Protocol deviations; tell-tale signs of patient non-adherence during clinical trials

When discussing patient adherence during clinical trials, most people will say that on average it is high and that the issue of non-adherence really surfaces during clinical care. But is this true or are we missing something in clinical trials? I think we are missing something, and the issue is far greater than we think it is.

Early drop-out during clinical trials has always been and still is an ongoing challenge, and despite all efforts its rate still is at a 30% average. This triggered me a few years ago to start working on a solution to improve patient adherence during clinical trials, as early drop-out is an extreme form of non-adherence. However, while it makes perfectly sense to use this definition of early drop-out I could never find hard data on patient adherence levels during clinical trials. There are numerous research reports and articles published on early drop-out, but I never found one on patient adherence during clinical trials. Until now.

Tell-tale signs hiding in your data

A few weeks ago, I had an interesting conversation with a good friend of mine, Charles Chen. His company, Zytonomy, is developing a team-centric platform to support clinical trial sites, site staff and patients alike. A long time ago we already agreed that running a clinical trial is very much about a true partnership and collaboration between all parties involved. For that to happen each partner needs to be able to get the right information and the right support at the right time. Charles referred to a published article on protocol deviations¹, which provided him inspiration on the challenges his platform could to provide a solution for. That same article finally provided me an insight into the potential level of patient non-adherence during clinical trials.

The authors of the article analyzed the protocol deviations of 1.387 smaller studies of which 126 (9%) were sponsored by pharma companies. In total 5.405 participants were involved of which 1.103 (20,4%) in pharma trials. Of these studies 73 reported a total of 447 protocol deviations.

And here is what they found:

• 402 (89,93%) deviations came from 63 pharma studies
• 266 (66,17%) of 402 related to Study Procedures
• 86 (21,39%) of 402 related to Safety
• 44 (10,94%) of 402 related to Informed Consent Document
• 6 (1,49%) of 402 related to Eligibility
• 281 (69,90%) of the 402 deviations were major.

Now some of you will say that these are small studies and in general the total number of protocol deviations for a larger, more complex trial is much higher. I agree with that, but what intrigued me is the ratio between the different types of deviations. I have discussed this with peers and experts in our industry and they seem to be on par with what we see in larger clinical trials. The interesting part is what measures were taken to prevent these deviations from recurring in the future.

A staggering 67% linked to patient non-adherence?

Only 26% of the measures to manage the major deviations were focused on the patient. The article does not mention what measures were taken for the remaining 124 minor deviations. But if we take a closer look at the details, then we see many deviations where the trial team had no full control over the protocol and patient behavior may have played a decisive role.

• Participant seen outside window period (126)
• Missed assessment (92) (related to missed visits?)
• Study procedure in protocol not followed (76) (patient and/or site induced?)
• Prohibited medication taken (4)
• Stopped study medication (2)
• Wrong version of diary used (2).

Following this rationale, 302 (67%) of the 447 deviations can be linked to patient non-adherence.

Does your data show the same or am I wrong?

To me this is a worrying conclusion. COVID has accelerated the adoption of virtual and decentralized clinical trials (VCTs and DCTs) out of necessity. One challenge may be an unexpected increase in patient non-adherence as trial teams cede some control to the emerging tools and platforms used to conduct these trials.  While VCTs and DCTs can offer convenience for the patients, the successful execution of a clinical trial protocol often has a very human element that emerges from the interactions and communication between investigators, patients, and site staff. In this new world of VCTs and DCTs, will digital interactions help reduce deviations and lead to faster, safer, more efficient trials? Or will we see an increase in deviations and non-adherence?

What does your data say? Have you seen the same ratio? Is my rationale wrong?

¹ An Audit of Protocol Deviations Submitted to an Institutional Ethics Committee of a Tertiary Care Hospital