January 10, 2020 Read Advisory Board's take: Do we need to rethink our entire approach to population health?

A program designed to reduce health care spending and improve health care quality among so-called "super-utilizers" did not result in lower hospital costs or fewer readmissions, according to a rigorous study published Thursday in the New England Journal of Medicine.

Your guide to implementing a successful super-utilizer program

Providers look to crack the code on so-called 'super-utilizers'

Super-utilizers are patients with serious illnesses who frequently visit the hospital and ED. They tend to be uninsured or enrolled in Medicare or Medicaid. Super-utilizers account for just 5% of the U.S. population but 50% of health care spending, the Associated Press reports.

Providers have been working to identify super-utilizers and address the root causes of their health problems in an effort to curb health care spending—a practice referred to as "hotspotting." For example, the Camden Coalition of Healthcare Providers 10 years ago launched a care-transition program aimed at reducing readmission rates and hospital costs among super-utilizers by providing the patients, who typically are very sick, with continued access to care and other resources after they are discharged from the hospital. Under the program, a team of community health workers, nurses, and social workers visited the homes of super-utilizers to coordinate their outpatient care and connect them with social services.

Early research suggested the program succeeded at reducing readmissions among super-utilizers, and the program gained national attention. Though the evidence demonstrating the program's initial success was weak, according to the New York Times' "The Upshot," the federal government lent support to further evaluate and expand the initiative, and the effort sparked similar programs throughout the United States.

New study shows program didn't reduce readmissions

Researchers set out to further evaluate the program's effectiveness and randomly assigned 800 patients who were hospitalized at least twice during a six-month period, had two long-term health conditions, and had at least two other circumstances complicating their lives—including homelessness, a mental health condition, drug misuse, difficulties accessing services, use of five or more medications, or no social support—into two groups after they were discharged. One group participated in the care-transition program for three months, which cost about $5,000 per patient. The other group received usual care, according to AP.

The patients in the care-transition program on average received 7.6 home visits and 8.8 phone calls from program staff during the three months following their hospital stay. In addition, the program's staff accompanied patients to an average of 2.5 visits with their physicians.

The researchers evaluated the outcomes of both patient groups six months after the patients' were discharged from the hospital and found similar results for each group. Overall, the researchers found patients in both groups had nearly the same chance of returning to the hospital and the same number of readmissions, spent nearly the same amount of time in the hospital overall, and generated nearly the same hospital costs.

Specifically, the researchers found both groups had a readmission rate of 62%, and there was no difference in total health care spending between the groups. The researchers also found that two of the program's key goals—providing patients with a home visit and ensuring patients visited a provider within one week of leaving the hospital—were achieved for just 28% of patients in enrolled in the program.

Why didn't patients enrolled in the program have better outcomes?

The researchers noted a few reasons why the program might have failed to show significant improvements when compared with the control group. For example, the researchers said usual care provided to patients in the control group might have improved. They also said homelessness and lack of access to a telephone among patients enrolled in the program might have inhibited its effectiveness.

In addition, the study might not have used a large enough sample size, which could have affected the results. It is possible outcomes varied across groups in small ways that could not be detected by the study's small sample size, "The Upshot" reports.

Kathleen Noonan, CEO of the Camden Coalition of Healthcare Providers, said the three-month evaluation period also might not have given the program enough time to help patients. She also noted there weren't enough places to help patients access care for substance use disorders or mental health conditions.

Jeffrey Brenner, a founder of the Camden Coalition of Healthcare Providers who now works on similar programs for UnitedHealthcare, said, "The bottom line is, we built a brilliant intervention to navigate people to nowhere." He continued, "The system is good at delivering care if you're an average patient on an average day," but "[t]hese patients have complex problems. If you don't meet their needs, they'll keep going to the hospital." (Daily Briefing is published by Advisory Board, a division of Optum, which is a wholly owned subsidiary of UnitedHealth Group. UnitedHealth Group separately owns UnitedHealthcare.)

Aaron Kesselheim, a health policy expert at Harvard Medical School who did not participate in the study, said the program was "clearly done with the best of intentions" and might have benefitted patients without reducing readmissions. Overall, Kesselheim said reducing readmission rates appears much more complicated than it might have seemed (Frakt, "The Upshot," New York Times, 1/8; AP/NBC News, 1/8; Finkelstein, New England Journal of Medicine, 1/9; Gorenstein/Walker, "Shots," Kaiser Health News/NPR, 1/8; Emery, Reuters, 1/8).

Advisory Board's take

Yesterday's publication, "Health Care Hotspotting—A Randomized Controlled Trial," has shaken many of us who follow population health research and innovation. Dissenting evidence regarding this seminal work may seem like a challenge to fundamentally held truths in population health. If high-touch interventions for super-utilizers don't impact outcomes or costs, does our industry need to rethink its entire approach?

Although surprising, the study is only one of many indicators that we still have a long, complicated journey ahead of us in improving population-wide health, and can be a helpful impetus to reexamine traditional approaches. Based on our research, we have two initial reactions:

  1. We can't solve structural problems with solutions designed for individuals.

    Growing evidence indicates that the majority of provider organizations aren't focused on the right ambitions. For many medically and socially complex patients, one-off referrals to social services won't seriously impact deeply entrenched behavioral patterns, trauma, and neighborhood conditions.

    Instead of relying on reactive, surface-level fixes, provider organizations need to target the root causes of poor health: intergenerational poverty and structural inequities. To take such a mandate seriously, providers must look far beyond 90-day hospital-based interventions. Structural change requires community-led coalitions working over decades. For instance, the 41 provider members of the Healthcare Anchor Network have moved in this direction with a combined $700 million pledged for social impact loans. 

  2. Short-term, targeted solutions can still help patients.
  3. Although this new research requires us to recalibrate our industry's ultimate goals, it doesn't mean we need to slash and burn our current strategy. For one, as well-known as the Camden model is, it is by no means the only evidence that supports high-risk care management or care transition support. Experimentation since the launch of the Camden model indicates that patients with this level of complexity require far more than three months to stabilize and alter behavioral patterns. Newer programs increasingly prioritize community-based staffing; cultural competency and humility; and preventive outreach prior to admission. The most promising interventions are those that measure success based on patient-centered measures like increased trust and activation levels.

Ultimately, population health doesn't have all the answers yet. We need new perspectives, a variety of models, and more randomized controlled trials. Let's use this uncertainty to avoid complacency and continue making progress.

For more of the tools that can help you succeed, like governance documents, needs assessments, and care decision guides, view our Population Health Resource Library.

Access the Library

Your guide to implementing a successful super-utilizer program

Use this guide to design or audit your super-utilizer clinic model across program elements such as patient eligibility criteria, staffing structure, patient management and care coordination processes, and funding.

Download Now
Download Now