CLIMATE AFFLICTIONS | SYNTHESIS REPORT Infected and Stressed by Climate Variability New Empirical Evidence from Bangladesh Iffat Mahmud, Wameq Raza, and Rafi Hossain 2021 Disclaimer This volume is a product of the staff of the International Bank for Reconstruction and Development/World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. CLIMATE AFFLICTIONS | SYNTHESIS REPORT Infected and Stressed by Climate Variability New Empirical Evidence from Bangladesh Iffat Mahmud, Wameq Raza, and Rafi Hossain iv | INFECTED AND STRESSED BY CLIMATE VARIABILITY Contents Acknowledgments vii Abbreviations viii Executive Summary ix CHAPTER 1 Introduction 1 CHAPTER 2 How Does Climate Change Impact Human Health? 3 2.1 Theoretical framework 3 2.2 Overview of the existing literature 6 CHAPTER 3 Data and Methods 11 3.1 Household panel data 11 3.2 Community profile 13 3.3 Definition of key terms used in this report 14 3.4 Limitations of the study 15 CHAPTER 4 Patterns of Climate Variability during the Surveys 17 CHAPTER 5 Infectious diseases 19 5.1 Prevalence of infectious diseases 19 5.2 Correlates of infectious diseases 25 CHAPTER 6 Mental Health 27 6.1 Prevalence of depression and anxiety 27 6.2 Correlates of depression and anxiety 30 CHAPTER 7 Recommendations for Public Policy 33 7.1 Documenting the known 33 7.2 Discovering the not-so-well-known 34 APPENDIX A Data and Methods 35 APPENDIX B Supplementary Tables on Demography, Socioeconomic Characteristics, and Disease Patterns by Location 41 REFERENCES 46 Contents | v LIST OF FIGURES Figure 1. Pathways by which climate change affects human health 4 Figure 2. Global trends in all: Case mortality and mortality from selected causes as estimated by the Global Burden of Disease 2017 for the 1990–2016 4 Figure 3. Vectorial capacity for dengue is increasing over time across the globe 5 Figure 4. Climate suitability for malaria, by region 6 Figure 5. WHO and World Meteorological Organization Framework on the interaction of meteorological and other determinants of dengue transmission cycles and clinical diseases 8 Figure 6. Relationship between incidence of dengue and minimum temperature, maximum temperature, and rainfall 9 Figure 7. Projected population distribution from sampled households 12 FIgure 8. Access to Services 14 Figure 9. Average weather variables, two months preceding each of the two rounds of surveys 18 Figure 10. Heat index measured in degrees Celsius 18 Figure 11. Prevalence of any illness, by season 20 Figure 12. Prevalence of vector-borne, waterborne, and respiratory diseases in monsoon and the dry season 20 Figure 13. Prevalence of infectious diseases across age groups 22 Figure 14. Prevalence of infectious diseases (excluding the common cold) by category, across age groups 23 Figure 15. Distribution of infectious diseases across socioeconomic status, any illness and by illness category 24 Figure 16. Equality of illnesses across socioeconomic status, monsoon, and dry season 25 Figure 17. Prevalence of depression and anxiety, by location, demographics, and seasonality 28 Figure 18. Prevalence of depression by location, age group, gender, and season 29 Figure 19. Prevalence of anxiety by location, age group, gender, and season 30 Figure 20. Sample PSUs by enumeration areas 37 Figure 21. Weather station locations in Bangladesh 39 vi | INFECTED AND STRESSED BY CLIMATE VARIABILITY LIST OF TABLES Table 1. Community profile 13 Table 2. Categorization of infectious diseases 15 Table 3. Infectious diseases (excluding the common cold) as a proportion of the total sample 21 Table A1. Household sample 35 Table B 1. Demographic characteristics of the sample at baseline 41 Table B 2. Socioeconomic characteristics of the sample at baseline 42 Table B 3. Infectious diseases across locations 42 Table B 4. Correlates of contracting any seasonal illness (excluding the common cold) 43 Table B 5. Correlates of contracting vector-borne, waterborne, or respiratory infections 44 Table B 6. Correlates of depression and anxiety 45 Acknowledgments The authors of the report are indebted to the Bangladesh Meteorological Department for sharing weather data and particularly for the cooperation extended by Bazlur Rashid, Meteorologist. The authors would like to recognize the team at Data International who collected data for the two rounds of the panel survey, specifically Nazmul Hossain and A.F.M. Azizur Rahman. The authors gratefully acknowledge contributions of Aneire Khan, Faizuddin Ahmed and Jyotirmoy Saha. The authors are also grateful for the collaboration extended by Syed Shabab Wahid with guidance from Prof. Brandon A. Kohrt of George Washington University for the analyses on mental health issues. Gail Richardson, Practice Manager of Health, Nutrition and Population, South Asia Region of the World Bank, provided oversight for this report, and the authors deeply appreciate her continued support and encouragement. The draft report was shared with the with the Climate Change and Health Promotion Unit (CCCHPU) and the Institute of Epidemiology and Disease Control Research (IEDCR) of the Ministry of Health and Family Welfare (MoHFW) of the Government of Bangladesh. The authors are thankful for their technical advice and collaboration. The authors express their gratitude to the peer reviewers, Anna Koziel (Senior Health Specialist), Stephen Geoffrey Dorey (Health Specialist) and Muthukumara Mani (Lead Economist), as well as Dhushyanth Raju (Lead Economist), Shiyong Wang (Senior Health Specialist), and Tamer Samah Rabie (Lead Health Specialist) for their valuable comments. The authors are grateful to Mercy Tembon, Country Director for Bangladesh and Bhutan, World Bank, who chaired an internal review meeting to seek expert inputs for finalization of the report. The authors are grateful for the financial support mobilized by the Global Facility for Disaster Reduction and Recovery (GFDRR) Multi-Donor Trust Fund as well as the Health Sector Support Project Multi-Donor Trust Fund, which is co-financed by the Embassy of the Kingdom of the Netherlands (EKN), Foreign, Commonwealth and Development Office (FCDO) of the United Kingdom, Gavi, the Vaccine Alliance, Global Affairs Canada (GAC), and the Swedish Development Cooperation Agency (Sida). vii Abbreviations BMD Bangladesh Meteorological Department BMRC Bangladesh Medical Research Council CCHPU Climate Change and Health Promotion Unit EA Enumeration Area ENSO El Niño Southern Oscillation GAD-7 Generalized Anxiety Disorder 7 GDP Gross Domestic Product IEDCR Institute of Epidemiology and Disease Control Research IPCC International Panel on Climate Change LMIC Low- and Middle-income Countries MoEFCC Ministry of Environment, Forest and Climate Change MoHFW Ministry of Health and Family Welfare NCD Noncommunicable Disease PHQ-9 Patient Health Questionnaire 9 PPS Probability Proportion to Size PSU Primary Sampling Unit WASH Water, Sanitation, and Hygiene WHO World Health Organization viii Executive Summary WHY THIS REPORT? Bangladesh’s extreme vulnerability to the effects of climate change is well documented. Through a complex pathway, climatic conditions have already negatively impacted human health worldwide. This is likely to escalate if predicted changes in weather patterns hold. Infectious disease transmission will change in pattern and incidence for certain vector-borne diseases such as malaria and dengue, and waterborne diseases such as diarrhea and cholera. The incidence of respiratory disease will be affected by extreme temperatures that exacerbate the effects of allergens and of air pollution (World Bank 2012). If global warming progresses toward a 4°C increase scenario—a scenario presented as the worst case at the 2015 Paris Climate Change Conference of Parties—stresses on human health can overburden the systems to a point where adaptation will no longer be possible (World Bank 2012). Hence the urgent need for the public sector to be better prepared to respond to the crisis. The consequences of climate change and/or climate variability are well documented and hypothesized. The literature linking climate change or climate variability and health, however, is less so. Climate variability refers to short-term changes in the average meteorological conditions over a month, a season, or a year. Climate change, however, refers to changes in average metrological conditions and seasonal patterns over a much longer time (Mani and Wang 2014). Compared to the availability of global evidence on this topic, the evidence from Bangladesh is far more limited. Among the studies available for Bangladesh, some require further substantiation because they are mostly regional one-off studies with a range of methodological limitations. For example, they often do not account for the representativeness of the population, or fail to use disease-specific data from hospital admission records, or fail to account for non-hospitalized cases, and use climatic conditions from a time that may not match the time period of the illness being investigated. ix x | INFECTED AND STRESSED BY CLIMATE VARIABILITY In an effort to fill this knowledge gap, this report and presents and analyzes climate change evidence from Bangladesh in two broad areas: 1. It quantifies the relationship between climate variability and infectious diseases using primary household-level data that is representative of urban and rural areas; and 2. It measures the prevalence and extent of mental health issues—notably, anxiety and depression—in a sample population using globally recognized standards,1 and establishes their relationship to climate variability. In doing so, the report responds to several key questions, summarized in this subsection. What it does not do is construct mathematical models for projecting the incidence and prevalence of infectious diseases and mental health issues based on predicted climate change patterns. Nor does it attempt to establish a causal relationship between climate change and the selected health conditions. The report uses primary data from a nationally representative sample of about 3,600 households surveyed during the monsoon and dry seasons. It links weather variables, the incidence of selected diseases, and health conditions in Bangladesh to ensure that the findings are, as much as possible, based on precise climate and health data. The recommendations, therefore, are context-specific and drawn from primary evidence. HOW DOES WEATHER AFFECT INFECTIOUS DISEASES? Climatic conditions directly impact the epidemiology of many infectious diseases. Furthermore, these climatic factors interact with behavioral, demographic, socioeconomic and other factors that influence the incidence, emergence, and distribution of infectious diseases (Watts et al. 2018). Climate suitability for climate-sensitive infectious diseases has increased globally (Watts et al. 2020). Vectorial capacity—a measurement of the efficiency of vector-borne disease transmission—is increasing for several climate-sen- sitive diseases, and this is occurring over a wide range of temperature and rainfall patterns. These diseases are most acutely experienced in low- and middle-income countries (LMICs) (Watts et al. 2019). The number of annual cases of dengue fever, which is spread by mosquitoes, has doubled every decade since 1990. One factor that has likely contributed to this increase is climate change (Watts et al. 2020). By comparison, climate suitability for malaria, another mosquito-borne disease, has remained the same for the Southeast Asia region, which includes Bangladesh. DOES A CHANGE IN SEASON MAKE PEOPLE SICK? On average, the likelihood of contracting an infectious disease is 19.7 percentage points lower in the dry season than during the monsoon. If disaggregation by disease type—vector-borne, waterborne, and respiratory diseases—is considered, this trend holds for vector-borne diseases such as dengue, malaria, and their associated symptoms: 25 percent of respondents suffered from vector-borne diseases during the monsoon season, compared to 14 percent in the dry season. For waterborne diseases and respiratory illnesses, the opposite is true: the incidence is higher in the dry season than during the monsoon. Executive Summary | xi HOW DO TEMPERATURE AND HUMIDITY LEVELS AFFECT THE SPREAD OF DISEASES? Humidity and mean temperature are negatively correlated with waterborne diseases but positively correlated with respiratory illnesses. A one percent increase in relative humidity reduces the likelihood of contracting a waterborne disease by 1.6 percentage points while an increase of 1°C in the mean temperature reduces its likelihood by 4.2 percentage points. For respiratory illnesses, higher temperatures are positively associated with respiratory illnesses: a one percent increase in humidity increases their likelihood by 1.5 percentage points, and an increase of 1°C in the mean temperature raises the likelihood of contracting such illnesses by 5.7 percentage points. For vector-borne diseases, an increase in temperature reduces the likelihood of contracting the disease by 1.4 percentage points, but this is not statistically significant. DO MEGA-CITIES EXPERIENCE A LARGER SPREAD OF INFECTION? Regardless of seasonality, monsoon or dry, a higher proportion of respondents residing in Dhaka and Chattogram cities reported experiencing an infectious disease compared to the averages for national, rural, and all urban areas, which include Dhaka and Chattogram cities. When disease disaggregation is considered, the proportion of incidence was higher in Dhaka and Chattogram (34 percent) compared to the national average (25 percent), rural areas (22 percent) and all urban areas (25 percent) during the monsoon, when vector-borne diseases are more prevalent. During the dry season, when waterborne diseases and respiratory illnesses are more prevalent compared to the monsoon, the cities of Dhaka and Chattogram report more respiratory illnesses compared to other areas, possibly due to higher exposure of their residents to air pollution. The incidence of waterborne diseases in Dhaka and Chattogram cities during the dry season is lower than other areas. IS AGE JUST A NUMBER, OR IS MORBIDITY LINKED TO IT? Across the seasons, monsoon and dry, the incidence of infectious diseases increases with a person’s age. Disaggregation by disease category reveals a different pattern. The prevalence of respiratory illnesses is the highest among the elderly—age 65 years and above—and this increases in the dry season—72 percent in monsoon and 83 percent in the dry season. Waterborne diseases are more common among children under 5. A different pattern presents for vector-borne diseases: of the various age groups, the prevalence is the highest among adults—ages 20 to 64 years—across the two seasons. HOW IS MENTAL HEALTH FARING? The findings in this study indicate nationally representative levels of depression and anxiety and identify their determinants. Overall, 16 percent of the respondents reported suffering from depression while a further 6 percent reported anxiety disorders. The most vulnerable to depression and anxiety are older, poorer, and disabled individuals. Additionally, while females are