This past July, Happify Health made headlines when it announced the launch of Ensemble, the first prescription digital therapeutic to treat both depression and anxiety. Along with excitement, there have been many questions about why and how we developed this landmark product, so I want to share some details and reflect on the journey we took to get here. 

The Ensemble story starts in the spring of 2018, I was brought on as Happify Health’s first Head of Digital Therapeutics (DTx) after a couple of years of working with the company as a consultant. My task: Explore how we might develop a prescription digital therapy (PDT) that would meet the FDA’s requirements to clinically treat anxiety and depression. 

By this time, Happify Health had developed a robust digital platform with 65 different therapeutic tracks (one-month programs) that had been on the market since 2013 and was regularly listed among the top wellness and mental health apps. It was popular with consumers—so popular, in fact, that Happify had recently expanded to partner with, and provide services to, large global and national employers and health plans including Cigna. By 2019, nearly 4 million people had used Happify, which was available in 8 different languages (now it’s available in 10). 

Happify had also recently completed a randomized controlled trial (RCT) and published the findings in the peer-reviewed International Journal of Wellbeing. The study included more than 1,000 patients and demonstrated that people who used Happify as recommended (completing at least 2-3 activities a week) reported a 25-30% reduction in anxiety and depression symptoms, as measured by PHQ-9 and GAD-7,after 8 weeks.

So in 2018 we asked the following questions: 

  1. Why would we want to create a PDT? 
  2. What would a PDT look like versus our existing product? 
  3. How would we go about creating the PDT?
     

HAP_016_Twitter_Graph-01Why would we want to create a PDT? 

While non-clinical meditation and happiness apps and mood trackers may be adequate for some, others simply need more support. Creating a PDT would enable us to go beyond self-care that addresses the causative factors of stress, depression, and anxiety (which our consumer app effectively does) to actually help treat two diagnosable conditions—Major Depressive Disorder (MDD) and Generalized Anxiety Disorder (GAD). 

We envisioned our PDT as something that could be prescribed on its own or as an adjunct therapy that could be used in conjunction with, and encourage better adherence to, other treatments. In other words, our PDT would not replace medication or in-person therapy, but rather serve as another means to the same end—to make it possible for more people to improve their mental health so they can live their fullest lives. 

Why do we need another way to treat mental health disorders? Consider the following: More than half of all U.S. adults with a diagnosis of mental illness receive no treatment at all2; and of those who access care for depression or anxiety, less than 35% receive appropriate treatment.3 This gap in treatment access and low quality of care is due to a number of factors like stigma, cost, provider shortages, long wait times,4 and low efficacy rates of antidepressant medications (ADMs).5

HAP_013_CBT_Chart_V7 (1) (1)

Cognitive Behavior Therapy (CBT) has been shown to be as effective as ADMs6; however, the way that in-person CBT is currently delivered puts a significant burden on the patient. A common 12-week CBT program requires at least 24-36 hours when accounting for travel time, waiting time, time in therapy, etc. Moreover, hour-long weekly appointments can be difficult to fit into one’s work and/or childcare schedule. Finally, there is a significant shortage of qualified therapists who practice CBT, of which only a few accept health insurance, and there are often month-long (or more) wait times to get access to these therapists. For these reasons, many people never start therapy or stop going before seeing any benefits.

Integrating digital therapeutics can redefine the member experience. Read our  guide to see how! →

Essentially, we saw a large unmet need for a PDT to treat symptoms of MDD and GAD, and assist in the clinical management of these disorders. And given that the Happify platform already was built on the evidence-based interventions of CBT, we were in a perfect position to create a more potent digital therapeutic for those living with depression and anxiety.  

What would a PDT look like versus our existing product? 

We knew we couldn’t just take the Happify app through an FDA process to get it cleared as a PDT. But we initially weren’t exactly sure how it would need to be modified. One thing was clear—we would have to find a way to maintain our industry-leading rates of activation and engagement. 

Happify was developed based upon human-centered design principles and was extensively A/B tested over many years to optimize the user experience. As a result, we see high levels of retention: 64% of Happify users are still active after 2 months, and after 6 months, 48% are still active. In comparison, the industry averages for continuous use/retention on health & fitness and medical mobile apps after 1 month is 4% and 3.5%, respectively.7  

HAP_012_Graphic_NoHeaderSo we knew we had to design a product that was more clinically potent without sacrificing engagement. Since we ultimately would be seeking FDA clearance, we initiated a pre-submission meeting with the FDA to share our ideas and get feedback. After that meeting, we had a better understanding of requirements regarding dosing, clinical integration, structure and order of the therapy, potency of the interventions, and patient monitoring and risk management. We were ready to get started. 


How would we go about creating the PDT? 

