EBP Flashcards
Five stages of evidence based practice
1 .convert the need for info in an answerable question
2 .find the best evidence to answer this question
3 . Critically appraise the evidence
4 . Apply evidence to the patient
5 . Evaluate your effectiveness and efficacy at step 1-4
Definition of EBP
EBP is the conscientious (careful) explicit and judicious use of current best evidence in helping individual patients make decision about their care in the light of their personal values and beliefs
What information can be found in guidelines
Usually based on 1 disease and describe : Diagnostic criteria Investigations Immediate treatment Subsequent treatment Monitoring treatment Discharge policy
When to use guidelines
Good enough quality AND
Relevant to your patient
Who produces guidelines that i can trust
NICE
SIGN- Scottish intercollegiate guiltiness network
How do I apply that information to my patient?
Best external evidence: guidelines -the stronger the evidence recommendation, the more weight this should be given
Individual clinical expertise : think about co- morbidities (guidelines usually only consider one disease)
Patients values and expectations
EBP
Evidence can come from
RCT
Meta analyses
Cohort studies
Case control studies
What’s the purpose of clinical trails
Designed to determine efficacy
Collect efficacy and safety data on new drugs
After which point does patient irreversibly stay in clinical trail?
After they have given consent
Before allocate to trt
What’s parallel group design
Treatment gp and control gp run in parallel
What does inclusion and exclusion criteria affect?
They affect the extent to which the results of the trail can be generalised
Randomisation is viewed as gold standard because?
Possible error is just chance
Randomisation avoid … bias
Benefit
Avoid allocation bias
Comparable groups with respect to measures and unmeasured risk factors
Eliminates confounding- trt is independent of outcome
Equipoise
Chi square
How likely it is that an observed distribution is due to chance
Double blind is ideal cause
It protects against information bias
Neither patient nor/or(if single blind) researcher knows the trt being taken
Study power =
Probably of detecting a true difference (usually >80%)
What is primary outcome
The outcome that the trial is designed to investigate
What is secondary outcome
Other outcomes that are of interest for example symptom relief, adverse events,quality of life
What is clinical endpoint
It reflects survival or symptomatic status of patient
What is Surrogate endpoint
Associated with clinical endpoint but do not directly reflect survival or a clinical status
Example: BP, cholesterol, tumour size
Non randomised method
2 types of control
Historical control - use the control from literature /databank and compare it with new drug group
Concurrent control - control gp at one centre, trt gp at another / same centre
Pros and cons of crossover design
Pros - eliminates between subject (person) variation from trt comparisons
- make more use of each subject
Cons- more complex
Difficult to interpret carry over (left over trt A + trt B) is present
Consequence of drop outs more serious
Date collection (when? What for? Types? Other purpose
Baseline date - prior to start of trt
Efficacy
Safety data - non/ serious ADR
Follow up status
Intention to treat -?
All subjects inc regardless of adherence
All outcomes must be determined of whether the trt has been discontinued
Objections to ITT
And response to objections
Sub who discountinue no longer receive the benefit from the trt
ADR occurred after trt stopped bias analysis
Study may lack of power where nonadherance is significant
Non adherence can be beneficial in non inferiority or equivalence trail
Non adherends reflect real-life usage
Excluding subject would produce biased estimates
Per protocol analysis
Include only those events that occur while participant is complying with intervention
Vulnerable to bias
Non inferiority design is
To evaluate new trt is not inferior to stnd trt ( less toxic?)
Trail design is more challenging
Terms used to Describe disease frequency
Incidence
Prevalence
What’s incidence
Measure of occurrence of new cases of disease
Incidence rate is equation
No of new cases of disease / mid year population -only inc those “at risk” of getting the disease
How to calculate incidence rate that take changes in population into account?
Number of new case / disease- free person-time at risk
2 case/ 24person- years
For dynamic population how to calculate incidence rate
The no of new case of disease within a specific time period (person-years: add up the years that each person lived for, when the test is carried out
Comparing incidence rates between calendar years can be misleading if the population structure has changed (ppl get old) What are the ways to overcome it
Standardisation -reduce distortion in the rate comparison
Compare rates by categories e.g. Age
Stratify -form into groups
Incidence is new case of disease but prevalence is focused on
Disease status
What’s point prevalence
The proportion of ppl who have a disease at a specific POINT in ti
time
Period prevalence
The number of ppl who gave the disease in a time period divided by the population at mid- point of the period
What’s the equation for prevalence
Number of cases / number of people in the population
Prevalence will always be between .. and …
0 and 1
Prevalence depends on two factors
Disease incidence
Duration
Prevalence may change over time with
Changes in incidence
Changes in disease duration
Intervention programmes
Changing classifications
What’s relative risk
How to calculate it
The ratio of risk between incidence (exposed gp) and incidence (unexposed gp)
Exposed outcome/(EO+exposed no outcome-total population)
Divide by
Unexposed outcome/ (UO+UNO)
What’s is risk?
