Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
gum as the indicator plant. Soil Sci. Soc. Am. Proc. 24:527- 528. Kubota, J., V. A. Lazar, L. N. Langan, and K. C. Beeson. 1961. The relationship of soils to molybdenum toxicity in cattle in Nevada. Soil Sci. Soc. Am. Proc. 25:227-232. Kubota, J., S. Rieger, and V. A. Lazar. 1970. Mineral composition of herbage browsed by moose in Alaska. J. Wildt. Manage. 34:565-569. Lagerwerff, J. V. 1971. Uptake of cadmium, lead, and zinc by radish from soil and air. Soil Sci. 111:129-133. Lazar, V. A., and K. C. Beeson. 1956. Mineral nutrients in native vegetation on Atlantic Coastal Plain soil types. J. Agric. Food Chem. 4:439-444. Loper. G. M., and Dale Smith. 1961. Changes in micronutrient composition of the herbage of alfalfa, medium red clover, Ladino clover, and bromegrass with advance in maturity. Wise. Agric. Exp. Stn. Res. Rept. 8. University of Wisconsin, Madison. Lowenstarn, H. 1964. Biological problems relating to the composi- tion and digenesis of sediments. In The earth sciences, T. W. Donelly (eel). Rice University Press, Houston. pp. 137-195. Mwozumi, M., T.V. Otow, and C. Patterson. 1969. Olemical con- centrations of pollutant lead aerosols, terrestrial dusts, and sea Sampling, ·Sample Preparation, and Storage for Analysis 97 salts in Greenland and Antarctic snow strata Geochirn. Cosmo- chirn. Acta 33 : 1247. Pierce, J. 0., and J. H. Meyer. 1971. Technical note: Sampling and analysis considerations in evaluating levels of atmospheric lead. Atrnos. Environ. 5:811-813. Schroeder, H. A. 1971. Losses of vitamins and trace minerals re- sulting from processing and preservation of foods. Am. J. Clin. Nutr. 24:562-573. Southern Cooperative Group. 195 1. Studies of sampling techniques and chemical analyses of vegetables. South. Coop. Ser. Bull. No. 10. Tanaka, A., and S. Yoshida. 1970. Nutritional disorders of the rice plant in Asia. Int. Rice Res. lnst. Tech. Bull. No. 10. International Rice Research institute, Manila. 51 pp. U.S. Geological Survey. 1972. Geochemical survey of Missouri: Plans and progress for fourth six-month period (January- June, 1971). U.S. Geol. Surv. Open-File Rept. U.S. Geo- logical Survey, Denver, Colo. Weiss, H. V., M. Koide, and E. D. Goldberg. 197la. Selenium and sulfur in a Greenland ice sheet. Science 172:261. Weiss, H. V., M. Koide, and E. D. Goldberg. 197lb. Mercury in a Greenland ice sheet. Science 174:692.
XII Experimental Design and Epidemiological Considerations JAMES E. BANTA, Co-Chairman HENRY L. LUCAS, Co-Chairman Kenneth C. Beeson, George K. Davis, Howard C. Hopps, Warren Winkelstein, Jr. Epidemiology has been variously described as "statistical study," "population survey," or "field survey." Actually, epidemiology seeks not only to describe but also to explain the distribution of various phenomena in populations. Ac· cording to Stall ones ( 1962), "Epidemiology is the descrip- tion and explanation of the differences in occurrence of events of medical concern in subgroups of a population where the population has been subdivided according to some characteristic believed to influence the occurrence of the event." It follows that the use of epidemiology to study the re- lationship of the geochemical environment to human health and disease implies interest in events related to health and in the manner in which they are influenced by the geo- chemical environment. The fust aim of the study is identi- fication of the characteristics of the individuals exposed or at risk, and the environment associated with the occur- rence or existence of events concerned with disease or health. This environment includes all pertinent components of the ecosystem of which the individual is a part; i.e., the air, soils, plants, animals, food supplies, litter, and water sources containing elements of potential concern. Associa- tions can be identified and tested by suitable statistical techniques to establish the magnitude of significance. "Significant association" does not imply the establishment of cause and effect, but it does provide an important lead in the identification of important factors related to the 98 occurrence of an event or disease (Banta and Fonaroff, 1969; Correa and Strong, 1972; Francis, 1960; Stallones, 1962). An epidemiological study should be based, ideally, either on probability samples or on the study of the total popula- tion under consideration, to permit generalization. If the total population is not studied, then it is important that the method of sampling be described and that relevant descriptive information of the total population be provided. The particular value of an epidemiological study is that all persons in a sample are examined without regard to the presence or absence of a particular disease or factor, thus avoiding some of the problems of biased selection. This method makes possible the identification of geochemical factors that differentiate meaningful groups; i.e., those with and without the disease or phenomenon. The most important task is the selection of an unbiased sample that is truly representative of the particular population being studied (Marienfeld, 1972). Epidemiological studies permit identification and study of multiple factors. Usually, a number of important factors concerning the relationship between the geochemical en- vironment and occurrence of disease will have been estab- lished by previous studies and these known factors can be examined during the analysis in relation to unknown fac- tors. Other important variables may be controlled by selecting homogenous groups with respect to age, sex, race,
