Quiz 4 Flashcards
Three types of literature reviews
Narrative, systematic, and Meta-analysis
What is a narrative review?
A review that summarizes in general what is in the literature on a given topic
Characteristics of a narrative review
- Often written by experts in the field.
- A good source for background information
- Fairly easy to write and read
______ review does not follow strict system methods like the other literatures
Narrative review
Which review is prone to bias and is low in the hierarchy of evidence?
Narrative review
How can narrative bias be seen?
- During the literature research
- During reporting of studies
- In discussion and conclusion
How can narrative bias be seen during literature research?
- Authors may be selective as to which articles are selected
- They may include articles that support their hypothesis & exclude those that do not
How can narrative bias be seen during reporting of studies?
- Rigorous appraisal methods are not used to evaluate included articles
How can narrative bias be seen in discussion and conclusion?
- The authors’ opinions may be mixed together with evidence
What is a systematic review?
A rigorous process of searching, appraising, summarizing, all the information on a selected topic
Systematic reviews most commonly address what type of questions?
Questions of effectiveness, but also diagnostic accuracy, prognosis, or other areas of research
What is the aim of a systematic review?
To find all studies addressing the review’s question using an OBJECTIVE & TRANSPARENT process
What is the gold standard of systematic reviews?
The cochrane collaboration
A systematic review is also called a
Study of studies
What is the process of a systematic review
- State the study objective
- Develop the protocol (set inclusion/exclusion criteria)
- Develop a search strategy
- Conduct the search
- Retrieve relevant papers
- Screen and select papers that meet established criteria
- Evaluate Methodogical quality of selected studies
- Analyze & synthesize findings
- Determine if statistical data are sufficient for further analyses (If not report results of systematic review) (if it does move to the next step)
- Analyze effects of size estimates
_ Report results of meta-analysis
The results of the included studies of a systematic review are ____ or ____
Qualitative or quantitative
What happens when the results of a systematic review is qualitatively synthesized?
Written information is merged
What happens when the results of a systematic review is quantitatively synthesized?
Data are merged
Most known diagnostic study quality?
QUADAS
Most known interventional study quality?
PEDro
Information collected from a research can include
The method, participants, funding sources, and outcomes
Short line in forest plot represents
Better, more valid information
Long line in the forest plot represents
Not as good, weaker information
The diamond in the forest represents
The combined results of all data included
The diamond is considered____ and ______ evidence
more reliable and better
What is a meta-analysis?
The statistical technique involved in extracting and combining data produce a summary result
When can/should you do a meta-analysis?
- When more than one study has estimated an effect
- When there are to no differences in the study characteristics that are likely to substantially affect outcome
- When the outcome has been measured in similar ways
What is the most common way to see the results of a meta-analysis summarized?
In a forest plot
What is a forest plot?
A type of graph often used in meta-analyses to illustrate the treatment effect sizes of the studies
Each study in a forest plot is represented by a black square that is ____
an estimate of their effect sizes. Ex: Cohen’s d
____ is appropriate with continuous data
Cohen’s d
_____ is appropriate when study’s outcome measure is dichotomous
Odds ratio (OR)
What is an odds ration (OR)
a comparison of odds of the outcome being present in the treatment group against the control group
What is a confidence interval?
a ratio of values that we are confident contains the population. Includes a point estimate
What is a point estimate?
A single value that represents the best estimate of the population value
_____ value on the confidence interval is probable
Not every
What is the line of no effect?
the line where there is no difference between the two sides of the forest plot
The whole length of the bar in the forest plot is the
Confidence interval
If the confidence interval includes the line of no effect, …
Then it is not statistically significant
What does a meta-analyses typically produce?
A weighted average for the treatment effect estimate
Small samples are more _____ than larger studies
susceptible
_____ studies are given less weight that ____ studies
Smaller & larger
Weighting is also based on____
study quality
The diagnostic process includes…
Pattern recognition, and logical reasoning
What are the components of pattern recognition?
History exam, and physical exam
What is diagnosis from the EBP perspective about
All about probabilities and limiting uncertainty
What are the key thresholds for diagnostic tests?
Test thresholds, and treatment threshold
What is a test threshold?
the probability BELOW which diagnostic test will not be ordered or performed because the possibility of the diagnosis is so remote
What is a treatment threshold?
The probability ABOVE which a diagnostic test will not be ordered or performed because the possibility of the diagnosis is so great that immediate treatment is indicated
What is PEcO
- Patient/Problem
- Exposure: to the diagnostic test under consideration
- c: rarely a comparison of multiple diagnostic tests ( can be with study clinical prediction rule)
- Outcome: diagnosis or detection of the target diagnosis
What is the gold standard study for a diagnostic test?
Cross-sectional study
What is the basic structure of a diagnostic study?
- Series of patients
- Index test
- Reference (gold) standard, - Compare the results of the index test with the reference standard (blinded)
What is a true positive?
