EBP Y1 Flashcards
What are randomised control trials good for?
Answering treatment and diagnosis questions.
They are considered the most reliable evidence when investigating the effect of an intervention.
?: volunteers are recruited from a sample of population, assesed for eligability and randomly assigned to groups. Investigation attempt to completely control the exposure.
What does PICOt stand for?
P: Patient/population/problem
I: intervention
C: comparison/control
O: outcome
When are randomised control trials usually used?
commonly used to verify the efficacy of a new treatment. EG: drug trials
What are the key features of a randomised control trial?
- Randomly allocated
- Always a control group
What are the issues with randomised control trials?
- Expensive
- Appropreate: not always the right study design to answer the question
- Ethical: is it ethihcal to randomly assign treatment?
What is the major difference between controlled and uncontrolled studies?
uncontrolled: all participants given the same new treatment no directly comparable group
cannot compare effects with any other group who didn’t recieve treatment
Controlled: presence of a comparison group. Allows for potential bias due to external influences during the study
What does concealment mean?
Sequence of treatment allocation is unknown, helps reduce posibility of interence with group allocations to maintain randomisation.
What does blinding mean?
Where the actual treatment allocation is unknown. Instead of groups placebo and drug its drug A and drug B
There is single blind trial and double blind trial
What is a parallel group trial?
Comparison of two treatments, random allocation. Can be two completely separate treatments or different doeses of the same.
What is a cross over trial?
What are teh advantages and disadvantages of this?
Patients recieve a sequence of at least two treatments during different time periods. They cross over treatments to another during the trial.
More balanced, participants recive same number of treatments, each paticipant will recive all treatments.
Advantages: Each participant is thier own matched control so it requires fewer participants, reduction in variability.
Disadvantages: Potential for carryover effect, response to previous treatment effectng response for current treatment, this causes bias.
washout period is the time between treatment periods.
What can be done about non-compliance?
- per-protocol : exclude data from the individual who didn’t comply
- As-Treated: reassign the individual to the control group
- Intention-to-treat: keep to individual in thier original randomised group
What are the two conditions of intention to treat?
- all randomised participants should be included in the analysis
- All participants are retained in the group to which they were randomised.
considered misrepresentitive of the data
What are the different types of bias?
- Selection: systematic differences in the baseline characteristics between treatment groups
- Performance: differneces in care or other exposures between treatment in groups than teh intervention of interest.
- Attrition bias: difference in the withdrawals between treatment groups
- Detection or assessor bias: difference between treatment groups and how outcomes are measured/determined
- Reporting bias: difference between the reported and unreported findings.
What is the ideal study design?
- randomisation of pateints
- a concurrent control group
- double-blinding of pateints and assesors
What is the number needed to treat?
The average number of pateints who need to be treated to prevent an additional negative outcome
if NNT = 1 : everyone from intervention group has improved
NNT = 1/ absolute risk reduction
If the outcome of a trial is continuous how is the difference expressed?
Difference in means
confidence intervals
strength of evidence “p”
What is the number needed to harm?
represents the number of pateints who would need to take the intervention for one pateint to experience a harmful event of interest. This number should be as high as possible
NNH = 1/ absolute risk increase
critical appraisal
RAMMbo
what is on the xfactor of making evidence based decisions?
- epidemiological evidence
- system features: economic legal, political, public health
- patient clinical circumstances: physical health, social, psycological
- values and preferences: person, family, community, practitioner
What increases the strength of the evidence?
The more its reviewed and summarised. IE not the original studies
what are the types of questions?
Background: what is hypertension? How is blood pressure regulated?
Foreground: What happens when pateints with hypertension don’t get treatment?
What are the benefits home home monitoring blood pressure?
State the clinical evidence pyramid from bottom up.
- Animal laboratory studies
- Case series and case reports
- Case control studies
- Cohort studies
- Randomised controlled trials
- Systematic reviews
- Meta-analysis
How do you formulate a clinical question?
In [Population], what is the effect of [intervention] on [Outcome], compared with [Comparison]
What are the descriptive studies?
- Qualitative
- Case report/series
- Survey (cross sectional)
What are the two types of analytical studies?
Observational or experiemental
What are the types of experiemntal studies?
Randomised parallel groups, randomised crossover or quasi-experimental
What are the types of observational studies?
cross sectional, case control, cohort
What is an observational case control study?
case study starts by looking at those with cases and those who don’t have the cases, then traces back to see who was exposed to certain things and who wasn’t.
What is an observational prospective cohort?
Look at a group of people ask if they have been exposed or not first then out of those groups who go the cases and who didn’t.
What is a retrospective observational study?
Same as prospective but instead of following the cases back to the start you’re there from the start watching it happen.
What is a parallel experimental study?
Two randomly assigned groups are either experimental or control and then an assesment of outcome is made after study.
What is an experiemental cross over study?
Where groups randomly assigned to group A or B with different intervention. There is then an assesment then a wash out period then they switch. Can be compared to themselves which is great.
What is an experimental quasi experiement?
where two groups are pre-assesed then exposed to intervention or control, then post assesed.
What is the diagnositic cut off?
Where the cut off point is placed will affect the number of false positives and false negative test. This will depend on the seriousness of the disease as well.
What is a critical appraisal checklist?
questions to help make sense of a diagnostic test study.
Three broad areas are: are the results of the study valid? what are the results? will the results help locally?
What is sensitivity?
refers to the tests ability to correctly identify the people with the disease, the true positives.
How do you work out sensitivity?
sensitivity = No. of people with the disease AND a positive test result / The total number of people with the diease.
What is specificity?
Refers to the preportion of people without the disease who will have a negative result.
How do you calaculate specificity?
Specificity = No. of people without the diease AND a negative result / The total number of people without the disease.
what is the helpful acronym for sensitivity and specificity?
SnNOUT = High sensitivity, negative results rule out
SpPIn = High specificity, positive tests rule in
What will effect sensitivity and specificity?
The cut off point
What is prevalence?
prevalance = total number of people with the disease/ total number of people in the population.
What is a positive predicitive value?
PPV = number of people with the disease and positive result / total number of people with positive test results
What is a negative predicitive value?
NPV = number of people without the disease / total number of people with a negative result
How do you calculate a positive likelyhood ratio?
LR+ = sensitivity / 1 - specificity
LR+ <1 means that a positive test is more likely in a person with the disease than without.
LR+ > 1 means that a positive test result is less likely in a person with the disease than without.
How do you calaculate a negative likelyhood ratio?
LR- = 1 - Sensitivity / specificity
LR- < 1 means that a neagtive test result is more likely in people without disease than with.
LR- >1 means that a negative test result is less likely in people without the diease.
what does estimating probability of disease look like?
