Prioritizing Risk Capture Compliance with NLP

Andrea Serra discusses risk capture complianceIn the past few weeks, several new stories have broken around risk capture compliance issues, particularly when it comes to Medicare Advantage. Individual DOJ lawsuits around MA risk capture and government agencies like the OIG weighing in on MA policy itself have led to a barnstorm of op-eds.

The crux of these stories, even the op-eds, zero in on unethical behavior within Medicare Advantage risk capture, sometimes assuming it is a systemic issue rather than a participant behavior issue. Similarly, scrutiny has been called for technology partners using artificial intelligence to “data-mine” for risk-adjustable codes, particularly those that charge based on additive codes.

In light of all of that, it feels like an excellent time to explain exactly how Health Fidelity supports its partners’ risk capture efforts, as well as share what we consider integral best practices to realize the full utility of our solutions.

What Does Lumanent Actually Do?

As a refresh, Lumanent represents the suite of technology and solutions provided by Health Fidelity. At a core level, it is powered by Lumanent Insights, an NLP and clinical inference engine. The NLP engine, designed explicitly for clinical use at Columbia University by NLP pioneer Dr. Carol Friedman, scans and processes all available clinical data and “translates” it from natural, narrative text into something more readily understood by traditional technology. The second layer, the inference engine, can then connect the dots between verifiable clinical evidence to present one of two data outputs:

  • Suggested Codes (suggestions)
  • Suspected Conditions (suspects)

Lumanent Retrospective Review and Lumanent Post-Encounter Review present suggestions to propose a more accurate code for submission or correction. This means all of our suggestions are connected to factual data within the patient record. In fact, in most cases, the supporting data is directly presented to coders for review.

Beyond the direct evidence link, there are additional mechanisms within Lumanent to support risk capture compliance. Coders are also presented with redaction opportunities. This means that if a code already submitted lacks sufficient clinical evidence, the system will surface those codes for review and potential deletion from the submission. This is done explicitly to reduce overpayments and help our partners remain compliant.

Lumanent Pre-Encounter Prep and Lumanent Suspects each present suspects for clinical validation and review. In these cases, highly likely but unconfirmed risk adjustable conditions are proposed to the care team for review prior to each encounter. In this way, any potential net addition to risk scores are validated within the next patient visit. In this regard, these tools support care through the lens of identifying risk adjustable conditions by surfacing patient needs just as much as they are technology built for RAF capture itself. This is especially true in light of how the more accurate documentation by clinicians aligns with the spirit of value-based care to begin with, more closely assessing and monitoring chronic conditions.

Medicare Advantage and Risk Capture Compliance

Risk adjustment strategy and best practices should not only include adding or escalating codes to submissions. Full stop. To lag in redaction of inaccurate codes or seeking to stymy any accidental upcoding is equally important. Risk capture compliance is intrinsic to risk capture. The goal is not, “find more” the goal is, “complete and accurate.” A complete submission includes proper documentation of all present conditions, to the exclusion of those that aren’t. An accurate submission means the code acuity matches the exact physician wording of present conditions, to the exclusion of those that aren’t.

One of the primary problems outlined independently by the OIG and the DOJ involved in-home assessments being used to allegedly inflate risk scores. The expansion of telehealth during the ongoing COVID-19 pandemic has complicated this, as the rules have relaxed to allow the documentation of time spent assessing patients and developing care plans, but there are still necessary diagnostic procedures that are required to fully capture many of the most common chronic conditions. Ask yourself and your risk adjustment partner: is your solution flagging encounters that lack necessary diagnostic evidence alongside time-based coding?

Chart review as called out as unethical or inappropriate is poorly defined and more complex to address. Frankly, the nature of documentation and risk capture for health plans and providers under value-based care means that the vast majority of risk adjustment work is going to be additive. The odds of a busy professional forgetting to check a box or document something when finalizing a patient visit is far more likely than unintentionally increasing the severity of a condition or diagnosing a disease that isn’t there at all. Still, a backstop needs to be there just in case. Much like the above example, it’s important to ask yourself and your risk adjustment partner: is there a redaction report available to you to help catch these potential inaccurate submissions?

When a technology partner presents themselves as experts providing you with your risk adjustment solution, that should mean they have the capability to both add and remove codes. From a strategic perspective, we view redaction not just as good practice for compliance, but also very pragmatic: when compliance workflows are utilized, care funding estimates are less likely to be inflated, which allows for better forecasting and, obviously, reduces the likelihood of lawsuits through any sort of errors.

If your organization would like more support in optimizing a compliant risk capture strategy, don’t hesitate to contact us today.