CAT Flashcards
What is heterogeneity?
Means variability in data - estimates how effective Tx different between studies
Clinical - differences in participants, interventions or outcomes
Methodological - differences in study design, risk of bias
Statistical - variation in intervention effects or results
Forest plot - overlapping CI’s
I^2 - 0% to 40%: might not be important
30% to 60%: moderate heterogeneity
50% to 90%: substantial heterogeneity
75% to 100%: considerable heterogeneity
Can’t do meta-analysis if heterogeneity is too high
Explore heterogeneity - done by subgroup analysis or meta-regression
Causes of publication bias?
Failure to publish/include results from valid studies, often because they show a negative or uninteresting result
Important in meta-analyses and SRs where studies showing negative results may be excluded
E.g., abstract not included
What is intention to treat analysis?
Analysis based on the initial treatment intended from allocation, not the treatment eventually given (e.g., if patient dropped out)
Keep individual in group, even if not treated
What is concealment of allocation?
The person randomising the patients does not know what the next treatment allocation will be, preventing selection bias
Prevent prediction
What is blinding?
Where some of the participants of trial (patient/clinicians/researchers) are prevented from knowing certain information that may lead to conscious or subconscious bias
Prevent conscious/unconscious bias
What is meta analysis?
Research process used to systemically synthesis or merge findings of single, independent studies, using stat. methods to calculate an overall or ‘absolute’ effect
Cohort studies are prone to…
Dropouts and non-completion > lack of follow > bias
Benefits of using medical records?
Objective information
Problems of rating scales
Tendency to overdiagnose mood disorders
How do studies try to tackle confounding
Excluding pt’s from sample
Measuring factors and adjusting
What is Zelens procedure
participants are randomly allocated and then approached and offered the group to which they were allocated
used firstly, to reduce disappointment bias in the conventional consent-randomization process, and secondly, to remove subjective bias in the recruitment process.
What is bias?
Systematic error in measurement
Methods of randomisation
Simple - allocated 1:1 or within each block of 6
Stratified - allocate randomly, often testing characteristics (i.e., severity of condition)
Minimisation - uses a computer algorithm to create balance in groups, according to characteristics
Why do SR?
single study may not provide definitive answer
combine single studies and may provide new evidence (may be unethical to do new primary research)
promotive a scientific (not subjective) approach to summarising information
reduce inaccuracies from bias and enhance replication
provide reliable and ‘speedier’ evidence
more statistical power
What is confounding?
variable that has relationship between both the exposure and outcome
How can confounding be addressed in design + analysis?
restriction (to entry of study)
matching (participants to study)
stratification (according to variable)
multiple variable regression (predict one variable from another)
What can go wrong in cohort study?
misclassification of exposure
differential follow-up between exposed/non-exposed
outcome assessment not blind to exposure category
failure to recognise and address possibility of confounding
What is treatment fidelity?
how accurately the intervention is reproduced from a protocol or model?
What is internal validity?
accuracy - how well the study was conducted, taking confounders into account and removing bias
What is external validity?
generalisability - how well the study can be applied to different scenarios
What is selection bias?
Error in assigning individuals to groups leading to differences that may influence the outcome. The subjects are not representative of the population
Sampling bias - selected subjects are not representative of the population
Volunteer bias - the volunteer subjects are not representative of the population
Non-responder bias - the respondents are not representative of the population
Particular problem in cohort studies
What is recall bias?
Difference in the accuracy of recollection of study participants, possibly due to whether they have the outcome or not
Problem in case-control studies
What is publication bias?
Failure to publish/include results from valid studies, often because they show a negative or uninteresting result
Problem in meta-analyses and SRs
What is work up bias?
Verification bias
Seen in studies trying to validate a new diagnostic test
Refers to the gold-standard test being performed more frequently in patients who already had a positive result from the new test
What is expectation bias?
Pygmalion effect
Only a problem in non-blinded trials
Observers may subconsciously measure or report data in a way that favours the expected study outcome
What is Hawthorne effect?
Group changing its behaviour due to the knowledge that it is being studied
What is procedure bias?
Subjects in different groups receive different treatment, others than just the interventions
What is length time bias
Screen over represents less-aggressive disease
e.g., aggressive tumours have a shorter asymptomatic period, so screening less likely to detect aggressive tumours, but more likely to detect slow-growing tumours with long asymptomatic periods. Slow growing tumour has greater survival, so appears screening improves survival
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 fatal disease many years alter when some patients may have died
Advantages of cohort studies
Best information about causation of disease, can work out incidences
Able to examine a range of outcomes from a particular exposure
Good for rare exposures - can select those exposed to certain things
Advantages of case control study
Simple/easy to conduct, do not require long follow up as outcome is already present
Good for rare outcomes - can select all patients with certain disease
Good for long latent periods - finding patients after disease developed, not waiting for it after exposure
Disadvantages of case control study
Bad for rare exposures - would require enormous sample size
Controls may not represent the population where the sample is from
Cases don’t represent full disease spectrum
Recall bias - different reporting of exposure from different groups
Confounders - not recognised/addressed
What is equipoise?
Where the researchers have no preference between the two interventions
Underpowered trials lead to…
insufficient no of participants so real effect cannot be established from random variation
Causes T2 statistical error - accepting null hypothesis when it is false
Why combine data in meta analysis
^ number of patients being analysed (increases sample size)
Improves precision and reduces the width of CIs
Can demonstrate a statistically significant result when none of the trials could do this individually
Challenges to measuring exposure in case controls
Recall bias
Variable exposure - patients environment may have changed (e.g., moved house)
Unavailable data - pt. can’t remember, medical records unavailable, incomplete, inaccurate
Limiting confounding in cohort study
Restriction - limit participants
Matching and stratification - make comparison groups (with and without confounder) - used for things like age and sex
Multiple variable regression - coefficients are established for the confounder groups, allows for better adjustment
What is likelihood ratio for a positive test result
How much the odds of the disease increase when a test is positive
What is likelihood ratio for a negative test result
How much the odds of the disease decrease when a test is negative
What is PPV
The chance that the patient has the disease if the test is positive
What is NPV
Chance the patient does not have the disease if the test is negative
What is sensitivity
Proportion of patients with the disease who get a positive test result
What is specificity
Proportion of patients without the disease who get a negative test result
