Predictive analytics could help states more effectively target HCBS and SDoH services in Medicaid Bookmark has been added
Predictive analytics could help states more effectively target HCBS and SDoH services in Medicaid
Health Care Current | April 30, 2019
This weekly series explores breaking news and developments in the US health care industry, examines key issues facing life sciences and health care companies, and provides updates and insights on policy, regulatory, and legislative changes.
Predictive analytics could help states more effectively target HCBS and SDoH services in Medicaid
By Jim Hardy, specialist executive, State Health Transformation Services, Deloitte Consulting LLP
State Medicaid agencies and managed care organizations (MCOs) have flexibility to offer optional or traditionally non-covered services to beneficiaries under the presumption that these services might improve the health and functionality of beneficiaries and reduce spending in hospitals, nursing homes, and other institutions. Two examples of non-covered services in Medicaid are:
- Home and community-based services (HCBS): One long-standing policy objective in covering home-based long-term services and supports (LTSS) has been that they might reduce the use of nursing homes. These benefits have been popular with beneficiaries and their families because people generally prefer to receive services at home as opposed to living in a nursing home or other institution.
- Coverage of services to address social determinants of health (SDoH): There is optimism that addressing the housing, nutrition, and other social needs of Medicaid members could result in fewer ambulance rides, fewer emergency room visits, and fewer hospitalizations. SDoH represents a newer set of interventions, but the underlying thinking is similar to HCBS programs.
HCBS coverage has been provided for many years, while SDoH coverage is in its early stages. What have we learned from HCBS programs so far? While these programs are popular with beneficiaries and advocates, enrollment growth has not led to a corresponding decline in the use of institutional care.1 I could envision this happening with SDoH services as well if we don’t target those services to the people who are most likely to benefit from them. My take is that predictive analytics could help Medicaid departments and MCOs more accurately target these services by using a state’s resources more efficiently and effectively.
A brief history of home and community-based care
Congress created the HCBS program in 1983 through section 1915(c) of the Social Security Act. HCBS waivers and new HCBS state plan options now exist in every state and provide services to about 4.6 million people at a cost of nearly $83 billion a year in state and federal funds.2 Many waivers are specific to certain populations, such as individuals with physical disabilities who need assistance with accomplishing daily activities, or beneficiaries who have specific diseases. Seniors and adults who have physical disabilities make up more than half of all waiver enrollment, followed by people with intellectual or developmental disabilities.
The waivers allow states to provide non-Medicaid covered services to targeted populations and allow them to limit the number of beneficiaries who can receive those services. But states must demonstrate that providing HCBS services would be less expensive than moving the beneficiaries into institutionalized care. States and CMS have been committed to re-balancing long-term spending by shifting more spending to less costly HCBS programs.
From one perspective, this re-balancing has already occurred. In fiscal year 2008, HCBS accounted for 43 percent of long-term care spending, and by fiscal year 2016, that percentage had increased to 57 percent.
These statistics are often cited to promote the success of states’ re-balancing initiatives. Upon closer inspection however, it becomes clear that the shift in long-term care services is the result of HCBS spending increases—rather than a corresponding decrease in institutional spending. In other words, while HCBS now accounts for a larger percentage of long-term care spending in Medicaid, it has not driven up overall long-term care spending.
Spending growth in HCBS and institutional long-term care
Why hasn’t the growth in HCBS led to lower nursing home use?
Increased HCBS spending reflects more people receiving services, people using more services, and rising costs for those services.
So why hasn’t the use of institutional care fallen? Possible explanations include:
- Demand for institutional services is growing faster than the supply of HCBS services.
- Institutional services tend to be driven by supply rather than demand, meaning people will take advantage of nursing homes if space is available.
Another factor that I think is important—and which could help explain why institutionalization costs haven’t decreased as HCBS spending has increased—is that existing assessment tools typically aren’t effective at identifying members who are most likely to need nursing home care. These tools tend to have an eligibility focus rather than a risk focus. They help Medicaid programs determine whether an individual meets the criteria for being admitted to an institutional setting. However, they don’t assess whether the individual is at a high risk of being admitted into an institution if HCBS services are unavailable.
In 2017, more than 700,000 beneficiaries were on HCBS waiting lists across 40 states—up 8 percent from the prior year.3 The fact that so many people are on the waiting list for HCBS services suggests that eligibility criteria might not be a good indicator of institutionalization risk.
Predictive analytics might help better target benefits
Some SDoH initiatives incorporate predictive analytics, which helps identify characteristics that are likely to be associated with the use of health care services. This can help MCOs and Medicaid agencies more effectively determine which beneficiaries should get referred to services. It also can help states decide which services to cover, and what other community resources might serve as effective interventions.
States might consider whether predictive analytics could be incorporated into HCBS programs to direct benefits to the people who are most likely to use institutionalized care. This could both effectively target HCBS spending and reduce institutional care. This technology could become part of the initial assessment to holistically determine the individual’s risk of institutionalization. Analytics might also be used to decide which beneficiaries should be promoted from waiting lists, or to direct HCBS resources more appropriately.
