3 ways cost models are a key to success under risk
As the year draws to a close, healthcare organizations will be examining their performance in various payment models, all of which are designed to usher physicians into some form of risk payment. Cost models can help healthcare organizations achieve success under risk in three key ways.
Until this year, physician groups and health systems had to choose between two options for participating in CMS’s value-based healthcare program for their attributed primary care patients. Those choices consisted of accountable care organization (ACO) participation or contracting with an insurer’s Medicare Advantage (MA) plan. Each incorporates different models for provider payment, with ACOs using a fee-for-service model followed by reconciliation to acost target with downside risk (and payback to Medicare) over that level. MA plans pay either on a fee-for-service or capitation basis.
Five primary care models introduced in May 2019 expanded the range of provider risk options for small and large physician groups. Under two Primary Care First (PCF) models, small practices can be rewarded financially for lowering hospital utilization, while Direct Contracting (DC) models provide capitation payments to cover all or some of health services for a population of patients.
Each model deploys a form of risk-based payment. CMS has been very clear that payment model reform is a primary objective of the transition to value-based care.
Cost model compatibility with risk-based payment models
Cost models are used to compare the results of payment with historical data and to guide future cost initiatives. These can be created for several purposes. First, cost models can compare actual or estimated costs (like staffing and other inputs into care) with revenues or targets of payment models. Insufficient data makes it difficult to package costs of care for patients because of unknown services outside the attributed provider, and partly because current provider systems are not designed to aggregate and attribute care costs. However, the promised access to CMS claims data can improve feasibility.
Second, cost models can compare estimated revenues under a fee-for-service and various payment models. This inquiry is a much easier task, and the results can inform providers about potential losses or gains under risk. It is still necessary to aggregate various sources of patient services and their revenues, but revenues are easier to obtain than costs. Again, claims data will improve the feasibility of this use of cost models.
Third, additional cost models can build from either of these methods, or claims, to examine costs or revenues by a variety of indicators, such as risk, diagnoses, procedures and distribution by categories of service. The point is to organize dollars such that total per patient per pear (PPPY) costs can serve as a current indicator to compare with targets of various payment models.
3 ways to create a path to success under risk
The comparison of PPPY costs or revenues with target or predicted revenues under payment models is critical. Constructing the cost models enables providers to create a path to achievement under risk in these three ways.
1. Decide on the best payment model for participation from a financial standpoint. Each payment model has a known history of costs associated with participating providers. ACOs have a claims history and a target for total expenditures. Direct contracting groups will receive similar claims history and capitation can be compared against historical payouts. PCF practices are likely to receive similar hospital utilization data.
Vendor technology is available to sort claims and cost data down to the patient and risk level, and to portray the costs against payment model expectations. This information gives physician leaders and administrators the confidence to set a direction and to create cost control priorities.
2. Create the foundation for examining variation in costs and establishing improvement initiatives. From PPPY totals, diagnosis and procedure data create averages and variations from the mean. Physicians are accustomed to looking at variations in care data, but cost is a new matter. Data is always questionable; there must be an open discovery process for physicians to participate in data review. Leaving that step out will damage credibility of cost models and value-based initiatives.
If carefully undertaken, physicians can guide health systems and practices in examining selective patient data and contributing solutions. They should participate voluntarily.
There will also be a need to address patient risk and its contribution to cost and outcomes. Recognition of social determinants of health (SDOH) is just emerging, but data collection is inconsistent, and programs to resolve gaps are still rare. Nevertheless, there is a need to address SDOH so that physicians can provide effective services.
3. Intervene and change the trajectory of individual patient’s cost. The most complex use of cost models is to change the cost trajectory for individual patients during their medical events. These cost models must rely on real-time patient data capture to provide alerts against predictive targets, especially for high-risk patients or high-cost events. Alerts should activate when the patient nears threshold, triggering medical, financial and social services reviews. Also needed are pre-determined review processes and agreed upon interventions.
Intervention cost models are akin to inpatient alert systems to affect utilization and inpatient scheduling of services, but they should be deployed across the continuum of care. We know that one of the main sources of high cost is patients whose medical situations have spun out of control from a variety of factors, who have been in multiple care settings and who have seen physicians in more than one specialty.
Iterative processes work to corral costs but must be sustained by data
Cost models can be a steering mechanism for successful achievement under financial risk. To make them effective, they must fuel processes that don’t weaponize cost results against physicians. Data is inherently imperfect, and actual costs will vary from predictions. Improvement of cost models result from an iterative process and will mature into reliability — resulting in more targeted solutions to control costs. But don’t mistake this for the final act, because the transition to provider risk is just beginning in earnest.