or other designated characteristics (Armstrong, 1971, 1972; Correa and Strong, 1972; Krueger, 1966). Because epidemiology is the study of the natural history of disease, and the population is uncontrolled or free-living in contrast to the laboratory experiment or controlled clin- ical study, it must exploit the experiment of opportunity. The experimental variables are from within the environment, as well as factors associated with the individuals constituting the population at risk. It is obviously impossible to assign individuals at random to particular groups, such as exposure versus no exposure, or treatment versus no treatment. AI· though it may restrict some statistical generalizations, a natural experiment approximates the experimental situation for many inferences. DESIGN OF SYSTEMS Epidemiological, ecological, and geochemical events occur in time. Therefore, in associating these events with a matrix or system, the temporal dimension must be considered. Data is gathered in a time frame that is either retrospective, pro- spective, or cross-sectionaL In the usual retrospective study, individuals are examined who have been exposed, or not ex- posed, to a particular risk factor, geochemical substance or element, and for this reason, either exhibit or are free of a particular sign of disease. The groups are then examined for various factors that differentiate them and may be suggestive of the cause, such as the exposure dosage of a trace element. In a prospective study, also called a cohort or longitudinal study, a population is selected for study without regard to disease status or exposure characteristics. AU individuals in the group are examined, and the disease status or other factor is determined. The population is then exposed to the risk factor (trace element). The characteristics of the expo- sure, magnitude and duration (i.e., dosage) must be taken into account. At the end of the exposure time, a re-exami- nation is conducted, the occurrence of the disease or other factor determined, and rates calculated to compare the differences in incidence between the groups. It is also use- ful to examine a number of factors; for example, sex, age, diet, blood pressure, etc. One of the disadvantages of the prospective study is the time that must elapse from the initial survey to the follow-up survey. The inherent prob- lem of multiple surveys is also a disadvantage (Banta and Fonaroff, 1969; Francis, 1960; Lilienfeld, 1965; StaUones, 1962). Cross-sectional Study Perhaps one of the more useful epidemiological approaches for the identification of geographic differences or deter- mination of geographic patterns, is the cross-sectional study. In this method the prevalence (presence or occurrence) of a disease or factor is determined and expressed as a rate; Experimental Design and Epidemiological Considerations 99 i.e., the number of events per unit of population within a defmed geographic region. The prevalence rates occurring within defmed geographic regions may then be compared. The distribution and magnitude of geochemical or trace elements within these defmed geographic regions may be compared similarly and significant correlations derived from the data (World Health Organization, 1959; Arm· strong, 1971 ; Banta and Fonaroff, 1969; Lilienfeld, 1965; Marienfeld, 1972;Sauer and Brand, 1971). Mortality data are special kinds of incidence data. Death certificates, which are recorded by aU political jurisdictions, are a valuable source of epidemiological data. The several state offices of vital statistics, as well as the National Center of Health Statistics and U.S. Department of Health, Edu- cation, and Welfare, are the primary sources of these data. Death rates may be compared by geographic region, as well as by such factors as sex, age at time of death, place of birth, place of death, and specific cause of death. Because deaths are coded in accordance with the International Classification of Diseases (National Center for Health Sta· tistics, 1967), geographic comparisons may be made with some confidence that the disease causing death has been classified in a standardized, comparable manner through- out the entire United States and Canada. International comparisons are also feasible where good national vital· statistics records are maintained, such as in the United Kingdom, Sweden, or Australia. In these countries, the International Classification of Diseases (World Health Organization, 1959) is also used for classifying the cause of death. ENVIRONMENTAL INFLUENCES Epidemiology recognizes the occurrence of disease as a manifestation of environmental influences and for this reason it has often been caUed "human ecology." It recognizes man as part of an ecosystem and as a biological entity, subject to biological adaptation in response to en- vironmental stresses (Francis, 1960). This model can be used to throw valuable light on the influence of the geo- chemical environment on human health (i.e., successful adaptation) and disease (i.e., maladaptation), and the model and its components-man, animals, plants, soil, rocks-can be studied to determine the dynamics of trace elements as they pass throughout the components of the ecosystem. Such a model ideally should have stability between the various components. This means a human population not subject to migration and mobility, a con- stant food (animals and plants) chain, linked to a stable geochemical environment in the soil, rocks, and water supply (Armstrong, 1971; Banta and Fonaroff, 1969; Marienfeld, 1972). The epidemiologist must be particu- larly mindful of the problem of human mobility and mi- gration in determining risk or exposure to a particular
