LATIN AMERICA & 92956 CARIBBEAN REGION Environment & Water Resources OCCASIONAL PAPER SERIES Environmental Health Costs in Colombia The Changes from 2002 to 2010 Environmental Health Costs in Colombia The Changes from 2002 to 2010 June 2014 Elena Golub, Consultant, World Bank Irina Klytchnikova, Sr. Economist, World Bank Gerardo Sanchez-Martinez, Consultant, World Bank Juan Carlos Belausteguigoitia, Executive Director, Centro Mario Molina © 2014 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpreta- tions, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, the governments they represent, or those of Centro Mario Molina. The World Bank does not guarantee the accuracy of the data included in this work. The estimations were made using the 2009 average exchange rate: 1 USD = 2,158 COP. 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. The Environment & Water Resources Occasional Paper Series is a knowledge product of the Environment Unit in the Latin America and Caribbean Region (LCSEN) of the World Bank. The publications in this Series were designed and produced under the supervision of Emilia Battaglini, Acting Sector Manager; Keiko Ash- ida Tao, Natural Resource Management Specialist; and Anna-Lena Sauer, Consultant (LCSEN). A list of the most recent papers is located at the end of this publication. For electronic copies of all our LAC Environment & Water Resources Occasional Paper Series please visit our web site at http://www.worldbank. org/LAC/LCSENSeries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any 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. All images of this publication are courtesy of Juan Pablo Ruiz and Anna-Lena Sauer. Front cover: Juan Pablo Ruiz, The World Bank. Table of Contents Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Acronyms and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 1.  Introduction and Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2.  Summary of the Technical Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Urban Air Pollution—Detailed Technical Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.  3.1  Urban air pollution and health. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.2  Baseline population, pollutant concentration and dose-response coefficients. . . . . . . . . . . . . . . . . . . . . . . . 10 3.3 Mortality and morbidity attributable to air pollution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.4  Health cost of urban air pollution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 4.  Inadequate Water Supply, Sanitation and Hygiene—Detailed Technical Findings . . . . . . . . . . . . . . . . . 19 4.1  Inadequate WASH and diarrheal illness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.2  Child mortality and morbidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 4.3  Child nutritional status and WASH health impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.4  Cost of inadequate WASH in Colombia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 5.  Indoor Air Pollution from the Use of Solid Fuels—Detailed Technical Findings . . . . . . . . . . . . . . . . . . . 28 5.1  Indoor smoke and health. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 5.2  Baseline mortality and morbidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 5.3  Mortality and morbidity attributable to indoor household solid fuel use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 5.4  Health cost of indoor household solid-fuel use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Conclusions and Policy Recommendations 6.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Annex 1.  Economic Basis for Choice of VSL and WTP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Air Quality Monitoring Data and Pollution Concentration Extrapolation Annex 2.  in Medium-Large Municipalities in Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Methodological Differences and Similarities, and Comparative Results with Previous Estimates.44 Annex 3.  List of Figures Figure 1 Health Costs of Selected Environmental Factors in Colombia in 2002 and 2010. . . . . . . . . . . . . . . . . . . . . 3 Figure 2  Environmental Health Cost in Colombia and in the Region (% of GDP). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Figure 3 Relative Change in PM10 Concentrations and Transportation Fleet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Figure 4  Monthly Average PM10 Concentration (Ug/M3) in Bogotá, 2006–12. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Figure 5 Percentage of Exposed Population and Attributable Mortality by City (With Monitoring Network). . . . . . . . 16 Figure 6 Percentage of Population with Access to Improved Sanitation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Figure 7 Estimates of Cause-Specific Non-Accidental Mortality Among Children Under Age Five in Colombia from Infectious and Parasitic Diseases, 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figure 8 Malnutrition Status (Percent) by Income Groups in Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 9 Estimated Annual Cost of Health Impacts Associated with Inadequate WASH. . . . . . . . . . . . . . . . . . . . . . . . 25 Figure 10  Annual Cases of Mortality Attributed to Indoor Air Pollution in Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . 31 List of Tables Table 1  Summary of the Environmental Health Costs in 2010. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Table 2 Change in Selected Urban Demographic Indicators in Colombia, 2002 Versus 2010. . . . . . . . . . . . . . . . . . 10 Table 3 Population and Weighted Average Concentration of PM10 and PM2.5 in Colombian Cities with over 100,000 Inhabitants Covered by Air-Pollution Monitoring Networks. . . . . . . 11 iii Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Table 4  Urban Air Pollution Dose-Response Coefficients. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Table 5 Baseline Mortality and Morbidity Data for Cities with PM Monitoring Data . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Table 6 Calculation of DALYs Per Case of Health Effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Table 7 DALYs Lost per Selected Health Effect Attributable to Air Pollution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Table 8 Estimated Health Impact of Urban Air Pollution in Cities with PM Monitoring Data. . . . . . . . . . . . . . . . . . . . . . . 16 Table 9 Estimated Annual Cost of Health Impacts (Billion Pesos). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Table 10  Estimated Unit Cost by Health End-Point. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Table 11  Baseline Data for Cost Estimation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Table 12 Water Supply and Sanitation in Colombia in 2010. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Table 13  Diarrheal Illness in Colombia among Children under Age Five in 2000–2009 . . . . . . . . . . . . . . . . . . . . . . 21 Table 14 Estimated Annual Cases of Diarrheal Mortality and Morbidity from Water-Sanitation-Hygiene (WASH) in Colombia, 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Table 15 Prevalence of Underweight in Children Under Age Five in Colombia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Table 16 Relative Risk of Mortality from Mild, Moderate and Severely Underweight in Children Under Age Five. . . 24 Table 17 Relative Risk of Illness from Moderate and Severe Underweight in Children under Age Five. . . . . . . . . . . 24 Table 18 Attributable Fractions of Under-Five Child Mortality and Morbidity from Underweight . . . . . . . . . . . . . . . . 24 Table 19 Estimated Deaths among Children under Age Five that could be Associated with Malnutrition in Colombia, 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Table 20 Estimated Annual Cost of Health Impacts (Billion Pesos) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Table 21  Baseline Data for Cost Estimation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Table 22 Household Fuels Used for Cooking in Colombia, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Table 23 Relative Risks of Health Effects from Cooking with Wood and other Biomass Fuels Applied to Colombia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Table 24  ARI Illness in Colombia among Children Under Age Five in 2000–2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Table 25 Estimated Annual Mortality from Household Use of Wood and Biomass for Cooking in Colombia, 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Table 26 Estimated Annual Cases of Morbidity from Household Use of Wood and Biomass for Cooking in Colombia, 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Table 27  Estimated Annual Cost of Health Impacts (Billion Pesos) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Table 28  Estimated Unit Cost by Health End-Point. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Table 29  Baseline Data for Cost Estimation in Rural Colombia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Table A.1 Economic Variables: “Plan Decenal de Descontaminación de Bogotá” (Thousands of Colombian Pesos, 2009) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Table A.2 Application of Benefit Transfer Approach to Estimate VSL for the Population Dying from Pollution in Colombia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Table A.3  Size of Cities With and Without Air Pollution Monitoring Network Coverage in Colombia. . . . . . . . . . . . . . 42 Table A.4 Proportion of Urban Population (Out of Total Population) in Colombian Cities With Over 100,000 Inhabitants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Table A.5 Comparison between 2004 Results for Health Effects of Urban Air Pollution and Current Analysis Including All Cities with Over 100,000 Inhabitants in Colombia . . . . . . . . . . . . . . . . . . . 45 Table A.6 Cost of Selected Health Services and Time Lost to Air Pollution-Related Health Outcomes . . . . . . . . . . . 46 Table A.7 Bogotá: Mortality and Morbidity Costs in the Baseline 2010 Case and in High and Low Pollution Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 List of Boxes Box 1  Bogotá’s Local Air Quality Management Policies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Box 2  The Association between Inadequate Water Supply, Sanitation and Hygiene and Prevalence of Malnutrition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 List of Graphs Graph A.1 Percentage of Urban Population in Cities > 100K Inhabitants in Colombia. . . . . . . . . . . . . . . . . . . . . . . . 43 iv Environmental Health Costs in Colombia Foreword The Latin America and Caribbean (LAC) Region has to contribute to global knowledge exchange on in- a unique mix of qualities and challenges when it novation in addressing environmental issues and comes to the environment. It is exceptionally en- the pursuit of greener and more inclusive growth. dowed with natural assets—diverse ecosystems in- The papers seek to bring to a broader public—deci- cluding the world’s greatest carbon sink in the Am- sion makers, development practitioners, academ- azon, globally significant biodiversity such as the ics and other partners—lessons learned from World Mesoamerican Barrier Reef, and valuable crops. Bank-financed projects, technical assistance and At the same time, the Region registers the highest other knowledge activities jointly undertaken with rates of urbanization in the developing world, wa- our partners. The Series highlights issues relevant ter and natural resources overuse, and increased to the Region’s environmental sustainability agen- pollution, with detrimental consequence for the en- da such as biodiversity conservation, natural and vironment and the health of people, especially the water resources management, irrigation, ecosys- poor. tem services, environmental health, environmental policy, pollution management, environmental insti- Over the past twenty years, the LAC Region has tutions and governance, environmental financing, made impressive gains in addressing these issues. and climate change and their linkages to develop- It leads the developing world in biodiversity conser- ment, growth and shared prosperity. vation, natural and water resource management, and is at the forefront in reducing urban pollution. In this particular paper, we present you the find- The World Bank often has the privilege to partner ings of an assessment of environmental health with countries in the Region to pioneer innovative costs in Colombia carried out in 2012. The World environmental policies and initiatives. Such initia- Bank’s 2005 Colombia Country Environmen- tives include improvement of fuel and air quality tal Analysis “Environmental Priorities and Pover- standards in Peru, carbon emission reduction in ty Reduction” estimated the costs of environmen- Mexico, payment for ecosystem services in Costa tal degradation for Colombia for the first time. The Rica, participatory and integrated water resources 2012 assessment was carried out as part of the management in Brazil, and new approaches to irri- broader program of technical assistance and an- gation management in Mexico. alytical support by the World Bank at the request of the Ministry of Environment and Sustainable The Environment & Water Resources Occasional Development and the National Planning Depart- Paper Series, a publication of the Environment Unit ment of Colombia in order to update and analyze of the Sustainable Development Department in the health impacts and the associated econom- Latin America and Caribbean Region (LCSEN) was ic costs caused by urban air pollution, indoor air launched in 2013. The objective of the Series is pollution from solid fuel use, and an inadequate v Environmental Health Costs in Colombia – The Changes from 2002 to 2010 supply of water and sanitation combined with increased complexity and an improved availabili- poor hygiene practices. These three environmen- ty of relevant data in Colombia. We hope that this tal problems consistently cause the highest health paper along with other publications under the Se- costs across Latin America and the Caribbean. The ries will make a contribution to knowledge sharing 2005 study concluded, among other things, that within the LAC Region and globally. the most costly problems associated with environ- mental degradation are urban and indoor air pol- Emilia Battaglini lution; inadequate water supply, sanitation, and Acting Sector Manager hygiene; natural disasters and land degradation. Environment Unit The present study updates and completes some of Sustainable Development Department the previous estimates, accounting for the much Latin America and the Caribbean Region vi Acronyms and Abbreviations ACS American Cancer Society IAP Indoor Air Pollution AF Population-Attributable Fractions ICU Intensive Care Units ALRI Acute Lower Respiratory Infections LAC Latin America and the Caribbean AMVA Valle de Aburrá Metropolitan Area LC Lung Cancer ARI Acute Respiratory Infection LPG Liquid Petroleum Gas ARI Acute Respiratory Illness MPS Ministry of Social Protection AURI Acute Upper Respiratory Infections NCHS US National Center for Health Statistics BMI Body Mass Index ORT Oral Rehydration Therapy BRT Bus Rapid Transit PEM Protein-Energy Malnutrition CB Chronic Bronchitis PDDB Bogotá Ten-Year Decontamination Plan  CEAs Country Environmental Analyses PM Particulate Matter COED Costs of Environmental Degradation RR Relative Risk Ratios COI Cost of Illness SD Standard Deviations COP Colombian Peso UAP Urban Air Pollution COPD Chronic Obstructive Pulmonary Disease VCO Volatile Organic Compounds DALYs Disability Adjusted Life Years VSL Value of a Statistical Life DANE Colombia’s National Statistical Authority WA Weight-For-Age ENDS National Demographic and Health WASH Water Supply, Sanitation and Hygiene Survey WAZ Weight-For-Age Z-Score GBD Global Burden of Disease WHO World Health Organization GDP Gross Domestic Product WTP Willingness to Pay HC Human Capital YLL Years of Life Lost HCA Human Capital Approach Currency Equivalents (Exchange Rate Effective December 31, 2010) Currency Unit = Colombian Peso 1 USD = 1867.3927 COP 0.000536 USD = 1 COP vii Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Acknowledgements This report presents the results of the study “Envi- (former Deputy Director for Environment and Sus- ronmental Health Costs in Colombia: Changes from tainable Development at the National Planning De- 2002 to 2010.” The study was carried out in 2012 partment) for their support throughout the prepa- as part of the program of analytical support and ration of this study, their helpful comments and technical assistance “Colombia: Strengthening En- discussion, as well as the participants of the work- vironmental Policies and Institutions”, implement- shops held in Bogotá in the context of the imple- ed by the team led by Juan Carlos Belausteguigoi- mentation of this study, and helpful comments by tia (former Lead Environmental Economist, World peer reviewer Ernesto Sanchez-Triana (Lead Envi- Bank, and currently Executive Director, Centro Ma- ronmental Specialist, World Bank) and Daniel Sel- rio Molina) and Irina Klytchnikova (Senior Econo- len (Sector Leader, World Bank). Assistance by Di- mist, World Bank). The estimates presented in this ana Guzmán and Geise Santos, editing by Janice study were carried out by Elena Golub and Gerar- Molina, and translation by Camila Sepulveda are do Sanchez (Consultants). The team would like to gratefully acknowledged. The study was complet- thank the Ministry of Environment and Sustain- ed under the overall guidance of Gloria Grandolini able Development of Colombia, especially Luz Hel- (Country Director), Ede Jorge Ijjasz-Vasquez (Sector ena Sarmiento (Minister of Environment), Pablo Director), Karin Kemper (former Sector Manager, Vieira (Vice Minister of Environment), Adriana Soto LCSEN), Daniel Sellen (Sector Leader), Geoffrey (former Vice Minister of Environment), Francisco Bergen (former Country Manager) and Issam Abou- Gómez (Director of Urban and Sectorial Affairs), sleiman (Country Manager) of the World Bank. Fi- Marcela Bonilla (former Director of Urban and Sec- nancial support by the multi-donor Trust Fund for torial Affairs), Sergio Hernandez, Paula Rodriguez Environmentally and Socially Sustainable Develop- and Olga Lucia Bautista; as well as Carolina Urrutia ment (TFESSD) is gratefully acknowledged. viii Abstract Despite considerable progress in the area of envi- costs were saved. Health costs associated with in- ronmental management over the last decade, Co- adequate water supply, sanitation access and hy- lombia still faces significant impacts from popu- giene have fallen to less than 0.7% of GDP in 2010 lation exposure to urban air pollution, inadequate but a large share of the costs still results from child access to water supply and sanitation, and indoor mortality, requiring continued policy attention to air pollution from solid fuel use. This study esti- improving environmental health conditions in this mates that the total health cost attributable to sector. The costs of indoor air pollution have fall- these three factors amounts to about 10.2 trillion en to less than a quarter of a percentage point of Colombian Pesos (COP) annually, or about 2% of GDP in 2010, but they continue to weigh the most GDP in 2010. In terms of mortality, about 7,600 an- on vulnerable groups and disproportionately affect nual premature deaths can be attributed to these women and children, often perpetuating the cycle environmental factors. This study updates some of poverty and predominantly bearing on the rural of the estimates of environmental health costs re- population. In an ideal setting, all the deaths and ill- ported in the 2005 Colombia Country Environmen- nesses can be avoided but public expenditures and tal Analysis “Environmental