at higher risk of depression than men, men are more susceptible to anxiety. Residents of urban centers are generally more anxious than their rural counterparts. Further analysis of the relationship between weather and depression and anxiety suggests that while temperature is negatively correlated with depression, anxiety is elevated by increases in temperature and humidity. WHAT DOES IT ALL MEAN? As corroborated by the global literature and the primary analysis undertaken for this report, climate variability and seasonal changes influence the prevalence of infectious diseases and affect people’s mental health. The groups most vulnerable to the adverse effects of climate change are children, the elderly, and those living in large metropolises. With climate further predicted to change, the deleterious effects on human physical and mental health are likely to escalate. The discussions point to the need for: 1. Improving data collection systems for improved predictability and localization of weather data, which will help in tracking the impact of climate variability on diseases; 2. Strengthening health systems to preempt and mitigate potential outbreaks of infectious and other emerging or reemerging climate-sensitive diseases; and 3. Ensuring the adequacy of response mechanisms for better adaptation to the effects of climate change. Both mitigation and adaptive measures need to be prioritized, as otherwise this can erode the progress Bangladesh has made over the past five decades. These mitigation and adaptive measures are further discussed in the following section. WHAT ARE THE MAIN RECOMMENDATIONS? Based on the findings, the recommendations included in this report focus on measures to increase the capacity to record accurate weather data at a more localized and granular level, and link it to health data. Such measures will assist in (i) tracing the evolution of climate-sensitive diseases; (ii) strengthening disease surveillance and establishing a climate-based dengue early-warning system that will use weather data to predict possible disease outbreaks; (iii) enhancing vector control measures through innovative approaches; (iv) addressing mental health issues through improved assessments and by facilitating the means to address the shortcomings; and (v) measuring air quality to tackle air pollution, which is an important compounding factor for the spread of diseases. HOW DO THE FINDINGS AND RECOMMENDATIONS SUPPORT THE POLICY DISCOURSE? Notwithstanding the limitations of the study, it is important to note that this report intends to establish a causal link between climate variability as well as seasonal variation and human health, and does not attempt to link human health with climate change, being cognizant of the distinction between the two concepts. The report thus aims to understand better how climate variability affects human health. The findings will assist practitioners and subject matter experts in policy dialogue that will contribute to moving forward the World Bank’s corporate commitment to climate change. The policy dialogue facilitated through this document will focus on supporting governments in further developing and implementing mitigation, adaptation, and resilience measures. Executive Summary | xiii Lastly, it is hoped that the report will pave the way for future research that focuses on building a stronger evidence base for the relationship between climate change and health, and fuel the need to strengthen health systems based on evidence. 2 | INFECTED AND STRESSED BY CLIMATE VARIABILITY increased by 9.5 percent globally since 1950 due to changing climatic conditions in dengue- endemic countries (Watts et al. 2018b). According to the International Panel on Climate Change (IPCC) Fifth Assessment Report (IPCC 2014), even though the present worldwide burden of ill health from climate change is not well quantified, it is likely that climate change has already negatively impacted health. The link between health and climate change is still relatively weak (Lancet Editorial 2019), hence the need for further research in this area (Watts et al. 2018a). In 2017, 43,000 articles were published in the general area of climate change, of which only four percent made any link to health, and fewer than one percent had a specific focus on health and climate change. Most of the scientific interest in health and climate change in 2017 focused on America and Europe. Fewer than 10 percent of the papers related to health and climate change in 2017 were about Africa or Southeast Asia, which includes Bangladesh (Watts et al. 2018a). With climatic conditions projected to worsen, severely climate-change affected countries such as Bangladesh are likely to bear a greater brunt of the adverse effects. Hence the need to understand better how the climate has changed over the years, and is changing, and to document its impact on human health. The existing literature on the linkages between climate change and health for Bangladesh will benefit from further substantiation because the existing studies mostly use small nonrepresentative data, present findings with a narrow regional focus, or are observational studies with limited analyses. A study on Bangladesh (Mani and Wang 2014) reports that the health impact of climate variability differs greatly between pre-monsoon and monsoon seasons. Using monthly surveillance data in regions with a high incidence of vector-borne diseases, the report identifies strong seasonal links between climate variability and vector-borne diseases but no clear trends over the past decade. A possible reason may be that because the authors used secondary health survey data retrofitted with historical weather data to analyze the relationship between climate variability and incidences of morbidity, this may have resulted in imprecisions in estimating this relationship, beyond reporting a marginal but positive relationship between the two. Notwithstanding the findings, the authors conclude—based on a review of existing literature on health and climate change, particularly the links between climate variability and infectious diseases—that this important area of research is still in nascent stages and merits further investigation. With the backdrop of evolving climatic conditions and its effect on health, particularly for Bangladesh, this report aims to establish the relationship between climate variability and infectious diseases and mental health, using nationally representative household-level panel data2 from 3,600 households. 2 How Does Climate Change Impact Human Health? This section begins by outlining the theoretical pathways for how climate change and variability affect health. This is followed by a thematic summary of the relevant literature on the effects of climate change/variability on infectious diseases and mental health as well as the lifecycle of mosquitoes that spread diseases. The summary first presents a global context of the relevant topics, followed by a focus on issues specifically pertinent to Bangladesh. 2.1 THEORETICAL FRAMEWORK The effects of climate change on human health can be direct and indirect, immediate or delayed (McMichael 2012). The main pathways and categories of the health impacts of climate change are shown in figure 1. The direct or immediate effects include risks associated with increased frequency and intensity of heatwaves and extreme weather events such as floods, cyclones, storm surges, droughts, and altered air quality (McMichael 2012). The indirect effects occur through changes and disruptions to ecological and biophysical systems, which may result in altered food production, in turn leading to undernutrition, water insecurity, air pollution, infectious diseases, mental health issues, and forced migration, with accompanying societal disruptions and further downstream effects (Patz et al. 2003; Takaro et al. 2013). Climatic conditions impact the epidemiology of infectious disease and interact with behavioral, demographic, and socioeconomic factors, among others, to influence the incidence, emergence, and distribution of infectious diseases (Watts et al. 2018a). Despite an overall trend of declining infectious disease-related mortality, it still accounts for 20 percent of the global burden of disease (Watts et al. 2018a). For instance, in 2016, deaths from dengue fever were the highest in the Southeast Asia region, which includes Bangladesh, and the overall trend is increasing, based on data from 1990 to 2016 (figure 2). 3 4 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 1. Pathways by which climate change affects human health Injuries, fatalities, Asthma, mental health impacts cardiovascular disease Severe Air weather pollution Malaria, dengue, Heat-related illness and deaths, encephalitis, hantavirus, cardiovascular failure Rift Valley fever, Lyme disease, ures erat Mo chikungunya, West Nile virus Extreme mp re Changes in te e heat vector ing xt rem ecology Ris e we evels ather O2 l gC Environmental Increasing Ris sin degredation allergens in a g re se Inc a le vel Forced migration, civil con ict, Respiratory mental health impacts allergies, asthma Water and food Water supply impacts quality impacts Cholera, Malnutrition, cryptosporidiosis, diarrheal disease campylobacter, leptospirosis, harmful algal blooms Source: World Bank Group and WHO (2018) FIGURE 2. Global trends in all: Case mortality and mortality from selected causes as estimated by the Global Burden of Disease 2017 for the 1990–2016 a. All causes b. Dengue fever c. Diarrheal diseases d. Environmental heat and cold exposure Death per 100,000 people Death per 100,000 people Death per 100,000 people Death per 100,000 people 2.0 2.0 1,250 1.5 100 1.5 1,000 1.0 1.0 750 50 0.5 0.5 500 0 0 90 00 10 90 95 05 10 16 10 16 90 95 00 05 16 95 00 05 90 95 00 05 10 16 19 19 20 20 20 20 19 19 20 20 20 20 19 19 20 20 20 20 19 19 20 20 20 20 e. Exposure to forces f. Malaria g. Malignant skin h. Protein-energy of nature melanoma malnutrition Death per 100,000 people 100 Death per 100,000 people Death per 100,000 people Death per 100,000 people 3 50 20 75 40 15 2 30 50 10 20 1 5 25 10 0 0 0 0 10 10 95 00 05 90 95 00 05 16 90 0 95 00 05 16 10 16 90 95 00 05 10 16 9 19 19 20 20 20 20 19 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 African region European region Southeast Asian region Eastern Mediterranean region Region of the Americas Western Pacific region Source: Watts et al. 2018a Notes: The Southeast Asia region, including Bangladesh, is depicted by light green lines. For infectious diseases (malaria and diarrhea are included in the figure), the mortality rate as measured by deaths per 100,000 people is declining over time in Southeast Asia, but for dengue fever it has been increasing in recent years. How Does Climate Change Impact Human Health? | 5 For several climate-sensitive diseases, vectorial capacity is likely to be positively associated with increasing exposure to temperature and rainfall (Watts et al. 2019). These effects are most acutely felt by LMICs across the world. Vectorial capacity is a measure of the average daily rate of subsequent cases in a susceptible population that result from one infected case. It is calculated using a formula that includes the following five variables: the vector to human transmission probability per bite, the human infectious period, the average vector biting rate, the extrinsic incubation period, and the daily survival period (Watts et al. 2019). Vectorial capacity, in other words, is “a measurement of the efficiency of vector-borne disease transmission” (Johns Hopkins Blomberg School of Public Health). Climate suitability for climate-sensitive infectious diseases has increased globally (Watts et al. 2020). Vectorial capacity for the transmission of dengue from Aedes aegypti and Aedes albopictus mosquitoes has increased significantly worldwide—by 3 percent and 6 percent respectively—compared with 1990 levels (figure 3). The number of cases of dengue fever recorded annually has doubled every decade since 1990, with 58.4 million apparent cases in 2013, accounting for more than 10,000 deaths (Watts et al. 2020). A factor that has contributed to this increase is climate change (Watts et al. 2020). Other emerging and re-emerging diseases, including yellow fever, chikungunya, mayaro, and zika viruses carried by A. aegypti and A. albopictus mosquitoes, are likely to be similarly responsive to the effects of climate change (Watts et al. 2020). Climate suitability for the Southeast Asia region for malaria has remained the same (figure 4), which implies that climate change is unlikely to alter the incidence and prevalence of this disease. Based on this evidence, a few infectious diseases have been purposely grouped and selected for further analysis in this report. These include vector-borne diseases such as dengue, malaria, and chikangunya; waterborne diseases such as diarrhea and dysentery, and respiratory illnesses such as pneumonia and severe acute respiratory infection, and their associated symptoms. FIGURE 3. Vectorial capacity for dengue is increasing over time across the globe 15 Change invectorial capacity (%) 10 5 0 -5 -10 1950 1960 1970 1980 1990 2000 2010 2020 Aedes aegypti Aedes albopictus Source: Watts et al. 2020 Notes: Aedes aegypti and Aedes albopictus transmit dengue and other emerging and re-emerging diseases, including yellow fever, chikungunya, mayaro, and zika viruses. Vectorial capacity is “a mea- surement of the efficiency of vector-borne disease transmission” (Johns Hopkins Blomberg School of Public Health). 6 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 4. Climate suitability for malaria, by region 4 Average number of months suitable for malaria transmission 3 2 1 0 1950 1960 1970 1980 1990 2000 2010 2020 African region Eastern Mediterranean region Western Pacific region Region of the Americas Southeast Asian region Source: Watts et al. 2020 Notes: Southeast Asia (including Bangladesh) is depicted by the light green line. Climate suitability for malaria transmission in Southeast Asia has remained relatively constant since 1950. 