Environmental Epidemiology: The Context
Many chronic diseases are of unknown etiology but may be related to environmental exposures. This report, the second of the Committee on Environmental Epidemiology of the National Research Council's Commission on Life Sciences, considers what information is needed to determine the prevalence of these diseases and whether they are a result of environmental exposures. The report highlights research opportunities and methodologic advances that will improve the scientific base of the evolving field of environmental epidemiology and examines the contribution of the so-called gray literature to our knowledge about possible links between environmental exposures and chronic diseases. The gray literature is defined as literature that is not ''white" (available and cataloged), and that is not "black" (not available, unknown, or not obtainable). Gray-literature reports are usually produced in small quantities, intended for limited audiences, and not widely known; they are not indexed in standard sources, such as Medline. In the field of environmental epidemiology, these gray-literature studies may include such items as state health-department reports and doctoral and master's theses. One of the committee's criteria for determining the quality of the information in papers in the gray literature was whether the papers would meet currently accepted peer-review standards for publication in journals. The twin focus on the gray literature and epidemiologic methods fulfills the charge to the committee for its second report to examine state health-department reports and similar analyses that are not generally available in the peer-reviewed literature. As undertaken by the committee, the examination of the gray literature involves developing criteria for
requesting a few state health departments to provide reports for examination, developing criteria that help to define adequate epidemiologic studies, applying those criteria to selected state studies, and assessing those studies.
Most people in the developed and developing worlds endorse environmental protection (Dunlap, 1992). Public concern about the environment and its relation to human health is demonstrated by the public reaction to reports of contamination at places like Love Canal, Times Beach, and Stringfellow Pits in the United States.
The nature of environmental-health concerns differs considerably between the developed and developing worlds (Doll, 1992). In the former, cigarette-smoking constitutes a major cause of illness and death, and occupational hazards, environmental tobacco smoke, lead, and other air pollutants adversely affect public health. Levels of lead thought to be safe a decade ago are now believed to produce permanent damage to children's intellectual potential (ATSDR, 1990).
In the developing world, concern focuses on basic sanitation, pure air, and clean water. The problems are traceable to a large extent to infectious agents, but exposure to toxic substances plays a role. The World Bank estimates that 1 billion people lack safe water, 1.7 billion are without adequate sanitation, 1.3 billion are exposed to unsafe soot and smoke, and 700 million women and children are exposed to severe air pollution from cooking fires (World Bank, 1992).
The incomplete understanding of causes of many common chronic diseases in both developed and developing countries fuels interest in identifying avoidable environmental hazards. Thus, more than 60% of all cases of birth defects are of unknown or poorly understood etiology (NRC, 1989a), as are many cases of degenerative neurologic diseases (NRC, 1992a), adult-onset asthma (NRC, 1992b) and other chronic respiratory diseases (NRC, 1989b), and renal and hepatic diseases. With respect to reproductive health generally, an array of end points are of concern, ranging from effects on offspring to reproductive health in males and females, including sexual maturation, onset of menses, menopause, sexual functioning, and endometriosis. Although these events are often discussed in an atmosphere of high public concern, suspect environmental factors must be studied with strict adherence to scientific canons of independent, verifiable research.
Although genetic predisposition and poor nutrition are important risk factors for many chronic diseases and other adverse health outcomes, the possible contribution of preventable or controllable environmental fac-
tors to these health outcomes needs to be clarified. Etiologic patterns in time and place of many diseases often cannot be determined, because appropriate data, methods, and theoretical frameworks have not been developed to evaluate such patterns. Thus, a top priority for environmental-health research is to identify better ways to look for environmental factors that may contribute to disease.
Definition of Environmental Epidemiology
Modern epidemiology, the study of disease patterns in populations, encompasses a broad array of subject matter, including subspecialties that concentrate on such domains as clinical trials of pharmaceutical agents; such outcomes as reproductive and developmental effects, infectious diseases, and chronic diseases; such risk factors as occupation, nutrition, and alcoholism; and special populations. Thus, epidemiology includes controlled clinical evaluations of different treatment methods; comparative assessment of lifestyle factors, such as smoking, drugs, and drinking habits; estimations of the risks of occupational factors; and cross-sectional and time-series analyses of factors that may affect health.