Priorities and Poverty investment resources are constrained and need to Reduction”. Using an updated methodology to re- be allocated efficiently. Specific policy recommen- flect income growth and better pollution monitoring dations and targeted interventions can be derived data compared with the earlier study and account- from future analysis of environmental health costs ing for population growth in cities, this assessment at subnational level, cost-benefit analysis of spe- finds that health costs of urban air pollution have cific policy interventions, and an analysis of the increased. Previously the second highest category burden of health costs disaggregated by popula- of costs, they have now shifted to the first place, tion groups and poverty levels. Disaggregated sta- rising by about a quarter of a percentage point to tistics on health outcomes, fuel use, access to in- 1.1% of GDP in 2010. An improvement in fuel qual- frastructure services, epidemiological studies, and ity and other policy measures implemented since air quality models (urban and industrial areas) are 2002 to improve urban air quality have helped re- required for such analysis. Disaggregated assess- duce the health damages. This study estimates ments and cost benefit analysis, recommended for that a reduction in PM10 levels from an average of future studies, will facilitate an evaluation of policy 66 µg/m3 in 2002 to 59 µg/m3 observed in Bogotá and investment outcomes in terms of their impacts in 2010 helped save an average of 200 lives per on the most vulnerable groups and the extent to year, and around 252 billion COP per year of health which they are well targeted and benefit the poor. ix Environmental Health Costs in Colombia 1. Introduction and Context This study estimates environmental health costs of these three factors has remained at a similar level as in 2002 but the relative magnitude of the costs pollution, updating an earlier assessment. has changed, reflecting population and income The World Bank’s 2005 Colombia Country Envi- growth, an improvement in access to improved san- ronmental Analysis, Environmental Priorities and itation, and growth in urban population in Colombia. Poverty Reduction, estimated the costs of envi- ronmental degradation (CoED) in Colombia for the The analysis in this study relies on large sets of first time. The study concluded that the most cost- statistics and data from various ministerial de- ly problems associated with environmental deg- partments, institutions, and institutes in Colom- radation were urban and indoor air pollution; in- bia. It also draws on Colombian and international adequate water supply, sanitation, and hygiene; research studies, and has benefited from various natural disasters and land degradation. The pres- methodological approaches applied by interna- ent report updates the previous estimates of the tional organizations such as the World Health Or- costs stemming from urban and indoor air pollution ganization, and is in accordance with all previous and from inadequate water supply, sanitation and related work within the World Bank’s Country Envi- hygiene, taking into account the income and popu- ronmental Analyses (CEAs). This study is aimed at lation growth and an improved availability of mon- deepening the understanding of the country’s ma- itoring data in Colombia since 2002. This study jor health challenges associated with environmen- presents an assessment of health costs as of 2010 tal pollution. Publicly available and easily traceable and compares them with the earlier estimates. information and indicators were used as much as possible in order to facilitate contrast and future Despite considerable progress in the area of envi- updates. All costs calculated in this report are ex- ronmental management in Colombia over the last pressed in monetary terms, and they include the decade, this assessment reveals that Colombia’s cost to society due to premature mortality, as well population still faces significant adverse impacts as the cost of healthcare provision to individuals from exposure to urban air pollution (UAP), inade- suffering from pollution-related illnesses and the quate water, sanitation and hygiene (WASH), and value individuals place on avoiding resulting pain indoor air pollution from solid fuel use (IAP). The and discomfort. Time losses or savings are valued total health cost attributable to these three fac- at the opportunity cost of time. tors amounts to about 10.2 trillion COP annually, or about 2% of GDP in 2010. In terms of mortali- Policy makers and stakeholders can use the esti- ty, about 7,600 annual premature deaths can be mates of premature mortality and illness and the attributed to these environmental factors. About associated economic costs as an input in the pro- 5,000 deaths are associated with UAP, around cess of setting environmental objectives and in- 1,600 with inadequate WASH and a further 1,000 vestment and policy priorities. The Government with IAP. The overall burden of health costs from of Colombia can use this information to examine 1 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 institutional capacity for sound environmental man- Lastly, this evidence base helps to track progress agement within the country and strengthen institu- with earlier reforms, as well as advancing further tions and governance to enhance environmental on the environmental protection and environmen- outcomes. The World Bank and other donors can tal health agendas. Ambitious and relevant policies use this study’s results to help set priorities in the will find adequate justification based on the large policy dialogue with the Government of Colombia. health and economic cost of environmental degra- Because preferences and values are expressed in dation in Colombian society. Addressing these en- monetary terms, the results can provide additional vironmental risks should continue to be a priority in guidance for the allocation of resources across the the environmental and public health policy agenda range of socio-economic development goals. in Colombia. 2 2. Summary of the Technical Findings Health costs of environmental degradation, updated at 2% of GDP—although Colombia’s GDP has more than doubled since. However, the ranking has shift- in this study, have increased at the same rate as the ed (Figure 1). The health costs associated with ur- economy and population have grown. ban air pollution have increased as a percentage of This study estimates that the total health costs at- GDP, rising to the first place, whereas the costs of tributable to urban air pollution, inadequate water indoor air pollution have remained unchanged and supply, sanitation and hygiene, and indoor air pol- the costs of inadequate water, sanitation and hy- lution from solid fuel use in Colombia amount to giene have dropped markedly relative to GDP. about 10.2 trillion COP (about 2% of GDP in 2010). The largest cost is attributable to urban air pollu- tion, to which a large share of the population is ex- Health costs of urban air pollution have increased posed, followed by inadequate water supply, sanita- tion and hygiene. The third is the cost attributable relative to other health costs, but better air quality in to the indoor use of solid fuels, which is a com- Bogotá helped temper that rise. mon practice in rural areas of Colombia. The cost Urban air pollution was previously estimated as of these three environmental factors combined the third highest in terms of the associated costs, has remained unchanged as a proportion of the following water supply, sanitation and hygiene, and GDP—an earlier assessment also estimated them natural disasters. In this round of evaluation the Health Costs of Selected Environmental Factors in Colombia in 2002 and 2010 Figure 1  2002 2010 1.2 1.2 1.04 1.12 1.0 1.0 0.8 0.78 0.8 % of 2002 GDP % of 2010 GDP 0.68 0.6 0.6 0.4 0.4 0.21 0.22 0.2 0.2 0 0 Urban AP WSH Indoor AP Urban AP WSH Indoor AP Source: Authors’ estimates for 2010 results; Larsen (2004) for 2002 results. Note: The estimates of the costs of urban air pollution for 2002 and 2010 shown above are not directly comparable as a slightly different methodology was used in this study as is takes into account significant income growth in Colombia over the last decade and an expansion of the air quality monitoring network (and populations of cities with monitoring data are included in the 2010 analysis, whereas in 2002 an extrapolation was used for cities without monitoring networks). About half of the costs are due to population exposure to air pollution in Bogotá, and air quality has improved there since 2002. Sensitivity analysis reveals that, keeping all other factors constant, health costs due to air pollution in Bogotá alone would have been 7% higher in 2010 had air quality not improved. 3 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 mean annual cost of urban air pollution for Colom- Indoor air pollution weighs the most on women bia is estimated at about 5.7 trillion COP, or 1.1% of 2009 GDP (Table 1). Mortality represents about and children, affecting the rural population and 79% of the total estimated cost. Without consid- perpetuating the cycle of poverty. ering the cost of natural disasters, this puts ur- The mean estimated annual cost of health impacts ban air pollution in first place, ahead of water sup- from indoor air pollution associated with the use of ply, sanitation and hygiene.1 An improvement in traditional fuels (mainly fuel wood) in rural areas fuel quality and other policy measures implement- of Colombia is 1.129 trillion COP (0.22% of GDP in ed since 2002 to improve urban air quality in Bo- 2010). This environmental factor weighs the most gotá alone have helped reduce the health damag- on the vulnerable groups and often perpetuates es that would have occurred in the absence of that the poverty cycle: mortality in children under age improvement. This study estimates that a reduc- five represents 6% of cost, and mortality in wom- tion in PM10 levels from an average of 66 µg/m3 in en over 30 years of age represents about 78% of 2002 to 59 µg/m3 helped save an average of 200 cost. Acute respiratory illness (ARI) in children and lives in 2010; and around 252 billion COP per year adult females and Chronic Obstructive Pulmonary of health costs were saved due to that improve- Disease (COPD) morbidity of adult females repre- ment in air quality. sent 16% of the cost. Table 1  Summary of the Environmental Health Costs in 2010 Factor Annual mortality Annual morbidity Associated monetary costs Number of DALYs COP Percentage of premature deaths (million) (billion) 2009 GDP Urban Air Pollution (UAP) 5,000 65 5,700 1.12 Water, Sanitation and Hygiene 1,600 20 3,450 0.68 (WASH) Indoor Air Pollution (IAP) 1,000 12 1,129 0.22 Total 7,600 97 10,279 2.00 Source: Authors’ estimates. Note: 1 USD = 1,817 COP. DALYs are disability-adjusted life years. 1  Direct comparison between estimates in 2002 and 2010 should be made with caution. Methodology of environmental health costs quantifica- tion significantly improved since 2002 in terms of capturing health impacts of pollution by PM2.5 and valuation of mortality for middle-income coun- tries. Furthermore, the air quality monitoring network expanded in Colombia, so it was possible to generate more accurate results in the current study using the revised methodology and more accurate estimates of population exposure to pollution based on monitored data instead of extrap- olations in cities with no monitoring data, as was done in the earlier study for 2002. Lastly, recent literature concurs that Human Capital Approach (HCA)—the valuation method used in the earlier study for an adult mortality risk valuation—should not be used in a situation with a substantial GDP per capita growth, which has occurred in Colombia since then. To assess the benefits of an improvement in air quality observed in Bogotá since 2002, sensitivity analysis was carried out. It shows the changes in mortality and morbidity and the associated costs in Bogotá in two cases: a high concentration scenario with the average levels of PM10 measured in Bogotá in 2002, and a low concentration scenario with the levels measured in 2012. It is assumed that the ratio of PM2.5 to PM10, with the former being the pollutant due to which the health damages occur—is between 40 and 60%, and the midpoint of 50% is used in this calculation. It is estimated that 200 additional mortality cases would have occurred in Bogotá had the pollution level remained unchanged compared with the 2002 level; and 440 fewer mortality cases would have occurred, ceteris paribus, with a concentration level on average equal to 48 µg/m3 (i.e., the level measured in 2012). Thus, the set of policy measures, including the intro- duction and enforcement of more stringent fuel quality standards in Bogotá and other measures, and which led to the lower measured levels of PM10 in Bogotá, resulted in a reduction of mortality cases in Bogotá by 7% in 2010 relative to what they would have been had air quality not im- proved. A further improvement in air quality in 2012 resulted in a reduction of mortality cases in Bogotá by 16% compared with the baseline sce- nario. For further details about the methodological differences between the 2002 and 2010 studies, see Annex 3. 4 in terms of lost Disability Adjusted Life Years (DA- Health costs of inadequate water supply, sanitation LYs)), the pattern is similar: nearly 70% of DALYs and hygiene have fallen but a large share of the are attributable to UAP, around 20% to WASH, and costs still results from child mortality. around 10% to IAP.3 The following levels of mortali- The mean estimated annual cost of health impacts ty and morbidity were estimated as attributable to from an inadequate supply of drinking water and UAP, IAP and WASH: sanitation and from poor hygiene in Colombia is 3.45 trillion COP (0.68% of GDP in 2010). Mortali- • About 5,000 premature deaths and almost 65 ty in children under age five represents 17% of the million DALYs can be attributed to urban air pol- cost, with morbidity accounting for the remaining lution each year in Colombia. Bogotá and the 83%. Diarrheal mortality and morbidity represent Valle de Aburrá Metropolitan Area (Área Met- about 89% of total cost and are estimated at about ropolitana del Valle de Aburrá, AMVA) account 3.05 trillion COP annually. Urban cost represents for over 75% of the attributable mortality. Near- about 77% of the total diarrheal cost. The reduc- ly 4,700 new cases of chronic bronchitis each tion of health costs in this sector is associated with year are also attributable to urban air pollution an improvement of public health measures, a re- in Colombia. Mortality represents about half of sulting reduction of background diarrheal mortali- the burden of disease attributable to air pollu- ty and morbidity burdens, and an increase of rural tion, and morbidity (i.e., diseases) accounts for population with access to improved sanitation over the other half. the 2002–10 period. • About 1,000 premature deaths and almost 12 million DALYs can be attributed to indoor Urban air pollution causes three times as many air pollution caused by solid fuel use each year deaths as inadequate water supply, sanitation and in Colombia. This burden of disease is almost completely restricted to rural areas, where hygiene, and 5 times as many deaths as indoor air nearly 50% of the population uses solid fuels pollution. for cooking and other household chores. An es- The health costs assessed in this study are de- timated 950 to 1,050 children and women died rived from the estimates of premature mortality prematurely from respiratory illnesses associat- and morbidity linked with pollution.2 As of 2010, ed with the use of wood and other biomass fu- about 7,600 premature deaths annually were at- els for cooking in Colombia in 2010. About 200 tributable in Colombia to urban air pollution, indoor of these deaths were among children under age air pollution from solid fuel use, and an inadequate five and 800 were among adult women. supply of water and sanitation combined with poor • About 1,600 premature deaths and almost hygiene practices. 20 million DALYs can be attributed to inade- quate water supply, sanitation and poor hy- Of the total number of deaths due to these three fac- giene each year in Colombia. About 1,000 of tors, UAP caused about 65% of premature mortali- these premature deaths occur in children un- ty, followed by WASH (around 20%) and IAP (about der age five due to infectious diseases ampli- 15%). In terms of burden of disease (measured fied by malnourishment. 2  These health impacts, in turn, must be interpreted cautiously. In the absence of relevant local epidemiological evidence, internationally accepted rapid assessment methods have been used for the calculations. Details about such calculations are explained in the report and technical annexes. 3  DALYs are the sum of years of potential life lost due to premature mortality and the years of productive life lost due to disability. 5 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Environmental health costs in Colombia are at the Recent efforts to strengthen air quality management lower end of the spectrum compared with other have been effective at preventing a worsening in air countries in the region. pollution despite the growing vehicle fleet. Similar analyses were conducted in other countries Significant progress has been made toward effec- in the LAC region over 2002–07 (Figure 2). The tive air pollution management in Colombia in the methodology in this report was shifted toward the last decade. The 2010 Air Pollution Control and application of value of statistical life (VSL) for the Prevention Policy notes some of the main areas of valuation of mortality due to urban air pollution (as progress: air quality assessment, monitoring, stan- opposed to the average between the human capital dardization of air quality inventories, fuel quality im- approach (HCA) and VSL in other studies). In prac- provement, and implementation of incentive pro- tice, this means a higher value compared with the grams for environmental control and monitoring. It combined HCA-VSL approach. In spite of this, Co- is important to note that a strong effort in this regard lombia is on the lower end of the environmental helped generated an ample evidence base that re- pollution burden compared to other Latin American vealed systemic weaknesses in air quality manage- countries. Urban air pollution by far outweighs other ment and areas for improvement. The action plan environmental health problems. At the same time, for the implementation of the 2010 policy set am- the improvement of water supply, sanitation and hy- bitious goals to address policy and investment pri- giene and the reduction of indoor air pollution re- orities. To illustrate specific improvements, in the main an important pending task that can effective- last 10 years PM10 concentrations decreased slight- ly reduce mortality and morbidity in children in rural ly in Bogotá and more significantly in Cali and Bu- and urban areas, women and the rural population. caramanga. They increased 5% in the Aburrá Valley Figure 2  Environmental Health Cost in Colombia and in the Region (% of GDP) 4.5 4.1 6,000 4.0 5,000 3.5 3.0 2.8 2.8 2.5 2.3 4,000 GDP per capita (USD) 2.5 % GDP 2.2 2.0 2.0 3,000 2.0 1.7 1.5 2,000 1.0 1,000 0.5 0 0 Bolivia Guatemala El Salvador Nicaragua Peru Honduras Colombia Colombia Ecuador (2005) (2006) (2006) (2007) (2006) (2006) (2002) (2010) (2004) WASH Urban AP Indoor AP GDP per capita (current USD) Source: Country Environmental Analyses for various years. For Colombia, this study for 2010 results and Larsen (2004) for 2002 results. Note: For Colombia, the estimates of the costs of urban air pollution for 2002 and 2010 shown above are not directly comparable as a slightly different methodology was used in this study as is takes into account significant income growth in Colombia over the last decade and an expansion of the air quality monitoring network (and populations of cities with monitoring data are included in the 2010 analy- sis, whereas in 2002 an extrapolation was used for cities without monitoring networks). About half of the costs are due to population ex- posure to air pollution in Bogotá, and air quality has improved there since 2002. Sensitivity analysis reveals that, keeping all other fac- tors constant, health costs due to air pollution in Bogotá alone would have been 7 percent higher in 2010 had air quality not improved. 