300 likes | 440 Views
What should we expect?. Choice of a control group FETP India. Competency to be gained from this lecture. Recruit a control group that will match the objectives of a case control study. Key areas. Study base Exposure probability Possibility to be a case
E N D
What should we expect? Choice of a control group FETP India
Competency to be gained from this lecture Recruit a control group that will match the objectives of a case control study
Key areas • Study base • Exposure probability • Possibility to be a case • Possibility to be recruited as a case • Time comparability • Addressing the objectives of the study
Need of a control group: Rationale • The frequency of exposure among cases is known • An exposure may be common among cases • The question becomes: • Is this exposure common in the population anyway? • Is this exposure specific of the cases? • The control group indicates what can be expected in the population in the absence of disease
Unmatched control group Cases Controls Bag of cases Bag of controls
Matched control group Cases Controls Sets of cases and controls that cannot be dissociated
1. Cases and controls must come from the same population • Cases and controls must be similar with respect to: • Demographic characteristics • Cultural background • Socio-economic group • Employment • Failure to recruit cases and controls from the same population lead to a bias Study base
Checking the comparability of cases and controls • Information must be collected to describe the basic characteristics of cases and controls • The first step of the analysis is to compare cases and controls with respect to these basic characteristics Study base
The study base is a concept referring to the study population • Real study base • Well defined population • Virtual study base(Hospital-bases studies) • Two options: • Hospital-based cases and controls • Hospital-based cases with community controls
Real study base • Cases and controls come from a population defined in terms of time, place and person • Example: • The employees of a company • The residents of a village • Key issue: • Find cases representative of all cases • Case ascertainment Cases Study base Controls
Virtual study base • The study base is defined by a concept • Example: • The persons who would seek health care in hospital X in case of a life threatening condition • Key issue: • Ensure that controls belong to the study base Study base
Hospital-based cases and controls The hospital The residence of the study participants Cases Controls Study base
Hospital-based cases, population based controls The residence of the study participants The hospital Cases Controls Study base
Example of a set of controls belonging to the same study base • Hospital-based study examining risk factors for cholera during an outbreak • Controls: • Patients admitted with meningitis into the same hospital • The two diseases are of equivalent severity • The population bases can be expected to be identical Study base
Example of a set of controls not belonging to the same study base • Hospital-based study examining risk factors for cholera during an outbreak • Controls: • Patients admitted with minor complaints at the outpatient clinic in the same hospital • The two diseases are not of equivalent severity • The population bases can be expected to be different • Catchment areas will be larger for more severe diseases Study base
2. Controls should have had an opportunity to be exposed comparable to cases • Controls are recruited to provide an indication of what happens in the general population • The possibility of exposure: • Should not be higher • Should not be lower Exposure probability
Examples of controls with higher probability of exposure • Case control study exploring risk factors for cholera • Including controls with typhoid: • Increases the proportion of exposed controls • Biases the odds ratio away from 1 Exposure probability
Example of controls with no probability of exposure (1) • Case control study exploring risk factors for gastro-enteritis on a cruise ship • Including controls who did not eat the contaminated meal: • Increases the proportion of unexposed controls • Biases the odds ratio away from 1 Exposure probability
Progressive restriction:The “cruise ship” approach • Recruit cases and controls • Ask about: • Attendance at specific meals • What was eaten at the meal attended • Run two analyses: • First analysis to identify the contaminating meal • Second analysis to identify the contaminated food among cases and controls who attended the contaminated meal Exposure probability
Examples of controls with no probability of exposure (2) • Case control study exploring risk factors for microsporidiosis among HIV infected patients • Men having sex with other men may be at higher risk • This exposure can only be examined in an analysis restricted to men • Women have a 0% probability to be men who have sex with men • Restriction for this variable only Exposure probability
3. Controls must have the theoretical possibility to develop the disease • Control-subjects need to have the possibility to acquire the disease being studied • If they cannot acquire the disease they could have been exposed just as much as the cases without becoming sick Becoming a case
Examples of controls who cannot acquire the disease • Men as control-subjects for case-patients with ovarian cancer • Control subjects immune to the infectious disease considered Becoming a case
4. Controls should be able to be recruited among cases if they were to develop the disease • Control-subjects can develop the disease • Control-subjects can then be recruited as cases in the case control study • Check if your controls could become cases: • Appropriate timing • Appropriate location • Appropriate person Becoming a case
5. Cases and controls are examined for the same referent exposure period • Cases were exposed during the referent exposure period • The referent exposure period must be identical for cases and controls • Same duration • Same time frame • If the effect of the exposure was time-dependent, some controls could report exposure but at a time when exposure was not associated with illness Time comparability
Time comparability: Example • Case control study examining risk factors for septicaemia in a hospital • Central venous line is a suspected risk factor • If an infected health care worker is a source of infection, the effect of exposure (central line) will vary over time (presence or not of the infected health care worker) Time comparability
6. The control group must be adapted to the objectives of the study • Research questions progress as an onion-peeling process • If some risk factors are already well identified: • Make your controls identical to your cases for these risk factors (matching) • The choice of controls must optimize the chances of finding an association if it exists Controls adapted to the objectives
Number of cases and controls • Does not need to be one to one • In an outbreak, it is not necessary to include all cases • Needs to be addressed as part of the sample size estimation • Case-to-control ratio flexible • If controls are easier to find than cases:Recruit more controls than cases • If cases are easier to find than controls:Recruit more cases than controls Controls adapted to the objectives
Relevance of recruiting multiple controls per case • Multiple controls per case can increase statistical power • The increase of statistical power has a plateau effect after 3 controls per case Controls adapted to the objectives
Dealing with imperfect control groups • Examine the limitations of your control group with respect to each criteria • Assess in which way the limitation will affect the odds ratio • Towards one • Away from one • Interpret your results in light of this review of limitations • Recruiting two control groups is an option Controls adapted to the objective
Characteristics of good controls • Come from the same population as cases • May be exposed like cases • Can develop the disease • Could be recruited as cases if diseased • Have exposure windows identical to cases • Are adapted to the study objectives