Prompt, full payments are essential for healthcare providers, who must constantly balance the need to provide quality care with the need to stay financially viable. Meanwhile, ensuring value-based care means considering a wider array of factors than past models have required.
Accurately predicting who will pay how much by what time is a critical ability for healthcare providers. Yet it can be increasingly difficult to calculate the probability of being paid without access to the right tools and data sets. As a result, many key healthcare conversations today center on propensity to pay.
Understanding Propensity to Pay
Propensity to pay can also be framed as a question: "How likely is a patient to pay his or her bill, and what is the probability that a provider will be reimbursed by a payer for the invoiced amount?"
Determining propensity to pay for any one payer can be difficult. One major hurdle, says CMS administrator Seema Verma, is the fact that "clinicians and hospitals have to report an array of measures to different payers." When multiple payers are involved, determining the likelihood that the various sources will contribute amounts that reimburse the total cost of service becomes even more complex -- especially when all these sources operate on their own internal schedules.
The question can be complicated by factors like capitated payments, or pre-arranged payments to provide services on a "per member, per month" basis, says Patrick C. Alguire, MD, director of education and career development at the American College of Physicians.
At the same time, practices must balance other factors that affect their bottom lines, like scheduling and utilization. When a patient misses an appointment, for instance, the provider misses the opportunity to provide and be paid for care. Inefficient scheduling can also lead to uneven utilization, overworking staff and facilities at certain times while leaving them nearly idle at others.
Addressing Financial Risks Beyond the Patient
Mergers and acquisitions, a common occurrence in healthcare, also come with a slate of financial risks that include risks surrounding payer behavior. "When two systems join together, replacing the contracts the acquirer has with the payer is not a quick, snap-of-the-finger type move," says Healthcare Finance associate editor Jeff Lagasse. When payers are insurance companies, the shift can create situations in which individual patients must be evaluated for their propensity to pay.
While mergers and acquisitions mean that organizations keep getting larger, it is nonetheless a mistake to think that predictive modeling for payment propensity only works for large-scale organizations, says Paul Bradley, chief data scientist at Waystar. "The reality is that it's just as effective for small to mid-size providers, perhaps more so since they have less margin for error in their financials."
To combat risk in the financial realm, some banks are turning to predictive analytics to help predict lifetime customer value, Raghav Bharadwaj writes at TechEmergence. Using artificial intelligence and machine learning algorithms, financial institutions are predicting customers' financial behavior, creating profiles for "ideal" customers and adapting operations to nurture customers with higher lifetime values.
A similar approach could be effective for healthcare providers, Welltok CEO Jeff Margolis says. "While the healthcare industry becomes increasingly adept at applying clinical and claims data to improve care, it has largely ignored other data sources that provide the greatest opportunity to positively impact health and cost at scale," he writes at Harvard Business Review.
This includes consumer data, which also plays a key role in determining the "when, how, and how much" of payments.
Which Payment Predictors Are Most Accurate?
One of the best ways to determine when, how, and how much a payer is likely to pay is by analyzing payments made in the past, say Eric Nilsson, chief technology officer at SSI Group. By using machine learning algorithms, providers can combine data from a number of sources to see how payers have responded to similar claims in the past, and thus to guess how the payer will likely respond in the future.
The confidence levels in these algorithms "can be 90 percent or greater in many cases," Nilsson says. "Being able to answer when your remit is to be paid can be particularly helpful when trying to forecast end of month revenue for your organization."
The medical conditions patients face also play a role in payment likelihood -- particularly as many insurance policies lapse in the face of more complex, difficult, or costly conditions, EconoSTATS' Wayne Winegarden notes at Forbes.
To understand how patients' conditions affect their propensity to pay, creating the right data set is crucial. A study published in the Journal of Hospital Medicine found that certain events during a patient's hospital stay highly correlate with their likelihood of being readmitted within 30 days. These included both complications that carried extra costs, like C. difficile infections, and those that did not, like vital sign instability upon discharge. Since readmission triggers additional costs, it factors into the propensity to pay equation and is worth spotting.
Spotting trends like these isn't always simple. While access to data is essential for technological solutions to provide accurate predictions, the shift to electronic health records is not yet complete. A 2018 study published in the Journal of Medical Internet Research found that most US hospitals, for instance, exist on a continuum between all-paper and all-electronic records. The researchers predicted that most hospitals will not function in an electronic-only environment until 2035.
To create more accurate tools for determining propensity to pay, it is important to ensure that the algorithms and models used by the system focus on the right data, says John Steensen at VISA. Examining broad trends, like the recent push away from payments for emergency care and hospital-based imaging to urgent care and stand-alone imaging centers, indicate broad behaviors but will have varying effects on payments at the individual provider level, notes Les Masterson at HealthcareDive.
When used correctly, these tools can analyze large quantities of real data to produce better predictions than human interpretation can provide.
Tools for Managing Risk and Payments
Patients are becoming more educated about healthcare choices and pricing, says Kimberly Zeltsar, executive director of revenue cycle for Kaiser Permanente, Hawaii Region. As a result, providers are improving their patient management systems with enhanced, automated services for patients and better data tracking for internal use. These tools empower patients to better understand and participate in the payment cycle.
Technology is helping healthcare providers manage payment-related risks, as well. In an article for HealthITAnalytics, Jennifer Bresnick advocates for the use of blockchain-based solutions for tracking payer data, allowing for a wide range of payment predictors to be weighted and analyzed in determining a patient's ability to pay. Smart contracts, in particular, could make it easier for providers to manage payments, Bresnick says.
A 2018 study in Computers and Society predicted that blockchain-based solutions for healthcare that address questions like propensity to pay would be more generalizable, and thus more effective, if they relied on a number of data points gathered across a variety of organizations.
PokitDok's Payment Risk solution enables lending institutions, payment solutions, health systems, and medical practitioners to calculate the financial risk of healthcare transactions before an episode of care, and to make new financing options available to patients for non-acute medical services. Our Payment Risk algorithm combines public, customer, and proprietary data to estimate the likelihood that a patient will pay their bill, and to calculate the probability that a provider will be reimbursed by a payer for a specific amount.
Images by: Christian Delbert/©123RF.com, Worrawit Saengrueang/©123RF.com, Marco Herrndorff/©123RF.com
Tags: Healthcare predictions