6 Metropolitan Area (Area Metropolitana del Valle de Relative Change in PM10 Concentrations Figure 3  Aburrá, which includes Medellin). At the same time, and Transportation Fleet the number of registered vehicles per 1,000 inhab- 70 160% itants grew about 105 to 140% in these metropol- 140% transportation fleet (change in %) 60 itan areas (Molina et al. 2008). In total, the trans- 50 120% from 2000 until 2007 portation fleet grew from 13 to 53% in 2000–07. 100% Figure 3 presents the relative change in PM10 con- 40 ug/m3 80% centrations over 2000–10 and the rate of increase 30 60% in transportation fleet over 2000–07 in these cities. 20 40% 10 20% In the case of Bogotá, the concentration of partic- 0 0% ulate matter has decreased consistently: from a Bogota Bucaramanga AMVA Cali yearly average of about 70 µg/m3 in 2007 it fell to PM10 concentration, 2000 around 50 µg/m3 in 2011. Furthermore, seasonal PM10 concentration, 2009–2010 variability and dangerous seasonal concentration Transportation fleet change peaks seem to be decreasing as well (Figure 4). Source: Authors’ estimates. PM10 concentrations are based on data from SISAIRE (http://www.sisaire.gov.co) for 2009–2010 and Larson (2002); transportation fleet estimated from Molina et Other subnational and local governments have al. (2008). also strengthened efforts through sectoral mea- sures with potential large benefits for air quality, such as urban transportation (Box 1). Following the license plate number at peak hours) schemes are example of Bogotá, car-free days, Bus Rapid Tran- also present now in major Colombian cities. A no- sit (BRT) systems and the promotion of non-mo- table improvement is the expansion of the air qual- torized transport are extending to other large ur- ity monitoring network. The population covered ban centers. The “Pico y Placa” (Peak and License by the network of monitoring stations as of 2010 Plate–a car-use restriction by the last digit of the was estimated at around 18 million people (up Figure 4  Monthly Average PM10 Concentration (Ug/M3) in Bogotá, 2006–12 110 100 90 80 70 60 50 40 2006–12 2007–02 2007–04 2007–06 2007–08 2007–10 2007–12 2008–02 2008–04 2008–06 2008–08 2008–10 2008–12 2009–02 2009–04 2009–06 2009–08 2009–10 2009–12 2010–02 2010–04 2010–06 2010–08 2010–10 2010–12 2011–02 2011–04 2011–06 2011–08 2011–10 2011–12 2012–02 Source: Bogotá Environmental Secretariat, pers. comm. 7 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Box 1  Bogotá’s Local Air Quality Management Policies The local government in Bogotá has made significant efforts in the recent years to improve air quality in the metropolitan area. Short of a full policy impact assessment, we cannot establish a direct causality between air quality management pol- icies implemented in Bogotá and the observed reductions in particulate matter. However, available indicators suggest that most measures and interventions adopted by the city government have likely had a large positive impact. These actions in- clude, but are not limited to: • Diesel fuel quality improvement. The maximum permissible sulfur content of diesel fuel sold in Bogotá in 2008 (500 parts per million) was reduced to 50 ppm in 2010, although real concentration of locally distributed fuel is around 30 ppm. As of the end of 2012, all diesel fuel in Colombia must be under 50 ppm. • Increased use of Bus Rapid Transit (BRT) system. Increased occupancy levels in the Transmilenio system have entailed comparatively large emission reductions. Current BRT emissions are about 0.058 grams of PM2.5 per passenger, com- pared with 0.311 grams in the rest of the current public transportation system in Bogotá (a comparative reduction of about 80%). Consequently, CO2 emissions of the BRT system are much lower than for other means of transportation as well. • Restriction on sales of two-stroke engine motorbikes in Bogotá and a later ban on their circulation in Bogotá (subse- quently overturned). According to the emissions inventory in the city’s ten-year decontamination plan (PDDB 2010) mo- torbikes contribute 25% of total mobile source emissions of PM in the city. Four-stroke engine motorbikes emit 40% less PM than two-stroke ones. The ban was only under application for two years, as a new mayor reversed the ban on two- stroke engines in 2012. Further progress in these areas is part of an ambitious 10-year air quality management plan currently under implementation. One important planned city-wide intervention is the integration of the public transportation system; although not yet imple- mented, this integration is projected to have a major impact on the city’s air quality. According to an ongoing study by the Uni- versidad de los Andes, the reorganization has the potential to reduce PM2.5 emissions from public transport by between 74 and 80%. That would entail health-related economic savings of 360 million USD over ten years (PDDB 2010). Source: Clean Air Institute for Latin American Cities (pers. communication). from 12.5 million in 2002, as reported in Larsen is no obvious decreasing trend, as 53% of rural (2004)) whereas the population living in cities with households used solid fuels in 2005 compared to greater than 100,000 inhabitants4 not covered by just over 50% in 2010 (ENDS 2010). In terms of air quality monitoring networks is close to 9 million a regional perspective, the Eastern and Caribbe- (about the same as in 2002). an regions have the highest incidence of solid fuel use: 17.7% and 20.1% respectively. The only region No systematic efforts have been made to reduce the where solid fuel use for household uses is negligi- ble is Bogotá. burden of disease from indoor air pollution. One area where little progress has been achieved over the last decade is in the management of in- Sanitation and diarrheal treatment have improved door air pollution. There have been no focused ef- forts to decrease exposure to indoor air pollution, in rural areas, reducing the incidence of diarrheal and the burden of diseases associated with indoor illness. air pollution continues to be overwhelmingly con- General malnutrition (low weight for age) in Colom- centrated in Colombia’s rural households. There bia decreased from 7% in 2005 to 4.5% in 2010. The cutoff point of 100,000 inhabitants is taken as an indicator that an urban setting is large enough to include substantial amounts of mobile 4  and point sources of pollution to represent a health hazard. 8 Severe malnutrition has also decreased slightly, Future studies should analyze environmental from 0.6 to 0.5% (ENDS 2010, p. 298). The use of some sort of oral rehydration therapy (ORT)—the health costs and carry out cost-benefit analysis main remedy used to treat diarrheal illness in chil- of investments and policy interventions at a dren—increased from 61% in 2000 to 70% in 2005 disaggregated level. and 74% in 2010 (ENDS 2010, p. 256). Although This assessment has revealed the continuing im- not proportional to GDP growth over the same pe- portance and high costs of the environmental riod, these improvements are still notable. Howev- health problems in Colombia. Specific policy rec- er, systematic differences remain between urban ommendations and targeted interventions can be and rural areas as well as among regions in terms derived from studies of environmental health costs of the awareness and care of diarrheal diseases in at subnational level, cost-benefit analyses of spe- children. cific policy interventions, and analyses of the bur- den of health costs disaggregated by population Rural-urban and regional differences still persist groups and poverty levels. Disaggregated statistics in the supply of safe drinking water and the provi- on health outcomes, fuel use, access to infrastruc- sion of the appropriate means of sanitation. At the ture services, and air quality models in urban areas same time, the national demographic and health are required for such analysis. Robust air quality surveys reflect that in the last ten years the share models at metropolitan level are critically needed of Colombian population without access to im- to enable an evaluation of alternative scenarios of proved sanitation decreased by 13%. The share of emission reductions from fixed and mobile sourc- rural population with improved sanitation reached es and the associated health benefits. Disaggre- 84%, and about 99% of urban households are con- gated assessments and cost-benefit analysis will nected to sewers. Inadequate WASH in rural areas facilitate an evaluation of policy and investment is still a serious environmental health and health outcomes in terms of their impacts on the most equity problem in Colombia that disproportionate- vulnerable groups and the extent to which they are ly affects the relatively more poor rural population. well targeted and benefit the poor. 9 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 3. Urban Air Pollution—Detailed Technical Findings 3.1 Urban air pollution and health (i) identification of air pollutants and determination Worldwide evidence on the health effects of urban of concentrations; (ii) population exposed and its air pollution has been substantial for decades now, baseline vulnerability; (iii) calculation of the health with extensive studies showing the association be- impact of exposure to air pollution based on epi- tween certain air pollutants and respiratory and demiological techniques; and (iv) estimation of the cardiovascular mortality, chronic bronchitis, respi- value of this health impact. ratory infections, and several other related disor- ders. Most studies show the strongest association 3.2 Baseline population, pollutant concentration between pollutants and health effects for inhalable particulate matter, particularly PM2.5 (smaller than and dose-response coefficients The proportion of the urban population was esti- 2.5 microns in diameter). To the comprehensive re- mated at 74% in 2010 (DANE 2011). Nearly 60% views in the late 1990s and early 2000s in Europe- of the Colombian population now lives in cities with an and North American countries, a growing body over 100,000 inhabitants and almost 30% lives in of evidence can be added from cities in developing cities with over 1 million inhabitants; Cartagena is countries of Asia (e.g., HEI 2008) and from cities in quickly approaching this threshold as well. These Latin American countries (Bell et al. 2006, O’Neill figures are in line with a long-term increasing ur- et al. 2008, Bell et al. 2011, among others). banization process in Colombia (see Table 2). The amount of information available in Colombia The population included in this analysis is the one has also increased greatly. Air-pollution monitoring nominally covered by air pollution measurement data, which in 2004 were available for four metro- networks as of 2010. This coverage has increased politan areas in Colombia, are currently available for greatly in the last decade, partly due to the govern- eleven large metropolitan areas comprising several mental effort to establish environmental monitoring municipalities (SISAIRE 2012). Monitoring of PM2.5 in areas surrounding ports and mining operations. has begun experimentally in some stations and In 2004, only four metropolitan areas (Bogotá, Bu- there are now local estimates of the PM10/PM2.5 ra- caramanga, Valle de Aburrá, and Cali) measured tio. There is more accurate information on and bet- ter coverage of demographic and statistical informa- Change in Selected Urban Table 2  tion, and health indicators are better disaggregated. Demographic Indicators in Colombia, Thus, the uncertainty of the analysis has been re- 2002 Versus 2010 duced, but its complexity has increased in propor- tion to the amount of decisions and generalizations Parameter DANE 2002 DANE 2010 to be made in order to obtain nationwide estimates. % of urban population 72 74 % living in cities with over 49 59 100,000 inhabitants With these considerations in mind, the analyti- % living in cities with over 28 29 cal approach to the estimation of the damage val- 1,000,000 inhabitants ue of air pollution follows the same main steps Source: National Administrative Department of Statistics (Depar- used in Larsen (2004) as well as in most Coun- tamento Administrativo Nacional de Estadística, DANE) http:// try Environmental Analyses including air pollution: www.dane.gov.co/. 10 PM10. Today, most of Colombia’s large metropolitan average concentrations for the metropolitan areas areas are covered by coordinated networks corre- covered by one or more monitoring networks. sponding to environmental jurisdictions. We have organized the information in areas correspond- This evaluation only includes the urban popula- ing to eleven metropolitan areas of various siz- tion living in municipalities with real-time, constant es. Unlike previous studies (Larsen 2004), we did monitoring of air pollution. Excluding a potentially not feel that partitioning land use into categories exposed population of 9 million (nearly 25% of Co- could provide valuable insights on exposure differ- lombia’s urban population) is a suboptimal analyt- entials. A quick overview of land use in Bogotá and ical choice, but the question remains about how to Medellín with a Geographical Information System estimate exposure without information on the pol- showed a completely mixed pattern of land use, lutants. The obvious option is to estimate their con- blurring any purported systematic differences. How- centrations based on those observed in Colombian ever, we weighted PM concentrations by the pop- cities of comparable size. Such was the approach ulation of the main urban setting in which each taken by Larsen (2004) in the absence of emis- monitoring station or separate network was embed- sions inventories from which to derive concentra- ded, assuming a relatively homogeneous disper- tions (a method with its own drawbacks). However, sion of pollutants. Table 3 presents the population there may be systematic differences between cit- and population-weighted PM10 and PM2.5 annual ies of comparable size according to their pollution Population and Weighted Average Concentration of PM10 and PM2.5 in Colombian Cities Table 3  with over 100,000 Inhabitants Covered by Air-Pollution Monitoring Networks Annual average Annual average Population in 2009 population weighted PM10 population weighted PM2.5 Urban area (million) concentration (µg/m3)* concentration (µg/m3)** Bogotá 7.26 60 30 AMVAa 3.25 56 28 Cali 2.22 22 11 Bucaramangab 0.90 33 17 Cucuta 0.61 60 30 Pereirac 0.56 53 27 Ibague 0.52 32 16 Soacha 0.44 48 24 Manizales 0.39 34 17 Palmira 0.29 44 22 Sogamosod 0.1 58 29 Yumbo 0.1 50 25 Source: Population figures are based on city specific estimates by DANE. Note: *Based on data from SISAIRE (http://www.sisaire.gov.co) for 2009–2010. **PM2.5/PM10 ratio of 0.50 is assumed. a Includes 10 conurbated cities in the Valle de Aburrá. b Includes Floridablanca and Girón. c Includes Dos Quebradas. d Includes Nobsa. 11 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 monitoring status (see Annex 2) that would render contrary, higher concentrations from mobile sourc- such an extrapolation highly uncertain. The force es and industry typically account for higher ratios, behind the establishment of air-quality monitoring so the ratios in cities smaller than Bogotá might networks—in Colombia and elsewhere—is precisely reasonably be expected to be equal or lower. We the presence of air-quality problems, which are in used a PM2.5/PM10 ratio of 0.50. turn frequently linked to the size and density of ur- ban settings.5 Therefore, it is expected that munici- Granted, inhalable particles are not the only palities with monitoring networks will be larger and health-relevant air pollutant. Many anthropogenic more “urban” than those without networks. The emissions have proved to be associated with ad- likely presence of relevant systematic differences verse health outcomes, including (but not limited (beyond mere size) between cities with and without to) sulfur oxides, nitrogen oxides, volatile organic monitoring networks provides additional uncertain- compounds (VCOs), carbon monoxide, lead, and ty to such a method for estimation. Furthermore, especially ozone. Tropospheric (i.e., ground-lev- the urban population not covered by a monitoring el) ozone can trigger a large number of respirato- network is shrinking rapidly in Colombia. Based on ry effects and aggravate certain chronic diseas- these considerations, we decided to drop altogeth- es, thus increasing outcomes, such as increased er the estimation of mortality and morbidity effects health care usage or absenteeism, with high costs of cities for which there were no data. Although we to society (US EPA 2012). An association between acknowledge that we are missing out on a poten- ozone concentrations and long-term mortality has tially substantial proportion of the overall health been found, but only when PM2.5 concentrations impact of urban air pollution in Colombia, we be- were taken into account (Jerrett et al. 2009). In lieve this approach will provide more robust results general, evidence shows that the strongest asso- for this and future updates. However, for compari- ciation and magnitude of effects in the interaction son purposes, we also report overall results includ- between air pollutants and premature mortality/ ing mortality in non-monitored cities (see Annex 3). health are related to particulate matter, particular- ly that with the smallest diameter fraction. Another divergence from previous studies on this matter is the PM2.5/PM10 ratio. The relative risks There has been a substantial improvement in avail- for mortality estimated in the literature most wide- able evidence on the links between air pollution ly used in this type of calculations (Pope 2002) and and mortality in Latin America in the last decade, in the previous Colombia Country Environmental although most studies have dealt with short-term Assessment relate to the concentration of PM2.5. effects. A recent study (O’Neill et al. 2008) ana- However, widespread monitoring of PM2.5 is still un- lyzed the effect of education on the association be- common. Colombia is no exception: as of today, tween PM10 concentrations and short-term mortali- only four measuring stations (all of them in Bo- ty in Mexico City, São Paulo and Santiago de Chile, gotá) measure PM2.5 systematically. A recent study and found total non-accidental adult mortality 1-day (PDDB 2009) for Bogotá places this ratio at 0.50. lagged increases of 0.39%, 1.04% and 0.61%, re- There is no reason to assume that other Colombi- spectively, for an increase of 10 µg/m3 in concen- an cities will have higher ratios than that; on the tration. In Brazil, studies have found associations 5  Excluding cases where point sources contribute a large proportion of the pool of pollution. 