Armed with the guidance that we received from the FDA to develop our PDT product, we  reviewed the literature on CBT for MDD and GAD and found that there was powerful support for creating one transdiagnostic product that would treat both disorders, rather than creating two separate products. Up to 50% of patients who have MDD will also meet the criteria for GAD,8 and many of the symptoms of the two disorders overlap.9 GAD diagnosis is often overlooked by clinicians, MDD and GAD diagnoses are often confused by clinicians, and the primary diagnosis can change from one to the other over the course of treatment. So Ensemble was designed to treat both MDD and GAD, delivering a sequence of interventions based on CBT, positive psychology and mindfulness; and this is why we chose to name the product Ensemble, “a group of items viewed as a whole rather than individually.”

Concurrent with this, we made a major strategic decision regarding our Quality Management System (QMS). Prior to developing our PDT, our existing quality and security program had passed audits by our health plan and employer customers, yet it didn’t satisfy all the requirements of the Good Manufacturing Practices (GMP) required by the FDA for a PDT. We realized that these GMP requirements would be substantial and unnecessary for our health plan and employer products, but we also didn’t want to have two different systems, so we standardized our products on one FDA GMP QMS.

Then in April 2020, shortly after the Covid-19 pandemic emerged, the FDA released an Enforcement Policy for Digital Health Devices For Treating Psychiatric Disorders During the Coronavirus Disease 2019 (COVID-19) Public Health Emergency for psychiatric disorders like MDD and GAD. This allowed us to bring our new PDT to market more quickly with the opportunity to collect clinical data to demonstrate safety and efficacy for subsequent FDA regulatory filings. This accelerated our efforts to finish the development of our PDT for MDD and GAD, and to bring it to market under this guidance. 

Here are the seven steps we went through to design, develop, test, and bring our new PDT to market. 

HAP_019_ProcessVisual_LinkedIN_V2

  1. Our clinical leaders created a framework based upon best clinical practices to treat MDD and GAD. This included types of therapies, sequence of interventions, key clinical objectives to achieve over the course of therapy, etc. 

  2. We selected elements from among our 3,000 activities that we thought were appropriate to treat MDD and GAD. We also identified gaps where we thought we would need to add to or modify existing activities.

  3. We added clinical instruments into the product to measure patients’ progress over time. We selected the PHQ-9 and GAD-7, utilized as outcome measures in our previous RCT, as they are validated tools designed for use in clinical practice.10 Patients are prompted to complete these assessments every two weeks. We knew that this would be a source of significant value to clinicians since less than 20% of them currently use these instruments on a regular and recurring basis with patients to measure and track the progress and impact of therapy.11

  4. Given that self-harm and suicidality are salient risks for this population, we added features to screen for this risk to provide relevant resources for the patient, and to also share this information with care teams to further support patients.

  5. We determined a way to provide clinical integration so that care teams could track patient progress via Electronic Medical Records (EMR). We developed reports detailing data and trends on engagement, risk, and clinical benefit.

  6. The American Medical Association (AMA) defines medication adherence as 80% compliance with dosing and duration. While consumers using the Happify app can use it as frequently as they like, the FDA requires clear dosing and duration guidance on PDT labels, so we determined two intervention activities per day (112 interventions total) could be completed in 8 weeks, but we would give patients 10 weeks to complete the therapy to be consistent with the 80% definition.

  7. We did all of this in a patient-centric way that leverages Happify’s many engagement features, game mechanics, game principles, and other tools and resources to maintain high user satisfaction and engagement.

Years ago when Happify first started, investors told us that you only know software works after it’s been tested on hundreds of thousands of people. A major strength of Happify Health compared to other PDT companies is that by initially focusing on the DTC market, we were able to test our products with millions of users in real-world settings, as opposed to a few hundred people in tightly controlled RCTs. Limited use makes it tough to know whether or not a product will deliver a user experience resulting in clinical benefit in the real world.

As with all SaMD (Software as a Medical Device), Ensemble is a living treatment that will continue to grow and evolve over time, but when we look back at the thousands of hours of work that went into this and the incredible team that created it, we are excited and humbled by the result. We maintained the personalization and high level of engagement found in the Happify wellness app, while building something that we have every reason to believe will meet FDA standards for approved medical devices.

By first establishing ourselves as a DTC company, we have been able to develop and refine a product that surprises and delights people. Our testing and experience allow us to create highly engaging and effective therapies that now can be delivered at a clinical grade to treat psychiatric disorders and potentially help millions of people. 

 

Request a Demo of Happify Health

 

Reference List

1 Parks et al., 2018 - link

2 2020 State of Mental Health in America [Mental Health America] - link

3 Young et al., 2008 - link; Young et al., 2001 - link

4 America’s Mental Health 2018 [Cohen Veterans Network & National Council for Mental Wellbeing] - link

5 Trivedi et al., 2006 - link

6 DeRubeis, Siegle & Hollon, 2008 - link; Butler et al., 2006 - link

7 Statista.com - link

8 Lamers et al., 2011 - link; Moffitt et al., 2007 - link; Kessler et al., 1999 - link

9 APA (2013). Diagnostic and statistical manual of mental disorders (5th ed.)

10 Kroenke et al., 2001 - link; Spitzer, Kroenke, Williams, & Löwe, 2006 - link

11 Jensen-Doss et al., 2018 - link