Disruption is business as usual for Amazon. Indeed, it’s practically CEO Jeff Bezos’ mantra. But the retailer’s decision earlier this year to enter the health-care arena was still big news. On Jan. 30, it announced a partnership with JPMorgan Chase & Co. and Berkshire Hathaway Inc. to look for “ways to address health care for their U.S. employees, with the aim of improving employee satisfaction and reducing costs.” The scale of employees and customers that the three companies could leverage suggested that something fearsomely game-changing was in the offing. Markets shuddered, with stocks for insurers like UnitedHealth Group and Anthem plunging.
But what, exactly, will be disrupted? Health care is an enormous, complex industry, and the three companies have been tight-lipped since their announcement, which made vague mentions of “technology solutions.” Add to that JPMorgan Chase CEO Jamie Dimon stressing the importance of “transparency, knowledge and control” for employees. Still, while a disruptive health-care product from Bezos and company doesn’t yet exist, the challenges the three companies promise to address are common and persistent.
A consortium like the one Amazon.com Inc. is leading provides opportunities to amass data that, in theory, can help employers and employees make better and more cost-effective health-care decisions. But much of the relevant data are inaccessible, locked down by middlemen. Employees can have a hand in providing information themselves, but inspiring employees — especially healthy ones — to voluntarily contribute that information has proven a tough behavioral-modification challenge. Clever and tech-savvy companies can find ways to extrapolate information, but doing so means entering a new frontier of privacy concerns.
“Amazon’s obvious strength is that it’s a behemoth company that has excelled in part by dramatically improving the consumer experience,” said David Goldsmith, chief strategy officer at WEGO Health and former CEO of Dossia, a now-defunct consortium that sought data-driven solutions to reduce health costs. “The question is, where are they going to focus? And is it possible that they can accomplish some of what they want to do without necessarily worrying about where the data’s going to come from, but focused much more on how they’re going to engage their employees in their health?”
The answer to that question has implications not just for the fate of the Amazon consortium but for how employers everywhere think about their employees’ health and well-being.
THE PROBLEM WITH CONSORTIA
Amazon isn’t the first large company that’s pooled resources of multiple companies to address health care. The Health Transformation Alliance (HTA), founded in 2015, now claims 46 member firms, including Fortune 100 companies like IBM Corp., Coca-Cola Co. and American Express Co. Members have agreed to make employee data available for robust analytics, and the alliance has brokered partnership deals that have produced some tangible benefits. Negotiated rebates and discounts have resulted in a 15% savings in prescription drug costs for member employees, according to HTA CEO Robert Andrews.
Andrews sees Amazon’s arrival less as a competitor entering a ring than as a validation of HTA’s thesis. “I see U.S. health care as a tug of war between people who want to retain the fee-for-service system and people who want to replace it with a fee-for-outcome or fee-for-value system,” he said. “We obviously favor the fee-for-value system. And, so, having more people pulling on our end of the rope is welcome news.”
Gathering more health data for large numbers of employees, groups like HTA arguably gain leverage to better negotiate with insurers and pharmacy benefit managers (PBMs), which control prescription medicines for health plans and promote more effective wellness programs.
Historically, though, that data has been hard to access, which was one of the main reasons for the downfall of Dossia, according to Goldsmith. Dossia was founded in 2006 with a goal of creating online personal health records for members’ employees, but insurance companies and PBMs resisted releasing that data, much to the frustration of Dossia leadership. The consortium lobbied federal agencies for increased access that would allow it to improve analytics that might reveal cost-saving opportunities, but to no avail. Dossia shuttered in 2016.
Gathering more health data for large numbers of employees provides leverage for consortia to negotiate fees.
Dearth of data also has sunk other efforts from large tech companies to help consumers manage their health information. Consider Google Health, which closed in 2011, and Microsoft’s HealthVault, which closed its line of mobile apps earlier this year. Such initiatives are only as good as the information they have to work with, the thinking goes, and the best information was unavailable. The size and economic might of the companies involved were unpersuasive. (Apple is seeking more success on this front: Earlier this year it announced an agreement with 13 health systems allowing individuals to download their health data to the company’s devices.)
“These other companies, or the health-care companies themselves, are also massive and have a lot of interest in protecting their data,” Goldsmith said. “Their view is, ‘This is one of our greatest assets, we’re going to protect it and we’re not going to assume that this is something that we should just be willing to share because we have a number of big companies telling us that we should.’”
Thus far, HTA has succeeded where Dossia hasn’t through relationship-building. For instance, its announced decrease in drug spending comes thanks in part to partnerships with two large PBMs, CVS Health and OptumRX.
“We don’t think the PBMs are a problem,” Andrews said. “We think that the prescription distribution system is a problem.” But if medical information won’t be easily accessible, that means Amazon will likely have to do something that HTA is doing: Rethink data.