12 between exposure to PM10 and low birth weight pollution in Colombia, a larger body of evidence is (Gouveia et al. 2004) and also with respiratory required to provide reliable estimates of health ef- mortality in the elderly (Martins et al. 2004). fects applicable at the national level. For the as- sociation between exposure to inhalable particu- In Colombia, the evidence base for health risks late matter and mortality, the coefficients of Pope of air pollution is still under development. Ibáñez et al. (2002) continue to be the most solid re- (2003) reviewed three studies that assessed the sults for long-term effects. Pope et al. (2002) uti- relationship between urban air pollution and health lized ambient air-quality data from metropolitan ar- effects in Bogotá, and provided dose-response co- eas across the United States for 1979–1983 and efficients for hospital respiratory admissions, child 1999–2000, and information on certified causes morbidity and respiratory mortality. A recent study of mortality of adults in the American Cancer Soci- (Aristizabal et al. 2009) studied the association ety (ACS) database over a period of 16 years. The between air pollution and acute respiratory infec- details of the study (which confirms previous obser- tion (ARI) in three municipalities within Bogotá and vations, such as those of Dockery et al. 1993, Pope found a higher incidence of objective symptoms et al. 1995) have been discussed extensively else- in children living in areas with higher exposure to where, and the results still stand as the best avail- PM10. That is, children living in more polluted ar- able evidence for the association between expo- eas of Bogotá are more likely to develop ARI; these sure to inhalable particulate matter and mortality. differences are statistically significant when con- trolled for other factors, such as cigarette-smoke Likewise, the morbidity coefficients (Ostro 1994, exposure. This study confirms previous observa- Abbey 1995) presented in Table 4 still represent tions (Arciniegas et al. 2006) and is contributing to highly relevant indicators of increased risk for the a growing body of local evidence that will allow for considered categories. These are extracted directly ever more relevant assessments in Colombian ur- from Larsen 2004, where the details of the studies ban areas. are discussed extensively. Although the mortality effects are based on associations with concentra- Although these studies contribute to a greater tions of PM2.5, the morbidity effects assessed in understanding of the health effects of urban air most worldwide studies are based on PM10. Table 4  Urban Air Pollution Dose-Response Coefficients Per 1 µg/m3 annual Dose-response average ambient Annual health effect coefficient concentration of: Mortality (% change in cardiopulmonary and lung cancer mortality) 0.8% PM2.5 Chronic bronchitis (% change in annual incidence) 0.9% PM10 Hospital respiratory admissions (per 100,000 population) 1.2 PM10 Emergency room visits (per 100,000 population) 24 PM10 Restricted activity days (per 100,000 adults) 5,750 PM10 Lower respiratory illness in children (per 100,000 children) 169 PM10 Respiratory symptoms (per 100,000 adults) 18,300 PM10 Sources: Pope et al. (2002) for the mortality coefficient; Ostro (1994) and Abbey et al. (1995) for the morbidity coefficients. 13 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 study on the prevalence of CB in Colombia (PRE- 3.3 Mortality and morbidity attributable to air POCOL: Caballero et al. 2008), there are still no pollution good data on the annual incidence of the disease. In order to ascertain the share of mortality that is The rates applied are those from the World Health attributable to air pollution, baseline data on cer- Organization (WHO 2001) and Shibuya (2001) for tain causes of mortality are required. These data the AMRO-B region6 of WHO in which Colombia is a are collected by Colombia’s national statistical au- part, modified with the known data for clinical prev- thority (DANE) and reported by department on an alence of COPD reported in the PREPOCOL study. annual basis. The categories included are cardio- The resulting incidence rate for the urban Colom- pulmonary causes and lung cancer (DANE catego- bian population over age 30 is 256 cases per ries 206, 301–309, and 605–608). Crude total 100,000 population in one year, compared to a val- and cardiopulmonary mortality rates are listed in ue of 205 for AMRO-B. Table 5. For the calculation of an attributable fraction, we Regarding nonfatal outcomes with known air-pollu- established a lower threshold level for PM2.5, be- tion associations, perhaps the most burdensome low which it is assumed there are no mortality ef- for patients and health systems is chronic bronchi- fects. Although there is much debate about the tis (CB). Although there is a rather complete recent usefulness of these lower limits (WHO recognizes Baseline Mortality and Morbidity Data for Cities with PM Monitoring Data Table 5  Crude mortality rate Cardiopulmonary and lung cancer Cities with PMMN (per 1,000 population)* deaths (% of all deaths)** Bogotá 4.2 34 AMVAa 5.1 36 Cali 5.6 32 Bucaramangab 4.8 34 Cucuta 5.0 37 Pereirac 5.8 36 Ibague 5.0 43 Manizales 5.4 39 Palmira 5.6 32 Sogamosod 4.5 42 Yumbo 5.6 32 Average 5.2 36 Sources: Based on DANE statistics. *Non-accidental based on departmental data. **Based on departmental data. a Includes 10 conurbated cities in the Valle de Aburrá. b Includes Floridablanca and Girón. c Includes Dos Quebradas. d Includes Nobsa. Member states of WHO are divided into six geographical regions. this region is further subdivided into sub-regions according to child and adult 6  mortality from A (lowest) to E (highest). The Americas conform one region (AMRO) and Colombia is one of the countries in the sub-region B. 14 that there is no safe threshold for inhalable parti- Calculation of DALYs Per Case of Table 6  cles), it is necessary for practical matters regard- Health Effects ing air-quality management. WHO (2002) recom- Average mended this threshold to be 7.5 µg/m3 in the World Disability duration of Health Report for mortality. However, a recent re- Health effect weight illness view by Krewski et al. (2009) lowers the threshold Mortality 1.0 (7.5 years lost) to 5 µg/m3. We applied the 5 µg/m3 threshold for Lower respiratory 0.28 10 days mortality effects. No threshold was used for mor- illness: children bidity effects. Respiratory 0.05 0.5 days symptoms: adults Restricted activity 0.1 1 day Aside from mortality, health end-points considered days: adults in this analysis are listed in Table 6. These specif- Emergency room visits 0.30 5 days ic health effects have become the standard health Hospital admissions 0.40 14 days* end-points considered in most of the worldwide Chronic bronchitis 0.2 20 years studies on air pollution. In order to facilitate mag- Source: Larsen (2004). nitude comparisons with other risk factors, health * Includes days of hospitalization and recovery period after effects can be converted to disability adjusted life hospitalization. years (DALYs, a combination of years lost due to premature mortality and years lost due to disabil- DALYs Lost per Selected Health Effect Table 7  ity). In order to do so, disability weights and aver- Attributable to Air Pollution age duration of each outcome are assigned to each health effect.7 We use the weights determined by DALYs lost per Larsen (2004) for the Latin America and the Carib- Health effect 10,000 cases bean (LAC) region. Years lost to premature mortal- Mortality 75,000 ity from air pollution were estimated from age-spe- Chronic bronchitis (adults) 22,000 cific mortality data for cardiopulmonary and lung Hospital respiratory admissions 160 cancer deaths, discounted at 3% per year. Emergency room visits 45 Restricted activity days: adults 3 Once health effects of air pollution are converted to Lower respiratory illness: children 65 DALYs, quick comparisons can be made between Respiratory symptoms: adults 0.75 different environmental risk factors. A calculation Source: Larsen 2004. of DALYs lost per 10,000 cases of the considered health end-points is presented in Table 7. population exposed, adjusting for age groups when The estimated health impact of urban air pollution necessary and extracting the fraction of these in Colombia8 is presented in Table 8. The values are health outcomes that is attributable to this specific calculated by applying relative risks and PM con- exposure. DALYs are calculated simply by multiply- centrations (minus thresholds, where applicable) to ing the number of cases by the factors in Table 7. 7  This approach is not free from controversy, since there is considerable uncertainty about duration estimates, and weights include a substantial subjective component. However, it is widely used for convenience in this type of calculations. 8  For cities covered by networks measuring concentrations of particulate matter. 15 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Table 8  Estimated Health Impact of Urban Air pollutant concentrations result in fewer attribut- Pollution in Cities with PM Monitoring able cases. However, with only three monitoring Data stations in the Cali Metropolitan Area (SISAIRE 2012), this result should be interpreted cautiously. Total Total Health categories cases DALYs Premature mortality 5,027 37,703 An important health outcome attributable to air Chronic bronchitis 4,675 10,285 pollution (mortality in children under age 15 from Hospital admissions 9,492 152 respiratory causes) is not included in this analysis Emergency room/ 186,208 838 since we lacked age-specific mortality by cause and outpatient hospital visits by city, but it should ideally be part of forthcoming Restricted activity days 32,748,479 9,825 updates. In order to properly link exposure to ef- Lower respiratory illness 374,314 2,433 fects with high confidence, it is necessary to collect in children health statistics on relevant outcomes within each Respiratory symptoms 104,225,594 7,817 environmental jurisdiction. This will allow public TOTAL 64,354 health and environmental authorities to track real Source: Authors’ estimates. progress in reducing environmental health threats to local communities. Percentage of Exposed Population Figure 5  and Attributable Mortality by City 3.4 Health cost of urban air pollution (With Monitoring Network) The estimated annual cost of health impacts from 60 urban air pollution is presented in Table 9. The 50 cost of mortality is based on the Value of a Statisti- cal Life (VSL). We are not reporting a cost based on 40 the Human Capital Approach (HCA), since we be- 30 lieve that an indicator based on foregone income 20 due to premature mortality severely underesti- 10 mates the true cost to society that excess mortal- ity represents in a rapidly improving context such 0 as that of urban Colombia. The total estimated an- Bogota AMVA Cali Bucaramanga Cucuta Pereira Ibague Soacha Manizales Palmira Sogamoso Yumbo nual health cost attributed to outdoor air pollution is about 5.7 trillion COP, or 1.1% of GDP in 2009. % of exposed population % of cases Mortality represents about 79% of the total esti- Source: Authors’ estimates. mated cost. The estimated cost per case of premature mor- The four largest urban centers (Bogotá, AMVA, Cali tality or specific health end-point is presented in and Bucaramanga) account for more than 80% Table 10. The VSL used in this assessment is that of the population exposed and attributable cas- utilized by the government in Bogotá for the Bo- es. There is good concordance between the ex- gotá Ten-Year Decontamination Plan (Plan Decenal posed population and the concentration of cases de Descontaminación de Bogotá, PDDB 2009). In (Figure 5), except in the case of Cali where lower this report, the value of statistical life was derived 16 Estimated Annual Cost of Health Impacts (Billion Pesos) Table 9  Health categories Total annual cost (billion pesos) Percent of total cost* (mean) Mortality 4,519 79 Morbidity Chronic bronchitis 40 1 Hospital admissions 56 1 ER visits/outpatient hospital visits 58 1 Restricted activity days (adults) 839 15 Lower respiratory illness in children 84 1 Respiratory symptoms (adults) 113 2 Total cost of morbidity 1,189 21 TOTAL COST (mortality and morbidity) 5,708 100 Source: Authors’ estimates. *Annual cost is rounded to nearest billion, and percentages are rounded to nearest percent. from: (i) Ortiz et al. (2009) in São Paulo, (ii) Hammit on consultations with health authorities and the and Ibarraran (2002) in Mexico City, and (iii) Bow- upper bound of the publicly listed prices that public land and Beghin (2001) in Santiago de Chile. The insurers pay healthcare providers, which in turn are PDDB does not specify which value is chosen or, deemed the most adequate reflection of the true in the case of a combination of the values from cost of treatment in Colombia (see Annex 3). The the three studies, which relative weight of pool- cost per case (comprising Cost of Illness plus the ing method was used. However, since the value proxy for WTP) is the basis for the estimation of the is not far off from the most recent and most rel- annual costs in Table 9 and multiplying each cost evant reference (Ortiz et al. 2009; see Annex 1), for the cases in Table 8. we found it most appropriate to use a locally esti- mated value for VSL. Likewise, the WTP proxy ap- Table 11 details the baseline data that were used plied is based on the ratio of Cost of Illness to Will- for the estimation of the cost of illness and the ingness to Pay (WTP) reported for Bogotá (PDDB costs of time lost to illness. For comparability with 2009). The calculated cost of treatment was based previous estimates (see Annex 3) we kept most of Table 10  Estimated Unit Cost by Health End-Point Total cost per case Cost of illness per WTP proxy Health categories (000 pesos) case (000 pesos) (000 pesos) Mortality 1,008,000 N/A N/A Chronic bronchitis 8,597 2,629 5,968 Hospital admissions 5,853 1,790 4,063 Emergency room/outpatient hospital visits 312 95 216 Restricted activity days (adults) 26 7.8 18 Lower respiratory illness in children 224 68 155 Respiratory symptoms (adults) 1.08 0.33 0.75 Source: Authors’ estimates. 17 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 the assumptions regarding duration of illness, rate 75% of average urban wage, and applied this cost and length of hospitalization, average time lost per both to working and nonworking individuals, based health end-point, frequency of doctor visits, and on the assumption of an equivalent opportunity discount rate. We also valued time lost to illness at cost for both categories. Table 11  Baseline Data for Cost Estimation Baseline Source Cost Data for All Health End-Points: Cost of hospitalization (pesos per day) 246,000 Per consultations with medical ser- Cost of emergency visit (pesos): urban 95,000 vice providers and health authori- ties (see Annex 3). Rounded to the Cost of doctor visit (pesos) (mainly private doctors): urban 29,000 nearest thousand pesos. Value of time lost to illness (pesos per day) 31,000 Based on urban wages in Colombia. Chronic Bronchitis (CB): Average duration of illness (years) 20 Based on Shibuya et al. (2001). Percent of CB patients hospitalized per year 1.5% From Schulman et al. (2001) and Average length of hospitalization (days) 10 Niederman et al. (1999). Average number of doctor visits per CB patient per year 1 Percent of CB patients with an emergency doctor/hospital 15% outpatient visit per year Estimated lost workdays (including household workdays) 2.6 Estimated based on frequency of per year per CB patient doctor visits, emergency visits, and hospitalization. Annual real increases in economic cost of health services 2% Estimate. and value of time (real wages) Annual discount rate 3% Applied by WHO for health effects. Hospital Admissions: Average length of hospitalization (days) 6 Estimates based on Larsen (2004). Average number of days lost to illness (after 4 hospitalization) Emergency Room Visits: Average number of days lost to illness 2 Restricted Activity Days: Average number of days of illness (per 10 cases) 2.5 Lower Respiratory Illness in Children: Number of doctor visits 1 Total time of caregiving by adult (days) 1 Estimated at 1–2 hours per day. Source: Authors. 18 4. Inadequate Water Supply, Sanitation and Hygiene—Detailed Technical Findings 4.1 Inadequate WASH and diarrheal illness with relevant health statistics is Colombia’s Na- Inadequate quantity and quality of potable water sup- tional Demographic and Health Survey (Encues- ply, sanitation facilities and practices, and hygiene ta Nacional de Demografía y Salud 2010 [ENDS conditions are associated with various illnesses both 2010]) by the Association for the Well-being of the in adults and children, as discussed in Larsen (2003). Colombian Family (Asociación Probienestar de la Diarrheal illness in children under age five (mortali- Familia Colombiana, Profamilia), the Ministry of ty and morbidity) and adults (morbidity) is the major Social Protection (Ministerio de Protección Social, burden of disease associated with inadequate WASH. Although diarrheal illness is generally not as serious Water Supply and Sanitation in Table 12  as some other waterborne illnesses, it is more com- Colombia in 2010 (percent) mon and affects a larger number of people. Table 12 Urban Rural presents the water supply and sanitation situation in Piped water 91.7 59.6 Colombia in 2010. About 98% of the urban popula- Well water 0.7 10.8 tion and 73% of the rural population have access to Surface water 0.3 19.8 improved water supply in Colombia. Rainwater 1.2 4.1 Tanker truck 0.1 0.5 In the last 10 years, the amount of the Colombi- Bottled water/demijohn 5.1 3 an population with unimproved sanitation has de- Other 0.9 2.2 creased significantly. Figure 6 below presents the Water within 15 minutes 99.7 93.6 percentage of the population with improved sanita- Flush toilet 98.1 80.3 tion in rural areas in 2000–2010. Pit latrine 0.5 3.8 No facility 1.3 15.8 Improvements in sanitation were one of the rea- Other 0 0.2 sons for substantial reductions in child mortality and diarrheal child mortality in Colombia. Reliable Source: ENDS 2010. data on the health and nutritional status of the Co- lombian population are for the most part readily Percentage of Population with Access Figure 6  available. Public health information systems con- to Improved Sanitation tain complete and reliable data on cause-specif- 90 ic adult and child mortality, child nutritional sta- tus, and incidence of infectious diseases. These 85 % rural population data were applied for estimating the health effects 80 or disease burden from environmental health risk 75 factors. This report uses reported DANE (Ministry 70 of Health) data, household surveys in Colombia (ENDS) and data reported by WHO that, combined, 65 provide indications of several dimensions of health 60 2000 2010 and nutrition in Colombia needed for this study of environmental health. The main household survey Source: ENDS 2000, ENDS 2010. 