PU520: Principles of Epidemiology Unit 5 Analytic Epidemiology: Types of Study Designs Flashcards
Briefly describe analytic epidemiology and descriptive epidemiology.
Analytic epidemiology is concerned with the etiology (causes) of diseases and other health outcomes.
Descriptive epidemiology classifies a disease or other health outcome according to the categories of person, place, and time.
Attached photo is overview for these flash cards and chapter.
What are the two subcategories of analytic studies?
Observational and experimental (interventional) studies.
What does observational studies include? (3)
Ecological, case-control, and cohort studies.
The investigator does not have control over the exposure factor. Additionally, the investigator is unable to assign subjects randomly to the conditions of an observational study.
What are the three types of cohort studies?
Prospective, retrospective, and historical prospective.
What are the two types of experimental (interventional) studies?
Clinical trials and community interventions.
Experimental designs enable the investigator to control who is exposed to a factor of interest (for example, a new medication) and to randomly assign the participants into the groups used in the study. Random assignment of subjects is used in pure experimental designs.
What provides a degree of control over confounding?
Random assignment of subjects to study groups.
When the results of a study have been distorted by extraneous factors, confounding is said to have taken place.
Information on the seven factors that characterize study designs
N/A
When characterizing study design, in what type of studies would you typically only make one observation?
What about studies where you’d make more than one observation?
One observation is typically the approach of cross-sectional, many ecologic, and most case-control studies.
Two or more observations are typically made in cohort and experimental studies.
Why does directionally of exposure vary in research design?
In a retrospective research design, information is obtained about exposures in the past and is used in case-control studies. People are queried about exposure that leads to the disease. This can be collected in other ways like medical records.
In a single point of time, as in a survey, can take a snapshot of a population. Single point in time is a time reference for a cross-sectional study.
In prospective approaches, information about study outcomes is collected in the future after the exposure has happened-experimental and cohort studies. Prospective cohort studies the investigator starts with disease-free groups for which exposures are determined first. The groups are then followed prospectively for disease development.
What are the two ways of data collection? (Answer is not a specific term)
Either from existing research or data created through new research.
Why is timing in data collection important?
Subjects may fall off when studies extend for longer periods of time, forgetfulness of previous exposures, etc.
Timing really can be whenever the investigator wants.
In epidemiologic research design, what is typically the unit of observation for almost all studies?
When is this unit different? (Specfic type of study)
Most designs employ the individual as the unit of observation.
Ecologic study designs employ the group as the unit of observation. (or called the unit of analysis)
What is the jist of availability of subjects in research design?
Some classes of people may not be available for epidemiologic research, to include ethical reasons.
What is a study in which the units of analysis are populations or groups rather than individuals?
Ecologic Study
The groups that are selected might be residents in a specific geographic location–state, country, census tracts, counties, etc.
What involves an assessment of the association between exposure rates and disease rates during the same time period?
Ecologic comparison study
In an ecologic study, information about both exposures (explanatory variables) and outcomes is collected at the group level. To illustrate, one could explore “… the relationship between the distribution of income and mortality rates in states or provinces.”
In the hypothetical example of cancer mortality, researchers might hypothesize that people who live in lower-income areas have greater exposure to environmental carcinogens than those who live in higher-income areas, producing differences in cancer mortality.
What is an association between two variables measured at the group level?
Ecologic correlation
In the figure, infant mortality rates and average number of children per women are calculated for some African countries, which are the units of analysis. The graph portrays the strong positive linear relationship between fertility and infant mortality in sub-Saharan Africa. Countries that tend to have high infant mortality rates tend to have high birth rates.
Recall one of the reasons why you’d conduct an ecological study.
Individual measurements may not be available for a variety of reasons, but group-level data can be obtained.
What is the data that has already been collected and stored in archives (group measurements) typically called?
The group measurements are called aggregate measures and they provide an overall measurement for the level (e.g., group or population) being studied.
Often, ecologic studies are helpful in revealing the context of health—how demographic characteristics and the social environment contribute to morbidity and mortality.
Examples attached.
What are some examples of examples of ecologic studies?
Ecologic analysis has been applied to the study of air pollution by examining the correlation of air pollution with adverse health effects such as mortality. Researchers measure the association between average levels of air pollution within a census tract (or other geographic subdivision) with the average mortality in that census tract.
Ecologic studies have also examined the association between water quality and both stroke and coronary diseases.
These studies have been used as well to examine the possible association between use of agricultural pesticides and childhood cancer incidence.
For example, a total of 7,143 incident cases of invasive cancer diagnosed among children younger than age 15 years were reported to the California Cancer Registry during the years 1988–1994. In this ecologic study, the unit of analysis was census blocks, with average annual pesticide exposure estimated per square mile. The study showed no overall association between pesticide exposure determined by this method and childhood cancer incidence rates. However, a significant increase in childhood leukemia rates was linked to census block groups that had the highest use of a type of pesticide known as propargite.
