The trial which allowed Cryolife to get the FDA to accept the lower INR range for the On-X was known as The PROACT Trial. The publication linked below contains data from the PROACT Trial. There were only 375 patients studied in the trial, with about half going to the the lower INR range of 1.5 to 2.0 and the other half, the control group, getting the range of 2.0 to 3.0. There ended up being significantly more TE and Thrombosis events in the 1.5 to 2.0 INR test group; 2.96/patient year vs 1.85/patient year in the control group. See link to Table 1 from the study linked below. That represents a 60% increase in events. Yes, the lower INR group had fewer bleeds, but many are of the opinion that they would rather have a bleeding event than a stroke.
Table 1
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6472691/table/table-1/?report=objectonly
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6472691/
Edit:
I came across this publication which was from a presentation of the interim results for the the PROACT Trial and wanted to add it to the above as it includes some additional data which I find interesting. See link below.
Some findings which I find noteworthy:
"No difference was noted between the low-dose and standard therapy arms at 1.6 years of follow-up, including mortality (2.1%/patient-year [PY] vs. 1.3%/PY, p = 0.45), major bleed (2.8%/PY vs. 4.2%/PY, p = 0.36), minor bleed (3.1%/PY vs. 4.8%/PY, p = 0.3), stroke (1.4%/PY vs. 0.3%/PY, p = 0.16), peripheral thromboembolic event (0.69%/PY vs. 0.32%/PY, p = 0.52), thrombosis (0% in both arms), or a major event, defined as major bleed, stroke, or thrombosis (4.2%/PY vs. 4.5%/PY, p = 0.84)."
They use the term "No difference" but there were, in fact, several differences. I believe that the proper phrasing would be to say: "The differences did not reach statistical significance." This is often the case when a study is under powered with low patient enrollment, and the enrollment was relatively low for PROACT.
One can only wonder if these differences would have reached statistical significance if the study had a higher power with more enrollees:
Mortality (2.1%/patient-year [PY] vs. 1.3%/PY, p = 0.45) This represents a 62% increase in mortality for the low INR group.
Stroke (1.4%/PY vs. 0.3%/PY, p = 0.16) This represents a 467% increase in strokes in the low INR group.
Peripheral thromboembolic event (0.69%/PY vs. 0.32%/PY, p = 0.52 This represents a 216% increase in the low INR group.
Major bleed (2.8%/PY vs. 4.2%/PY, p = 0.36) The low INR group had 33% fewer major bleeds.
Of all of these, the one which I pay the most attention to is mortality. I find a 62% increase in mortality very noteworthy, even if the p value did not reach statistical significance. Was the strategy to keep the enrollment sufficiently low in order to be able to say "No difference in mortality"?
Interesting that the FDA gave approval, given there were so few enrollees. In my view, there is good reason for the controversy over PROACT, as well as the decision to allow for the lower INR range by the FDA. I would also be very interested to know how many of the strokes and other clotting events left the patients with permanent damage.
https://www.acc.org/Latest-in-Cardiology/Clinical-Trials/2014/04/17/21/22/PROACT