EPIDEMIOLOGY

Epidemiology is the study of the frequency, distribution, and determinants of disease in humans. Its aim is the prevention or effective control of disease. The term originated in the study of epidemics, rapidly spreading diseases that affect large numbers of a population (from the Greek epi meaning upon and demos meaning people). Epidemiology touches on ethics in two key areas: The need for competent and honest use of its information, and questions of responsibility raised by the global picture it presents of the health of humanity.

Speculation about the nature and causes of disease dates back to antiquity. The formal history of epidemiology, like that of statistics, begins with the systematic official recording of births and deaths in the seventeenth century, proceeding to the quantitative investigation of diseases with the emergence of scientific medicine in the nineteenth. Based on the theory of probability, statistical inference reached maturity in the early-twentieth century and gradually spread into a wide range of disciplines. Its application to medical research gave rise to biostatistics and contemporary epidemiology.

There is no clear division between the two fields. Epidemiology focuses more on public health issues and the need for valid population-based information, but it uses the theory and methods of biostatistics. Its practitioners tend to be individuals with primary interest and training in medicine or a related science, whereas biostatisticians come from mathematics. They work together as members of the medical research team, in the dynamic context of scientific advances and the latest information technology.

Modern Epidemiology

The mathematical approach to medicine, with the methodical tabulation of patient information on diseases and treatment outcomes, was introduced in the 1830s by the French physician Pierre C. A. Louis (1787–1872). As a notable result of his researches in Paris hospitals, his Numerical Method revealed the uselessness of bloodletting. Inspired by Louis, his British student William Farr (1807–1883) became the central figure in the development of vital statistics in England and the use of statistics to address public health concerns. Farr worked with John Snow (1813–1858), the physician who investigated the cholera epidemic sweeping through London in 1854. Snow's finding that the cholera poison was transmitted in contaminated water from the Broad Street pump was a milestone event in epidemiology and public health. Farr also provided guidance in statistics for Florence Nightingale (1820–1910) to support her work in hospital reform.

The existence of microbes was discovered in the late-seventeenth century by the Dutch lens grinder Antonie van Leeuwenhoek (1632–1723), who saw "animalcules, more than a million for each drop of water" through his microscope (Porter 1998, p. 225). The role of germs as causes of disease was established by Louis Pasteur (1822–1895), French chemist and founder of microbiology. Pasteur invented methods to isolate and culture bacteria, and to destroy them in perishable products by a heat treatment now called pasteurization. He found that inoculation by a weakened culture provided immunity, protection against the disease. This explained the earlier discovery of the English physician Edward Jenner (1749–1823) that vaccination with the milder cowpox protected against smallpox. (Vaccination comes from the Latin vacca meaning cow.) The German physician Robert Koch (1843–1910), founder of bacteriology, further developed techniques of isolating and culturing bacteria. He identified the germ causing anthrax in 1876, tuberculosis in 1882, and cholera in 1883. He contributed to the study of other major diseases, including plague, dysentery, typhoid fever, leprosy, and malaria.

Extensive public health measures of hygiene and immunization, along with the introduction of the sulfonamide drugs in the late 1930s and antibiotics in the 1940s, brought most infectious diseases under control. Attention turned to chronic diseases, by then the leading causes of morbidity and mortality—multicausal diseases with a long latency period and natural course. Two historic discoveries of the mid-twentieth century were tobacco use as a cause of lung cancer, and risk factors for heart disease. From the study of infectious and chronic diseases epidemiology has evolved into a multidimensional approach, defined by disease, exposure, and methods, with focus on new developments in medical science. Its many specialties include cancer, cardiovascular, and aging epidemiology, environmental, nutritional, and occupational epidemiology, clinical and pharmaco-epidemiology, and molecular and genetic epidemiology. With the sequencing of the human genome,

genetics is assuming increasing importance across all lines of inquiry. In its principles of studying human populations, epidemiology is related to psychology, sociology, and anthropology, all of which employ statistical inference.