State Medicaid programs might want to look at the reasons institutional care costs have not been offset by growth in HCBS spending and use new tools to increase the value of HCBS programs. States that implement SDoH programs might also want to determine how to ensure spending is aimed at individuals who are most at risk of a decline in health status so that they can leverage the lessons learned from the HCBS experience.
1 CMS/medicaid.gov, Medicaid Expenditures for Long-Term Services and Supports in FY 2016 (https://www.medicaid.gov/medicaid/hcbs/index.html)
2 Medicaid Home and Community-Based Services Enrollment and Spending, April 4, 2019, Kaiser Family Health Foundation (https://www.kff.org/medicaid/issue-brief/medicaid-home-and-community-based-services-enrollment-and-spending/)
3 Key Questions About Medicaid Home and Community-Based Services Waiver Waiting Lists, April 4, 2019, Kaiser Family Health Foundation (https://www.kff.org/medicaid/issue-brief/key-questions-about-medicaid-home-and-community-based-services-waiver-waiting-lists/)
In the News
HHS initiatives encourage value-based payments in primary care
On April 22, the US Department of Health and Human Services (HHS), the US Centers for Medicare and Medicaid Services (CMS), and the Center for Medicare and Medicaid Innovation (CMMI) released the CMS Primary Cares Initiative. This initiative, serviced through CMMI, gives providers five payment-model options under two paths, Primary Care First (PCF) and Direct Contracting (DC).
PCF has two payment models
- Primary Care First
- Primary Care First (High Needs Populations)
The payment models under this path seek to reduce hospitalizations and determine if performance-based payments to primary-care providers will reduce Medicare expenses, promote care quality, and improve patient outcomes. Under the models, providers will receive monthly payments. The High Needs Populations model offers higher payments to practices that specialize in care for patients who have complex, chronic conditions. PCF is scheduled to begin in January 2020 and will be tested for five years.
DC has three payment modes
- Direct Contracting—Global
- Direct Contracting—Professional
- Direct Contracting—Geographic
The DC payment model options address primary care and seek to engage a wider variety of provider organizations, such as Accountable Care Organizations (ACOs), Medicare Advantage (MA) plans, and Medicaid managed care organizations. Participants in the Global model option will bear full financial risk, and Professional model participants will share risk with CMS.
(Source: HHS, HHS To Deliver Value-Based Transformation in Primary Care, April 22, 2019)
RELATED: According to CMS, there are 41 ACOs in the Next Generation ACO Model for 2019—a decline from 51 in 2018. In March 2018, seven ACOs left the model, which brought the total down from an initial 58. CMMI launched the Next Generation ACO model in 2016, and the model will end in 2020. According to the National Association of ACOs (NAACOs), some participants left the Next Generation model to join the Medicare Shared Savings Program (MSSP), which offers similar benefits.
Trust Fund could be depleted in seven years, Medicare Board of Trustees predicts
The Hospital Insurance (HI) Trust Fund is expected to deplete in seven years, according to an annual report from the Medicare Board of Trustees. The HI fund—which funds Medicare Part A hospital insurance—will continue paying full benefits until 2026, when it is expected to be depleted. The Supplementary Medical Insurance (SMI) Trust Fund—which funds Part B and Part D—is expected to be financed for the entire 75-year projection period. Several other projections emerged from the report, they include:
- Total Medicare costs are projected to increase to 5.9 percent of gross domestic product by 2038 and will likely reach 6.5 percent of GDP by 2093.
- SMI costs are projected to increase from 2.1 percent of GDP in 2018 to 3.7 percent of GDP by 2038.
Seema Verma, CMS administrator and secretary of the Medicare Board of Trustees, said the recent projections are “a dose of reality in reminding us that the program’s main trust fund for hospital services can only pay full benefits for seven more years.”
(Source: CMS, Medicare Trustees Report shows Hospital Insurance Trust Fund will deplete in 7 years, April 22, 2019)
(Source: CMS, CMS Administrator Seema Verma Statement on the 2019 Medicare Trustee's Report, April 22, 2019)
CMS encourages adoption of integrated-care models for dual-eligibles
In an April 24 letter to state Medicaid directors, CMS outlined several models that states can use to address the health needs of their dual-eligible populations (individuals who are enrolled in both Medicare and Medicaid).
- The first option would allow Medicaid programs to enter into joint contracts with CMS and with private payers to cover services at a capitated rate. In a prepared statement, CMS Administrator Seema Verma explained that some of the previous administrative burdens associated with operating a Medicare-Medicaid plan have been removed, which could make this option more attractive to states.
- A second model is a fee-for-service payment approach in which CMS and a state’s Medicaid office would enter into an agreement allowing the state to retain some of the shared savings for Medicare integrated care. Though some states have been testing such models, Verma noted this model could negatively impact state balanced-budget constraints and therefore might not be effective for all state programs.
