Social determinants of health

Generative artificial intelligence can be used to pull social determinants of health data, like housing or employment status, from clinician notes to identify patients who need additional support, according to a study published in January 2024.

Large language models trained by researchers could identify 93.8% of patients with adverse social determinants of health, while official diagnostic codes include that data in only 2% of cases. 

The finely tuned models were also less likely than OpenAI’s GPT-4 to change their determination when demographic information like race or gender was added. Algorithmic bias is a major concern for AI use in healthcare, amid fears the technology could worsen health inequities. 

Large technology firms like Microsoft, Google, Oracle and Amazon have invested heavily in generative AI over the past year. Many firms have honed in on using the tools, which can create original text and images, to lessen administrative burden for providers, through use cases like automatically documenting conversations between patients and doctors. 

But using large language models to extract social determinants of health data is another area of potential promise for the technology, according to the study published in npj Digital Medicine. 

Social determinants, which include information like access to transportation, stable housing and healthy food, are key to health outcomes. But the data often isn’t systematically organized in health records, making it challenging to pull that information into databases or easily identify patients who may need help.  

“Algorithms that can notice things that doctors may miss in the ever-increasing volume of medical records will be more clinically relevant and therefore more powerful for improving health,” Danielle Bitterman, a study author and faculty member in the Artificial Intelligence in Medicine Program at Mass General Brigham, said in a statement.

Researchers set out to build language models that could extract that data from health records. They reviewed 800 clinician notes from 770 patients with cancer who received radiotherapy, tagging sentences that referenced social determinants like employment status, housing challenges, transportation issues, parental status and social support. 

The researchers then tested the models with an additional 400 notes from patients who had been treated with immunotherapy. 

Their finely tuned models could consistently find references to social determinants, even though sentences could be linguistically complex and few sentences in the notes referenced social issues at all, researchers said. 

The study’s authors also created synthetic clinical notes using ChatGPT, and used that data to compare their models’ performance against the general language model GPT-4.

Though both model types changed their determinations about social determinants when demographic and gender information was added, the discrepancy rate for the finely tuned models was half that of GPT’s.

Fine-tuning language models could be one strategy to lessen the risk of bias, but more research is needed, Bitterman said in a statement. 

The CMS is exploring programs that would pay social or community health workers to address patients’ social needs in a bid to invest more heavily in food, housing, transportation and other social determinants of health, according to agency officials.

“We are looking at that. For example, in maternal health, thinking about the role of doula and community health workers,” Liz Fowler, director of the Center for Medicare and Medicaid Innovation, said on Thursday during the CMS’ inaugural health equity conference.

Social determinants of health are a significant driver of health outcomes and medical costs, especially for vulnerable populations. Despite widespread acknowledgment of the importance of addressing SDOH, investments tend to be modestly funded or temporary, and stalled during the COVID-19 pandemic, data shows.

The cost for primary care practices to screen patients for social needs is substantial, but pales in comparison to the cost of interventions, according to one recent study.

The study estimated that existing federal funding covers less than half of the cost of providing food, housing, transportation and care coordination assistance.

Advocates have been calling for greater investment in the nation’s underfunded social care programs, to address downstream factors that result in poor health outcomes. Though it might seem backward for healthcare programs to address things like food or housing insecurity, the reality is that they have more money to spend, Fowler said.

However, legal requirements stipulating who Medicare is allowed to pay — essentially, traditional healthcare providers — can hamstring the creation of more social-focused programs, said Meena Seshamani, CMS Medicare director.

As a result, regulators in the Biden administration are trying to find ways to expand or work around those restrictions. A new CMMI model announced Thursday gives funding to primary care physicians, but encourages them to use it to contract with a community-based organization, Seshamani said.

“We’re also exploring, particularly in our accountable care organizations — traditionally, you had to see a primary care physician to become part of one of these models. And we’re thinking about, what about nurse practitioners? What about a social worker? Are there other people who are involved in someone’s care on a regular basis where it would make sense to say, OK, that becomes the hub for these models?” Seshamani said.

CMS officials made their case for fundamentally changing healthcare programs like Medicare to refocus around health equity — a key priority for the Biden administration — at the conference this week.

