Download - Quantitative Synthesis II: Interactive Quiz Thomas Trikalinos, MD, PhD Joseph Lau, MD Tufts EPC
Quantitative Synthesis II:Quantitative Synthesis II:Interactive QuizInteractive Quiz
Thomas Trikalinos, MD, PhDThomas Trikalinos, MD, PhDJoseph Lau, MDJoseph Lau, MD
Tufts EPCTufts EPC
Interactive Quiz: InstructionsInteractive Quiz: Instructions
Open this presentation as a slideshow. This will activate the hyperlinks.
When you come to a question, choose the red box corresponding to your choice. If you are correct, you will be directed forward. If you are incorrect, you will be directed back to the question to choose again.– Click on forward hyperlinks (Red Boxes) to follow
through based on your responses.
– Click on home hyperlink (Blue House Icon) to go back to the last correct step in the series.
Participation Quiz: Participation Quiz: Question 1Question 1
Assume that you are performing a meta-analysis of RCTs on vitamin D supplementation and mortality in the institutionalized elderly. You want to explore whether the effect differs across trials according to their maximum followup. Maximum length of followup is
A patient-level covariate
A study-level covariate
Q1: A Study-level Covariate Q1: A Study-level Covariate
Correct! Study-level covariates pertain to the whole study, and do not vary across patients in the same study. Maximum followup is the same for all patients in a study.
[SELECT RED BOX]
Q1: A Patient-level CovariateQ1: A Patient-level Covariate
Not exactly…– Patient level covariates differ across
patients in the same study or in the same study arm.
– Maximum followup length is the same for all patients in a study.
[Click on Blue Box to Go Back]
Participation Quiz: Participation Quiz: Question 2Question 2
What if you wanted to explore whether the effect differs across trials by protocol-specified vitamin D dose? The protocol specified dose is
A patient-level covariate
A study-level covariate
Q2: A Study-level Covariate Q2: A Study-level Covariate
Correct! All characteristics of the protocol or the study design are study-level covariates.
[SELECT RED BOX]
Q2: A Patient-level CovariateQ2: A Patient-level Covariate
Not exactly…– Characteristics of the protocol or the study
design apply to the whole study (the protocol does not differ across participants)
[Click on Blue Box to Go Back]
Participation Quiz:Participation Quiz:Question 3Question 3
What if you wanted to explore whether the effect differs across trials by the received vitamin D dose (assume that you actually know this information)? The mean dose of vitamin D that was actually received is
A patient-level covariate
A study-level covariate
Q3: A Study-level CovariateQ3: A Study-level Covariate
Not exactly… – Strictly speaking, the received dose is a
patient-level covariate, as it may be different than the protocol-specified dose.
– Whenever the covariate is a mean level of a quantity or a proportion of patients with some characteristic, it is something that varies at the patient level.
[Click on Blue Box to Go Back]
Q3: A Patient-level CovariateQ3: A Patient-level Covariate
Correct! Whenever the covariate is a mean level of a quantity or a proportion of patients with some characteristic, it is something that varies at the patient level
[SELECT RED BOX]
Participation Quiz:Participation Quiz:Question 4Question 4
Because of the risk of ecological fallacy, meta-regressions on patient-level covariates should never be performed.
I agree
I disagree
Q4: I agreeQ4: I agree
Never say never. – Meta-regressions of patient-level covariates are
not by definition false. For example, the average received vitamin D dose in a well conducted RCT may be very close to the protocol dose and may be practically representative of the dose received by each patient.
– Further, meta-regression is about exploration and forming of hypotheses. If an association from a meta-regression on the mean of a patient-level covariate is biologically plausible, it merits further study.
[Click on Blue Box to Go Back]
Q4: I disagreeQ4: I disagree
Correct! Meta-regression is about exploration and forming of hypotheses. If an association from a meta-regression on the mean of a patient-level covariate is biologically plausible, it merits further study.
[SELECT RED BOX]
Participation Quiz:Participation Quiz:Question 5Question 5
Communication or Communication or MisinterpretationMisinterpretation
[Go to the next slide][Go to the next slide]
Question 5: Vitamin D and Question 5: Vitamin D and Effects on HealthEffects on Health
Ann Epidemiol 2009
Which is more likely to hold?
These conclusions are based on a well conducted meta-regression (see on the right) and are definitely valid
The analyses and conclusions are suspect
Mean 25(OH)D concentration
Q5: Analyses and Q5: Analyses and Results Are ValidResults Are Valid
As a general rule, one should be suspicious of conclusions that are too strong and results that seem too good to be true… Take another guess…
[Click on Blue Box to Go Back]
Q5: Analyses and Q5: Analyses and Results Are SuspectResults Are Suspect
Correct! This is a problematic analysis.– Ecological fallacy may operate here. (Why?)
– The meta-regression is not correctly specified. You have to read the study and the papers cited therein to know this for sure, but there is at least one telltale sign of a mis-specified meta-regression: the scale of the Y-axis should be logarithmic for the OR and the RR (it should be linear for RD). Here, it is linear for the OR, a bizarre choice. (Can you find more?)
[SELECT RED BOX]
Congratulations!Congratulations!
You have successfully completed the quiz!