In our last post on protocol deviations, we looked at the mechanics of categorizing and classifying PDs. In this post, we’ll look at strategies for categorizing and classifying.
As we discussed in an earlier post, we categorize protocol deviations (group them into topics) so we can assess compliance. We classify protocol deviations (assess their severity) to help us respond to non-compliance appropriately, to identify which deviations are subject to reporting to IRBs/IECs, and to determine which subjects should be removed from the per-protocol analysis. We need to categorize and classify in ways that will support those study-specific objectives.
Some teams find it useful to categorize deviations by topic – e.g., informed consent, laboratory assessments, imaging, questionnaires, con meds, etc. Others find it more useful to categorize by process – e.g., eligibility, missed observation or procedure, out-of-window observation or procedure. The team needs to decide what method would work best for evaluating compliance for that particular study: which method will produce a better signal-to-noise ratio?
Most sponsors divide protocol deviations into two categories, which may be called Major and Minor, Important and Other, or another variation. Definitions for these categories are frequently difficult to pin down. Sometimes, classifications are tied to categories: for example,
Other classifications schemes are independent of categories: for example,
Some sponsors incorporate ICH E3’s (Structure and Content of Clinical Study Reports) definition of “important” into their classification definitions. E3 states that “all important deviations related to study inclusion or exclusion criteria, conduct of the trial, patient management or patient assessment should be described” in the clinical study report. It goes on to say, “In the body of the text, protocol deviations should be appropriately summarized by center and grouped into different categories, such as:
Other sponsors focus on ICH E9’s (Statistical Principles of Clinical Trials) definition of the “per-protocol” analysis which includes only subjects who are “more compliant” with the protocol, such as
In our opinion, protocol deviation classification during study conduct should be based on a scheme that will help the study team to respond to non-compliance in real time, not on schemes that are intended for post-hoc reporting or analysis. The study team should consider which conditions should generate a flag that would result in some kind of corrective or preventative action. Another option is to forego classifying individual deviations throughout the study, but rather classify trends.