Analytic Epidemiology: types of study designs
information about both exposures (explanatory variables) and outcomes is collected.
use group rather than individual because indiv. measurements might not be available but group ones are (archives, )
Common outcome variables: mortality, cause specific mortality, types of morbidity, etc.
ex. introduction of a preventive or therapeutic regimen or an intervention designed to test a hypothesized relationship
randomized control trails
Birth or age cohort (baby boom generation)
work cohort (employers studied for occupational exposures)
the investigator does not have control over the exposure factor.
the experimenter is not able to randomly assign subjects
control group = do not have the disease
can only examine a single outcome or a limited set of outcomes
experimenter interviews cases and controls about past exposures
1. can be used to study low prevalence conditions
2. relatively quick and easy to complete
3. usually inexpensive
4. involve smaller number of subjects
1. measurement of exposure may be inaccurate
2. representativeness of cases and controls may be unknown
3. provide indirect estimates of risk
4. the temporal relationship between exposure factor and outcome cannot always be ascertained
Permit direct observation of risk
expose factor is well defined
can study exposures that are uncommon in the population
the temporal relationship between factor and outcome is known
expensive and time consuming
complicated and difficult to carry out
subjects may be lost to followup during the case of the study
exposures can be misclassified
may provide information about the context of health
can be performed when indv. measurements not available
can be conducted rapidly and with minimal resources
the ecologic fallacy
imprecise measurement of exposure
2. How many observations are made?
3. what is the directionality of exposure?
4. What are the methods of data collection?
5. What is the timing of data collection
6. What is the unit of observation?
7. How available are the study subjects?
Sometimes more observations are made. Cohort studies and experimental studies
investigator starts out with someone that already has the disease and asks them about exposures that may have led to disease
Ecologic Studies use existing data
in other studies, subjects may be followed (into the future) over a period of time. If subjects drop out of the study the outcome variable may be lost.
ex. sex, socioeconomic level, age, income inequality, race, prevalence of physicians, unemployment, etc
(AD)/(BC) = OR
OR of more than one suggests a positive association
when it is 1.0 there is no association
the ratio of the incidence rate (of the disease or health outcome) in an exposed group to the incidence rate (of the disease or health outcome) in a non exposed group.
incidence = risk of occurrence of an outcome
PRD = incidence in the total population – incidence in the non-exposed segment
community trails are expensive, complex and time consuming
some studies may select subjects by taking a sample of convenience – or by using random samples +
ex. subjects need to be randomized, appropriate and reliable measurements need to be taken.
B- no disease, exposed
C- with disease not exposed
D- no disease, not exposed