Vol. 132, No. 1

AMERICAN JOURNAL OF EPIDEMIOLOGY

Copyright © 1990 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved

Printed in U.S.A.

THE EPIDEMIOLOGIC FIELD INVESTIGATION: SCIENCE AND JUDGMENT IN PUBLIC HEALTH PRACTICE RICHARD A. GOODMAN,1 JAMES W. BUEHLER,2 AND JEFFREY P. KOPLAN3 Goodman, R. A. (CDC, Atlanta, GA 30333), J. W. Buehler, and J. P. Koplan. The epidemiologic field investigation: science and judgment in public health practice. Am J Epidemiol 1990;132:9-16. Epidemiologic field investigations are often done in response to acute public health problems. When outbreaks of disease occur, there usually is an urgent need to identify the source and/or cause of the problem as a basis for control. Alternatively, the identification of environmental or occupational hazards frequently demands evaluation of exposed persons and assessment of the risks of disease. In this commentary, the authors present a perspective on the publicsector practice of epidemiology by considering the factors that influence epidemiologic investigations in the field; contrasting epidemiologic field investigations with prospectively planned studies; and examining the complexities of the relations between epidemiology and public health practice. disease outbreaks; environmental exposure; epidemiologic epidemiology; health policy

The art of epidemiologic reasoning is to draw sensible conclusions from imperfect data (1, p. 206). George W. Comstock

Epidemiologic field investigations are often done in response to acute public health problems (2). When outbreaks of disease occur, there usually is an urgent need to identify the source and/or cause of Received for publication July 21, 1989, and in final form December 15, 1989. 1 Epidemiology Program Office, Centers for Disease Control, Atlanta, GA. 2 Center for Infectious Diseases, Centers for Disease Control, Atlanta, GA. 3 Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control, Atlanta, GA. Reprint requests to Dr. Richard A. Goodman, Epidemiology Program Office, Centers for Disease Control, 1600 Clifton Road NE, Atlanta, GA 30333. The authors are indebted to the following persons for their review of this manuscript and suggestions: Drs. Paul Blake, Claire Broome, Andrew Dean, David Fraser, Sander Greenland, Michael Gregg, Robert Gunn, Richard Hoffman, Gregory Istre, James Marks, Michael Osterholm, Jeffrey Sacks, Robert Tauxe, and Stephen Thacker.

methods;

the problem as a basis for control. Alternatively, the identification of environmental or occupational hazards frequently demands evaluation of exposed persons and assessment of the risks of disease. Regardless of the nature of the problem, there exists an immediate need to investigate, to recommend control and preventive measures, and to convince the affected community to accept public health recommendations. Thus, because epidemiologic field investigations compress within a limited time frame the application of scientific methods, the formulation and implementation of health policy, and the art of public relations, they represent a unique form of the practice of epidemiology. Acute disease outbreaks and hazardous exposures often require immediate action. (For the purposes of this article, an outbreak is the occurrence of a number of cases of a disease (or condition) in excess of the number expected in a given time and place; in some situations, a single case may constitute such an unusual occurrence.) This

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

10

GOODMAN ET AL.

