Biostats Flashcards
3 types of analysis in randomised consent trials.
- Per-protocol analysis
- Intention-to-treat analysis
- As-treated analysis
What does Per-protocol analysis mean.
Only individuals that completed the intervention/remained compliant/remained in their original group to which they were randomly assigned are included in analysis.
All others are discounted/ not inlcuded in analysis.
What those “As treated” analysis mean.
Analysis of individuals is based in terms of the intervention/ treatment they actually received, even if it was not what they were originally randomised to.
What does “intention to treat” analysis mean.
Individuals are analysed within the group to which they were intended to be i.e. the group to which were assigned initially- regardless of whether or not they remained within that group/ received the intended intervention/treatment.
It in a sense ignored non-compliance.
2 disadvantaged of Per-protocol
- Can overestimate treatments effects e.g. people who stopped treatment because they did not see any benefit are not included.
- Can misrepresent the “true” study results.
Advantages/disadvantages of intention to treat
- Advantages:
o Preserves intent of randomisation.
o It is more reflective of real-life clinical practice as not everyone adheres to treatment- therefore it can be seen as representative of “best case scenario” in terms of a treatment regimen.
o It eliminates bias caused by non-completion. - Disadvantages:
o It can result in underestimation of true efficacy/benefits associated with a treatment by not analysing data pertaining to the actual treatment received; can dilute treatment effect
advantages of as-treated
- Commonly used to assess side effects to interventions
o More accurately represents side effects to specific intervention by considering the actual treatment received.
o Can allow for exploration of dose-response relationships.
o Can better assess safety profile of intervention as can assess rare or unexpected side effects to intervention
What is meant by a cluster-randomised trial
- Clusters of people rather than individuals are randomised to an intervention and control group and outcomes are measured on individuals within those clusters.
o E.g. patients in a certain clinic, pupils in a certain school.
3 reasons why CRTs might be used
o Studying interventions that affects groups rather than individuals.
o When interventions are given to healthcare professionals with the aim of impact of it on patient outcome. E.g. education of HCPs, different working hours, longer breaks etc.
o May be cheaper and easier to organise clusters rather than individuals.
3 factors to consider when choosing CRT method
- Similar Outcomes within Clusters: When people are grouped together in clusters (like schools or neighborhoods), they tend to be more alike in their outcomes. For example, students in the same school might have similar test scores. This means we can’t assume their outcomes are independent, or completely separate from each other.
- Standard Approaches Don’t Work: The usual methods we use to plan and analyze studies might not be accurate when dealing with clustered data. We need special methods to account for the fact that outcomes within a cluster are related to each other.
- Ethical Considerations: When doing research with clusters, we have to think about more than just individual consent. We might need to involve families or communities in the decision-making process because interventions could affect them as a group.
Define the 2 types of reliability
Internal reliability: Refers to the degree to which a measurement tool is actually measuring the construct it was set out to measure. i.e. are all the items on a questionnaire suitably assessing for the same thing: the consistency within the questionnaire.
External reliability: Refers to whether a measurement tool/ study can achieve the same results if done again under the same conditions.
What is meant by reproducibility?
Reproducibility refers to the ability to obtain consistent results when measurements or experiments are repeated under changing conditions. These changing conditions may include different measurement methods or instruments being used, or measurements being made by different observers
What is meant by validity? List the 4 types
Validity refers to whether or not a tool is measuring what it is meant to measure.
- Face validity
- Construct validity
- Content validity
- Criterion validity
Distinguish between the 4 types of validity
Face Validity: whether a measurement or test appears, on the surface, to measure what it is supposed to measure. It’s like asking, “Does this test look like it’s measuring what it claims to measure?”
Construct Validity: whether a test actually measures the theoretical construct or concept it claims to measure. It’s like checking if a test accurately captures the idea or theory it’s supposed to represent. For example, if a test claims to measure intelligence, construct validity would involve examining whether it truly measures intelligence and not something else.
Content Validity: the extent to which a test covers all the aspects of the concept or construct it’s supposed to measure.
Criterion Validity: This involves comparing the results of a test to some external criterion or standard. It’s like checking if the scores on a test correlate with some other measure that is already established as a valid measure of the same concept.
List 3 types of bias and outline what each means.
Selection, recall and non-response bias.
- Selection bias means that there is a systematic exclusion or over-representation of a certain cohort in the way that participants are selected.
- Recall bias means that there are inaccuracies in participants’ responses to questions which do not reflect the truth.
- Non-response bias occurs when those who agree to participate differ to those who refuse in relation to the characteristic being studied.
