Definitions

Question 1

What is bias

Answer 1

systematic error (rather than a random error or chance) leading to a difference between the comparison groups which leads to an incorrect estimate of the association between the exposure and the risk of disease

Need to be minimised in study design and analysis - more difficult to adjust for in analysis so design of the study is key

Question 2

What is selection bias

Answer 2

systematic differences between the groups

Question 3

Construct validity

Answer 3

Does the test measure the concept that it’s intended to measure?

Question 4

Content validity

Answer 4

Is the test fully representative of what it aims to measure?

Question 5

Face validity

Answer 5

Does the content of the test appear to be suitable to its aims?

Question 6

Criterion validity

Answer 6

Do the results accurately measure the concrete outcome they are designed to measure?

Question 7

Sampling error

Answer 7

In an unbiased study the difference between the sample values and the population value occur due to chance.

Question 8

How do you control confounding

Answer 8

In the design stage: restrictions, matching and randomisation.
In the analysis stage: multivariate analysis, stratification.

Question 9

What are the features of a case in case control study

Answer 9

● A case definition is used
○preferable to use newly diagnosed cases (incident cases) than existing cases (prevalent cases)
○ Risk factors identified may be related to survival of the prevalent cases (which may be related to the degree of exposure) than development of the disease.
○ easier to establish the temporal sequence.
○ all included cases to be based on the same case definition.
○ Case definition should include time, place, person or be based on an existing classification
● Selection of cases is either population based, or hospital based
○ Hospital-based cases may not be representative of the whole population

Question 10

What are the features of a control in case control study

Answer 10

Controls should be from the same population as cases and matched for certain characteristics that aren’t tested
○ Population-based
■ May not recall events as well as patients who are ill
■ Less motivated to participate and less likely to be available in the day
■ Costly and time consuming
○ Hospital-based
■ More likely to recall events before being admitted
■ More cooperative
■ If in the same hospital likely to have similar backgrounds (like area) although this may different depending on catchment areas for different specialties.
■ Ill people are likely to be different to the healthy population e.g. more likely to smoke.
■ Have to ensure not in hospital because of a condition that could be related to the exposure being studied.
tested
○ can have more than 1:1 ratio but after 1:4 no additional power to study

Question 11

what are the advantages of case control study

Answer 11

● Rapid and cheap
● Ideal for rare diseases
● Useful for diseases with long latent periods
● Can examine a large range of exposures
● Good for populations where follow-up is difficult
● Don’t need as many participants and existing records can be used
● Can examine multiple exposures simultaneously

Question 12

what are the disadvantages of case control study

Answer 12

● Selection bias - the disease and exposure has already occurred, need to ensure the groups are representative of the wider population and that controls are taken from the same population as the cases
● Temporal relationships difficult to establish → reverse causality
● Recall bias of the exposure or disease
● Observer bias - if no which participants are cases
● Can only look at one outcome
● Poor for rare exposures
● Can’t compare incidence rates
● Risk of misclassification of disease or exposure status

Question 13

how to mitigate bias in a case control study

Answer 13

Use standardized and objective diagnostic criteria for identifying cases.
Ensure that those assessing disease status are blinded to exposure status.
Minimize recall bias by collecting exposure information in a uniform and systematic manner from both cases and controls.
Use standardized questionnaires and clear definitions of exposures.
Ensure that interviewers are blinded to the case or control status during data collection.
Use objective measures of exposure whenever possible (e.g., biomarkers).

Question 14

what is a case -crossover study?

Answer 14

Cases and controls are the same people - but two time periods are compared, one where the disease is about to occur, and another long before the disease (to see if the exposures differed)
Here each case is its own control, but recall bias still present

Question 15

why are case-crossover studies used?

Answer 15

used to investigate the association between transient exposures (events or exposures that are brief and not sustained) and the occurrence of acute outcomes or events, especially when those outcomes are rare

Question 16

what is a nested case-control study

Answer 16

A case-control study nested within a cohort study - a cohort is agreed at the beginning and as cases develop disease a control is found from within the cohort population

Question 17

what are the advantages of a nested case-control study

Answer 17

○ Useful if too expensive to test an entire cohort
○ Avoids selection bias as both cases and controls from same cohort
○ Cost-effective
○ Can avoid recall bias by using data collected before the onset of disease
○ Reduces issues with establishing temporal bias as exposure info collected before disease develops

Question 18

what are the disadvantages of a nested case-control study

Answer 18

○ Need to enroll enough people in the cohort to ensure cases do develop over time and reduce effects of loss to follow up in the disease-free group

Question 19

Critical Appraisal of Case-Control Studies (long answer)

Answer 19

Summary
● Type of study and subject
● Journal, data
● Authors, conflicts of interest, funding
● Ethics

Aim
● Clearly focussed question - use of PICO
● Important and relevant
● What the study adds to the current literature

Population / Setting
● Exclusion criteria

Methods (Internal Validity)
● Design
○ Appropriateness
○ Subjective or objective measurements
○ Validated measurements
○ Reliable measurements
● Sample
○ Size (power calculation?)
○ Allocation
○ Recruitment
○ Loss to follow-up
○ Were cases defined precisely
○ Sufficient number of cases included
○ Selection bias
○ If high non-response is this to do with the sample
● Intervention / exposure
○ Clear, valid, reliable
○ Did the two groups receive the same care/treatment aside from the intervention (i.e. treated equally)
● Outcome
○ Clear, valid, reliable
○ Did measurement method reduce bias
● Analysis
○ Appropriate statistics
○ Were loss to follow ups accounted for or adjusted for - were those lost-to-follow up systematically different
○ Any missing or incomplete data
○ Absolute and relative effects reported
○ Was the follow up long enough
○ Were confounding factors taken into account

Result
● Chance
○ P-values
○ Confidence intervals (precision)
○ Type 1 error, type 2 error
● Bias
○ Selection bias
○ Measurement bias
● Confounding
○ Baseline characteristics - were the groups similar at the start of the study
○ Adjustments in the statistics such as sensitivity analysis
○ Randomisation
● Bottom line results reported including effect size

Generalisability (External Validity)
● Based on population, setting and internal validity
● Are study participants similar to your population

Discussion
● Limitations
● Conclusions justified?

Relevance
● Applicability to public health practice, if observational unlikely to be enough evidence to implement a treatment of policy, does other evidence support it
● Reproducibility - costs, resources, sustainability needed to implement the treatment or intervention

Question 20

what are the features of a cohort study

Answer 20

Observational, analytical
A group of individuals who do not have the outcome of interest are selected and followed up to see who develop the outcome, according to their exposure.
● Prospective or retrospective (exposures and outcomes recorded in past records)
● For common exposures a population sample can be used, limiting selection bias
● For rare exposures, the cohort may be chosen on the basis of exposure e.g. selecting a group from a certain factory where likely to be exposed, and a group from the same factory who is not involved in manual work (risk of the healthy worker effect)
● Groups must be compared to ensure similarity of potential confounding factors
● Outcomes must be measured in the same way between groups

Question 21

what are the uses of cohort study

Answer 21

● Investigate role of lifestyle factors in disease e.g. Framingham Heart study
● Investigate cause of infectious disease based on exposure to a risk factor e.g. food poisoning
● Occupational health - if employees exposed to a risk factor develop disease
● Prognostic studies - associations between an exposure on survival or disease