Verma said CMS is also open to states developing their own integrated-care models to address dual-eligibles.
(Source: CMS, Three New Opportunities to Test Innovative Models of Integrated Care for Individuals Dually Eligible for Medicaid and Medicare, April 24, 2019)
CMS’s proposed rule could mean lower payments to urban hospitals, continued payments for CAR-T
CMS has proposed changing its payment model so that rural hospitals receive larger Medicare payments. But the change could translate to lower payments for urban facilities. On April 23, CMS released its proposed Inpatient Prospective Payment System (IPPS) and Long-Term Care Hospital (LTCH) Prospective Payment System (PPS) rule for fiscal year 2020. CMS proposes increasing the wage index—which sets the nation's hospital payments—to address disparities in urban and rural hospital payments. Specifically, the proposed rule seeks to change the formula for rural hospitals, which tend to have low reimbursement rates. Additional proposals include:
- Calling for hospitals with a wage index value below the 25th percentile to receive an increase of "half the difference between the otherwise applicable wage index value for that hospital and the 25th percentile wage index value across all hospitals." Further, the agency proposed decreasing the wage index for hospitals that are above the 75th percentile, which CMS says could help reduce Medicare spending. The agency projects that Medicare spending on inpatient hospital services will increase by $4.7 billion in 2020.
- Continuing add-on payments for two types of chimeric antigen receptor (CAR) T-cell therapies, which treat certain kinds of cancers and can cost as much as $1 million per patient. The American Hospital Association (AHA) has expressed support for this proposal.
- Launching a new add-on payment pathway for “breakthrough” devices, as classified by the US Food and Drug Administration (FDA). The breakthrough pathway seeks to expedite the development process for new drugs or devices that treat serious illnesses or address unmet needs and are considered better than existing treatments. However, CMS’s add-on policy might make it difficult to determine if a product is better than one already on the market. To address this, CMS proposes classifying products that were included in one of FDA’s expedited programs and received market authorization as being “new” and therefore dissimilar to existing products.
- Increasing uncompensated-care payments to hospitals by $216 million compared to 2019, to $8.5 billion in 2020. The agency also proposed changing certain quality-care programs, such as the Hospital Readmission Reduction Program (HRRP) and the Hospital Inpatient Quality Reporting Program (HIQRP).
- Adopting two new opioid-related clinical quality measures for the HIQRP, starting in 2021.
Last fall, HHS's Office of Inspector General said the wage index formula led to more than $140 million in overpayments between 2014 and 2017 and recommended an overhaul of the wage index system.
(Source: CMS, Fiscal Year (FY) 2020 Medicare Hospital Inpatient Prospective Payment System (IPPS) and Long Term Acute Care Hospital (LTCH) Prospective Payment System Proposed Rule and Request for Information, April 23, 2019)
Wireless sensors innovate care for premature babies
Researchers and clinical teams have made great strides in improving outcomes for premature babies. Multiple wires are typically used to help monitor babies and alert the care team when interventions are necessary. But, these wires and adhesive patches can damage fragile skin and prevent parents from holding and nurturing their babies.
A team from the Northwestern University Feinberg School of Medicine has been working for five years to develop a wireless sensor system that uses Bluetooth technology to monitor babies in the neonatal intensive care unit (NICU). The new system appears to be as precise and accurate as traditional monitors. Two lightweight wireless patches (they weigh about as much as a raindrop) are attached to the baby's chest and foot. The sensors collect a wide range of data, including temperature, respiratory rate, EKG, oxygen saturation, and blood pressure. The patches are flexible and gentle on a newborn's skin. The wireless device allows for more physical contact between baby and parents. Growing evidence shows the importance of skin-to-skin contact in improving outcomes for premature babies. These lightweight wireless sensors allow for that bonding to take place.
The wireless monitoring system has been tested on 90 babies to date and could begin appearing in US hospitals in the next few years. Already, about 20,000 sensors are being planned for medically underserved nations like Zambia, Pakistan, and India. Because the sensors can be linked by smartphone, they are less expensive than the traditional system.
Approximately 300,000 premature babies are delivered each year in the US. While care has improved over the last few decades, some preemies face developmental delays and could experience health problems throughout their lives. About 15 percent of babies admitted to a NICU will have some developmental delay or learning disability by the time they reach school. However, researchers are not able to tie these delays to any physical problems in the brain, so the consensus is that the delays are related to environment. It is possible that the loud noise, lights, and constant handling of the babies in the NICU creates a state of stress that affects development. These wireless sensors allow the care team to mimic the quietness of the womb, while allowing for beneficial contact between the babies and their parents. The researchers note that the sensor system could be sent home with a patient, so monitoring can continue beyond the hospital if necessary.
(Source: Amy Paller, John Rogers, Hany Aly, March 1, 2019, Science)