I have pre/post data on a very few number of patients (n=5). Measurement X (continuous) was collected at random intervals starting from baseline (resting position) and with increasing heart rate. Each measurment was collected for each patient at variable ranges.
E.g at heart rate 60, 75, 86, 100... for pt 1 pre
at heart rate 62, 66, 77, 109 for pt 1 post
at heart rate 70, 78, 91, 98 for pt 2 pre
...
About 5 measurements were collected at each test. Data was collected as a change from resting position for each patient, since the resting heart rate was variable per person.
I want to test the hypothesis that the magnitude of the change in measurement X from resting postion (baseline) was LESS after the intervention as compared to pre. The measurement does decrease with inreasing heart rate, but this is just a nuisance and we do not care about change with heart rate (dont care about differences in slope). So normally, I believe I would need to fit a mixed model to this data. But with only 5 patients, I do not think this is feasible. Is there any simplified test anyone could suggest I use to supplement a case overview? Since heart rate is not important here, should I consider:
(1) Just testing difference in X between pre and post, as tested from the maximum obtained heart rate for each patient (even if they are very variable)? Is there a way to standardize?
(2) I could just a frequency that everyone had data for and use that to test post /pre change to be consistent, but one of the subjects only got to a very small increase in heart rate post intervention so I am hesitant.
Any suggestions?
