newcehnheader
[ home ] [ resource guide ] [ training manual ] [ get involved ] [ publications ] [ index ] [ about CEHN ]

 


What's New

Donate
 Donate Now

Links

 

 
   
 

Epidemiology

Susan K. Cummins, MD, MPH and Jane Lipscomb, RN, DrPH

              Return to Contents >
Many have called epidemiology one of the basic sciences of medicine. Epidemiologic research informs all aspects of clinical practice and provides current knowledge of important disease risk factors. In addition, it is the foundation of public health and preventive medicine.

The application of epidemiology to children's environmental health, however, poses many challenges. Epidemiological studies help identify preventable environmental risk factors, but they generally have not considered youth populations. Children are not little adults and epidemiological studies conducted among adult populations often underestimate the health risk of an environmental exposure for children.

This module provides a primer for teaching health care professionals basic epidemio-logic concepts critical to children's environmental health and outlines a structured approach to journal article review in which these concepts can be applied. The recommended learning methods are brief lectures, followed by work in journal clubs or small group sessions to reinforce the concepts introduced in the lecture.

This module outlines the concepts that should be covered; faculty will want to present their own illustrative case examples.

Learning Objectives

After completing this module, faculty will be able to teach students and residents to:

  • Define a core set of basic epidemiologic terms
  • Describe and contrast basic epidemiologic study designs
  • Discuss methods for assessing exposure to a risk factor, with an appreciation for the
  • difficulties inherent in assessing environmental exposures.
  • Implement a strategy for reading epidemiologic literature.

General Principles of Epidemiology

Epidemiologic Terms

Epidemiology is the study of the distribution and determinants of health-related states or events - diseases, injuries, etc. - in specified populations, and the application of this study to control health problems. This science has historically been concerned with the causes of disease epidemics - usually an infectious agent. In recent years, however, epidemiologists have focused on understanding the underlying distribution of and risk factors for chronic diseases and conditions such as asthma, cerebral palsy, and cancer.

Epidemiologic research can be:

  • Descriptive, evaluating the occurrence of adverse health effects in a defined population
  • Analytic, exploring causal relationships between various factors and the risk of adverse health effects
  • Experimental, studying the impact of treatments or other interventions on the occurrence, progression, or exacerbation of adverse health effects
  • Meta-analysis, using statistical methods to combine and summarize, in a systematic way, the results of different epidemiologic studies of a particular problem.

A sample is a selected subset of a population. A sample may be randomly or non-randomly selected and may be representative or non-representative. Ideally, one attempts to sample a representative subset of the population, so as to enhance the validity of extrapolating findings from the sample study to the target population.

A risk factor is an aspect of personal behavior or life-style, an environmental exposure, or an inborn or inherited characteristic that is known, on the basis of epidemiologic evidence, to be associated with the risk of adverse health effects. Epidemiologic research informs clinical practice by identifying risk factors that can be reduced, modified, or eliminated by appropriate intervention.

A confounding factor distorts the apparent effect of the risk factor of interest due to its association with both the risk factor and the adverse health effect. For example, exposure to deteriorating lead-based paint is both more common in impoverished children and a risk factor for childhood lead poisoning. Thus, poverty is a confounding factor in the association between lead-paint exposure and lead poisoning.

Bias is any deviation from truth in the collection, analysis, interpretation, publication, or review of data. Parents of children who are born with birth defects, for example, may have a much more complete recall of environmental exposures during the pregnancy than parents of children born without health problems. This is an example of information or recall bias.

A hypothesis is a supposition for explaining observed facts, expressed in a form that will allow it to be tested and thus accepted or rejected.

Hypothesis testing is the process of applying statistical tests to data to assess the likelihood that the data are consistent with a proposed hypothesis. In the course of hypothesis testing, the researcher first specifies both null and alternative hypotheses. The null hypothesis is a statement specifying that a variable is not associated with another variable or set of variables, or that two populations do not differ from each other by more than chance variation. The alternative hypothesis, in contrast, specifies that the variables are associated with each other or that the study populations significantly differ from each other.

Type I and II errors result in a false or mistaken conclusion in a study. A type I error is a "false positive" error; that is, the error of rejecting a null hypothesis when it is true. In contrast, a type II error is a "false negative" error; the error of accepting the null hypothesis when it is false.

Statistical power is the ability to demonstrate an association between variables, if it exists. Power is dependent on the difference to be detected, the intrinsic variability of study factors, the study design, and the size of study populations.

Incidence rate is the rate at which new events occur in a population. The numerator is the number of new events that occur in a defined period; the denominator is the population at risk of experiencing the event during this period.

