Breaking the Blind in Clinical Trials & Reporting to Health Authorities, Investigators & IRBs/Ethics Committees

Jan 30, 2013
Bart Cobert

Pharmacovigilance, Drug Safety and Regulatory Affairs Author & Expert

This is a topic that has caused much consternation in pharmaceutical companies and has, in my experience, caused some clinical trials and even NDAs to fail due to an inadequate number of analyzable patients.

The issue here is the breaking of the blind on individual patients in clinical trials for the purpose of submitting an expedited (15 or 7 day) report to FDA, EU member states and other health agencies.

Let’s first review the background and the regulatory requirements for unblinding.

For many years after the concept of expedited or alert reports (or as FDA calls them, IND safety reports) of individual cases was introduced, the agencies accepted blinded reports. Companies rarely broke the blind – the usual reason when this was done was for immediate treatment of the patient in an emergency situation. When the blind was broken the expedited report had the unblinding included and the FDA and other agencies were notified as were the investigators and IRBs/ethics committees. How these cases were interpreted, how signals were analyzed, how benefit-risk analyses were done is unclear. Data Monitoring Committees (DMCs) were the exception rather than the rule in the past so this level of oversight was not present. In practice, unless there was a cluster or bunching of similar SAEs/SARs/SUSARs (however you wish to call them), there was little benefit-risk analysis done during the trials. Presumably the patients were at greater risk.

However, in the last several years the FDA, EMA and other health agencies have made it clear that they now expect all expedited reports to be submitted with the blind broken. This is consistent with the general (and laudatory) trend to minimize patient risk and maximize benefit-risk analysis by doing so in an ongoing and continuous manner by multiple groups: the company’s safety and clinical research groups, the IRBs, the investigators, DMCs and adjudication committees amongst others and, of course, the health agencies.

Aiding in this change has been the improvement of drug safety databases – both commercial and home grown ones. In the early days, databases could only produce one MedWatch/CIOMS I form per case at any one time and it was either blinded or unblinded. Furthermore, it was often not possible to limit access to different people who logged onto the database. That is, everybody saw all or nothing. Now with much more sophisticated databases, it is possible to have multiple versions (i.e. blinded and unblinded) of the same case and it is possible to limit the access to the unblinded case to only certain people. This has made the operational aspects of unblinding somewhat easier.

By the way, some companies prefer to use the terms masking or unmasking rather than blinding or unblinding. This has usually been by companies making ophthalmologic products. They felt that talking about “blinding” of studies was insensitive and inappropriate.

The Regulations

FDA/Competent Authority Notification

FDA has made it clear that they now expect all expedited reports submitted to them to be unblinded. However, as FDA notes in 21CFR312.32, if patient safety can be assured without breaking the blind, FDA encourages sponsors to discuss alternative reporting arrangements with FDA..and described in the protocol. That is, if the company anticipates the need to break the blind frequently, a mechanism to keep the blind in place or to have very limited unblinding and safety analysis (e.g. by a DMC) should be put in the protocol before starting the study.

So, for the FDA, the blind should be broken and the case reported unblinded to FDA if it meets the expediting criteria (serious, unexpected, possible relation to the study drug).

In the EU the regulations are explicit:

  • Treatment codes should be broken by the sponsor before reporting a SUSAR to the competent authority (that is, the health agency) and the Ethics Committee/IRB.
  • When a serious AE comes in and if it might be a SUSAR, the blind should be broken only for that patient by the sponsor even if the investigator has not broken the blind. Then 3 possibilities occur:

1. The patient took the test product: this is an expedited report.
2. The patient took a marketed comparator: the SAE should be assessed for expectedness according to the labeling (IB or SPC as appropriate) for that product. If it is unexpected then the SUSAR is expedited; otherwise it is an expected SAR and not expedited.
3. The patient took placebo: Events associated with placebo will usually not satisfy the criteria for a SAR and therefore are not expeditable.

Now things get interesting. How are the IRBs and investigators handled in the US and EU?

Previously, the investigators and the IRBs were sent the case as it was sent to FDA – usually blinded. Many investigators and IRBs complained that a blinded case was useless in terms of assessing risk. Others presumed it was always the study drug even though this might not have been the case. With the changes in the regulations, things are now more complex.

Investigator Notification

In the US: The sponsor must report to any participating investigators under all open INDs, including those held by the sponsor and Investigator Initiated Research where the sponsor provides the investigational drug. Follow-up reports should be sent to investigators about an important serious adverse reaction, if it significantly affects the care of the subjects or conduct of the study. Minor refinements that do not significantly affect care of subjects or conduct of the study must be sent to FDA but need not be sent to investigators. Such information may be communicated to investigators in a routine update of the IB. The timing is usually 15 calendar days with the clock start the same as the clock start for the FDA report.

So this is a bit nebulous; it suggests that all initial reports should be sent to all investigators though it is not explicit about unblinding. Follow ups should be sent if important. This then puts the decision on what to send in the hands of the sponsors.

In the EU: A bit more explicit and prescriptive. The sponsor must inform all investigators concerned of findings that could adversely affect the safety of study subjects. The information can be aggregated in a line listing of SUSARs accompanied by a summary of the evolving safety profile of the product. In the case of blinded trials the line listing should present data on all SUSARs, regardless of the medication administered (e.g. active/placebo), maintaining the blind. A significant safety issue, whether an ICSR, aggregate data or a safety issue that impacts on the course of the study, should be communicated asap to all investigators. Note that this does not actually specify 15 days for communication to investigators.