Whatever the subject of epidemiologic evaluation, the basic theoretical and general principles are concerned with evaluating the statistical and biologic importance of variations in the frequency of occurrence of illnesses and related phenomena of health and health care. Epidemiologic study involves examination of the extent to which observed rates of a given phenomenon differ significantly from those that would be expected under specified conditions (Miettinen, 1985).
Interest in the application of epidemiology to the study of environmental hazards is increasing because epidemiologic studies can validate the models used in predicting hazards and can characterize the actual and potential health effects of such exposures.
The same definition of environmental epidemiology is used here as in the first volume (NRC, 1991), that is, the study of the effect on human health of physical, biologic, and chemical factors in the external environment. By examining specific populations or communities exposed to different ambient environments, environmental epidemiology seeks to clarify the relation between physical, biologic, and chemical factors and human health.
Purview of This Report
As in the first report, this volume excludes or limits the attention it gives to some environmental-epidemiology topics that have been the subject of other National Research Council committees. Thus, little attention
is given to issues related to toxic air pollutants and pesticide exposure of children. We do not deal with voluntary behavioral exposures, such as smoking, or with subjects, such as radiation, that have been dealt with in recent NRC reports.
This report considers the needs for research in environmental epidemiology in general, extends the assessment of information on the health effects of exposures from hazardous wastes started in the first volume up to the middle 1990s, and includes selected reports in the gray literature. Most of the gray literature examined here comes from state-generated studies or, if they are available to the public, analyses conducted by researchers for use in legal proceedings. Some of these contain critical information on subjects of interest to the committee. Also, as part of our review of the gray literature, we consider recent reports from the World Bank and the Pan American Health Organization regarding the health consequences of environmental contamination in central Europe, Central America, and China. The committee recognized its obligation to subject such studies to its own peer-review. Thus, at least two committee members, expert in the relevant field of a study but not authors of the study, examined each of the reports from the gray literature that are cited in this report.
The decision to include gray literature arises because the committee found in preparing its first report that reports from a substantial number of relevant studies had not appeared in the conventional peer-reviewed literature. As discussed in more detail in chapter 7 of this volume, the committee solicited certain key reports from selected states on the epidemiologic study of hazardous-waste exposures. The committee then reviewed these to see how well they met its criteria for acceptable gray literature. The criteria are that a report be about an epidemiologic study involving a community or residential population, that it have some sort of peer-review or other evidence of quality control, that the authors tried to collect exposure data, and that it have been published since 1980.
The first report of the committee considered findings in the published literature about hazardous waste in relation to the legislative mandates of the Environmental Protection Agency (EPA) and the Agency for Toxic Substances Disease Registry (ATSDR). The committee concluded that, as a practical consequence of these mandates, EPA and ATSDR tend to focus chiefly on the general estimation of exposure from and management of hazardous-waste sites and only indirectly consider potential public health implications of exposures from such sites. As this committee and the US General Accounting Office (GAO, 1991) have observed, this is because inadequate resources have been devoted to characterizing human health risks possibly associated with exposure to hazardous wastes. Of more than $4 billion spent annually on the Superfund Program, less than
1% has been applied to the development of an active public-health program. In particular, exposures potentially critical to public health have been sparsely documented.
Structure of This Report
Chapter 2 reviews basic epidemiologic methods that can be applied to environmental problems and comments on some well-recognized problems that environmental epidemiology faces, including small numbers of persons exposed, agents that have not been well characterized, and concern with small increases in risk. It states principles for inferring causation and discusses the types of evidence needed for environmental-epidemiology studies.
Chapter 3 expands on the needs for improving exposure-assessment information for environmental epidemiology and discusses these vis-à-vis specific epidemiological-study designs.
Chapter 4 identifies health outcomes that should be subjected to environmental-epidemiologic study, ranging from chronic diseases of poorly understood etiology to those for which some causes are known. It identifies research opportunities for using biologic markers to study environmental factors that may be relevant to several chronic diseases, as well as for improving exposure-assessment information.
Chapter 5 considers existing data systems that are relevant to the research needs of environmental epidemiology.
