Critically Appraised Topic Flashcards
What is selection bias?
= error is assigning indidivuals to groups leading to differences that may influence the outcome
The subjects are not representative of the population
Selection bias: what is sampling bias?
= selected subjects are not representative of the population
Selection bias: what is volunteer bias?
= the volunteer subjects are not representative of the population
Selection bias: what is non-responder bias?
= the respondants (subjects) are not representative of the population
Selection bias may be a particular problem in which type of observation study?
= cohort studies
What is a cohort study?
= involves finding a sample that do not have the outcome (e.g., lung cancer)
Group them according to exposure
Follow them forward to see who develops the outcome
What is a case control study?
= find a sample that HAS the outcome AND, find a control sample that doesn’t have the outcome
Look to the past to find out who was exposed to the exposure
What is recall bias?
= difference in the accuracy of recollection of study participants, possible due to whether they have the outcome of not
(e.g., patient with mesothelioma may try harder to remember exposure to asbestos than a control patient)
In which type of observational study is recall bias a particular problem?
= case-control study
What is publication bias?
= failure to publish/ include results from valid studies, often because they show a negative or uninteresting result
In which type of studies may publication bias be a problem? (2)
- meta-analysis
- systematic reviews
(studies showing negative results may be excluded)
What is work-up bias (verification bias)?
= this refers to the gold-standard test being preformed more frequently in patients who have already had a positive result from the new test
e.g.,those with a +ve d-dimer may be assessed further using ultrasonography, whereas those with a -ve d-dimer will be assessed with 3-month follow up. Patients who had a DVT, but neg. d-dimer may not have been diagnosed by 3-month follow-up. This risks underestimating the no of false negatives, and thus may overestimate the sensitivity of a new test
What is an expectation bias (pygmalion effect)?
And how is it tackled?
= observers may subconsciously measure or report data in a way that favours the expected study outcome
= avoided by blinding
What the hawthorne effect?
= group changing it’s behaviour due to knowledge that it is being studied
What is procedure bias?
= subjects in differed groups receive different treatment, other than just interventions
(e.g., elderly patients could receive human contact along with intervention, whereas control patients get no extra human contact)
What is length-time bias?
= screening over-represents less-aggressive disease
e.g., aggressive tumours have a shorted asymptomatic (screening) period, so screening is less likely to detect aggressive tumours
What is lead-time bias?
= early diagnosis appears to prolong survival
What is late-look bias?
= gathering information at an inappropriate time
e.g., studying a fatal disease many years later when some patients may have died
Advantages of a cohort study (3)
- best information about causation of disease, as can work out incidence
- able to examine range of outcomes from a particular exposure
- good for rare exposures - can select those exposed to certain things
Disadvantages of a cohort study? (8)
- long follow-up period, expensive & time consuming
- bad for rare outcomes - would require enormous sample size
- bad for long latent periods (have to wait ffor disease to develop)
- misclassified exposure (lying)
- different follow up for exposed/ non-exposed
- outcome assessors not blind to exposre category
- selection bias - subjects not representative of the population as a whole
- confounders not recognised/ addressed
Advantages of case control studies? (3)
- simple/ easy to consuct, does not require long follow up
- good for rare outcomes - can select patients with certain disease
- good for long latent periods - finding patients after disease developed, not waiting after exposure
Disadvantages of a case control study? (5)
- bad for rare exposures - would require an enormous sample size
- controls may not represent the population where the sample is from
- cases don’t represent full disease spectrum (e.g., fatal vs. non-fatal lung cancers)
- recall bias - different reporting of exposure drom different groups
- confounders not recognised/ addressed
What is equipoise?
= the situation where the researchers have no preference between the 2 interventions being studied in the trial
Why is equipoise important?
= important as it is unethical to run a trial when you alread known one intervention is inferior, becasue you are exposing those patients to sub-optimal treatment, and perhaps unnecessary risk of harm
What kind of bias may incomplete randomisation lead to?
= selection buas, where there are significant baseline differnces
Ideal method of randomisation?
= random computer-generated numbers
Is randomisation based on date of presentation, order of presentations, letter of name or DOB adequate methods of randomisation?
Why?
= no,
may allow the allocating clinician to predict which group the patient might be in before deciding to recruit them into the study
What might you do to improve randomisation?
= stratify
What is stratified randomisation?
= with powerful confounders, patients can be first split, or stratified, into different groups before randomisation so there will be the same number of pateints with and without the confounder in each arm of the trial
What is power?
= power is equal to the chance of detected a true difference when there really is one
(80-90% is usually an acceptable power value)
What is an approximately acceptable power?
= 80-90%
Why might a trial be underpowered?
= not enough participants, therefore th real effect (or lack of effect) of a treatment cannot be established from random variation
What kind of error might an unpowered trial lead to?
= type II (false negative)
What is subversion bias?
= refers to the deliberate manipulation of patient recruitment by clinicans in order to influence trial results
(may be done actively, for example shining a light through an envelope to determine allocation group, or subtly, when a clinican decides not to recruit to a trial a patient who they feel might not do well)
Benefits of combining data in a meta-analysis? (3)
- increases size of study sample
- improves precision, and reduces the width of the CIs
- can demonstrate a significant result when none of the trials could do this individually
Meta-analysis: what happens if you combine data that is too heterogenous?
= the analysis can provide data that is meaningless or misleading
How can you limit confounding in a cohort study? (3)
- restriction
- matching & stratification
- multiple cariable regression
What is multiple variable regression?
= co-efficients are established for the confounder groups. Allows for better adjustment
Challenges to measuring exposure in case controls? (3)
- recall bias
- variable exposure
- unavailable data
Which of the following types of studies provides the best evidence to answer a question on DIAGNOSIS?
- cross-sectional analytic study
- cohort study
- case-control study
- meta-analysis
- RCT
- systematic review
- cross-sectional analytic study
Which of the following types of studies provides the best evidence to answer a question on AETIOLOGY? (2)
- cross-sectional analytic study
- cohort study
- case-control study
- meta-analysis
- RCT
- systematic review
- cohort study
- case-control study
Which of the following types of studies provides the best evidence to answer a question on PROGNOSIS?
- cross-sectional analytic study
- cohort study
- case-control study
- meta-analysis
- RCT
- systematic review
- cohort study
Which of the following types of studies provides the best evidence to answer a question on TREATMENT? (2)
- cross-sectional analytic study
- cohort study
- case-control study
- meta-analysis
- RCT
- systematic review
- RCT
(systematic review of RCTs)
Which of the following types of studies provides the best evidence to answer a question on EVALUATION? (2)
- cross-sectional analytic study
- cohort study
- case-control study
- meta-analysis
- RCT
- systematic review
- systematic review
- meta-analysis
What is treatment fidelity?
= how accurately the intervention is reproduced from a protocol or model
What is internal validity?
= how well study was conducted, taking confounders into acount + removign bias