Basic Concepts and Methods

Epidemiology may be descriptive or analytic. Descriptive epidemiology reports the general characteristics of a disease in a population. Its methods include case reports, correlational studies (to describe any association between potential risk factors and disease in a given database) and cross-sectional surveys (to determine prevalence of a disease and potential risk factors at a given point in time). Analytic epidemiology uses observational and experimental studies. The latter are clinical trials to test the effectiveness of interventions to treat or prevent a disease. But experimentation on humans is not ethically feasible for studying causes of disease. Observational research designs are thus the primary tools of epidemiology, the main types being case-control and cohort studies. After definition of some basic terms, these are discussed further below.

MEASURES OF MORBIDITY AND MORTALITY. Some basic concepts of epidemiology are listed in Table 1. It is important to distinguish between the prevalence of a disease and its incidence. Prevalence signifies the amount of disease present at a point in time, such as the proportion of people with adult-onset diabetes in the United States on January 1, 2005. Incidence refers to new cases diagnosed during a given period of time, such as the proportion of U.S. adults diagnosed with diabetes in 2005. The denominator of incidence rate is person-time, a useful concept that allows for inclusion of subjects with different starting dates and lengths of time observed in a study. Causes of a disease can be investigated by observing incidence in a well-defined group of subjects without the disease, and patterns of disease incidence can be compared over time or populations.

Mortality is measured in terms of crude death rate, the actual proportion observed, or the standardized death rate, which involves adjustment for some characteristic. The example shows age-adjusted cancer death rates for the states of Alaska and Florida. Alaska has a much lower crude death rate than Florida, but its population is much younger. Both breast and prostate cancer are associated with older age, but after age-adjustment the two states are seen to have similar death rates for these two sites, both lower than the national average. The adjusted figures are meaningless in themselves, but provide for valid comparison of rates across groups and time. U.S. cancer death rates have been adjusted using the 2000 U.S. age distribution to make them comparable back to 1930 and ahead to the future.

OBSERVATIONAL RESEARCH DESIGNS: CASE-CONTROL AND COHORT STUDIES. A case-control study is retrospective: It identifies a group of people with the disease (cases) and selects a group as similar as possible to the cases but without the disease (controls). The aim is to determine the proportion of each group who were exposed to the risk factor of interest and compare them. Table 2 shows results of the case-control study of lung cancer and smoking reported in 1950 by Sir Richard Doll (b. 1912) and Sir Austin Bradford Hill (1897–1991), British pioneers of epidemiology and biostatistics. They identified 649 men and sixty women with lung cancer in twenty London hospitals and matched them with controls of the same age and sex but without lung cancer. The information they collected on all participants included their smoking history. The observed association, measured by the so-called odds ratio (the odds of smoking in cases over the odds of smoking in controls), was clearly statistically significant.

A cohort study is usually prospective. (It may be historical, if based on recorded past information.) It identifies a large group (cohort) of individuals who do not have the disease but for whom complete information is available concerning the risk factor(s) of interest; the cohort is then observed for the occurrence of the disease. A noted cohort design was the Framingham Heart Study, initiated by the U.S. Public Health Service in 1948 to identify risk factors for heart disease. Over 5,000 adult residents of Framingham, Massachusetts, men and women with negative test results for cardiovascular disease, agreed to join the study and undergo repeat testing at two-year intervals. The age and test measures at the start of each two-year period were used to classify subjects. Results of a thirty-year follow-up evaluation (part of a multivariate analysis including other risk factors and cardiovascular outcomes) are shown in Table 3, demonstrating a strong association between systolic blood pressure and incidence of coronary heart disease. Other suitable groups for cohort studies are members of professional groups, like doctors and nurses.

There are advantages and disadvantages pertaining to each research design, and the choice depends on the circumstances of the scientific question of interest. Any observed association then requires careful interpretation.

Association or Causation?

