There is a fair bit of talk about information bubbles and echo chambers - mostly in reference to our current political climate. Reading two pieces - most recently today's Rock Health 1H 2017 Digital Health Funding Midyear Report - and the accompanying Twitter reactions, it is clear that echo chambers aren't reserved for politics alone. By virtue of the fact that this post will express an opinion, it contributes to that echo chamber. My point here is that if we rely on sound bites and 140 character interpretations of complex data, we may be tempted to think that either (1) digital health doesn't work, or (2) digital health is the clear path forward. The truth is somewhere in the middle - for digital health to truly impact cost of care delivery and improve patient outcomes one needs to clearly articulate and deeply understand the problem that is trying to be solved, and for whom.
On June 26, 2017, JAMA Internal Medicine e-published a study from Volpp et al from the University of Pennsylvania evaluating the impact of technology on medication adherence and outcomes in over 1,500 post acute myocardial infarction (AMI) patients. The study concluded:
"A compound investigation integrating wireless pill bottles, lottery-based incentives, and social support did not significantly improve medication adherence or vascular readmission outcomes for AMI survivors."
If one stopped there, the temptation would be to consider this a reasonably sized study that demonstrated no real impact of digital health or patient engagement tools on medication adherence, cost or outcomes. Digital Health is doomed - proponents of the view get to do the Neener Neener Dance, while digital health evangelists double down, citing data to the contrary.
Digging a Little Deeper
The 12-month University of Pennsylvania Health System (UPHS) study enrolled 1,509 patients from predominantly northeast (~36%) and southern (~40%) geographies, across 5 different payers. AMI survivors were prescribed between two and four medications (statin, aspirin, beta blocker, anti platelet agent) and ranged in age from 18 to 80 years old; 34%/37% were female (control/intervention). Primary outcomes were time to re-hospitalization for a vascular event or death, and secondary outcome was medication adherence & total medical cost. The compound intervention included any of the following
- Up to four electronic pill bottles (Vitality GlowCaps) for prescribed cardiovascular medications
- Daily lottery incentives with a 1 in 5 chance of a $5 payout and a 1 in 20 chance of a $50 payout based on medication adherence the previous day
- The option of enlisting a friend or family member who would be automatically notified if a patient failed to use the electronic pill bottle 2 out of the 3 previous days
- Access to social work services
- A staff engagement advisor to provide close monitoring, feedback, and adherence reinforcement
Data from the GlowCaps was sent to Way to Health, an NIH-funded, UPHS software platform that "facilitates patient engagement". Additional details of the engagement intervention/adherence reinforcement were not provided. The analysis was performed of the "compound" intervention on a discreet set of medications - there was no breakdown referencing the number/type of patients opting for each intervention(s), and how/if each intervention differed by outcome or the number of medications prescribed ("between 2 and 4" is fairly broad given side effect profiles of each). Most importantly, there was no available description of the patient population beyond age and gender. Certainly in older populations, it is not uncommon for patients to have multiple co-morbid conditions - the classic cluster is heart disease, diabetes, and hypertension - each requiring its own set of medications. In this study, only patients with metastatic cancer, dementia, and end stage renal disease on dialysis were excluded. The point is that the study participants are heterogenous, the number of medications prescribed was heterogenous, and the intervention was heterogenous and unstratified. As the authors state in the study limitations: "We could not design the trial with sufficient power to detect small differences in adherence or costs because of the impracticably large sample size that would have been required" (Enter Evidation Health ...)
Based on how the study was designed and analyzed, the interventions had no impact relative to a control group of routine care. The investigators got the answer to the question they asked. But the reality is that patients are not the same one to another. Just as, even for a given condition, therapeutic intervention is personalized to an individual's lab values & imaging, history, co-morbid conditions, and (more recently) genomics, so must the strategies that enable adherence and engagement. THAT is where the secret sauce lies - that is where AI, data and analytics smartly applied can have real impact on outcomes and cost. That is yet another level of digital health - the enabling technology that links right patient, right intervention, right time.
Following the Money
Which brings me to the Rock Health report, which triumphantly declared Q2 2017 a "record-shattering quarter" for digital health. In the spirit of full disclosure, I am a huge fan of Rock Health. And not just because it is a female founded and led venture and one dedicated to companies at the intersection of technology and healthcare way before it was a thing to do. Rock Health provides critical information against a backdrop of policy, healthcare & technology, and financial markets. I haven't found another organization quite like it. That said, today's report fell a little flat for those tilted towards the patient- and healthcare cost side of the equation ... the side of the equation contemplated by the UPHS study. The Rock Health report also asks us to more carefully consider what we label as "digital health": of the year's two "largest digital health deals ever", one is a consumer health information platform ($500 M) that brings patient information to waiting and exam rooms, and the other is a connected exercise bicycle ($325 M). The Rock Health team quickly points out that the size of the deals skew the top funded "digital health" categories - suddenly weighted towards "consumer health information" and "connected fitness equipment".
If you're looking for trends, of the top 2016 deal digital health categories - genomics & sequencing, analytics/big data, wearables/biosensing, telemedicine, digital medical devices, and population health management - only analytics/big data made it to 2017's top six. Even if you take out those deals greater than $100 M, you're still only left with analytics/big data and wearables/biosensing as categories consistently engendering venture funding.
The question the report raises: what is the relationship between what stimulates capital investment and what truly solves a healthcare problem? Moreover, what is the opportunity for "digital health" in this environment, and how do we define what that actually encompasses? Genomics saw huge investments this year - Grail alone closed over $900M in Q1. Despite being heavily analytics & big data driven, they did not make Rock Health's cut. If they had, genomics would be the top funded category two years in a row, especially when one considers the many smaller genomics deals, also heavily weighted towards analytics (and not represented in the report). The problem these companies are trying to solve is complex and will dramatically impact a high cost, high need problem for patients and providers alike - early detection of cancer to foster effective, personalized therapy.
When we think about what contributes to our country's exorbitant healthcare expenditures, accounting for 20% of our economy, the answer is complex and multifactorial. But what it isn't is lack of access to connected exercise equipment. It isn't lack of in-office patient education. It is complex chronic disease, lack of operational & clinical efficiency in our health system, an imbalance between real quality care and the cost to deliver that care, and imperfect care coordination and communication. Smart digital health can actually alter the conclusions from the Volpp paper - personalized, accessible solutions based on smart health analytics can dramatically impact care. These are the real digital health companies - and while the financial return may be slower than that of connected exercise equipment, the impact (and return) will ultimately be far greater.