Final Flashcards
What is evidence based practice
“Evidence-based practice (EBP) is essentially a clinical decision making framework that encourages clinicians to integrate information from high quality quantitative and qualitative research with the clinician’s clinical expertise and the client’s background, preferences and values when making decisions” (OT seeker)
Client centered approach to care
Appraises interventions to see if they can be improved
Barriers to EBP
Time Access How to find the literature Cost Misconception that EBP is used over clinic knowledge
Concepts in Evidence-Based Rehab
Awareness
Consultation
Judgment
Creativity
Concepts in Evidence-Based Rehab: Define Awareness
Awareness -“The clinician must be aware of the evidence that has to do with practice and maintain focused awareness”
Concepts in Evidence-Based Rehab: Define Consultation
Consultation- “Specialized set of skills and knowledge and the ability to communicate well as a educator and service provider”
Concepts in Evidence-Based Rehab: Define Judgment
Judgment- “The practitioner differentiates between cases about how t apply recommendations of EBP, tailed to the clients situation.”
Concepts in Evidence-Based Rehab: Define Creativity
Creativity- “EBR requires creativity and insight because the practice and application of the best available evidence is not always straightforward.”
What are the 5 steps in the EBP process?
Clinical question Appraisal of evidence Application of evidence Consider the clients needs Evaluate the clinic outcomes
Why should we be using EBP
Identifies best practices including tx interventions Supports a client centered approach Based on ongoing self-directed learning Supports our services Makes us relevant Reimbursement
Qualitative Study Designs
Purpose- To develop a deep understanding usually through narrative description
Used for descriptive or exploratory research
Open ended questions, interviews, observations
To describe the state of conditions or to explore associations, formulate theory, generate hypotheses
LOE (Level of Evidence):
Level I
systematic reviews, meta-analyses, randomized controlled trials
LOE (Level of Evidence):
Level II
2 groups, non-randomized studies (cohort, case-control)
LOE (Level of Evidence):
Level III
1 group, nonrandomized (before and after, pretest and post test)
LOE (Level of Evidence):
Level IV
Descriptive studies that include analysis of outcomes (single subject design, case series)
LOE (Level of Evidence):
Level V
case reports and expert opinion that include narrative literature reviews and consensus statements
Meta-Analysis:
A systematic review that uses quantitative methods to summarize the results
combines data for individual studies and performs statistical tests
-uses statistical results to provide a treatment effect
Systematic Review:
Authors have systematically searched for, appraised, and summarised all of the medical literature for a specific topic.
comprehensive review of the literature
- critically evaluates the relevant articles - synthesizes the information
Critically Appraised Topic:
Authors evaluate and synthesize multiple research studies.
Critically Appraised Articles
Authors evaluate and synopsize individual research studies.
Randomized Controlled Trials:
Include a randomized group of patients in an experimental group and a control group. These groups are followed up for the variables/outcomes of interest.
Cohort Study:
: Identifies two groups (cohorts) of patients, one which did receive the exposure of interest, and one which did not, and following these cohorts forward for the outcome of interest.
Case-Control Study
Identifies patients who have the outcome of interest (cases) and control patients without the same outcome, and looks for exposure of interest.
Background Information/Expert Opinion
Handbooks, encyclopedias, and textbooks often provide a good foundation or introduction and often include generalized information about a condition. While background information presents a convenient summary, often it takes about three years for this type of literature to be published.