Chapter 6 discusses several areas where improvements in methodology will advance the field of environmental epidemiology.
Chapter 7 presents a summary of selected gray-literature reports on hazardous wastes. It reviews several state studies of reproductive outcomes to illustrate the constraints on state health departments in carrying out environmental-epidemiologic research.
Chapter 8 recapitulates the major conclusions and recommendations of the committee.
Special Issues for the Study of Environmental Epidemiology
As indicated in the first report, an optimal analysis of potential public-health consequences of suspect exposures proceeds from an assessment of past, current, or future exposures to the formulation of testable hypotheses of effects to be studied in one or more specific populations. An ideal environmental-epidemiology assessment considers all possible adverse health effects in exposed and unexposed persons and includes relevant contributing factors, including those that could confound, or interfere
with, the interpretation of results. Few studies meet this ideal; this limitation is also common in peer-reviewed, published reports.
The committee relied on a combination of evidence from different sources to assess the impact on public health of exposures suspected of causing symptoms or disease. The types of information from these sources are discussed at length in chapter 2 of volume 1.
Small Relative Risks, but Large Numbers of Cases
Increasingly, environmental epidemiology concerns the search for factors that might moderately affect the risk of common multifactorial diseases. The effect of an individual environmental exposure on the relative risk of a disease may be small, but this does not mean that it is inconsequential; it can affect very large numbers of people and thus be associated with large numbers of cases of disease. For example, the risk of death in males aged 45-74 years with a diastolic blood pressure of 95 mm Hg in the Framingham study was only about 1.15 times the risk in those with a diastolic blood pressure of 85 mm Hg, yet the amount of disease that could be prevented in the population by reducing diastolic pressures to 85 mm Hg would be substantial. Increased use of hypertension medication, along with improvements in diet and exercise, is thought to be responsible for some part of the substantial decline in cardiovascular mortality in the last 20 years.
Large sample sizes and long-term followup studies are generally necessary to demonstrate potentially serious effects that involve small increases in relative risk. The chronic effects of ozone exposure and the acute effects of particulate air pollution are instances in which the relative risk may be small, but the population disease burden may be substantial. Ozone levels are often excessively high in many urban and coastal areas in the summer, when millions of people are outdoors, so that even a 10% increase in relative risk will produce a large number of cases of disease. For further discussion on this point, readers are referred to chapter 4.
The Need for More-General Monitoring
Monitoring systems utilizing existing data sources, as we discuss in chapter 5, will be of increasing importance. The first report also noted that monitoring may be the only way to determine the extent to which disease rates have changed as a result of changes in environmental contamination.
The use of aggregate statistics is also critical to detect trends or patterns in environmental pollution and health consequences that are not apparent at a local level. For example, what was initially thought to be laboratory drift in the measurement of blood lead concentrations in the
Second National Health and Nutrition Examination Survey was later identified as a 37% decrease in the population mean blood lead level as a result of a decrease in the use of lead in gasoline (Annest et al., 1983). This 37% decrease nationwide was used to justify the elimination of lead from gasoline.
Detecting the effects of monitored changes in disease and exposure also often depends on the alertness of researchers. The 2.5-fold increase in mortality in the London smog episode in 1952 became apparent when an investigator compiled weekly mortality data. Similarly, hospital admissions for asthma in children were cut in half in a Utah valley when a steel mill closed down, and admissions returned to their previous level when it reopened; this was also not detected by clinicians or the state health department but required the examination of hospitalization data by an investigator (Pope, 1991).
Developing Relevant Exposure Gradients
Exposures to synthetic organic chemicals and other modern products cannot be accurately segregated by source, such as air, water, or soil. Rather, modern exposure scenarios often model multimedia, multi-temporal levels of many complex chemical compounds. Further, physical and biologic characteristics of other environmental factors can influence uptake and total dose of chemicals. Thus, heat, meteorologic conditions, humidity, and particle size affect the extent of uptake of airborne contaminants, and water hardness, pH, acidity, volatility of contaminants, and other natural background factors affect exposure to materials in water.