need for expediency sometimes means that practical recommendations (table 1) (2-5). the investigation must be done with less These published guidelines have focused on than the desired scientific rigor. A common the practical and operational aspects of feature of the relation between epidemiol- field investigations. In general, however, ogy and public health practice is that spe- they have not addressed the role of science cific actions must sometimes be taken with- and the pressure of setting policy in deterout satisfying all the important criteria that mining whether an investigation will be establish causal relations. done, the extent of the investigation, and The key issue for the investigative epi- appropriate recommendations for control. demiologist and public health decisionmaker is: To what extent must an acute AGGRESSIVENESS AND SCIENTIFIC RIGOR IN THE FIELD INVESTIGATION health problem be epidemiologically defined before action should be initiated? In Several factors influence the aggressivethis commentary, we present a perspective ness, extent, and scientific rigor of an epion the public-sector practice of epidemiol- demiologic field investigation. The primary ogy by considering the factors that influ- reason for doing an investigation is usually ence epidemiologic investigations in the to provide a basis for implementing control field; contrasting epidemiologic field inves- measures to terminate a threat to the pubtigations with prospectively planned stud- lic's health, the magnitude of which is ies; and examining the complexities of the determined by characteristics such as the relation between epidemiology and public severity and size of the problem and the health practice. potential for further spread. Other reasons include opportunities for research about a T H E EPIDEMIOLOGIC FIELD problem, training, public concern, legal obINVESTIGATION ligations, and such program considerations as available resources. While field investigations of acute problems share many characteristics with proTABLE 1 spectively planned epidemiologic studies, Guidelines for epidemiologic field investigations* they may differ in at least two important respects. First, because field investigations 1. Prepare for field work (e.g., administration, often start without clear hypotheses, they clearance, travel, contacts, designation of lead investigator, etc.). may require the use of descriptive studies 2. Confirm the existence of an epidemic. to generate hypotheses before analytical 3. Verify the diagnosis. studies are conducted to test these hy4. Identify and count cases or exposures. potheses. Second, when acute problems oc—Create a case definition. cur, there is an immediate need to protect —Develop a line-listing. the community's health, as well as address 5. Tabulate and orient the data in terms of time, place, and person. its concerns; these responsibilities drive the 6. Take immediate control measures (if indiepidemiologic field investigation beyond cated). the confines of data collection and analysis 7. Formulate hypotheses. and into the realm of public health action. 8. Test hypotheses. 9. Plan additional systematic study(ies). The concepts and techniques used. in 10. Implement/evaluate control and preventive field investigations derive from clinical medicine, epidemiology, laboratory science, 11. measures. Initiate surveillance. decision analysis, skilled communications, 12. Communicate findings. and common sense. Guidelines for con—Summarize investigation for requesting authority. ducting epidemiologic field investigations —Prepare written report(s). have been developed to reflect the needs for urgency, multifaceted methods, and *See also references 2 and 3.

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

11

THE EPIDEMIOLOGIC FIELD INVESTIGATION

Developing control measures The impetus for many epidemiologic field investigations is the need to develop control measures to terminate an outbreak. In the prototypic investigation, control measures are formulated only after a series of other steps have been carried out (2) (table 1). In practice, however, decisions regarding control measures may be appropriate at any step in the sequence. For most outbreaks of acute disease, the scope of an investigation is a function of the levels of certainty about 1) the etiology of the problem (e.g., the specific pathogen or toxic agent) and 2) the source and/or mode of transmission (e.g., waterborne or airborne). When the problem is identified, the levels of certainty for the etiology and source/mode of spread may range from known to unknown (figure 1). These basic dichotomies are illustrated in figure 1 by four examples that probably represent the extremes. In many situations, control measures can be implemented empirically, while in others, interventions are appropriate only after exhaustive epidemiologic investigation. Often, preliminary control measures can be

instituted on the basis of limited, initial information and then modified as needed as the investigation proceeds. For example, the occurrence of a single case of hepatitis A in a child day-care setting can lead to the administration of immune globulin prophylaxis to an entire cohort of exposed children and staff (6). This response recognizes the known epidemiologic relations between asymptomatic and symptomatic cases, and thus directs efforts toward prophylaxis of exposed persons while minimizing the need for an extensive investigation to specifically identify infected persons. In this instance, the response is predicated on routine policy and guidelines that have been developed by experts on the basis of studies and previous outbreak experience, as well as virtual certainty about both the etiology of the problem and its mode of spread. More commonly, there is some degree of uncertainty about the etiology or about the source/mode of spread (figure 1). In most outbreaks of gastrointestinal disease, the control measures selected will depend on knowing whether transmission has resulted from person-to-person spread or from a common-source exposure and—if the latter is the case—identifying the source. For ex-

