Exploring Socio-Economic Impact of Dengue Fever in Dhaka City: A Statistical Modeling Approach

There was a sudden increase in dengue affected people in Dhaka city during 2018 and 2019, considering the seriousness of the disease, this study attempted to investigate the socio-economic impact of dengue fever instead of its biological significance. The study considered a primary dataset of 235 affected and 235 unaffected participants from Dhaka city. The impact of dengue infection on the monthly expenditure of the patient was determined by the multiple linear regression models. The impact of Dengue on the human productivity of the respondents was assessed by another multiple linear regression model; the dependent variable absence (number of days absent from work) was applied as a proxy for measuring the productivity of the patient. Moreover, an important objective was to find out potential determinants of dengue in Dhaka city. Binary logistic regression applied for detecting the factors which were responsible for occurring dengue disease. The study found no significant association of family cost with dengue incidence but the loss of productivity turned out as statistically significant. People who lived alone were identified to experience the disease more, which might occur due to their insincerity about this disease. So, living alone persons need to increase their consciousness considering the seriousness of this disease. It was highly recommended by respondents to use mosquito repellent and net during sleeping, changing the water regularly from plant container, providing regular mosquito spray, and developing a drainage system in Dhaka city.


INTRODUCTION:
Contagious diseases have emerged as a main indicatory factors of poverty and poor health status in developing and under-developed countries (Kumar et al., 2007). In 2018 & 2019, dengue viral epidemic in Bangladesh affected the ability and strength of working people to carry on with day to day activities. In many countries, dengue is a public health problems (Okanurak et al., 1997). The term "Dengue" was first originated in Zanzibar naming "Denga" during 1870 epidemic (Mahmood & Mahmood, 2011). The dengue virus usually spreads by day biting female Aedes mosquitoes, primarily Aedes aegypti and Aedes albopictus (Sharmin et al., 2015). Climate change will increase severity of dengue in the future (Naish et al., 2014). lack medical documentation system and modern diagnostic facilities, continuous surveillance for dengue disease is a vast challenge here (Organization, 2015). In 1964, dengue was first reported in Dhaka, hence named after as "Dacca Fever" (Aziz et al., 1967). The fever remained epidemic since then (Ahmed et al., 2001).
Dhaka had been experiencing a large-scale dengue outbreak in every year since 2000 (Akram, 2019). The dengue epidemic in 2019 at Dhaka city created serious public health problem, causing significant absence in working place. Fig 1 shows how dengue emerged on a sudden large scale. The disease caused a huge economic and social burden in the city. In this paper, the socio-economic impact was measured through the number of absent days in working place of the affected people and percentage of monthly family income spend to monthly family expenditure for occurring dengue disease. The significant covariates for occurring this disease were identified and various suggestions for knocking the disease off had been analyzed, discussed and concluding remarks made.

MATERIALS AND METHODS:
Primary data were collected from October 9 to November 20, 2019. For this purpose, a questionnaire was designed and data were collected by direct interview (face-to-face) method. Respondents were conveniently selected from Dhaka city where dengue outbreak had occurred. The study covered 235 dengue affected and 235 unaffected respondents.
The questionnaire involved seven major sections: personal (demographic) information of the respondent's, economic condition, environmental condition, disease incidence, preventive measure, treatment, and suggestions. Regression analysis is an important statistical method to investigate the relationships that exist between a dependent variable and a set of independent variables (Draper & Smith, 1998;Zeileis & Hothorn, 2002). This widely used analysis examines which factors matter most, which factors can be neglected, and how these factors influence each other (Fahrmeir et al., 2013;Montgomery et al., 2012). Regression techniques for making predictions is drive by three techniques mostly: type of dependent variables, number of independent variables, and shape of the regression line (Seber & Lee, 2012). However, two widely used regression model, multiple linear regression and binary logistic regression, had been considered here. If independent variable is more than one and dependent variable is continuous type having linear relationship with independent variables and follows some other specific assumptions then multiple linear regression model can assess the impact of independent variables on dependent variable (Nathans et al., 2012). For the vector of covariates and vector of response variable , the form of the multiple linear regression model in matrix notation becomes (Brown, 2009) = ′ + , Where, and be the vector of regression coefficient and random error term, respectively. If the response vector be of binary type, that is, referring to whether an event of interest has occurred or not, binary logistic regression is used for modeling purpose, which has the following functional form (Sarkar & Midi, 2010).
Where, ( ) represents the conditional mean of given i.e., ( | ). In both models, the unknown parameters ( ) are estimated by the method of maximum likelihood estimation (Albert & Anderson, 1984;Myung, 2003). Here, the former model applied for relating family cost and productivity to dengue incidence, while the factors for occurring dengue disease was identified by the later model. Also, multiple response analysis had been incorporated to assess various suggestions from patients.