Probability that individual WILL BE diagnosed with outcome over follow up period
What does it mean when RR=1
There is no difference in incidence exposed and unexposed
Measures of variability
Standard deviation qualifies…
Used for …
Quantified spread of a set of individuals in population
Used for description
Stand error quantifies…
Describes
Used for…
Spread of sample means
Precision of means
Estimating and calculate confidence interval
How are populations selected in a cohort study
People WITHOUT the disease being studied
2 ways to measure an effect give examples for
Absolute
Relative
Absolute risk difference, Attributable risk , NNT, NNH
Relative risk, risk ratio, hazard ratio, odd ratio
How to calculate attributable risk
=absolute risk reduction
Incidence exposed - incidence unexposed
What does NNH measures?
The ADR impact of a trt
The no of ppl who when treated results in one additional harmful outcome
How to calculate NNH?
Results should be round up/ down?
1/ absolute risk difference
Always round down the nearest while number
what’s NNT
No of ppl who must be treated with new trt to achieve one more success than an old trt
What does it mean when NNT=1
New trt always effective
How to calculate NNT
1/ absolute risk difference
Hazard means
The risk of getting outcome in following short time interval (given that outcome has not already occurred)
Hazard can quantify …
Differences in survival
What’s hazard ratio?
Will it be constant over time?
Compare hazards bw gps over follow up time
Hazard may change but hazard ratio is constant
What’s confidence interval
A range of values which we can be confident includes the true value (e.g. Point estimate
How to calculate 95% confidence interval
Mean+- (1.96x SE)
What does SE represent
SD?
The variability of sample MEANS
Variability of a population
Definition of 95% confidence jntervak
If you repeated the study 100 times, on 95 of those times the point estimate (mean) will be within the confidence interval
Point of no difference for absolute and relative difference when value is no difference to interval
Absolute =subtraction so 0
Relative division so 1
Definitions for null hypothesise
Alternative hypothesis
Null: there is NO DIFFERENCE in outcomes comparing treatment with no treatment
Alternative: there is A DIFFERENCE
Definition of p value
Probability of HAVING OBSERVED our data or more extreme data when the NULL HYPOTHESIS IS TRUE
P=…. is usually the value to compare p values against
0.05
How to work out SD
Root mean square Work out mean Mean - value , square it Add up the square values Divide by no of values - mean Square root the mean
What fixed/ open cohort studies
Fixed- exposures are defined
New subjects not included
No movement bw gps
Open- movement can occur in and out of the cohort
Hills aspects of association (for causation of disease)
CSS -consistency strength of association specificity
BBC -biological gradient and plausibility coherence
EAT - experimental evidence analogy temporality
Limitation and use of large simple trails
Reduced the info collected to the minimum required to test hypothesis
Used to test the risk of ADR
Is observation studies experimental or non experimental
Are there any control over exposures/ population
Can confounding occur
Non experimental - no risk to patients
No control over exposures assigned but can select subjects included
Confounding can occur
Steps in the design process of case control study
Define research question Design study To find - group of interest (cases) - comparison group (controls) Take histories (identify exposures) Analyse results Draw conclusions
RCT- 4 aspects of measuring outcomes
Unequivocal (no doubt) death Measured BP, hba1c Clinical assessment (objective) swollen joints, X-ray results Patient opinion (subjective) pain, s/e
In parallel group design for non randomised trail -may have a run-in period
Patient presents - eligibility - consent -run in then -(irreversibly) allocate to trt
Test tolerance to cardiac suppression drugs
Determine disease stability e.g. Copd
Test patient adherence to intervention
Ensure zero plasma level of old trt
Case control study-what cases should be included
Cases with outcomes of interest
CCS- controls - are there any req?
Controls should be found from hopstial neighbourhood community
CCS- match and unmatched studies can be used to
Often match on .. and …
Overcome confounding ( males more likely to have MI) Match on age and sex
Odds:
Probability of event occurring divided by probability of event not occurring
Odds ratio:
How to calculate:
Compares odds that an outcome will occur, between exposed and absence of exposure groups
Odds of exposure in case and control gps
OR= odds of CASES having exposed/ odds of CONTROLS having exposed
A/C (a- cases exposed, c- cases unexposed)
Divided by
B/D (b- control exposed, d- control unexposed)
Cohort vs case control studies Population sampling- How to obtain data - Exposure / event status known or unknown- Compare results - Analytic calculation-
Cohort: Take a gp from population who do NOT have the outcome of interest / case control: take a gp WITH outcome of interest (cases)
Follow them forward in time see what happens/ retrospectively (consider past exposure) identify an appropriate comparison group
EXPOSURE is known / EVENT STATUS is know
Compare OUTCOMES in exposed and unexposed gps/ compare EXPOSURES in case and control
Calculate RELATIVE RISK/ calculate ODD RATIOS
What is p value
Probility of finding the observed, or more extreme, results when the null hypothesis is true (results is due to chance)
What tools help you to critically appraise?