THE PROBLEM WITH DATA
Health care is getting smarter, with more power to not just make diagnoses about a patient’s current illness but to better predict diagnoses that are likely to happen without intervention. For instance, Sanford Health, a South Dakota-based health-care network, recently announced that it was using algorithms that process patient data to predict the likelihood of a diabetic’s unplanned medical visit with 80% accuracy.
Medical data, by one estimate, accounts for 30% of all the world’s electronic data. And it’s ever-growing as more people get comfortable with wearable fitness trackers like Fitbits and Apple Watches. Plus, more companies are getting more sophisticated with analytics. HTA, for instance, uses its partnership with IBM and its Watson supercomputer to begin doing the kind of number-crunching required to do predictive diagnoses. No PBM data? No problem.
“Rather than relying on PBMs to understand the efficacy of Copaxone for [multiple sclerosis] or Metformin for diabetes, we want to find that out for ourselves,” Andrews said. “And with the 7-million-life cooperative which we now have, we are slowly but surely — with IBM Watson — building our own analytics capabilities, and we can answer those questions ourselves.”
To test that idea, HTA is focused on a handful of the most common health-care issues — type 2 diabetes, back pain, and hip/knee replacements — to search for common threads among people with those diagnoses.
“What Watson can do is reverse-engineer those post diagnosis facts,” Andrews said. “Say [a hypothetical employee named] Fred is now diagnosed as type-2 diabetic. What did Fred look like a year ago? What did a thousand Freds look like a year ago?”
Lacking direct access to Fred’s patient data, HTA might consider a host of other pieces of information it does have at hand: The number of sick days Fred took before the diagnosis, Fred’s use of the employee benefit gym membership, public-sector information about trends in the census tract where Fred lives.
As a result of the data Fred unwittingly provided, a company might be able to then contact Fred with guidance about his options or contact another hypothetical employee unsolicited and warn her that she’s at risk. But such robust analytics carries privacy risks, according to Priya Nambisan, associate professor and chair of the Department of Health Informatics and Administration at the University of Wisconsin-Milwaukee’s College of Health Sciences.
“What if you can’t exercise? What if you’re extremely obese and you really can’t go to a gym for exercise?” she asked. “There are a lot of issues — and some of these are addressed by the Americans with Disabilities Act — that you cannot discriminate against employees for not going to a gym or having any kind of health issues … I don’t think employers should be tracking Fitbit info or lapsed company gym memberships. That is a privacy issue.”
Andrews conceded that this is “fraught with cultural and HIPAA and privacy issues. You don’t want to stalk Fred. But our people are CHROs who live with this problem every day, of knowing how to interact with employees in a way that’s balanced and appropriate.”
THE CRITICAL MASS SOLUTION
If a consortium like Amazon’s runs into problems using employee data passively, it also can encourage employees to actively provide it. The company that mastered collaborative filtering and one-click ordering has plenty of expertise at nudging consumer behavior, and it could conceivably succeed where previous personal health-record efforts like Google Health have failed.
“I think the big lesson learned from Google Health and HealthVault as well is that just having a repository for your health data doesn’t do you a lot of good,” Goldsmith said. “Most of us make a decision on our health based on a very specific trigger. That trigger is often a new diagnosis.”
That information can still be useful for consortia, but it’s incomplete, slanting the pool of data toward treatment-based solutions, not preventive ones. Moreover, higher employee engagement with information can get complicated. Amazon presents consumers with plenty of information about a book, stroller or set of power tools via customer and professional reviews. Transferring that sensibility to health may result in higher engagement, Nambisan said. And as anybody who’s witnessed a Facebook food fight over politics, high engagement isn’t necessarily a blessing.
“Consumer knowledge is actually increasing,” she said. “But then that consumer knowledge is a double edged sword. Once consumers increasingly read health-related journal articles, they might have a reduced trust with the health-care providers.”
Another potential solution, HTA’s Andrews said, is to find ways to gamify the information gathering in ways that Google and dating sites and Amazon itself have used. Consider Fred again: Perhaps Watson has found a potential correlation between diabetes and joining a beer club, Andrews suggested. Employers might find out who most commonly joins such beer clubs, then take the question to workers. “You’d find a subtle way to ask, ‘Fred, have you joined a beer club?’”
That, too, presents privacy issues, but Andrews is confident they can be resolved. He’s also confident that HTA can keep scaling up: The alliance has a goal of having 20 million lives in its system. But to successfully use the kind of analytics HTA is looking for, it needs depth as well as breadth. It requires enough people in every kind of American community to gain the kind of understanding it needs. And that may be the ultimate difference-maker for Amazon, JPMorgan Chase and Berkshire Hathaway.
“Medical is local,” Andrews said. “A hospital system in Phoenix is really not that impressed that you have millions of people in other markets. They want to know what you have here. If you basically tripled the number of lives you have, we believe that in every market we would have sufficient critical mass to have an impact on the provider’s economics, which is how you change the provider’s behavior part of how you change it. It opens the door.”
Mark Athitakis is a freelance writer for WorldatWork.