The CMMI, created by the Affordable Care Act, tests new payment models to try to push the industry toward more value-based care. The agency, which receives $10 billion every decade to invest in new models, paused or ended a number of models a few years ago following watchdog and congressional criticism that few models to date have resulted in cost savings or better care quality.

Since then, the CMMI has realigned its models, which touch tens of millions of Americans, to focus more heavily on accountable care and health equity.

Now, all of CMMI’s models require providers to collect social determinants of health data. The CMMI has also committed to include patient-reported outcomes measures in all of its models, and is currently looking at other factors it can incorporate to accurately capture disparities, Fowler said.

For example, regulators currently define underservedness based on dual-eligibility status and area deprivation index, but are considering including other factors like life expectancy, Fowler said.

The CMS is also working to align internal metrics around health equity. Currently, different Medicare programs in fee-for-service and Medicare Advantage have their own set of measures.

“It presents a really good opportunity to become more granular with these measures, so that we can really see across different types of people how are things being done, so that you can really drive equity in everything,” Seshamani said.

The CMS is also working to expand a CMMI concept allowing providers from rural and underserved communities to apply for upfront funding to invest in health-related social needs to the Medicare Shared Savings Program, Seshamani said.

Telehealth firm Amwell is partnering with Nestlé Health Science to create digital health programs focused on nutrition, the two companies announced in October 2023.

The partners’ first offering, the Advanced Surgical Nutrition Automated Care Program, pairs Amwell’s virtual companion service with Nestlé’s Impact Advanced Recovery drink and is available for patients undergoing major elective surgery in the U.S. 

Enrolled patients receive the drink while the digital program offers patient education and guidance, evaluates progress and adherence and can prompt clinicians to assist if necessary.

The program could “help strengthen patients’ pre-operative journeys and foster better care plan compliance,” according to an Amwell spokesperson. The companies didn’t disclose financial terms behind the deal, but said they will work on other initiatives that aim to improve outcomes and lower costs for health systems.

Amwell, one of the nation’s largest telehealth companies by market share, is in the midst of migrating its clients to its Converge software, a newer platform that aims to consolidate its offerings and integrate third-party tools. 

During its second quarter earnings call in August 2023, the company said 43% of visits during the period took place on Converge. But Amwell also lowered its revenue guidance for the year after missing analyst expectations and racking up an impairment charge.

The vendor recently added a new lab offering within the Converge platform through a partnership with data exchange company Health Gorilla, allowing providers expanded access to order labs and view results from patients’ lab networks.

Earlier in the spring 2023, Amwell inked a deal with digital chronic condition management firm DarioHealth to offer a cardiometabolic health program through its platform. 

The latest program “is the first of several initiatives” that the telehealth company is working on with Nestlé, Murray Brozinsky, chief strategy officer at Amwell, said in a statement.

Amwell’s automated care programs use virtual companions to provide information like discharge instructions and can escalate users to a higher level of care if needed. Nestlé Health Science makes consumer health, medical nutrition products and nutritional pharmaceutical therapies. 

Dive Brief:

• Uber will soon be able to deliver groceries and other over-the-counter items to patients of its payer and provider clients, in the latest benefits expansion of its Uber Health platform.
• Uber Health, which already provides non-emergency medical transportation and prescription delivery, has been expanding beyond medical transportation to care coordination across multiple benefits, Catilin Donovan, the head of the division, said in an interview. Soon, provider users will have access to patient benefit data and eligibility files from their health insurers, so they can prescribe services they know are covered.
• Uber is not yet giving specifics of Uber Health’s financial performance, but Uber Health grew bookings by 75% year over year in the first quarter. Donovan declined to say whether Uber Health was profitable.

Dive Insight:

Uber created Uber Health in 2018 as an NEMT dashboard allowing healthcare workers to book rides for patients to and from medical appointments. The ride sharing app has spent the past five years expanding its capabilities, and in March added same-day prescription delivery.

The company’s expansion into grocery and OTC item delivery is meant to help clients improve value-based care and population health programs without adding extra burden on patients, and represents Uber Health’s pivot from an NEMT provider to a broad web-based tool that allows doctors’ offices, health plan workers, case managers, social workers and other providers to request things for patients based on their health needs, Donovan said.

Other delivery platforms have launched food-as-medicine initiatives, including Instacart, joining a number of startups with the goal of linking medically tailored meals with convenient access. Food-as-medicine, which has historically been viewed as an important but niche area of healthcare — and one that faces large systemic problems — has been attracting more attention. Last year, the White House committed more than $8 billion to fight malnourishment.