Animal Research/Lab Studies
Information begins at the bottom of the pyramid: this is where ideas and laboratory research takes place. Ideas turn into therapies and diagnostic tools, which then are tested with lab models and animals
PICO stands for
Patient/Population
Intervention
Comparison
Outcome
The purpose of a problem statement is:
to identify the problem leading you to propose a plan of action and provides rationale for further research in this area
Main Types of Research Design
descriptive, correlational, quasi-experimental and experimental
Descriptive research design
describes data and characteristics of what is being studied
Correlational research design
examines relationships between two or more variables
Quasi-experimental research design
investigates causal relationships between variables without random assignment to experimental groups
- not randomly allocated
Types: nonequivalent control group, interrupted time series, combined design
-not ethical to withhold treatment
Experimental research design
manipulates variables to determine causality between factors
for example randomized controlled trial Types: posttest only design (no pretest), solomon four-group design, factorial designs (two or more IV), counterbalance design (cross over) -randomization -control group -manipulation of IV
Qualitative research
descriptive, observations, interviews
Types: phenomenology, ethnographic, heuristic, narrative
Quantitative research
true experiment (cause and effect), randomization, control group, manipulation of IV Types: retrospective, quasi-experimental, pre-experimental, nonexperimental
Mixed methods research
collecting both qualitative and quantitative info
- can be very informative - narrative stories and statistical data - pro: to provide a better understanding of problem - con: can be time consuming, relating different types of data
Non-experimental research designs
survey design, ex post facto design (retrospective), observational designs
Observational studies
- for example…no treatment group (aka no intervention), case reports (1 patient), case series (several patients), case control studies (with/without conditions), cohort studies (with/without exposure)
Cohort study (prospective)
study of a group of individuals, some of whom are exposed to a variable of interest (i.e. drug or environmental exposure), in which participants are followed up over time to determine who develops the outcome of interest and whether the outcome is associated with the exposure
Cohort study (retrospective)
data is gathered for a cohort that was formed sometime in teh past. Exposures and outcomes have already occurred at the start of the study. You are studying the risk factor and see if you can associate a disease to it. Individuals split by exposure.
Case control study
patients who already have a specific condition or outcome are compared with people who do not. Researchers look back in time (retrospective) to identify possible exposures. They often rely on medical records and patient recall for data collection. Individuals split by disease.
Randomization (answer yes or no for the following) True-Experimental Quasi-Experimental Non-Experimental Pre-Experimental
True-Experimental- yes
Quasi-Experimental- yes
Non-Experimental- no
Pre-Experimental- no
Control Group? (answer yes or no for the following) True-Experimental Quasi-Experimental Non-Experimental Pre-Experimental
True-Experimental - yes
Quasi-Experimental- no
Non-Experimental- maybe
Pre-Experimental-maybe
Manipulation? (answer yes or no for the following) True-Experimental Quasi-Experimental Non-Experimental Pre-Experimental
True-Experimental- yes
Quasi-Experimental- yes
Non-Experimental- maybe
Pre-Experimental-no
N =
n =
N = population; n = sample
p =
probability or likelihood that a result could have occurred by chance
-the lower the p-value….
the better, because that means there was less of a probability that the results occurred by chance
-p=0.05 means there is a
95% chance that the intervention tested caused the change in results, aka the results are statistically significant, which tells the reader that the study is important
-p=0.01 is better than 0.05
true or false?
true
Confidence interval (CI)
an interval in which a measurement or trial falls within or corresponds with a given probability, aka what is the likelihood that the intervention caused the outcome
Independent variable
represents the value being manipulated by the researcher, aka the intervention/treatment the authors are manipulating in the study
Ex - fall prevention program or weighted vest Think “I goes with I”...independent = intervention
Dependent variable
the event studied and expected to change when the independent variable is manipulated, aka what the authors are measuring. It reflects the output resulting from manipulation of the independent variable
Ex - # of falls at followup, attention Think this is what’s dependent on the intervention
Parametric Variables Types
(hard numbers)
-interval variables
allow us not only to rank order the items that are measured, but also to quantify and compare the sizes of differences between them. Do not have a true zero, meaning the value of the variable cannot be 0.
-ex - age, weight, muscle power (in lbs), blood pressure, time
-ratio variables
same as interval variables, but they do have an identifiable absolute zero.
- ex - temperature, ROM, exam grade - none of these can be 0 * most statistical data analysis procedures do not distinguish between the interval and ratio variables.
Non-parametric Variables Types:
nominal variables
allow for only qualitative classification.
Ex - demographic info like gender, race, color, city, etc.
Non-parametric Variables Types:ordinal variables
allow us to rank order the items we measure in terms of which has more or less of the quality represented by the variable but not exactly by how much.
Ex - we can’t measure how much Heidi likes the 4 types of candy, but we can rank the order of which she likes best, second best, least, and so on, socioeconomic status, level of agreement, developmental level
Descriptive statistics
used to provide a description of a given variable
-used in univariate analysis
Ex - averages, ranges, and degree of variation
Inferential statistics
used to draw conclusions about a population from a sample by
Estimation - to determine the true value of the parameter from a sample
Hypothesis testing - to determine whether a difference exists in a parameter between groups of data
Inferential statistics : Estimation
to determine the true value of the parameter from a sample
Inferential statistics : Hypothesis testing
to determine whether a difference exists in a parameter between groups of data
Central tendency
refers to what is typical or average; usually reported as descriptive statistics
What are the 3 measures of central tendency?
-Mean: mathematical average
Ex - the average of {1, 3, 5, 7, 9} is 5
Typically expressed as M or X in research papers
-Median: mathematical half point
Ex - the median of {1, 2, 4, 7, 9} is 4
-Mode: most frequently occurring value
Ex - the mode of {2, 3, 3, 5, 9} is 3
Measures of variability
reflects the degree of spread or dispersion that characterizes a
group of values
Standard deviation (sd or SD)
amount that a score varies from the mean or average of scores
- 1 SD is _____of the whole
- 2 SDs is _____ of the whole
- 3 SDs is ____ of the whole
- 1 SD is 68% of the whole
- 2 SDs is 95% of the whole
- 3 SDs is 99% of the whole
Variance (s squared)
represents the square of the standard deviation, measure of how far a set of numbers are spread out from each other
-Range
represents the difference between the highest and lowest score in a distribution
T-test
used to compare the means of 2 groups of measurements (i.e. comparing average exam scores between section 1 and section 2) including:
Independent samples:
comparison of outcomes of control vs. experimental groups in
experimental designs. Independent T is an individual score.
Paired samples:
comparison between pretest and posttest scores in a single group before-and-after design (a.k.a. paired samples t-test). Paired T is pre- and post-measures compared to each other.
-Reported as a t = #.##, p
Mann-Whitney U test (“U”)
- nonparametric
- equivalent of independent samples t-test that can be used to analyze ordinal data -two-group independent groups design when measurement is at least ordinal,
**aka it is comparing 2 non-parametric groups
- non parametric equivalent of independent samples t-test
- Analyses the degree of separation
- Reported as “U’ or “U”’
Wilcoxon signed-ranks test (T)
- nonparametric test
- nonparametric version of aired samples t-test when measurement is at least ordinal
- Nonparametric test that is the nonparametric version of paired samples t-test
- It is represented by “t”’
**aka it compares 2 non-parametric individuals (pre- and post-test)
Analysis of Variance (ANOVA) -
- more complex version of a t-test
- allows comparison of 3 or more groups when there are at least 2 outcomes being analyzed,
Parametric data More complex version of T test, - compares 3 or more groups. One way ANOVA is used if there is only one outcome or independent variable being analyzed. Reported as F (df, df), = #,##, P
One way ANOVA is used if
there is only one outcome or independent variable being analyzed
Two-way ANOVA for more than
1 independent variable. However, when we are dealing with more than 1 dependent variable, the term changes to Multivariate ANOVA or MANOVA.
-Reported as: F (df, df) = #.##, p
The Kruskal-Wallis H-test
Used for nonparametric with 3 or more groups
• Is similar to the ANOVA but is used for nonparameteric
(ordinal) measures when 3 of more independent groups
are involved
• Just as how the ANOVA extends the independent samples
• t-test, the H-test extends the Mann-Whitney U-test
Friedman ANOVA test
• is also similar to the repeat measures ANOVA (i.e.,
comparing multiple points of ordinal data from the same
sample)
• As such it may be compared to the Wilcoxon signed rank
test but for a comparison of 3 or more variables
Analysis of Covariance (ANCOVA)
Used for nonparametric pre/post
Similar to the repeated ANOVA measure
Compared to Willcox and signed Ranks, extends.
• In factorial designs, there is a potential for co-effect
among variables.
ANCOVA is used as a statistical
procedure comparing two or more groups while
controlling for the effect of covariates.
When we are
dealing with more than one dependent variable, the
term used is Multivariate ANCOVA or MANCOVA.
