This study (retrospective) involved examination of heart rate to QT interval (an EKG measurement) in five treatment groups. Data was collected on multiple individuals in each group (anywhere between 4 to 15 individuals per treatment group). The goal was to look at QT with respect to different heart rates grouped into 10 beats per minute segments (30-39 beats per minute, 40-49 bpm, so on).
A simple approach, had it been prospective, would be to measure a few QT's for each heart rate segment, take the average per individual as an n=1, then do ANOVA with all the groups (for each heart rate group) with the data expressed as mean, SD, N (using sigmastat/ sigmaplot), where mean is the average QT of all the individuals per treatment group.
But this was not the case. So I pooled all the heart rates/QT's for each treatment group, then separated them into the different HR segments. Some segments had 12 HR-QT measurements, some had 300 (most fell around 50 measurements). I am not sure how to do the stats on this. I would like to compare each treatment group at each HR segment. With the data pooled, the graph looks nice, two of the treatment groups start to diverge from the rest at faster heart rates. How would i do the stats on this, is it still an anova, if so, how do I express the data (so that I can appropriately fit it into sigmastat)? Thanks for the help.
Advertise on Talk Stats