Critical literature journal appraisal
[meteor_slideshow slideshow=”adssa” metadata=”height: 126, width: 630″]
Please pull the following article : Granger, Christopher B., et. al.” Apixaban versus warfarin in patients with atrial fibrillation.” New England Journal of Medicine 365.11 (2011):981-992. Please do critique and answer the following questions regarding Measurement/Statistics: 1) Measurement of the intervention/exposure
a)Was the intervention/exposure for the treatment group clearly defined and would it be reproducible (eg. dosing, titration time, distinct procedures in an intervention)?
b)Was the intervention/exposure isolated or were additional interventions initiated that could be responsible for the treatment effects?
c) How was the intervention given to the treatment group? Was it provided consistently across subects, across settings, and consistent over time?
d) Was there any direct measure for the implementation of the intervention to assure the intervention was implemented (er. compliance measures?)
e)For observational studies, how was exposure defined and would there be the possibility for a measurement bias (eg. surveillance or misclassification bias)?
f) For observational studies, is there concern for confounding?
2) Measurement of the outcomes
a) Was blinding implemented properly in all concerned entities (patients, providers, data analysts) or do you see any indication for a measurement bias?
b) Was the selection of outcomes appropriated ? were all important outcomes considered?
c) Are the measures/definitions for these outcomes appropriate and valid?
d) Was measurement reliable or was their chance for large variation in measrement that could obscure the results?
e) Was the follow-up time sufficient to find changes in the outcomes?
3. Other biases
a. Are there any other biases that need to be considered?
4. Statistical analysis
a. Does the article report a power calculation? if yes, were the variables (baseline values and effect size) reasonable? if not, would you consider the study large (and long enough to shoow significant changes in endpoints or would you consider the study too large and essentially overpowered?
b. Are the statistics computed appropriate for the type of data and design?
[meteor_slideshow slideshow=”best” metadata=”height: 126, width: 630″]