Hello,
I have a randomized trial's data with 2 groups. During research were made several steps where some hemodynamic parameters were recorded. My question: If I compare hemodynamic parameters between groups on each steps, will I have a multiple comparisons problem? And if I compare several hemodynamic parameters at one step between groups?
thank you.
sincerely yours,
Dmitrii Volkov
All depends what is the the aim of your investigation. If you follow some, for example, consecutive measurements in time-intervals (effects of some medication), and have control group without administration of the same medication, it will be good to compare each time-point between groups. But, you can compare all parameters within the group(s) and between the groups.
This is a multiple comparisons problem. The assumption needed for testing between the two groups at each time point is that the readings are independent of each other. This is obviously not true. A person who is tachycardic at t0 is more likely to be tachycardic at t1.
As far as I am aware there are some strategies to deal with this.
1. Use some sort of correction (Bonferroni etc.) This increases the chance that your p values won't be significant, but that shouldn't worry you.
2. As a corollary to point (1), you should reduce the number of tests you do. For example, if you are looking at hypertensive response to intubation and are measuring blood pressure every minute, there's no point in comparing the groups after every minute. Domain knowledge should inform you that the peak hypertensive effect occurs at about 3-5 minutes after the laryngoscopy. Restrict tests accordingly.
3. A third strategy is to calculate the area under the curve for each group and compare them once. But the question asked has now changed.
4. If you are not interested in the precise amount of change in hemodynamics, but are only interested in, say, the number of hypotensive episodes in a given period, please define that as the primary outcome. For example, SBP reduced to 80% or lower of the baseline for 2 minutes or more will be counted as one episode of hyypotension. Then you can compare the number of episodes between the groups. Or you may be interested in the highest SBP recorded. Choose you primary outcome clearly taking into account the domain knowledge.
5. You can also think of doing a time series analysis, but you will definitely need a statistician to help you with that.
So ultimately, the answer, as usual, is it all depends.
Sorry for the long post.
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress