Here are some numbers from my simulation based on survival curves and MK study enrollment numbers.
For each set of parameters, the simulated study was run 25000 times. Each run the approximately 400 "patients" were randomly assigned a lifetime in the study based on the survival curve. This simulation was a statistical exercise only and comes with no guarantees of correctness!
1) SUCCESS PROBABILITY (defined here as end date OAS MK group >=10% better than SOC group)
What percent of the studies were successful for Taiwan and CVM curves at dropouts of 0, 2.5 and 5%? Note DO=dropout %
Taiwan survival curve,
__________0% DO__2.5% DO__5% DO
JUL 2019: ___87%____76%____63%
AUG 2019: __91%____81%____69%
SEP 2019: __93%____85%____74%
OCT 2019: __95%____88%____79%
NOV 2019: __96%____91%____83%
DEC 2019: __97%____93%____86%
CVM survival curve,
__________0% DO__2.5% DO___5% DO
JUL 2019: ___100%____100%____99.9%
AUG 2019: __100%____100%____99.9%
SEP 2019: __100%____100%____100%
OCT 2019: __100%____100%____100%
NOV 2019: __100%____100%____100%
DEC 2019: __100%____100%____100%
Second for each month, I recorded the bell curve of outcomes of OAS improvement. In a given month (assuming the study ended at that time) I list the min, mean and max AOS improvement (or detriment -- meaning fewer survived in MK group). For example, CVM reported about 20% improvement in their phase 2 study. Keep in mind that the min and max numbers are at the tails of the bell curve -- they should look a bit shocking. However this does give an idea how wide the range of outcomes may be.
2) RANGE IN OAS IMPROVEMENT (OR DETRIMENT IF NEGATIVE)
Taiwan survival curve, 2.5% dropout rate, 25000 study simulation
__________min___mean__max
JUL 2019: -17%____17____62%
AUG 2019: -16%___18.7___64%
SEP 2019: -15%___20.3___64%
OCT 2019: -13%___21.8___65%
NOV 2019: -12%___23.3___68%
DEC 2019: -12%___24.8___74%
CVM survival curve, 2.5% dropout rate, 25000 study simulation
__________min___mean__max
JUL 2019: 9.6%____59____130%
AUG 2019: 10%____62____131%
SEP 2019: 14%____64____136%
OCT 2019: 16%____67____139%
NOV 2019: 18%____70____139%
DEC 2019: 21%____72____141%
3) CHANGE IN OAS IMPROVEMENT PER MONTH
At the time range I am looking at (2nd half 2019) how does the mean OAS improvement (detriment) vary per month. In other words from month to month how much is the bell curve of outcomes sliding to the right?
Taiwan, all dropout rates, 2nd half 2019
mean improvement in OAS increases about 1.5% every month.
CVM, all dropout rates, 2nd half 2019
mean improvement in OAS increases about 2.7% every month.
4) TIME RANGE FOR SUCCESS PROBABILITY TO GO FROM 0 TO 100%
During the early years of the study (if it were to have ended) there is virtually 0 propability of success. On a long enough timeline the study not ending virtually guarantees success, but what is the expected range?
Taiwan, 2.5% DO
About 3.25 years
0% until Nov 2017, virtually 100% on Jan 2021 (>99% since June 2020 but very slowly increasing)
CVM, 2.5% DO
Just over 2 years
0% until May 2017, virtually 100% on June 2019
Hope I did not over number you. Also hope that CVM picked the best survival curve to match their patients. If so things are looking very good. With the Taiwan curve, we are better than 50% but not high 90's yet. No one likes waiting, but that means MK works better and your ultimate investment success is more likely.
Rogue2