2.2 OVERVIEW OF THE EXISTING LITERATURE Infectious diseases Climate change, which includes alterations in one or more variables such as temperature, rainfall, sea-level elevation, wind, and duration of sunlight, influences many cli- mate-sensitive infectious diseases through the survival, reproduction, or distribution of disease pathogens and hosts, as well as the availability and means of their transmission environment (Wu et al. 2016). Human behaviors such as crowding and displacement amplify the risk of infection (McMichael 2012). An agent (pathogen), a vector (host), and favorable transmission environment are the three components essential for the spread of an infectious disease (Wu et al. 2016). There is a limited range of climatic conditions that constitutes the climate envelope within which an infective agent or vector species can survive and reproduce (Patz et al. 2003). At this point in time, there is sufficient observational evidence of the effects of meteorological factors on the incidence of vector-borne, waterborne, airborne and foodborne diseases. A more contemporary concern is the extent to which changes in disease patterns will occur under the conditions of global climate change (Patz et al. 2003). For that reason, the correlation between meteorological factors and the components of transmission cycles—such as parasite development rates, vector biting, and survival rates, or the observed geographical distribution of disease—have been used to generate predictive models (Campbell-Lendrum et al. 2015). These models link projections of future scenarios of climate change with other determinants such as gross domestic product—as a measure of socioeconomic and technological development— and urbanization. However, because of uncertainties in climate projections and future How Does Climate Change Impact Human Health? | 7 development trends, as well as the compounding effect of natural climate variability over short to medium timescales—from years up to two decades—the models are highly approximate and are only able to comment on broad trends. In Bangladesh, very few studies have explored the relationship between environmental variables and infectious diseases. Temperature and precipitation changes have been found to impact the dynamics of vector-borne diseases such as malaria, dengue, visceral leishmaniasis—commonly known as Kala-azar—cholera and diarrheal diseases (Rahman et al. 2019; Banu et al. 2014; Hossain et al. 2011; Reid et al. 2012; Hashizume et al. 2007). Although the country has made progress in controlling communicable diseases in recent years, dengue cases have surged, as have chikungunya and zika cases more recently, posing major threats to the health of the population. Higher temperatures are expected to increase the transmission and spread of vector-borne diseases by increasing mosquito density in some areas and increasing their replication rate and bite frequency (Costello et al. 2009). This, in turn, will likely increase the incidence of malaria, dengue, and tick-borne encephalitis (Costello et al. 2009). Mental health Global evidence of the effects of climate change or climate variability on mental health is limited but is steadily increasing. Extreme weather events brought on by climate change have been identified as one of the triggers of a host of mental health issues (Berry et al. 2010). These include major depressive disorders and other forms of depression, anxiety, post-traumatic stress disorder, grief and bereavement, survivor guilt, recovery fatigue, substance abuse, suicidality, and vicarious trauma in first responders (Berry et al. 2008; Berry 2009; Berry et al. 2010; Bourque and Willox 2014; Willox et al. 2013a; Willox et al. 2013b; Willox et al.2015; Doherty and Clayton et al. 2014; Clayton et al. 2017; Coyle and Susteren 2012; Weissbecker 2011; Swim et al. 2009). Most of the evidence on the topic, however, comes from high-income countries. Based on the limited insights available from low- and middle-income countries (LMICs), mental health issues in these countries are likely to be aggravated, given their existing vul- nerabilities and their limited capacity to address them (WHO 2009). In Bangladesh, natural disasters and environmental degradation on account of climate change, climate variability, or both are known risk factors that can affect the psychological health of vulnerable populations, especially those in coastal areas— although this has not been documented well in the local context. The majority of Bangladeshis who live in coastal areas are low-income agricultural workers, many of whom are landless and are relatively asset-poor (Government of Bangladesh 2008a; Paul et al. 2009). They are frequently affected by natural disasters yet have insufficient resources to protect themselves, adequately rebuild their lives after the event, or access medical services when needed (Nahar et al. 2014). The initial response to a natural disaster is to ensure that the survivors receive the most basic necessities, such as shelter, food, safe water, and sanitation. However, once this acute, emergency phase has passed, many of the affected populations or climate refugees are still left with some level of psychological or mental health problems (Nahar et al. 2014). These can include post-traumatic stress disorders, depressive symptoms or major depressive disorders, anxiety or generalized anxiety disorders, and more general mental health problems such as sleep disruption, substance abuse, or aggression (Norris 2005; Paul et al. 2009). 8 | INFECTED AND STRESSED BY CLIMATE VARIABILITY Climate variability and mosquitoes Dengue is one of the most critical mosquito-borne diseases affected by climate variability, and it continues to spread globally throughout the tropical and subtropical regions (Costa et al. 2010). Dengue, chikungunya, and Zika virus are spread by the same mosquito species, Aedes aegypti (Lowe et al. 2017). Figure 5 presents the pathways by which dengue transmission cycles are altered by weather variables and other factors. Ebi and Nealon (2016) have summarized the lifecycle of a mosquito: female mosquitoes lay eggs on the side of water-holding containers while humans provide the blood meals necessary for egg development. The female mosquitoes usually rest in cool, dark places and generally bite humans indoors. After a flood or rain, the eggs hatch into larvae. Within a week or so, under favorable environmental conditions, the larvae transform into pupae and evolve into adult mosquitoes. It takes between 5 and 33 days, with a mean of 15 days, at 25°C for the viruses spread by these mosquitoes to multiply, mature, and travel to the salivary glands of the mosquito before it can transmit the virus to a person by biting. The variability in climatic conditions such as temperature, precipitation/rainfall, and humidity brought on by climate change is affecting the biology of mosquito vectors as well as the risk of disease transmission (Costa et al. 2010). Colón-González et al. (2013) conclude that dengue transmission rapidly increases when the minimum temperature rises above 18°C, based on data from Mexico on laboratory-confirmed dengue cases from 1985 to 2007, along with weather data (monthly averages for minimum temperature, maximum temperature, and rainfall). Colón-González et al. conclude that the minimum temperature has the biggest impact on dengue—with zero risk below 5°C but a rapidly increasing risk when the average minimum temperature exceeds 18°C. FIGURE 5. WHO and World Meteorological Organization Framework on the interaction of meteorological and other determinants of dengue transmission cycles and clinical diseases Temperature and precipitation Population size and distribution Vector Social and ecology Individual virological and Epidemic ecological Aquatic Vector or endemic immunological factors context breeding capacity disease Community infrastructure sites and feeding and behavior oppertunities Epidemiolog y Control policies Vector density Clinical and services (and fitness and longevity) Dengue disease serverity Vector control Community action Dengue transmission Herd immuntity Source: Ebi and Nealon 2016 How Does Climate Change Impact Human Health? | 9 The maximum temperature also influences dengue independently of the minimum temperature; Colón-González et al. found that dengue cases increase particularly rapidly when the maximum temperature is in the 25°C–35°C range, with a peak at 32°C. At temperatures above 32°C, the risk of dengue begins to decrease as adult mosquitoes begin to die at temperatures above 35°C. With respect to rainfall, dengue cases increase particularly rapidly in the range of 200 and 800 millimeters of rainfall, with a peak at 550–650 millimeters. The authors also found a higher incidence of dengue in the wet season from May to October in Mexico (figure 6). Zhang et al. (2019) conclude that periods of increased temperatures can trigger dengue epidemics. FIGURE 6. Relationship between incidence of dengue and minimum temperature, maximum temperature, and rainfall a. Minimum b. Maximum c. Rainfall temperature temperature 30 8 5 4 6 Incidence Incidence Incidence 20 3 4 10 2 1 2 0 0 0 5 10 15 20 25 15 20 25 30 35 40 0 200 400 600 800 1000 1200 Degrees Celsius Degrees Celsius Precipitation (millimeters) Source: Colón-González (2013). Note: Incidence of dengue is on the vertical axis and weather data on the horizontal axis. Dotted lines are standard deviations. 3 Data and Methods Several sources of data have been used in this report. These include primarily collected panel data over two time periods—August to September 2019, and January to February 2020; localized weather data from the Bangladesh Meteorological Department (BMD) covering conditions from the preceding two months of each survey, and secondary analysis of data available from various sources. 3.1 HOUSEHOLD PANEL DATA The first round of household panel data was canvassed on 3,610 households comprising 15,383 individuals between August and September 2019, immediately past the peak of the monsoon season. The follow-up round collected the same information from the same households between January and February 2020 during the dry season. The second round collected information from 3,480 households comprising 14,474 individuals, with an attrition rate of three percent. The timings of the two rounds were deliberately chosen to identify seasonality and variations in the outcomes of interest. Households were tracked for any change of their residence between the two survey periods. The sample is representative of urban and rural areas. Furthermore, the sampling design allows for assessing heterogeneity across urban areas, such as major city centers and other urban areas. A structured questionnaire, directed toward the primary female member of the household, was used to collect information on an array of issues. The cascading questions first asked whether any member of the household had fallen ill followed by whether they had visited a doctor for the illnesses and had received a medical diagnosis. The subsequent set of questions collected detailed information about symptoms of illnesses in the event that they had not acquired a medical diagnosis. This report considers three sets of primary outcomes—infectious diseases, persistent or chronic illnesses, and mental health. During the survey, respondents were asked about their morbidities. A detailed description of the survey, sampling strategy and methods are provided in annex A. 11 12 | INFECTED AND STRESSED BY CLIMATE VARIABILITY Figure 7 shows the projected national population distribution from the sample, and table B-1 (in appendix B) provides details of demographic profiles of the sample. The sample is equally distributed between genders and remains consistent across the geographic strata. The average age is approximately 28 years. Almost half of the population on average are married. The approximate educational attainment is 4.9 years of schooling, with the urban population better educated than their rural peers. A high proportion of household heads (92 percent) are male; that proportion is 4 percentage points lower in Dhaka and Chattogram cities than in rural areas. Similarly, the average age of the household heads is nearly twice the national average, and they are better educated in urban than in rural areas. The largest proportion of households (24 percent) falls in the lowest quintile of the socioeconomic index while the smallest (16 percent) is in the highest quintile. Table B-2 (in appendix B) outlines the socioeconomic conditions of the sampled households at baseline. The poorest are most heavily represented in the rural areas (28 percent) than all urban, with the cities of Dhaka and Chattogram at 10 and 6 percent respectively. Inversely, the richest reside in Dhaka and Chattogram (45 percent) and all urban areas (35 percent). This is reflected in other household characteristics as well. Compared to 22 percent in rural areas, none of the households in Dhaka and Chattogram as have mud or straw as the primary building material of their walls, and only 4 percent of households in all urban do. In the rural areas, 90 percent of the houses use tin as a roofing material, compared to 48 percent in Dhaka and Chattogram cities and 68 percent in all urban areas. Access to electricity is nearly universal across the urban space, compared to 86 percent in rural areas. Similarly, 97 percent of households in Dhaka and Chattogram cities use clean stoves, compared to half of households in all urban areas, and only FIGURE 7. Projected population distribution from sampled households 75+ 71 to 75 66 to 70 61 to 65 56 to 60 51 to 55 46 to 50 Age group 41 to 45 36 to 40 31 to 35 26 to 30 21 to 25 16 to 20 11 to 15 6 to 10 0 to 5 20 15 10 5 0 5 10 15 20 Population in millions Male Female 14 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 8. Access to Services 100% Improved water 100% 96% 90% 80% 99% Market 70% Primary school 84% 60% 100% 50% 81% 40% 30% 20% 10% 84% Secondary Fire station 25% 0% 3% school 81% 21% 37% 46% 49% 85% 62% Police station Post office Bank Urban Rural 3.3 DEFINITION OF KEY TERMS USED IN THIS REPORT Climate is defined as “the mean and variability of relevant atmospheric variables such as temperature, precipitation and wind. Climate can thus be viewed as a synthesis or aggregate of weather” (Fouque and Reeder 2019). Climate variability refers to “the day-to-day change in meteorological parameters including temperature, precipitation, humidity, and winds. Extreme weather events are significant deviations of meteorological variables, such as floods caused by excessive rainfall, droughts, storm surges, and heat waves (extreme temperature). Climate variability refers to short-term changes in the average meteorological conditions over a time scale, such as a month, a season, or a year” (Mani and Wang 2014). Climate change refers to “changes in average metrological conditions and seasonal patterns over a much longer time horizon, often over 50 or 100 years” (Mani and Wang 2014). Infectious diseases are defined as any seasonal disease that was experienced in the 30 days preceding the survey. Infectious diseases include diseases diagnosed by a medical professional as well as symptoms the respondents experienced for which they did not seek any medical help. During the analysis, these symptoms were tagged to specific diseases. Table 2 provides details of the categorization used for seasonal illnesses. These infectious diseases are likely to be climate-sensitive. Data and Methods | 15 TABLE 2. Categorization of infectious diseases CLASSIFICATION DIAGNOSED DISEASES SYMPTOMS CLASSIFIED WITH DISEASES BUT UNDIAGNOSED (1) (2) (3) Common cold Common cold Fever with runny nose, chills, sore throat Runny nose, chills, sore throat, cough (without fever) Vector-borne diseases Dengue (classical) Fever with body aches, pain in small joints, retro-orbital pain, rash Dengue (hemorrhagic) Fever with body aches, chills, pain in small joints, retro-orbital pain, rash, hemorrhages Chikangunya Fever with body aches, joint pain, radiating joint pain Malaria Fever with chills, body aches Respiratory illnesses Pneumonia Influenza-like illness Fever with