Do You Have a Healthcare Fraud Prevention Strategy Ready for COVID-19?

Artificial Intelligence allows health plans to more nimbly address FWA while reducing false positives

To make an obvious understatement, the new coronavirus (COVID-19) pandemic is an event like we’ve never seen. While essentially the whole country is shut down, the healthcare system is both slowing and accelerating at the same time. All of us in the FWA business know that 99.9% of providers are working tirelessly to treat affected patients, calm those with symptoms, and manage an unprecedented healthcare event. At the same time, both intentional and unintentional changes may spur a spike in healthcare fraud, waste and abuse.

I have recently seen personal examples of how changes to treatment and patient care during this time will change, with both my primary care and multiple specialists moving to Telehealth visits.

You have probably already seen that the Department of Health and Human Services (DHHS) and the Center for Medicaid and Medicare Services (CMS) have eased some of the coverage and benefit restrictions for many enrollees to allow for less administrative issues on the part of providers as well as quicker time to treatment. The Coronavirus Preparedness and Response Supplemental Appropriations Act, 2020 (§H.R.6074) was signed into law on March 6 along with authorized federal spending to help with the pandemic. There are several aspects to the bill but perhaps most pertinent to the SIU involves the ability for the DHHS secretary to temporarily waive certain Medicare requirements for Telehealth services.

Foremost among the Telehealth changes (Telehealth Services During Certain Emergency Periods Act of 2020) that initially affect Medicare beneficiaries by allowing more lenient guardrails around Telehealth include:

• Waiving originating site (such as hospital, physician office, etc.) and the geographic requirement
• To qualify for the waiver, the provider must have treated the patient within the past 3 years or have been in the same practice as a rendering provider
• Telehealth patients can use their Smartphones to receive the Telehealth visit

Medicaid changes and reimbursement will, obviously, be on a state-by-state basis, but several U.S. House Representatives are requesting their states relax Telehealth requirements similarly to CMS. We are already seeing some of our commercial payer customers changing Telehealth reimbursement (CPT 99201-99215; HCPCS G0425-G0427; G0406-G0408; G2012, G2010) to reflect a typical office visit (CPT 99210-99215). In addition, we should expect that not only will many commercial plans duplicate the CMS recommendations, but we should expect some copays may be waived as well.

If some estimates are accurate, as much as 80% of the visits in the coming weeks and months will be via Telehealth. Therefore, payers need to be prepared to identify not only potential fraud, waste and abuse, but also to recognize when the encounter is a false positive due to the pandemic.

Because our Pareo Fraud solution incorporates multiple layers and scoring modalities, there are 4 primary areas where AI can assist in quick and accurate recognition.

Look at more than a just query of CPT, ICD-10 codes

If you’re using only a rules-based system for detection, queries into codes alone will show a limited range of views into an encounter and, more importantly, a trend. Since the pandemic is, by its nature, time-based and characterized by specific symptoms, a multi-faceted view is important.

An example of this would be using Artificial Intelligence to dig deeper and wider to look at metrics such as specialty, dates of service, patient acuity, and past treatment history of the individual patient in coordination with each other. The model then compares like providers based on those metrics, as well as like patient history, past SIU history of the provider, and other key indicators. The model continues to learn and evolve as the system reviews more claims so what might not be a trend today, could be recognized as one tomorrow. This helps SIUs instantly recognize an issue and take immediate action.

Spikes in volume from week-week, month-month, especially with highly likely specialties

Monitoring spikes in claim volume based on dollars paid or number of unique patients is always important and common among most SIUs. But COVID-19 identification and treatment are moving so fast that it may be difficult to keep up with a spike. Therefore, your detection should be able to determine what was “normal” at what points in time, when did it become “abnormal” and for how long. This may seem like a retrospective approach, but AI enables you to simultaneously look at historical normal, recent and current spikes, and predict what the “new normal” should look like.

The AI models in Pareo Fraud combine the time frequency, frequency of codes billed during the time frame, volume and frequency of length of time that the provider has been seeing the affected patient population, any previous payment integrity audits on the codes, and others. As of March 13, 2020, the AMA created a new CPT code to help streamline Coronavirus testing, so the utilization of dates and billing frequency becomes even more important.

Scope of Practice and Specialties

Most states have requirements around Telehealth and who can provide these services. The medical boards of 49 states, plus those of the District of Columbia, Puerto Rico, and the Virgin Islands, require that physicians engaging in telemedicine be licensed in the state in which the patient is located. Twelve state boards issue a special purpose license, telemedicine license or certificate, or license to practice medicine across state lines to allow for the practice of telemedicine.

However, the symptoms of coronavirus may not be the virus itself but, instead, other illnesses such as flu, bronchitis, pneumonia or simply a common cold. This lack of clarity makes identifying actual COVID-19 cases and legitimate testing and treatment difficult to differentiate from a provider trying to take advantage of the system. Paying careful attention to the specialty of the provider, combined with their history and the patient’s history is important.

The AI models in Pareo Fraud can combine the following to help identify excessive coronavirus testing and treatment outside of the normal:

• Specialty. We should expect to see, with varying degrees of frequency, all primary care, Pediatrics, ER, Pulmonology, Allergy & Immunology, Infectious Disease
• Dates of service. When did billing for the common coronavirus CPT and ICD-10 codes start, peak and tail off?
•  Geography of the provider’s patient population
• Peer-to-peer analysis of provider vs. same specialty
• Place of Service to determine ER, hospital, office, Telehealth or elsewhere.

Again, these metrics are incorporated into a single model and, when combined with other AI models, give a fuller picture during a specific time frame (the pandemic).

What about false positives?

Dealing with false positives will always be a challenge when fighting healthcare fraud, waste and abuse. Due to the extreme nature of the coronavirus outbreak, severity of symptoms and overall lack of testing equipment, there is still the concern of people flooding the healthcare system when even just one of the many COVID-19 symptoms are present. This will create enormous volumes of claims that are at least somewhat related to the outbreak. So how does an investigator differentiate from a concerned patient and a diligent provider versus a potential suspect?

By using the AI models described above, much of the differentiation can be accomplished within the technology itself, before an investigator must make a judgment call. Because of the self-learning nature of the models, they can determine a one-off from a trend, and look at multiple dimensions of providers, members, claims, and overall encounters during a finite (hopefully!) period. This helps alleviate inconveniencing a provider about legitimate claims and focus on the those trying to take advantage of a once-in-a-lifetime situation.

Let’s hope that this is a temporary challenge but also be vigilant in helping those who need it while identifying those trying to profit off some very trying times.