Session 4 Lecture 1: Critical appraisal of research evidence: appraisal of specific designs

Question 1

What are we looking for when deciding if the study is a case control study ?

Answer 1

• the cases and controls are selected based on outcome which is always single
• cases can be incident or prevalent cases
• retrospective design: having individuals recall their exposure
• outcome measure is an odds ratio: null value = 1

Question 2

When appraising a case control study, what are we most concerned about ?

Answer 2

• Selection bias
• Recall bias
• Misclassification bias
• Interviewer/observer bias
• Confounding

Question 3

State 3 of the key questions we should be asking when we read a case control study ?

Answer 3

• how are the controls defines and how are they selected ?
• what do we know about response rate and drop out rate for cases and controls ?
• do the cases represent the population of interest ?

Question 4

What are we looking for when deciding if the study is a cohort study ?

Answer 4

• participants selected on exposure status
• a longitudinal prospecting design
• allows for Temporal sequence to be determined
• multiple exposure and outcome can be measured (compared to case control where only a single outcome and exposure is measure)
• outcome measure can be odds ratio, risk ratio and incidence rate ratio (taking into account time)

Question 5

When appraising a cohort study , what are we most concerned about ?

Answer 5

• selection bias
• response bias
• measurement bias
• healthy worker effect
• loss to follow up/ Attrition bias
• confounding

Question 6

State 5 of the key question we should be asking when we read a cohort study ?

Answer 6

• can we be sure that the participants were outcome free at that start of the study (baseline)?
• was the follow up period long enough for the outcome to have occurred ?
• was follow up of participants adequate ? How may participants were lost to follow up?

Question 7

What should we be looking for when deciding if the study is a cross sectional study?

Answer 7

• snapshot in time so can’t determine temporal sequence
• determines the burden of disease in a population, e.g prevalence
• can be descriptive or analytical
• no long follow up period

Question 8

When appraising a cross sectional study , what are we most concerned about ?

Answer 8

• selection bias and how representative the sample is of the population
• response bias
• measurement bias
• confounding

Question 9

State the 2 of the key questions we should be asking when we read a cross sectional study ?

Answer 9

• What do we know about non respondent in the study ?

- how has the data been collected ?

Question 10

State the specific issues looked out for when appraising ecological studies

Answer 10

• population should be clearly defined
• sources of information must be well described and must also be reliable, accurate, complete and allow for meaningful comparison
• confounding issues
• ecological fallacy

Question 11

State 5 of the key questions that needs to be asked when reading any of the studies

Answer 11

• how representative is the sample?
• how has the data been collected ? Tools used; validated or reliable tools?
• is confounding discussed?
• what is the key finding and how precise is it?
• how applicable are the results to your population ?
• how does the study fit with other research ?