What is a small, relatively permanent statistical subdivision of a county delineated by a local committee of census data users for the purpose of presenting data?
Census tracts.
Census tracts nest within counties, and their boundaries normally follow visible features, but may follow legal geography boundaries and other nonvisible features in some instances.
These census tracts ideally contain about 4,000 people and 1,600 housing units.
What is a statistical area bounded by visible features, such as streets, roads, streams, and railroad tracks, and by nonvisible boundaries, such as selected property lines and city, township, school district, and county boundaries?
Census block
Another example attached.
What is defined as an erroneous inference that may occur because an association observed between variables on an aggregate level does not necessarily represent or reflect the association that exists at an individual level?
Ecologic (or ecological) fallacy
Review: Advantages and Disadvantages of Ecological Studies
Advantages
- May provide information about the context of health.
- Can be performed when individual-level measurements are not available.
- Can be conducted rapidly and with minimal resources.
Disadvantages
- Ecologic fallacy
- Imprecise measurement of exposure
N/A
What type of study is one in which subjects are DEFINED on the basis of the presence or absence of an outcome of interest?
Case-control study
These studies, since the outcome such as a disease is the criterion for being included in the case group, can examine only a single outcome or a limited set of outcomes.
Case-control studies use a retrospective approach to collect information about exposure to a factor, in which the exposure occurred in the past. One method to determine past exposure is for the investigator to interview cases and controls regarding their exposure history. An advantage of case-control studies is that they can examine many potential exposures, such as exposure to toxic chemicals, use of medications, or adverse lifestyle characteristics. In some variations of the case-control approach, it may be possible to conduct direct measurements of the environment for various types of exposures.
What is a cohort study design were subjects are classified to their exposure to a factor of interest and are then observed over time to document the occurrence of new cases (incidence) of disease or other health events?
Prospective cohort study
At the inception or baseline of a prospective cohort study, participants must be certified as being free from the outcome of interest. For this reason cohort studies are not helpful for researching dis-eases that are uncommon in the population; during the course of the cohort study, only a few cases of the disease may occur.
What is a measure of association used in cohort studies called?
Relative risk (RR), the ratio of the incidence rate of a disease or health outcome in an exposed group to the incidence rate of the disease or condition in a nonexposed group.
As noted previously, an incidence rate may be interpreted as the risk of occurrence of an outcome that is associated with a par-ticular exposure. The RR provides a ratio of two risks—the risk associated with an exposure in comparison with the risk associated with non-exposure.
We may interpret relative risk in a manner that is simi-lar to that of the odds ratio. A relative risk of 1.0 implies that the risk (rate) of disease among the exposed is not different from the risk of disease among the nonexposed. A relative risk greater than 2.0 implies that the risk is more than twice as high among the exposed as among the nonexposed. In other words, there is a positive association between exposure and the outcome under study.
Sometimes a relative risk calculation yields a value that is less than 1.0. If the relative risk is less than 1.0 (and statistically significant), the risk is lower among the exposed group; for example, a relative risk of 0.5 indicates that the exposure of interest is associated with half the risk of disease. This level of risk, i.e., less than 1.0, sometimes is called a protective effect.
What, in a cohort study, refers to the difference between the incidence rate of a disease in the exposed group and the incidence rate in the nonexposed group?
Attributable risk difference
Returning to the calculation example shown in Table 5, the incidence rate (expressed as rate per 1,000) in the exposed group was 28.04 (rounded off) and the incidence rate (expressed as rate per 1,000) in the nonexposed group was 3.32 (rounded off). The attributable risk is the difference between these two incidence rates (28.04 per 1,000 − 3.32 per 1,000) and equals 24.72 per 1,000. This is the incidence rate associated with exposure to the solvent.
What provides an indication of the benefit to the population derived by modifying a risk factor, which is a measure that assesses differences in rates?
Population risk difference
Defined as the difference between the rate of disease in the non-exposed segment of the population and the overall rate in the population.
What is defined as “[a] in which subjects are randomly allocated into groups, usually called test and control groups, to receive or not to receive a preventive or a therapeutic procedure or intervention?
Randomized controlled trial (RCT)
The results are assessed by com-parison of rates of disease, death, recovery, or other appropriate outcome in the study control groups.
In com-parison with observational studies, a randomized controlled trial is considered the most scientifically rigorous study design and to have the highest level of validity for making etiologic inferences; an RCT can control for many of the factors that affect study designs, including assignment of exposures and biases in assessment of study outcomes.
RCTs are limited to a narrow range of applications; they are not as helpful for study-ing the etiology of diseases as are observational designs. For obvious ethical reasons, it is not possible for an investigator to run experiments that determine whether an exposure causes disease in human subjects.
Components of RCT
N/A
