Digital Therapeutics Face Challenges With Control Conditions in RCTs

As a result of the potential of digital therapeutics to increase access to personalized care, many choose less stringent minimal control than typically seen in randomized controlled trials (RCTs).

For digital therapeutics (DTx) looking to gain regulatory approval, randomized controlled trials (RCTs) are critical; however, this technology faces challenges with traditional RCTs.

In a new review published in Frontiers in Digital Health, researchers highlight different approaches to control conditions and patient blinding in RCTs for DTx, specifically for those with psychosocial, cognitive, or behavioral content.

“DTx have the potential to revolutionize modern medicine by improving access to evidence-based, personalized treatments that may for some indications even evolve into first-line therapy,” the authors explained. These therapies “often translate face-to-face treatment modalities…into mobile application or web-based interventions.”

However, the DTx field is relatively new and there remains a need to standardize evidence generation to support medical claims and regulatory clearance, they added. Later-phase pharmacological RCTs use stringent placebo control conditions with pills that look exactly like the active treatment. Trials for psychological interventions are more difficult to design because it is challenging to identify which components of cognitive behavioral therapy and related psychosocial therapies can be included without producing a therapeutic effect. Placebo therapies in this space can have “low face-validity,” which runs the risk of unblinding participants.

The researchers conducted a narrative review to explore control conditions in DTx RCTs. A total of 14 RCTs were reviewed with most deploying 2 arms, with the exception of 2 studies, which ran 3 arms. Half of the trials used an unblinded waitlist or treatment-as-usual (TAU) control group and the other half used different forms of sham controls. However, 1 trial with 3 arms used an active comparator in one arm and TAU as the main comparison.

Among the studies using sham control, different terms were used including digital control, placebo, sham, and attention-matched placebo control. The design of these sham controls was also different, including fake but plausible therapies, general disease-agnostic well-being tips, and content missing the key aspects hypothesized to drive efficacy.

The authors noted that all the DTx with gamified cognitive training or remediation interventions used some form of sham control, such as controls that looked similar to consumer grade apps or video games.

None of the trials studied reported blinding checks. “In general, low rates of reported blinding checks are a known issue in nonpharmacologic treatments which DTx,” the authors wrote. It was often not described in the studies if participants were blind-to-assignment (ie, they knew they would either receive an active treatment or a placebo) or blind-to-hypothesis (ie, they expected to receive 1 of 2 potential treatments).

The researchers note that given the breadth of DTx, a one-size-fits-all approach for control condition design is unlikely. They came up with a list of recommendations for choosing control conditions:

  • Define the appropriate degree of stringency for digital controls—stringency of control conditions should reflect the risk profile and novelty of the DTx
  • Define minimal control level—similar to drug trials, control conditions should control for the incident effects of being in a trial or bias related to repeated assessments.
  • Inactive digital control conditions—any features and components designed to deliver therapeutic content should not be included in the control condition. Features related to motivated and regular engagement should be assess for any shared aspects with known treatment activities that may affect the outcome.
  • 3-arm studies—a third arm can help “elucidate real-world effectiveness of DTx while also providing the potential to compare to a more stringent or even active control condition.”

In addition, when it comes to describing control conditions, they recommended describing digital control conditions in detail and describing additional characteristics related to expectation setting and blinding.

Given the challenges identified, the low risk profile, and the potential to increase access to personalized care, many DTx choose less stringent minimal control or waitlist control, which can overestimate treatment effects, but may be a viable option in large-scale real-world studies, the authors noted.

“Indeed, in light of the iterative nature of software development, there has been a call for more innovative real-world data approaches to evidence generation for DTx beyond classical RCTs,” the authors concluded. “DTx are uniquely positioned to collect real-world data, engagement patterns, user reported outcome data, and/or clinically relevant digital phenotypes directly through their software application to assess their real-world engagement and effectiveness.”

Reference

Lutz J, Offidani E, Taraboanta L, Lakhan SE, Campellone TR. Appropriate controls for digital therapeutic clinical trials: a narrative review of control conditions in clinical trials of digital therapeutics (DTx) deploying psychosocial, cognitive, or behavioral content. Front Digit Health. 2022;4:823977. doi:10.3389/fdgth.2022.823977