How many other tools are there?
CASP- critical appraisal skills program
7
What does PICO stand for
Population
Intervention
Comparator
Outcomes
What is alpha level
Is the significance level - probability of making wrong decision when the null hypothesis is true
Usually use alpha level 5%
What’s bias
It can be in terms of
Can result in … risks
Systemic error in data/ study
At the pint of study design
Cannot be eliminated at point of analysis
In terms of selection of study subjects, exposure classification, outcome classification
Over/ underestimated risks
Misclassificarion have 2 gps
Non differential - misclassification to same degree for all groups
Differential- misclassifcation different between groups
Confounding
The distortion of RISK ESTIMATE due to the mixture of the ppl in the study population
Confounder
A confounder is a RISK FACTOR for the disease and is CORRELATED with the EXPOSURE INDEPENDENT OF DISEASE
How to control confounding?
Can be implemented at design and analysis stage
1 randomisation 2 restriction 3 matching 4 statified analysis 5 multivariate analysis RRMSM
Two ways of analysis of data to control confounding
Stratified analysis- separate out factors, so any mixture of their effect is removed, used for 1-2 factors
Multivariate analysis - take into acc of annulled of factors
Methods to search a paper for systemic review
1 define the question PICO 2 identify search terms 3 use literally searching 4 review papers 5 summarise results - descriptively for systemic review - quantitatively for meta analysis ( check for heterogeneity
Define a standard method for extracting data from a paper in systemic review
Need study details (type: cohort/ case control/ RCT) Location No of patient included Patient demographics Exposure Covariates Overall result
When to do a systemic review when to do a meta analysis?
Descriptive summary - systemic
Results can be combined and quantified - meta analysis
What type of Meta analysis to use when variation is due to sample variation only?
If there is heterogeneity
Fixed effect meta analysis
Examine this assumption using a test of heterogeneity
Then use a Random effect meta analysis
What is marginal utility?
In economics, decisions are made at the margins i.e. At the point when the utility (satisfaction) not ‘worth’ the 80p you spend
Marginal utility- at the margin of your satisfaction (about to become unsatisfied)
What is opportunity costs
You have less resources (money, time) to do things (opportunity)
What is health economics
The study of attempts to allocate limited health care resources among unlimited wants and needs to achieve the max health benefit for society
What is pharmacoeconomics
The application of the theories, techniques and concepts of health economics to the study of pharmaceutical interventions (treatment)
Medicine regulations QSEC
Quality
Safety
Efficacy
Cost effectiveness
What’s is economic evaluation (in term of pharmacoeconomics)
Comparison of two or more alternative treatments in terms of costs and consequences
Quantify the differences
Ensure max utility from scares resources
What’s QALY
Quality- adjusted life-years
Number of years lived x utility
What can be used to measure the QOL for all health conditions?
EQ-5D scores
Compare with state values
Gives a utility scores between 0 (worst state) and 1
Different types of economics analysis CMA
1 cost minimisation analysis - assume effectiveness is equal,
compares cost of trt with two or more drugs/ interventions
Whichever gives desired outcome at the LOWEST price is preferred
Different types of economics analysis CEA
Cost effectiveness analysis
- NO assumption on equal effectiveness
-Benefit = non monetary terms (usually QALY)
-the greatest effectiveness for the lowest cost
- only consider NHS costs
- main measure in CEA is the ICER
ICER= C1-C0/ E1-E0
C1,E1 the cost and effect in study trt group
C0,E0 the cost and effect in study control group
how to calculate ICER
Compare two different trts
Diff in treatment costs/ diff in effectiveness
Additional cost for each additional patient effectively treated
Different types of economics analysis CBA
Cost -benefit analysis
To compare the outcomes of a medical intervention expressed in monetary terms - not easy so use
Willingness to pay approach (assign monetary value to health consequences)
How much you like to pay for additional 2 yrs of life? (Depends on the age)
Economics is about how to make choices …
Make choices in the context of scare resources
Economic decision are made at the …..? What the outcome of interest
Margin
The incremental cost effectiveness
Incremental analysis
Assessment of the relative cost and the benefit of each marginal addition or reduction in production or consumption
What’s the CEBM levels of evidence
The centre for evidence based medicines
CEBM - does this trt help?
Level 1 (best) systematic review of RCT 2 RCT OR observational study with dramatic effect 3 non Randomised controlled cohort/ follow up study 4 case series, CCS, historically controlled studies 5 mechanism- based reasoning
CEBM - what are the common harms?
1 systematic review of RCT or Observational study with dramatic effect
2 RAndomised trail or ob study with effect
3 non randomised CT/ follow up study
4 case control studies, historically CS
5 mechanism- based reasoning
CEBM- graded may go down on basis of :
Grade may go up if
Study quality
Imprecision of result ( large 95CI)
Indirectness (study PICO not question PICO)
Absolute effect size is small
- there is a large effect size
