No one wants to leave money on the table. But that’s often what happens when health plans don’t recover payments for claims that are someone else’s financial responsibility. A health plan’s successful recovery of injury-related claims depends upon a fine-tuned and optimized subrogation process. In our three-part series, “Next-generation subrogation solutions,” we offer effective strategies for the identification, optimization, and measurement of the subrogation process to help your health plan contain costs and maximize results.
Identifying the right subrogation cases is the first step in recovering injury-related payments made by the health plan. Many of the currently utilized subrogation identification practices are outdated and unrefined, resulting in more or missed cases, but not the right cases to maximize results. There are several best practices that can help health plans accurately identify these cases, without wasting resources on unnecessary investigations. Implementation of these best practices, which leverage today’s technology, predictive data analytics and scoring, can truly optimize your subrogation results.
“If you don’t identify the cases, you can’t recover or cost-avoid,” Liz Longo, Discovery Health Partners General Counsel, said in a recent webinar. “It’s a delicate balance between identifying too many cases, which results in false positives or non-recoverable cases, and identifying too few cases, which results in missed opportunities.”
Fine-tuning your identification process can help you achieve optimum accuracy in pinpointing the right subrogation cases, resulting in a cost savings to your bottom line.
Best practice: scoring and predictive analytics
Current practice: subjective, static
Diagnosis code lists have long been the starting point for uncovering potential subrogation cases. However, these lists of codes tend to be subjective and not frequently reviewed. Scoring and predictive analytics can help you uncover which cases will drive your recoveries. Predictive analytics means that there is a continuing analysis of the codes used (diagnosis, procedure, revenue codes), in connection with varying demographics (age, location, presence of other medical conditions), compared against the data on recoveries achieved. When looking at all of this together, you can identify and constantly refine which combinations are more likely than not to result in a recovery. The goal is to learn from the data then use what you learn.
Looking at diagnosis codes in isolation from other claims information often misses the mark. Looking at diagnosis codes along with demographic information will identify the relationships that lead to recoveries and allow the plan to prioritize recovery efforts.
For example, fractured femur and fractured ankle and foot codes are typically included in the subjective and static identification lists commonly used. Fractured femurs are very common in older populations and more often than not are not related to any accident or injury for which there is a recovery source. Similarly, fractured ankles and foot bones are common among diabetics unrelated to any accident or injury. By utilizing scoring and predictive analytics, there will be a fine tuning and continuous sharpening of which fractured femur, foot and ankle codes, in combination with other codes and demographics, are likely to yield a recovery and which codes are not.
Plans want to make sure that those relationships and trends that result in recoveries are continuously identified and that subrogation identification is refined based on what has been learned to drive improved accuracy.
Best practice: deeper dive analysis
Current practice: limited review
It is critical that plans leverage all available claims data in the identification process and not just consider one or two diagnosis codes. A deep dive into all available claims data reveals important details that might be overlooked in a more limited review. Diagnosis codes must be reviewed along with other codes, including procedure codes and revenue codes. If you only look at the first code on a diagnosis claim, you will likely miss valuable identification information.
For example, “E” codes describe external causes of injury in the place where those occurred (i.e., a car accident on a highway). Because “E” codes are not revenue-generating codes, they usually never appear as the first code on a claim. You might see that a member has sustained a sprained neck by looking at the first diagnosis code on a claim. Without further review of the second or third code, you might overlook the code that indicates that the sprained neck was sustained as a result of a car accident.
The importance of analyzing all available claims data is further illustrated in the case of mass torts. Identification of mass torts is very different from traditional subrogation cases and requires a unique analysis and identification process.
With respect to the failed hip and knee devices (Stryker, Depuy, Biomet, etc.), diagnosis codes alone are not selective enough to identify those revisions or other medical conditions that may be caused by a failed or defective device. Procedure codes along with diagnosis and revenue codes are the only effective way, solely from a data mining perspective, to identify these opportunities, making a deeper dive analysis of all claims data a critical step.
Best practice: table driven and flexible
Current practice: hard-coded and not customizable
One size does not fit all for purposes of identification. Different populations and demographics require different identification. Therefore, it is important that your identification be table driven and flexible.
In current practices, identification is often hard-coded and not customizable. This rigidity results in the inability of the plan to adjust its identification criteria for a select group or population leading to over or under identification. Oftentimes, ASO groups may have identification needs different from other populations such as excluding or flagging the identification of medical malpractice cases. Without table driven identification, such select identification exclusion is not possible.
As another example, firefighter populations illustrate the need for flexible and table driven identification. Many states have enacted cancer, cardiac and lung presumption laws, meaning that these conditions are presumed to be work-related diseases. Accordingly, flexibility in identification is required to pinpoint these cancers, cardiac and lung conditions that we would actively seek to avoid identifying in other populations. Your identification needs to be flexible enough to accommodate the differing identification needs of varying populations.
Best practice: scoring and tracking
Current Practice: ignores chronic condition
Another critical component of an effective identification process is the ability to account for and track member chronic conditions. When a member or other source indicates to the plan or the plan’s vendor that the member has a chronic condition, e.g., a bad back, knee, shoulder, that condition for that member should be flagged. Going forward, then, that condition, alone, should not be reinvestigated in the future.
This chronic condition scoring is a win-win for everyone. It results in reduced member abrasion by minimizing unnecessary member outreach and saving costs associated with needless investigations. This identification and tracking of chronic conditions substantially reduces cost and waste for the plan.
Using these innovative subrogation best practices for identification, Discovery Health Partners has significantly improved the subrogation process to deliver better recoveries. The first step to these recoveries is accurate identification, but the process doesn’t stop there.
The next installment of our three part-series on “Next-generation subrogation solutions” will look at the next two steps in subrogation recovery process: optimization and measurement.