Note to readers: An earlier version of this story stated that the FDA may be concerned about the conduct of the CHAMPION trial. This statement could be construed as an implication of ethical misconduct or negligence on the part of the trial sponsors or investigators. I apologize if I conveyed this impression in the earlier story. This story has been further updated below, as indicated in the text.
Next month’s meeting of the FDA’s Circulatory System Devices Panel has stirred controversy, though it may be a tempest in a teapot. On December 8 the committee is scheduled to review the PMA for the implantable CardioMEMS HF Pressure Measurement System (HF System), which provides daily pulmonary arterial pressure measurements for the purpose of guiding heart failure treatment. The potential controversy involves the possible unblinding of patients during the pivotal CHAMPION trial. (Note: The earlier version of this story stated that concerns had been raised about the unblinding of the trial investigators. As the trial publication clearly states, CHAMPION was a single-blind trial in which only the patients but not the investigators were blinded to their treatment assignment.)
On Wednesday morning Wells Fargo analyst Larry Biegelsen reported that
…the FDA has recently visited a number of sites that participated in the CardioMEMS pivotal trial because the FDA has concerns with how the trial was conducted. It is our understanding that FDA is concerned that physicians in the study may have been coached on how to treat the patients in the treatment group and this may have led to bias in the study.
Here is the relevant text from the Lancet publication of the trial:
To maintain patient masking, all patients were asked to take pressure readings every day. Standardised clinician communication scripts were provided for telephone calls to patients about changes to drugs. Sites were required to balance the number of contacts between patients in the treatment and control groups. Patients were masked to their assignment group. The masking of patients was maintained until analysis of the 6-month data was complete for the entire patient population.
According to Biegelsen, however,
Both groups of patients (the treatment and control groups) received calls from physicians so that the frequency of contact with patients was similar between the 2 groups. To ensure that physicians did not inadvertently disclose the patient’s randomization assignment, standardized clinician communication scripts were provided for use during patient contact. However, we were told that the conversations were very different with each patient group because physicians were forced to call patients in the control group for no reason other than to maintain a balance in the number of calls to each group.
Biegelsen reports that trial investigators have said that this does not represent a serious criticism of the trial, and notes that the FDA had signed off on the trial protocol.
Update: Senior management at CardioMEMS has assured me that they are not aware of any extraordinary problems associated with the blinding process in CHAMPION. Although difficulties in trial design and blinding often occur in device trials, CardioMEMS management is not aware of any serious problems that occurred in this area in the trial. The protocol and blinding process was worked out in details with the FDA and the steering committee. CardioMEMS also stated that it believes the FDA has only been conducting routine site audits and that the company is unaware of any serious problems associated with these audits.
Here is the description in the trial protocol of the procedures for handling phone contacts with trial subjects:
A matching phone contact will be generated to a CONTROL group subject when a contact is made to a TREATMENT group subject. The script for both contacts will be identical with the exception of the medication adjustment made to the TREATMENT group subject.
“Hello, (insert subject’s name), this is (insert clinician’s or coordinator’s name) from (insert institution). Thank you for taking your HF pressure measurements. At this time, we would like you to (insert treatment plan, i.e., increase lasix to 40 mg bid, etc.). Please continue to take your daily HF pressure measurements. Thank you.”
“Hello, (insert subject’s name), this is (insert clinician’s or coordinator’s name) from (insert institution). Thank you for taking your HF pressure measurements. At this time, we are not making any changes to your medications. Please continue to take your daily HF pressure measurements. Thank you.”
When a TREATMENT group subject is contacted, the site personnel will randomly select a subject from the CONTROL group and will contact that subject as soon as reasonably possible.
Finally, a blinding survey found that the vast majority of patients in the trial did not know the group to which they had been assigned, and there was no significant difference between the groups in correct guesses.