100 THE RELATION OF SELECTED TRACE ELEMENTS TO HEALTH AND DISEASE geochemical environment. This is one important advantage of using indicator nonhuman animals, both wild and do- mestic, for determining geochemical effects (Banta and Fonaroff, 1969). They may not be as subject to severe problems of migration and mobility as man (Banta and Fonaroff, 1969; Francis, 1960; Lilienfeld, 1965; Marien- feld, 1972;Sauer and Brand, 1971). Because epidemiological studies are concerned with the incidence and prevalence of diseases in population groups rather than in individuals, a reliable index of disease must be employed to measure the incidence. The method should be objective, such as measurable physiological phenomena. It could be, for example, blood pressure, electrocardio- graphic data, blood-serum levels of substances such as glu- cose, enzymes, as well as microscopically identifiable histo- pathologic changes. These measurements are more useful than the assignment of a diagnostic category, which is either negative or positive and not quantitative (Pollock and Krueger, 1960; Rose and Blackburn, 1968). If trace elements from the geochemical environment provoke a physiological response, the response is likely to be a dosage-response phenomenon, graded in relation to the magnitude of the dosage. Trace elements in human tissues and body fluids must be measured. In an epidemiologic study, it might be possible to detect dosage response within defmed population groups, depending on graded exposure to particular trace elements. At any rate, stan- dardization of criteria and methodology become exceed- ingly important when a variety of data are to be collected from many different persons and geographic sites. Atten- tion to field collection and laboratory methodology is equally critical to ensure comparability and quality control of the data (Kubota, 1972; Laitinen, 1972). Experience in conducting epidemiologic research on cardiovascular diseases a decade ago indicated that criteria and methodol- ogy must be standardized. This research effort necessitated several conferences, sponsored by such agencies as the World Health Organization and the National Heart Institute of the U.S. Public Health Service. These conferences resulted in manuals on classification, criteria, and methodology that are very useful today both intrinsically and as models (Na- tional Center for Health Statistics, 1967; World Health Organization, 1959; Pollock and Krueger, 1960; Rose and Blackburn, 1968). The complexity of epidemiological studies relating the geochemical environment to human health and disease seems to dictate the need for interdisciplinary plarming, execution, and analysis of any studies undertaken. The presence of an epidemiologist and biostatistician should improve the likelihood of a study demonstrating a rela- tionship. In summary, the epidemiological approach is inseparable from the study of the relationship of the geochemical en- vironment to health and disease. Its salient points can be summarized as follows: 1. Epidemiology is a useful method for demonstrating relationships between the geochemical environment and human health and disease. The method is most useful when complemented by controlled laboratory and field experi- mentation. 2. Because of the diversity of skills required in studying the geochemical environment as it relates to health, epi- demiological studies should be planned and conducted by interdisciplinary groups. 3. The total ecosystem should be examined, and each component of the system (i.e., soil, plants, animals, man) defmed and examined separately. 4. Epidemiological studies should be conducted on se- lected, relatively simplified ecosystem models with few variables. Areas with defined geographic boundaries, un- mobile populations, and a minimum of external food sources would be suitable. 5. Attention must be given to comparability of criteria and methodology to ensure standardization among dif· ferent studies. Use of standard procedure manuals (i.e., American Hospital Association, American Public Health Association, World Health Organization, etc.) is advisable where possible. If unstandardized procedures are used, they should be related to standard procedures. 6. Studies can employ standard vital statistics collected by the usual health jurisdictions. The National Center for Health Statistics can provide important health and demo- graphic data for the United States. 7. Point prevalence of particular physiologic or patho- logic phenomena can be compared between defined geo- graphic locations according to some geochemical variable. This method will help to detect significant effects that are geographically distributed and may be the result of geo- chemical influence. The kind of studies employed to demonstrate the rela- tions between geochemical environment and health are listed below. SW'VeyS For norms, trends, fluxes For ascertaining associations For testing models Field Experiments Laboratory Experiments Physiological reactions Balances Mathematical Modeling Work Subsystems Systems The investigator should finally proceed to the point of
