Here are 10 considerations when doing evidence-based literature review
Process helps with making best drug decisions
Before hospital pharmacists make medication change recommendations to their hospitals' pharmacy and therapeutics (P&T) committees, they need to do a thorough evaluation of the literature.
An expert recommends that they do this evaluation with 10 major considerations. Here's what they are:
1. Was the power set and met?
"Was the power set and met with the appropriate number of people needed in the study?" says Patrick J. Bryant, PharmD, FSCIP, director of the drug information center and a clinical professor in the division of pharmacy practice and administration in the school of pharmacy at the University of Missouri-Kansas City.
2. Was the dosage or treatment regimen appropriate?
For instance, if investigators used minimal doses of the comparative agent and a strong/higher dose of the study drug, then that might make the study drug look better than it otherwise might, Bryant says.
"What we're looking for is whether these are dosages you would normally see in a clinical setting and whether they're equitable between the two drugs being studied," he adds.
3. Was the length of the study appropriate?
The study's length needs to be appropriate to what is needed to show a true intervention effect.
For example, with antidepressant drugs, these sometimes take eight weeks or longer to demonstrate a maximum effect, Bryant notes.
"So if you did a study that's only three weeks long then you probably won't see how well these drugs will perform," he explains.
4. Is the inclusion/exclusion criteria adequate to picking out the population of people you want to study?
Using the antidepressant drug study example, it would be important to make certain that if a hospital P&T is considering including the drug for treating anxiety, that studies that enrolled anxiety diagnoses are reviewed. It might be less helpful to review only the literature pertaining to the drug's treatment of major depression disorder, Bryant says.
"You don't want to mix apples and oranges," he says. "So you need to make sure it defines what you're looking for, and this ties back to that clinical question."
5. Is the exclusion criteria adequate?
"You want to make sure that you're excluding people that the drug could harm," Bryant says. "Perhaps people with certain disease states shouldn't be given this drug because it might exacerbate their disease state."
The goal is to find studies that list adequate exclusion criteria so a pharmacist can assess the risk more thoroughly.
6. Was blinding present?
"This is important because if you have a blinded study, meaning neither the investigator nor patient know what the patient's getting, then you can eliminate a lot of the bias," Bryant explains. "You can have a double-blinded study where neither person knows or a single-blinded study where the physician does know but the patient doesn't, or it could be unblinded where they both know which drug is given."
Each of these scenarios will result in a different bias, he adds.
7. Randomization resulted in similar groups.
"Randomization makes sure everyone has an equal chance of getting into one or the other group," Bryant says. "That's important because the way you can tell how a randomization schedule works is if you look at the demographics of each group."
For instance, generally the first table in an article is the demographics table.
"Go down the table to see if the ages were about the same, if there were the same amount of men and women," he explains.
"If there is a difference in the groups when you start out, then when you get to the end of the study and you see a difference in results, how would you know if the difference is due to the drug or whether it's due to the difference in people between the study drug and the control drug? Bryant says. "You need to start out with similar groups up front so if you see a difference in blood pressure at the end of the trial, then you can be reasonably certain the groups were at least the same in the beginning."
8. Biostatistical tests appropriate for the type of data analyzed?
Data can be categorized into four types: interval data, ratio data, nominal data, and ordinal data, Bryant says.
"Each are a different kind of data that has a different distribution, and so most of these will require a different type of test," he explains. "Ratio and interval data have the same type of test."
9. Measurements, standards, validation and accepted practice.
"Did you use a test that is generally accepted and/or validated to be able to measure something that is also generally the way you'd practice in a clinical situation?" Bryant asks.
"For example, was the blood pressure measured with a sphygmomanometer, which is the accepted tool?" he says. "It's what we use in clinics."
If a study had investigators using a different blood pressure measurement, then that might raise a red flag, or at least a question.
10. What are the authors conclusions?
"You want to see that the author's conclusions are supported by the results," Bryant says. "Use your logic."
If a pharmacist sees some questionable items in the study and it appears the author is over-reaching with his or her conclusions, then it's time to look more closely at potential biases in the data, he adds.