The WHO country profiles on environmental burden of disease (Figure 1) are composed of three parts. Part 1 (a) provides "exposure-based" estimates for three risk factors, i.e. based on globally available country exposures, and for which quantitative methods for disease burden estimation have been published [4]. These three risk factors include 'unsafe water, sanitation and hygiene', 'indoor air pollution from solid fuel use' and 'outdoor air pollution'. Part 1 also includes main malaria and other vectors that are present in the country and cause certain health risks. Part 2 (b) is a preliminary estimate of the total environmental burden of disease for the country, based on a review of the evidence completed by expert opinion. Part 3 (c) presents a breakdown by disease group for the estimate provided in Part 2.
In addition, the country profiles list a number of socio-economic parameters such as the GNI (Gross National Income), population, population living in poverty, urbanization and others in order to better understand the country's situation. Also, selected environmental exposures are presented to provide additional information regarding prevention opportunities [11]. These additional exposures were selected because they are both globally available and relevant for quantification of various health outcomes: lead (e.g. associated with cognitive impairment), malnutrition (associated with most infectious diseases, and water, sanitation and hygiene) and crowding (e.g. associated with the transmission of tuberculosis). The impact of environment-related malnutrition has already been incorporated in the EBD estimates (see [12] for additional details relating to water, sanitation and hygiene in particular). No comprehensive database was, however, available for blood lead levels, and no methods for estimating the disease burden due to crowding had been developed at the time of estimation. Such country-level exposures can be used for refined estimates at country level.
Both exposure-based estimates (Part 1) and evidence reviews completed by expert opinion-based methods (Part 2 and 3) use population attributable fractions, but differ in its estimation method. To obtain the environmental burden of disease, these attributable fractions are then multiplied by the total burden of disease (in deaths or DALYs) of the relevant disease (see [13] or Volume 1 of [4] for additional information on the Global Burden of Disease concept). The population attributable fraction is defined as the proportional reduction in disease or death that would occur if exposure to the risk factor were reduced to zero (see Chapter 1 of [14] or Volume 1 of [4] for references and explanations).
Attributable disease burden as estimated here – based on attributable fractions – is in principle not equivalent to preventable burden. Rather, preventability also depends on the technical, social, economic, psychological and ethical dimensions of a situation [15]. This being said, such burden points to the impact of causes, and the potential for prevention under certain circumstances. To some extent, the preventability of the attributable burden is also influenced by its method of estimation (in particular the counterfactual, or reference value, against which the impact of current exposure burden was estimated, or in the case of expert evaluation the formulation of the question; see sections (a) and (b) below).
While the results are expressed in terms of total deaths or a summary measure of population health, more detailed information is sometimes available (such as by age group or region, provided the underlying epidemiological information was specific enough).
The following sections provide additional details on the methods underlying the Parts 1 to 3 of the country profiles.
(a) Exposure-based estimates (Part 1)
The exposure to the three risks 'unsafe water, sanitation & hygiene', 'indoor air pollution from solid fuel use' and 'outdoor air pollution' has been assessed or estimated globally, and the databases with country exposures are publicly accessible [16–18]. These data are mainly based on household surveys for water, sanitation and solid fuel use, and measurements for outdoor air pollution. Missing data were modeled. For example, the proportion of households using solid fuels was available for 93 of 181 countries; modelled for 36 countries based on proportion of rural population and gross national income (GNI); and assumed as less than 5% of households using solid fuels for the 52 higher income countries with GNI above US$ 10,500. The diseases taken into account include the following:
• diarrhoea for water, sanitation and hygiene;
• acute lower respiratory infections (in children under 5 years), chronic obstructive pulmonary disease and lung cancer (in adults above 30 years) for solid fuel use; and
• respiratory mortality (in children under 5 years), cardiopulmonary mortality and lung cancer (in adults above 30 years) for outdoor air pollution.
The disease statistics used for this analysis have been compiled by the World Health Organization, by country, gender and age group [19]. The methods used to combine exposure data and disease statistics into attributable disease burden have been published in the environmental burden of disease series [20–22], based on previous global analyses [1, 23–25]. Detailed information on the input data and methods can be found in the respective sources, which are all publicly accessible.
The method for obtaining exposure-based estimates can (for the risks studied here) be summarized as combining exposure with risk measures to obtain a population attributable fraction (simplified e.g. from [26, 27]):
Where: i is the exposure category, Pi is the proportion of the population in exposure category i, and RRi is the relative risk at exposure category i compared with the reference level.
