Hi Anoop and the list,
These are excellent questions and there are two answers for each one.
First, I'd give the traditional standard statistical answer. When you do a study you are supposed to have only one hypothesis (primary outcome) and any other (secondary) outcomes should be 'hypothesis generating' only, that is the results should only be used as a hypothesis for a future study. Also, you are supposed to find out what happened to every person in the study and count their outcomes that actually occurred, regardless of their compliance with the studied therapy. If a person in the study does not complete the study, they cannot be counted in the Intention to Treat analysis.
Now, I'd be remiss if I didn't say that there are other ways of doing this. You can have several "primary" outcomes, but in order to have the level of significance (alpha) set at 0.05, you must use some sort of correction (such as the Bonferroni correction = alpha / number of outcomes) to compare p to in order to have statistical significance. That is still a clean and reasonable way as you are pre-specifying the number of outcomes that you will be studying and not doing that post-hoc (after the data is obtained).
As for the Intention to Treat, there are several ways to deal with patients who have dropped out of the study and are not available for analysis. If a patient dropped out and you can find out what their outcome was, they are perfect to be included in an ITT analysis. However, frequently they are not available for any follow up. Then, probably the most sensible way to analyze their outcomes is using the "Best case - Worst case" scenario. You anaylze the results as if the patients who dropped out all had the worst outcome and then all had the best outcome and compare those numbers. If they are very close to each other, you can expect that the actual result is very robust and is not very 'sensitive' to outlier patients. On the other hand, more likely they will be far apart and you will have to put the results into perspective when you discuss the results, leading to more hypothesis generation. Another way of doing this that is becoming more popular is to use some form of adjustment to model or "predict" the outcomes in those patients who dropped out or were otherwise lost to follow up.
I hope that this helps and will eagerly await the other comments that you question generated.
Best wishes,
Dan Mayer, MD
Professor of Emergency Medicine
Albany Medical College
________________________________________
From: Evidence based health (EBH) [[log in to unmask]] on behalf of Anoop Balachandran [[log in to unmask]]
Sent: Sunday, September 21, 2014 11:10 AM
To: [log in to unmask]
Subject: Intent to Treat and Secondary Outcomes
Hi everyone,
I had two questions about Intent to treat anlaysis and secondary outcomes:
When you write a hypothesis, do you include the secondary outcomes too? If yes, is it of any value since the sample size was based on the primary outcome usually. Now most of the studies have multiple number of secondary outcomes. But i haven't seen anyone elaborate on this.
Also, can we only call a study Intent to treat analysis unless we contact and post test ALL the subjects in the study. I finished an exercise study and we tested all the subjects in their assigned groups, except for the ones who dropped out and lost in contact. Can I still call my study intent to treat?
Thank you so much
Anoop
-----------------------------------------
CONFIDENTIALITY NOTICE: This email and any attachments may contain confidential information that is protected by law and is for the sole use of the individuals or entities to which it is addressed. If you are not the intended recipient, please notify the sender by replying to this email and destroying all copies of the communication and attachments. Further use, disclosure, copying, distribution of, or reliance upon the contents of this email and attachments is strictly prohibited. To contact Albany Medical Center, or for a copy of our privacy practices, please visit us on the Internet at www.amc.edu.
|