Digital placebo: Why it matters and what makes a good one?

Traditionally a placebo is an inert substance or treatment which is not designed to have a therapeutic value. In clinical trial it serves as a control and patients usually aren’t aware if they are taking it or the experimental therapy (this condition is called blindness).

Should digital medicine consider a digital placebo too?

Why placebo matters in clinical development?

Believing that you’re being treated may have a positive impact on your health status. Expectations and enthusiasm themselves often have a positive impact on clinical outcomes.

Before digital therapeutics, placebo took form of “sugar pill”, saline injection and so on. But with digital therapy the solution isn’t so straightforward.

Do digital therapeutics need placebo?

First of all, let’s consider if a placebo-controlled study is needed for DTx too or not.

Two systematic reviews (and meta‐analysis) of randomized controlled trials on smartphone-based mental health interventions may help us answering this question. The first one considered apps aimed to reduce anxiety, the other one considered apps for depressive symptoms reduction.
Experimental apps appeared moderately effective in reducing depressive symptoms in studies where smartphone apps were compared with wait-list control groups, but only a small effect was observed when active control conditions were used.

The diminished effect sizes of smartphone interventions when active control groups are involved, suggests the role a digital placebo may play in these settings, consistently with what happens also in non-digital clinical studies.

Thus, a digital placebo may be beneficial.

What makes a good placebo? (Digital or not)

To properly work, a placebo must be credible.

The patient shouldn’t be able to tell if (s)he’s taking the experimental treatment or placebo, otherwise “placebo effect” will vanish. Without a credible digital placebo, it becomes the same as comparing DTx solutions to a waiting list.

How a digital placebo is made?

While for drugs, it’s sufficient to keep excipients without adding the active ingredient, you can’t just remove the content/interactivity from a digital tool and pretend users (patients) won’t notice it.

Of course DTx companies may capitalize on already existing architectures (the one of their DTx candidate or pre-existing ones) but they’ll probably have to put efforts on development.

Digital placebo examples

Great examples comes from two digital therapies we already discussed here on DTxExplained.

Video-game placebo

Akili’s project Evo is a digital therapeutic for ADHD which looks like a video-game. «When your treatment looks like a video game, you ask yourself what kind of placebo can be used,» said Vincent Hennemand, Akili’s Vice President of Strategy. «If it’s a randomized, double blind trial, you have to show something that is also a video game. We couldn’t find a product on the market that would fill the role of placebo, so we developed our own: it’s an educational word game like Boggle. It’s a video game with a cube with letters; you “shake” it, and there’s a random distribution of letters, and you have to invent words.»

A placebo for online cognitive behavioural therapy

Sleepio offers a web-based cognitive behavioural therapy (CBT) course to treat insomnia. The course is delivered through the online platform by an automated virtual therapist: “The Prof”. They compare it in clinical trials «with a credible placebo; an approach required because web products may be intrinsically engaging, and vulnerable to placebo response».

The same application platform, design and execution principles as for digital CBT, but with no known active therapeutic ingredient, was used as placebo. DTx and placebo were both delivered by “The Prof”. Sleepio creators also used a credible name for it, even in case trial participants would have googled it: “imagery relief therapy”.

Take-home messages from the first digital placebos

The above-mentioned examples represent two clever ways to overcome the need for an appropriate control group, which is a primary aspect to consider while designing clinical trials.

It is to be noted that digital placebo usage doesn’t prevent the possibility to combine additional therapies, like may happen in traditional trial. In the Sleepio example, indeed, usual care continued in both groups, consistent with a real-world setting. What’s important is that participants (and possibly researchers too) cannot determine if they are included in the experimental or control group

Digital placebo development

Creating sham apps that can serve as placebos seems to be the best solution. But digital placebo development is challenging and won’t be for free. A budget has to be allocated and a sufficient amount of time for development has to be considered.

Obviously we haven’t to reinvent the wheel. If a DTx is to be tested in a clinical trial, a digital platform / app architecture already exists. The first step, at least to evaluate, should be considering keeping the same “wrapper” and filling it with different contents. For example a virtual coach may be replaced just with basic disease info and recurring motivational quotes.

Or instead of also delivering a digital therapy, a placebo version could just collect patients reported outcome (PRO), without providing any customized feedback on them.

Digital placebo development: so far so good, but…

Since in digital health it’s not easy to know in advance which features can be “digital active ingredients” and which make just inactive but credible placebos, we have to carefully plan and design a digital placebo. If, on one side, it is mandatory to develop a credible sham app, on the other side there is the risk to develop an inadvertently active digital treatment. This would erroneously reduce the perceived efficacy of the tested DTx.

Which are the available alternatives to digital placebo?

In some fields (such as when behaviour is involved) it is difficult to deliver credible sham therapy, so non-placebo control groups are often involved. Active control groups (e.g. treatment as usual) should be more reliable than inactive ones (e.g. wait-list), but both types don’t control for participant expectations of high-tech treatments and the empowerment allowed by these self-management apps.

Head to head trials may represent an additional alternative to placebo-controlled trials. However it could be limitative. Indeed, head to head studies have to demonstrate at least the same efficacy of another treatment already approved by regulatory bodies for that indication (non-inferiority trials).

Head to head studies can be more demanding, requiring an higher efficacy than the one which could be sufficient for these novel therapies. For example, it could be fine to have both a standard treatment with, say, “therapeutic effect 100” and a DTx with “therapeutic effect 80”, considering also it will be probably safer and cheaper; clinicians then would have an additional therapeutic option and could prescribe the most appropriate treatment for each patient. A head to head study, however, would reject the DTx in this scenario.

Going further

Kaia example: back pain DTx vs guideline recommended intervention

Kaia Health was even more audacious: their app-based back pain treatment outperformed the treatment currently recommended by clinical guidelines.

Their control group consisted of standard physiotherapy combined with an online program with high-quality educational content accompanied by motivational messages. This is a remarkable strength in their study design. Indeed, the structured education regarding back pain (recommended as first line treatment by current guidelines) was very probably even more emphasized in the control group than in usual care conditions.

Summing up

Whenever possible, a credible digital placebo should be developed and used to ascertain how much of the efficacy can be attributed to the treatment itself.

However, sometimes it could be not feasible to develop a credible one. In that case a comparison with standard care or a waiting list could be fine.

As well as in case DTx companies are so confident to challenge recommended treatments (like in Kaia example).

In each case, please go ahead generating high-quality evidence to validate DTx!