John Snow and Modern-Day Environmental Epidemiology

Dale P. Sandler

From the Epidemiology Branch, National Institute of Environmental Health Sciences, P.O. Box 12233—Mail Drop A3-05, 111 T. W. Alexander Drive, Research Triangle Park, NC 27709.

What does an anecdote about John Snow have to do with modern-day epidemiology? And why use it to introduce an issue of the Journal highlighting the challenges of studying disease risks associated with low dose environmental exposures?

In this issue, Lilienfeld describes John Snow giving expert-witness testimony on behalf of industry (1Go). Besides being interesting on a historical basis, this incident raises several issues that are pertinent today. Lilienfeld's paper and the accompanying commentary by Vandenbroucke (2Go) deal directly or indirectly with the role and responsibilities of expert witnesses, the extrapolation of data on health effects from high dose exposures to low dose exposures, the importance of epidemiology to the development of public health policy, the current debates on environmental justice (3Go), and the use of the precautionary principle (4Go) in standard-setting. Furthermore, if faced with an issue similar to that faced by Snow—namely, local residents' being worried about health consequences associated with emanations from factories—would modern-day environmental epidemiologists be any better positioned to carry out appropriate studies and reach sound conclusions?

Snow can be seen at once as victim and perpetrator of sins that are common in epidemiology in general and in environmental epidemiology in particular. Was Snow victimized by the medical establishment, including The Lancet, for expressing views that were not commonly held by the scientists of the day? Were his peers outraged because of the reactionary social position he was taking (as suggested by Vandenbroucke)? On the other hand, was he as guilty as proponents of the miasma theory for trying to apply his theory of disease transmission to all situations without allowing for the possibility of multiple disease pathways? Did he fall into the trap of equating the absence of data with an absence of effect?

When Snow contended that emanations from the bone-boiling factories were not causing ill health in the community at large, he invoked arguments that are often raised when unexpected health effects are encountered following supposed low dose exposures. One argument is that such health effects are implausible given what we know about high dose exposures. In this instance, Snow noted that the factory workers were not dying and therefore health effects in the community at large were not plausible. A related argument is that, even if workers are dying or suffering other health effects, because of the distance from the exposure source, the exposure levels in the community are probably too low to plausibly affect health.

Health effects of low dose exposures are often seen as implausible, even in the face of accumulated consistent evidence. Such arguments have frequently been invoked in environmental epidemiology. Examples of low dose exposures that have been deemed implausible contributors to disease risk based on what is known about high dose exposures include passive smoking, residential radon exposure, childhood lead exposure, electromagnetic fields, and residence near nuclear facilities. If one begins with a fixed idea of what is plausible, arguments regarding susceptible subgroups, inverse dose rate, hormesis, multiple pathways, multifactor etiologies, and complex exposures (e.g., the different constituents of sidestream and mainstream smoke) are untenable.

But how do we know that the factory workers were not dying or suffering other ill effects? Snow cited no studies. All too often the absence of data is argued as proof of no effect. This issue becomes especially difficult when regulatory decisions are being made. In the absence of evidence, can something be considered safe? While science is important, it is ultimately social forces, as much as science, that guide regulators in decision-making.

Snow's statements and the questions that were put to him call to mind some of the fundamental difficulties inherent in environmental epidemiology. Today, there are numerous examples of residents who live near potential environmental hazards claiming health effects that can never be proven beyond a reasonable doubt. Although the "gold standard" is an unbiased risk estimate with precise confidence limits, studies focused on overt health effects are invariably underpowered because of the small numbers of residents in the neighborhoods of interest. Other creative approaches to assessment of subclinical health effects are more costly and difficult to implement, but even these studies are often too small for conclusive results. Yet, what is the right thing to do? If we wait for strong scientific evidence before we act—if we require proof that workers are dying or evidence of overt illness in the community—have we waited too long? Few clusters are ever resolved with the identification of a causal link between some localized exposure and disease. While many apparent clusters may be artifacts, what is the real cost of the true hazards that cannot be proven? These were the issues facing Parliament when Snow testified on behalf of industry.