19 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 MPS) and the US Government through USAID. This urban and rural areas is the same as that estimated survey includes information on child nutritional sta- for Colombia as a whole. tus, and estimates of the cause-specific structure of child and adult mortality in Colombia by WHO The reported data suggest that diarrheal mortality (2009). The reference year for this study is 2009. decreased more than 50% in relative terms. At the same time, diarrheal prevalence decreased only 4.2 Child mortality and morbidity 10%. The estimate of diarrheal cases per person According to ENDS 2010, the under-five child mor- is presented below. ENDS 2010 contains import- tality rate in Colombia was around 22 per 1,000 live ant information on the prevalence of diarrhea and births in 2010 (25 in rural areas and 21 in urban ar- symptoms of respiratory infections in children un- eas). It decreased about 9% from 2000. Based on der age five. ENDS 2010 reports a two-week diar- statistics of the under-age-five population in Colom- rheal prevalence rate of 11.6 to 15.2% in urban-ru- bia (DANE 2009), from a population of 3 million chil- ral areas of Colombia. The annual incidence of dren under age five in urban and 1.3 million in rural diarrhea per child per year is calculated based on areas, an estimated 12,600 children under age five the number of prevalence periods in a year and is died in Colombia in urban areas and 6,400 in rural adjusted for the duration of the diarrheal illness. areas in 2009. WHO and DANE provide estimates of The average duration of diarrhea is assumed to be cause-specific child mortality in 2008–2009 for Co- 3 to 4 days. The incidence of diarrhea is therefore lombia (WHO 2010; DANE 2010). According to these 2.4 to 3.6 cases per child per year in urban-rural ar- estimates, about 13 to 16% of mortality among chil- eas, according to ENDS 2010. dren under five in Colombia was from infectious and parasitic diseases, and 84 to 87% was from oth- Although diarrheal prevalence decreased in the er causes (Figure 7). For purposes of this report, it last decade, it is still higher than average in chil- is assumed that the structure of child mortality in dren from lower-income households. Thus, the im- pact of inadequate WASH is higher in these income Figure 7  Estimates of Cause-Specific Non- groups. For diarrhea, 88% of cases globally are at- Accidental Mortality Among Children tributed to water, sanitation and hygiene (Prüss et Under Age Five in Colombia from al. 2002; Prüss-Ustün et al. 2004). None of the sur- Infectious and Parasitic Diseases, 2008 veys reports diarrheal disease among the popula- tion aged five years and older. Results from house- hold surveys in other countries indicate that the 20% 20% incidence rate in children under age five is around 7 to 10 times higher than among the population 7% aged five years and older. If this is also the case 53% in Colombia, there are 2.4 to 3.2 cases of diar- rhea per person/per year in the population aged five years and older, totaling over 13 million cas- es. Thus, in total there were over 24 million cases Diarrhea Meningitis Preumonia Other inflectiones and of diarrhea in Colombia in 2009. About 21 million parasitic diseases of these cases are attributable to inadequate water Source: Produced from DANE estimates of mortality among chil- supply; sanitation and hygiene, representing a loss dren in Colombia in 2009 (DANE 2010). of over 29,100 DALYs (see Table 14). 20 Table 13  Diarrheal Illness in Colombia among Children under Age Five in 2000–2009 Urban Rural Total Children under five mortality, 2000 24 36 28 (per 1,000 live births) Children under five mortality, 2009 21 25 22 (per 1,000 live births) Diarrheal mortality, 2000 (%) 7.3% Diarrheal mortality, 2009 (%) 3.5% Diarrheal morbidity, 2000 13.2% 15.6% 13.9% (prevalence last 2 weeks) By household wealth index Lowest 17.5% Second 15.9% Middle 12.9% Fourth 10.4% Highest 10.2% Diarrheal morbidity, 2009 11.6% 15.2% 12.6% (prevalence last 2 weeks) By household wealth index Lowest 16.1% Second 14.4% Middle 11.3% Fourth 10.6% Highest 7.4% Source: ENDS 2000, ENDS 2010. Estimated Annual Cases of Diarrheal Mortality and Morbidity from Water-Sanitation- Table 14  Hygiene (WASH) in Colombia, 2009 Annual cases of Attributable Annual cases DALYs diarrhea fraction from WASH from WASH from WASH Urban Rural Urban Rural Urban Rural Children under age 5: 440 220 400 200 13,500 6,800 mortality Children under age 5: 7.2 4.0 6.3 3.5 2,540 1,400 88% morbidity (million) Population aged 5+: 9.1 3.8 8.0 3.3 10,750 4,440 morbidity (million) Total morbidity (million) 16.3 7.8 14.3 6.9 13,290* 5,840* Source: Authors’ estimates. Note: *DALYs from diarrheal morbidity. 21 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 of piped water supply and sewage systems. The 4.3 Child nutritional status and WASH health impact programs are justified on the grounds that the ben- Studies in different low-income countries with sim- efits are concentrated primarily among the poor. ilar water supply, sanitation and hygiene problems Hygiene programs have estimated benefits far in suggest that measures to reduce environmen- excess of costs and should receive the highest pri- tal damages are justified in a number of areas on ority. The same applies to programs aimed at en- cost-benefit grounds as well on grounds of benefit- couraging the disinfection of drinking water. All in- ing the poor. For water supply and sanitation, im- terventions to improve WASH also have the benefit provements in facilities in rural areas yield benefits of reducing the burden of malnutrition. in excess of costs under most assumptions. In ur- ban areas, the focus should be on the monitoring Commonly used indicators of poor nutritional sta- of drinking-water quality and on the rehabilitation tus in children are underweight, stunting and wast- ing. Underweight is measured as weight-for-age relative to an international reference population.9 Prevalence of Underweight in Table 15  Stunting is measured as height-for-age, and wast- Children Under Age Five in Colombia ing is measured as weight-for-height. Underweight Urban Rural Total is an indicator of chronic or acute malnutrition or Moderate and severe 2.9% 4.7% 3.4% a combination of both. Stunting is an indicator of underweight 2009 chronic malnutrition, and wasting an indicator of Severe underweight 2009 0.5% 0.9% 0.6% acute malnutrition. Underweight status among Moderate and severe children under age five is most commonly used in underweight by household assessing the risk of mortality and morbidity from wealth index poor nutritional status (Fishman et al. 2004). Lowest 7.7% Second 5.0% A child is defined as moderately underweight or Middle 3.4% stunted if his or her weight or height is in the range Fourth 3.3% of –2 to –3 standard deviations (SD) below the Highest 1.6% weight or height of the median child in the inter- Moderate and severe n.a. n.a. n.a. underweight, 2000 national reference population, and severely under- Severe underweight 2,000 n.a. n.a. n.a. weight or stunted if the child’s weight or height is Moderate and severe –3 SD below the weight or height of the median underweight by household child in the reference population. The standard de- wealth index viations are also called z-scores and noted as WAZ Lowest 9.3% (weight-for-age z-score). ENDS 2010 is used here Second 9.0% to provide some perspectives on the nutritional sta- Middle 5.6% tus among children under age five in Colombia. Fourth 4.4% Highest 3.0% Malnutrition status improved in all income groups, Source: ENDS 2010. with the greatest improvement achieved in the 9 The recently published WHO international reference population (representing a diverse group of countries) is increasingly replacing the interna- tional reference population defined by the US National Center for Health Statistics (NCHS). 22 second-lowest income group (about 4% fewer chil- Malnutrition Status (percent) by Figure 8  dren under age five were malnourished in Colom- Income Groups in Colombia bia by 2009). 10 9 Measuring the burden of disease and subsequent 8 7 economic costs from environmental health risks is 6 important in helping policy makers to better inte- 5 grate environmental health into economic develop- 4 3 ment, and specifically in their decisions related to 2 resource allocation among various programs and 1 activities to improve child health. Building on previ- 0 Lowest Second Middle Fourth Highest ous estimates, and due to the linkages among en- Ends 2010 Ends 2000 vironmental health, malnutrition and disease, WHO recently revised the burden-of-disease estimates, Sources: ENDS 2000, ENDS 2010. taking into account malnutrition-mediated health impacts associated with inadequate water and The basic method applied to estimate the conse- sanitation provisions and improper hygiene practic- quences of malnutrition in terms of health impact es (Fewtrell, Prüss-Üstün et al. 2007). from infectious diseases in children under age five consists of the following steps (Blössner and de The new WHO estimates reveal that the environ- Onis 2005; Fishman et al. 2004): mental health burden in children under age five is substantially higher when all linkages through • Estimation of the number of children with a WA malnutrition, especially in those subregions where below –1 standard deviations (SD) of the mean; malnutrition and poor environmental conditions • Estimation of fractions of mortality due to di- coexist, are incorporated. In a study of the linkage arrheal disease, malaria, measles, lower re- between the global disease burden and the envi- spiratory infections, other infectious diseases ronment (Prüss-Üstün and Corvalán 2006), it was (besides HIV) and PEM that are attributable to estimated that 50% of malnutrition is attributable malnutrition, based on relative risks from the to the environment, essentially to water, sanitation literature; and hygiene (pooled expert opinion based on liter- • Calculation of the disease burden attributable ature review). to malnutrition by multiplying mortality, inci- dence and DALY statistics with attributable Blössner and de Onis (2005) presented a method- fractions. ology to quantify the burden of disease associated with malnutrition. To quantify the impact of malnu- Fishman et al. (2004) present estimates of in- trition, it is necessary to factor in population data creased risk of cause-specific mortality and all- of weight-for-age (WA) in children and the disease cause mortality in children under age five with mild, burden (deaths, incidence and DALYs) of infectious moderate and severe underweight from a review of diseases and protein-energy malnutrition (PEM). available studies. Severely underweight children For Bolivia, such information may be obtained from (WA <–3 SD) are 5 times more likely to die from DHS 2008 and WHO deaths, incidence and DALY measles, 8 times more likely to die from acute low- tables from Global Burden of Disease (GBD) 2008. er respiratory infections (ALRI), nearly 10 times 23 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 more likely to die from malaria, and 12 times more the population-attributable fractions (AF) of mortal- likely to die from diarrhea than non-underweight ity from underweight in children under age five: children (WA >–1 SD). Even mild underweight dou- n n bles the risk of death from major diseases in early AF = ( ∑ P RR − 1 /∑ P RR (1) ( i −1 i i i −1 i i childhood (see Table 16). where RRi is the relative risk of mortality for each Child underweight also increases the risk of ill- of the four WA categories (i) in Table 5; and P is the ness. Fishman et al. (2004) present estimates of percentage of children in each of the four catego- increased risk in children under age five with mod- ries (i). erate and severe underweight (WA <–2 SD). The largest increased risk of illness is for pneumonia/ In addition to these malnutrition-related mortalities, ALRI. No increased risk of measles is confirmed Fishman et al. (2004) include 100% of PEM mortal- (see Table 16 and Table 17). ity and a share of mortality from perinatal condi- tions (low birth weight associated with low maternal The WA prevalence rates and the relative risks of pre-pregnancy body mass index [BMI <20 kg/m2]). cause-specific mortality can be used to estimate About 9% of infants had low birth weight (<2,500 g) in Colombia in 2008.10 Relative Risk of Mortality from Mild, Table 16  Moderate and Severely Underweight ENDS 2010 data, needed to estimate the preva- in Children Under Age Five lence of child underweight in Colombia using the NCHS reference population, were not available at Weight-for-age <–3 –2 to –1 to >–1 the time this report was being prepared. Applying (WA) SD –3 SD –2 SD SD the assumption (Blossner, Ortiz 2005) about nor- Pneumonia/ALRI 8.1 4.0 2.0 1.0 mal distribution of malnourished children under Diarrhea 12.5 5.4 2.3 1.0 age five, mild underweight children under age five Measles 5.2 3.0 1.7 1.0 were estimated for Colombia. Malaria 9.5 4.5 2.1 1.0 All-cause 8.7 4.2 2.1 1.0 Attributable Fractions of Under-Five Table 18  Source: Fishman et al. (2004). Child Mortality and Morbidity from Underweight Relative Risk of Illness from Table 17  Mortality Morbidity Moderate and Severe Underweight in Children under Age Five Pneumonia/ALRI 5.7% 0.3% Diarrhea 7.4% 0.9% Weight-for-age (WA) <–2 SD >–2 SD Malaria 6.3% 0.3% Pneumonia/ALRI 1.86 1.0 Measles N/A N/A Diarrhea 1.23 1.0 Protein-energy malnu- 100% 100% Measles 1.00 1.0 trition (PEM) Malaria 1.31 1.0 Other causes 6.0% N/A Source: Fishman et al. (2004). Source: Estimated in this report. 10  Source: ENDS 2010. 24 The application of the relative risks of illness and Estimated Annual Cost of Health Figure 9  WA malnutrition rates to (1) indicates that about Impacts Associated with Inadequate 6% of pneumonia/ALRI mortality and 7% of diar- WASH rhea mortality in children under age five in Colom- 3,000 bia are from malnutrition (Table 18). 2,500 Table 19 presents the deaths among children un- 2,000 Billion COP der age five that could be associated with malnu- 1,500 trition. Since mortality and incidence from diarrhe- 1,000 al and lower respiratory diseases in children were 500 already counted as an impact of inadequate WASH and of indoor air pollution, only other diseases 0 Diarrheal Diarrheal ARI Malnutrition were included in the costs of malnutrition. Morbidi- morbidity mortality morbidity mortality ty was not included because no data were available Rural cost Urban cost on the prevalence of other diseases. Source: Authors’ estimates. 4.4 Cost of inadequate WASH in Colombia and Table 19) are multiplied by an average value The estimated annual cost of health impacts from of mortality case using the HC approach.11 Morbidi- urban air pollution is presented in Table 20 and ty is valued using the cost of illness (COI) approach Figure 9.The cost of mortality among children un- der age five is based on using the human capital (HC) approach (discounted lifetime income of the Estimated Annual Cost of Health Table 20  average person in Colombia). Annual mortality cas- Impacts (Billion Pesos) es associated with inadequate WASH (Table 14 Urban annual Rural annual cost (billion cost (billion Health categories pesos) pesos) Estimated Deaths among Children Table 19  Mortality: under Age Five that could be Children under age 140 70 Associated with Malnutrition in 5: diarrheal mortality Colombia, 2009 Children under five: 232 148 Number of malnutrition- malnutrition-related related deaths mortality Morbidity: Post-neonatal deaths 59 Diarrheal illness 2,214 629 Measles 3 ARI illnessa 10 2 Protein-energy 400 malnutrition (PEM) TOTAL COST 2,596 849 (mortality and Other causes 560 morbidity) Total 1022 Source: Estimates by the authors. Source: Authors’ estimates. * Annual cost is rounded to nearest billion. a Malnourishment-related illness. 11  351 million COP 25 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 (Annex 1). The WTP proxy applied is based on the of time lost to illness. For purposes of comparability ratio of cost of illness to willingness to pay reported with previous estimates (see Annex 3), this analysis for Bogotá (PDDB 2010). The cost per case (com- kept most of the assumptions regarding duration of prising cost of illness plus the proxy for WTP) is illness, rate and length of hospitalization, average the basis for the estimation of the COI for diarrhea time lost per health end-point, frequency of doctor (Annex 1) by multiplying each cost for the cases in visits, and discount rate. It also valued time lost to Table 14. illness at 75% of the average urban wage, and ap- plied this cost both to working and nonworking indi- Table 21 details the baseline data that were used viduals, based on the assumption of an equivalent for the estimation of the cost of illness and the costs opportunity cost for both categories. Table 21  Baseline Data for Cost Estimation Baseline Source: Percent of diarrheal cases treated at medical facilities 37–46% ENDS 2010. (children < age 5) and with medicines Percent of diarrheal cases treated with ORS (children < age 5) 68–77% ENDS 2010. Percent of diarrheal cases treated at medical facilities 30–35% Estimated from a (population > age 5) and with medicines combination of INS data and ENDS 2010. Average cost of doctor visits (urban and rural): pesos 29,000 Per consultations with Average cost of medicines for treatment of diarrhea: pesos 3,600–12,600 pharmacies, medical service providers, and Average cost of ORS per diarrheal case in children: pesos 1,200 health authorities. Average duration of diarrheal illness in days (children and adults) 3–4 Assumption. Hours per day of caregiving per case of diarrhea in children 2 Assumption. Hours per day lost to illness per case of diarrhea in adults 2 Assumption. Value of time for adults (caregiving and ill adults): pesos/hour 1,500–3,600 Based on urban and rural wages in Colombia. Hospitalization rate (% of all diarrheal cases): children under age 5 0.75% Adjusted based on Hospitalization rate (% of all diarrheal cases): children under age 5 0.50% evidence from Egypt (Larsen 2004). No data available for Colombia. Average length of hospitalization (days) 2 Adjusted from Egypt (Larsen 2004). Time spent on visitation (hours per day) 4 Assumption. Average cost of hospitalization (pesos per day) 180,000 Per consultations with hospitals. Percent of diarrheal cases and hospitalizations attributable to water, 88% Prüss-Ustün et al. sanitation and hygiene 2004. Source: Authors. 26 Box 2  The Association between Inadequate Water Supply, Sanitation and Hygiene and Prevalence of Malnutrition Malnutrition in children under five years of age makes infections worse and often more frequent. Substantial research con- firms that early childhood repeated infections, including diarrhea, may account for at least one third of weight gain retarda- tion in children under five years of age, thus contributing as much as half of disease burden from malnutrition. WHO study (Fewtrell et al. 2007) reports that malnourished children have a risk of dying from various diseases that is several times higher than non-malnourished children. Then the disease burden from inadequate water supply, sanitation and hygiene (WASH) as much as doubles compared to only considering the direct effect of inadequate WASH on diarrhea. The methodol- ogy proposed by WHO incorporates not only direct health risks from environmental factors (such as diarrheal disease burden from poor water and sanitation), but also seeks to include the indirect risks (concentrating on WASH and its indirect impact on mortality through malnutrition). Thus, while a traditional burden of disease calculation would associate WASH with only di- arrheal diseases (see figure below), the inclusion of the indirect path implies the need to include all diseases attributable to malnutrition (as 50% of the consequences of malnutrition are, in turn, attributed to poor WASH). The Health Effects of Environmental Risks Factors Poor water Inadequate sanitation Improper personal and household hygiene Indoor air pollution Inadequate WRM Diarrhea Acute lower respiratory Direct infections Malari ct Malnutrition Diarrhea (WSH) Dire (50% from bad sanitation ALRI (IAP) and repeated infections Malaria (IWRM) 50% from food availability) Tuberculosis (IAP) Protein-energy malnutrition Indirect (WSH) Diarrhea (food) ALRI (non IAP) Infections non-environmentally related (excluding HIV) Source: World Bank (2008) Environmental Health and Child Survival. Epidemiology, Economics. Experiences. Washington DC.; Fewtrell, L., A. Prüss-Üstün, R. Bos, F. Gore and J. Bartram. 2007. Water, Sanitation and Hygiene- Quantifying the health impact at national and local levels in countries with incomplete water supply and sanitation coverage. Environmental Burden of Disease se- ries, No. 15. Geneva, World Health Organization. 