testing hypotheses, the final goal in scientific inquiry. The question that is fmally asked in the present context is, Is there a relation between man~ health and the geochemical environment? The premises on which a positive answer to this question is based are listed below: 1. The deficiency or toxicity, or both, of a number of elements-such as iodine, fluorine, lead, and iron-have been demonstrated in man. 2. Because of the communality of man and animals with respect to basic biochemical phenomena, and from demon- strated results with animals, it may be assumed that other elements could cause deficiency, toxicity, or both to occur in man-even though such results may not yet have been observed (e.g., for copper, molybdenum, cadmium, zinc, and selenium). 3. The ultimate source of the elements is the geochem- ical environment. While substantial evidence indicates a relationship be- tween man's health and the geochemical environment, specific quantitative proof presents a challenge to the investigator. It would be prudent to consider the factors discussed below to increase the likelihood of successful research. Some Difficulties The Relation Is Complicated Under field conditions, a fairly strong relation between local geochemical environ- ment and health of animals has been demonstrated. Under laboratory conditions, very strong relations between ani- mal health and dietary levels of the elements have been demonstrated. The relations under both field and labora- tory conditions are complex; they involve multiple inter- actions between the elements, and of the elements with other factors. Distant Environments Affect Man The relation to local geochemical environment for man is strongly conditioned by the fact that much of man's intake, orally or otherwise, of the elements-including those in liquids of various forms-can be from materials produced in distant geo- chemical environments (e.g., many processed foods and drinks). This factor will reduce in varying degrees the strength of relation to local geochemical environment below that observed in animals in the field . If it were pos- sible to conduct laboratory work with humans as critical as that done with animals, strength of relation of human health to dietary levels of the elements should be about that found in laboratory animals. The nature of relation for humans could, of course, differ from that for animals even as does the nature of relation between animals of different species. One must also clearly distinguish be- Experimental Design and Epidemiological Considerations 10 I tween nature and strength of relation: nature has to do with the biological mechanisms involved, which become reflected in the mathematical forms or shapes of curves; strength is concerned with how well one can predict cer- tain variables from other variables. Kinds of Studies That Should Be Made Proper experimental and survey designs and careful statis- tical analysis of all data are important in all studies cate- gorized here. ⢠Health surveys seeking associations with geochemical environment ⢠Health-response surveys observing health changes caused by increasing or decreasing the intake of the ele- ments (assessment to be made clinically, by measurements, on blood, skin, hair, and enzyme activity) ⢠Balance experiments on man, with different levels and combinations of the elements and associated biochemical changes (possibly using radioforms of the elements) ⢠Balance studies on animals and plants ⢠Balance studies on soil and water ⢠Mathematical modeling work on facets of the soil- plant-animal-man system The Systems Approach The so-called "systems approach" involves the following steps: I. definition of the system; 2: identification of components of the system, points of interaction, flows of matter and energy between compo- nents of the system, points of this interaction, flows of matter and energy between components, inputs to the sys- tem and outputs from it (usually includes a system diagram identifying these components); 3. quantification (expression in mathematical state- ments-mathematical modeling) of how components of the system change with time of fluxes between compo- nents, of output fluxes as related to state of the system and input fluxes (usually shown by differential or different equations with probabilistic features); 4. analysis of the mathematical model of the system by numerical, or nonnumerical means, or both, to ascertain the important features of behavior of the model and how realistic they are; and 5. comparison of the predictions yielded by the model with data from surveys and experiments. Sound accomplishment of steps in the systems approach depends critically on the data and the concepts developed from the kind of studies listed.