Possible reasons for an observed statistical association are listed in Table 4. Chance is simply the meaning of

the P-value, the probability that the association is due to chance. Bias refers to systematic errors that do not cancel out with larger sample size, but distort the results in one direction. For example, in a case-control study patients with the disease may be more likely to recall exposure to the risk factor than the controls, leading to recall bias. Bias is a serious problem in observational studies and needs to be assessed in the particular context of each research design. Confounding is the effect of an extraneous variable that is associated with the risk factor, but is also an independent risk factor for the disease. For example, an association between birth rank and Down's syndrome, the genetic disorder Trisomy 21 (an extra copy of chromosome 21) does not imply causality; the confounding variable is maternal age, which is associated with birth rank and is a known risk factor for the disease. There may also be confounding variables as yet unknown, but their potential effects must always be considered.

The establishment of causation is a long-debated problem in the philosophy of science. In the practical field of medicine, where life-and-death decisions must be made every day, there are guidelines to help assess the role of agents in the etiology of disease. When microbes were being identified as causes of devastating diseases in the late-nineteenth century, Robert Koch formulated postulates to prove that a particular microbe causes a given disease. Anticipated by his teacher Jacob Henle (1809–1885), these are also called Henle-Koch Postulates. They are shown in Table 5, along with current updates using molecular biology. The original version claims only necessary causation, not sufficient; the microorganism needs a susceptible host. Even more general, the molecular guidelines are expressed in terms of statistical association. But they are the organizing principle in contemporary studies of microbial etiology, crucial for the identification of newly emerging pathogens that may pose serious threats to public health.

Guidelines for establishing causality in observational studies are listed in Table 6. Formulated by Sir Austin Bradford Hill, they are based on criteria employed in the 1964 U.S. Surgeon General's Report to show that smoking causes lung cancer. Applied in a wider context, they are to be used primarily as an aid to exploration. In general there is no necessary or sufficient condition to establish causality from an observed association. Such conclusions result from a consensus of the scientific community.

Epidemiology and Ethics

The complex, probing methods of epidemiology yield tentative, partial, often conflicting results, replete with qualifications. Taken out of context by interest groups or the media, they can mislead and have harmful consequences. Their correct use requires professional competence and integrity. But beyond these issues of immediate concern, epidemiology plays a larger role. With its adjusted measures allowing comparison of health patterns over space and time, it provides a quantitative aerial video of the globe. Some of the images it presents are troubling.

There are now more obese than undernourished people living on earth, and their number is increasing rapidly in developing nations. According to a 2000 estimate of the World Health Organization (WHO), there are 220 million adults with Body Mass Index (BMI) <17, classified as undernourished, and over 300 million with BMI > 30, defined as obese. (BMI is weight in kilograms divided by height squared in meters.) This global epidemic of obesity, called globesity, brings with it the related conditions of diabetes, hypertension, and heart disease, and the problem is equally serious for children.

The harmful effects of tobacco have been known for half a century, and while the prevalence of smoking has been slowly declining in most industrialized nations, it has been rising steadily in the developing world. It is estimated that the number of smoking-related premature deaths worldwide, 5 million in 2000, will rise to 10 million per year by 2030, with 70 percent occurring in developing countries. Tobacco use will kill more people than the combined mortality due to malaria, pneumonia, tuberculosis, and diarrhea.

In the area of infectious diseases, after decades of exuberant optimism reality set in with the appearance of acquired immune deficiency syndrome (AIDS) in the 1980s. Homo sapiens lives in a sea of microbes and will never have total control. Vigilance for the emergence of disease-causing strains must be the aim, to detect outbreaks, identify pathogens and their mode of transmission, and seek control and prevention. Knowing the cause may not eliminate the disease, even when possible in principle, if (as with smoking) it hinges on human behavior. AIDS, for example, is preventable. Ongoing threats include new diseases from mutation or isolated animal reservoirs (Ebola, West Nile, severe acute respiratory syndrome [SARS]), resurgence of older strains, drug-resistance, targeted release through bioterrorism, and rapid spread through global travel.

At a WHO conference held in Geneva in November 2004, experts issued an urgent appeal for greater international cooperation, and called on governments to make pandemic preparedness part of their national security planning. Of particular concern was the new bird influenza strain A(H5N1), which could mutate and cause a pandemic on the scale of the influenza epidemic of 1918 that killed more than 20 million people. It is estimated that a new pandemic virus could spread around the world in less than six months, infecting 30 percent of the population and killing about 1 percent of those infected. The drug industry would have to prepare billions of doses of the influenza vaccine within weeks of an outbreak to halt its course. There are questions of what could possibly be feasible technologically, the huge investment needed, and the driving force to motivate the effort when it cannot be a matter of fiscal gain.