cough, sore throat, body aches, headache Severe acute respiratory Fever with cough, sore throat, body aches, headache, difficulty in infection breathing Waterborne diseases Diarrhea Dysentery Note: The table shows the classification of infectious diseases (column 1). Column 2 is a list of medically diagnosed illnesses reported by the respon- dents. Column 3 presents symptoms that were grouped together (with IEDCR’s guidance) and assigned to the disease category for patients who did not seek medical care. 3.4 LIMITATIONS OF THE STUDY The survey data were collected during the monsoon and dry seasons on the incidence of selected diseases from a representative sample of households. Although the survey is representative of the country’s urban–rural population distribution, the linearized error terms of some of the illnesses exceed the recommended level of dispersion—more than 15 percent. While it may not be representative at the sub-outcome level, the results suggest trends that are worth noting. A larger sample size could have helped to enhance the robustness of the findings as well as its reliability. The study presents certain other limitations which should be investigated in future research to estimate the burden of disease. Not all illnesses were medically diagnosed, and disease categories were inferred from a wide array of symptoms as reported by the respondents. Also, the severity of the disease or the symptoms was not considered. The prevalence of air pollution, which is directly linked to several diseases including respiratory diseases, cardiovascular damage, fatigue, headaches, and anxiety, was not considered. Finally, the correlation between different health outcomes was not evaluated. This is particularly important for malnutrition, which is related to the occurrence of several diseases. Climate variables used in this study were based on data collected from only 43 weather stations of BMD. For enumeration areas where a weather station was not available, BMD provided the closest approximation. This may have resulted in imprecisions in the measurement of local climate conditions and, consequently, in the estimated impact of climate variability on the incidence of diseases. The scope of the questionnaire was broadened to meet the needs of users. The structured questionnaire was fairly substantial—15 pages long, and containing some open-ended questions. As a result, some of the respondents were unwilling to give adequate time to the interviewers and this may have impacted their responses and consequently some of the findings. For questions regarding expenditure, respondents were not asked to provide any documentary evidence, and some had difficultly recalling the amount of money spent. 4 Patterns of Climate Variability during the Surveys This section discusses climatic conditions related to the survey, covering the monsoon from August to September 2019 and the dry season from January to February 2020. Data on three key weather variables—maximum temperature, minimum temperature, and rainfall—were collected from BMD’s 43 weather stations during the two months preceding each round of the survey (details in appendix A). These weather variables are reported for Dhaka and for Chattogram, and also as national averages, for May, June, November and December 2019 (figure 9). Between May and June, the maximum temperature decreased significantly across the country, while the minimum temperature increased slightly. Rainfall reported in Dhaka was substantially higher than in Chattogram, which is unusual since historical analysis of weather data shows that rainfall in Chattogram is usually higher. While rainfall increased between May and June in Dhaka and nationally, it decreased for Chattogram. The weather variables for November and December 2019 show that the usual temperatures—both maximum and minimum—fell significantly compared to temperatures in May and June, and that no rainfall was recorded in these two months. The heat indexes for Dhaka and Chattogram cities are presented in figure 10. Heat index is a measure arrived at by factoring relative humidity in with the actual air temperature.3 Evidence from Bangladesh indicates a higher risk of cholera two days after heatwaves during the rainy season (Wu et al. 2018). The heatwaves can increase the risk of transmission of Nipah virus to humans because the extreme heat conditions put the bats that carry the virus under physiological stress, which can trigger prolonged viral shedding (Rahman et al. 2019). Given the potential impact of heat waves on health conditions in general—not just infectious diseases—a heat index was constructed for the months of May, June, July, August, November, and December 2019 and January and February 2020. May to August 2019 covers the period of data collection during the monsoon as well as two months prior to it, while November 2019 to February 2020 covers the dry season—data collection period plus two months prior. There was no variation in the heat indices of May and June 2019 for Dhaka and Chattogram cities. However, in 17 18 | INFECTED AND STRESSED BY CLIMATE VARIABILITY July and August 2019, it was higher for Dhaka city than Chattogram city by 5°C to 6°C. The heat index for Dhaka city during the dry season was lower than Chattogram city, except for January 2020. FIGURE 9. Average weather variables, two months preceding each of the two rounds of surveys 40 300 35 34.5 33.5 33.7 34.6 33.9 33.2 250 231 242 30.3 30.6 30.3 Temperature (degrees Celsius) 30 26.5 26.5 26.7 26.8 26.2 Rainfall in millimeters 25.6 25.7 25.0 25.4 200 25 183 21.7 20.8 20 155 19.7 150 16.0 123 15.4 15 114 14.2 100 10 75 50 5 0 0 0 14 0 0 0 May June May June May June Nov Dec Nov Dec Nov Dec Dhaka Chattogram National Dhaka Chattogram National MONSOON DRY SEASON Minimum temperature Maximum temperature Rainfall FIGURE 10. Heat index measured in degrees Celsius 60 50 50 49 49 49 50 50 44 44 40 36 37 33 29 29 28 30 25 25 20 10 0 May '19 Jun '19 Jul '19 Aug '19 Nov '19 Dec '19 Jan '20 Feb '20 MONSOON DRY SEASON Dhaka Chattogram 5 Infectious diseases This section analyzes illness patterns across the two seasons covered by the survey— the monsoon when data were collected in July and August 2019 and the dry or post-monsoon of January and February 2020. The results are presented with varying degrees of disaggregation across areas of geographic representation, demographic and socioeconomic status, and household water, sanitation, and hygiene (WASH) practices. As Lowe et al. (2017) indicate, both rainfall and drought can increase the availability of potential habitats for mosquitoes—containers with stagnant clean water—and therefore the availability of adequate WASH facilities is an important compounding factor for dengue, for example. 5.1 PREVALENCE OF INFECTIOUS DISEASES This subsection discusses the prevalence of infectious diseases defined as individuals who have reported experiencing any infectious diseases or illness within 30 days preceding the survey across the monsoon and dry seasons. For the purposes of analysis, these infectious diseases, excluding the common cold, are classified as vector-borne, waterborne, or respiratory diseases. Disease-specific figures are presented as a proportion of reporting any infectious disease, excluding the common cold, within these three categories. Common cold is reported separately. Overall, the common cold is more prevalent than other kinds of infectious diseases across seasons and locations, except for Dhaka and Chattogram cities during the monsoon. On average, at the national level, the likelihood of reporting an infectious disease, excluding the common cold, is 1.2 percentage points higher during monsoon than in the dry season: 5.7 versus 4.5 percent (figure 11). While overall urban and rural areas are generally comparable and remain so across the seasons, Dhaka and Chattogram cities report the highest number of infectious diseases, excluding the common cold, across the seasons. Respiratory illnesses are the highest reported, across seasons nationally. At the subnational level, the rates in all urban and rural areas are comparable over the two seasons,. But in Dhaka and Chattogram cities, the rates significantly lower during 19 20 | INFECTED AND STRESSED BY CLIMATE VARIABILITY monsoon (43.8 percent) but the highest during the dry season (66 percent). Figure 12 provides a breakdown of the three categories of illnesses across the seasons and by area. FIGURE 11. Prevalence of any illness, by season 14% 12% 11.4 11.6 10.4 10.7 9.9 10% 9.1 8.1 8% 6.2 5.7 5.6 5.7 5.7 6% 4.8 4.5 4.7 4.5 4% 2% 0% l l l l na an an d ra na an an d ra at io rb a ities Ru at io rb a ities Ru N All u k a c N All u k a c Dh am Dh am gr gr atto atto Ch Ch MONSOON DRY SEASON Infectious disease (excl. common cold) Common cold Note: The figure shows weighted averages across geographical clusters. FIGURE 12. Prevalence of vector-borne, waterborne, and respiratory diseases in monsoon and the dry season 100% 14 15 20 14 25 22 25 34 80% 60% 62 63 62 61 62 66 44 61 40% 20% 21.9 23 23 24 14 15 14 15 0% l l l l na an an d ra na an an d ra at io rb a ities Ru at io rb a ities Ru N All u k a c N A ll u k a c Dh am Dh am g r g r a tto a tto Ch Ch MONSOON DRY SEASON Waterborne disease Respiratory disease Vector-borne disease Note: The categories of infectious disease (vector-borne, waterborne, and respiratory diseases) are a subset of contract- ing any infectious disease (excluding the common cold) in the 30 days before a survey. Infectious diseases | 21 Vector-borne diseases are more prevalent in Dhaka and Chattogram cities across the two seasons, monsoon and dry, than in the national, all urban, and rural areas (figure 12). Subnationally, vector-borne diseases are reported by 22 percent of all urban areas, and 25 percent in rural areas, during the monsoon. The rates are the highest in Dhaka and Chattogram cities (34 percent). During the dry season, while the rates are consistent across all urban and rural areas—between 14 and 15 percent—the prevalence of vector-borne diseases remains highest in Dhaka and Chattogram cities, at almost 20 percent. Across all locations, the prevalence of vector-borne diseases is lower in the dry season than in the monsoon. Finally, the lowest proportion of the ill report contracting waterborne diseases (figure 12). At the national level, while 14 percent contract a waterborne disease during the monsoon, this increases to 23 percent during the dry season. Subnationally, the rates across all urban (15 percent) and rural areas (14 percent) are comparable to each other: but both rates increase to 23 percent during the dry season. Dhaka and Chattogram cities deviate from the rest of the country across the seasons with respect to waterborne diseases. While the prevalence is higher during monsoon in these cities—22 percent compared to national (14 percent), all urban (15 percent), and rural areas (14 percent)—it reduces significantly in the dry season to 15 percent, and is the lowest among all the geographical areas. To demonstrate the scale of morbidity, the data on disease prevalence are alternatively presented as a proportion of the total population, as opposed to a subgroup of those who reported an illness. As shown in table 3, vector-borne and respiratory diseases are significantly higher in the Dhaka and Chattogram cities than in rural areas in the dry season. Figure 13 shows the prevalence of infectious diseases and the common cold separately across age groups. Nearly one in ten children under 5 years of age and the elderly ages 65 or older report a seasonal illness, excluding the common cold, during the monsoon. The likelihood of reporting a seasonal illness is the lowest among adolescents ages 6–9 years (4 percent), followed by adults ages 20–65 years (6 percent) in the same season. Though overall rates of morbidity fall during the dry season, they remain proportional to the rates in the monsoon. In the dry season, the elderly ages 65 or older report the highest incidence of infectious diseases, excluding the common cold. TABLE 3. Infectious diseases (excluding the common cold) as a proportion of the total sample MONSOON DRY SEASON Dhaka and Dhaka and All Chattogram All Chattogram Urban cities Rural Difference Urban cities Rural Difference Disease Category (1) (2) (3) (3)-(1) (3)-(2) (4) (5) (6) (6)-(4) (6)-(5) Waterborne 0.9% 1.4% 0.8% n.a. * 1.1% 0.8% 1.1% n.a. n.a. Respiratory 3.5% 2.7% 3.5% n.a. n.a. 2.9% 3.8% 2.8% n.a. * Vector-borne 1.3% 2.1% 1.4% n.a. n.a. 0.7% 1.1% 0.6% n.a. * Sample 15,409 14,758 Note: The table shows the geographic prevalence of diseases across seasons. Figures presented are weighted means. Tests (columns labeled as difference) show significance levels from a weighted t-test.* p<0.1; ** p<0.05; *** p<0.01. n.a. = not applicable. 22 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 13. Prevalence of infectious diseases across age groups 25% 21 20% 16 15% 14 10 11 10% 10 10 10 9 9 8 7 6 5 5% 4 3 0% ar s ar s ar s rs ar s ar s ar s rs ye ye ye yea ye ye ye yea 0– 5 19 4 5+ 0– 5 19 4 5+ 6– –6 6 6– –6 6 20 20 MONSOON DRY SEASON Infectious diseases (excl. common cold) Common cold Note: Figure shows weighted means across age groups and seasons. Figure 14 presents the prevalence of infectious diseases by category. Respiratory diseases remain the most pervasive across the two seasons, with prevalence significantly higher during the dry season, except for children—ages 0 to 5 years—in the dry season when they experience more waterborne diseases. The prevalence of respiratory illnesses is the highest among the elderly, at 72 percent, in the monsoon season and 83 percent in the dry season. Nearly 65 percent of children ages 5 or less report a respiratory disease during the monsoon season, but this number markedly reduces to 37 percent during the dry period. The prevalence of respiratory illnesses among the younger population ages 6–19 years and adults ages 20–64 years remains comparable at 59 percent during monsoon, although it increases considerably for both groups during the dry season. In the monsoon, the reported rates of waterborne diseases are highest among children ages 5 or less and the elderly ages 65 years or more—22 and 24 percent respectively—while the lowest for younger populations from ages 6 to 19 years at 6 percent. Overall, rates of waterborne diseases increase for everyone during the dry season, except people ages 65 years and above. Incidences increase by nearly three-fold to 56 percent among children ages 5 years or less in the dry season. The prevalence among younger population from ages 6 to 19 years and the adults from ages 20 to 64 years remain comparable between 17 and 18 percent in the dry season (figure 14). During the monsoon, the prevalence of vector-borne diseases is the highest among the younger population ages 6–19 years at 35 percent, followed by adults ages 20–64 years at 28 percent (figure 14). The rate is 13 percent among children ages 5 years or younger, and lowest among the elderly ages 65 years or older (4 percent). During the dry season, the prevalence is generally lower, except for the elderly. The younger Infectious diseases | 23 FIGURE 