Much work in the past has relied on assumed dichotomous, yes-no exposures, but it is not always possible to find and study populations that are entirely unexposed to some environmental agents. Epidemiologists must work closely with exposure analysts to generate meaningful gradients of exposures for such populations, including the use of models to improve exposure estimates. These models need to include environmental and biologic fate, population activity patterns, biomonitoring, and biomarkers. Wherever possible, models should be validated by monitoring carefully selected subsamples of the population under study. This will allow more-refined estimates of individual exposure to be used in population-based studies, as is discussed in chapters 3 and 6.
The Role of Public-Health Departments in Environmental-Epidemiology Research
Many issues in environmental epidemiology are in the domain of departments of public health. A group of residents near a hazardous-waste
site, for example, may become concerned about odors from the site, seepage into their yards or basements, or various symptoms that they may attribute, in the absence of other information, to chemicals from the site. These concerns are likely to come to the notice of the local public-health department, usually with requests for reassurance in the form of a study. Thus, public-health departments may be asked to study diseases of unknown etiology where there may be insufficient evidence to incriminate the hazardous-waste site or an alternate source and for which there has often been insufficient time since initial exposures (if indeed the exposures were from a waste site) for the presumed latent period to be exceeded. The public tends to have unrealistic expectations of what an epidemiologic study can produce.
Unfortunately, most public-health departments are ill prepared to conduct epidemiologic research (Ozonoff and Boden, 1987), or other factors impede research:
- The staff of the public-health department may have little training in environmental epidemiology, environmental toxicology, or exposure assessment. Public concerns notwithstanding, most public-health practitioners will have been trained to cope with other important matters, such as the study of infectious-disease outbreaks, immunization, improvement of maternal and child health, or even the prevention of cancer, but these rarely have any direct relevance to the environment. These deficiencies in personnel and training can foster simplistic approaches to potential environmental-health problems that inappropriately apply the basic concepts of infectious-disease epidemiology rather than those appropriate to chronic disease.
- Even if outside specialists are found to perform an appropriate investigation, the resources available to such departments for investigation are usually insufficient. This may result in a limited investigation that is inconclusive, with the suspicion that, if a more thorough study had been conducted, a more definitive answer might have been achieved.
- Resources may also be lacking for adequate measurement of exposure. Regulators often do not take measurements for the primary purpose of assessing human exposure, but for other purposes, such as to determine compliance with an administrative requirement. A result is that the measurements, good for their intended purpose, may be inappropriate as a basis of studies of human health effects.
- Political and social considerations can impede the conduct of research. One group seeks assurance that no problem exists, while another seeks validation of its health concerns. Or the source of a possibly hazardous pollutant is an economic mainstay of a community.
- Even though it may be clear, as in a chemical spill from a railroad car in a populated area, that the public has been exposed to a possible
- hazard, proprietary or "trade-secret" concerns may block access to information that is needed for prompt and appropriate remedial action.
- Finally, as emphasized in the first report, many exposed populations are so small, the period of observation so short, the exposure so poorly measured, or the outcomes so poorly defined that a verdict of "not proved" is all that can be attained. The situation is usually poorly suited for the conduct of research regardless of the efforts that are expended and the skills of the investigators.
Nevertheless, given the great and increasing variety of chemical, physical, and biologic pollution in the environment, the first indication of a hazard from a particular chemical or group of chemicals may still follow an investigation of some event, or cluster, by a public-health department or concerned citizens. Reports from state and local public-health groups may then enter the gray literature and in some cases be the starting point for research that is published in peer-reviewed journals. These locally initiated studies, although using small populations with poorly characterized exposures, may suggest an effect. Efforts to develop databases of such studies might serve 2 purposes: other parties interested in similar exposures could learn what studies are in progress, so as to increase sharing of information on study design; and appropriate analysis of the combined study findings might eventually become feasible, as discussed further in chapter 7.
The focus of this report is on environmental-health issues in the developed world, but it is recognized that in developing countries environment-related diseases occur along with the more predominant infectious and chronic diseases. Severe indoor particulate air pollution has been documented in the developing world at levels 100 times higher than the US standard of 150 µg/m3 for particles less than 10 µm in diameter (PM,10), and outdoor pollution is also sometimes extreme. These high levels of air pollution, coupled with other disadvantages in the developing world, may account for the fact that acute respiratory disease is the second leading cause of death in children under 5 years of age in countries of the developing world (Leowski, 1986). Important risk factors for both infectious and chronic diseases include basic sanitation, living conditions and urban infrastructure, housing, air and water pollution, and working conditions. All of these are threatened by the sanitary burden that is being accumulated in developing countries. These are the areas in which hygienists have had a great impact in the past in developed countries.