Source/transmission mode Known Investigation Control

Unknown +

+++

Example: Hepatitis A in day care (6)

LU

Investigation Control

Investigation +++ Control

+++

+++

+

Example: Salmonellosis and marijuana (7)

Investigation

+++

Control +

Example: Parathion

Example: Legionnaires'

poisoning (8)

disease (9)

FIGURE 1. Relative emphasis of investigative and control efforts (response options) in disease outbreaks as influenced by levels of certainty about etiology and source/mode of transmission. "Investigation" means extent of the investigation; "control" means the basis for rapid implementation of control measures. Pluses show the level of response indicated: +, low; ++, intermediate; +++, high.

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

12

GOODMAN ET AL.

ample, a multistate outbreak of Salmonella muenchen infection in 1981 required an extensive epidemiologic field investigation, including an analytical (case-control) study, before the mode of spread was found to be personal use of or household exposure to marijuana (7). The converse situation (i.e., in which the source is presumed, but the etiology is unknown) is illustrated by an outbreak of acute organophosphate poisoning in Sierra Leone in 1986 (8). In that outbreak, contaminated bread was initially implicated as the source of exposure, thereby enabling the epidemiologists to make recommendations for preventing subsequent outbreaks and the laboratory investigators to identify parathion as the etiologic agent. Finally, as was illustrated by the Legionnaires' disease outbreak in 1976, an extensive epidemiologic field investigation can fail to identify either the cause or the source/mode in time to control the acute problem, but still enable advances in knowledge that ultimately lead to preventive measures (9). Characteristics of the problem The severity and magnitude of the problem influence the urgency and course of an investigation. Severity is indicated by correlates such as the degree and nature of complications (e.g., mortality), the duration of illness, the need for treatment and hospitalization, and the economic impact. For example, virtually all cases of rabies occurring among humans in the United States trigger extensive epidemiologic investigations because of the vital need to quickly identify other exposures and the animal source. Similarly, clusters of nosocomial infection—especially those involving postsurgical or immunocompromised patients—are often investigated both because of the potential for serious complications and greatly prolonged hospitalization and because of the possibility of iatrogenic illness, which has its own special urgency as an unnecessary medical event.

Research opportunities Because disease outbreaks are "natural experiments," they present opportunities to address questions germane to both basic and applied epidemiology. Even when a clear policy exists for control of a specific problem, investigation may still provide opportunities to identify risk factors for infection or disease, define the spectrum of manifestations of disease, measure the impact of control measures or clinical interventions, or assess the usefulness of microbiologic markers. Some outbreaks that initially appear to be "routine" may lead to important epidemiologic discoveries. For example, in 1983, by aggressively pursuing a cluster of cases of diarrhea, investigators were able to trace the chain of transmission of a unique strain of multi-antibiotic-resistant Salmonella backward from the affected persons to hamburger they had consumed, to the meat supplier, and ultimately to the source animal herds (10). This investigation played a key role in clarifying the linkage between antibiotic use in the cattle industry and human disease. Training By analogy to clerkships in medical school and postgraduate residencies, outbreak investigations provide opportunities for training in basic epidemiologic skills. Just as clinical training often is accomplished at the same time patients are being cared for, training in field epidemiology often simultaneously assists in disease control and prevention. For example, the Epidemic Intelligence Service of the Centers for Disease Control has provided assistance to state and local health departments while simultaneously training health professionals in the practice of applied epidemiology (11). Political and public concern Although the public's perceptions of hazards may differ from those of epidemiolo-

