BKT, yes I have a dynamic simulation running that updates itself each day and to which I can add sites as they open. This 2nd version is more refined then the first one that I used in the beginning and in which I had to program a non-linear centre opening estimation (based on Shan's hockey stick statement).
Furthermore there were a lot of little parameters to support assumptions for the start up period which is always different from when routine is installed and all systems at the CRO are oiled in.
For this version I grouped all centres that opened with a same listing update at clinicaltrials.gov and added the date in the model. That means I know, given a small margin of error due to reporting delay, when centres where opened and therefore how many days they are open. I was planning to now port my detailed centres list and do that by individual centres and adding an overwrite column to enter data collected from the field which is always centre specific. And now that we know all centres it would also have allowed me to put in a parameter related to their location, size, reputation, ect to tune some. However, I am not going to do the effort because then I am going to start deploy more ways to gather that information and I think PPHM would prefer that not to happen and they want those centres to be left alone. I know what I need to now so I put not to much extra time in that.
The current simulation has three main parameters.
The first one is an on/off saying if the centres opened and had immediately one patient available. That parameters was historical because as long as there was no prove of the hockey stick it could be that PPHM opened the centres based on who had patients ready for treatment and who not, which would have made sense. This parameter is now set to off (0 patients available at opening centre).
The second parameter is the number of patients per day/centre which allow to precisely set the number of patients. Eg: 0.0167 = 0.5 patients per month of 30 days, so 1 patients every 60 days. I do it this ways to simplify the implementation.
The third parameter is the number of patients PPHM MUST enrol per day, not taking in account the # centres, to stay on track. That is 0.7973 patients a day if we use JAN 2013 till DEC 2015 or 730 days. In that case the LAST patients of the 582 would be enrolled in DEC 2015. In the beginning they will have for sure behind on that schedule, now with 138 centres open one may expect they can catch up and have 2 patients every 3 days enrolled somewhere. I simulate this LINEAR because I do not have the parameters of PPHM estimation that makes them say they are On Track.
The last parameter is the date overwrite. If it is 0 it uses the system date (today) and if is not 0 it represents the number of days in the future that I want a simulation for based on the number of centres open. So I can easily see when the trial is fully enrolled.
What I do NOT have any-more (I removed it) is a part that simulates the opening of the remaining centres in the future and their patient enrolment. So technically the simulation gets better each time I add centres.
The results of the simulation per 22/Sept/2014 with the above mentioned parameters is:
- Simulation type 1: PPHM Linear On Track for DEC 2015
---> 208 patients need to be enrolled
- Simulation 2: Patients/month/centre estimated
---> 1...: 484 patients should be enrolled (Done OCT 2014)
---> 0.5 : 242 patients should be enrolled (Done JAN 2015)
---> 0.25: 121 patients should be enrolled (Done OCT 2015)
So if there would be NO other centres to be opened and we assume only 1 patients EVERY 4 month per centre (0.25) then we would be fully enrolled in OCT 2015 instead of DEC 2015. At 1 patients every 2 month (0.50), which is still my favourite estimate, we could be fully enrolled in JAN 2015. I like to add some error margin, some announcement delay etc and therefore stay on the March/April/May/Jun 2015 window for us to be officially told. Of course if PPHM opens another 15 centres the coming months then all this improves.
I consider EVERY earlier announcement of being fully enrolled a TREMENDOUS success and a bonus-grade achievement. But I think no one will have a problem with that, will they?
Peregrine Pharmaceuticals the Microsoft of Biotechnology! All In My Opinion. I am not advising anything, nor accusing anyone.