The disease burden from water, sanitation and hygiene and indoor air pollution reflects the total attributable burden, in other words the burden that could be avoided if water, sanitation and hygiene, and solid fuel use could be improved to the point of not causing any health impact. The estimates for outdoor air pollution however reflect the disease burden that could be prevented if pollution levels were reduced to WHO guideline values for particulate matter, although it is acknowledged that adverse health effects occur even below this value [28].
The exposure-based assessment has been carried out for only these three risk factors because they were the only ones with both a method for quantified estimation of health impacts and global databases of exposure assessment available. While for some additional risk factors such global quantification may become possible with development of additional methods (e.g. for second-hand smoke, crowding), it is more difficult for risk factors for which population exposure is more cumbersome to assess (such as for ionizing radiation). We therefore proceeded to complete exposure-based assessments with literature review/expert-based estimates (Part 2 and 3).
Results are expressed in premature deaths (as compared to standard life expectancy) and in DALYs (Disability-Adjusted Life Years), a combination of death and disability (Additional file 1 contains mainly deaths, whereas the full country profiles contain both deaths and DALYs for each risk factor). DALYs for a disease are the sum of the years of life lost due to premature mortality (YLL) in the population and the years lost due to disability (YLD) for incident cases of the health condition [29]. DALYs therefore reflect the age of death, which means that deaths occurring mostly at more advanced age produce fewer DALYs than deaths occurring primarily in childhood.
Exposure-based estimates of disease burden should, in principle, not be added up across risk factors, as disease could be prevented by acting on different risks. For example, diarrhoea deaths could be prevented by improving nutritional status, water, sanitation and hygiene, or food safety. To avoid double-counting, one practical approach consists in avoiding to sum up disease burden within one disease category altogether, or otherwise to carefully consider the potential for double counting in the specific cases.
(b) Preliminary estimate of the total environmental burden of disease for the country (Part 2)
The data presented in the second part of the country profiles represent the disease burden that could be avoided by modifying the environment as a whole. The "modifiable environment" includes pollution of air, water and soil; radiations; noise; occupational risks; the built environment, including housing and road design; land use patterns; agricultural methods and irrigation schemes; man-made changes to the climate and ecosystems, and behaviour related to the environment (such as hand-washing or personal protection). Excluded from the definition are individual choices, such as alcohol and tobacco consumption, drug abuse, diet; natural environments or ecosystems that cannot reasonably be modified (rivers, etc); unemployment (provided that it is not linked to the degradation of the environment); natural biological agents (e.g. pollen); person-to-person transmission that cannot reasonably be prevented by environmental interventions.
The total environmental burden has been estimated on a disease-by-disease approach, rather than by risk factor. For 102 diseases or injuries, the literature has been systematically reviewed in terms of attribution to or preventability by environmental improvements, and completed by a survey of over 100 experts worldwide [2]. In the survey, experts were asked to focus on the modifiable part of the environment, hereby indicating the disease burden that could potentially be prevented (or shifted to health risks that lie outside the environmental area). Experts were selected on the basis of their international expertise in the area of each disease or risk factor of concern. They provided a best estimate and a "confidence" interval. These probability distributions were pooled, giving equal weighting to each distribution (i.e. to each expert reply), to obtain a combined probability distribution for the attributable fraction. Additional details on methods and results of this review have been published previously [2, 30]. The resulting attributable fractions, specified by region/age group where applicable, have been combined with WHO disease statistics by country, gender and age group[19] to result into the estimate of the environmental burden of disease by country. The attributable environmental fractions used here are regional fractions, and not national as is the case for the indicators used in the first part of the country profile. The approach by disease rather than by risk factor avoids overlaps or double-counting.
Results are expressed in total number of deaths and DALYs per capita, and percentage of the national burden of disease attributable to the environment.
(c) Environmental burden by disease category (Part 3)
Part 3 of the profile is a breakdown by disease group of the information given in Part 2, applying the same methods. It indicates the country's yearly number of DALYs per capita attributable to environmental factors by disease group.
For comparison, the world's lowest and highest country rates are provided for each disease group. The bar chart provides an indication of the country's situation for a particular disease group in comparison to other countries: A full bar indicates that the country's environmental DALYs per capita for the particular disease category is close to the highest rate encountered in any country, and a small bar indicates a rate close to the lowest encountered.