102 THE RELATION OF SELECTED TRACE ELEMENTS TO HEALTH AND DISEASE Comment Many facets of the system defined implicitly by the Sub- committee on the Geochemical Environment in Relation to Health and Disease in Geochemical Environment in Re- lation to Health and Disease (Hopps and Cannon, 1972) are known. There is, however, much that is not known, and much of that is critical to realistic quantification and practical decision making. A valuable feature of the systems approach is the fol- lowing: In the construction of a system diagram, and in the mathematical modeling steps, macro and micro facets of the system and its behavior are put into perspective. Things known are more clearly seen, but-more importantly- things not known and the degree of their criticality are brought sharply into focus. This then provides important guidance with respect to the kind of experiments and sur- veys needed and the priorities to be placed on them. It would be presumptuous to assume that this brief exam- ination of the design of experiments, surveys, and data anal- ysis provides a comprehensive guide. However, it suggests a beginning and at least an orientation to one of the more im- portant aspects of elucidating the relationship of the geo- chemical environment to health and disease-satisfactory scientific methodology. REFERENCES Armstrong, R. W. 1971. Medical geography and its geologic substrate. In Environmental geochemistry in health and disease, H. L. Can- non and H. C. Hopps (eds). Geol. Soc. Am. Mem. No. 123. Geo- logical Society of America, Boulder, Colo. pp. 211-219. Armstrong, R. W. 1972. Is there a particular kind of soil or geologic environment that predisposes to cancer? Ann. N.Y. Acad. Sci. 199:239-248. Banta, J. E., and L. S. Fonaroff. 1969. Some considerations in the study of geographic distribution of disease. Prof. Geogr. 31: 87-92. Correa, Pelayo, and J. P. Strong. 1972. Atherosclerosis and the geo- chemical environment: A critical review. Ann. N.Y. Acad. Sci. 199:217-228. Francis, T. 1960. The epidemiological approach to human ecology in comparative medicine in transition. The University of Michi- gan School of Public Health, Lansing. pp. 385-394. Hopps, H. C., and H. L. Cannon (eds). 1972. Geochemical environ- ment in relation to health and disease. Conference on Geochem- ical Environment in Relation to Health and Disease, October 4- 6, 1971, New York. New York Academy of Sciences, New York. 352 pp. Krueger, D. E. 1966. New numerators for old denominators-multiple causes of death. Epidemiological study of cancer and other chronic diseases. Natl. Cancer lnst. Monogr.19:431-443. Kubota, Joe. 1972. Sampling of soils for trace element studies. Ann. N.Y. Acad. Sci. 199:105-117. Laitinen, H. A. 1972. Analytical methods for trace metals: An over- view. Ann. N.Y. Acad. Sci. 199:173-181. Lilienfeld, A. N. 1965. Epidemiological concepts applied to studies of chronic diseases. In Genetics and the epidemiology of chronic diseases. U.S. Public Health Serv. Publ. No. 1163. U.S. Govern- ment Printing Office, Washington, D.C. pp. 87-102. Marienfeld, C. 1. 1972. A field study relatq geochemical environ- ment to health and disease. Ann. N.Y. Acad. Sci. 199:335-348. National Center for Health Statistics. 196 7. Eighth revision, Inter- national Qassification of Diseases. Adapted for use in the United States by the National Center for Health Statistics. U.S. Public Health Serv. Publ. No. 1693. U.S. Government Printing Office, Washington, D.C. PoUock, H., and D. E. Krueger (eds). 1960. Epidemiology of cardio- vascular diseases methodology: Hypertension and arteriosclerosis. Am. 1. Public Health Suppl. 50(10). American Public Health Association Rose, G. A., and H. Blackburn. 1968. Cardiovascular survey meth- ods. WHO Monogr. Ser. 56. World Health Organization, Geneva. Sauer, H. I., and F. R. Brand. 1971. Geographic patterns in the risk of dying. In Environmental geochemistry in health and disease. H. L. Cannon and H. C. Hopps ( eds). Geol. Soc. Am. Mem. No. 123. Geological Society of America, Boulder, Colo. pp. 131- 150. Stallones, R. A. 1962. Introduction to chronic disease epidemiology. Contin. Educ. Monogr. No. 2, Cardiovascular diseases. American Public Health Association, Western Branch, San Francisco. pp. 2G-3l. World Health Organization. 1959. Hypertension and coronary heart disease: aassification and criteria for epidemiological studies. WHO Tech. Rept. Ser. World Health Organization, Geneva. p.168.
