番茄社区

Certain services from the famous "hotspotting" pilot program in Camden, New Jersey, failed to reduce readmissions in a randomized study of "superutilizers" with unusually high healthcare use and costs, researchers reported.

An intensive program for people with medically and socially complex conditions in Camden, one of the nation's poorest cities, the program provided home visits, scheduling and accompaniment to primary and specialty care visits, patient coaching, and assistance for applications to social services and appropriate behavioral health programs, among other services.

Yet hospitalized patients wound up with statistically similar rates of hospital readmission in the 180 days after discharge whether they had been randomized to the experimental care-transition program or usual care (62.3% vs 61.7% controls, P=0.81), according to a research group led by economist Amy Finkelstein, PhD, of Massachusetts Institute of Technology, Cambridge, reporting in the .

Of note, though, it wasn't clear that the "usual care" control group was getting real-world usual care, as Finkelstein and colleagues acknowledged that a number of city-wide initiatives on post-discharge care were underway during the study period.

Still, Jeffrey Brenner, MD, who founded the Camden Coalition of Healthcare Providers and its pilot program, conceded that the intervention wasn't as effective as hoped.

"The key here is that navigation and coordination are important, but insufficient in and of themselves to improve outcomes. You've got to improve care," said Brenner, who won a in 2013 for his on high-cost, high-needs patients.

"If you're an average person on an average day, you get average care," Brenner said in an interview. "Average services don't work for complex patients.... You need very specialized training to take care of people with trauma."

Ongoing Efforts to Solve an "Existential Problem"

"When we set out 10 years ago to figure this out, the dominant opinion both nationally and in Camden was that the problem in the patient's experience was one of disorganized and confusing care," according to Brenner. Now, after all this time helping to organize and navigate care for patients, it appears that it's actually the care itself that doesn't work for these patients, he said, citing problems such as incorrect diagnoses, unnecessary testing, and inappropriate treatment.

For example, homeless people put in shelters actually get sicker, so his team started a program where the homeless get their own apartments and receive help moving in, getting groceries and furniture, and accessing local care.

With this additional component, the Camden model continues to operate and evolve since the version tested in the study by Finkelstein and colleagues, according to Mark Humowiecki, JD, senior director of national initiatives for the Camden Coalition.

Ultimately, it's an "existential problem" for superutilizers, Brenner suggested. "These patients have really high rates of early-life trauma, sexual and physical abuse. The outcome of all that toxic stress in childhood is that it rewires your brain." This is how people end up with a lot of medical complexity including high rates of addiction, mental illness, he added.

There is still a role for care coordination in improving health outcomes, but the healthcare system should also remain humble about what can be expected from those interventions without deeper, structural changes to promote health, according to Margot Kushel, MD, of the University of California San Francisco.

"The participants in this study are facing multiple challenges -- profound housing instability, deep poverty, trauma, the insidious effects of structural racism -- etc. The solutions to this will likely come from addressing the underlying structural conditions -- housing, economic security, trauma," Kushel commented in an email.

Brenner left the Camden Coalition 3 years ago to work for UnitedHealthcare, where he continues to work on housing for medically complex patients in other parts of the country.

Today, the Camden Coalition works with an adult population and also runs a maternal health program for mothers and babies, according to Humowiecki.

What Happened in Camden?

Included in the trial by Finkelstein's group were 800 hospitalized superutilizers identified as having been hospitalized 1.8 times on average in the 6 months before their index admission.

Study participants had been recruited from Cooper University Hospital and Our Lady of Lourdes Hospital in Camden. They comprised a cohort with chronic conditions and complex needs -- the sickest patients ever enrolled in a care-management intervention study, according to Brenner.

Patients were generally highly engaged with the Camden program, with 95% of those in the treatment group having at least three encounters with program staff and the group as a whole participating in the program for a median 92 days.

Notably, the rate of hospital admissions had risen sharply in the 6 months before the index admission, compared with the previous 30 months, and then fell rapidly afterward in both intervention and control groups. Finkelstein and colleagues emphasized, however, that the average readmission rates in the succeeding 12 months were substantially higher, in both groups, than in months -36 to -6 relative to the index admission.

The researchers noted that usual care in Camden was "evolving" and improving during the trial period, given the establishment of multiple other care-management programs, including a city-wide campaign -- run by the Camden coalition -- to connect patients with primary care within a week of hospital discharge.

Regression to the mean is also likely at play: people who use a lot of care in one year don't tend to use much in the ensuing years. This is a "large problem" that has led to misinterpretation of many social experiments that rely on pre-post data, Kushel noted.

"This underscores the importance of using randomized evaluations for understanding the impact of health care policies; issues like regression to the mean and other sources of bias can produce very misleading results in observational studies," Finkelstein told 番茄社区.

Half the study cohort were men, and over half in the 45-64 year age range. By race, over half were black, and 30% Hispanic. Nearly all were unemployed and on Medicaid or Medicare.

Finkelstein's team acknowledged that their study was not powered to analyze the effects of the Camden model by patient subgroup.

"The data did not permit evaluation of potential non-tangible benefits such as improved relationships with providers. Nor did the data allow comparison of outpatient care for the treatment and control groups," the investigators added.

Comment