Descriptive and Analytic Epidemiology

Descriptive Studies

Descriptive studies are designed to describe data on health outcomes, such as disease incidence, prevalence, and mortality according to three variables: person, place, and time. Person variables describe the people who develop disease in terms of their personal characteristics, such as age, gender, race, marital status, blood type, immune status, occupation, socioeconomic status, and so on, and where and when they were exposed to the agent causing the disease. Place variables may include any or all three sites: where the individual was when disease occurred; where the individual was when he or she became infected from the source; and where the source became infected with the etiologic agent. Time variables may refer to duration, age, calendar time, birth cohort, or time trend (secular trend, period trend, and seasonal trend). All these belong to what questions, because they tell us what happened. Graphical methods and descriptive statistics are commonly employed in descriptive studies.

Descriptive studies can be used to generate hypotheses about the causal association between exposure and outcome. For example, for a given disease, detailed plotting of the duration from exposure to death (time variable) against age and sex (person variable) may prompt generation of the hypothesis of higher mortality being associated with older males. The putative association or causal relationship may then be confirmed or refuted by testing this hypothesis using an analytic study design such as a prospective cohort study. A well-known historical example of a hypothesis-generating descriptive study is the exploratory analysis of the mortality rates from stomach cancer and from colon cancer of ethnic Japanese living in Japan and California. A comparison of cancer mortality rates of Japanese living in these two localities and between firstand second-generation Japanese immigrants in California and Hawaii suggested the hypothesis that diet and lifestyle (environmental factors) were more important risk factors than heredity (genetic factors) for this type of cancer.

Besides the main purpose of generating hypotheses, descriptive studies can also be used to assess the health status of a population and to plan public health programs. Descriptive studies include the following types:

• 1.
  Prevalence surveys are sample surveys conducted for the purpose of estimating the prevalence of a health condition or outcome or exposure to risk factors in a population, at a particular point in time.
• 2.
  In a cross-sectional study, both the health condition (or outcome) and exposure to risk factors are [Page 262]measured on the same subjects at the same time, so that the joint distribution is also available. Crosssectional studies can be used to calculate the prevalence ratio (referring to the proportion of existing cases), but not the incidence ratio (referring to the rate of occurrences of new cases), and can generate causal hypotheses but not to draw causal inferences.
• 3.
  A case report describes a single case and often focuses on unusual aspects of a patient's disease/condition or unusual association between the diseases and exposures, while a case series is a study of a series of case reports with a common health outcome of interest.
• 4.
  Surveillance studies refer to the ongoing systematic collection, analysis, and interpretation of outcome-specific data: These studies can be very useful, for instance, in alerting public health officials if many cases of a previously rare disease are occurring in a particular area, which would suggest the need for further investigation.
• 5.
  Analysis of secondary administrative data is not a true study design but refers to analysis of routinely collected data, such as those from population census, vital registrations, and tumor registries with the usual demographic characteristics of age, sex, race, and so on, and region. The above example of cancer study on Japanese migrants using mortality rates pertains to this type.

Analytic Studies

Analytic studies are the other main type of research design in epidemiology. They are designed to make comparisons and to test statistically hypothesized causal relationships. Ecological (group-level) data may be used for descriptive studies for the purpose of generating hypotheses, but for testing hypotheses, analytic studies are generally employed that require individuallevel data. Analytic studies consist of experimental (intervention) studies, quasi-experimental studies, and nonexperimental (observational) studies.

...