IRB/Ethics Committee Notification:

In the US: Investigators are required to report promptly “to the IRB… all unanticipated problems involving risks to human subjects or others,” including AEs that should be considered unanticipated problems (CFR56.108(b)(1), 312.53(c)(1)(vii), and 312.66). The investigators may rely on the sponsor’s assessment and provide to the IRB a report of the unanticipated problem prepared by the sponsor. The sponsor and IRB may agree that the sponsor report directly to the IRB copying the investigator(s).

So now the question becomes what is an unanticipated SAE? In general, an AE is considered unanticipated and reported to the IRB, only if it were unexpected, serious, and would have implications for the conduct of the study which means requiring a significant, and usually safety-related, change in the protocol, such as revising inclusion/exclusion criteria or including a new monitoring requirement, informed consent, or IB.

An individual AE ordinarily does not meet these criteria. The major exceptions to the general rule are SAEs that are uncommon and strongly associated with drug exposure, SAEs that are not commonly seen with drug exposure, but are otherwise uncommon, multiple occurrences of an A or, an AE with an increased specificity or severity.

In the EU: The sponsor shall inform all concerned investigators about SUSARs that could adversely affect the safety of subjects. This may vary by Member State. MS’s may require that an Ethics Committee receive only cases from that MS. In this case, it is strongly recommended that all SUSARs from other Member States and from third countries, are periodically reported to the ethics committees at least every 6 months as line listings and a brief report on the main points of concern.

Clear? Not really.

The Realities in the Workplace

So what is the reality in the workplace? Since the US and EU rules are not really harmonized any more, the situation may arise where a case will be sent to the EU and not the US or vice versa. Several methodologies have arisen to try to work through this:

  • The most common method I have seen is to break the blind in the company (or CRO) drug safety group for every case that could be expedited (that is, serious, possibly related, unlabeled/unexpected). If the drug is the placebo, then no expedited report in the US or in the EU (usually). If the comparator, choice of reporting to the manufacturer of the comparator or (most common) report to the health agency as an expedited report without notifying the other manufacturer. If it is the sponsor’s study drug, expedited report with notification to all “concerned” investigators (this is defined in various ways in each company) and some or all IRBs (usually all). In general, all investigators and IRBs are informed.
  • The next variant is the same as above but not informing investigators or IRBs unless the case will alter the study by changing the protocol (inclusion, exclusion, new test, change of regimen etc). Thus the rest of the company as well as the investigators and IRBs are still blinded or not informed of the case at all. This can be tricky to do as the expedited report sent to the health agencies must be unblinded but must not “escape” into the rest of the sponsor’s team (clinical research, regulatory, CROs etc.). This preserves the blind and keeps the patient valid if the patient continues on the study. If the patient drops out of the study due to the SAE then that case is sent to the investigators and IRBs.
  • Unblinded reporting to the health agencies and line listings or periodic summaries to investigators and IRBs/ethics committees.
  • Another alternative involves unblinding for the Data Monitoring Committee but no other groups (IRB etc.) or only a limited number of people in the company.
  • Another method, is to let the CRO (if there is one) handle the blind break and informing the IRB, investigators etc. as agreed in the protocol, and inform the company (perhaps only the head of PV) of the unblinding and keep everyone else at the company/sponsor blinded. Someone, either at the CRO or the company, must still do the safety signaling and trend analysis.

Comments & Opinion: So what should you do?

  • If you unblind too many cases and if these cases become invalid for the statistical efficacy analysis in the clinical trial and NDA/MA, the entire trial and NDA/MA may lose statistical significance and fail. Companies should thus be very sure that the case must be unblinded for reporting purposes before unblinding.
  • If the company does have to unblind but the case does not meet the threshold for reporting to investigators and/or IRBs, then the blind should be maintained within the company as best as can be to keep the case valid for the statistical analysis. Set up a system to keep the case unblinded only to a limited number of people in drug safety. Do not let the unblinding “escape” from the safety database – be sure access to the case is limited so that those who shouldn’t see, don’t see.
  • Set up, where feasible, Data Monitoring Committees and consider writing into the protocol a mechanism to allow the DMC to take more responsibility in safety (e.g. trend analysis on unblinded data while the company does trend analysis on blinded data) where feasible and where the health authorities allow this.
  • Build into the trial sufficient patient numbers so that even if some patients drop out, quit or have SAEs that require unblinding, there is a sufficient valid number of patients left in the trial for the efficacy analyses. This may be more expensive up front but might save the company from having to repeat a clinical trial!

And yet another comment/opinion:

This incredible complexity is consistent with the path that drug safety and pharmacovigilance is taking (or perhaps more aptly is being forced upon us), namely, the deharmonization and increase in complexity in the operational aspects of clinical research and drug safety. No doubt the intentions are good and noble: patient protection and public health. But this bureaucratization and complexity is making the business of clinical research and pharmacovigilance very difficult for regulators and companies. And there is not a lot of good evidence that all this complexity is helping the patients and bettering public health.