What is the role of the epidemiologist in this quagmire? In Snow's London, the living conditions of people near the factories were likely to have been dismal. There were no doubt residents who perceived their symptoms as being related to the smells—smells that, if nothing else, impacted the quality of life. Policy-makers must balance "doing the right thing" with regard to human suffering and quality of life with the financial costs of doing so. Epidemiology can only go so far in providing the answers.

It is this political and social tug-of-war that makes environmental epidemiology especially difficult. On the one hand, there are well—funded industries with a financial stake in the outcome of such research. As Vandenbroucke notes (2Go), these industries often are in a position to exploit the many weaknesses that epidemiologists are trained to identify in their own studies and in the work of others to cast potentially damaging results in a more favorable light. On the other hand, there are environmental groups committed to proving that a particular environmental exposure can be linked to a variety of personal complaints; these groups may be motivated by the possibility of effecting social change through science or by the prospect of receiving needed medical attention or financial compensation. Those who attempt to work in this arena often find themselves and their research attacked from all directions.

Environmental epidemiology is difficult to conduct today for other reasons as well. Adequate tools with which to measure and quantify exposures are lacking. Studies are often unable to detect meaningful effects because exposures are low, infrequent, or difficult to measure with certainty. How many investigators are willing to tackle this problem? In the case of the bone-boiling factories, would research linking questionnaire data on symptoms to factory releases be believed? Would a study relating distance from the factory to disease be sufficient evidence of effect? What health effects would be plausible based on known biologic mechanisms? How well could those effects be measured, and could they be measured objectively? Is there a biomarker of exposure? If a biomarker exists, does it measure relevant past exposures? Is the measure unaffected by current health status—particularly the disease under study?

In addition to Lilienfeld's historical report and Vandenbroucke's commentary, this issue of the Journal features papers that illustrate various aspects of the difficulties faced in studying health effects of environmental exposures. Several of these include innovative attempts to improve the quality of such research.

The paper by Viel et al. (5Go) may come closest to what many may think of as environmental epidemiology. The authors have examined the spatial distribution of soft tissue sarcomas and non-Hodgkin's lymphomas around an incinerator with high dioxin emissions. Their results are suggestive but need to be followed by studies incorporating more rigorous exposure assessment—perhaps a biologic measure of exposure such as that used in the study of polychlorinated biphenyls and breast cancer reported by Zheng et al. (6Go). Other studies described in this issue used a variety of approaches to exposure assessment. Rondeau et al. (7Go) linked estimates of levels of aluminum and silica in drinking water to risks of dementia and Alzheimer's disease. Laden et al. (8Go) used questionnaire data on use of electric blankets to estimate exposure to electromagnetic fields, and Gustavsson et al. (9Go) used questionnaire data and expert assessment by industrial hygienists to classify environmental and occupational exposures. Radiation workers are one of the few groups for which historical records of personal exposure typically are available. Dupree-Ellis et al. (10Go) took advantage of such records to estimate cumulative external radiation exposure.

Several of the papers evaluate methods for assessing exposure. For example, Oglesby et al. (11Go) average individual-level annoyance scores to estimate community-level exposure to air pollution. The authors propose that this measure better accounts for exposure variability than data from fixed-site monitoring stations. This is an interesting twist in a field where much work is based on linking data from monitoring stations with population-level mortality statistics. The measure seems to be easy to operationalize, and it correlates well with monitoring station data, although its ultimate utility may be limited. The real gold standard—a more precise direct measure of individual exposure, rather than another indirect measure—is what is needed. Hwang et al. (12Go) propose an alternative modeling approach whereby air pollution monitoring station data are used to ascribe exposures to individuals with and without school absences due to respiratory disease. Auvinen et al. (13Go) compare several possible methods for measuring and classifying exposure to electromagnetic fields. This is a topic that has been hurt by the lack of consensus on the best and most appropriate exposure measure, and results tend to vary for studies employing different exposure metrics. The paper by Karagas et al. (14Go) attempts to link a biologic measure, arsenic in toenails, with an environmental measure of arsenic in water. The toenail measure is likely to reflect total body burden, but it appears to correlate with water only when water levels are high. This presents an interesting regulatory dilemma. The best epidemiologic research may be based on a direct measure of body burden such as levels in toenails, whereas it is water levels that need to be regulated. Studies of toenail arsenic levels may not shed direct light on the link between water levels and disease.

