Care Redesign
Targeting, Tailoring, and Trimming Chronic Illness Care (14:47)

“Any American effort to meaningfully improve the value of care needs to get better at managing the longitudinal trajectory of those with chronic illness,” says Arnold Milstein, Director of the Clinical Excellence Research Center at Stanford University. With 86% of health care spending directed toward chronic illness, we cannot ignore this trajectory. Fortunately, America is beginning to learn from a sequential failure of experiments in managing chronic illness care, says Milstein, which date back approximately 40 years.

Managing care wasn’t discussed much prior to 1980, according to Milstein. Instead, choice of care consisted of a 15-minute doctor visit, a phone call responded to hours later, or an answering machine message instructing the patient to go to the emergency room. “It’s self-evident that that did not work well for people with multiple chronic illnesses, which almost inevitably begin to go off track on evenings and weekends,” says Milstein.

When health insurance claims data analysts observed that 10% of people spend 70% of the money each year, an adjustment to choice of care was made: having someone from a remote location contact the patient, coordinate care, and support the patient’s self-management. “On the face of it, it seemed like a good idea,” says Milstein. But there was no evidence that this change meaningfully impacted either health or health care costs. That, says Milstein, was the first failure.

The second failed experiment occurred from about 2000 to 2010, with a new wave of entrants into the case and care management industry. Case managers were embedded in health care delivery organizations but rarely in the clinical teams themselves. “It was interesting to see these entrepreneurs come in with absolute certainty that they were going to save Medicare a fortune, improve quality,” recalls Milstein. And yet, 33 out of 34 attempts went bankrupt. “The problem here is that most of these interventions did not involve much in the way of personal face-to-face contact between the patients and the care managers, and the care managers and the clinicians. It was still relatively remote,” Milstein explains.

But, he says, “I don’t consider this collectively a failure. I consider this an illustration of a not-too-well-organized national learning system that could learn a whole lot faster than we do. It’s been a 40-year learning process. But I think out of the rubble of these failures, one can extract a few pearls about what might work better as we now gear up to think about care and case management 4.0.”

  1. Better targeting. The world is moving away from risk stratification tools that use traditional regression-based analysis and toward supervised machine learning–based mechanisms that help determine who is more likely “to get into serious health trouble in the near term.” As we learned in the 1990s, says Milstein, of the 10% of people who will spend 70% of next year’s dollars, 60% of them are patients who were not that sick in the baseline year. “It’s a substantial challenge. You have to predict the ‘cost blooms.’ You can’t simply rely on predictive models that try to figure out who among last year’s cost blooms will persist because you won’t be addressing 60% of the patients who will be in next year’s 10%.”
  2. Better tailoring: “We have to tailor our care plans if we want to have a hope of being successful,” says Milstein, “and the list [of variables], I have to say, is daunting. It really is an invitation to computer scientists to think about how we might integrate these things.” This includes traditional physical threats to health, as well as behavioral health risks, social risks, and medical neighborhood issues. Quantifying patients’ level of activation in being able to care for themselves is another important factor in tailoring a care plan. “It’s one thing to say to a patient before they leave, ‘here’s a list of the six things you have to do,’ but if the patient is highly self-efficacious why waste money supporting that patient at home? On the other hand, if the patient’s score is quite low that’s a situation where it’s probably a wonderful investment,” he says.
  3. Better trimming: Many of the failed experiments did succeed at lowering rates of emergency room use and hospitalization, but the amount of resources spent to prevent these expensive events far exceeded the net savings generated. And the tools they primarily used were expensive, sophisticated nurse case managers. “Does the nature of the problem you’re trying to solve for patients who struggle to self-manage at home always require a sophisticated case management nurse, who at least in the Bay Area, including benefits probably costs somewhere between $150,000 and 200,000 a year?” asks Milstein. “I don’t think so. The evidence suggests that for many patients, having a psychologically minded, intuitive BA-level trained medical assistant work with that patient to improve their level of activation is probably more than enough to substantially elevate and improve the patient’s health trajectory and lower their health spending trajectory in the near term as well as over time.”

“So, yes, we have sequentially failed, as occurs in other industries,” Milstein says. “The challenge to us is to learn more quickly from our failure and to evolve as a national health system.”

From the NEJM Catalyst event The Future of Care Delivery: Relentless Redesign at Providence St. Joseph Health, January 19, 2017.

More From Care Redesign

Improving Care and Cutting Costs: Implementation of a Laboratory Formulary to Facilitate Better Laboratory Ordering Practices

Can a formulary system help to prioritize cost-effective lab tests in the same way it has done for prescription drugs?

Data Graphic: Real-Time Communication Is Key to Improving Post-Acute Care Transitions

The NEJM Catalyst Insights Council weighs in on the best opportunities for post-acute care transitions.

Collaborative Care for Depression in a Safety-Net Health System

Integrating depression treatment into primary care in New York City’s public system.

The Waiting Game — Why Providers May Fail to Reduce Wait Times

Waiting has emotional effects on patients, so it’s ironic that physicians often cite long waiting times as evidence of their excellence.

No Stories Without Data, No Data Without Stories

We must remember to listen to the stories of the human beings on the receiving end of the policies we develop.

From Co-Located to Integrated Teams: How Utah’s Neurobehavior HOME Program Changed Its Culture

University of Utah Health incentivized coordination through integrated teams to provide better care at a lower cost for patients with developmental disabilities.

What’s More Valuable Than a Healthy Choice? Making Lifestyle Medicine Standard Practice.

A framework for embracing the health benefits of lifestyle choices in medicine.

The Other Victims of the Opioid Epidemic

The opioid epidemic is a national crisis that should not be underestimated. But its solution will require development of meaningful interventions.

Population Health — What’s in a Name?

Physicians and executives may agree on the concept but differ on how to define it.

We Need More Geriatricians, Not More Primary Care Physicians

Geriatricians are among the most satisfied specialists, so why don’t we have more of them?


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »


Resetting the Nation’s Health Care Quality…

Poor measures of care quality have consequences. The National Quality Forum is essential to creating…

Care Integration

51 Articles

Collaborative Care for Depression in a…

Integrating depression treatment into primary care in New York City’s public system.

Social Needs

44 Articles

Walking with Our Patients in the…

Mobile integrated health, a disruptive home-based delivery model, repurposes existing resources and addresses critical care…

Insights Council

Have a voice. Join other health care leaders effecting change, shaping tomorrow.

Apply Now