This article first appeared in NEJM Catalyst. It is co-authored with Nirav R. Shah, MD, MPH, and Charles M. Farkas, MBA.

The consolidation craze continues, but vertical integration has yet to demonstrate real progress toward the Triple Aim. Health care leaders would do well to consider innovative approaches that are working in other industries, including the tech-enabled full stack model.

As in many industries, health care has experienced waves of consolidation in the name of growth. Hospitals have bought physician practices, doctors are opening ambulatory surgical centers, and there is a buy free-for-all across all parts of the continuum of care and new forms of payment as never before. Such vertical integration purportedly drives financial efficiencies with scale, improves clinical quality by reducing problems of care transitions, encourages teamwork, and calibrates scope of practice to services provided. Yet the jury is still out. Costs continue to rise as the market power of health systems grow with lots of finger pointing, readmissions and problems in patient handoffs persist, anticompetitive practices emerge, and quality remains highly variable. Perhaps this lack of progress toward the Triple Aim suggests that we should reconsider our assumptions that vertically integrated systems will achieve productivity, patient- centeredness, and will deliver on a scalable version of Kaiser Permanente. Perhaps we should revisit the so-called rules of health care.

New business models, such as full stack tech-enabled clinical services meaning the complete delivery of the clinical care made scalable and lower-cost by applying technology, are showing promise.”

In fact, other industries have shown what’s possible with different approaches to integration. iPhones are assembled to stringent requirements set by Apple using parts from hundreds of suppliers from around the world, to ultimately build a branded, high-margin and high-quality product labeled “Made in California by Apple.” In reality, the suppliers of each component know their job is to innovate constantly: Next year’s camera is going to be smaller, higher resolution, draw less power, and cost less than this year’s . . . and it is the suppliers’ job to focus of their efforts on this task and to raise the capital required for innovation, take on supply chain risk, etc.. Similarly, when Toyota partners with its suppliers, everyone benefits from faster cycle times for new products, lower costs, and improved quality — one form of a learning network.

In health care, however, the argument is that we can’t do that, as approximately of costs relate to inflationary wages, the regulatory burdens continue to grow, and health care is just different.

Some Glimmers that Health Care May Be More Like the Rest of the Economy

New business models, such as full stack tech-enabled clinical services meaning the complete delivery of the clinical care made scalable and lower-cost by applying technology, are showing promise. In rural areas, much of radiology is outsourced to companies such as Teleradiology Solutions or VRad whose job as suppliers is to turn around high-quality results as fast as possible, leveraging large independent contractor armies of radiologists working remotely. They are able to innovate their part of the health care ecosystem — e.g., using artificial intelligence to assist reads — because they have achieved scale in their niche in ways that traditional vertically integrated delivery systems cannot. Health systems have little incentive to maximize radiologist productivity with AI tools and better workflow because radiologist labor costs are often fixed; hospitals get no additional revenue from these investments and do not have queues of unread images or the ability to grow volumes of images like outsource partners.

Many examples of scalable tech-enabled full stack services exist for chronic diseases that have also grown into highly successful businesses, including Flatiron for oncology, Omada for pre-diabetes, Virta and Livongo for type diabetes, Lyra for behavioral health, and Aspire for end-of-life care. These businesses manage many aspects of care including symptom monitoring, disease maintenance, and clinical follow-up and have built models with much lower labor costs, better service and outcomes, and lower prices compared to their traditional health care delivery counterparts. For example, a study by Omada assessing outcomes in privately insured pre-diabetics demonstrated weight loss nearly greater than that of comparable interventions without the technology in the index population. Omada’s remotely accessible educational curriculum producing these outcomes was delivered at costs lower than the face-to-face, one-on-one standard of care for promoting lifestyle modifications. Additionally, according to a model created by Thorpe and Yang, if scaled, this level of weight loss could produce Medicare savings of up to . billion over years.

These businesses all have broken a lot of so-called rules of traditional medicine.”

These businesses all have broken a lot of so-called rules of traditional medicine. Does care have to be part of a health system? Can care models outside of electronic health records EHRs avoid duplicative testing and fragmentation of care? Can new technology create more reliable care models? Can these new models be compliant with regulations? Will health systems be fast followers, copy, and squash these start-ups? The results speak for themselves: These models seem to work very well in all dimensions and also appear to be uncopiable by health systems.

Quality for all of these can surpass the outcomes of vertically integrated health systems operating in a vacuum. Flatiron’s oncologists are much better equipped to adhere to guidelines and use genetically guided treatments when supported by the company’s clinico-genomic database. As discussed above, Omada — as well as Livongo and Virta — can deliver better objective and subjective patient-reported outcomes than previously documented in health systems alone. Studies using Virta’s continuous care platform have exhibited superior outcomes across a number of endpoints at and years compared to patients undergoing conventional care orbiting around health systems. Livongo’s application promoted better psychological self-efficacy in diabetics compared to the control population subject to traditional care, in addition to insurance cost reductions of , after one year of use.