14. Prevalence of infectious diseases (excluding the common cold) by category, across age groups 100% 4 7 10 13 15 17 28 35 80% 37 60% 72 65 68 65 83 59 40% 59 56 20% 22 24 17 18 14 6 7 0% s ar s ar s ar s ar s ar s ar ar s ar s ye ye ye ye ye ye ye ye 9 0–5 19 64 5+ 0–5 6–1 64 5+ 6– 20 – 6 20 – 6 MONSOON DRY SEASON Waterborne disease Respiratory disease Vector-borne disease Note: Figure shows weighted averages across age groups. The categories of infectious disease, excluding the common cold (vector-borne, waterborne, and respiratory diseases), are a subset of contracting any infectious disease in the last 30 days. population ages 6–19 years, and adults ages 20–64 years maintain the highest rates (15 and 17 percent, respectively). Vector-borne diseases in the elderly (ages 65+) is 6 percentage points more in the dry season than in the monsoon. Figure 15 presents the likelihood of infectious diseases across socioeconomic quintiles. Morbidity declines with increasing socioeconomic status and the trend holds across the monsoon and dry seasons. Trends in disease types across socioeconomic groups reflect intuitive trends. For instance, respiratory diseases are the most reported diseases, particularly during the dry season. The highest rates of respiratory-relat- ed morbidity are reported by the highest quintile. Compared to the monsoon, the difference between the bottom and top quintiles increases to 10 percentage points in the dry season. On average, the likelihood of reporting a waterborne disease is lower during monsoon than in the dry season. The difference in the prevalence rates between the quintiles is limited during monsoon, ranging between 13 and 16 percent. While the overall prevalence rates increase by up to 10 percentage points during the dry season, except for the richest quintile where it decreases. The rates of waterborne diseases among the richest are significantly lower in comparison to the poorest—9 versus 24 percent. Vector-borne diseases are more prevalent among lower socioeconomic groups in the monsoon—28 percent among the poorest in contrast to 19 percent among the richest. While the prevalence rates are marginally lower during the dry season, the most dramatic drops are experienced by those in the lower three quintiles—17 percentage points on average. The proportions among the richest remain comparable 24 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 15. Distribution of infectious diseases across socioeconomic status, any illness and by illness category a. Any illness 7 7 7% 6 5 6% 5 5 5 5% 4 4 4 4% 3% 2% 1% 0% Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 MONSOON DRY SEASON b. Illness category 100% 12 7 12 17 19 21 18 28 24 30 80% 60% 66 63 61 67 55 66 73 59 62 55 40% 20% 24 25 27 26 13 13 16 16 15 9 0% Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 MONSOON DRY SEASON Waterborne disease Respiratory disease Vector-borne disease Note: The figure shows weighted averages across socioeconomic quintiles constructed using an asset and facilities index. The categories of illnesses (vector-borne, waterborne, and respiratory diseases) exclude common cold, and are a subset of contracting any illness in the last 30 days. Q1 represents the poorest quintile and Q5 the richest. across the seasons; this could be because data collection during monsoon overlapped with Eid holidays when a significant proportion of the rich travel abroad and thus may have been less exposed to vector-borne diseases. Figure 16 shows the concentration of illnesses across socioeconomic status during monsoon and the dry seasons. Concentration curves are used to identify whether socioeconomic inequality in the share of illnesses exists and whether it is more pronounced at one point in time than another. The curves represent the cumulative share of illnesses according to the cumulative share of the population, ranked from poorest to the richest. The further a curve is above the reference equality 45-degree line, the more the corresponding health variable—that is, share of illnesses— Infectious diseases | 25 FIGURE 16. Equality of illnesses across socioeconomic status, monsoon, and dry season a. Monsoon b. Dry season 1.0 1.0 Cumulative share of illnesses in monsoon 0.8 0.8 Cumulative share of illnesses in winter 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 Cumulative share of ill (poorest first ) Cumulative share of ill (poorest first) Rural Urban Equality Rural Urban Equality Note: The figure shows concentration of illnesses across the range of a household’s relative socioeconomic status, stratified by urban and rural areas. is concentrated among the poorest households. When the concentration curve lies under the 45-degree line, the corresponding health variable is more concentrated among the richest households. While no differences exist across the socioeconomic percentiles during monsoon, the concentration curve corresponding to the rural areas lies above the 45-degree line during the dry season, suggesting that the rates of morbidity are more concentrated among the poor during this time.4 The concentration curve for the urban areas during the dry season is very close to the 45-degree line, which indicates very little association between share of illnesses and socioeconomic status. 5.2 CORRELATES OF INFECTIOUS DISEASES Weather is correlated with illness. An increase in humidity and mean temperature by one unit in two months preceding the survey reduced the likelihood of contracting an illness by 0.2 and 1.4 percent, respectively. The likelihood of contracting an infectious disease is 19.7 percentage points lower in the dry season than during monsoon. Seasonal differentials suggest younger people ages 6–19 years are 1.6 percentage points more likely to suffer from an infectious disease during the dry season, while similar is true for adults ages 20–64 years, at 2 percentage points. Overall, younger people ages 6–19 years and adults ages 20–64 years are, respectively, 5 and 3.9 percentage points less likely to experience an infectious disease than children ages 5 or younger, or the elderly 65 years or older. Table B-5 in appendix B shows correlates of contracting any infectious disease, excluding the common cold. Coefficients presented are derived 26 | INFECTED AND STRESSED BY CLIMATE VARIABILITY using a weighted linear specification. The models control for a host of additional individual- and household-level characteristics, including education, household characteristics, gender, disability status, availability of WASH facilities, infrastruc- ture-based connectivity, and so on. Condition-specific correlates are presented in table B-5 in appendix B, which are also estimated using a linear specification. With regards to weather outcomes, humidity and mean temperature are negatively correlated with waterborne diseases by 1.6 and 4.2 percent. The opposite is true for respiratory illnesses: an increase in humidity or mean temperature increases the likelihood of suffering from a respiratory illness by 1.5 and 5.7 percent, respectively. There is some heterogeneity across seasons and age groups. Adolescents, adults, and the elderly are less likely to experience waterborne diseases during the dry season by 19, 28 and 49.5 percentage points, respectively, than during monsoon. In contrast, all three groups are more likely to experience respiratory illnesses during the dry season than in monsoon, with 27.8, 32.4 and 36.8 percentage point increases for adolescents, adults and the elderly, respectively. Disabled individuals are 8.8 percentage points less likely to experience respiratory infections but 5 percentage points more likely to contract a vector-borne disease during the dry season than in monsoon. While morbidity rates among households based in Dhaka and Chattogram cities do not on average vary much compared to other urban or rural locations, the likelihood of contracting waterborne diseases fall during the dry season, but—in line with the literature—the chances of contracting a respiratory illness increases by 19.4 percentage points. 6 Mental Health 6.1 PREVALENCE OF DEPRESSION AND ANXIETY Based on the same set of panel data used in the previous section, analyses of mental health issues are presented in this subsection. The analyses of depression are based on the Patient Health Questionnaire-9 (PHQ-9) score, a commonly used depression screening tool. For anxiety, we used the Generalized Anxiety Disorder-7 (GAD-7). A clinical cutoff score of 10 (further details in appendix A) was used to establish the presence of depression and anxiety for both the PHQ-9 and the GAD-7, based on best practice standards in the literature. Both depression and anxiety are significantly more prevalent in Dhaka and Chattogram cities than in the rest of the country. The weighted prevalence of depression was the same across rural and urban locations, at nearly 16 percent. The prevalence of anxiety, however, varied—7.9 percent in urban areas compared to 5.5 percent in rural areas. The prevalence estimates specific to age, gender, location, and seasons are presented in figure 17, which shows that depression is more prevalent than anxiety across all these dimensions. Both depression and anxiety increase with age. Almost two in five older people ages 65 years or more reported being depressed, and 13 percent suffered anxiety. The prevalence of depression between male and female was the same, at approximately 16 percent, but men reported more anxiety than women. With respect to seasonality, more people were depressed in the dry season compared to monsoon—14 percent compared to 18 percent—while anxiety levels were similar, with a difference of just one percentage point. Across all categories, the older-age cohorts presented with more severe depressive symptoms (figure 18). Younger females, ages 26–40 years, were more vulnerable to depression than their male counterparts, ages 26–40 years, and females of other age groups, as well as urban-dwelling males ages 41–65 or older. In Dhaka and Chattogram cities, females ages 26–40 years experienced more depression than any other age groups. In rural settings, males and females in the 26–65 age group had similar levels of depressive symptoms, but males ages 65 and older experienced more depression 27 28 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 17. Prevalence of depression and anxiety, by location, demographics, and seasonality 60% 50% 13.3 40% 30% 6.7 20% 7.9 7.0 5.5 6.0 5.5 38.9 5.3 6.5 5.4 10% 22.1 16.3 16.2 16.4 3.0 16.1 16.4 18.3 11.9 14.3 6.3 0% National Rural Urban 15–25 26–40 41–65 65+ Female Male Monsoon Dry years years years years Location Demographics Gender Season Depression Anxiety Note: Depression analysis is based on PHQ-9 scores and anxiety analysis on GAD-7. than females of the same age. Weather also played a role. Depressive symptoms were elevated during the dry season compared to the monsoon, with the most pronounced increases among the two oldest-age cohorts. The pattern was slightly different for younger females: more females ages 26–40 years were found to be depressed in the monsoon than in the dry season. Figure 19 shows the results for anxiety levels based on GAD-7 scores across the same parameters of age, gender, location, and age cohorts. The most pronounced difference was between urban and rural populations, with urban residents having substantially higher anxiety rates. Males ages 26–65 years in rural areas appeared more anxious than males and females ages 26 to 65 years or more. In urban areas, by contrast, younger females—those ages 15–40 years—were found to be more anxious than males and older women (26–65 years or more). During monsoon, females in Dhaka and Chattogram cities were more anxious than in the dry season. Anxiety levels for all urban and rural areas were similar in monsoon and the dry season. The prevalence estimates of depression and anxiety in Bangladesh indicate that substantial morbidity of these mental disorders remains under-addressed and is in urgent need of preventive and treatment initiatives. The most vulnerable for depression and anxiety are older, poorer, and disabled individuals. Accordingly, these populations should be accorded the highest priority for mental health services. Mental Health | 29 FIGURE 18. Prevalence of depression by location, age group, gender, and season ALL URBAN DHAKA AND CHITTAGONG CITIES RURAL 20 DRY SEASON 10 0 PHQ 9 SCORE 20 MONSOON 10 0 15–25 26–40 41–65 65+ 15–25 26–40 41–65 65+ 15–25 26–40 41–65 65+ years years years years years years years years years years years years Female Male Notes: Figure shows the distribution of PHQ 9 scores across location (vertical axis) and seasons (horizontal axis). The results are further broken down by age groups (box plots). The horizontal lines within each box plot show the median, while the top and bottom ranges show the 25th and 75th percentiles, respectively. The scatterplots represent gender (blue for male and pink for female). The orange line at score 10 represents the cutoff line for PHQ 9, beyond which the individual is likely to be associated with experiencing clinical depression. 30 | INFECTED AND STRESSED BY CLIMATE VARIABILITY FIGURE 19. Prevalence of anxiety by location, age group, gender, and season ALL URBAN DHAKA AND CHITTAGONG CITIES RURAL 20 15 DRY SEASON 10 5 GAD 7 SCORE 0 20 15 MONSOON 10 5 0 15–25 26–40 41–65 65+ 15–25 26–40 41–65 65+ 15–25 26–40 41–65 65+ years years years years years years years years years years years years Female Male Notes: Figure shows the distribution of GAD-7 scores across location (vertical axis) and seasons (horizontal axis). The results are further broken down by age groups (box plots). The horizontal lines within each box plot show the median, while the top and bottom ranges show the 25th and 75th percentiles, respectively. The scatterplots represent gender (blue for male and pink for female). The orange line at score 10 represents the cutoff line for GAD-7, beyond which the individual is likely to be associated with experiencing clinical depression. 