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

THE EPIDEMIOLOGIC FIELD INVESTIGATION

gists, these perceptions sometimes drive the political process that mandates investigations or actions. In some cases, public concerns mandate investigations that are premature or unlikely to be fruitful from a scientific perspective but critical in terms of community relations. At the other extreme, attempts at more thorough epidemiologic investigation can be misinterpreted as community experimentation or bureaucratic delay. In a large outbreak of Escherichia coli enteric disease at Crater Lake National Park in 1975, a 1-day delay in implementing control measures in order to obtain more epidemiologic data resulted in a Congressional hearing and charges of a "cover-up" (12, 13). The 1-day "delay" represented a difference of judgment among decision-makers regarding the weight of evidence necessary to support an action as substantive as closing a national park. Small clusters of disease (e.g., leukemia or adverse fetal outcomes) are an example of problems that frequently generate great public concern. Small clusters often occur by chance alone and only occasionally yield new research information when investigated (14). However, because community members may perceive a health threat and because certain clusters do represent specific preventable risks, some public health agencies have developed standard procedures for investigating such clusters even though the likelihood of identifying a remediable cause is low. Legal obligations

Some epidemiologic investigations are likely to be used as testimony in civil or criminal trials. In these situations, investigations may be carried further than they would otherwise have been if court testimony were not likely. For example, in one investigation of a cluster of cardiac arrests in an intensive care unit, the investigation went to the unusual length of attempting to determine the contents of medical records that could not be located (15, 16).

13

Program considerations Certain disease-control programs at national, state, and local levels have specific and extensive requirements for epidemiologic investigation. For example, as part of the measles elimination effort in the United States, a measles outbreak is considered to exist if one case of measles is confirmed in a community (17). Accordingly, every case of measles is investigated to identify and vaccinate susceptible persons and to evaluate other control strategies, such as the exclusion from school of children who cannot provide proof of immunity. A potentially detrimental effect of such policies is that costly investigations may yield limited public health benefits (18). Because field investigations can be costly in terms of personnel time and resources, they may detract from other activities. Thus, the capacity to conduct epidemiologic field investigations may be limited by competing demands of other programs within an agency, whether at national, state, or local levels. Under these circumstances, failure to investigate a specific problem could result in a public health problem of greater magnitude which, if controlled earlier, would have caused less economic loss. Conversely, there may be reluctance to initiate investigations for problems when options for control or intervention are not likely to be identified (e.g., cancer clusters); under these circumstances, resources committed to field investigations could undermine efforts to address remediable problems. CAUSATION AND THE FIELD INVESTIGATION

The need to determine whether a statistical association also supports a causal relation is as essential to an epidemiologic field investigation as it is to a prospectively planned study. Thus, criteria to assess causal associations (19-21) are integral to the scientific framework of epidemiologic field investigations. The challenge to the

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

14

GOODMAN ET AL.

public-sector epidemiologist is to balance the need to assess causality through the process of scientific inquiry with the potentially conflicting need to intervene quickly to protect the public's health. Few epidemiologic field investigations appear to address the criteria explicitly; when the criteria are applied to such investigations, they may be only partially met. The usefulness of individual criteria (e.g., temporality, the strength of association, the biologic gradient, consistency, plausibility) varies (20). In any outbreak, different groups of people representing diverse viewpoints may be exposed, affected, or involved in some respect. Because of differences in knowledge, beliefs, and perceived impact of the outbreak, each group may draw from the same information different conclusions regarding causality. For example, in a restaurant-associated foodborne outbreak, each of several interest groups (e.g., restaurant patrons, management, media, attorneys, and local health officials) is likely to have a different reason and threshold for judging the food from the restaurant to be the source of disease. In this situation, the epidemiologist's concerns might focus on the strength of association and the biologic gradient between exposure to a certain food item and illness, while a restaurant patron's primary concern may simply be plausibility.

UNIQUE CHALLENGES TO EPIDEMIOLOGISTS IN FIELD INVESTIGATIONS

Epidemiologists investigating acute problems in the field face unique challenges that sometimes constrain the ideal use of scientific methods. In contrast to prospectively planned studies, which are based upon carefully developed and refined protocols, the investigative response to acute problems such as outbreaks must rely on data sources that are less readily controlled and that may literally change with each successive hour or day.

