Glossary of Epidemiology: 15) Control Selection - Secondary Study Base

Start Studying! Add Cards ↓

Major problem for controls in secondary study base
If the Probability of being a control differs for the E+ and E- groups, then the measures of effect will be biased (SELECTION BIAS)
4 common methods for control selection in a seconday study base
1)population lists. 2)neighborhood controls. 3)Random-digit dialing. 4)Hospital controls
Control selection and population lists
If lost participants from the population list then there is SELECTION BIAS. Eg. (of 1330 eligible controls only 998 participated)
Neighborhood controls
Study base is divided into mutually exclusive neighborhoods where all cases are from. Controls are randomly sampled from the neighborhood of each case at the time disease is diagnosed.
When best to use neighborhood sample
1) recent or current disease identification. 2) Sufficient exposure variation within neightborhoods.
Disadvantages of Neighborhood samples
Lack of exposure variation loses statistical efficiency
Random Digit Dialing
Random numbers likely within secondary study base. Strict procedures for qualifying controls. Keep track of tries to get control
Problems with random dialing
1) Subjects without phones. 2) Correspondence between telephone numbers and persons. 3) Nonresponse.
Hospital controls - timing
Cases are identified continuously in real-time so controls can be selected from patients in the hospital at the same time as cases.
Identifying hospital controls
Disease should NOT have the probability of admission related to E. To reduce this some use several diseases
Biggest mistake for hospital controls
Using the same organ system to "make controls like the cases"
When and How to initiating procedures for hospital controls
Start when a case is identified, have a defined procedure, keep track of tries.
Hospital controls and index date
Tx is set at oset of the symptoms for the disease and a comparable time for the controls
4 advantages for hospital controls
1) Convenience. 2) High participation rates (especially for acute diseases). 3) Controls in hospital likely to have same source of information bias (have the same data). 4)Same access to medical care.
2 ways hospital controls are valid
1) Treatment and referral patters are same for control and case diseases (often in network of hospitals). 2) Control diseases not associated with exposure.
3 Althernative control selection methods
1) Medical practice controls. 2)Friend or relative control. 3) Controls outside study base.
Medical practice controls
Managed care patients. Can lead to bias (eg. caffeine and colon cancer)
Advantages of Friend/Relative control
1) Convenient. 2) Possibly Reduce social class bias
Disdavantages of Friend/Relative control
1) Does not meet study base principle. 2) Differential selection related to exposure is likely
Controls from ouside study base
Comparable to using historical controls in CT or external E-
Multiple control groups
Try to reduce bias. Usually a dark alley (what if results differ between the groups!)

Add Cards

You must Login or Register to add cards