May 2, 2019

ROI in Digital Health: Is Anyone Making Money?

Chandana Fitzgerald

Headlines for the last 6 years have been about the amount of money being invested. Now we are starting to be asked about returns.What is the ROI on Digital Health? Return on investment (ROI) is king in every sector, healthcare not spared. Jargon like “soft ROI” and “social ROI” are simply not satisfactory. We want to see demonstrable benefits in real dollars.

In this week’s blog, I am not going to apply economic theories as I have no formal training in economics besides the token modules in my MPH. However, we will make this a qualitative discussion because ROI is fundamentally important in driving the adoption of such technologies in healthcare and helping us get a bang for our buck as an investor, pharma company, or as a consumer.

The Problem

In digital health, we use a whole lot of undefined and non-uniform terminology to describe value - behavior change, patient engagement, beyond the pill, care pathway optimization - the list is pretty long! Do we even understand what a lot of these really mean? How much does ‘engagement’ cost, what returns will it give us? I am certainly one to look at returns on monetary health investments beyond just the dollars - there are clinical returns like faster remission, reduced recurrence, etc; there are epidemiological returns like reduced incidence and prevalence of a certain condition in a population, there are social returns like improved quality of life, etc. However, the bottomline is we also need financial returns on top of all of the above - nobody wants to pay tax into vague concepts being reimbursed, or capital into wishy-washy foundations. Digital has successfully reduced cost and improved service in other industries.

Much of what is reported so far in the sector of digital health by financial analysts consists of predictions of revenue, market share, and potential growth. This is predicated upon adoption rates (real or estimated), on potential reach within a population, on known disease burden/ epidemiological studies. That is a fair assessment, however, just not true or proven yet.

Apples to Apples

What all must we account for while measuring a monetary return on our digital health investment? Cost of solution A vs solution B (where B can be a drug, another tech intervention, or no intervention). Labor/ staff costs, implementation costs - training staff, patients, marketing, etc. Technology costs - iterations to the tech as we go. Easy, right? The problem is, unlike a drug and diagnostic tests, a lot of the solutions don’t have a price on the labels yet. The problem in calculating ROI may very simply be because we don’t know how to put a cost price on it!

Biases from other industries

In many other industries, number of clicks, number of shares, number of downloads and daily active users would all qualify as significant engagement metrics. But for a healthcare app (there are too many, we know), would the same metrics be valid? Are we satisfied with 1 million downloads as an indication of patient traction and engagement? How many of us downloaded an app and never looked at it again? I do accept Daily Active Users (with time spent on app) is a pretty good proxy to indicate regularity and compliance to some extent. But if it is a digitally delivered CBT, and I just switch on the module without really engaging in it, the above metric will be collected incorrectly. How many of us bought a wearable that we just wore for a few days, or bought a health assessment kit during a ‘healthy phase’ and forgot all about it? If the makers of those products used your download or purchase as a qualifying metric, then they’d all claim huge successes. Technically, they did get an ROI, but is that sufficient as an ROI for us in healthcare? A point purchase versus sustained health outcomes?


Maybe if we set some clear benchmarks that we want to aspire to, that could give us some semblance of an ROI. For example, some initiatives that exist do this more broadly for providers and patients, but collectively picking some of this for health technologies may be a first step.  CMS is incentivizing the hospitals that provide high-quality, low-cost care (lower readmissions, for eg) with bonuses. Hence, patient experience and quality of care are metrics that we can achieve. Ditto with Hospital Consumer Assessment of Healthcare Provider and Systems Survey (HCAHPS) which measures patients’ perceptions of hospital care and allows for direct comparison of the patient experience in hospitals across the country. So, combining metrics for measuring ROI with existing, established benchmarks may be the simpler approach for starters.

Take, for example, population health initiatives like screening or vaccination programmes. If a digital health solution is augmenting this offering, increasing its outreach, improving its delivery - these are easier to measure because of known organizational costs sans the intervention, versus with. The outcomes (numbers screened, or vaccinated) are also easily measurable. So, setting more concrete benchmarks that can be quantified is of the essence.

Measuring ROI for patients and providers must be approached differently, because the outcomes they are seeking are different. For example, for a provider, the benefit of a digital (or any) solution comes down to driving efficiency and productivity. For patients, the benefit is the direct impact on their health, and the price they pay to get back to good health. So, depending on who you are selling to, ensure you make the ROI attractive for them.

I was reading this paper by Welldoc (Bluestar), a chronic disease management platform and digital therapeutic, who projected their cost savings (not ROI, but could serve as a measure for potential ROI) for patients with diabetes as a function of the reduction in HbA1C (glycosylated hemoglobin, an indicator of the amount of blood glucose in the body, and therefore the severity of the diabetes) levels in their blood. While we can measure the cost of uncontrolled diabetes (and the savings achieved by controlling it, by reducing the HbA1C to normal levels), it will not include the correlations with reduced risk of stroke, heart attack, and early death. However, the cost-savings associated with these reductions are often nonspecific and do not apply evenly to all patients: in other words, not all drops in A1C are economically equivalent - digital or no digital. The only advantage existing drugs have is that there is observational data collected over the years from where to capture the reduced risk of comorbidities. Maybe we will have the same for a digital intervention in the years to come, thus making this measurement less erroneous. So, back to the Bluestar paper - they hypothesised that the cost savings associated with decreased A1C levels would be a function of both the change in A1C as well as the starting point in A1C. The higher the starting point and the higher the drop, the larger the economic savings potential. They have published some interesting projections -

They have indicated near-term cost savings for a high risk population for the disease. And that is the ROI on their digital health intervention because it saves the above dollars versus standard of care.

But, promoting good health is not just about ROI

It is only fair to question (depending on who you are) if limiting a health outcomes discussion to finances is even fair. But, in a world that is putting in big money into this space, we must accept that monetary returns on investment will be questioned. As we continue to define our expectations more clearly, and collect richer data, we will be able to answer this conundrum more effectively in the coming years.