Data sources Field investigations often use information abstracted from a variety of sources, such as hospital, outpatient medical, or school health records. These records vary dramatically in completeness and accuracy by patient, care provider, and facility, since entries are made for purposes other than conducting epidemiologic studies. Thus, the quality of such records as sources of data for epidemiologic investigations can be substantially less than that of information obtained, for example, from standardized, pretested questionnaires. Small numbers In a prospectively planned study, the epidemiologist determines appropriate sample sizes on the basis of statistical requirements for power. In contrast, outbreaks can involve a relatively small number of persons, thereby imposing substantial restrictions on study design, statistical power, and other aspects of analysis. These restrictions, in turn, place limitations on inferences and conclusions derived from an epidemiologic field investigation. Specimen collection Because the field investigator often arrives on the scene "after the fact," it is not always possible to collect necessary environmental or biologic specimens. For example, implicated food items may have been entirely consumed or discarded, an implicated water system may have been flushed, or ill persons may have recovered, thereby precluding the collection of acutephase specimens. Under these conditions, the epidemiologist is dependent on the diligence of health-care providers who first see the affected persons, as well as the recall of affected persons, their relatives, or other members of the community. Publicity Acute disease outbreaks often generate considerable local attention and publicity. In this regard, media coverage can assist

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

THE EPIDEMIOLOGIC FIELD INVESTIGATION

the investigation by helping to develop information or identify cases or by assisting in implementing control measures, thereby improving the speed and completeness of the investigation. On the other hand, such publicity may cause affected persons and others in the community to develop preconceptions about the source or cause of an outbreak, which in turn can lead to potential biases in comparative studies or failure to fully explore alternate hypotheses. Parenthetically, media representatives in pursuit of the most current information on an investigation may monopolize epidemiologists' time to the detriment of the field investigation itself. Reluctance to participate While virtually all health departments are empowered to conduct investigations and gain access to records, voluntary and willing participation of the involved parties is more conducive than forced participation to a successful investigation. Persons whose livelihoods or related interests are at risk may be reluctant to cooperate voluntarily. This may often be the case for common-source outbreaks associated with restaurants and other public establishments, in environmental or occupational hazard investigations, or for health-care providers suspected of being sources for transmission of infectious diseases such as hepatitis B. When involved parties do not willingly assist, delays may compromise access to and quality of information. Conflicting pressures to intervene Epidemiologists conducting field investigations must weigh the need for further investigation against the need for immediate intervention. However, the strong and varied opinions of affected persons and others in the community can interfere with the optimal scientific approach.

15

sent "quick and dirty" epidemiology. This perception may reflect the inherent nature of circumstances for which rapid responses are required. However, these requirements for action do not provide a rationalization for taking epidemiologic shortcuts. Rather, they underscore for the practicing epidemiologist the importance of combining good science with prudent judgment. A better descriptor of a good epidemiologic field investigation would be "quick and appropriate." In judging an epidemiologic field investigation, paramount consideration should be given to the quality of the science. However, this should not be the sole standard; rather, the full range of limitations, pressures, and responsibilities imposed on the investigator must also be taken into account. The goal of the epidemiologist should be to maximize the scientific quality of the field investigation in the face of applicable limitations and competing interests. The epidemiologist in the public sector must reconcile the multiple competing/ conflicting interests and develop a scientifically optimal study design. Sound judgment combined with a rigorous scientific effort will result in a field investigation that meets the public's needs and also provides scientific data of the highest quality. Thus, standards for an epidemiologic field investigation are that it: 1) address an important public health problem in the community, as defined by standard public health measures (e.g., attack rates, serious morbidity, or mortality) or community concern; 2) be timely; 3) examine resource needs early enough in the investigation and commit an appropriate amount of public resources; 4) use appropriate methods of descriptive and/or analytical epidemiology; 5) probe causality to the degree required to enable identification of the source and/or etiology of the problem; and 6) establish immediate control and long-term interventions.