As these papers demonstrate, technological advances are making possible a wide range of new study designs and strategies to better assess both exposures and outcomes. Although progress has been made, research in environmental epidemiology is far from perfect. As epidemiologists face pressures and criticisms from industry, regulatory bodies, and other scientific disciplines, it is important to not lose sight of the lessons from John Snow.

NOTES

Reprint requests to Dr. Dale P. Sandler at this address (e-mail: sander{at}niehs.nih.gov).

REFERENCES

  1. Lilienfeld DE. John Snow: the first hired gun? Am J Epidemiol 2000;152:4–9.[Abstract/Free Full Text]
  2. Vandenbroucke JP. Invited commentary: the testimony of Dr. Snow. Am J Epidemiol 2000;152:10–12.[Free Full Text]
  3. Foreman CH Jr. The promise and peril of environmental justice. Washington, DC: Brookings Institute, 1998.
  4. Horton R. The new new public health of risk and radical engagement. (Editorial). Lancet 1998;352:251.[ISI][Medline]
  5. Viel J-F, Arveux P, Baverel J, et al. Soft-tissue sarcoma and non-Hodgkin's lymphoma clusters around a municipal solid waste incinerator with high dioxin emission levels. Am J Epidemiol 2000;152:13–19.[Abstract/Free Full Text]
  6. Zheng T, Holford TR, Tessari J, et al. Breast cancer risk associated with congeners of polychlorinated biphenyls. Am J Epidemiol 2000;152:50–8.[Abstract/Free Full Text]
  7. Rondeau V, Commenges D, Jacqmin-Gadda H, et al. Relation between aluminum concentrations in drinking water and Alzheimer's disease: an 8-year follow-up study. Am J Epidemiol 2000;152:59–66.[Abstract/Free Full Text]
  8. Laden F, Neas LM, Tolbert PE, et al. Electric blanket use and breast cancer in the Nurses' Health Study. Am J Epidemiol 2000;152:41–9.[Abstract/Free Full Text]
  9. Gustavsson P, Jakobsson R, Nyberg F, et al. Occupational exposure and lung cancer risk: a population-based case-referent study in Sweden. Am J Epidemiol 2000;152:32–40.[Abstract/Free Full Text]
  10. Dupree-Ellis E, Watkins J, Ingle JN, et al. External radiation exposure and mortality in a cohort of uranium processing workers. Am J Epidemiol 2000;152:91–5.[Abstract/Free Full Text]
  11. Oglesby L, Künzli N, Monn C, et al. Validity of annoyance scores for estimation of long term air pollution exposure in epidemiologic studies: The Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA). Am J Epidemiol 2000;152:75–83.[Abstract/Free Full Text]
  12. Hwang J-S, Chen Y-J, Wang J-D, et al. Subject-domain approach to the study of air pollution effects on schoolchildren's illness absence. Am J Epidemiol 2000;152:67–74.[Abstract/Free Full Text]
  13. Auvinen A, Linet MS, Hatch EE, et al. Extremely low-frequency magnetic fields and childhood acute lymphoblastic leukemia: an exploratory analysis of alternative exposure metrics. Am J Epidemiol 2000;152:20–31.[Abstract/Free Full Text]
  14. Karagas MR, Tosteson TD, Blum J, et al. Measurement of low levels of arsenic exposure: a comparison of water and toenail concentrations. Am J Epidemiol 2000;152:84–90.[Abstract/Free Full Text]
Received for publication March 17, 2000. Accepted for publication March 29, 2000.


Related articles in Am. J. Epidemiol.:

John Snow: The First Hired Gun?
David E. Lilienfeld
Am. J. Epidemiol. 2000 152: 4-9. [Abstract] [FREE Full Text]  

Invited Commentary: The Testimony of Dr. Snow
Jan P. Vandenbroucke
Am. J. Epidemiol. 2000 152: 10-12. [Extract] [FREE Full Text]