Dr Simon Peck
I have spent most of my career tracking down Fraud, Waste and Abuse in Health Insurance. Some cases have been simple, others more complex. Schemes range from simple double-billing to organised patient sharing rings of providers (provider is the industry term for doctors/hospitals) that churn patients amongst themselves or even perform unnecessary operations against the best interest of the patient. Wasteful use of resources for financial gain is not uncommon and substantial savings can be made from tackling such waste.
For two decades I ran a fraud investigation team that was considered to be an industry leader. The team used in-house exception reports and analyses to find anomalies. Quite often, when showing our findings to the company (who ends up footing the bill for treatment), I was met with the response, “That’s so obvious – how did we miss that for so long?”
If I had a pound every time I heard that, I would be wealthy man.
If fraud is indeed obvious, why do fraudsters get away with it? The simple answer is that traditional labour-intensive methods of claim integrity management have failed to keep up with increasing data volumes. Furthermore, simple one-claim-at-a-time based reviews look at the claim in question not the context and history of the provider or customer and they often fail to spot increasingly sophisticated fraud patterns as fraudsters find new ways to exploit the control systems.
To help understand the problem better, let’s look at two key functions in the claim management process – Claim Assessors and Special Investigators.
Claim Assessors
After an insurance claim has been made, it is received by a Claim Assessor. The assessor decides whether the insurance company should pay the claim or not, with each decision typically taking no more than 5 minutes.
Experienced claims assessors are irreplaceable in the fight against fraud. I once found a string of fake identities courtesy of a claims assessor who noticed that a customer spelled his wife’s name incorrectly. On that occasion, the assessor realised something was amiss and sent the claim for investigation. However, as fraudulent claims are inherently deceptive, most sail through normal eligibility and reimbursement checks.
Even when found, a fraudulent claim on its own can appear perfectly normal. If someone bills for a complex treatment with complications, that’s nothing to be concerned about per se. However, if a doctor bills all treatments at the highest complexity and all their patients have complications, that needs to be looked at. The doctor could of course be an expert who takes the most difficult cases. However, she could also be exaggerating the complexity of the claims.
An assessor rarely has the time for deep analysis of a claim. Nor do they have the information to place the claim in its proper context (other claims submitted by the same Doctor, the medical history of the patient etc). The sheer volume of data is also a challenge: bad transactions are hidden amongst millions of others. For a human being, it can be difficult to know where to look.
Special Investigators
When a Claim Assessor thinks a claim needs more thorough analysis, they can escalate the claim to the Special Investigation Unit (SIU). The SIU investigation is detailed and thorough, which looks at not just the claim on its own but also its context. This process is however relatively costly and time-consuming. Hence, escalation and recovery often only make sense when the claims involve large sums of money. For most low-value claims, it is not cost effective to investigate after payment has been made by the insurers. This is a fact that fraudsters know well, and losses from high-volume, low-value fraud can be significant.
I recall a high-volume, low-value fraud case I investigated back in 2008. It was well known in the industry that certain procedures are prone to exaggeration. One common example is gastroscopy. In 2008, there used to be only two standard codes for gastroscopy – one simple (diagnostic gastroscopy) and a second more complicated (therapeutic gastroscopy). At the time, therapeutic gastroscopy was around 25% more expensive than diagnostic. Hence doctors had a financial incentive to bill for the more expensive treatment whenever they could.
Several insurers agreed to share data on the ratio of diagnostic to therapeutic procedures by practice. When this was done, quite a few practices had always billed therapeutic gastroscopy but never for diagnostic. These practices were written to, asked to check their billing was correct and reminded of the correct way to bill. Astonishingly, many of the practices wrote in afterwards, admitted they had misbilled and offered repayments.
Investigations were conducted on a small number of practices which did not change their billing and in two cases significant misbilling was uncovered. These cases were referred to the regulator and two doctors were suspended from the medical register. The first doctor also resigned his role when he was found to have billed for £85,000 dishonestly.
A press release was issued in the first case following the verdict (ref [1] at the foot of the article). A second doctor challenged the verdict in the High Court. The appeal was lost. For those interested, the full ruling can be obtained by following the link in reference [2].
This was an important investigation as it showed that misbilling for financial gain may have serious consequences on a doctor’s career. It also shows the power of simple outlier analysis in a fraud strategy. However, it is also clear that back in 2008, performing outlier analysis could only be undertaken on a case-by-case basis due to the high cost of collecting and analysing the data. Today, on the other hand, we have software platforms that aggregate data and automate analysis. With the right tools, specialist analysis like this can now be routinely performed.