For many chronic, degenerative diseases of potential interest in environmental epidemiology, data on rates of occurrence (incidence) in de-
fined populations are not routinely collected. In addition, exposure is rarely assessed in a manner compatible with the needs of epidemiologic investigation. Thus, it is often impossible to determine whether the incidence of a particular disease has changed in response to a new or changing environmental exposure. Where a gradient of exposure can be determined, the risk of disease can sometimes be related to dose in a specially designed study. However, in many instances, diseases of possible environmental etiology cannot be examined in relation to environmental factors until baseline disease incidences have been determined and appropriate measures or estimates of exposure have been developed.
Study of the health outcomes associated with environmental exposures suffers from a lack of sophisticated technology for assessing chronic effects, from basic methodologic limits of study designs, and from the highly charged climate in which such studies are at times conducted. Moreover, cross-sectional study designs, rather than the more-reliable cohort or case-control studies, are often required in environmental epidemiology. Until quite recently, federal and state support has focused on the need for rapid health assessments that do not necessarily comply with the requirements for environmental epidemiology. Few academic departments of epidemiology have concentrated on refinements in the techniques needed to improve environmental epidemiology, and those that do so must struggle with limited resources. The remainder of this volume outlines strategies that can remedy these problems.
Annest, J.L., J.L. Pirkle, D. Makuc, J.W. Neese, D.D. Bayse, and M.G. Kovar. 1983. Chronological trend in blood lead levels between 1976 and 1980. N. Engl. J. Med. 308:1373-1377.
ATSDR (Agency for Toxic Substances and Disease Registry). 1990. Toxicological Profile for Lead. ATSDR/TP-88/17. Atlanta, Ga.: Agency for Toxic Substances and Disease Registry, Public Health Service, US Department of Health and Human Services.
Doll, R. 1992. Health and environment in the 1990s. Am. J. Pub. Health 82:933-940.
Dunlap, R.E., G.H. Gallup, Jr., and A.M. Gallup. 1992. The Health of the Planet Survey. Princeton, NJ: The George H. Gallup International Institute.
GAO (US General Accounting Office). 1991. Superfund: Public Health Assessments Incomplete and of Questionable Value. Report to the Chairman, Subcommittee on Oversight and Investigations, Committee on Energy and Commerce, House of Representatives. GAO/RCED-91-178.
Leowski, J. 1986. Mortality from acute respiratory infections in children under 5 years of age: global estimates. World Health Stat. Q. 39:138-144.
Miettinen, O.S. 1985. Theoretical Epidemiology: Principles of Occurrence Research in Medicine. New York: John Wiley & Sons. 359 pp.
NRC (National Research Council). 1989a. Biologic Markers in Reproductive Toxicology. Washington, DC: National Academy Press. 395 pp.
NRC (National Research Council). 1989b. Biologic Markers in Pulmonary Toxicology. Washington, DC: National Academy Press. 179 pp.
NRC (National Research Council). 1991. Environmental Epidemiology. Public Health and Hazardous Wastes. Washington, DC: National Academy Press. 282 pp.
NRC (National Research Council). 1992a. Environmental Neurotoxicology. Washington, DC: National Academy Press. 154 pp.
NRC (National Research Council). 1992b. Multiple Chemical Sensitivities: Addendum to Biologic Markers in Immunotoxicology. Washington, DC: National Academy Press. 202 pp.
Ozonoff, Boden, 1987. Truth and consequences: health department responses to environmental problems. Science, Technology and Human Values, 12:70-77.
Pope, C.A., 3rd. 1991. Respiratory hospital admissions associated with PM10 pollution in Utah, Salt Lake, and Cache valleys. Arch. Environ. Health 46:90-97.
World Bank. 1992. World Development Report 1992: Development and the Environment. New York: Oxford University Press. 308 pp.