6.2 CORRELATES OF DEPRESSION AND ANXIETY Correlates of depression and anxiety are presented in table B-6 (appendix B). Figures are marginal effects from an adjusted logit model. In determining the inclusion of predictor variables, the models were informed by social determinants (Allen et al. 2014) and a biopsychosocial approach (Engel 1977) to mental illness. Accordingly, the gender, age, and marital status of the respondent, as well as the gender of the household head, household size, education level, urban or rural location, and socioeconomic status as measured by asset quintiles have been used. Due to sluggish progress and the continued vulnerability of the Bangladeshi population to sudden economic, health, or environmental shocks, these measures, or shocks, have also been included as predictors of mental health. Any recent illness was also considered, as this outcome Mental Health | 31 can impact family members’ mental health and wellbeing. Disability is a strong predictor for developing mental health comorbidity, and accordingly was included in the models. In addition, seasonality—monsoon and dry season—has been incorporated to explore any effects on mental health and controlled for specific weather factors such as humidity and temperature. Lastly, the models control for individual WASH practices and community characteristics. Depression An increase in temperature lowers the probability of depression by 1.6 percent. Increasing age is significantly associated with increased probability of depression, with older populations substantially at higher risk (appendix B; table B-6; column 1). Compared to the youngest age group of 15–25 years, individuals ages 26–40 years are more likely to be depressed—5.2 percentage points higher. Similarly, the probability of developing depression is higher for the 41–65 year age group—9.6 percentage points—and for those over 65 years—16.3 percentage points compared to the 15–25-year-old age group. Overall, men are less likely to have depression—2 percentage points less—than women. Each additional year of education lowers the probability of depression by a modest 0.6 percentage points in absolute terms. Compared to people without a disability, individuals with a disability have a probability of developing depression that is higher by 14 percentage points. Compared to healthy people, the presence of illness increases the probability of developing depression by 5.7 percentage points. Increasing socioeconomic status, as measured by asset quintiles, indicate that those with more access to wealth and resources have lower probability of developing depression. As compared to the poorest quintile (Q1), individuals of Q2 were less likely to be depressed—2.7 percentage points—with lower likelihood for members of Q3 at 3.9 percentage points, Q4 at 4.8 percentage points and Q5 at 6.1 percentage points. A one-point increase in the WASH index has a protective effect of lowering the probability of depression by 1 percentage point. The experience of a household shock—environmental, health, or financial—increases the probability of depression by 4.6 percentage points. Living in a community that disposes of general waste in a designated place increases the probability of depression by 3.3 percentage points. Also observed was an increased likelihood of depression, at 1.9 percentage points, among those in the middle category of the community connectivity index, compared to the least connected communities. Anxiety Weather variables have a slightly different impact on anxiety than depression. Increases in mean humidity and mean temperature raises the probability of having anxiety by 0.3 percent and 0.8 percent, respectively. Age, however, has a similar effect on anxiety as depression, with older individuals more vulnerable to developing anxiety. Compared to those ages 15–25 years, an increase of 2.1 percentage points exists in the probability of being anxious for those ages 26–40 years, an increase of 1.9 percentage points for the age group 41–65, and an increase of 4.1 percentage points for those 65 and older (appendix B; table B-6; column 3). Men are more likely, by 1.1 percentage points higher, to have anxiety than women. Urban residents are more likely to be anxious (2.2 percentage points) than their rural counterparts. Having a disability increases 32 | INFECTED AND STRESSED BY CLIMATE VARIABILITY the likelihood of having anxiety by 3.4 percentage points. Like depression, increasing socioeconomic status has a protective effect against the probability of having anxiety— Q2: 2 percentage points; Q3: 2.1 percentage points; Q4: 2.5 percentage points; and Q5: 3 percentage points, compared to the base Q1. Experiencing a shock makes it 2.3 percentage points more likely for someone to become anxious. Living in a community with a designated general waste disposal location increases the probability of anxiety by 3.5 percentage points, while living in high- connectivity location decreases its likelihood by 3.2 percentage points. 7 Recommendations for Public Policy Based on the survey findings, the following public policy recommendations can be implemented with the assistance of existing institutions and platforms, and through the collaboration of decision makers, implementers, academia, and technical experts. Policy options are classified under two broad approaches: the first is documenting the knowns in order to be better informed with evidence and analyses of the effects of climate change and climate variability that are known to have occurred. This category includes both short- and medium-term actions and will help predict and mitigate risks. The second set of policy options consist of discovering the unknown, or not- so-well-known, in order to explore impacts or effects that have not been adequately researched and, hence, help in adaptation to climate change over the longer term. 7.1 DOCUMENTING THE KNOWN Strengthen surveillance of diseases. Surveillance involves two strands of information—entomological data collection—the study of vector or mosquito lifecycles—and the information system that links epide- miological data on disease incidence with climatic data associated with transmission ecology. The MoHFW needs to strengthen its routine surveillance mechanisms to track the incidence and prevalence of diseases. Second, the existing literature highlights the importance of using localized climatic conditions to predict the evolution of infectious diseases, after considering other important compounding factors such as the population’s immunity status, internal migration or mobility patterns, and implementation of vector control measures. In Ecuador, for example, the peak dengue season has shifted from the first trimester to the second trimester, although dengue transmission remains high in the hot and rainy seasons (Lowe et al. 2017). Hence, the need for the MoHFW to set up a climate-based dengue early-warning system that will use climate data to track potential disease outbreaks. The effectiveness of such systems is dependent on the capacity to collect accurate climate information and use it to forecast patterns (Lowe et al. 2017). This points to the imperative for 33 34 | INFECTED AND STRESSED BY CLIMATE VARIABILITY enhanced coordination between MoHFW and BMD to develop a climate-based, dengue early-warning system. Measure air quality to tackle an important compounding factor. As discussed in this report, air pollution is an important determinant of respiratory illnesses and other diseases. BMD is well positioned to track air pollution levels using instruments in the field as well as satellites. Satellites in the Geostationary Operational Environmental Satellites (GEOS) R series, and the joint polar satellite system, monitor particle pollution in the atmosphere. They track smoke particles from wildfires, airborne dust during dust and sandstorms, urban and industrial pollution, and ash from erupting volcanoes. The existing sources of information, coupled with localized ground-level information, can assist the government in tracking air pollution levels in Bangladesh. This information should be used to analyze how patterns of diseases are correlated and, accordingly, implement measures to reduce air pollution. Give urgent attention to mental health issues. Due to cultural norms, mental health issues such as depression and anxiety are largely ignored in Bangladesh and merit better understanding. Once the underlying situation is better understood, effective mitigation measures should be implemented accordingly. While women are in general at higher risk than men for depression, men are in general more susceptible to anxiety. Traditional gender norms and social values therefore indicate the necessity of gender-sensitive programmatic efforts. In addition to a more mainstreamed response through the provision of mental health services, community-based solutions for prevention and treatment are well warranted. Creation of peer support groups, for instance, could help mitigate feelings of loneliness, depression, and anxiety among the elderly. Additionally, training nonspecialists to detect and treat common mental disorders has been shown to be effective in many low- and middle-income settings. Such types of supported task-sharing approaches for mental health are worth consideration. 7.2 DISCOVERING THE NOT-SO-WELL-KNOWN Undertake further research to document accurately how climate change affects health and other sectors. Establishing the causal effects of climate change on health outcomes will entail the collection and analyses of long-term, household-level data and localized climatic conditions. The Ministry of Environment, Forest and Climate Change (MoEFCC) may be better placed to undertake such multisectoral research using household panel surveys over longer periods of time for all relevant sectors. Given its experience in dealing with large household-level surveys, the Bangladesh Bureau of Statistics may be able to assist. Such surveys would help quantify and eventually project the effects of climate change accurately, mitigate imminent risks, and identify and adapt to emerging issues. For example, it is not known if the outbreak of the novel coronavirus disease in 2020 is a direct or indirect effect of climate change. Similarly, the effects of heat on human health in Bangladesh need to be understood. Also to be prioritized is more detailed research on air pollution and zoonotic diseases. APPENDIX A: DATA AND METHODS SURVEY DESIGN AND SAMPLING STRATEGY The survey was designed using a two-stage stratified random sampling. The primary sampling units (PSUs) in the first stage were selected using probability-proportion- al-to-size (PPS) methods, using the 2011 Population and Housing Census. The 150 PSUs at the first stage were selected based on three strata to account for levels of congestion. Map A1 presents the distribution of the 150 PSUs across the country. The first stratum represents rural areas comprising 90 PSUs, the second representing Dhaka and Chattogram city corporation areas comprise 24 PSUs, and the third strata consists of 36 PSUs, representing all urban areas (table A1). The second stage of the selection process in each of the enumeration areas (EAs) began with a listing exercise. For very large EAs, a smaller section was delineated for the listing. The second level of stratification is defined as i) households with women and children under five years of age; ii) households with elderly people (65+ years); and iii) households with mixed demographics. Households were randomly selected from each stratum, with the predetermined ratio of 16:2:2. For EAs where the ratio could not be attained because of the absence of households in certain strata, replacements were obtained from the first category to arrive at a final number of 20 observations per EA. Table A1 provides more detail. TABLE A1. Household sample HOUSEHOLDS CLUSTER ROUND 1 ROUND 2 (AUG-SEP 2019) (JAN-FEB 2020) Dhaka and Chattogram cities 24 580 517 All urban areas (including Dhaka 36 866 838 and Chattogram) Rural areas 90 2,164 2,125 National 150 3,610 3,480 35 36 | INFECTED AND STRESSED BY CLIMATE VARIABILITY Sampling weights are calculated in two stages. For the first stage, the probability of selection of the sample was calculated separately for each stage and EA, using the following specification in equation 1: ∑ where the probability of the ith EA being selected from the jth stratum is presented by Pij1, and nj represents the number of selected EAs within each stratum. The number of households in the ith EA in the jth stratum is represented by Mi j while ∑Mi j represents the total number of households in the stratum. The probability of the second stage of the selection process is estimated using the following specification in equation 2: 2 = where the probability of a household being selected for the sample is represented by P ij2, and ht i j is the number of households of type t in EA i in stratum j selected to be surveyed from a total number of households (Ht i j) within each EA and the particular category. The overall likelihood of a particular household being selected from a particular stratum is therefore represented as the product of the two aforementioned probabilities, calculated as shown in equation 3: 1 2 × × ij The weight is subsequently constructed as the inverse of the likelihood of a particular household being selected (1/P i j). The first stage involves the calculation of the probability of each cluster being sampled within each stratum. The probability of selection of households within each stratum is calculated from the total number of listed households from each of the strata in each EA. The overall likelihood of a household being selected from each stratum is, therefore, calculated as the product of the two probabilities. The household weights are subsequently constructed as the inverse of the likelihood of a household being selected. DATA A structured questionnaire, directed toward the primary female member of the household, was used to collect information on an array of issues. The cascading questions first inquired as to whether any member of the household had fallen ill, followed by whether they had visited a doctor for the illnesses and had received a medical diagnosis. The subsequent set of questions collected detailed information about symptoms of illnesses in the event they had not obtained a medical diagnosis. This report considers three sets of primary outcomes: infectious diseases, persistent or chronic illnesses, and mental health. During the survey, respondents were asked about their morbidities. The types of diseases were grouped as infectious diseases and persistent illnesses. Recommendations for Public Policy | 37 FIGURE 20. Sample PSUs by enumeration areas Details were collected on healthcare-seeking behavior, conditional on reporting either infectious or persistent illnesses. Questions related to whether healthcare was sought, the duration between the illness and the care received, details on the type and location of the care provider, mode and duration of travel time, and the direct costs (consultation, medication, diagnostics) and indirect costs (transportation) of care. Data on the status of common mental disorders, namely depression and anxiety, were collected. Depression is measured using the Patient Health Questionnaire-9 (PHQ-9), a commonly used depression screening instrument, comprising 9 items on a 4-point Likert response scale. The scale has demonstrated good sensitivity and specificity in identifying depression in both clinical and nonclinical settings (Levis et al. 2019; Spitzer et al. 1999). The PHQ-9 has been validated for use in Bengali-speak- ing populations (Chowdhury et al. 2004), and has been widely used in Bangladesh 38 | INFECTED AND STRESSED BY CLIMATE VARIABILITY (Arafat et al. 2018; Islam et al. 2020; Islam et al. 2015; Mamun et al. 2019a; Moonajilin et al. 2020; Roy et al. 2012). For anxiety, the Generalized Anxiety Disorder-7 (GAD-7), a 7-item, 4-point, Likert-style anxiety screening scale was used. The GAD-7 has good reliability and validity for measuring generalized anxiety disorder (Spitzer et al. 2006) and has also been used previously in clinical and nonclinical research settings in Bangladesh (Hossain et al. 2019; Islam et