27 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 5. Indoor Air Pollution from the Use of Solid Fuels—Detailed Technical Findings 5.1 Indoor smoke and health outcomes. The risks are generally reported rela- Household air pollution from use of solid fuels for tive to the risks of health effects from the use of cooking and other purposes is associated with sub- liquid fuels (e.g., LPG). The evidence from stud- stantial health effects, particularly among young ies around the world is summarized in meta-anal- children and adult women, because these groups yses by Desai et al. (2004), Smith et al. (2004), tend to spend the most time in the household envi- Dherani et al. (2008), Kurmi et al. (2010), and Po ronment. Combustion of solid fuels generates fine et al. (2011) and include elevated risks of acute particulates (smoke) and other pollutants harmful lower respiratory infections (ALRI) in children un- to human health. Combustion of biomass (straw/ der age five, and chronic obstructive pulmonary shrubs/grass, agricultural crop residues, and ani- disease (COPD), chronic bronchitis (CB), lung mal dung) tends to generate the most smoke, fol- cancer, and tuberculosis in adult women.12 Ezza- lowed by wood and coal/charcoal. Fossil fuels (e.g., ti and Kammen (2001, 2002) also document ele- liquid petroleum gas [LPG], biogas, kerosene) are vated risks of ALRI in adult women and acute up- the cleanest and generate the least smoke. per respiratory infections (AURI) in children and adult women. Studies of the health effects of out- About 50% of rural households in Colombia use door ambient particulate matter (PM) have found wood/charcoal and other solid fuels (e.g., agricul- that exposure to PM increases the risk of cardio- tural residues, straw), according to ENDS 2010. vascular mortality (Pope et al. 2002). A recent The rate was about 60% in 2000 (ENDS 2000). study in Guatemala found that cooking with wood Only about 2% of urban households use these fu- on open fires, compared to cooking with wood els (Table 22). and using an improved chimney stove, is associ- ated with higher systolic blood pressure among Combustion of solid fuels for cooking is a ma- adult women (McCracken et al. 2007). Elevated jor source of household air pollution in develop- systolic blood pressure is associated with an in- ing countries. Combustion of these fuels is as- creased risk of cardiovascular disease and mor- sociated with an increased risk of several health tality (Lawes et al. 2004). Table 22  Household Fuels Used for Cooking In light of the evidence from these studies, the rel- in Colombia, 2010 (percent) ative risks of health effects from the use of wood and other biomass fuels for cooking applied in this Urban Rural study to Colombia are presented in Table 23. These Electricity 4.5 2.7 relative risks are applied to children under age five LPG, natural gas 90.9 45 and to adult women in households using these fu- Kerosene 0.2 0.2 els, because these are the household members Coal, lignite 0.1 0.9 who are most exposed to air pollution from cook- Firewood, straw 1.9 49.3 ing. Only acute respiratory infections and COPD No cooking in household 2.5 1.7 were considered as health impacts of indoor air Source: ENDS 2010. pollution in Larsen (2002). 12  CB is a subset of COPD. 28 Relative Risks of Health Effects from Cooking with Wood and Table 23  other Biomass Fuels Applied to Colombia Population group Health outcome Relative risk ratios (RR) Children < age 5 Acute respiratory infection (ALRI) 2.0 Women ≥ age 30 Chronic obstructive pulmonary disease (COPD) 2.8 Chronic bronchitis (CB) 2.4 Ischemic heart disease 1.19 Cerebrovascular disease 1.26 Hypertensive heart disease 1.51 Other cardiovascular disease 1.12 Lung cancer (LC) 1.5 Source: Based on Desai et al. (2004), Smith et al. (2004), Dherani et al. (2008), Kurmi et al. (2010), Po et al. (2011), and estimates of cardiovascular disease risks based on McCracken et al. (2007) and Lawes et al. (2004) as presented in Larson (2012). To calculate the fraction of health outcomes asso- are then multiplied by the estimated annual base- ciated with the use of wood and other biomass for line mortality in Colombia for each health outcome cooking, the following attributable fraction formu- in order to arrive at annual premature mortality as- la is applied: sociated with the use of wood and biomass. n n AF = ( ∑ P RR − 1 /∑ P RR (7.1) ( i −1 i i i −1 i i 5.2 Baseline mortality and morbidity Pi=1 and Pi=2 is the share of the population cook- Baseline mortality is estimated from DANE 2010, ing with wood or other biomass and with other which presents estimates of cause-specific mor- types of fuels, respectively; and RR is the relative tality in Colombia by age group. ARI mortality and risk of morbidity and mortality from indoor cooking morbidity in children under age five is presented in with wood and biomass. The attributable fractions Table 24 below. Table 24  ARI Illness in Colombia among Children Under Age Five in 2000–2009 Urban Rural Total ARI mortality, 2000 (%) 7.5% ARI mortality, 2009 (%) 7–12% ARI morbidity, 2000 (prevalence last 2 weeks) 12.3% ARI morbidity, 2009 (prevalence last 2 weeks) 6.1% 5.7% 6.0% By household wealth index Lowest 6.5% Second 6.9% Middle 6.7% Fourth 4.7% Highest 3.8% Sources: ENDS 2000, ENDS 2010. 29 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 ARI prevalence among children under age five de- baseline cases of morbidity. Baseline cases of ARI creased by more than 50% in the last 10 years. in Colombia are estimated from ENDS in the same ARI mortality is at about the same level; the differ- manner as for diarrheal prevalence. The baseline ence may be explained by reporting errors and the prevalence of COPD is estimated from Caballero et uncertainty of WHO estimates of deaths by cause al. (2008) and Shibuya et al. (2001). among children under age five. Estimated annual cases of ARI and CB from the use of wood and other biomass fuels are about 3 mil- 5.3 Mortality and morbidity attributable to indoor lion in 2009, representing a loss of about 6,400 household solid fuel use DALYs (see Table 25).14 An estimated 950 to 1,050 children and women died prematurely from the use of wood and other Annual cases of mortality attributed to indoor air biomass fuels for cooking in Colombia in 2009 (see pollution are presented in Figure 10. Table 25). About 200 of these deaths were among children under age five and 800 were among adult 5.4 Health cost of indoor household solid-fuel use women. The deaths represent 11,600 years of life The estimated annual cost of health impacts at- lost (YLL) per year, of which about 60% are among tributed to indoor air pollution in rural areas is pre- children under age five.13 Morbidity from the use sented in Table 27. The cost of mortality is based of wood and other biomass fuels can be estimat- on the Value of a Statistical Life (VSL) for adult ed in the same manner as for mortality by multiply- women and the HC approach for children under ing the attributable fractions from equation (1) with age five. Estimated Annual Mortality from Household Use of Wood and Biomass for Cooking Table 25  in Colombia, 2009 Attributable Annual Annual fraction from mortality from DALYs (YLL) Population baseline use of solid use of solid from use of Health outcome group mortality fuels fuels solid fuels ALRI children u-5 603 33% 201 6,837 COPD females 30+ 442 47% 209 1,255 Ischemic heart disease females 30+ 2,320 9% 201 1,208 Cerebrovascular disease females 30+ 1,427 12% 164 985 Hypertensive heart disease females 30+ 556 20% 113 678 Other cardiovascular females 30+ 950 6% 54 323 disease Lung cancer females 30+ 310 20% 62 350 Total 6,608 n/a 1,004 11,635 Source: Authors’ estimates. 13 Years of life lost (YLL) are estimated according to WHO’s calculation of DALYs using age weighting and a three-percent discount rate. 14 Cases of cardiovascular disease and lung cancer are not estimated due to data constraints. DALYs are estimated according to WHO’s calcula- tion of DALYs. 30 The estimated cost per case of premature mor- Annual Cases of Mortality Attributed Figure 10  tality or specific health end-point is presented in to Indoor Air Pollution in Colombia Table 10. Adult mortality is valued with the VSL ap- 250 proach. The VSL used in this assessment is that uti- 200 lized by the government in Bogotá for the Bogotá Annual cases Ten-Year Decontamination Plan (Plan Decenal de 150 Descontaminación de Bogotá, 2010). In that re- 100 port, the value of statistical life was derived from: 50 (i) Arigoni et al. (2009) in São Paulo, (ii) Hammit 0 and Ibarran (2002) in Mexico City, and (iii) Bow- COPD Ischaemic heart disease ALRI Cerebrovascular disease Hypertensive heart disease Lung cancer Other cardiovascular disease land and Beghin (2001) in Santiago de Chile. The PDDB does not specify which value is chosen or, in the case of a combination of the values from the three studies, which relative weight of pool- ing method was used. However, since the value is not far off from the most recent (and perhaps Source: Authors’ estimates. Estimated Annual Cases of Morbidity from Household Use of Wood and Biomass for Table 26  Cooking in Colombia, 2009 Health Population Annual Attributable fraction Annual cases from DALYs from use outcomes group baseline cases from use of solid fuels use of solid fuels of solid fuels ARI children u-5 1,825,000 33% 608,300 1,004 ARI females 30+ 1,148,000 33% 382,698 2,679 COPD females 30+ 2,500 47% 1,184 2,664 Total   2,975,500 n/a 992,182 6,347 Source: Authors’ estimates. Table 27  Estimated Annual Cost of Health Impacts (Billion Pesos) Health categories Total Annual Cost (billion pesos) Percent of total cost* Mortality: Children under age 5 73 6% Adult females over age 30 882 78% Morbidity: COI 46 4% WTP to avoid illness 128 11% TOTAL COST (mortality and morbidity) 1129 100% Source: Authors’ estimates. *Annual cost is rounded to nearest billion, and percentages are rounded to nearest percent. 31 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 most relevant) reference (Arigoni et al. 2009; see kept most of the assumptions regarding duration of Annex 1), we found it most appropriate to use a lo- illness, rate and length of hospitalization, average cally generated value for VSL. The mortality of chil- time lost per health end-point, frequency of doctor dren under age five is valued using the human cap- visits, and discount rate. We also valued time lost ital approach (discounted lifetime income of the to illness at 75% of the average urban wage, and average person in Colombia). Morbidity is estimat- applied this cost both to working and nonworking ed using the COI approach (Annex 1). Likewise, the individuals, based on the assumption of an equiva- WTP proxy applied is based on the ratio of cost of lent opportunity cost for both categories. illness to willingness to pay reported for Bogotá (PDDB 2010). The cost per case (comprising cost The mean estimated annual cost of health impacts of illness plus the proxy for WTP) is the basis for the from indoor air pollution associated with the use estimation of the annual costs in Table 27 by multi- of traditional fuels (mainly fuelwood) in rural areas plying each cost for the cases in Table 26. of Colombia is 1.129 trillion COP (0.22% of GDP in 2009). Child mortality represents 6% of cost; fe- Table 29 details the baseline data that were used male mortality represents about 78% of cost. Acute for the estimation of the cost of illness and the respiratory illness (ARI) in children and adult fe- costs of time lost to illness. For purposes of compa- males and COPD morbidity of adult females repre- rability with previous estimates (see Annex 3), we sent 16% of the cost. Table 28  Estimated Unit Cost by Health End-Point Health categories Total cost per case Cost-of-Illness per WTP proxyb (000 pesos) case (000 pesos)a (000 pesos) Mortality: women over age 30 1,100,000 N/A N/A Mortality: children under age 5 351,000 351,000 N/A COPD 8,060 2,500 5,560 ARI: children under age 5 300 90 210 ARI: women over age 30 420 130 290 Source: Authors’ estimates. a Per treated case. b Per all cases attributed to indoor air pollution. 32 Table 29  Baseline Data for Cost Estimation in Rural Colombia Baseline Source Acute Respiratory Illness (ARI): Percent of ARI cases treated at medical facilities 44.5% ENDS 2010 (rural children). (children < age 5) Percent of ARI cases treated at medical facilities 35–40% Estimated from a combination of INS (adults > age 15) data and DHS 2000. Average cost of doctor visits in rural areas 29,000 Per consultations with pharma- (mainly primary healthcare centers): pesos cies, medical service providers, and Cost of medicines for treatment of ARI: pesos 4,000–24,000 health authorities. Percent of ARI cases treated with medicines 44.5% ENDS 2010. Value of time for adults (caregiving and ill adults): 1,500 Based on rural wages in Colombia. pesos/hour Chronic Obstructive Pulmonary Disease (COPD): Average duration of Illness (years) 20 Based on Shibuya et al. (2001). Percent of COPD patients hospitalized per year 1.5% From Schulman et al. (2001) and Average length of hospitalization (days) 10 Niederman et al. (1999). Average number of doctor visits per COPD 1 patient per year Percent of COPD patients with an emergency 15% doctor/hospital outpatient visit per year Estimated lost workdays (including household 2.6 Estimated based on frequency of workdays) per year per COPD patient doctor visits, emergency visits, and hospitalizations. Cost of hospitalization (pesos per day) 213,000 Per consultations with medical Cost of emergency visit (pesos): rural 72,000 service providers and health authorities. Cost of doctor visit (pesos) (mainly primary 29,000 health clinic): rural Value of time lost to illness (pesos per day) 12,000 Based on rural wages in Colombia. Annual real increases in economic cost of 2% Estimate. health services and value of time Annual discount rate 3% Applied by WHO for health effects. Source: Authors. 33 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 6. Conclusions and Policy Recommendations The 2005 Colombia Country Environmental Analy- as well as to further environmental protection and en- sis, Environmental Priorities and Poverty Reduction, vironmental health agendas. Ambitious policy inter- concluded that: “The analysis of the cost of environ- ventions can find adequate justification in the large mental degradation conducted shows that the most health and economic cost of environmental degra- costly problems associated with environmental deg- dation in Colombia. The results show that address- radation are urban and indoor air pollution; inade- ing these environmental risks should continue to be quate water supply, sanitation, and hygiene; natu- a priority in the environmental and public health pol- ral disasters and land degradation.” Colombia has icy agenda of Colombia. The mean estimated annual made substantial progress in the last years in reduc- cost of health impacts from indoor air pollution asso- ing the population exposure to urban air pollution, ciated with the use of traditional fuels (mainly fuel- inadequate water and sanitation, and indoor air pol- wood) in rural areas of Colombia is 1.129 trillion COP lution from solid fuel use. However, these forms of (0.22% of GDP in 2009). Child mortality represents environmental degradation continue to have a sig- 6% of cost; female mortality represents about 78% nificant impact on the Colombian society in terms of of cost. Acute respiratory illness (ARI) in children and premature mortality and disease. adult females and COPD morbidity of adult females represent 16% of the cost. The analysis in this report relied on large sets of statistics and data from various ministerial depart- Both the health impact assessment and the eco- ments, institutions, and institutes in Colombia. It also nomic valuation can be utilized by policy makers and has drawn heavily from Colombian and international stakeholders in the process of setting environmen- research studies, and benefited from various meth- tal objectives and priorities. The World Bank and odological approaches applied by international or- other donors could also use the results to estab- ganizations such as the World Health Organization. lish support priorities. The report may be useful for Publicly available, easily traceable information and the Colombian Ministry of Finance and the National indicators were used as much as possible, in order Planning Department because preferences and val- to facilitate contrast and upcoming updates. The es- ues are expressed in monetary terms, thus the re- timation of the cost of environmental damage includ- sults can provide additional guidance for the allo- ed many aspects, both economic and otherwise, al- cation of resources across diverse socio-economic though effects considered were only those related development goals. to the three mentioned factors (UAP, WASH, IAP). All costs calculated in this report are expressed in mone- At the request of the Government of Colombia, the tary terms, and they include the cost to society due to methodology used in this study as well as other ap- premature mortality, as well as the cost of healthcare proaches to environmental valuation and case stud- provision to individuals suffering from pollution-relat- ies from the region were presented to a group of ed illnesses and the value individuals place in avoid- technical experts and decision-makers at the “Ande- ing resulting pain and discomfort. Time losses or sav- an Workshop on the Costs of Environmental Degra- ings are valued at the opportunity cost of time. dation” in November 2012, and intended to transfer the methodology to the local experts and govern- In addition, the results of the health costs analysis are ment agencies, to enable them to carry out similar useful to track progress made by policy interventions periodic assessments. 34 Blössner, M., de Onis, M. (2005). Malnutrition: References quantifying the health impact at national and local Abbey, D. et al. (1995). Long-Term Ambient Concen- levels. Geneva, World Health Organization, 2005. trations of Particulates and Oxidants and Develop- (WHO Environmental Burden of Disease Series, ment of Chronic Disease in a Cohort of Nonsmok- No. 12). Global Burden of Disease (GBD) 2008. ing California Residents. Inhalation Toxicology, Vol 7: 19–34. Bogotá (2012). Bogotá local government Depart- ment of Transportation website. Available online at Alberini, A., Krupnick, A. (2000). Cost-of-Illness and http://www.Bogotá-dc.com/trans/bog-tra.htm. Willingness-to-Pay Estimates of the Benefits of Im- proved Air Quality: Evidence from Taiwan. Land Eco- Bowland, B., Beghin, J. (2001). Robust estimates of nomics, Vol. 76, No. 1, pp. 37–53. value of a statistical life for developing economies. Journal of Policy Modeling: 385–389. Arciniegas, A., Rodríguez, C., Pachón, J., Sarmien- to, H., Hernández, L.J. (2006). Estudio de la mor- Caballero, A. et al. (2008). Prevalence of COPD in bilidad en niños menores a cinco años por enfer- Five Colombian Cities Situated at Low, Medium, medad respiratoria aguda y su relación con la and High Altitude (PREPOCOL Study). Chest 2008; concentración de partículas en una zona industri- 133; 343–349. al de la ciudad de Bogotá. ACTA NOVA; Vol. 3, No. 2, junio 2006. Cifuentes A. et al. (2005). Urban air quality and hu- man health in Latin America and the Caribbean. Aristizabal,G. et al. (2009). Asociación entre la con- IADB, Washington DC, USA. taminación del aire y la morbilidad por enfermedad respiratoria aguda en menores de 5 años en tres DANE (2011). Departamento Administrativo Nacio- localidades de Bogotá. nal de Estadística de Colombia. Available online at http://www.dane.gov.co/ Baron, G. (2007). Cuentas de salud de Colombia, 1993–2003, Bogotá. Desai, M.A., Mehta, S., and Smith, K. (2004). In- door Smoke from Solid Fuels: Assessing the Envi- Bell, M.L. et al. (2006). The avoidable health ef- ronmental Burden of Disease at National and Local fects of air pollution in three Latin American cities: Levels. Environmental Burden of Disease Series, Santiago, São Paulo, and Mexico City. Environmen- No. 4. World Health Organization. tal Research, 100: 431–440. Dherani, M., Pope, D., Mascarenhas, M., Smith, K., Bell, M.L., Cifuentes, L.A., Davis, D.L., Cushing, Weber, M., and Bruce, N. (2008). Indoor air pollu- E., Telles, A.G., Gouveia, N. (2011). Environmen- tion from unprocessed solid fuel use and pneumo- tal health indicators and a case study of air pol- nia risk in children aged under five years: a system- lution in Latin American cities. Environ Res. 2011 atic review and meta-analysis. Bulletin of the World Jan;111(1):57–66. Epub 2010 Nov 13. Health Organization, 86:390–98. 