Uber claims it’s differentiated from other healthy food delivery or NEMT vendors because it also addresses medication adherence and benefits coordination.

“Now we’re a one-stop shop,” Donovan said. “If you think about what NEMT providers do today, or what supplemental benefit providers do in general today, their work is incredibly important. But it’s incredibly siloed.”

Healthcare companies partner with NEMT players to reduce patient no-shows, shifting the reliance on emergency services to upstream preventive primary care, especially for vulnerable low-income populations. The downstream costs of missed medical visits in the U.S. are estimated at roughly $150 billion annually, and 3.6 million people miss appointments due to lack of transportation each year, according to research.

Uber competitor Lyft has also been building out its NEMT division after launching in 2016.

Uber and Lyft directly compete with other medical transportation companies and brokers in the NEMT market. Some NEMT players have accused the large rideshare companies of not having the healthcare experience or specialization of vehicles and services to address the needs of some populations.

Uber Health, which has grown steadily since its founding, is paid by its more than 3,000 healthcare customers in a few different ways — per ride or visit, or a fee schedule or per-member per-month charge for higher volume clients. Generally, patients don’t have to pay out of pocket for Uber Health services, according to Donovan.

Dive Brief:

• Kroger, the network of grocery stores, has announced its third Medicare Advantage plan offering in partnership with a health insurer.
• Kroger’s new MA plans with Select Health, the insurance subsidiary of nonprofit hospital operator Intermountain Health, are meant to focus on health and wellness by providing customer savings on groceries and prescription medications at Kroger stores and pharmacies, according to the Tuesday press release.
• The plans, which are targeted at Medicare-eligible Kroger shoppers, will be offered in select counties in Colorado, Idaho, Nevada and Utah for the 2024 plan year.

Dive Insight:

Like other major retailers, including CVS, Walgreens and Walmart, Kroger has been moving more into care management and delivery, looking to leverage its extensive network of groceries, pharmacies and retail clinics to connect with patients in the ongoing shift to community and home healthcare.

Kroger took its first step into MA plans in 2021, when the Cincinnati-based grocery chain partnered with Anthem Blue Cross Blue Shield (which has since rebranded to Elevance Health) on co-branded MA plans in Ohio, Kentucky, Georgia and Virginia.

Last year, Kroger announced its second co-branded MA plan with Priority Health in select counties in Michigan.

Select is a nonprofit plan that currently covers more than one million members in Utah, Idaho and Nevada. The Intermountain subsidiary was interested in partnering with Kroger because many Medicare beneficiaries shop and fill prescriptions at its stores, Jason Worthen, Select Health vice president of government markets, said in a statement.

Enrollment has steadily climbed in Medicare Advantage plans, where private health insurers contract with the government to cover the care of Medicare seniors.

The plans, which can offer nontraditional benefits such as nutrition services, in 2023 grew to cover more than half of all Medicare beneficiaries, despite concerns over skyrocketing federal spending on MA.

Health insurers have inked agreements with retailers to offer members access to a lower-cost site of care and differentiate their offerings in the crowded MA market. Walmart, for example, offered a joint MA plan with Clover Health in Georgia in 2021.

Kroger has been expanding its offerings in healthcare, even creating a sector-specific division, called Kroger Health.

Kroger Health includes 2,200 retail pharmacies in 35 states and a medical clinic network of almost 230 locations in nine states, along with a specialty pharmacy and infusion service.

In January, the healthcare division launched a clinical trial site network to find and enroll patients in clinical trials, and help research organizations, pharmaceutical companies and health systems conduct select trials.

Kroger Health delivered positive sales growth in Kroger’s first quarter results released in June, despite the effects of a contract dispute with pharmacy benefit manager Express Scripts. Kroger decided to take its pharmacies out of network with the Cigna subsidiary last year, after determining it would be filling prescriptions below cost under Express Scripts’ contract proposal.

In December, Kroger said the contract termination would reduce sales by about $100 million in the one month that remained of its fiscal fourth quarter.

Kroger did not respond to a request for comment on its healthcare business strategy.

Medicare patients treated in acute hospital-at-home programs had low levels of mortality and rarely needed to return to facilities for care, according to a study published in Annals of Internal Medicine in January 2024.