al. 2020; Moonajilin et al. 2020). Informed by best practice standards in existing literature (Levis et al. 2019; Manea et al. 2015; Spitzer et al. 2006), a clinical cutoff score of 10 has been used in this report to establish the presence of depression and anxiety for both PHQ-9 and GAD-7 scores. Background information was collected also on a host of related issues. A roster amassed information on each household’s demographic composition, while disability status was compiled using the short version of the Washington Group Questions.1 A composite socioeconomic index accounting for unique urban and rural charac- teristics was created from physical household attributes—wall and roof material, area per capita, and availability of a separate kitchen—and a roster of durable assets was created using methods outlined in NIPORT 2016. Quintiles of the continuous index were used in the rest of the report. A similar index for water, sanitation, and hygiene (WASH) was constructed using WASH attributes of the household, such as accessibility to sanitary facilities, type of water used, hygiene-oriented behavior such as hand washing, and so on. The WASH index was categorized as low, medium, and high, with low indicating access to fewer WASH facilities and lower use, and so on. A community questionnaire was administered with the goal of assessing commu- nity-level characteristics such as relative poverty scenario, connectivity, and cleanliness. A connectivity index was constructed using community-level outcomes such as its size, construction material of the main road, average hours of electricity availability, whether a community-level system is available for collecting refuse, experience of waterlogging the previous year, and access to community level services such as transport, a post office, schools, and banks. Community-level information was collected from key respondents deemed to have a good understanding of the issues surrounding the locality, typically the community leaders. A range of information was collected related to socioeconomic status of average households, issues faced such as water logging or road conditions, access to basic services such as schools, police, post offices, markets, financial services, and so on. CALCULATION OF CORRELATION COEFFICIENTS Correlation coefficients presented for infectious diseases and mental health were estimated using the weighted linear specification given in equation 4: = α + β1θ + β2θ * + + ∅ + ε where Y it represents the physical or mental health outcome for the ith individual at time t living in the zth PSU. β2 represents the coefficient of the interaction term identifying the seasonal differentials. PSU-level fixed effects are represented by ∅z, while tt reflects time trends. The decision to use PSU levels as opposed to exploiting the panel nature of the data through individual- or household-level fixed effects in the models was primarily because of the absence of positive illness outcomes in the sample. Approximately five percent of the total sample reported an illness during the two surveys. Similarly, the issue precluded further assessment of heterogeneity Recommendations for Public Policy | 39 through the interaction of demographic or socioeconomic and weather variables. The use of PSU-level fixed effects captures and controls for locational trends over the two seasons in the models. The idiosyncratic error term is represented by εitz. Interpretation of binary explanatory outcomes are in percentage changes (for example, 0.03 = 3 percentage points) and in percent for linear outcomes (0.03 = 3 percent). WEATHER DATA Weather data from 1976 to 2019 were collected from BMD. The BMD collects data from 43 weather stations across the country. Figure 21 shows the distribution of these stations relative to population density. These 43 stations have been set up over an extended period of time, which is why data for some of the earlier years, between 1976 and 1990, are from fewer weather stations. FIGURE 21. Weather station locations in Bangladesh 40 | INFECTED AND STRESSED BY CLIMATE VARIABILITY APPROVAL AND CLEARANCE PROCESSES Approval from the Bangladesh Medical Research and Council (BMRC) was obtained before starting the survey. All ethical protocols and standards of BMRC were adhered to during fieldwork. The procedures listed below were followed: • Ensuring the privacy of information collected by keeping it anonymous, without recording respondents’ names or attaching their names to any data, and using instead a unique identity number to the household; • Obtaining the written informed consent of the interviewee; • Not collecting data on nationality or religion; • Excluding respondents’ personal information from data files; and • Presenting results in aggregate form, without identifying any individuals. A concept note outlining the overall objectives and structure of the report was reviewed and approved by World Bank senior management. The note was also shared with the Climate Change and Health Promotion Unit (CCCHPU) and the Institute of Epidemiology and Disease Control Research (IEDCR) of the Ministry of Health and Family Welfare (MoHFW) of the Government of Bangladesh before finalizing the survey instrument and sample size. The draft report was also shared with CCHPU and IEDCR before finalization. For quality assurance, the report was reviewed at an internal World Bank meeting, chaired by Mercy Tembon, Country Director for Bangladesh and Bhutan. The review was organized to discuss the methodology and findings of the report and the potential implications of the conclusions and recommendations for Bangladesh. Based on detailed discussions during the internal review and extensive comments provided by the reviewers, including six World Bank experts—Dhushyanth Raju (Lead Economist), Shiyong Wang (Senior Health Specialist), Tamer Samah Rabie (Lead Health Specialist), Anna Koziel (Senior Health Specialist), Stephen Geoffrey Dorey (Health Specialist) and Muthukumara Mani (Lead Economist)—the report was finalized. APPENDIX B: SUPPLEMENTARY TABLES ON DEMOGRAPHY, SOCIOECONOMIC CHARACTERISTICS, AND DISEASE PATTERNS BY LOCATION TABLE B 1. Demographic characteristics of the sample at baseline URBAN Dhaka and NATIONAL All Chattogram RURAL TEST Urban- Cities- Mean SD Mean SD Mean SD Mean SD Rural Rural VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Male/Female (1/0) 0.50 0.50 0.50 0.50 0.52 0.50 0.50 0.50 n.a. n.a. Age 28.0 19.5 28.3 18.9 27.8 17.5 27.9 19.6 n.a. n.a. MARITAL STATUS Married (1/0) 0.51 0.50 0.51 0.50 0.49 0.50 0.51 0.50 n.a. n.a. Never married (1/0) 0.23 0.42 0.24 0.43 0.26 0.44 0.23 0.42 n.a. * Other (1/0) 0.26 0.44 0.25 0.43 0.24 0.43 0.26 0.44 n.a. n.a. Years of education (cont.) 4.92 4.46 5.98 4.91 6.06 5.08 4.64 4.30 *** *** Sex of hh head male/female (1/0) 0.92 0.28 0.91 0.28 0.88 0.33 0.92 0.28 n.a. *** Age of hh head 46. 13.3 46.3 13.1 44.3 12.8 46.2 13.4 n.a. *** Years of education of hh head (cont.) 4.51 4.69 5.80 5.32 6.07 5.66 4.18 4.45 *** *** Individual member with a disability (1/0) 0.13 0.34 0.14 0.35 0.15 0.36 0.13 0.34 n.a. n.a. N = 15,383 Note: Table shows weighted means. Test (columns 9 and 10) shows significance levels from a weighted t-test. * p<0.1; ** p<0.05; *** p<0.01. n.a.= not applicable; hh = household; cont. = continued. 41 42 | INFECTED AND STRESSED BY CLIMATE VARIABILITY TABLE B 2. Socioeconomic characteristics of the sample at baseline URBAN Dhaka and NATIONAL All Chattogram RURAL TEST Urban- Cities- Mean SD Mean SD Mean SD Mean SD Rural Rural VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) WALL MATERIAL: Straw/ mud 0.19 0.34 0.04 0.14 0.00 0.00 0.22 0.37 *** *** Tin 0.47 0.50 0.32 0.46 0.13 0.33 0.51 0.50 *** *** Brick/cement 0.34 0.48 0.64 0.48 0.87 0.33 0.27 0.44 *** *** ROOF MATERIAL: Tin 0.85 0.35 0.68 0.47 0.48 0.50 0.90 0.30 *** *** Brick/cement 0.13 0.34 0.31 0.46 0.52 0.50 0.08 0.28 *** *** Has access to electricity (1/0) 0.88 0.32 0.98 0.13 1.00 0.06 0.86 0.35 *** *** Stove type: clean(1) / unclean (0) 0.20 0.40 0.56 0.50 0.97 0.17 0.10 0.30 *** *** Rooms per capita 0.54 0.28 0.54 0.31 0.46 0.31 0.54 0.27 n.a. *** Has separate dining room (1/0) 0.29 0.45 0.27 0.44 0.21 0.41 0.30 0.46 *** *** ASSET QUINTILES: Q1: Poorest 0.24 0.43 0.10 0.30 0.06 0.24 0.28 0.45 *** *** Q2 0.23 0.42 0.14 0.35 0.08 0.27 0.25 0.43 *** *** Q3 0.20 0.40 0.17 0.38 0.15 0.35 0.21 0.41 *** *** Q4 0.17 0.38 0.23 0.42 0.26 0.44 0.15 0.36 *** *** Q5: Richest 0.16 0.37 0.35 0.48 0.45 0.50 0.11 0.32 *** *** N = 15,383 Note: Table shows weighted means. Test (columns 9 and 10) shows significance levels from a weighted t-test.* p<0.1; ** p<0.05; *** p<0.01. n.a.= not applicable. TABLE B 3. Infectious diseases across locations URBAN Dhaka and NATIONAL All Chattogram RURAL TEST Urban- Cities- Mean SD Mean SD Mean SD Mean SD Rural Rural VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Any infectious disease (1/0) 0.06 0.24 0.06 0.24 0.08 0.27 0.06 0.23 n.a. ** Waterborne disease (1/0) 0.14 0.35 0.13 0.34 0.19 0.40 0.14 0.35 n.a. n.a. Respiratory disease (1/0) 0.64 0.48 0.65 0.48 0.50 0.50 0.63 0.48 n.a. * Vector-borne disease (1/0) 0.22 0.42 0.22 0.42 0.30 0.46 0.22 0.42 n.a. n.a. Individual had a common cold (1/0) 0.09 0.29 0.09 0.29 0.06 0.25 0.10 0.29 n.a. *** Note: Table shows weighted means. Waterborne, respiratory and vector-borne diseases are conditional on whether an individual reported an infec- tious disease (excluding the common cold) in the four weeks preceding the survey. Whether an individual had a common cold represents the sample population. Test (columns 9 and 10) shows significance levels from a weighted t-test. * p<0.1; ** p<0.05; *** p<0.01. n.a.= not applicable. Recommendations for Public Policy | 43 TABLE B 4. Correlates of contracting any seasonal illness (excluding the common cold) COEFFICIENT STD. ERROR (1) (2) Season (1=dry/monsoon=0) -0.197*** 0.034 AGE (BASE: 0 TO 5 YEARS) 6 to 19 years -0.050*** 0.007 20 to 64 years -0.039*** 0.006 65+ years -0.011 0.010 AGE (BASE: 0 TO 5 YEARS) * DRY SEASON 6 to 19 years * Dry season 0.016* 0.009 20 to 64 years * Dry season 0.020** 0.008 65+ years * Dry season 0.010 0.014 Gender (1=male/0=female) 0.005* 0.003 Individual is disabled (1=y/0=n) 0.018*** 0.004 Never married (1=y/0=n) -0.008* 0.004 Education (continued years) -0.001** 0.000 Male head of household (1=y/0=n) -0.011** 0.005 Household size -0.005*** 0.001 SOCIOECONOMIC QUINTILES (BASE: BOTTOM 2 QUINTILES) Q3 -0.015*** 0.004 Q4 -0.008** 0.004 Q5 (Richest) -0.010** 0.005 WASH INDEX (BASE: LOWEST) WASH Index (mid) -0.012*** 0.003 WASH Index (high) -0.009** 0.004 Household experienced a shock (1=y/0=n) 0.010* ** 0.003 COMMUNITY CHARACTERISTICS: Number of households in community_ihs 0.064 0.106 Trash deposited in a single location (1=y/0=n) -0.063* 0.034 CONNECTIVITY INDEX (BASE: LEAST) Connectivity index (mid) 0.005 0.059 Connectivity index (high) -0.085* 0.046 Average hours of electricity available_ihs -0.438 0.649 LOCATION (BASE: RURAL AREAS) Cities of Dhaka and Chattogram -0.053 0.067 Cities of Dhaka and Chattogram* dry season 0.018 0.012 WEATHER (MEANS FROM 2 MONTHS PRECEDING SURVEY) Humidity (2-month mean, pre-survey) -0.002* 0.001 Mean temperature (2-month mean, pre-survey) -0.014*** 0.003 R2 = 0.03 N = 30,091 Note: Table shows coefficients from a weighted linear specification. Dependent variable is whether the individual experienced any seasonal illness. Variables with suffix “_ihs” refer to inverse hyperbolic sine transformations. The models account for PSU-level heterogeneity. *, ** and *** indicate significance at 10 %, 5% and 1% per cent respectively. 44 | INFECTED AND STRESSED BY CLIMATE VARIABILITY TABLE B 5. Correlates of contracting vector-borne, waterborne, or respiratory infections WATERBORNE DISEASES RESPIRATORY INFECTIONS VECTOR-BORNE DISEASES Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error (1) (2) (3) (4) (5) (6) Season (1=dry season/monsoon=0) -0.124 0.247 0.337 0.313 -0.214 0.255 AGE (BASE: 0 TO 5 YEARS) 6 to 19 years (adolescents) -0.169*** 0.046 0.021 0.058 0.148*** 0.047 20 to 64 years (adults) -0.084** 0.038 -0.009 0.049 0.093** 0.040 65+ years (elderly) 0.035 0.056 0.101 0.071 -0.136** 0.058 AGE (BASE: 0 TO 5 YEARS) * DRY SEASON 6 to 19 years * Dry season -0.190*** 0.064 0.278*** 0.081 -0.087 0.066 20 to 64 years * Dry season -0.280*** 0.053 0.324*** 0.067 -0.044 0.054 65+ years * Dry season -0.495*** 0.079 0.368*** 0.100 0.127 0.081 Gender (1=male/0=female) -0.005 0.019 0.013 0.025 -0.008 0.020 Individual is disabled (1=y/0=n) 0.038 0.026 -0.088*** 0.033 0.050* 0.027 Never married (1=y/0=n) 0.022 0.037 -0.056 0.046 0.034 0.038 Education (cont. years) 0.001 0.003 0.002 0.004 -0.003 0.003 Male head of household (1=y/0=n) -0.041 0.037 0.001 0.047 0.040 0.038 Household size 0.003 0.006 -0.002 0.008 -0.001 0.007 SOCIOECONOMIC QUINTILES (BASE: BOTTOM 2 QUINTILES) Q3 0.028 0.028 -0.048 0.036 0.020 0.029 Q4 0.049 0.031 -0.007 0.040 -0.042 0.032 Q5 (Richest) -0.049 0.036 0.083* 0.046 -0.034 0.037 WASH INDEX (BASE: LOWEST) WASH Index (mid) 0.006 0.027 -0.034 0.035 0.028 0.028 WASH Index (high) -0.035 0.033 0.023 0.042 0.012 0.034 Household experienced a shock (1=y/0=n) -0.005 0.024 -0.057* 0.031 0.062** 0.025 COMMUNITY CHARACTERISTICS: Number of households in community_ihs -0.682 0.608 0.625 0.772 0.057 0.628 Trash deposited in a single location -0.279 0.297 -0.048 0.377 0.327 0.307 (1=y/0=n) CONNECTIVITY INDEX (BASE: LEAST) Connectivity index (mid) -0.240 0.341 0.417 0.434 -0.176 0.353 Connectivity index (high) -0.186 0.348 0.600 0.442 -0.414 0.360 Average hours of electricity available_ihs 4.862 3.719 -6.092 4.726 1.230 3.842 LOCATION (BASE: RURAL AREAS) Dhaka and Chattogram cities 0.306 0.408 -0.369 0.518 0.063 0.421 Dhaka and Chattogram cities* Dry season -0.156* 0.086 0.194* 0.109 -0.038 0.089 WEATHER (MEANS FROM 2 MONTHS PRECEDING SURVEY) Humidity (2-month mean, pre-survey) -0.016** 0.007 0.015* 0.009 0.000 0.007 Mean temperature (2-month mean, -0.042* 0.022 0.057** 0.028 -0.014 0.023 pre-survey) R2 0.24 0.22 0.24 N 1,604 1,604 1,604 Note: Table shows coefficients from a weighted linear specification. Dependent variable is whether the individual experienced waterborne, respiratory or vector-borne illnesses. Variables with suffix “_ihs” refer to inverse hyperbolic sine transformations. The models account for PSU-level heterogeneity. *, ** and *** indicate significance at 10, 5 and 1 percent respectively. Recommendations for Public Policy | 45 TABLE B 6. Correlates of depression and anxiety DEPRESSION ANXIETY Coefficient Std. Error Coefficient Std. Error (1) (2) (3) (4) Season (1=dry season/0=monsoon) -0.177 0.109 0.073 0.066 AGE (BASE: 15 TO 25 YEARS) 26 to 40 years 0.052*** 0.017 0.021** 0.010 41 to 65 years 0.096*** 0.019 0.019* 0.011 65+ years 0.163*** 0.024 0.041*** 0.013 Gender (1=male/0=female) -0.020* 0.011 0.011* 0.006 Education (cont. years) -0.006*** 0.001 -0.001 0.001 Never married (1=y/0=n) 0.013 0.027 -0.014 0.012 Location (1=urban/0=rural) 0.004 0.012 0.022*** 0.008 Male head of household (1=y/0=n) -0.013 0.020 0.003 0.008 Household size 0.001 0.003 -0.001 0.002 Individual is disabled (1=y/0=n) 0.140*** 0.018 0.034*** 0.010 Any household illness (1=y/0=n) 0.057*** 0.021 0.005 0.011 SOCIOECONOMIC QUINTILES (BASE: POOREST QUINTILE) Q2 (Poorer) -0.027* 0.016 -0.020** 0.008 Q3 (Middle) -0.039** 0.016 -0.021*** 0.008 Q4 (Richer) -0.048*** 0.016 -0.025*** 0.009 Q5 (Richest) -0.061*** 0.020 -0.030*** 0.011 WASH Index -0.010*** 0.003 -0.002 0.002 Household experienced a shock (1=y/0=n) 0.046*** 0.012 0.023*** 0.007 COMMUNITY CHARACTERISTICS: Number of households in community 0.000*** 0.000 0.000*** 0.000 Trash deposited in a single location (1=y/0=n) 0.033*** 0.012 0.035*** 0.007 Connectivity index (base: least) Connectivity index (mid) 0.019* 0.011 0.004 0.007 Connectivity index (high) -0.020 0.016 -0.032*** 0.007 Average hours of electricity available 0.000 0.001 0.000 0.001 WEATHER (MEANS FROM 2 MONTHS PRECEDING SURVEY) Humidity (2-month mean pre-survey) 0.000 0.002 0.003*** 0.001 Mean temperature (2-month mean pre-survey) -0.016* 0.009 0.008* 0.005 Pseudo R^2 0.13 0.08 N 7086 7086 Note: Table shows marginal effects from weighted logit models. Dependent outcomes are whether an individual suffers from depression or anxiety. Models account for PSU-level heterogeneity. *, ** and *** indicate significance at the 10, 5 and 1 percent respectively. 