35 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Dockery, D.W., Pope, C.A. III, Xu, X., et al. (1993). Fishman, M.S., L.E. Caulfield, M. De Onis, M. Blossner, An association between air pollution and mortality A.A. Hyder, L. Mullany, and R.E. Black. (2004). Child- in six US cities. New England Journal of Medicine, hood and Maternal Underweight. In Ezzati, M., A.D. Lo- 329:1753–1759. pez, A. Rodgers, and C.J.L. Murray (Eds): Comparative Quantification of Health Risks—Global and Regional ECON (2007). Mitigating the cost of environmen- Burden of Disease Attributable to Selected Major Risk tal impacts from inadequate water supply and san- Factors. Vol. 1. Geneva: World Health Organization. itation and hygiene in rural Qena, Egypt. Commis- sioned by the World Bank. Giedion, U. et al. (2010). Los sistemas de salud en Latinoamérica y el papel del seguro privado. Fun- ENDS (2010). Encuesta Nacional de Demografía y dación MAPFRE, Madrid, España. Salud 2010. Asociación Probienestar de la Familia Colombiana (Profamilia), Bogotá, Colombia. Avail- Gouveia, N., Bremner, S. and Novaes, H. (2004). able online at http://www.profamilia.org.co/en- Association between ambient air pollution and cuestas/Profamilia/Profamilia/ birth weight in São Paulo, Brazil. J Epidemiol Com- munity Health, 58:11–17. Ezzati, M., and D. Kammen. (2001). Quantifying the effects of exposure to indoor air pollution from bio- Hammitt, J., Ibarraran, M. (2002). Estimating the mass combustion on acute respiratory infections economic value of reducing health risks by improv- in developing countries. Environmental Health Per- ing air quality in Mexico City. Molina Center for En- spectives 109 (5): 481–88. ergy and the Environment, LaJolla, California, USA. Available online at http://mce2.org/newsletter/ Ezzati M., and D. Kammen. (2002). The health im- nwsltr_2/espaniol/econvalue.htm pacts of exposure to indoor air pollution from sol- id fuels in developing countries: Knowledge, gaps, HEI (2008). Public Health and Air Pollution in Asia and data needs. Environmental Health Perspec- (PAPA): Key Results from Bangkok, Hong Kong, tives, 110 (11): 1057–68. Shanghai, and Wuhan. Health Effects Institute, Boston, Massachusetts, USA. Fewtrell, L., A. Prüss-Üstün, R. Bos, F. Gore, and Bartram, J. (2007). Water, Sanitation and Hygiene: Ibáñez, A.M. (2003). Benefits and Costs of Environ- Quantifying the Health Impact at National and Lo- mental Policy in Colombia: A Review of Valuation cal Levels in Countries with Incomplete Water Sup- Studies. June 25, Bogotá. Prepared for the World ply and Sanitation Coverage. Environmental Bur- Bank. den of Disease Series 15. Geneva: World Health Organization. Jerrett, M. et al. (2009). Long-Term Ozone Expo- sure and Mortality. New England Journal of Medi- Fewtrell, L., R.B. Kaufmann, D. Kay, W. Enanoria, cine, 360;11, 1085–1095. L. Haller, and Colford, J.M. (2005). Water, Sanita- tion and Hygiene Interventions to Reduce Diarrhea Krewski, D., Jerrett, M., Burnett, R.T., Ma, R., in Less Developed Countries: a systematic review Hughes, E., Shi, Y., Turner, C., Pope, C.A., Thurston, and meta-analysis. The Lancet Infectious Diseas- G., Calle, E.E., Thunt, M.J. (2009). Extended fol- es 5:42–52. low-up and spatial analysis of the American Cancer 36 Society study linking particulate air pollution and effects in elderly people: an analysis of six re- mortality. HEI Research Report, 140. Boston, MA: gions in São Paulo, Brazil. J. Epidemiol. Community Health Effects Institute. Health 58, 41–46. Kurmi, O.P., Semple, S., Simkhada, P., Smith, W.C., McCracken, J.P., Smith, K.R., Díaz, A., Mittleman, and Ayres, J.G. (2010). COPD and chronic bron- M.A., Schwartz, J. (2007). Chimney stove inter- chitis risk of indoor air pollution from solid fuel: a vention to reduce long-term wood smoke expo- systematic review and meta-analysis. Thorax, 65: sure lowers blood pressure among Guatemalan 221–228. women. Environmental Health Perspectives, 115(7): 996–1001. Larsen, B. (2004). Cost of environmental damage in Colombia: A Socio-Economic and Environmental MINAM (2010). Política de prevención y control de Health Risk Assessment. World Bank, Washington la contaminación del aire. Ministerio de Medio Am- DC, USA. biente, Vivienda y desarrollo territorial, Bogotá, Co- lombia. Larsen, B. (2003). Hygiene and Health in Develop- ing Countries: Defining Priorities through Cost-Ben- MIPS (2011). Rendición de Cuentas 2010–2011. efit Assessments. International Journal of Environ- Ministerio de la Protección Social, República de Co- mental Health Research. Vol 13: 37–46. lombia. p. 22. Larsen, B. and Skjelvik, J.M. (2012). Environmen- Niederman, M. et al. (1999). Treatment Cost of tal health in Sindh province of Pakistan: An eco- Acute Exacerbations of Chronic Bronchitis. Clinical nomic assessment of health effects and benefits Therapy, 21(3): 576–91. and costs of select interventions. Prepared for the World Bank. O’Neill, M.S., Bell, M.L., Ranjit, N., Cifuentes, L.A., Loomis, D., Gouveia, N., Borja-Aburto, V.H.(2008). Lawes, C.M.M., Vander Hoorn, S., Law, M.R., Elliott, Air pollution and mortality in Latin America: the role P., MacMahon, S., Rodgers, A. (2004). High blood of education. Epidemiology. Nov;19(6):810–9. pressure. In: Ezzati, M. et al. (eds): Comparative quantification of health risks: global and regional Ortiz, R.A., Markandya, A., and Hunt, A. (2009). burden of disease attributable to selected major Willingness to Pay for Mortality Risk Reduction As- risk factors. World Health Organization. sociated with Air Pollution in São Paulo. RBE, Rio de Janeiro, V. 63 n. 1, p. 3–22, Jan-Mar 2009. Lozano, N. (2004). Concentration-response ap- proach for air pollution in Bogotá. Desarrollo y so- Ostro, B. (1994). Estimating the Health Effects of ciedad. Facultad de Economía, Universidad de los Air Pollution: A Method with an Application to Ja- Andes, ISSN 1900–7760(54):133–177. karta. Policy Research Working Paper, World Bank. Martins, M.C.H., Fatigati, F.L., Ve´spoli, T.C., Mar- Ostro, B. (2004). Outdoor Air Pollution - Assessing tins, L.C., Pereira, L.A.A., Martins, M.A., Saldiva, the environmental burden of disease at national P.H.N., Braga, A.L.F. (2004). Influence of socioeco- and local levels. Environmental Burden of Disease, nomic conditions on air pollution adverse health Series, No. 5. Geneva: WHO. 37 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 PDDB (2009). Elementos Técnicos del Plan Dece- Ezzati, M. et al. (eds): Comparative quantification of nal de Descontaminación de Bogotá. Parte 1: Car- health risks: global and regional burden of disease acterización de Material Particulado y Modelos Re- attributable to selected major risk factors. World ceptores de la Alcaldía Mayor de Bogotá, elaborado Health Organization. por la Universidad de los Andes. Schulman, Ronca and Bucuvalas, Inc. (2001). Con- PDDB (2010). Plan Decenal de Descontaminación fronting COPD in North America and Europe: A Sur- de Bogotá. Secretaría Distrital de Ambiente, Gobi- vey of Patients and Doctors in Eight Countries. erno Distrital de Bogotá, Colombia. Shibuya, K., Mathers, C., and Lopez, A. (2001). Po, J.Y.T., FitzGerald, J.M., and Carlsten, C. (2011). Chronic Obstructive Pulmonary Disease (COPD): Respiratory disease associated with solid bio- Consistent Estimates of Incidence, Prevalence, mass fuel exposure in rural women and children: and Mortality by WHO Region. Global Programme systematic review and meta-analysis. Thorax, 66: on Evidence for Health Policy. World Health Orga- 232–239. nization. Pope, C.A. III, Burnett, R.T., Thun, M.J. et al. (2002). SISAIRE (2012). Sistema Nacional de Información Lung cancer, Cardiopulmonary mortality, and Long- Sobre la Calidad del Aire de Colombia. Available on- term exposure to fine particulate air pollution. Jour- line at http://www.sisaire.gov.co:8080/faces/indi- nal of the American Medical Association, 287: ceIndicador/indiceNacional.jsp 1132–1141. Smith, K., Mehta, S., and Feuz, M. (2004). Indoor Pope, C.A. III, Thun, M.J., Nambudiri, M.M. et al. air pollution from household use of solid fuels. (1995). Particulate air pollution as a predictor of In: Ezzati, M. et al. (eds): Comparative quantifica- mortality in a prospective study of US adults. Amer- tion of health risks: global and regional burden of ican Journal of Respiratory and Critical Care Medi- disease attributable to selected major risk factors. cine, 151: 669–674. World Health Organization. Prüss, A., Kay, D., Fewtrell, L., and Bartram, J. US EPA (2012). United States Environmental Pro- (2002). Estimating the Burden of Disease from tection Agency website on ground-level ozone. Water, Sanitation and Hygiene at the Global Lev- Available online at http://www.epa.gov/ground- el. Environmental Health Perspectives. Vol. 110, levelozone/health.html No. 5. WHO (2001). Global Burden of Disease 2001. The Prüss-Ustün, A., and C. Corvalan. (2006). Prevent- World Health Organization, Geneva, Switzerland. ing Disease through Healthy Environments—To- wards an Estimate of the Environmental Burden of WHO (2002b). The World Health Report 2002. The Disease. Geneva: World Health Organization. World Health Organization, Geneva, Switzerland. Prüss-Ustün, A., Kay, D., Fewtrell, L., and Bartram, WHO (2010). World Health Statistics 2010. World J. (2004). Unsafe water, sanitation and hygiene. In: Health Organization, Geneva, Switzerland. 38 Annex 1. Economic Basis for Choice of VSL and WTP The evidence base for willingness to pay (WTP) to For cardiovascular disease (CVD) causes, the avert mortality and morbidity risks from air pollution values are 5.5 days and 11 days, respective- in Colombia is still scarce. However, there are some ly (AHA 2008). important studies that derive estimates from other • One emergency room (ER) visit entails 1 lost examples in the Latin American context. One such workday and 3 additional days of average dis- study was conducted as a technical background for ability duration. the Bogotá Ten-year Decontamination Plan (Plan • One visit to an ARI room entails 0.5 lost work- Decenal de Descontaminación de Bogotá, 2010). day and 1.5 total average disability duration. In this report, the value of statistical life from was • Average duration of Intensive Care Unit derived from: (i) Arigoni et al. (2009) in São Pau- (ICU) hospitalization is 8 days, with an average lo, (ii) Hammit and Ibarran (2002) in Mexico City, total duration of disability of 24 days. and (iii) Bowland and Beghin (2001) in Santiago de • Symptoms from a respiratory disease entail an Chile. The PDDB does not specify which value is average total duration of disability of 3 days. chosen or, in the case of a combination of the val- ues from the three studies, which relative weight of The resulting unit costs by health end-point from pooling method was used. For the cost of illness, the PDDB are listed below: the following assumptions were made by analysts in the PDDB evaluation: The most recent reference used by the PDDB re- port is a study by Ortiz et al. (2009), which was • 90% of all urgent cases that are treated at a considered for this analysis and later discarded healthcare facility are in Acute Respiratory Ill- in the presence of a locally generated VSL. This ness (ARI) rooms. study aims to estimate the population’s willing- • Average duration of the hospitalization for re- ness to pay (WTP) to reduce risks of death asso- spiratory causes is 5 days. Average disabili- ciated with “typical” air pollution policies and con- ty duration for respiratory causes is 10 days. sequently the value of a statistical life (VSL) in São Economic Variables: “Plan Decenal de Descontaminación de Bogotá” (Thousands of Table A.1  Colombian Pesos, 2009) Cost of medical Lost Cost of illness Willingness to Item care productivity (COI) pay (WTP) VSL N/A N/A N/A 1,008,000 Hosp. adm.: respiratory causes 1,290 470 1,760 4,000 Hosp. adm.: CVD causes 1,370 520 1,980 4,600 ER care: respiratory causes 310 130 440 670 Care in ARI rooms 50 60 110 N/A Care in ICUs 6,600 1,000 7,600 N/A Respiratory symptoms (< age 5)   95 95 400 Source: Plan Decenal de Descontaminación de Bogotá (2010). 39 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Paulo, Brazil. Uniquely for that country, the study evenly over the ten-year period, which makes it uses a methodology that has previously been test- possible to avoid discounting the respondents’ an- ed in several industrialized countries (US, Japan, nual payments. Canada, South Korea, England, France and Ita- ly) and involves a computer-based contingent valu- The values of a statistical life estimated from 1-in- ation survey. This survey instrument was adapted 1,000 risk reductions are much higher than those to the Brazilian context and was used to elicit will- estimated using the 5-in-1,000 risk reduction. This ingness to pay measures of reductions in risk of is purely due to the lack of proportionality between death in Brazil. the willingness to pay estimates regarding the dif- ferences in the size of risk reductions. It is sug- Key features of the survey instrument involve elicit- gested that the VSL estimates derived from mean ing the health status of the respondents and their and median willingness to pay estimates for a 5-in- families; explaining basic concepts of probabili- 1,000-risk reduction are of greater policy relevance ty and proposing simple practice questions to fa- since they represent more conservative estimates miliarize the respondents with the probability con- than those estimated using willingness to pay esti- cepts introduced; presenting the leading causes of mates for 1-in-1,000-risk reduction. Thus, for poli- death for a Brazilian individual of the respondent’s cy assessments in São Paulo conservative values age and gender, and setting these in the context of a statistical life ranging between 0.77–1.31 mil- of common risk-mitigating behaviors; and asking lion USD (VSL estimate based on median and mean about the individual’s willingness to pay for risk re- 5-in-1,000 risk reduction correspondingly) are sug- ductions of a given magnitude that occur at a spec- gested. An upper value of 1.3 million USD was ad- ified time. opted as a higher bound to apply benefit transfer to Colombia. The mean and median willingness to pay values were estimated using the interval data model that When compared with European and North Amer- can be generated from the dichotomous choice ican estimates, these values seem to be higher with a follow-up question format. The responses than expected. Given the close link between will- to willingness to pay and follow-up questions were ingness to pay estimates and the population’s in- combined to generate intervals in which the unob- come, lower willingness to pay values for develop- servable respondents’ willingness to pay is to be ing countries might be expected. A possible reason found. The Weibull probability distribution was se- for the high WTP and VSL estimates found in the lected for the random variable of willingness to current study might have been the “cooperative” pay. The statistical willingness to pay model us- behavior observed in many of the respondents. It ing the Weibull distribution is estimated using the is possible that those respondents tried to be “co- maximum likelihood method. The corresponding operative” or helpful by saying “yes” to every ques- values of a statistical life were estimated using tion. It was believed that the relatively high figures both median willingness to pay estimates (conser- in this valuation exercise may be partly due to this vative estimates) and mean willingness to pay val- bias. The value of a statistical life excluding pos- ues. They were obtained by dividing the willingness sible “yeah-say” responses using parametric esti- to pay figures by the corresponding annual risk re- mation of mean and median annual willingness to duction being valued. It was assumed that re- pay in 2003 (Weibull distribution, 95% CI) is 0.4 spondents implicitly considered the risk reduction to 0.5 million USD (VSL estimate based on median 40 and mean 5-in-1,000 risk reduction corresponding- Applying a similar analysis, in the “Plan Decenal de ly) and in 2009 0.45 to 0.56 million USD.15 A low- Descontaminación de Bogotá” (PDDB 2009) a VSL er value of 0.4 million USD (VSL based on median at the level of 0.47 million USD was applied. Thus, 5-in-1,000 risk reduction, excluding “yes-”saying the midpoint of all estimates described above is respondents) was adopted as a lower bound to ap- 1.09 billion COP or 0.5 million USD. The latter esti- ply benefit transfer to Colombia. mate was applied in the report (Table A.1). Application of Benefit Transfer Approach to Estimate VSL for the Population Dying from Table A.2  Pollution in Colombia Median 5-in- Mean 5-in- 1,000 risk 1,000 risk Plan de reduction, reduction, Median Desconta- Value of excluding excluding 5-in- Mean 5-in- minación Statistical Life “yes” saying “yes” saying 1,000 risk 1,000 risk de Bogotá, Midpoint (VSL) respondents respondents reduction reduction 2009 estimate VSL estimates Ortiz et 0.45 0.56 0.77 1.46 from Brazil al. 2009 (USD million) Colombia GDP WDI 235 235 235 235 (USD billion), 2012 2009 Colombia GDP WDI 5.09E+05 5.09E+05 5.09E+05 5.09E+05 (COP billion), 2012 2009 Population WDI 45.65 45.65 45.65 45.65 (million) in 2012 2009 GDP per capita 5,166 5,166 5,166 5,166 (USD) in 2009 Exchange rate 2,156 2,156 2,156 2,156.3 (year average 2003) VSL in Colombia 0.28 0.35 0.54 0.91 0.47 0.51 million USD VSL in Colombia 0.61 0.76 1.16 1.97 1.008* 1.1 billion COP *Source: Plan Decenal de Descontaminación de Bogotá (PDDB 2009). Note: GDP per capita (current USD): Brazil 8,251, Colombia 5,166 (Source: WDI 2012). 15  Adjusted using CPI presented by US Department of Labor, Bureau of Labor Statistics (2012).ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt. 41 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Annex 2. Air Quality Monitoring Data and Pollution Concentration Extrapolation in Medium-Large Municipalities in Colombia A large number of Colombians potentially exposed below summarize basic size and urban population to air pollution live in cities that lack systematic parameters for cities with and without monitoring measurements of air pollutant concentrations. The networks in Colombia. question for the analyst estimating the health im- pacts of air pollution is whether to drop those cities With regard to the “urban” character of both groups from the analysis or to extrapolate measurements of cities, the question remains whether the propor- from cities that do have monitoring networks. For a tion of the urban population within those unmon- fair extrapolation, cities with and without measure- itored municipalities is comparable to the pro- ments should otherwise be largely comparable. A portion in cities where air pollution is monitored. complete analysis of this comparability is beyond According to Table A.4, the average proportion of the scope of this report, but some basic variables the urban population seems to be lower in cities can shed light on the uncertainties involved. In this without a monitoring network. With regard to the and other CEA-related work, the size and propor- significance of that difference, the dissimilarly tion of the urban population of municipalities are sized, small and non-normally distributed samples used as proxies for exposure to significant concen- preclude the use of a t-statistic contrast, an accept- trations of inhalable particles. The minimum cutoff able alternative being a non-parametric test (e.g., a point of 100,000 inhabitants is explicitly set as a Mann-Whitney U test 16 featured below): proxy for exposure to significant concentrations of particulate matter (PM) in outdoor air. Therefore, it • U1 = 561,0 is expected that the municipalities with monitoring • U1 = n1n2 + 0.5(n1) (n1 + 1) – R1 networks are by definition larger than those without • U2 = 267,0 such networks. Moreover, the force behind the es- • U2 = n1n2 + 0.5(n2) (n2 + 1) – R2 tablishment of air-quality monitoring networks (in Colombia and elsewhere) is precisely the presence Level of significance 5%; Ucrit = 287. Null hypoth- of air-quality problems, in turn commonly linked esis (no differences between the mean of the two to the size of urban settings. Tables A.3 and A.4 samples) is rejected at the set level of significance. Table A.3  Size of Cities With and Without Air Pollution Monitoring Network Coverage in Colombia Avg. municipality Max. municipality Avg. urban Type of municipality Sum population population population population With M.N. 17,943,138 780,136 7,259,597 754,120 Without M.N. 8,617,545 239,376 1,179,098 210,914 Source: Authors’ estimates. 16  Where n1, n2 are the sizes of samples 1 and 2, and R1, R2 are the sums of the ranks for samples 1 and 2. 42 Graph A.1 below represents the samples’ respective access to health services) might show systemat- point estimates of proportion of urban population. ic differences between the two groups. The uncer- tainty in such an extrapolation, along with the rap- Therefore, we cannot rule out the possibility that idly shrinking size of the population not covered by the two groups of cities have a different proportion monitoring networks, are the reasons why we decid- of urban population. Similarly, other health-rele- ed to exclude from the analysis the cities for which vant baseline variables (e.g., population structure, there were no measurement data as of 2012. Proportion of Urban Population Table A.4  Percentage of Urban Population Graph A.1  (Out of Total Population) in in Cities > 100K Inhabitants in Colombian Cities With Over 100,000 Colombia Inhabitants % % Urban % Urban 100 population cities population >100,000 with cities >100,000 80 Parameter M.N. without M.N.* 60 Average 91.12 82.76 40 Minimum 68.76 39.31 Maximum 99.78 99.86 20 SD 7.80 15.48 0 N (S. size) 23 36 Cities with M.N Cities without M.N Source: Authors’ estimates. Source: Authors’ estimates. *The city of Uribia was considered an outlier and excluded from the analysis. 43 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Annex 3. Methodological Differences and Similarities, and Comparative Results with Previous Estimates One objective of this report is to update results total cases considered. Therefore, caution should from a previous study (Larsen 2004) on the eco- be exercised when comparing these percentages nomic cost of environmental degradation in Colom- with previous reported figures. For instance, we bia. As such, and for purposes of comparability, it cannot conclude that total premature mortality at- draws heavily from that study in terms of structure, tributable to urban air pollution in 2009 (a total of references and assumptions. In terms of structure, 5,027 premature deaths/37,703 lost DALYs) rep- it is identical: the steps toward this update are se- resents an improvement over the previous esti- quential, starting with the existing evidence on en- mate of 2004 (6,040 deaths/45,300 lost DALYs). vironmental exposures, re-estimating the mortality In fact, when cities without monitoring networks and morbidity effects attributable to such expo- are included in the analysis, assuming an expo- sure, and valuing their socioeconomic impacts. Be- sure of the average PM2.5 and PM10 concentrations low are the main points of disagreement and, of cities with measurement data, there is a net in- where relevant, continuity. crease in the total reported cases and DALYs (see Table A.5). Most of this increase can be explained Mortality: In previous estimates for the effect of through the increase in the urban population and urban air pollution (Larsen 2004), and because consequently of the population exposed. of the admitted underestimation that registered deaths supposed with regard to the total amount Human Capital Approach: We are not reporting a of deaths in DANE statistics, crude mortality rates cost based on the Human Capital Approach, since were estimated for each city, using a regression we believe that an indicator based on foregone in- equation that included urban population share, come due to premature mortality severely underes- population share above age 50, external death rate timates the true cost to society that excess mortal- and child mortality rate. However, reporting and ity represents in a rapidly improving context such statistical methodologies have improved substan- as that of urban Colombia. However, if we apply tially in the last decade; DANE no longer reports an the Human Capital Approach, then mortality relat- issue with differences between registered and esti- ed health cost of urban air pollution is estimated mated death rates, so data are drawn directly from at about 420 billion COP. Then total annual health its departmental databases without further trans- cost associated with urban air pollution is estimat- formation. ed at about 1.6 billion COP. An annual estimate consistent with the estimation in the Larson, 2002 Exposed population: For purposes of comparison report where for mortality valuation an average be- with previous results (Larsen 2004), it is important tween the value estimated at VSL and the Human to mention once more that we are not taking into Capital Approach value would be at about 3.55 bil- account those mortality and morbidity end-points lion COP annually, or 0.72% of GDP. occurring in cities for which there are no data on urban air-pollutant concentration. In Larsen 2004, Willingness to pay proxy: Although we also translate the population in that category of cities represent- different health effects from environmental risks ed 42% of the exposed population and 35% of the into DALY metrics for purposes of comparability, we 44 Comparison between 2004 Results for Health Effects of Urban Air Pollution and Current Table A.5  Analysis Including All Cities with Over 100,000 Inhabitants in Colombia   2012 Update Including Cities Without M.N. 2004 results Health categories Total cases Total DALYs (000) Total cases Total DALYs (000) Premature mortality 7,147 53,601 6,040 45,300 Chronic bronchitis 6,604 14,528 7,410 16,300 Hospital admissions 14,806 237 12,970 210 ER/outpatient hospital visits 290,451 1,307 255,000 1,150 Restricted activity days 51,081,865 15,325 42,000,000 12,640 LRI in children 583,864 3,795 585,000 3,800 Respiratory symptoms 162,573,589 12,193 135,000,000 10,100 Total N/A 100,986 N/A 89,500 Note: ER: Emergency room; LRI: Lower Respiratory Infections. are discontinuing in this report the approach based the solidity of such a proxy for WTP value. By using on supplementing cost-of-illness (medical costs this WTP/COI ratio instead of the DALY-GDP proxy, and costs of time lost to illness) values with a proxy the cost per case is bound to differ substantially in of willingness to pay (WTP) multiplied by GDP per this study compared to the previous 2004 report. capita to avoid or reduce the risk of illness. There is One especially drastic contrast is the cost per case sound evidence that individuals place a much high- of chronic bronchitis. Although we agree on placing er value on avoiding pain and discomfort associat- a larger cost in a burdensome illness such as CB, ed with illness than that reflected solely in medi- we would rather apply consistent WTP values for cal costs; this has also been observed specifically all air pollution-related outcomes in the absence of in connection with air-pollution risks (Alberini and specific information for CB. Krupnick 2000, Arigoni et al. 2009). However, us- ing DALYs and per capita GDP as components for a Baseline data for cost estimation: We kept most of proxy to lost value is too far-fetched a step. Instead, the assumptions regarding duration of illness, rate we apply the WTP proxy used in the Bogotá ten-year and length of hospitalization, average time lost per air-quality management plan (PDDB 2009), which health end-point, frequency of doctor visits and dis- in turn is based on Inter-American Development count rate. We also valued time lost to illness at Bank (IADB) estimates (Cifuentes et al. 2005) and 75% of the average urban wage, and applied this a local study (Lozano 2004) which used benefit cost both to working and nonworking individuals, transfer to calculate the WTP value. For morbidi- based on the assumption of an equivalent oppor- ty outcomes, the WTP/COI ratio is 2.27, which is tunity cost for both categories. a figure remarkably similar to the upper estimate reported by Alberini and Krupnick (2000) between In the context of updating the results from the Lars- 1.61 and 2.26 times, depending on pollution lev- en (2004) study, the differences between the es- els. The latter also note that such ratios are simi- timated costs in 2002 and those in 2010 might lar to those for the United States, despite the differ- seem surprising. It is certainly counterintuitive to ences between countries, which further reinforces obtain unit costs of healthcare services that are 45 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 lower a decade later, currency value shocks exclud- controlled pricing) of healthcare delivery in Bo- ed. After the passage of the healthcare reform law gotá are likely larger than in the rest of the coun- (Law 100 of 1993), healthcare financing in Colom- try’s urban settings. Real urban wages in Bogotá bia relies in essence on a public insurance scheme are on average 29% higher than the mean wage funded mainly through social security contribu- of all urban areas of Colombia combined (DANE- tions, taxes and subnational transfers. Virtually ev- GEIH 2011), which suggests a lower value of time eryone in Colombia receives health services in one lost to illness at a national level. However, whereas of three possible ways: (i) Régimen contributivo, a the private medical salaries in Bogotá might reflect mandatory healthcare insurance linked to and paid a similar difference with the rest of urban areas, through deductions from salaries of those who are other inputs to the cost of medical care would not formally employed. Coverage provided is compre- necessarily follow the same patterns. The average hensive; (ii) Régimen subsidiado, a less-compre- cost of hospitalization resulting from the PDDB es- hensive insurance scheme for those whose pay- timates would be around 258,000 COP per day. We ment capacity is lower; and (iii) Those without any found that the values used in the PDDB were quite payment capacity whose healthcare is provided al- close to the upper bound of the public prices (SOAT most exclusively by the network of public hospitals 2009) of the most common treatment and health- and healthcare facilities, and paid for by depart- care usage scenarios for health outcomes attribut- mental governments. As of 2010, the percentage able to air pollution. The result of applying Larsen’s of the population covered by publicly paid health in- (2004) assumptions to these values is summa- surance was around 93% (Ministerio de la protec- rized in Table A.6 below: ción social—MIPS 2011). Cost of chronic bronchitis (CB): There is still lit- Public and private healthcare providers compete in tle evidence of healthcare usage and costs of pa- the provision of services, most of which are funded tients regarding the cost of a new case of chron- by public entities that act effectively as insurance ic bronchitis. Moreover, estimates of incidence providers themselves. Private voluntary insurance are rare in general. In Colombia, a recent study represents only about 16% of healthcare payment in Colombia, largely concentrated in higher-income Cost of Selected Health Services Table A.6  groups (Baron 2007). Unsurprisingly, reliable esti- and Time Lost to Air Pollution- mates of private healthcare costs are hard to ob- Related Health Outcomes tain and difficult to contrast. Private healthcare providers, both independent professionals and en- 2002 values tities, represent about 60% of healthcare expen- (Larsen Update 2004) 2012 diture, with services largely paid for by these pub- Cost of hospitalization 280,000 259,601 lic insurance entities at fixed rates (Gideion et al. (pesos per day) 2010). Therefore, the publicly listed prices that Cost of emergency visit 90,000 71,635 public insurers pay providers are considered an ad- (pesos): urban equate reflection of the true cost of illness. Cost of doctor visit 40,000 28,654 (pesos): urban Another approach considered was to use Bo- Value of time lost to 20,000 31,320 illness (pesos per day) gotá costs (PDDB) for whole of urban Colombia. However, true costs (i.e., unsubsidized or without Source: Authors’ estimates. 46 (PREPOCOL 2009) estimates the prevalence of PM2.5 to PM10, with the former being the pollutant clinical COPD at 3.2% in adults over age 40, but we due to which the health damages occur—is between could not find data either on incidence or regarding 40 and 60%, and the midpoint of 50% is used in usage and cost of healthcare by patients. There- this calculation. It is estimated that 200 addition- fore, we relied on the same information used be- al mortality cases would have occurred in Bogotá fore—by Schulman et al. (2001) and Niederman et had the pollution level remained unchanged com- al. (1999)—from the United States and Europe and pared with the 2002 level; and 440 fewer mortali- applied it to Colombia. The estimate of lost work- ty cases would have occurred, ceteris paribus, with days per year is based on the frequency of estimat- a concentration level on average equal to 48 µg/m3 ed medical treatment plus seven additional days (i.e., the level measured in 2012). Thus, the set of for each hospitalization and one extra day for each policy measures, including the introduction and en- doctor and emergency visit. These days are add- forcement of more stringent fuel quality standards ed to reflect the time needed for recovery from ill- in Bogotá and other measures, and which led to the ness. The estimated cost of a new case of CB as- lower measured levels of PM10 in Bogotá, resulted sumes a 20-year duration of illness over which in a reduction of mortality cases in Bogotá by 7% medical costs and value of time experience an an- in 2010 relative to what they would have been had nual real increase of 2%, and costs are discounted air quality not improved. A further improvement in at a three-percent rate per year, a value commonly air quality in 2012 resulted in a reduction of mor- applied by WHO for health effects. tality cases in Bogotá by 16% compared with the baseline scenario. The percentage savings in costs Estimates of cost savings due to changes in pollu- due to a reduction in mortality and morbidity are ap- tion levels in Bogotá: The average PM10 concentra- proximately the same. If one compared the annual tion in Bogotá declined from 66 µg/m3 in 2002 to health costs and mortality cases avoided in Scenar- 59 in 2010 and then further to an average of 48 in io 2 compared with Scenario 1, one could interpret 2012. For decision-making purposes, it is import- the findings as follows: there has been a reduction ant to assess the health benefits that this decline of annual mortality and morbidity costs (and a re- in pollution levels in Bogotá brought about. To as- duction of mortality cases) by 23% in Bogotá alone. sess the benefits, sensitivity analysis is carried out If the costs of the policy interventions and invest- that shows the changes in mortality and morbidi- ments were available, it would be possible to car- ty and the associated costs in Bogotá in two cas- ry out cost-benefit analysis of the interventions car- es: a high concentration scenario with the aver- ried out throughout that period; and other cities and age levels of PM10 measured in Bogotá in 2002, metropolitan areas can also be included in the sce- and a low concentration scenario with the levels nario analysis with some additional data. measured in 2012. It is assumed that the ratio of 47 Environmental Health Costs in Colombia – The Changes from 2002 to 2010 Bogotá: Mortality and Morbidity Costs in the Baseline 2010 Case and in High and Low Table A.7  Pollution Scenarios Scenario 1: Scenario 2: 2010 baseline High concentration Low concentration case 1/ 2/ Air pollution, PM10 (average annual 59 66 48 concentration, µg/m3) PM2.5 (estimated annual average 24–36 27–40 20–29 concentration, µg/m3) Estimated mortality range due to UAP in 2,580–3,000 2,830–3,150 2,060–2,640 Bogotá (number of cases) Midpoint estimate of mortality due to UAP in 2,790 2,990 2,350 Bogotá (number of cases) Mortality increase (reduction) over baseline 200 (440) (number of cases) Cost of mortality and morbidity (billion COP) 3/ 3,515 3,767 2,961 Increase (reduction) of morbidity and mortality 252 (554) costs over baseline (billion COP) Increase (reduction) of morbidity and mortality 7% –16% costs over baseline (%) Source: Authors’ estimates. Notes: 1/ Scenario 1 uses the average PM10 concentration of 66 µg/m3 observed in Bogotá in 2002. 2/ Scenario 2 uses the average PM10 concentration of 48 µg/m3 observed in Bogotá in 2012. 3/ Morbidity costs are estimated as 26% of mortality costs, based on the estimates in this study. 48 Publications from the LCSEN Environment & Water Resources Occasional Paper Series ■■ Framework for Conducting Benefit-Cost Analyses of Investments in Hydro-Meteorological Systems (2014) ■■ Developing Offsets for Loss of Biodiversity: Experience from Colombia (2014) ■■ Metropolitan Pollution Management in Brazil: Taking Stock (2014) ■■ Developing a Program for Contaminated Site Management in Low and Middle Income Countries (2014) ■■ Restoring the Coastal Environment in Cartagena, Colombia (2014) (Available in English and Spanish) ■■ Climate Change Impacts on Water Resources and Adaptation in the Rural Water Supply and Sanitation Sector in Nicaragua (2013) (Available in English and Spanish) ■■ Climate Change Impacts on Water Resources Management: Adaptation Challenges and Opportunities in Northeast Brazil (2013) ■■ El Futuro del Riego en el Perú: Desafíos y Recomendaciones (Volumen I: Informe de Síntesis y Volu- men II : Informe Principal) (2013) ■■ Empowering Women in Irrigation Management: The Sierra in Peru (2012) ■■ Environmental Health in Nicaragua: Addressing Key Environmental Challenges (Originally Published in 2010, Republished in 2012) (Available in English and Spanish) ■■ Expanding Financing for Biodiversity Conservation: Experiences from Latin America and the Caribbean (2012) (Available in English and Spanish) ■■ Overcoming Institutional and Governance Challenges in Environmental Management. Case Studies from Latin America and the Caribbean Region (2012) ■■ Policy and Investment Priorities to Reduce Environmental Degradation of the Lake Nicaragua Water- shed (Cocibolca) (Originally Published in 2010, Republished in 2012) (Available in English and Span- ish) ■■ Uncertain Future, Robust Decisions; The Case of Climate Change Adaptation in Campeche, Mexico (2012) To find copies of these publications, please visit our web site: http://www.worldbank.org/LAC/LCSENSeries 49 LATIN AMERICA & CARIBBEAN REGION Environment & Water Resources OCCASIONAL PAPER SERIES