Researchers found that 0.5% of patients in the study died while receiving acute care in their homes, while 6.2% had to go back to the hospital for at least 24 hours.

The results could help make the case to continue supporting hospital-level care delivered at home, researchers said. A waiver that expanded the programs at the height of the COVID-19 pandemic is set to expire at the end of 2024.

The CMS launched the Acute Hospital Care At Home program in 2020, looking to help facilities increase their capacity as they struggled to manage a surge of COVID-19 patients. Thousands of patients from 300 hospitals have been treated in their homes thanks to the waiver, according to the study.

Researchers analyzed clinical characteristics and outcomes from more than 5,800 patients who received care under the Acute Hospital Care at Home waiver, using traditional Medicare claims filed between July 2022 and June 2023. 

Most of the patients were older — more than 40% were at least 80 years old — and had complex medical conditions. About 43% had heart disease or chronic obstructive pulmonary disease, while around 22% were diagnosed with cancer. 

But patients experienced low rates of mortality while undergoing care in hospital-at-home programs, and findings were consistent across demographic groups that frequently suffer worse outcomes, like Black or Latino patients or those who were dually eligible for Medicare and Medicaid. 

Researchers also followed patients shortly after their home care programs ended. Within 30 days of discharge, 3.2% of patients died, 15.6% were readmitted and 2.6% went to a skilled nursing facility.

Hospital-level care at home could be beneficial because providers can show patients how to care for themselves in their homes once they are discharged, said corresponding author David Michael Levine, clinical director for research and development for Mass General Brigham’s Healthcare at Home. They are also more likely to be able to stand up and move around while at home compared with a hospital setting. 

“In addition, the clinical team has a greater ability to educate and act on the social determinants of health that we see in the home. For example, we can discuss a patient’s diet right in the kitchen or link a patient with resources when we see the cupboards are bare,” Levine said in a statement. 

But more research is needed to compare home care with brick-and-mortar hospitals, the study’s authors said. Rural patients, who are at greater risk for poor health outcomes and frequently need to travel further for care, were also underrepresented in the study.

Dive Brief:

• More than 20% of people without a personal vehicle and with only poor or fair access to public transportation said they had foregone healthcare needs because of difficulty finding transportation, according to a report this week from the Urban Institute.
• Adults with a disability, Black adults, people with low incomes and those on public insurance were more likely to report going without needed healthcare because of transportation issues, it found.
• Overall, 5% of U.S. adults said transportation barriers kept them from care they needed, the Urban Institute said.

Dive Insight:

Payers and providers in recent years have bolstered their efforts to address social determinants of health such as lack of transportation. Missed medical appointments can cost the U.S. healthcare industry as much as $150 billion annually.

Insurers, particularly in the Medicare Advantage program, have expanded benefit offerings to include a ride to a doctor’s appointment in some cases. Providers have partnered with companies such as Uber and Lyft to provide transportation.

Lyft has been in the non-emergency medical transportation business since 2016 and launched an additional patient request option in 2021. Uber started Uber Health in 2018 with an NEMT platform. The company most recently expanded into same-day prescription delivery. Still, some research has questioned the effect Uber and Lyft have on helping people make appointments.

“For policymakers, these findings highlight a potential return on investments in public transit in the form of improved access to health care, as well as the gaps that may need to be filled by Medicaid coverage of nonemergency medical transportation or expanded access to telehealth when public transit options are limited,”  the report said.

Telehealth services have seen an uptake during the COVID-19 pandemic as people stayed home and avoided doctor’s offices. Federal regulations were relaxed to allow for this, but some are still considered temporary.

Dive Brief:

• Medicare Advantage's quality bonus program has had no observable difference in plan quality despite costly federal investments, a new study suggests, bolstering calls from some stakeholders to substantially revise or even eliminate the program altogether.
• The study published in Health Affairs Monday analyzed nine claims-based measures of quality for MA beneficiaries versus a control group of commercial enrollees. Program participation was associated with significant quality improvements among MA beneficiaries on four measures, significant declines on four other measures and no significant change in overall quality.
• The findings jibe with past research suggesting the quality bonus program has had little-to-no effect on quality, despite the government awarding MA plans about $30 billion in quality bonus payments through the program over its first seven years — hinting the American people may not be getting enough bang for their buck.