When a clinical test is positive and a condition is present
What is a false positive?
When a clinical test is positive and a condition is absent
What is a false negative?
When a clinical test is negative and a condition is present
What is a true negative?
When a clinical test is negative and a condition is absent
Key methodological aspects of a diagnostic study
- Gold standard
- Blinding
- Spectrum of patients
- Everybody should have both index test and gold standard test
What is a spectrum bias?
a lack of sufficient heterogeneity of subjects
What is a Work up/ verification bias?
a study that will distort properties of the proposed diagnostic test
What are the three steps in appraising a diagnostic test?
- Are the results valid
- What are the results?
- Will they help care for my patient?
What is the 2 by 2 table: sensitivity of a diagnostic accuracy?
The proportion of people with the disease that have a positive test result.
a/ a + c
A= true negative C = false negative
What are the important features of a 2 by 2 table that should be read?
Sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio.
What is the 2 by 2 table: specificity of a diagnostic accuracy?
The proportion of people without the disease who have a negative test result.
d/b+d
B = false positive D = true negative
Specificity is helpful for ______ the condition ( when test is positive)
ruling in
Sensitivity is helpful for _______ the condition (when negative test)
ruling out
What does SpPin stand for?
A test with HIGH specificity, that is positive, helps rule a condition In
What does SnNout stand for?
A test with HIGH sensitivity, that is negative, helps rule a condition Out
Postive and Negative predictive values look at…?
Positive: true positive & false positive
Negative: false negative and true negatives
Why arent Postive and Negative predictive values really used anymore?
It is highly influenced by prevalence of condition
What are likelihood ratios (LRs)?
The sensitivity information combined with specificity information
What yields a posttest probability?
Pretest probability of having a condition combined with LR yields a post-test probability
How is a nomogram used?
- The pre-test probability is estimated
- Test LR is obtained from an article
- Draw a line between the pretest and the LR, extending to the posttest probability
Where does the pre-test probability value come from?
- Intuition based on patient demographics and nature of patient’s complaint or clinical experience
- Find published values for prevalence of the condition in a population of patients similar to your patient
Standard interpretations of LRs
- Small: rarely important: LR+ is between 1-2 & LR- is between 0.5-1
- Small: sometimes important: LR+ is between 2-5, LR- is between 0.5-0.2
- Moderate shift: LR+ is between 5-10, LR- is between 0.1-0.2
- Large and often conclusive: LR+ > 10, LR- < 0.1
Where to find sensitivity, specificity, and LRs?
Pubmed, EBP textbooks
It is not really accurate to use LRs in series due to ____
Shared variance (overlap of prediction)
What is the best way to deal with a continuous diagnostic test variable?
With a receiver operator characteristics (ROC) curve
What is an ROC curve?
A curve that plots the probability of true positives (sensitivity) against probability of false positives (1-specificity) for all possible cut scores
Area under curve (AUC) is
the overall measure of sensitivity, specificity
What is a best cut score?
A score on the continuous scale that yields the highest true positive rate while still minimizing the false positive rate
What is the process for a clinical decision development process?
Derivation study, validation studies, and impact analysis
Steps of a derivation study
- Is a clinical decision instrument(CDI) needed or possible?
- will the CDI be simple or complex?
- coming up with candidate variables (the larger you have, the larger the population examined)
- Statistical analysis: Measure candidate variables in a population, and follow the population to determine outcomes
Reasons to validate a CDI
- Confirming results
- Make sure that CDI works in a different population than the one used in the derivation study
- Need to make sure that CDI can be collected properly
How to validate a CDI?
Take the CDI, and apply to the population of patients for which it was intended then follow them forward to determine if the outcomes are diagnostic or prognostic
Types of validation studies
Internal, external, narrow or diverse
What is an internal validation study?
Studies that are usually done on the same population that was used for the derivation study
What is an external validation study?
Studies that are done on a different population than that from the derivation study
What is a narrow validation study?
Studies that involve a specific group of patients that may or may not reflect patients seen in practice.
What is a diverse validation study?
Studies that involve a broader range of clinical situations
Things to look for when reading a validation study
- Make sure that patients were selected in a non-biased fashion
- Make sure that whoever was using the clinical instrument don’t know the outcome of the patient (blinding)
- Make sure that whoever is measuring the gold standard doesnt know the outcome.
- Any subjective measures should have been shown to have good inter-rater reliability
- Follow up: make sure that the researchers followed up with the participants of the study
Reasons for an impact analysis
- Make sure that the study works in a real world
- Make sure that CDI does what its supposed to in the real world
- Make sure that CDI is useable
How to do an impact analysis
- Randomization of clinicians/institutions to usual care (what they normally do) or to use the CDI, look at benefits harm, and cost of the CDI. (best one)
- Non randomized design of looking at pre and post CDI. Can result in a bias