Lyra’s care model delivers improved access to behavioral health care in acute scenarios beyond the clinical setting such as in suicidal ideation and tracks outcomes that cannot be monitored in individuals “invisible” to conventional health systems by failing to seek in- person care. Finally, Aspire’s care protocol has produced hospice transition rates up to higher than comparable initiatives undertaken at integrated health systems, while simultaneously reducing end-of-life costs in the last year of life by , relative to control populations. All of these platforms can simultaneously deliver care to far more people than is feasible under the staffing resource constraints of traditional health systems.

Why Is It Hard for Health Systems?

Health systems are at high risk of following in the footsteps of Sears and Kodak. They have years of experience building highly successful businesses that maximize local market scale, delivering high volumes of very complex high-quality care, with management systems reliant on high-cost labor and inflexible, expensive enterprise software. Moreover, recent years have been among the most successful ever financially for most health systems. Between , total health spending increased at a rate times that of other consumer expenditures, with growth that was particularly concentrated for the inpatient services with price increases of per admission and outpatient services with cumulative expense increases of that are provided by health systems.

While most health system leaders talk about futures where they manage total cost of care and leverage modern technology, the impetus to get to this end state lacks a burning platform.”

The rate of health spending growth accelerated at the end of this window — again, in particular for health system–related services — after several years of Affordable Care Act– and recession-associated stabilization. According to IBISWorld, industry revenues now exceed trillion, with growth over the next years at a projected rate of . annually compared to . for the broader economy. As a result, while most health system leaders talk about futures where they manage total cost of care and leverage modern technology, the impetus to get to this end state lacks a burning platform.

For example, the creation of larger systems may give lip service to better care coordination, but that has not been the reality on the ground. We use the language of scale deals and scope deals. When one hospital buys another, that is a scale deal. In theory, they can consolidate the back office finance, IT, HR, etc.. As a larger combined entity they should be able to procure some of their supplies and equipment at a lower cost. But all too often, two considerations have driven much of the consolidations: more scale to leverage pricing negotiations with payers and a larger geographic footprint to potentially offer narrow network products. Many of the scope acquisitions have been physician practices. These often are justified as important to care transitions or as necessary to broaden an Integrated Delivery System’s geographic footprint. While that may be true, the other perspective is that such scope acquisitions are just to build a funnel to drive procedure and inpatient volume into anchor hospitals.

Moreover, the path from the current business to the future model seems risky, financially painful, and organizationally challenging. All future models have less emphasis on hospitals and specialty and more investments in primary care, behavioral health, and home-based care. These models shift revenue from hospitals and proceduralists to risk- bearing doctors. These models also call for investments in very different technologies than electronic health records and data centers, relying instead on more agile consumer technologies and cloud computing. To date, where health systems have turned to outsourcing it has been driven more by the need to access resources — Emergency Department physicians, radiologists, and some have been driven by access to capabilities or someone else’s scale — like the outsourcing of labs to Quest and LabCorp.

The perceived pain of truly embracing disruptive models leads health systems to do toe-in- the-water pilots that are seldom resourced with the talent, capital, and steadfastness needed to build care models that rival the start-ups whose existence depends on success. Moreover, the payment incentives, governance friction, and bureaucratic constraints health systems offer their innovators are far less compelling than the equity, clean-sheet- of-paper, and rapid decision-making offered by start-ups, which makes attracting the talent needed to succeed to traditional health systems extremely unlikely. Because entrepreneurial and tech talent is critical, and health systems struggle to recruit, they are destined to struggle at innovation.

To become innovators beyond simply aggregators, provider systems need to learn how to coordinate all the resources that are required to deliver on the Triple Aim without having to own all the resources.”

Taken together, health systems would be wise to be early-adopter customers of promising clinically disruptive businesses. While mergers and acquisitions activity by provider systems in reached its highest value billion since , an early-adopter approach contrasts the observed pattern of mega deals to incorporate relatively mature corporations. Provider systems that tend to build in-house would be wise to acquire great teams by purchasing start-up businesses that are relatively affordable — such as with Harris Healthcare’s acquisition of Iatric Systems to augment its existing EHR systems, and Premier’s acquisition of Stanson Health to “realize Premier’s vision of an enterprise analytics and performance improvement platform.”

Additionally, in order to become innovators beyond simply aggregators, provider systems need to learn how to coordinate all the resources that are required to deliver on the Triple Aim without having to own all the resources. This is a strategy employed successfully by Johns Hopkins Hospital through its Guided Care model for primary care, which has since been licensed to many health systems nationally. This model uses comprehensive, patient- specific needs assessments to determine care requirements, integrate behavioral health and social services, facilitate care access, and coordinate care communication — all through leveraging “local resources” beyond the health system in a vacuum.

While early, these experiments of decoupling parts of a hospital show promise in achieving Triple Aim outcomes: including improved outcomes, decreased costs, and enhanced access. Additionally, as seamless, clinically oriented information technology is a basic pillar of these programs, continuous learning and improved provider satisfaction are literally built in given that burnout has been directly linked to EHR burdens in excess of face-to-face clinical care. This potentially resolves the fourth tenet of the IHI’s recently issued Quadruple Aim clinician satisfaction. Nonetheless, in the short term, it may be the buyers of health care that drive more of the change than the integrated networks.