Alina Schnake-Mahl, MPH, ScD, Evaluation Scientist, Cityblock Health and Pooja Mehta, MD, Women’s Health Lead, Cityblock Health
In 2011, a new healthcare program based in Camden, New Jersey, gained national attention and made “healthcare hotspotting” a common term in the industry. The program was built on the idea that people who frequently access emergency or hospital care often experience overlapping medical conditions and challenges in their daily lives — impeding health and driving up costs.
Recently, though, a rigorous study evaluating the impact of the Camden Coalition program showed that people with two hospital admissions in a six-month period who were randomly assigned to a care transition intervention did not use less hospital care than similar “super-utilizers” who did not receive the intervention. Though it was not the result the Coalition hoped for, at Cityblock, we’re actually encouraged by the lessons we can glean from their work in Camden.
The Camden Coalition, like Cityblock, is embedded in lower-income communities and provides team-based care to individuals (whom we call “members”) with complex health and social needs. These social factors — such as the building in which one lives, a history of trauma during childhood, or not knowing where one’s next meal is coming from — are not always obvious to providers of traditional care in the rushed minutes of a medical visit.
One potential explanation for why the study intervention did not work, raised by the report’s authors and stakeholders, is “regression to the mean.” That is, the statistical concept that individuals with extreme values (like being admitted to the hospital at rates far higher than the majority of the population) at one point in time will naturally, and without intervention, return to average values. We’ve also reflected on other factors, related to the build and design of the intervention, that we believe help explain the results. Here’s what we’re taking back to our work.
Rather than screen and refer, we must truly integrate and reshape primary care delivery with, not just for, the members we serve.
Cityblock is a provider organization that delivers team-based care to our members in their homes and communities using value-based payment arrangements. As such, we were struck by one of the take-home points from the study: that “putting a patient with numerous chronic conditions in touch with a primary care doctor for 15 minutes” might be “coordinating to nowhere.” These results are a call to action for us to bring primary care and behavioral health providers — specifically, those who accept Medicaid — to communities where they are not currently available.
These results are also a call to redefine the care that primary care and behavioral health providers offer around member needs, outside of the limitations imposed by a fee-for-service system. At Cityblock, this looks like care specifically designed for those with complex health and social needs. We match our members with a designated Community Health Partner who serves as their ally on their health journey, and who we consider the quarterback of our care teams in coordinating and managing care. They work with clinicians at the frontline together, in homes and convenient, welcoming Neighborhood Hubs instead of hospitals — connecting with members well beyond the boundaries of a 15-minute appointment in a doctor’s office.
Furthermore, we believe that the most effective way to provide truly integrated care delivery is by working directly with payers in contracting arrangements that allow flexibility in how and where care happens. With this payment structure in place, we can bring members and their families directly into the design process for their own care plans. As the experts on their own lives and experiences, we ask members to tell us what they need, how they want care provided, by whom, and over what period of time. When members have a say in their own care, we believe we can iterate effectively towards other critical outcomes — such as reducing hospital admissions and inpatient stays.
We have to look beyond individual social determinants of health towards structural change to meaningfully improve outcomes.
For those who live with structural challenges related to long-term neglect and discrimination — such as lack of access to affordable housing, poor access to nutritious food, or limited public transit systems — providing immediate social support and short-term referral to social services following hospitalization may not have a significant impact on acute utilization. The Camden study identified potential study participants by reviewing care needs captured in traditional electronic health records. Individuals who had factors indicating medical and social complexities — such as multiple medications, lack of social support, homelessness, prior difficulty accessing services, mental health comorbidities, or active use of illicit or licit drugs — were invited to participate and then randomized to receive or not receive the study intervention. Yet we know that much, if not most, social complexity is not captured in the typical EMR (though we’re encouraged by Z-codes!), and so we must identify members who would benefit from our services using additional data sources.
With appreciation for what we have learned from Camden, we are deeply investing in building robust social screenings so we can see a 360-degree picture of members' stories — to understand social determinants of health like members’ housing situation, and also what brings them joy and helps them build resilience. This critical need to address social factors is clear in the Camden Coalition’s evolving response to their work, which, like our Cityblock model, now emphasizes the presence of health hubs, health information exchanges, and relationships with payer and policy partners that can enable broader investments in community infrastructure.
As providers, we have opportunities to mitigate the impact of social challenges on inequitable health outcomes. We can bring our socially isolated members in need of companionship together at community events. We can create local economic development and effective, empathetic therapeutic relationships by hiring care teams from the communities we serve, creating professional pipelines where none may exist. We can advocate for policy change at the state level to expand Medicaid access to the uninsured, and also access to high-quality food, education, workplace protections, and housing.
We believe all of this can have a real impact, but it takes time and thoughtful work with those who are experts in a community leading the way, and with partners like payers, community-based organizations, and policymakers.
This work is complicated, and it necessitates supporting those facing challenges at various points in their care journey — before the crisis point.
The months following a hospital admission are critical points of engagement with primary care. Previous experimental research shows post-transition care can effectively reduce readmissions. The timeframe of the Camden coalition — six months maximum with an average engagement of 90 days — is an impressive start, particularly with the high number of touchpoints that participants received during this time. However, given how deeply entrenched the challenges are, we may have to allow more time and more highly individualized approaches to improve health and reduce acute utilization.
To meaningfully move the needle on outcomes that matter to our members, including but not limited to hospital readmissions, we need to engage those with “rising risk” before the crisis point — taking time to develop longitudinal relationships between care teams and members. The Camden Coalition selected individuals during a hospitalization, a definitionally extreme point of utilization. The study and subsequent coverage have highlighted the likely regression to the mean among both the intervention and control groups, due to selection in the hospital. This selection criteria in part reflected the intervention focus — care transitions.
Primary care models like Cityblock, however, cannot limit our cohorts to individuals who are currently in the hospital. We need to intentionally select heterogeneous populations with varied risk-levels, not just the top 1% of utilizers, because we know that the top utilizing group is not consistent year to year. We have further to go to better understand who exactly the “rising-risk” population encompasses, but we believe that longer interventions tailored to preventing and actively managing existing conditions allow us to not only help people break a cycle of high utilization, but also prevent a broader population from getting into that cycle in the first place. We’ve also built assumed regression to the mean into our contracts, and estimate the efficacy of our model based on reduction in costs, utilization and improvements in health, and net of mean regression.
In closing, the work must continue.
We are appreciative of the Camden Coalition and the MIT Poverty Action Lab for these valuable lessons to guide us, and we also acknowledge the risks of over-generalizing the results of one study. The study measured the effect of a specific intervention, at a specific time, in a specific context, with a specific population, and we should be cautious of directly applying these findings to other models addressing both social and medical needs.
Perhaps the most important result of the trial was that the Camden Coalition learned what didn’t work — and they have already begun to adapt their model. We can all apply their findings to improve our models and test new ways to improve health for the members and communities we serve. We need to not just measure the overall impacts of our model, but also conduct rigorous research and evaluation designs to truly understand the impact and to share our results. By doing so we avoid learning in silos, and instead, collectively build towards a system that costs less, drives better outcomes, and contributes to better health in communities across the country.
Pooja Mehta leads Women’s Health at Cityblock. She is an obstetrician-gynecologist, health services researcher, and health policy expert.
Alina Schnake-Mahl leads evaluation at Cityblock. She is a social epidemiologist and health services researcher and has a doctorate and MPH in social and behavioral sciences in public health.