Chi-Square
• Used to analyze tallies or frequencies of nominal or
categorical data. It may be used to:
Analyze the “goodness of fit” of a single group sample
Example: Given a sample of students, is there a specific preference
for teaching styles
Test the degree of independence of 2 or more groups (also
known as Pearson’s Chi Square)
Example: Do audiences who watch more TV per week more
obese?
Correlation
• Used to measure the extent to which two variables are
associated (i.e., when X changes, Y tends to change also)
Correlations should NOT be interpreted as a cause and effect
relationship, i.e., X does not cause Y to change or vice versa.
• Reported as r = .##, p
Pearson’s r
• Also known as Pearson’s Product Moment Correlation
Coefficient is used for parametric data.
• Represents quantitatively the extent to which scores
on two variables occupy the same relative position.
Spearman’s Rho (rs)
• This correlation coefficient is appropriate when
either of the following two conditions are met:
One variable is an ordinal scale and the other is an ordinal
scale or higher.
One of the distributions is markedly skewed.
In either case, both scales must be converted to ranks
Regression
• Is a type of analysis designed to predict the levels of
another variable.
• A common use for this statistical procedure is to identify
risk factors as a predictors of a certain condition.
• Reported as R2 = .## for each independent variable that
is significantly predictive.
• In other words, only those variables that could “account
for” as predictive of a condition are included.
Example: In one study (Halfon et al, 2001), length of stay, age, and
morbidity predisposition were identified as predictive factors in falls
within a hospital setting. Gender and type of surgery were minimally
correlated but were not considered predictive of falls in the hospital
setting.
Logistic regression
• a type of regression analysis that is used to predict the outcome of categorical variables (dependent variables).
• A common use of logistic regression is when the outcome is
dichotomous (e.g., fell, did not fall). This procedure is termed as
binomial logistic regression.
Predicts the odds associated with the presence or absence of the dependent
variable/outcome
Results presented as odds ratio (OR) – how much more or less likely it is for a
participant to belong to the target group than the reference group
• When there are more than two categorical possibilities (e.g., better, no
change, worse), the procedure is known as multinomial logistic
regression.
Cronbach alpha coefficient
• is a coefficient of reliability or internal consistency; a
measure of how closely related a set of items are as a
group.
Intraclass correlation coefficient (ICC)
• is a measure of the reliability of measurements or
ratings; often used to determine inter-rater reliability
Types of bias:
Selection bias
how participants are unrepresentative of the population you are studying
how participants are entered into the studies. Subjects in the sample are unrepresentative of the population you’re interested in.
Ex - volunteer or referral bias, seasonal bias, attention bias
Types of bias: Measurement bias:
problems related to how the outcome of the interest was measured
Ex - lack of “independent” evaluation, recall or memory bias
Types of bias: Researcher bias
This occurs when a researcher’s personal beliefs influence the choice of methodology or the research question.
Types of bias: Researcher bias
-Interviewer bias
type of bias in which an interviewer’s knowledge may influence the structure of questions and the manner of presentation, which may influence responses
Types of bias: Researcher bias
-Recall bias
type of bias in which those with a particular outcome or exposure may remember events more clearly or amplify their recollections
Types of bias: Researcher bias
-Observer bias
type of bias in which observers may have preconceived expectations of what they should find in an examination
Types of bias: Researcher bias
-Intervention bias:
how the tx or intervention was carried out
Types of bias: Researcher bias
-Publication bias
Research with positive results is more likely to be published than research that shows no effect. This can make interventions seem more effective than they actually are.
According to the study design pyramid, how do the study types rank from top to bottom (best to worst)?
- systematic reviews and meta-analysis
- RCT double blind study
- cohort study
- case control study
- case series
- case report
- ideas, editorials, opinions
Bias - any systematic error in the design, conduct, or analysis of a study.
Identify the type of risk from Level I to level IV
Level I evidence has a low risk of bias
Level II has a moderately low risk of bias
Level III has a moderately high risk of bias
Level IV has a high risk of bias
Intervention bias
how the tx or intervention was carried out.
Ex - contamination, co-intervention, timing of intervention, site of tx, different therapists