SmartLight Analytics

Reducing healthcare costs for self-funded employers

Analytics

By

SmartLight Analytics named “Outstanding Service Innovation” winner in the D CEO 2020 Excellence in Healthcare awards

SmartLight Analytics is honored to be recognized by D CEO as a 2020 Excellence in Healthcare Award winner for our work to reduce healthcare costs for self-funded employers. The work to identify and eliminate fraud, waste and abuse in healthcare claims this year has become especially important as employers continue to provide coverage for employees during the pandemic.

For the team at SmartLight Analytics, however, the opportunity to make an impact in making healthcare more affordable is our true reward. We also want to thank Health Transformation Alliance, who nominated SmartLight for this award. Our partnership with HTA has broadened our ability to help self-funded employers across the country to eliminate wasteful and fraudulent spending in their healthcare plans. Moving into 2021, we are ready to partner with more companies to reduce healthcare costs without disruption to the quality of care and coverage for their employees. Making healthcare more affordable is our continued mission.

D CEO 2020 Excellence in Healthcare winners

By

In healthcare claims, does distance matter?

Can the distance between your medical provider and the lab used to process your test results tell you anything about potential fraud, waste, or abuse? In fact, it can.  

In reviewing healthcare claims, data specialists use geospatial analysis to compare distances traveled for lab services. This strategy is used in one of many inferential analysis models that is not searching for a specific medical code within claims, but uses statistics and probability to determine if an observed action differs from an expected action. This can lead to a great deal of information inside of healthcare claims.  

How? It begins when analysts calculate a mathematical score that evaluates the actual distance traveled for lab services compared to what is the expected distance. One claim reviewed by the SmartLight Analytics team expected a distance of seven miles for lab work, the actual distance was two miles from the provider. That claim was then compared to another claim where the expected distance for lab work was 10 miles, but the actual distance was 2,976 miles. This drastically different mileage might indicate a pass-through billing lab scheme and at the very least requires further investigation.  

When working with businesses across the country, data analysts have found instances where providers have sent lab work or even bill for services across thousands of miles and sometimes even across state lines unnecessarily. Geospatial analysis has also been used to find claims where services were billed hundreds of miles from where the member employee could have reasonably been assumed to be that day. For example, a claim was flagged because it contained billing for an office visit to a nurse practitioner on a specific day as well as billing for tests conducted at a rural hospital 350 miles away on the same day. To believe this occurred as billed, the member would have had to travel more than 350 miles from his or her home on the same day for these tests. What likely happened was that the member saw the provider near home and the provider sent the claims to the rural hospital for billing. Such claims would then be investigated further to determine whether or not it was fraud or just an unusual case or a very sick patient who needed a specialized service. 

Geospatial analysis is just one of 40 different proprietary models used by SmartLight analysts to find fraud, waste, and abuse within healthcare claims.  

By

How does looking at pharmacy claims combined with medical claims help find fraud?

When it comes to healthcare claims analysis, combining both medical and pharmacy claims data provides a more thorough picture of what is happening for specific members (or dependents) and adds an additional analytical layer to help find different types of fraud, waste or abuse that might exist.

A case study from SmartLight Analytics provides an example of how the combined data can uncover abuse. The case uncovered prior authorization abuse in pharmacy claims resulting in thousands of dollars of on-going fraud.

 SmartLight analysts noticed a set of pharmacy claims for one member with several red flags:  the member was receiving multiple controlled substances and using multiple pharmacies (at least three) to fill and repeatedly refill prescriptions for Oxycodone and Fentanyl. The Fentanyl prescription was for Subsys, which is a spray specifically approved for breakthrough pain in cancer patients. Dr. Franklin Baumann, SmartLight Analytics CMO and co-founder, called this a “very specific drug used in very specific cases.”

The analysis found that for over a year, every single month the member was receiving a prescription for Subsys at the cost of $10,907 per month. The employer’s plan was continuing to pay for this large claim. It was verified that the employer’s plan allowed this drug with a prior authorization only in cases where an employee has cancer. Once SmartLight’s statistical team noticed these pharmacy claims, they turned their research over to the clinical team (including board certified physicians) to compare these findings with the member’s medical data.

“We know there are very strong policies around this particular Fentanyl. It is only used for cancer and, specifically for cancer related breakthrough pain. In this case, it received prior authorization which means the prescribing physician indicated that in the records.”

Having looked at the bigger picture of this member’s medical and pharmacy claims data, there was no indication of any kind of cancer diagnosis.

The clinical team performed detailed historical research on all diagnosis and procedure codes submitted for this member and found “very clearly” that there was no indication of current cancer, current treatment for cancer and probably no prior treatment for cancer.

“We can say that with some confidence because of the way medical codes work.” Dr. Baumann explained. “Generally, when you have a serious cancer and residual effects of cancer, it’s going to get carried along in your medical diagnosis codes.”

SmartLight then brought its findings to the insurance carrier. “The way we bring this up with the carrier (who paid the claims) is that we don’t see it on our end (a cancer diagnosis). Based on everything we saw, it’s not there. When they looked at it, they agreed with us.”

Dr. Baumann said the repeated claims had fallen through the cracks after a prior authorization was issued initially. SmartLight partnered with the carrier to stop the payment on these claims and continues to monitor the case for the employer.

By

The Differentiator: Inferential Analytics

What is inferential analytics? This analytic method is used by the SmartLight team to find fraud, waste, and abuse in employee healthcare claims. Simply put, it does not rely only upon algorithms, which are a set of rules created to solve specific problems. Instead, it looks beyond the expected “rules” to find outlier patterns of behavior.

Inferential Analytics

Algorithms look for known issues and targets. For example, an algorithm might be set to look for a specific rule — all providers who bill two x-rays for every patient – which could possibly be an indicator of fraud or overbilling. However, that specific rule could be overlooking a more creative or different fraud scheme. In contrast, inferential analysis does not rely on rules but instead uses evidence, reasoning, and content knowledge to look for anomalies in the data. It models “normal behavior” such as the average spend per patient then compares all the data against that model.

When using algorithms as your primary tool, the downside is you are never ahead of the curve. As soon as a provider finds that claims are getting denied using one code, then the provider will switch those codes to something else. With inferential analytics, data analysts are not searching for a specific code but using statistics and probability to determine if an observed action differs from an expected action.

As an example, geospatial analysis used by SmartLight compares distance traveled for services such as lab processing. Analysts calculate a mathematical score that evaluates the actual distance traveled between a provider and a lab for routine lab services, compared to the expected distance.  One claim reviewed had an expected distance of seven miles for lab work, and the actual distance was two miles from the provider. The claim was then compared to another claim where the expected distance for lab work was 10 miles, but the actual distance was 2,976 miles. This drastically different mileage might indicate a “Pass-through Lab Scheme” and at the very least requires further investigation.

Inferential analytics allows employers to see beyond the surface of their employee healthcare claims and prevent wasteful spending.