EMB and Study Design

Question 0

Randomisation

Answer 0

Different from random sampling as both groups are the same in all ways
Breaks confounding
Eg random numbers, block allocation, stat softwear

Question 1

Randomised controlled trials

Answer 1

Evaluate safety and effectiveness of an intervention by comparison with a control group
Randomisation-> breaks confounding
Eg random numbers, block randomisation, stat softwear
Allocation concealment
Blinding

Question 2

Allocation concealment

Answer 2

Masking randomisation process before a patient is entered into the study
Study team don’t know which group next patient will be in
Avoids selection bias

Question 3

Blinding

Answer 3

Concealment after allocation 
Reduces ascertainment bias
Single-> patient
Double-> doctor and patient 
Triple-> outcome assessor as well
Means causation can be assessed

Question 4

Dealing with withdrawal

Answer 4

Causes attrition bias as sample no longer random

Intention to treat or per protocol

Question 5

Intention to treat analysis

Answer 5

Analysed based on initial treatment allocation and not treatment eventually received
Avoids attrition bias and cross over
Doesn’t require observation of compliance status
Includes those who drop out in results
Provides information about the potential effects of treatment policy rather than specific treatment
Missing data-> make assumptions or last value carried forwards

Question 6

Per protocol analysis

Answer 6

Only take results from patients who completed the trial

Effected by attrition bias

Question 7

Cohort study

Answer 7

To measure incidence of disease
Traditionally prospective-> study pop is disease free at beginning of study -> asses temporal relationship
Exposure variables tend to be common
Risk is compared
Multiple diseases and/or exposures can be examined
Time consuming and expensive
Vulnerable to attrition bias, selection bias and confounding
Real world-> causality
Inefficient for rare diseases
Sample then separate in to exposure/non exposure

Question 8

Case control studies

Answer 8

Asses risk factors for rare diseases
Source pop-> cases and controls-> sample of each-> exposure
Select participants based on presence or absence of disease, ideally incidence cases
Odds ratio
Retrospective
Less people need to be studied as you can be sure to recruit enough
Investigation of multiple exposures
Can’t calculate prevalence as not a random sample
Only 1 disease
Vulnerable to selection and recall bias, confounding
Real world

Question 9

Bias

Answer 9

Systematic distortion from the true value

Avoid by-> randomisation, matching, post hoc adjustment

Question 10

Error

Answer 10

Non random variation around the true value
Not all the data
Avoid by large sample and lowering significance level
Minimised not corrected

Question 11

Meta analysis

Answer 11

A systematic review with quanta give evidence synthesis
Examine distribution of evidence
Improve generalisability

Question 12

Best study design to answer question

Answer 12

Therapy/prevention-> rct
Diagnosis-> case control
Cause/risk-> cohort/case control
Prognosis-> cohort

Question 13

Qualitative research

Answer 13

Any research that uses data that do not indicate ordinal values
Focuses on people and processes-> understand human beliefs and interactions
Searching for participants meanings
Detailed description
Explores real world detail
Flexible designs
Inductive reasoning
Useful when little is known or few potential participants
Support quantitative research-> triangulation

Question 14

Discourse analysis

Answer 14

Discourse is central
Not interview bases
Transcript captures all features of interaction
Language explained in terms of construction and function
Analyse written, vocal or sign language or any semiotic event

Question 15

Content analysis

Answer 15

Establish codes then count the number of times they occur in the data
Codes are deductive-> derived from pre existing theory and research
Can be manifest or latent
Analyses writing, images, recordings and cultural artefacts
Qualitative and quantitative

Question 16

Theme analysis

Answer 16

Content analysis but with analysis of meaning as well
Themes identified in data
May be manifest or latent
Systematic analysis of the content themes occur in and the links between them

Question 17

Interpretative phenomenological analysis

Answer 17

Phenomology-> personal experience is central-> study of structure of experience and conciousness
Hermeneutics-> how we interpret experiences
1) summaries the experiences described by participant
2) think creatively about what this means
3) how did the participant make sense of the experience

Question 18

Grounded theory

Answer 18

Researcher engages simultaneously in data collection and analysis
Constant comparative method
Theory grounded in data
Theoretical sampling-> discovery of theory through analysis of data

Question 19

Mixed research methods

Answer 19

Mixture of qualitative and quantitative
Draw on strengths and minimises the weakness of both
Expansive and creative
Induction-> searching for patterns
Deduction-> testing theories
Abduction-> going beyond original hypothesis to formulate new ideas

Question 20

Critical appraisal of quantitative research

Answer 20

Theoretical framework, method and context made explicit
Sampling described and justified
Detailed fieldwork description
Description and justification for analytical method
Triangulation used 
Method repeatable
Repeated to increase reliability 
Does raw data support analysis