27, 531–543.
Morin, C. W., Comrie, A. C., & Ernst, K. C. (2013). Climate and dengue transmission: The Times of India, 11th April 2017. https://timesofindia.indiatimes.com/city/delhi/
Evidence and implications. Environmental Health Perspectives, 121, 1264–1272. delhi-chikungunya-dengue-arrive-big-early-this-year/articleshow/58118536.cms.
http://dx.doi.org/10.1289/ehp.1306556. [Accessed on 19th November, 2017].
Mullahy, J. (1997). Instrumental-variable estimation of count data models: Applications The World Bank (2017). The world bank database. https://data.worldbank.org/indicator/
to models of cigarette smoking behavior. Review of Economics and Statistics, 79, PA.NUS.FCRF?end=2016&locations=IN&name_desc=false&start=1960&view=
586–593. chart. [Accessed on – 22nd September 2017].
National Vector Borne Disease Control Programme (2007). Long term action plan for pre- World Health Organization (2016). Dengue and severe dengue, fact sheet. http://www.who.
vention and control of dengue and chikunguniya. Directorate General of Health Services, int/mediacentre/factsheets/fs117/en/. [Accessed on– 22nd October 2016].
Ministry of Health & Family Welfare http://www.nvbdcp.gov.in/Doc/Final_long_ Wooldridge, J. M. (1997). Quasi-likelihood methods for count data. In M. Pesaran, & P. B.
term_Action_Plan%20.pdf. [Accessed on 19 September 2017]. Schmidt (Eds.). Handbook of applied econometrics, volume II: Microeconometrics.
National Vector Borne Disease Control Programme (2015). Directorate general of health Malden, MA: Publishers Ltd.
services, ministry of health & family welfare. http://www.nvbdcp.gov.in/den-cd.html. Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge,
[Accessed on 29 December 2015]. MA: The MIT Press.
National Vector Borne Disease Control Programme (2016). Operational guidelines for urban Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data.
vbds scheme (2016). Directorate General of Health Services, Ministry of Health & Massachusetts London: The MIT Press Cambridge.

188