Estimating episodic target prices for each patient in the Oncology Care Model (OCM) can be challenging and time consuming. Applying that time to quality-focused care management tactics, based on observed utilization and patient outcomes, may wind up being more valuable, and help to reduce unnecessary spending.
This article has been co-authored by Alyssa Dahl, principal healthcare informatics analyst, DataGen.
The Oncology Care Model (OCM) is a 5-year alternative payment model (APM) for oncology practices and independent practitioners that began on July 1, 2016. Participants, whose episode cycles run for 6 months, received their fourth performance data feed from CMS in December 2017.
OCM is very different from traditional bundled payment models like Bundled Payments for Care Improvement and the Comprehensive Care for Joint Replacement, because CMS views oncology care as life-encompassing. Because providers are managing not just the cancer, but a complex set of factors affecting the patient, each patient essentially has their own unique target. This is based on their personal risk factors and specific events that occurred within the episode time period.
In an ideal world, OCM participants would know these targets early in the patient’s care journey, so they can refine their OCM strategy. However, knowing a patient’s episode target price with any kind of acceptable margin of error is impossible.
The desire to estimate an episode’s target price is understandable—how else can one plan budgets and forecast spending? However, there are so many unknowns and variables in healthcare that those estimates can result in wildly inaccurate numbers. Here are 3 reasons why you shouldn’t waste time and effort estimating individual episode target prices:
1. Performance period episodes need to be approximated. Because oncology patients are likely to receive several opinions upon diagnosis, it’s likely that they’ll receive care at multiple practices throughout their treatment. This makes it unclear to identify which practice was ultimately most responsible for the patient’s care and should “own” the episode. CMS’ solution to this challenge is an attribution method based on visit frequency—something that can’t be immediately determined at the forefront of an episode.
Attribution is unknown until the reconciliation process begins about a year after the performance period ends. This affects any totals, since there could be overlap between practices in the same system and the calibration of when the episode actually started. Practices will also have to estimate beneficiary eligibility, as some patients are likely to drop out due to changes in Medicare coverage, or may not initiate care until a later date. Another thing to consider is that actual expenditures may be underestimates, so there’s a lack of completeness in knowing what was actually spent on each patient when making a target price comparison.
Episode construction has to happen no matter what before you can make any sort of cost comparison to a target price. However, OCM participants would need to make some big assumptions even before they have an idea of when to define their episodes while using the claims data in CMS’ feedback report.
2. Covariates—even some of the ones that are known—may change, and others can only be abstracted. While some covariates—such as patient age, sex, hospital referral region, and dual eligibility status are known at the time a potential episode is initiated, others, such as episode length, could change based on the date identified to trigger the episode. Even the patient’s cancer type could change, because it’s based on coding plurality rules during the 6-month episode period.
When it comes to estimable covariates that can be abstracted from the claims data, practices can’t yet know all of the factors needed at the start of an episode to complete the regression model calculation to generate the target price. These covariates require the presence of complete claims information during an episode to be determined with certainty, and include:
Other covariates are complete unknowns in the performance period data. Without adequate patient history and gaps in the feedback report claims data, it’s impossible to know when the patient last had chemotherapy prior to episode initiation, what their hierarchical condition categories are, or their institutionalization status (which cannot be determined from claims data). Some of these factors significantly influence the magnitude of an episode’s target price.
3. There are other unknowns. If all that isn’t enough, there are additional unknowns that impact the ability to estimate target prices:
It’s understandable that OCM participants will want to try to look at target pricing, but OCM is such a complex bundled payment model that any estimate made today might cause practices to make inaccurate assumptions on financial targets anyway.
It is, however, possible to simulate an episode with the information currently at hand—knowing that it will definitely be inexact and that the magnitude of inaccuracy will vary across episodes. Another track to take is to wait until attribution happens for the period to try to calculate a target price. The result will have more precision in regard to the timing of the episode and claims run out, but it will be less time sensitive because you can’t do this until about a year after the performance period ends. At that point, CMS will be providing the real target price about 2 months later, so the exercise is likely not worth the time and effort spent.
There are so many factors that impact care for cancer patients that taking the time and energy to estimate flawed target prices simply isn’t worth the effort. Rather, applying that time to quality-focused care management tactics based on observed utilization and patient outcomes may wind up being more valuable, and help to reduce unnecessary spending. To learn more about the OCM program and other bundled payment initiatives, visit datagen.info.