IMPLICATIONS

There is a perception in the public health community that field investigations repre-

REFERENCES 1. Comstock GW. Vaccine evaluation by casecontrol or prospective studies. Am J Epidemiol

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

16

GOODMAN ET AL.

1990;131:205-7. 2. Gregg MB. The principles of an epidemic field investigation. In: Holland WW, Detels R, Knox G, eds. Oxford textbook of public health. Vol 3. New York: Oxford University Press, 1985:284-99. 3. Galbraith NS. The application of epidemiological methods in the investigation and control of an acute episode of infection. In: Holland WW, Detels R, Knox G, eds. Oxford textbook of public health. Vol 4. New York: Oxford University Press, 1985:3-21. 4. Dixon RE. Investigation of endemic and epidemic nosocomial infections. In: Brachman PS, Bennett JV, eds. Hospital infections. Boston: Little, Brown and Company, 1979:73-93. 5. Kelsey JL, Thompson WD, Evans AS. Epidemic investigation. In: Methods in observational epidemiology. New York: Oxford University Press, 1986:212-53. 6. Recommendations for protection against viral hepatitis. MMWR 1985;34:313-16. 7. Taylor DN, Wachsmuth K, Yung-Hui S, et al. Salmonellosis associated with marijuana: a multistate outbreak traced by plasmid fingerprinting. N Engl J Med 1982;306:1249-53. 8. Etzel RA, Forthal DN, Hill RH Jr, et al. Fatal parathion poisoning in Sierra Leone. Bull WHO 1987;65:645-9. 9. Fraser DW, Tsai TR, Orenstein W, et al. Legionnaires' disease: description of an epidemic of pneumonia. N Engl J Med 1977;297:1189-97. 10. Holmberg SD, Osterholm MT, Senger KA, et al. Drug-resistant Salmonella from animals fed anti-

microbials. N Engl J Med 1984;311:617-22. 11. Langmuir A. The Epidemic Intelligence Service at the Centers for Disease Control. Public Health Rep 1980;95:470-7. 12. Rosenberg ML, Koplan JP, Wachsmuth IK, et al. Epidemic diarrhea at Crater Lake from enterotoxigenic Escherichia coli: a large waterborne outbreak. Ann Intern Med 1977;86:714-18. 13. Crater Lake National Park. Hearing before the Committee on Interior and Insular Affairs, US Senate, 94th Congress, Medford, Oregon, September 6, 1975. Washington, DC: US GPO, 1975. 14. Schulte PA, Ehrenberg RL, Singal M. Investigation of occupational cancer clusters: theory and practice. Am J Public Health 1987;77:52-6. 15. Sacks JJ, Stroup DF, Will ML, et al. A nurseassociated epidemic of cardiac arrests in an intensive care unit. JAMA 1988;259:689-95. 16. Sacks JJ, Aung HK, Sniezek JS. The epidemiology of missing records. (Letter). JAMA 1988;259:685. 17. Measles prevention. MMWR 1987;36:409-26. 18. Weissman JB, DeWitt WE, Thompson J, et al. A case of cholera in Texas, 1973. Am J Epidemiol 1975;100:487-98. 19. Hill AB. The environment and diseases: association or causation? Proc R Soc Med 1965;58:295300. 20. Rothman KJ. Modern epidemiology. Boston: Little, Brown and Company, 1986:16-20. 21. Hennekens CH, Buring JE. Statistical association and cause-effect relationships. In: Mayrent SL, ed. Epidemiology in medicine. Boston: Little, Brown and Company, 1987:30-53.

Downloaded from https://academic.oup.com/aje/article-abstract/132/1/9/77979 by University of Tasmania Library user on 04 January 2018

The epidemiologic field investigation: science and judgment in public health practice.

Epidemiologic field investigations are often done in response to acute public health problems. When outbreaks of disease occur, there usually is an ur...
596KB Sizes 0 Downloads 0 Views