Patient Attribution Key Component to APMs, Value-Based Contracts

Patient attribution timing

Patient attribution methods are either prospective or retrospective, the Society of Actuaries explained.

Under prospective patient attribution methodologies, providers know the population they are at financial risk for prior to the start of a performance period. Payers provide healthcare organizations with a list of attributed members at the beginning of the performance period and no new members are added during the period.

With retrospective patient attribution methodologies, payers determine the at-risk population after the performance period ends. Retrospective patient attribution is the most commonly used method in APMs and value-based contracts because payers have the claims data needed to assign patients to specific providers.

However, retrospective models present challenges to providers. The methodology is “a rear-view window approach with a lag time on demonstrating through claims data how well they managed their attributed patient panel in accordance with the risk contract terms,” the research paper stated.

Providers participating in APMs and contracts with retrospective patient attribution should run their own attribution calculation at the start of a performance period to help determine who is at financial risk under the model and estimate the costs associated with that population, the actuaries advised.

Then, providers should run an attribution calculation every quarter to update assignments.

Patient choice, geography, or visit-based attribution

Layered on the timing of the patient attribution model is the method for assigning patients. Payers and other stakeholders attribute patients to providers under APMs and value-based contracts through patient choice, geography, or visits.

The simplest method is patient choice. Under the patient choice method, patients decide which provider they want to be held accountable for the quality and cost of their care. The method is ideal if patients frequently see a particular provider.

But the patient choice method can create difficulties for providers. Enforcement is a challenge for low-cost patients who have low utilization rates, which can lead to a low capture rate for the APM.

As a result, patient choice is rarely used for patient attribution.

Geography-based attribution is more commonly used because the method assigns patients to a provider using the assignment of a network or use of zip code or residence county data. The method leads to a high capture rate.

However, geography-based attribution lacks the sensitivity for showing the care utilization across the population, the actuaries pointed out.

Payers and stakeholders also assign patients under an APM or value-based contract using visit-based attribution.

“Visit-based attribution is an algorithm-based approach that uses claims experience, which has been the more universally trusted source for the demonstration of performance in quality and efficiency,” the research paper explained. “A variant of visit-based methodology is episode-based attribution or bundling, which has a clearer alignment when a discrete number and level of services are delivered over a defined period of days and in return for a single cost.”

The method typically uses a hierarchy of criteria to assign patients. For example, payers may first attribute patients to a primary care provider seen during a recent period for specific codes. If there is no primary care provider seen for those codes, then the payer may then assign patients to a medical subspecialist.

The visit-based method typically attributes about three-quarters of patients in a specific population, the Society of Actuaries reported.

However, the method is administratively complex, especially with the recent healthcare merger and acquisition trend under which providers frequently change identification numbers. The method is also highly dependent on the quality of data collected at the provider level.

Evaluating patient attribution models

Not all patient attribution models are created the same. With different types of methods used for assigning patients to a provider under an alternative payment model or value-based contract, providers should be assessing which models are the most appropriate for their practice.

Providers should work with actuaries and consultants to evaluate patient attribution models during contract negotiations, the research paper stated.

Providers, consultants, and actuaries should answer the following questions to truly evaluate a patient attribution method:

• What is the best estimate of future performance of an attributed panel of patients in the arrangement?
• What methodologies can be used to understand and act on patient attribution accountability and churn?
• What are the chances of savings based on the attributed panel?
• What is the downside financial risk in the arrangement based on the attributed panel and what is the risk of loss for the population?
• How can the payment model arrangement impact the range of attribution possibilities?

As alternative payment models and value-based contracts increase in popularity, the models are likely to mature and evolve. Providers should be doing their due diligence and working with experts to evaluate patient attribution models as the value-based arrangements grow in complex, the actuaries stressed.

Key considerations for patient attribution models

In addition to understanding the type of patient attribution model used, providers should also consider how the assignment methodology accounts for medically complex patients and unassignable patients, the actuaries advised.

Medically complex patients tend to run up healthcare costs. The most expensive one to two percent of patients account for about 30 percent of total healthcare costs, and the top five percent represent about one-half of costs.

Patient attribution models should exclude the outliers, the research paper stated.

“Although identification and exclusion of complex patients is intended to mitigate their impact on the variability of patient costs in a population, small variations in the prevalence of other high-cost patients, especially if a panel of patients is small, can cause marked changes in total experience,” the actuaries wrote.

Patient attribution models should also address unassignable patients. A portion of patients will not receive services during the attribution period or the services delivered are not part of the assignment methodology. Low-cost patients who rarely seek care are also difficult to attribute.

However, providers should know how a patient attribution model handles unassignable patients, so they can estimate costs and services for their assigned at-risk population.

“Attribution methods may deal with these members by alternate means, such as by auto-assigning them to providers, or they may be excluded from the value-based contract,” the actuaries warned providers. “It is important to understand the size of this population and whether they are included in development of base measures.”

As alternative payment models and value-based contracts mature, payers are developing accurate patient attribution methods to ensure providers are appropriately at risk for the care quality and costs of patients.

Current methods may challenge payers and providers, but healthcare leaders are working to improve patient attribution.

“More attribution methods are not necessarily better, but targeted ways for plans to attribute patients to their most appropriate provider groups for performance measurement and fiscal risk accountability will continue to evolve,” the paper stated.