What are the 2 types of question in a clinical setting?
- background
- foreground
What are the 5 Steps to the EBP process
ask, access, appraise, apply, assess
What is a background question?
- general questions
- can come from the doctor or patient
What are foreground Qs?
- answerable clinical questions
- from patient and/or doctor
- ie. PICO
What are the realms of foreground questions
(PTHD)
- Prognosis - what was the observed outcome?
- Therapy - will a modality help?
- Harm - will exposure have an adverse effect?
- Diagnosis - will a particular test help me?
When is PICO formatting required?
for creating a foreground question
What does PICO stand for?
- Population/problem
- Intervention
- Comparison intervention
- Outcome of interest
What kind of studies help to answer foreground Qs
Primary studies
- RCTs for therapy related
What is a more practical source of information (other than Dr. Bhalaero)
Pre-appraised literature
What sites are useful for pre-appraised literature
- Dynamed
- TRIP
- Physiotherapy Evidence Database (more applications for Chiro)
When assessing the patient what are some questions you need to ask when it comes to pre-appraised lit.
- Did it help?
- Did it hurt?
What is the EBP Sandwich…
Meat/cheese - the take home question
Bread - how effective is the treatment
Bread - how good is the evidence
What is the pyramid of evidence for research? (from top to bottom)
GOD ->Systemic reviews ->RCT ->Cohort Studies ->Case-control ->Case-study/cross sectional ->Expert opinion ->animal studies
What are best answers for Therapy Qs?
RCT
What are the best answers for etiology and risk factor?
RCT, cohort studies, or case-control studies
What are the best answers for frequency and rate Qs?
cohort and cross sectional studies
What are the best answers for Diagnostic Qs?
cross-sectional studies
What are the best answers for Prognosis and Prediction Qs?
Cohort studies
What is an Observational study?
One in which the researcher is passive
What is an Experimental study?
One in which the researcher is active
What are observational cross-sectional studies
exposure and outcome are measured at the same time and once, like a snap shot in time
what are observational case-control studies
subjects with a specific outcome are matched with those without the outcome and info is obtained about their past exposure to a factor under study
Which study type is best for assessing risk factors or rare conditions?
case-control
What are observational cohort studies
data is obtained via exposure of 2 groups over time
What is considered the primary study in an Experimental Study?
RCT
What are the key components of a RCT?
- Allocation
- Randomization
- Baseline measurement
- Blinded intervention
- Blinded assessors measure outcomes
If the exposure/intervention was randomly allocated, the study is a ______________?
RCT
If the outcome was determined some time after the intervention, the study type was a ____________?
cohort study
If the outcome was measured at the same time as the exposure, the study was a _____________?
cross-sectional (snap shot in time)
If the outcome was measured before the exposure was determined, the study was a ______________?
case-control
A RCT requires what?
- large population sample
- randomization of treatment and control groups
- baseline measurements
- a treatment of interest
- monitoring
What are the key components of a RCT?
- inclusion criteria/exclusion criteria
- concealed allocation/baseline measurements
- treatment initiation/ongoing outcome assessment
- final outcome assessment/determination of treatment effect
Inclusion criteria
- defined characteristics that participants must have in order to be included in the RCT (ie. superpowers…sorry Ben Affleck Batman)
Exclusion Criteria
Characteristics that each participant cannot have in order to be included in the RCT (ie. liking vampires)
Randomization
makes sure there is a well distributed sample of the population in the groups of study
Monitoring Outcomes
moving forward in time; drop outs common
Realms of Appraisal
- How well has the study been conducted (internal validity)?
- What is the treatment effect?
- Is the treatment effect due to chance?
- Is the treatment effect clinically useful, especially given the patient’s unique perspectives and values?
Internal Validity
Are all the RCT parts there?
How well is each part done?
If a part of study is missing, does it fatally flaw the study?
Is it well done that you want to read more about the results?
ABCDFIX - “A”
Allocation
What happens when concealment allocation isn’t done?
40% overestimation!
Is concealment the same as randomizatio?
NO
Allocation best achieved by using what?
a centralized off-site computer allocation process
THE BEST allocation is done by using what?
a sealed opaque envelope
ABCDFIX - “B”
Blinding
The Big Three Blinded
Patients
Providers
Outcome Assessors
The Little Two Blinders
Statisticians
Adjudicators
ABCDFIX - “C”
Comparisons
What is important when doing a comparison?
