The adherence illusion: Digging into the root cause of variability in clinical trials

To a large extent clinical trials can be thought of as an exercise in managing variables: the underlying purpose is to understand how, under the same controlled conditions, different responses might be triggered by defined inputs, whether in terms of active dose or placebo.

It is important to remind ourselves, however, that these conditions are not always controlled to the extent that we might assume. In certain situations, the framework or methodology for the trial might allow sources of variability to hide in plain sight, unknowingly introducing the potential for unreliable output data to be generated and, as a consequence, misguided conclusions to be made.

?

Take as an example the Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist (TOPCAT) trial[1]. Here, investigators identified significant differences in the clinical profiles, event rates, and responses to spironolactone between trial patients who were enrolled in the Americas (United States, Canada, Brazil and Argentina) and those enrolled in Russia and Georgia.

?

To investigate these disparities, further analysis was carried out on serum samples taken during the trial period from patients who were assigned to receive spironolactone and who self-reported taking the study drug at the 12-month visit. The purpose of this analysis was to establish concentrations of canrenone, an active metabolite of spironolactone, in order to provide robust evidence of true drug exposure levels among trial participants from the two regions. In short, an absence of canrenone would indicate non-adherence, counter to the participants’ self-reported claims.

The results were eye catching. They showed that among participants from the US and Canada, canrenone concentrations were not detected in just three of the participants who reported taking assigned spironolactone, equating to an adherence level of 96%. However, for Russian patients who reported taking assigned spironolactone, the adherence level was recorded at a far lower level of 55%, with canrenone concentrations undetectable in 30 of 66 participants.

?

Further analysis was conducted to investigate median canrenone concentration levels among both the self-reporting populations from the two regions and the populations where the presence of canrenone was independently verified. These findings were broken down according to the spironolactone dose reported by the participant (15mg, 30mg or 45mg).

?

Across all patients where canrenone was detected, there was correlation between dose amount and concentration level. The same was not true for the self-reporting group, however. Here, correlation was strong among 75% of the US population who self-reported being adherent, leaving 25% of the patients out of the analysis. On the other hand, there was no evidence of correlation among the Russian population who self-reported being adherent, as 45% (30/66) did not do so.?

?

At the level of this trial alone, these findings provide an interesting insight into discrepancies between analysis methods. Excluding 25% of the US population because they self-reported to be nonadherent and including unrecognized nonadherence based on self-report in Russia are two unsatisfactory approaches to analyse trial data resulting in confusing findings.

?

Undoubtedly more significant, however, is the revelation that flaws in trial methodologies can impair the findings and corrupt the validity of a study if undetected. Claims of self-reporting, while potentially reliable – as in the case of the US participants – also have the potential to be highly unreliable. Indeed, they can fundamentally undermine the purpose of the study, with investigators in this case stating that the results obtained “do not reflect the true therapeutic response to spironolactone”.

?

This is deeply concerning given that adherence to the dosing regimen specified in the protocol should be regarded as one of the critical variables to be controlled during a clinical trial. Indeed, it is only by truly knowing drug exposure that sponsors can truly rely on the data outputs and consequent judgements on efficacy and safety. And this can only really be achieved by avoiding flawed self-reporting methods for adherence measurement and adopting more reliable digital technologies instead.

?

For trial co-ordinators and sponsors, this serves as an important warning that while the existence of compliant patients can create the illusion of controlled conditions, the process is not truly ‘watertight’ if non-compliance can also exist undetected. When it comes to self-reported adherence, all is not necessarily as it seems.

?

Establishing hard evidence for your trial…??

AARDEX has a best practice methodology, independent of any device package or software platform. Utilising our expertise and experience in medication adherence and patient compliance we acquire, monitor, analyse, guide and interpret data to deliver absolute clarity and bring confidence to sponsors, trialists, and ultimately, patients.???

AARDEX is the only mature, robust and proven adherence solution on the market today, one that maximises the reward, mitigates the risk and delivers resolution for your clinical trial. All delivered with the clarity, integrity and certainty you need to proceed with complete confidence in the exposure-response.???

[1] https://biolincc.nhlbi.nih.gov/studies/topcat/