Prevalence rate is the number of events (e.g., existing cases of disease) in a given population at a specified point or period in time. Prevalence is conceptually related to incidence as follows: 

Prevalence = Incidence x Duration

Epidemiologic Study Designs

Descriptive Designs

The ecologic study, which is commonly used in environmental epidemiology, is a correlational study in which the units of analysis are populations or groups of people rather than individuals. In a typical environmental ecologic study, populations of specific geographic regions are characterized by their exposure to an environmental factor and then compared for the risk of health problems.

The prevalence or cross-sectional study examines the relationship between health problems and other variables of interest as they exist in a defined population at one particular time.

Analytic Designs

The case-control study starts by identifying a set of individuals with the health problem of interest (cases) and an appropriate set of individuals without the problem (controls). The frequency of past exposure to particular risk factors is then compared among cases and controls to identify those factors that increase or decrease the risk of adverse health effects. Case-control studies are retrospective in design.

A cohort study is a longitudinal study that divides a defined population into those who are exposed and those who are not exposed to a particular risk factor. Study subjects are then followed over time to compare health problems in the two groups. The goal of a cohort study is to mimic an experiment, had one been possible.

Experimental Designs

The randomized control trial is an epidemiologic experiment in which subjects in a population are randomly assigned into groups (usually called "study" or "treatment" groups and "control" groups) to receive or not to receive an experimental preventive or therapeutic intervention. The results are assessed by rigorous comparison of rates of health problems, death, recovery or other appropriate outcome. The randomized control trial is generally regarded as the most scientifically rigorous method of hypothesis testing available in epidemiology. These trials are often conducted in a double-blind fashion, so that neither the researcher nor the subject knows which group the subject is in.

Meta-Analysis

Meta-analysis is a research design that uses statistical methods to combine and summarize, in a systematic way, the results of different epidemiologic studies of a particular problem. Meta-analysis has both a qualitative component (assessing the completeness of data and absence of biases) and a quantitative component (the integration of the numerical information). Meta-analyses often are helpful in setting health policy because they summarize large bodies of data.

To conduct a meta-analysis, the researcher compiles all published and unpublished research on the question of interest. These studies are reviewed systematically. Studies to be included in the final analysis are selected based on specified inclusion and exclusion criteria. Data from each study included are summarized statistically into a single measure of the risk factor-health effect relationship. In the course of compiling this summary, individual studies are weighted by the size of their sample, with larger studies carrying more weight.

Definitions of Common Epidemiologic Risk Statistics

Relative risk, or risk ratio, is a measure derived from the risk of a particular health effect occurring in the exposed group, divided by the health risk in the unexposed group. Relative risk may only be calculated from prospective studies. In theory, the relative risk can range from close to zero to infinity. A relative risk of greater than 1 denotes a factor associated with increased risk of a health problem. A relative risk of less than 1 denotes a factor associated with reduced risk of a particular health problem. The magnitude of the relative risk is important. Relative risks between 1 and 2, common in studies of multifactorial chronic disease, may be real or spurious. Relative risks are often expressed with their 95% confidence interval.

The confidence interval is a range of values that represent the uncertainty surrounding an estimate for a variable of interest, such as a blood measurement. It includes a range of values for the variable of interest. The confidence interval is calculated so that the range of values for the variable has a specified probability (usually 95%) of including the true value of the variable in the underlying population. If the confidence interval includes the value 1, then the factor is not significantly associated with adverse health effects.

Again, relative risk may only be calculated from prospective studies; however relative risk can be estimated in prevalence surveys by a prevalence ratio and in case-control studies by an odds ratio.

The prevalence ratio is analogous to the relative risk, and consists of the ratio of the prevalence in the risk group compared to the prevalence in the comparison group.

The odds ratio is technically defined as the ratio of the odds of health problems in the group with the risk factor, compared to the odds of health problems in the group without the risk factor. The odds ratio always overestimates the true relative risk, but the overestimate is relatively small if the health effect is rare. Thus, the odds ratio is a good estimate of the relative risk for rare health problems (those that occur in less than 1% of the population), but not for common ones. Like the relative risk, the odds ratio is often reported with a 95% confidence interval.