In March 2005 the British medical journal Lancet published four articles reporting on the appalling state of global infant health care. Four million babies die each year in the first month of life, nearly all in low- and middle-income nations. The highest numbers occur in south-central Asian countries, while the highest rates are generally in Sub-Saharan Africa. It is estimated that three-quarters of these deaths could be prevented with low-lost interventions. A similar number of babies are stillborn and 500,000 mothers die from pregnancy-related causes each year. The moral implications of this public health tragedy are overwhelming.

The problems humanity faces at the start of the twenty-first century are inseparable from dominant worldviews and the interplay of powerful economic and political forces. Epidemiology provides health-related information as a guide to action. Its proper use is an essential component of the Ethics of Evidence, proposed for dealing with the uncertainties of medicine in the framework of contemporary culture (Miké 1999, 2003). The Ethics of Evidence calls for integrating the best evidence of all relevant fields to promote human well-being, anchored in an inescapable moral dimension. Looking to the future, it urges all to be aware, to be informed, and to be responsible.

VALERIE MIKÉ

SEE ALSO Biostatistics;Health and Disease.

BIBLIOGRAPHY

Altman, Lawrence K. (2004). "Experts Urge Greater Effort on Vaccine for Bird Flu." New York Times, November 13,
p. A3.

American Cancer Society. (2004). Cancer Facts and Figures 2004. Atlanta, GA: Author. Includes report on worldwide effects of tobacco use.

Brooks, George F.; Janet S. Butel; and Stephen A. Morse. (2001). Jawetz, Melnick, & Adelberg's Medical Microbiology, 22nd edition. New York: McGraw-Hill.

Doll, Richard., and Austin Bradford Hill. (1950). "Smoking and Carcinoma of the Lung: Preliminary Report." British Medical Journal 2: 739–748.

Gail, Mitchell H., and Jacques Benichou, eds. (2000). Encyclopedia of Epidemiologic Methods. New York: John Wiley & Sons.

Hill, Austin Bradford. (1965). "The Environment and Disease: Association or Causation?" Proceedings of the Royal Society of Medicine 58: 295–300.

Hennekens, Charles H., and Julie F. Buring. (1987). Epidemiology in Medicine. Boston: Little, Brown.

Lawn, Joy E.; Simon Cousens; and Jelka Zupan. (2005). "Neonatal Survival 1: 4 Million Neonatal Deaths: When? Where? Why?" Lancet 365: 891–900. First report in the four-part series.

Lilienfeld, David E.; Paul D. Stolley; and Abraham M. Lilienfeld. (1994). Foundations of Epidemiology, 3rd edition. New York: Oxford University Press.

Miké, Valerie. (1999). "Outcomes Research and the Quality of Health Care: The Beacon of an Ethics of Evidence." Evaluation & the Health Professions 22: 3–32. Commentary by Edmund D. Pellegrino, "The Ethical Use of Evidence in Biomedicine," is included in this issue.

Miké, Valerie. (2003). "Evidence and the Future of Medicine." Evaluation & the Health Professions 26: 127–152.

Porter, Roy. (1998). The Greatest Benefit to Mankind: A Medical History of Humanity. New York: W. W. Norton.

Rothman, Kenneth J., and Sander Greenland. (2000). "Hill's Criteria for Causality." In Encyclopedia of Epidemiologic Methods, eds. Mitchell H. Gail, and Jacques Benichou. New York: John Wiley & Sons.

Stokes, Joseph III; William B. Kannel; Philip A. Wolf; et al. (1989). "Blood Pressure as a Risk Factor for Cardiovascular Disease: The Framingham Study—30 Years of Followup." Hypertension 13(Supplement I): 113–118.

INTERNET RESOURCE

Controlling the Global Obesity Epidemic. World Health Organization (WHO). Updated September 2003. Available from http://www.who.int/nut/obs.htm. From the organization's Nutrition pages.