How Bots Will Change the Doctor-Patient Relationship

6

Executive Summary

The doctor-patient relationship is health care’s choke point. There’s no technical reason why a variety of common medical conditions-high blood pressure, diabetes, high cholesterol-can’t be managed by a bot and overseen by a nurse with support from a physician only if needed. The lesson from other industries such as retail and banking where self-service has become commonplace is that transformational change requires productivity change, and taking humans out of the loop as needed. In health care that means getting over the sanctity of the doctor-patient relationship and handing a lot of doctors’ high-touch work off to bots. This will free up doctors to focus on cases where human touch and the doctor-patient relationship are most important.

Tim Robberts/Getty Images

It seems nearly everyone believes U.S. health care needs some transformative change to improve quality, expand access, or lower costs. Many of the contemporary approaches toward that change involve making it easier for patients to see doctors, particularly primary care doctors. While that seems intuitive, we think it is the wrong path.

Imagine it’s 1970 and commercial bank executives are deciding how to help their customers get the banking services they need. One executive remarks, “Most of our customers engage with us through our bank tellers-even if they’re later referred to someone behind a desk. To help our customers get the banking services they need, we must make it easier for them to get in front of bank tellers.”

Whether or not discussions like that really happened, that wasn’t the direction banks took. Instead, banks introduced automated teller machines to improve customer service. As a result of this unshackling of banking from tellers, 25-year-olds today find it unimaginable that their parents contorted their schedule to get cash during the Monday to Friday, 9 a.m. to 3 p.m. window. In the age of Venmo, they can’t imagine the need for cash in the first place.

And yet, fifty years later, health care leaders continue to discuss how to get more patients in front of primary care providers, or in general to make it easier for patients to see doctors. The value-add from the technical knowledge and skills of primary care and specialist clinicians is greater than that from bank tellers, but the principles limiting the value of this strategy are the same. True transformations come from enhancing productivity, and productivity is enhanced by decreasing personnel effort, not increasing it.

If we continue to define health care as a service that happens when patients see doctors, we limit our possible productivity gains.

Writing in The New England Journal of Medicine, we argued that the doctor-patient relationship is health care’s choke point. There’s no technical reason why a variety of common medical conditions-high blood pressure, diabetes, high cholesterol-can’t be managed by a bot and overseen by a nurse with support from a physician only if needed. And as our experience and the supporting evidence increase, more conditions might be directed by guidelines, allowing physicians to direct more of their time to where they’re really needed. Much is made of the comforting aspects of personal relationships between patient and clinician. But do we really need that soft touch to manage hypertension? Maybe sometimes, but certainly not always.

So why is it heretical to suggest replacing some of health care with the facilitated self-service that has transformed the financial, retail, and travel industries? Why was it relatively easy to abandon bank tellers, travel agents, and tax preparers with the introduction of ATMs, travel websites, and tax software, but we get push back when nearly identical approaches are suggested for health care?

We think the resistance reflects our social conventions rather than our technical limitations. The technical challenges of safely introducing driverless cars are far more daunting than the technical challenges of introducing bots to manage hypertension and diabetes, yet the prospect of driverless cars seems to be awaited with excitement. We need to solve three problems:

First, the insurance industry-government and commercial-must get better at its job. Because they are ill-equipped to determine whether care was truly needed or appropriately delivered, they use proxy process measures: Was the care face to face? Was the right amount of time spent or the right number of facts documented? Was the right kind of clinician present? Without a measure of what’s good, insurers have found it easier to insist that care be delivered in traditional ways. It’s hard to explore new and potentially better models of care when only old approaches get reimbursed.

Second, state-based regulation of insurance and clinician licensure must be replaced by a system that recognizes that health care is not always best delivered locally. Facilitated self-service creates efficiencies across state lines. It’s likely that occasionally some of Wyoming’s 600,000 residents would benefit from care delivered by someone outside the 1,000 physicians practicing in that state-perhaps by a bot with second-line back up from a nurse or a physician elsewhere. State licensure of physicians and insurance regulations reflect federalist principles harder to justify in a connected world.

Third, we should require the same standards of safety and efficacy for automated approaches to health care that we have come to assume for the safety of pharmaceuticals. Whether that regulation comes from the FDA or elsewhere, it needs to be ramped up to address the volume of potential new approaches. Even if we think a bot can help manage hypertension, it doesn’t mean any bot can do that. The organizations that credential clinicians might find they are well suited to credential robots.

If there is a fourth problem, it is our sense of nostalgia. Norman Rockwell’s paintings of what he saw as wholesome and right reflected a time when doctors attended to the whole family, knew some from the cradle and some into the grave, and were paid with a basket of eggs. Facilitated self-service health care doesn’t challenge the appeal of this image, but it does shift it toward those elements of care that can’t as easily be handled by machine.

The health care changes we want, or at least the opportunities to try them, are held back by a combination of technical limitations and social conventions. But our social conventions present the greater obstacles. The lesson from other industries is that transformational change requires productivity change. And in health care that means we must find ways to move past approaches to facilitate care with doctors toward approaches that facilitate care without them.