Exposure Assessment

Exposure can be measured in a variety of ways, including an interview, chart review, or biologic sampling or measurement. The best exposure assessment provides a biologic measurement of exposure for each individual study subject, such as urine cotinine for an estimate of quantitative tobacco exposure over a specific time period. In environmental epidemiological studies, individual exposure assessments often are not possible because exposure data are taken from public health monitoring stations, such as air quality stations. The following terms are often used in discussing exposure assessment:

  • A biomarker is a laboratory-derived, cellular or molecular indicator of exposure, health effect, or susceptibility. The blood lead level is an example of a biomarker of exposure, whereas cholinesterase depression following exposure to organophosphate is a biomarker of an effect.
  • Latency is a delay between exposure to a disease-causing agent and the appearance of the disease. The latency period for expression of leukemia following exposure to ionizing radiation averages five years; that for the development of mesothelioma following asbestos exposure is thought to be 20-30 years.
  • Measurement error is a systematic error arising from inaccurate measurements or misclassification. Measurement error can occur when measuring adverse health effects or exposure and can be non-differential (the same in all study groups) or differential (varying between study groups). Non-differential misclassification results in a bias towards the null hypothesis (no effect) whereas differential misclassification can bias the study toward either the null or alternative hypothesis and is therefore more threatening to the study's validity.

Challenges in Exposure Assessment

There are several difficulties inherent in conducting epidemiologic studies of environmental exposures.

  • Environmental exposures are often poorly defined and/or measured. Individual-level measurements of contaminants in the air, water and soil may be crude or completely unavailable.
  • Little knowledge currently exists regarding the health consequences of exposure to a mixture of chemicals. This lack of understanding severely limits the ability to study real-life environmental exposures, which usually consist of a mix of chemicals.
  • The fact that community-based environmental exposures often affect a relatively small number of individuals limits the usefulness of epidemiological study designs that depend on large study populations.
  • Communities are composed of individuals with varying degrees of susceptibility to environmental exposures. A dearth of knowledge about the underlying causes for this variability severely limits the ability to study and detect health effects among the most vulnerable segments of the population.
  • The long latency period between exposure to environmental hazards and evidence of chronic health problems such as cancers makes it particularly difficult to study these outcomes, which are often of greatest interest to concerned communities.

Reading the Literature

In order to maintain a current knowledge base in pediatrics, health care practitioners must learn to read the medical literature carefully. A brief strategy for reading research articles follows. In addition, the bibliography provides a number of published papers on the topic of reviewing research articles.

  • Define the research question, risk factors of interest, and health outcomes of interest. Ask yourself, "Is this an important question? Is it relevant to my patients?"
  • Identify the study design (cross-sectional study, case-control study, cohort study, ecologic design, randomized control trial, etc.).
  • Identify the study subjects, including the sampling method for obtaining them, the inclusion and exclusion criteria, and the exposure assessment methodology. Was the response rate high? Were the inclusion and exclusion criteria and the case ascertainment methods appropriate? Was the exposure assessment strategy adequate? Did the authors use a biologic measure of exposure, if available?
  • Evaluate the summary statistics. Was an appropriate method for summarizing the data used? Was there adjustment for confounding factors? Was there a statistically significant result? Was it interpreted properly?
  • Draw conclusions about the quality of the research and its relevance to your patients.

Learning Methods

  • Provide a brief lecture that teaches each set of epidemiologic concepts, followed a journal club during which students review articles that illustrate the points made in the lecture.
  • Conduct a "Call for Submissions" exercise, in which students act as the hypothetical Review Committee of a conference. Students review abstracts submitted for poster or paper presentations and reach an acceptance decision based on abstracts' epidemiologic merits. Students can review the abstracts alone or in small groups. Abstracts from current articles in the medical literature can be used. Clearly, abstracts published in the literature would be "acceptable" for the hypothetical conference, so it might be useful to alter summary statistics to include some "unacceptable" abstracts for students to catch.
  • Case studies such as those developed by the Centers for Disease Control and Prevention's Epidemic Intelligence Service may be useful for laboratory or small group work. These can be obtained from the American Teachers of Preventive Medicine web site (http://www.atpm.org/eis/).
  • The questions outlined in "Reading the Literature" (above) can be used as a hand-out to accompany articles for review in a journal club, "Call for Submissions," or other exercise.
  • Small groups could also compare and contrast the evaluation of disease clusters caused by toxic exposures versus infectious diseases. For example, what epidemiologic methods would be most useful to evaluate a cancer cluster suspected to have been caused by water contamination? What epidemiologic methods would be most useful to evaluate a hantavirus cluster suspected to have been caused by exposure to field mice?

Evaluation Methods

Knowledge of specific epidemiologic concepts may be assessed by pre- and post-tests, review of handouts prepared by students for their journal club or by discussion around a case study presented for laboratory or small group work.

Resources

The Epidemiology Monitor is a monthly newsletter that provides a wide range of brief articles on various aspects of epidemiology. Each year's January issue includes a listing of short courses and meetings in epidemiology offered during the year. To subscribe, write to 2560 Whisper Wind Court, Roswell, GA 30076, or call (770) 594-1613.