46 | INFECTED AND STRESSED BY CLIMATE VARIABILITY REFERENCES Allen, J., R. Balfour, R. Bell, and M. Marmot. 2014. “Social Determinants of Mental Health.” Int Rev Psychiatry 26, no. 4: 392–407. doi:10.3109/09540261.2014.928270. Arafat, S.M.Y., A. Shahoriar, A. Faruq, Z. Sojib Bin, and A. Amin. 2018. “Depression in Spinal Cord Injury Patients: A Cross-Sectional Observation with PHQ-9 in a Rehabilitation Center of Bangladesh.” J Behav Health 7, no. 1: 36–40. doi: 10.5455/jbh.20171004071405. Banu, S., W. Hu, Y. Guo, C. Hurst, and S. Tong. 2014. “Projecting the Impact of Climate Change on Dengue Transmission in Dhaka, Bangladesh.” Environment International 63 (February): 137–142. https://doi.org/10.1016/j.envint.2013.11.002. Berry, H.L., B.J. Kelly, I.C. Hanigan, J.H. Coates, A.J. McMichael, J.A. Welsh, and T. Kjellstrom. 2008. Rural Mental Health Impacts of Climate Change. Commissioned Report for the Garnaut Climate Change Review. Canberra: The Australian National University. http://www.garnautreview.org. au/CA25734E0016A131/WebObj/03-DMentalhealth/$File/03-D%20Mental%20health.pdf. Berry, H. 2009. “Pearl in the Oyster: Climate Change as a Mental Health Opportunity.” Aust Psychiatry 17, no. 6: 453–56. https://doi.org/10.1080/10398560903045328. Berry, H.L., K. Bowen, and T. Kjellstrom, 2010. “Climate Change and Mental Health: A Causal Pathways Framework.” Int J Public Health 55, no. 2 (April): 123–32. doi: 10.1007/s00038-009-0112-0. Bourque. F., and A.C. Willox. 2014. “Climate Change: The Next Challenge for Public Mental Health?” Int Rev Psychiatry 26, no. 4 (August): 415–422. doi: 10.3109/09540261.2014.925851. Campbell-Lendrum, D., L. Manga, M. Bagayoko, and J. Sommerfeld. 2015. “Climate Change and Vector-Borne Diseases: What Are the Implications for Public Health Research and Policy?” Phil. Trans. R. Soc. B 370: 20130552. http://dx.doi.org/10.1098/rstb.2013.0552. Chowdhury, F.R., Q.S.U. Ibrahim, M.S. Bari, M.M.J. Alam, S.J. Dunachie, A.J. Rodriguez-Morales, and M.I. Patwary. 2018. “The Association between Temperature, Rainfall and Humidity with Common Climate-Sensitive Infectious Diseases in Bangladesh.” PLoS ONE 13, no. 6: e0199579. Clayton, S., C. Manning, K. Krygsman, and M. Speiser. 2017. Mental Health and Our Changing Climate: Impacts, Implications, and Guidance. Washington, DC: American Psychological Association and ecoAmerica. Costello, A. M. Abbas, A. Allen, S. Ball, S. Bell, R. Bellamy, and S. Friel, et al. 2009. “Managing the Health Effects of Climate Change.” The Lancet 373, no. 967 (May): 1659–1734. Coyle, K.J., and L. Van Susteren, 2017. The Psychological Effects of Global Warming on the United States: And Why the US Mental Health Care System is Not Adequately Prepared. National Wildlife Federation. http://www.climateaccess.org/sites/default/files/NWF_Psychological%20Effects.pdf. Doherty, T.J. and S. Clayton, 2011. “The Psychological Impacts of Global Climate Change.” Am Psychol 66, no. 4 (May–June): 265–276. https://www.apa.org/pubs/journals/releases/amp-66-4-265.pdf. Engel, G. L. 1977. “The Need for a New Medical Model: A Challenge for Biomedicine.” Science 196, no. 4286, 129. doi:10.1126/science.847460. Fouque, F., and J. C. Reeder. 2019. “Impact of Past and On-Going Changes on Climate and Weather on Vector-Borne Diseases Transmission: A Look at the Evidence.” Infectious Diseases of Poverty 8, no. 51. Government of Bangladesh. 2008a. Cyclone Sidr in Bangladesh: Damage, Loss and Needs Assessment for Disaster Recovery and Reconstruction. Dhaka: GOB. Government of Bangladesh, Ministry of Environment and Forests. 2008b. Bangladesh Climate Change Strategy and Action Plan 2008. https://www.sdnbd.org/moef.pdf. Hashizume, M., B. Armstrong, S. Hajat, Y. Wagatsuma, A.S. Faruque, T. Hayashi, and D.A. Sack. 2007. “Association between Climate Variability and Hospital Visits for Non-Cholera Diarrhoea in Bangladesh: Effects and Vulnerable Groups.” International Journal of Epidemiology 36: 1030–1037. Hossain, M., E. Noiri, and K. Moji. 2011. “Climate Change and Kala-Azar,” in Kala Azar in South Asia, 127–137. Dordrecht, Netherlands: Springer doi: 10.1007/978-94-007-0277-6_12. IPCC. 2014. Climate Change 2014: Synthesis Report. A Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Edited by the Core Writing Team, R.K. Pachauri, and L.A. Meyer. Geneva: IPCC. Islam, S., R. Akter, T. Sikder, and M.D. Griffiths. 2020. “Prevalence and Factors Associated with Depression and Anxiety Among First-Year University Students in Bangladesh: A Cross-Sec- tional Study.” International Journal of Mental Health and Addiction. (March). doi:10.1007/ s11469-020-00242-y. Recommendations for Public Policy | 47 Islam, S.M.S., L.B. Rawal, and L.W. Niessen. 2015. “Prevalence of Depression and Its Associated Factors in Patients with Type 2 Diabetes: A Cross-Sectional Study in Dhaka, Bangladesh.” Asian Journal of Psychiatry 17: 36–41. doi:10.1016/j.ajp.2015.07.008. Lancet editorial. 2019. “Health and Climate Change: Making the Link Matter.” The Lancet 394, no. 10211 (November): 1779-1878, e36. https://doi.org/10.1016/S0140-6736(19)32756-4. Levis, B., A. Benedetti, and B.D. Thombs. 2019. “Accuracy of Patient Health Questionnaire-9 (PHQ-9) for Screening to Detect Major Depression: Individual Participant Data Meta-Analysis.” BMJ 365 (April), l1476. doi: https://doi.org/10.1136/bmj.l1476. Lowe, R., A.M. Stewart-Ibarra, D. Petrova, M. Garcia-Díez, M.J. Borbor-Cordova, R. Mejía, M. Regato, et al. 2017. “Climate Services for Health: Predicting the Evolution of the 2016 Dengue Season in Machala, Ecuador.” The Lancet Planet Health 1, no 4 (July): e142–51. https://doi. org/10.1016/S2542-5196(17)30064-5. Manea, L., S. Gilbody, and D. McMillan. 2015. A Diagnostic Meta-Analysis of the Patient Health Questionnaire-9 (PHQ-9) Algorithm Scoring Method as a Screen for Depression. General Hospital Psychiatry 37, no. 1: 67–75. https://doi.org/10.1016/j.genhosppsych.2014.09.009. Mani, M., and L. Wang. 2014. Climate Change and Health Impacts: How Vulnerable is Bangladesh and What Needs to Be Done? Washington DC: World Bank Group. http://documents.worldbank. org/curated/en/836581468004786339/Climate-change-and-health-impacts-how-vulnerable-is- Bangladesh-and-what-needs-to-be-done. McMichael, T. 2012. “Health Risks, Present and Future, from Global Climate Change.” BMJ 344 (March): e1359. doi:10.1136/bmj.e1359. Moonajilin, M.S., M.E. Rahman, and M.S. Islam. 2020. “Relationship between Overweight/Obesity and Mental Health Disorders among Bangladeshi Adolescents: A Cross-Sectional Survey.” Obesity Medicin. 18, no. 100216. https://doi.org/10.1016/j.obmed.2020.100216. Nahar, N., Y. Blomstedt, Y., B. Wu, I. Kandarina, L. Trisnantoro, and J. Kinsman. 2014. “Increasing the Provision of Mental Health Care for Vulnerable, Disaster-Affected People in Bangladesh.” BMC Public Health 14, no. 708. https://doi.org/10.1186/1471-2458-14-708. Norris, F.H. 2005. Psychosocial Consequences of Natural Disasters in Developing Countries: What Does Past Research Tell Us about the Potential Effects of the 2004 Tsunami? Washington DC: National Center for PTSD and U.S. Department of Veterans Affairs. Paul, B.K., 2009. “Why Relatively Fewer People Died? The Case of Bangladesh’s Cyclone Sidr.” Nat Hazards 50: 289–304. 10.1007/s11069-008-9340-5. Patz, G.A., J.P. McCarty, S. Hussein, U. Confalonieri, and N.D. Wet. 2003. Climate Change and Infectious Diseases, Climate Change and Human Health: Risks and Responses. Geneva: World Health Organization, 103–32. Rahman, M.M., S. Ahmad, A.S. Mahmud, M. Hassan-uz-Zaman, M.A. Nahian, A. Ahmed, Q. Nahar, et al. 2019. “Health Consequences of Climate Change in Bangladesh: An Overview of the Evidence, Knowledge Gaps and Challenges.” WIREs (Wiley Interdisciplinary Reviews) Climate Change, e601. https://doi.org/10.1002/wcc.601. Reid, H.L., U. Haque, S. Roy, N. Islam, and A.C. Clements. 2012. “Characterizing the Spatial and Temporal Variation of Malaria Incidence in Bangladesh, 2007.” Malaria Journal 11, no. 170. Roy, T., C.E. Lloyd, M. Parvin, K.G.B. Mohiuddin, and M. Rahman. 2012. “Prevalence of Co-Morbid Depression in Out-Patients with Type 2 Diabetes Mellitus in Bangladesh.” BMC Psychiatry 12, no. 1: 123. doi:10.1186/1471-244X-12-123. Spitzer, R.L., K. Kroenke, and J.B. Williams. 1999. “Validation and Utility of a Self-Report Version of PRIME-MD: The PHQ Primary Care Study. Primary Care Evaluation of Mental Disorders, Patient Health Questionnaire.” JAMA 282, no. 18: 1737–1744. doi:10.1001/jama.282.18.1737. Swim, J., S. Clayton, T. Doherty, R. Gifford, G. Howard, J. Reser, P. Stern, et al. 2009. Psychology and Global Climate Change: Addressing a Multi-Faceted Phenomenon and Set of Challenges. A Report by the American Psychological Association’s Task Force on the Interface between Psychology and Global Climate Change. Washington: American Psychological Association. Takaro, T.K., K. Knowlton, and J.R. Balmes. 2013. “Climate Change and Respiratory Health: Current Evidence and Knowledge Gaps.” Expert Rev. Respir. Med. 7, no. 4: 349–361. Watts, N., M. Amann, N. Arnell, S. Ayeb-Karlsson, K. Belesova, H. Berry, T. Bouley, et al. 2018. “The 2018 Report of the Lancet Countdown on Health and Climate Change: Shaping the Health of Nations for Centuries to Come.” The Lancet 392, no. 10163: 2479–2514. http://dx.doi.org/10.1016/ S0140-6736(18)32594-7. 48 | INFECTED AND STRESSED BY CLIMATE VARIABILITY Watts, N., M. Amann, S. Ayeb-Karlsson, K. Belesova, T. Bouley, M. Boykoff, P. Byass, et al. 2018b. “The Lancet Countdown on Health and Climate Change: From 25 Years of Inaction to a Global Transformation for Public Health.” The Lancet 391: 581–630. Watts, N., M. Amann, N. Arnell, S. Ayeb-Karlsson, K. Belesova, M. Boykoff, P. Byass, W. Cai, et al. 2019. “The 2019 Report of the Lancet Countdown on Health and Climate Change: Ensuring That the Health of a Child Born Today is Not Defined by a Changing Climate.” The Lancet. 394: 1836–78. https://doi.org/10.1016/S0140-6736(19)32596-6. Watts, N., M. Amann, N. Arnell, S. Ayeb-Karlsson, J. Beagley, K. Belesova, M. Boykoff, et al. 2020. “The 2020 Report of the Lancet Countdown on Health and Climate Change: Responding to Converging Crises.” The Lancet 397, no. 10269 (December): 129 –170. https://doi.org/10.1016/ S0140-6736(20)32290-X. Weissbecker, I., ed. 2011. Climate Change and Human Well-Being: Global Challenges and Opportunities. Berlin: Springer Science + Business Media. https://doi.org/10.1007/978-1-4419-9742-5. Willox, A.C., S.L. Harper, V.L. Edge, K. Landman, K. Houle, and J.D. Ford. 2013a. “The Land Enriches the Climatic and Environmental Soul: On Change, Affect, and Emotional Health and Well-Being in Rigolet, Nunatsiavut, Canada.” Emotion Space Soc. 6 (February): 14–24. https://doi.org/10.1016/j. emospa.2011.08.005. Willox, A.C., S.L. Harper, J.D. Ford, V.L. Edge, K. Landman, K. Houle, S. Blake, et al. 2013b. “Climate Change and Mental Health: An Exploratory Case Study from Rigolet, Nunatsiavut, Canada.” Climatic Change 121, no. 2: 255–70. Willox, A.C., E. Stephenson, J. Allen, F. Bourque, A. Drossos, S. Elgarøy, M.J. Kral, et al. 2015. “Examining Relationships between Climate Change and Mental Health in the Circumpolar North.” Reg Environ Change. https://doi.org/10.1007/s10113-014-0630-z. World Bank. n.d. Climate Change Knowledge Portal. Washington, DC: World Bank Group. https:// climateknowledgeportal.worldbank.org. World Bank. 2012. Turn Down the Heat: Why a 4°C Warmer World Must Be Avoided. Potsdam Institute for Climate Impact Research and Climate Analytics Report No. 74455. Washington, DC: World Bank Group. World Health Organization. 2003. Climate Change and Human Health: Risk and Responses. Geneva: WHO. World Health Organization. 2009. Mental Health Systems in Selected Low- and Middle-Income Countries: A WHO-AIMS Cross-National Analysis. https://www.who.int/mental_health/evidence/ WHO-AIMS/en. World Health Organization and the United Nations Framework Convention on Climate Change. 2015. Climate and Health Country Profile 2015 Bangladesh. Geneva: WHO. World Health Organization. 2016. “An Estimated 12.6 Million Deaths Each Year are Attributable to Unhealthy Environments.” News release (March 16). http://www.who.int/mediacentre/news/ releases/2016/deaths-attributable-to-unhealthy-environments/en. Wu, X., Y. Lu, S. Zhou, L. Chen, and B. Xu. 2016. “Impact of Climate Change on Human Infectious Diseases: Empirical Evidence and Human Adaptation.” Environment International 86: 14–23. https://doi.org/10.1016/j.envint.2015.09.007. Wu, J., M. Yunus, M. Ali, M.E Escamilla. 2018. “Influences of Heatwave, Rainfall, and Tree Cover on Cholera in Bangladesh.” Environment International 120 (November): 304–311. Zhang, Q., Y. Chen, Y. Fu, T. Liu, Q. Zhang, P. Guo, and W. Ma. 2019, “Epidemiology of Dengue and the Effect of Seasonal Climate Variation on its Dynamics: A Spatio-Temporal Descriptive Analysis in the Chao-Shan Area on China’s Southeastern Coast.” BMJ Open 9, no. 5 (May): e024197. doi: 10.1136/bmjopen-2018-024197; https://bmjopen.bmj.com/content/bmjopen/9/5/ e024197.full.pdf. Recommendations for Public Policy | 49 NOTES 1 In the report, depression is measured using a locally contextualized version of the Patient Health Questionnaire 9 (PHQ-9). Anxiety is measured using a locally contextualized version of the Generalized Anxiety Disorder 7 (GAD-7). 2 Data are representative of both urban and rural areas. The sampling distinguishes between two kinds of urban areas: major metropolises such as Dhaka and Chattogram cities, given their distinctive nature, and other urban areas. To account for seasonality, data from the same households are collected, first during the monsoon and, next, during the dry seasons. 3 The heat index is available at the National Weather Service Weather Prediction Center website https://www.wpc.ncep.noaa.gov/html/heatindex.shtml. 4 Tests suggest that the rural curve during the dry season dominates the “equity” line and is significantly below the 5 percent level.