- did they report a baseline
- did they correct for imbalances
Uneven attention in comparison
- # of treatments
- Amount of attention
- The Hawthorne Effect (more interaction with patients) and Placebo Effect
ABCDFIX - “D”
Drop Outs
Why is considering drop out important
- drop outs may be different
- could be a side effect
- stat power based on group size
- the randomized balance is lost
- it changes the math
5-20% rule
- under 5%: very limited risk of bias
- over 20%: high risk of bias
ABCDFIX - “F”
Follow Up
What is an ideal follow up period
- long enough for the study to be useful clinically
ABCDFIX - “I”
Intention to treat
The intention to treat principle
Once subjects are randomized they should be analyzed in the group they were first randomized, even if 1) they never received treatment, 2) discontinued the trial, 3) or crossed over to the other group
Why do an intention to treat?
prevents an overestimation of the treatment effect
ABCDFIX - “X”
X-factor…minus Simon Cowell
What 2 questions do you ask yourself once you decide to see if your RCT will help to answer PICO
- What do the results show?
- Could the results be due to chance?
Stats involve 2 key principles
- estimation
- inference
Estimation
- how big or small something is
- how big or small something is compared to something else
Concepts with Estimation
mean, median, mode - central tendency
standard dev. - measures of spread
Measures of Central Tendency
Mean - “the average”
Median - halfway point in a list on numbers
What is an outlier
data that doesn’t follow the pattern
What is skewness?
data not normally a perfect bell curve or normally distributed
WHat are the 2 outcome measures?
- Dichotomous outcomes - yes or no outcomes that either happen or don’t
- Continuous outcomes - outcomes that are potentially limitless and vary on a continuum
Reporting of results with dichotomous outcomes
- odds ratio or relative risk (RR)
- absolute risk reduction
- relative risk reduction
- number needed to treat
Reporting of results of continuous outcomes
effect size - difference between the treatment and control group means tells us how large or small the effect is
Inference
drawing conclusions (Bob Ross style)
Hypotheses
statement about the world that could be tested to see whether t is true or false
Confidence Intervals
more informative than p values
measure of spread or precision
an est. of the range of values that are likely to include the results of the study
P-value
>0.05 = stat insignificant - nothing going on <0.05 = stat significant
Confidence Intervals
- the narrower the better - larger studies have narrower CIs
- 95% confidence interval
- should not overlap the point of no effect (point where the null hypothesis is true)
Statistical Significance vs. Clinical Significance
Intervention only considered useful:
- 95% CI stat and clinically significant - clinical significance = size of the effect and the 95% CI in relation to a minimum effect that is clinically important (see graphs!!!!) (pg 10, 8/20/2013 packet)
Relative Risk
risk = chance of something happening RR = comparison of one risk to another
RR in a clinical world
is a comparison of risk outcome, desirable or not, in a treatment group compared to the risk of the same outcome in the control group
(= treatment group / control group)
What does an RR of 1.0 mean?
no difference between control and treatment
treatment neither decreases or increases the risk of event
What does >1.0RR mean?
- treatment group has a higher risk of the outcome than the control
- treatment increases the risk of event
What does <1.0RR mean?
- treatment group has a lower risk of the outcome than control
- treatment reduces the risk of the event
When would you reject the null hypothesis?
when a p value is less than 0.05
When would you accept the null hypothesis?
when the p value is above 0.05
Interpret RR = 0.5 in a RCT testing sunscreen on preventing skin cancer in Pheonix, AZ
- People who use sunscreen have less than a half chance of develop. melanoma
- people who do not use sunscreen have 2x greater risk of developing melanoma
- the rate of melanoma with sunscreen is 50% of the rate without
What is the relative risk reduction?
- the proportional decrease in event rates achieved by therapy
- impressively larger than the risk difference when the event rates are low
What is Absolute Risk Reduction?
the arithmetic difference in event rates achieved by therapy
What is Number Needed to Treat (NNT)?
# of patients needed to treat in order for only one of them to benefit ****the lower the NNT, the more effective the treatment
Calculating NNT (QUESTION)
100 / Absolute Risk Reduction (if given as a %)
1 / Absolute Risk Reduction (if given as a decimal)
Rule of thumb for NNT (QUESTION)
single digit NNTs for treatment applicable to chiro practice are considered effective treatments
Application questions
Is the treatment feasible?
what else do i need to apply this evidence?
what alternatives are available?
External validity
the application of evidence to individuals
External Generalizability
nature of your patients condition and how it relates to the study participants:
- phase/severity of injury
- age etc
What is PARQ
Procedures
Alternatives
Risks
Questions
Assessment
it is critical that you assess the effect on the patient and whther it 1) help? 2) hurt?
will you use it again?
the entire EIP process - was it easy? hard?, how can it be streamlined?