Model 1
The impact of dengue infection on the monthly expenditure of the patient was determined by the first model (multiple linear regression models). In this model, dependent variable family cost [percentage of monthly income spent on expenditure] was used as a proxy for measuring monthly expenditure of the patient. The main independent variable was dengue incidence and related covariates were also considered. The model had the form as - Where, denotes regression coefficient of the covariates and is the random error term of the ℎ respondent.

Model 2
The second model was also a multiple linear regression model which determined impact of dengue infection on human productivity. The dependent variable absence [number of days absent from work] was used as a proxy for measuring productivity of the patient. The main independent variable was dengue incidence with other related covariates. The model took the form as - Where, and bear the same meaning as before.

Model 3
Binary logistic regression applied for detecting risk factors of dengue disease. The acting dependent variable dengue incidence [whether dengue disease occurred or not], which was of binary type. This variable and related independent variables took the functional form as -= 0 + 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + , = 1, 2, ⋯ , ; Where, and bear the same meaning as before.  Multiple linear regression and binary logistic regression models had been employed to analyze dengue patients' data. The obtained results were shown in Table 1, Table 2, and Table 3. Table 1 presented the factors related to family cost.  Table 2, it is noticed that the covariates incidence, living status, and treatment cost were significantly affecting productivity, while family size, age, marital status and gender turned out as insignificant at 5% level of significance. Dengue incidence had significant effect on productivity of the patients with p-value less than 0.001. It hampered approximately 6 working days, keeping all other covariates fixed. Living status was significant at 10% significance level. Patients living with family remained 1.261 days less absent from workplace than patients living alone, keeping all other covariates constant. The positive value of the coefficient under treatment cost denoted more absence in workplace with the higher treatment cost. That is, the more a patient had to pay for dengue, the more he remained absent in workings (Shazeed-Ul-Karim, 2019).
The risk factors that are acting behind to occur dengue fever were shown in Table 3. The variables living status, drainage system, providing spray, and area were found significant in Table 3 while gender, age, marital status, and education status were insignificant. The respondents living with family had 2.048 times significantly less odds of occurring dengue disease at 10% significance level than respondents living alone. The respondents provided with good drainage system had experienced 87.9% fewer odds compared to bad drainage system and this is highly significant at 1% level of significance having p-value 0.002. Providing spray in the locality was significant at 1% significance level with p-value 0.001 bringing 54.1% less odds of experiencing dengue disease compared to from not providing spray. The patients belonging to residential, VIP, and commercial area had significantly 1.653, 2.038, and 1.716 times less odds, respectively, of having this disease at 5% level of significance than patients from university area. Multiple Response Analysis -Multiple response analysis is an analysis of frequency when more than one response can be obtained from each participant. Multiple responses had been arranged in three portions in this study: preventive approaches of dengue, suggestions about people's steps on prevention of dengue, suggestions about government's steps on prevention of dengue.

CONCLUSION AND RECOMMENDATIONS:
A self-collected dataset was analyzed through sophisticated statistical modeling approaches to suspect socio-economic impact of dengue. The analysis found no direct effect of dengue on family cost, but it affected productivity which in terms might affect economy. The analysis revealed that alone living group are in greater risk of dengue (Kularatne, 2015). It might be the case that living alone people didn't take enough protection against mosquito bite, hence experienced the fever. People with bad drainage system experienced dengue disease, which might happen for blocked drains that grew and reserved mosquitos (Singh, 2007). The result expressed that providing mosquito spray decreased dengue incidence which comply with previous literature (Chadee, 2013). The probability of being affected varied significantly by different areas (Pathirana et al., 2009). Cleanliness of living areas is a must for preventing the disease which was proven in literature (Pai et al., 2006). From the multiple response analysis, it was observed that a few number of respondents used no preventive approach, while majority used mosquito net as preventive approach. Using mosquito net is effective, as this practice was suggested in literature (Nalongsack, Yoshida, Morita, Sosouphanh, & Sakamoto, 2009). The respondents suggested mostly to use mosquito repellent and net, keep drain free from blockage, and regular changing water from plant container. They suggested government to provide mosquito spray and develop drainage system regularly.
Alone living people were more prone to experience the disease, which might occur due to their insincerity about this disease. So, the living alone persons need to increase their consciousness considering the seriousness of this disease. Drainage system is an important issue for a high densely populated capital like Dhaka city. Government should focus to have well developed systems for neat and clean drains in the capital. City Corporation should provide mosquito spray to a regular basis as it had significant influence to decrease dengue incidence. The magnitude of dengue was different at different areas. Respondents highly recommended using mosquito repellent and net during sleeping, changing water regularly from plant container, providing regular mosquito spray and developing drainage system in Dhaka city. People should become conscious enough about the disease remembering all of its hazards.

ACKNOWLEDGMENT:
First and foremost, the author is grateful to Almighty Allah. The author is also thankful to anonymous reviewers and editors for their helpful comments and suggestions.