Dive Insight:

The government has publicly reported MA plan quality in the form of a star ratings system since 2007. Ratings range from one to a high of five stars based on plans' calculated performance across dozens of measures relating to preventative care, chronic disease management, patient experience and more.

Medicare introduced the quality bonus program in 2012. The program links financial bonuses to commercial insurers' quality performance in the increasingly popular MA program, which covers roughly 42% of all Medicare beneficiaries, as measured by star ratings.

Unlike other value-based initiatives in traditional Medicare, which are designed to be budget-neutral or lower spending, the quality bonus program actually sets aside additional funding for payers. The incentives are pricey, totaling $6 billion in 2019 alone and making up 2.3% of aggregate payments to private insurers in MA, according to the Medicare Payment Advisory Commission.

But despite the major investments in the program (and some overhauls to its incentive structure and performance measures), its efficacy has largely been unclear.

The new study analyzed insurance claims from 2009 to 2018, comparing changes in performance between millions of MA beneficiaries and a control group of commercial enrollees. It's the first evaluating the quality bonus program using a broad swath of performance measures and a control group, researchers said, and they found disappointing results.

The findings were consistent with other studies, including one from two years ago, finding the quality bonus program was associated with minor or no changes in quality.

"Given the massive incentives in the quality bonus program and plans' focus on star ratings, it is perhaps surprising the program was not associated with improvements in quality," the University of Michigan researchers said.

There are myriad possible reasons. Plans may be more focused on the nonclinical administrative measures that contribute to star ratings, such as customer service, that they have more control over and still contribute 25% of the weight for star ratings calculation. Or, it may be difficult for plans to pass down incentives to physicians, who may have a more direct impact on quality of care.

The study backs arguments from some Medicare watchdogs, including MedPAC, that the quality bonus program could use a substantial facelift to address key deficits in the current rewards model and move it toward budget neutrality.

That recommendation is further supported by other recent studies finding that doubling bonuses in certain counties didn't increase enrollment or quality, but did result in payment disparities to plans.

Winnowing unnecessary spending in the Medicare program is looking increasingly important as the trust fund that finances Medicare's hospital benefit is expected to run dry in just five years, absent legislative action. Currently, program spending is expected to balloon from 4% of the country's gross domestic product to 6.2% by 2045.

According to Congressional Budget Office estimates, eliminating benchmark bonuses would reduce Medicare spending by $94 billion between 2021 and 2028.

Dive Brief:

• Most physicians in group practices owned by health systems are paid based on the volume of care they provide rather than value, despite payer efforts to move toward more value-based care, according to a study published Friday by the nonprofit Rand Corporation and published in JAMA Health Forum.
• Volume-based compensation was the most common type of base pay for more than 80% of primary-care doctors and more than 90% of specialists at medical practices owned by health systems, according to a survey conducted between November 2017 and July 2019.
• Systems still used financial incentives for quality and cost performance measures, but the percentage of total physician compensation based on quality and cost was just 9% for primary-care providers and 5% for specialists, according to Rand.

Dive Insight:

Public and private payers in recent years have pushed for payment reforms encouraging providers to improve care quality and slow spending growth by making care more value based. Still, physicians overall still are paid primarily based on volume rather than the quality and value of healthcare services provided, according to the Rand study.

"The payment systems that are most-often in place are designed to maximize health system revenue by incentivizing providers within the system to deliver more services," Rachel Reid, the study's lead author and a physician policy researcher at Rand, said in a statement.

Researchers examined physician payment structures used in 31 physician organizations affiliated with 22 health systems located in four states. They also interviewed physician organization leaders and surveyed practices to characterize compensation arrangements for both primary care and specialty physicians.

Physicians most commonly reported increasing the volume of services performed as a way to increase their compensation, with about 70% of practices following those plans, and in those cases, incentives based on volume accounted for more than two-thirds of their compensation.

While incentives based on clinical quality, cost, patient experience and access to care also were commonly included in physician pay, those payments represented a small fraction of doctors' total compensation and are "likely to only marginally affect physician behavior," the study said.

"For the U.S. healthcare system to truly realize the potential of value-based payment reform and deliver better value for patients, health systems and provider organizations will likely need to evolve the way that frontline physicians are paid to better align with value," Reid said.