References

Bennett KJ, Sackett DL, Haynes RB, Neufeld VR, Tugwell P, Roberts RA. Controlled trial of teaching critical appraisal of the clinical literature to medical students. JAMA 257:2451-2454 (1987).

Fowkes FGR, Fulton PM: Critical appraisal of published research: introductory guidelines. BMJ 302:1136-40 (1991).

Gehlbach SH. Interpreting the Medical Literature: A Clinician's Guide. Lexington, MA: DC Heath and Company, 1982.

Guyatt GH, Sackett DL, Cook DJ. Users' guides to the medical literature. II. How to use an article about therapy or prevention. B. What were the results and will they help me in caring for my patients? Evidence-Based Medicine Working Group. JAMA 271:59-63 (1994).

Guyatt GH, Sackett DL, Sinclair JC, Hayward R, Cook DJ Cook RJ. Users' guides to the medical literature. IX. A method for grading health care recommendations. Evidence-Based Medicine Working Group. JAMA 274:1800-1804 (1995).

Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ Sackett DL. How to keep up with the medical literature: VI. How to store and retrieve articles worth keeping. Ann Intern Med 105:978-984 (1986).

Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: II. Deciding which journals to read regularly. Ann Intern Med 105:309-312 (1986).

Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: III. Expanding the number of journals you read regularly. Ann Intern Med 105:474-478 (1986).

Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: V. Access by personal computer to the medical literature. Ann Intern Med 105:810-816 (1986).

Haynes RB, McKibbon KA, Fitzgerald D, Guyatt GH, Walker CJ, Sackett DL. How to keep up with the medical literature: IV. Using the literature to solve clinical problems. Ann Intern Med 105:636-640 (1986).

Haynes RB, McKibbon KA, Fitzgerald D, Guyatt, GH, Walker CJ, Sackett DL. How to keep up with the medical literature: I. Why try to keep up and how to get started. Ann Intern Med 105:149-153 (1986).

Hayward RS, Wilson MC, Tunis SR, Bass EB, and Guyatt G. Users' guides to the medical literature. VIII. How to use clinical practice guidelines. A. Are the recommendations valid? The Evidence-Based Medicine Working Group. JAMA 274:570-574 (1995).

Hulley SB, Cummings SR. Designing Clinical Research: An Epidemiologic Approach. Baltimore, MD: Williwns and Wilkins, 1988.

Jaeschke R, Guyatt GH, Sackett DL. Users' guides to the medical literature. II. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients? Evidence-Based Medicine Working Group. JAMA 271:703-707 (1994).

Last JM. A Dictionary of Epidemiology (3rd Edition). New York, NY: Oxford University Press,1995.

Laupacis A, Wells G, Richardson WS, Tugwell P. Users' guides to the medical literature. V. How to use an article about prognosis. Evidence-Based Medicine Working Group. JAMA 272:234-237 (1994).

Levine M, Walter S, Lee H, Haines T, Holbrook A, Moyer V. Users' guides to the medical literature. IV. How to use an article about harm. Evidence-Based Medicine Working Group, JAMA 271:1615-1619 (1994).

Oxman AD, Cook DJ, Guyatt GH. Users' guides to the medical literature. VI. How to use an overview. Evidence-Based Medicine Working Group. JAMA 272:1367-1371 (1994).

Oxman AD, Sackett DL, Guyatt GH. Users' guides to the medical literature. I. How to get started. The Evidence-Based Medicine Working Group. JAMA 270:2093-2095 (1993).

Richardson WS, Detsky AS. Users' guides to the medical literature. VII. How to use a clinical decision analysis. B. What are the results and will they help me in caring for my patients? Evidence Based Medicine Working Group. JAMA 273:1610-1613 (1995).

Rothman KJ. Modern Epidemiology. Boston, MA: Little, Brown and Company, 1986.

Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology: A Basic Science for Clinical Medicine, 2nd ed. Boston, MA: Little Brown and Company, 1991.

Sackett DL, Rosenberg WM, Gray JA, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn't. BMJ 312:71-72 (1996).

Sackett DL. Applying overviews and meta-analyses at the bedside. J Clin Epidemiol 48:61-70 (1995).

Sackett DL. Inference and decision at the bedside. J Clin Epidemiol 42:309-316 (1989).

Sackett DL. Rules of evidence and clinical recommendations for the management of patients. Can J Cardiol 9:487-489 (1993).

Wilson MC, Hayward RS, Tunis SR, Bass EB, and Guyatt G. User's guides to the medical literature. VIII. How to use clinical practice guidelines. B. What are the recommendations and will they help you in caring for your patients? Evidence-Based Medicine Working Group. JAMA 274:1630-1632 (1995).