Replies to post #249081 on Innovation Pharmaceuticals Inc (IPIX)

[BonelessCat]

Do we know the severity of UP for each patient? I haven’t seen that revealed other than inclusion criteria that stated there was a global assessment involved. The b cohort could have enrolled a higher number of severe cases over all than the first and last. Also note that on the poster there is one outlier in results for each group, where a and c’s outlier had shorter time retention, while b had one outlier at >4 hours. All that can easily be explained by UP severity and accompanying damage or weakness to the colon with cohort b assigned a higher number of those patients.

[PlentyParanoid]

Nice problem. I give it first a probability treatment to see if any other explanation is really needed.
We have 17 subject out which 6 have median enema retention time less than one hour. 4 of these short retention (SR) subjects are found in cohort B which has 6 subject. How likely this is if subjects are otherwise similar?

This is called k combination out of set N problem.
There are 12376 unordered ways to take 6 subjects out of 17 (cohort B out of all subjects) or
Factorial(17) / [Factorial(6)*Factorial(17-6)] = 12376
There are 15 unordered ways to take 4 subjects out of 6 (4 out of SR subjects to cohort B) or
Factorial(6) / [Factorial(4)*Factorial(6-4)] = 15
There are 55 unordered ways to take 2 subjects out of 11 (2 out of non-SR subjects) or
Factorial(11) / [Factorial(2)*Factorial(11-2)] = 55

Probability that cohort B would have exactly 4 short retention subjects is:
(15*55)/12376 ~ 0.0666.

The result would not cut the threshold for significance, but, still, does ask for some explanation. Can we find one? According to my reading of the poster the trial was open label dose escalation study with retention time evaluation based on patient diary during outpatient visits. No mention of any randomization (possibly none) which will keep us in the dark how subjects were distributed among cohorts. So ...

I tend to agree with Karin and LilKahuna - there are several benign ways how the results could have happened especially in an open label phase 1 study using evaluation performed by patients - could be differences in enema administration times, overall physical activity, eating times, overall health etc...

But let's stay with probabilities for a while.
It seems that subjects had mild to moderate UP/UPS. It is possible that cohort B just had more moderate subjects than other groups. For instance, assuming that there were 8 moderate and 9 mild subjects and no randomization based on disease severity then the probability that cohort B had at least half of the moderate subjects is about 25 % - not impossible occurrence.