BIOL 437 Week Eight P.1 (Statistical Methods in Analytical Epidemiology) Flashcards
analytic study
- attempts to answer why and how a health-related event occured
- test specific a priori hypotheses
- comparision group
- observational and experimental
analytic observational studies
- researchers observe relationships between variables
- may be exploratory or analytic
analytic experimental studies
-a portion of the participants are assigned the intervention
types of analytic observation studies
-case-control
-case-crossover
-nested case-control
-cohort
>prospective
>retrospective
observational exploratory
- useful for identifying clues as to cause-effect relationships
- variety of associations examined
observational analytic
- evaluations of associations between exposure
- outcome variables starts with a specific a prior hypothesis
case-control study
-outcome is always identified prior to the exposure
>looking back in time
steps to a case-control study
- Identify cases
- Identify controls
- Investigate whether the cases are more or less likely than controls to have had past experiences, lifestyle behaviours or exposures
selection of cases
- Establish the diagnostic criteria
- Definition of disease
- may consist of new cases (incidence) or all cases (prevelance)
cases come from
- records from public health clinics
- physician offices
- health maintenance organizations
- hospitals
- industrial and government sources
- should be representative of all persons with the disease
selection of cases types
- Sampling
2. Restriction
sampling
-representation requires random selection with a sufficiently large sample size
restriction
- may improve validity (advantage)
- may limit generalization (disadvantage)
control subjects should
-look like case subjects, with the exception of not having the disease
-representative of the general population
>same possibility of being selected or exposed
selection of controls places
- general population
- hospital
- family, friends, relative
hospital advantages
- easily identified
- more likely to cooperate
- more likely to be aware of exposures
hospital disadvantage
-don’t accurately represent the exposure distribution
general populaiton advantage
-represent population
general population disadvantages
- more costly and time consuming
- may be difficult to contact
- may have poorer recall
- less motivated to participate
special groups advantages
- healthier than hospitals
- more likely to cooperate
- provide more control
special groups disadvantages
-if exposure is similar, an underestimation of true association could result
info about exposure status may be obtained
- medical records
- interviews
- questionnaries
- surrogates (spouses, sibilings, employees)
time window for determining exposure
- important
- influenced by understanding of potential causal factors associated with the disease
appropriate measure of association
- depends on nautre of data
- when dichotomous:
1. Odds ratio (OR) (relative odds): case-control
2. Risk ratio (relative risk): cohort
3. Rate ratio: cohort
2x2 contingency table
-used to summarize relationship between exposure and health outcomes
OR=1
-no association between exposure and outcome
OR>1
-positive association between exposure and outcome
OR<1
-negative association between exposure and outcome
OR range
- from 0 to infinity
- mathematical properties allows calculation of the OR using logistic regression
bias
-systematic error in the collection or intrepretation of epidemiological data
bias results in
-inaccurate (over or under) estimation of the association between exposure and disease
avoid bias at the design stage
-paramount because of the difficulty identifying and accounting for it later
types of bias in case-control studies
- Selection bias
- Observation
>recall bias
>interview bias - Misclassification
- Confounding
selection bias and observation
-will always result in mis-classificaiton
selection bias
- cases and controls based on exposure
- relationship among exposure and disease in participants differ
- recruiting all cases in population avoids it
types of selection bias
- Berkson’s bias
2. Prevelance-incidence bias
Berkson’s bias
-hospital-patient bias
-controls from hospital may be more likely to have confounding conditions
-minimize by doing randomization
>not possible in observational analytic studies
recall bias
-differential accuray of recal between cases and controls
interviewer bias
- interviewer probes cases differently than controls
- interviewers must be blinded
