- case control
o compares 2 groups of people
§ with disease or condition (cases)
§ similar group of people who do not have disease / condition
o medical and lifestyle history
o purpose to establish association
o observational
- design
o subjects selected on basis of whether or not they have disease of interest
o cases compared to controls in terms of their exposure to hypothesized causal factor
- major concepts
o selection of cases and controls
· diagnostic criteria for defining case must be explicit
· should be newly diagnosed
§ controls
· ideally, should be representative of general population without disease
o random sample from same population that gave rise to cases
· to avoid confounding variables, controls often matched
o pair matching
o frequency matching.
§ sometimes à cases and controls selected from within cohort study
· nested case control
§ potential bias
· systematic – over or under representation of one or more cells
o mistaken estimate of relationship between exposure and disease.
· selection bias
· information bias
o inaccurate classification of study subjects with regard to disease / control status.
§ differential
· misclassification different in cases and control
§ non-differential
· amount & direction of misclassification is same for cases and controls.
o causes of wrong classification
§ inaccurate recall
§ inaccurate reporting
§ interviewer effect
§ inaccurate measurement of disease etc
§ inaccurate diagnosis
o Analysing case control studies
§ comparison of case and control
· frequency of exposure to hypothesized causal
· RR is estimated by Odds Ratio
o how much more likely it is that cases were exposed to risk factor and the controls were not exposed.
§ if exposure is associated with disease, would expect the odds of exposure to be higher for cases than for controls.
· Odds Ratio
o comparing whether the probability of a certain event is same for two groups
o not possible to determine risk of disease amongst subjects exposed to particular factor in comparison to those not exposed.
§ for large sample and rare disease à RR and OR approx the same
o defined as
§ ratio of odds of development of disease in exposed person to odds of development of disease in non-exposed persons.
o range from 0 to some positive number
o Advantages
§ cheap
§ valuable for studying rare / uncommon conditions
§ fewer subjects required than in cross sectional studies
§ odds ratio can be calculated à good approx of relative risk
o Disadvantages
§ usually inefficient if exposure is rare
§ do not establish time sequence
§ selection of control group can be difficult
§ prone to potential bias as compared to other designs.
No comments:
Post a Comment