Evidence-based practice
Explain the difference between theory and evidence.
Theory = why something should work
evidence = if something works
Explain single subject design
repeated measures
subject serves as their own control
limits intersubject variability
enhances intrasubject variability
various treatment designs (ABA, ABC) treatment/baseline
We like PEDro scores > ___ and QUADAS scores > ___ to indicate high quality studies.
8, 7 (or 10 ideally)
True or false: We want number needed to treat (NNT) to be low.
True - no cap
True
Long duration = P > I
Name 2 barriers that PTs have to utilizing evidence-based practice.
1. low confidence in ability to search
2. low confidence in ability to interpret
3. lack of research skills/training
4. sticking with what they learned in school is easier
5. time
6. limited access to evidence-based resources
7. not prioritized by employer
When do we use non-parametric statistics? (2 reasons)
non-normally distributed data
ordinal scale
What does PICO stand for, and what do we pair for our initial searches?
Patient or problem
Intervention (or cause, prognosis)
Comparison or control
Outcomes
Pair P & O or I & C
The Giordano test has a sensitivity of 95 and specificity of 45 for detecting sarcasm. If the test is negative, what does that tell us about the individual being tested?
We can rule out sarcasm.
An example of a _____ would be: how many people experience a wrist fracture from FOOSH compared to wrist fracture from sports?
Ratio
What are the 3 pillars (or 3 contributing factors) to evidence-based practice?
Research evidence, clinical expertise, and patient values
Which of the following are non-parametric statistical tests?
1. t test
2.chi square
3. ANOVAs
4. shapiro-walk
5. Mann-whitney U test
6. wilcoxon rank sum test
chi square, mann-whitney, and wilcoxon
What are 4 essential requirements of informed consent?
Purpose
methods
risks
benefits
alternatives
compensation
option to withdraw
*questions
What are two approaches we can take with data when we have drop outs?
Completer analysis - take out all data points
intention-to-treat analysis - last known value carries
assume the worst for dichotomous values
"What percentage of individuals who sustain a concussion experience dizziness?" is an example of a _____
Proportion
pros: good randomization minimizes bias, easier to blind that observational studies, good statistical value, clearly identified populations, higher level of evidence
cons: time/money expense, volunteer bias, correlation not causation
more and richer data
follow-up Q
non-verbal behaviors
better for more difficult subjects
controlled response rate
What 4 sections should be part of your first pass when reviewing an article?
1. last paragraph of intro (purpose statement)
2. last paragraph of discussion (conclusion)
3. figures/tables
4. methods (experimental protocol)
1. effect size
2. clinical significance
3. statistical significance
2. noticeable difference in the pt's function (MCID)
3. P< alpha (MDC - greater than standard error)
RRR is associated with ______
1. prospective/cohort studies
2. retrospective/case control
cohort studies = prospective
Rank and briefly describe 5 different types of studies in terms of level of evidence.
1. SR w MA
*CPG
2. RCT
3. cohort (prospective)
4. case-control (retrospective)
5. case series
6. case study/case report
7. expert opinion
ethnography - long term association of researcher w group, but researcher is distinctly different. describes cultures, interactions.
data collection>analysis>dissemination.
grounded theory - data collection and analysis occur simultaneously. theory grounded in observation, not preconceived hypothesis.
List and briefly explain the 3 research principles we uphold.
1. Beneficence (do no harm, do good)
2. Justice (equitable selection, fair treatment)
3. Respect for Persons (autonomy, informed consent)
In the mathematical equation to calculate specificity, what is not accounted for, which directly relates to how we utilize specific tests?
a) true positives
b) false positives
c) true negatives
d) false negatives
does not account for False negatives, so we cannot rule out with a negative test
Give an example of a"rate"
There are X people per 1000 people per year
Rate of new disease over a particular unit of time.
"45 new concussions per year while playing intramural athletics"