The Need for Automation
We now know a bit more about Claim Assessors and Fraud Investigators. Each of them performs an essential role in managing the integrity of claims. However, they are challenged by the volume of claims and the need to spot complex patterns in large datasets sometimes spanning multiple years of data.
As human beings, we excel at spotting visual patterns. Once we know what to look for, our brains can combine patterns and information very efficiently. However, even the best claims assessors cannot be vigilant all the time. The time pressure and the need to make quick decisions on limited information means that errors and omissions happen often, leaving weaknesses for fraudsters to exploit.
This is where effective anti-fraud software comes in. A good anti-fraud software solution must do the following and more:
- Empower claims management personnel to do their work, not try and replace them
- Enable claim assessors to quickly identify claims that need looking at
- Extract the most high-value cases for investigators
- Spot behavioural patterns that are spread across a large number of claims
- Minimise the number of false positives (nobody likes a wild goose chase)
- Fit within the complex workflows of health insurers
- Speak the language of its users
- Enable information-sharing across the business
With the right software, both assessors and investigators can focus their energy where it matters. Quite often, all the software needs to do is to show the professional where to look. As per the headline, fraud is obvious when you find it. However, it’s exactly the finding part that us human beings can use a bit of help with.
Let’s look at some examples of real-life uses where software has helped find insurance fraud:
Example 1 – Drug Diversion
In this case (outside of the UK), an individual claimed for several prescriptions in a series of small claims. Zooming out on the case, we found that all the prescriptions were for the same drug. The prescriptions were simultaneously obtained from multiple doctors. Digging in deeper, the claim was a controlled drug and as it turned out, the drugs obtained were sold on in the black market. This is classic drug diversion. It often flies completely under the radar in many jurisdictions, mostly because each individual transaction is too small to be looked at on its own.
Example 2 – Limit Surfing
Having limited information and vast data volumes poses a challenge for claim assessors today. As the industry moves towards digital billing and automatic adjudication, professional fraudsters adapt and learn to hide their activities. They work out what triggers alerts or controls and learn to bill “correctly”. In other words, they are gaming the system to find out how they can break it.
As an example, consider the following: for low-value claims, many insurance companies set a minimum amount threshold per claim. Below this threshold, it is not economical to investigate the claim. Hence any claim below the threshold is automatically paid. Once the threshold is set, fraudsters will quickly work it out. I have often shown people how transactions cluster just below the scrutiny threshold – we call this limit surfing.
Using outlier analysis and peer-to-peer comparisons, Kirontech has been able to detect multiple instances of limit surfing. Once detected, the insurer was able to prevent inappropriate billings. These high-volume, low-value issues can lead to substantial losses and can be dealt with if they can be spotted prior to payment. The key is to spot them at the pre-payment stage since each claim on its own is too small to be economically recoverable.
About Kirontech Health Insurance Platform
Kirontech Health Insurance Platform (HIP) is custom built to tackle Fraud, Waste and Abuse in Medical Insurance. Unlike traditional AI/ML software providers, we combine our software solution with an expert team of medical and fraud experts. Our in-house experts validate the software solutions, and also work hand in hand with our customers to help them retain a competitive edge in an industry that is rapidly adopting new technologies.
Our system is adopted and being adopted by industry leaders in the UK and outside, and we have a proven track record of delivering value and helping identify complex cases. In addition to a library of rule-based checks, Kirontech HIP can automatically flag claims that fit existing, wider fraud patterns. This allows claim assessors to instantly narrow down on the most relevant claims and makes sure no time is wasted.
Kirontech HIP can run thousands of checks simultaneously, placing each claim in its context. Advanced pattern-recognition techniques minimise false positives.
For Claim Assessors, Kirontech HIP easily integrates with existing claim systems and helps make quick and correct choices. The integration can either happen at pre-payment stage (providing real-time alerts to aid decision-making) or post-payment (as evidence for investigation and recovery).
For Special Investigators, Kirontech HIP has an easy-to-use UI with information combined in a single package. After cases are automatically detected, they show up the investigator’s Case Inbox, enabling easy visualisation and data consolidation.
If you would like to find out how Kirontech can help you and your insurance business tackle Fraud, Waste and Abuse, please get in touch with us on our contact us page.
Case references
[2] Sethna Saverymuttu vs General Medical Council Case reference [2011] EWHC 1139 (Admin)
Dr Simon Peck, Chief medical Advisor, Kirontech UK Ltd