Impossible or Improbable? Claim Analysis in Wider Context

Dr Simon Peck

One of the primary uses of Kirontech software is to identify unusual patterns of behaviour in healthcare claims. When looked at individually, most claims look perfectly normal. It is only when we analyse at large groups of claims over time that we begin to see patterns of behaviour. By using traditional rule-based methods that often focus on single invoice at a snapshot in time, insurance payers often miss these patterns entirely.

An effective program to deal with Fraud, Waste and Abuse (FWA) looks not only at individual claims but also at the underlying history of the providers, practitioners, and of course, the patient's history.

Impossible: The Case of the Missing Cervix

One of my first fraud cases involved a gynaecologist who billed patients for colposcopy (cervical examination) after claiming for radical Wertheim's hysterectomy for cancer. Wertheim’s hysterectomy on its own is a perfectly reasonable medication procedure. Colposcopy is standard procedure and medical professionals recommend that a cervical screening is carried out regularly.

If a claims processor received an invoice for Wertheim’s hysterectomy one day and an invoice for colposcopy the day after, he might conclude that everything is in order. However, there is a catch: Wertheim’s hysterectomy includes the removal of the cervix. Hence, there is no cervix to examine, and the two operations are incompatible. In analytical terms, any colposcopies must be carried out before Wertheim’s hysterectomy, or otherwise the treatment path is invalid.

In the case of the gynaecologist above, no one spotted this impossible combination for many years. Eventually the pattern was discovered by an astute claims investigator The rest is history – the doctor concerned repaid a sum of money and he was erased from the medical register having been found guilty of fraud.

The cervical case is an example of an impossible patient pathway. Once spotted the pattern is clear and difficult to argue against.

In contrast with the clear-cut example above, many other claim patterns are possible but implausible. One the first big cases I worked on was a surgeon who claimed for complications and additional services for nearly every patient he saw. Soon after the surgeon's claims began to arrive, it was fairly obvious that the sheer volume of additional complications was unlikely. Once this pattern was identified by a member of our team, a review of medical notes was carried out. The review showed that every single bill analysed included services not documented in the underlying records. The eventual investigation was a major one, involving a full review of over 300 sets of patient notes.

The doctor in question was very well-known and had some high profile supporters who lobbied vocally on his behalf. This pattern is by no means unusual. In the case of the surgeon, he was medically competent and his surgeries were mostly very successful. Hence his patients would often feel a degree of gratitude, and were willing to go public with their expressions of gratitude and support. However, the issue in this case was not with the quality of care as such but the fraudulently inflated medical claims consistently submitted by the surgeon. In addition, his patient did not cover the bill of treatment (the insurance company did). However, insurance is not a victimless crime - any amount of funds / resources misallocated leads to higher costs of medical treatment whether in terms of premiums (private healthcare) or taxes (public healthcare).

In this particular case, all the fraudulently obtained monies were eventually repaid the money and the surgeon was charged with 19 counts of fraud. On the grouds of his health, he was found unfit to stand trial (1), and following his acquittal he left the UK. After leaving the UK, he worked overseas but ended up being investigated again for similar behaviour after only a few months.

In both of the above cases it was possible to find objective evidence of misbilling. There are cases where finding any objective evidence may not be possible. Consider for example a clinic which routinely charges for 30 different blood tests in every patient. When challenged, the clinic will already have performed the tests. Individually, it will be very difficult to challenge the tests - for any single individual one can find at least a semi-plausible medical reason for carrying out blood tests. However, common sense tells us that if every patient (or a majority of patients) routinely receives 30 blood tests, then something is well outta whack.

When confronted with the "Definitely Maybe" case, many insurance companies feel a little lost, and in fact some feel they cannot manage such matters at all. In traditional, invoice-centric (single-claim) analysis, it is indeed difficult to produce enough evidence to take action in the cases that are not clear-cut. However, armed with the right aggregation and analytics tools, we can in fact achieve huge cost savings for the insurer whilst ensuring that our precious medical resources are not used inappropriately, especially at a time when the entire medical industry is enduring cuts to funding across the board.

All of the above cases were confined to single providers and were relatively easy to spot as they could be demonstrated by statistical comparisons with peers.

However, sometimes wrongful practices pervade an entire industry or sector of an industry and providers may their seek to argue that the behaviour is "normal". These sorts of aberrant behaviour cannot be demonstrated by outlier analysis against peers – rather it is necessary to compare against accepted standards of medical care. Our software has the ability to do this as well.

Recently for example we saw examples of routine screening for thyroid cancer. This may on first glance seem innocuous – most people will agree that looking for cancer is a generally a good thing. However, the issue is far from being black and white. Unnecessary tests and non-evidence-based screening can be a cause of harm to patients if it leads to unnecessary interventions. In South Korea, screening for thyroid cancer has led to an epidemic of thyroid cancer diagnoses, with the rate of diagnosis increasing from 6.4 cases per 100,000 to 40 cases per 100,000 (2). The death rate in the population from thyroid cancer has remained exactly the same, so this increased detection has not translated into any measurable benefit. Likely, screening simply highlights tumours that would not have caused any problems.

The British Medical Journal recently called for regulation of non-evidence-based screening programmes where tests with no evidence of benefit are routinely performed causing unnecessary anxiety and sometimes harm to patients from unnecessary interventions (3).

It is worth remembering that overuse of tests is not always driven by providers. On occasion it is the payors themselves that drive these behaviours. If payments are structured in a way that incentivises overuse, then overuse is the inevitable result.

The overuse of blood testing in the UK private healthcare market is one such example. The contracts in this space typically specify service rates which allow huge profits to be made. It is no surprise then that the providers are tempted to bill more than is medically necessary, given that every billed operation results in a guaranteed profit. Where tests can be billed at 20 or more times the cost of performing the test, the temptation to print money by adding extra tests can be hard to resist.

Dealing with the types of issues described above represents a very powerful opportunity to make cost savings but it requires thinking outside the box. The conventional model used by most insurers of challenging bills at the claim level will not work as the provider will simply justify the behaviour in any one case. Indeed, a common finding when I investigated such issues was cut-and-paste justifications in the medical notes designed to deal with any casual inquiries by purchasers.

It is essential that all fraud managers understand that there are three types of issues they are likely to encounter:

1. Those that can be tackled at the claim level by simple denying, recovering or part paying a claim.
2. Those that are best tackled at the entity level – perhaps by sanctioning a provider or taking them out of network.
3. Those that need to be tackled at the infrastructure level for example by changing business practices, altering contracts or altering rules.

The behaviours referred to this article generally need solutions in categories 2 and 3 above.

The simplest problem in claims is improbable or dysfunctional behaviour by a single provider or by a small group of providers. Many of these can be solved by targeting the provider in question directly. When I was working in the field, I would often start with a simple negotiation that showed the provider how far their behaviour was from the norm. In most cases, once the provider realised their behaviour has noted, they would volunteer to return any excess monies paid and self-correct their behaviour.

However, there are always going to be some bad apples; some providers will categorically deny every, refuse to amend their behaviour, and dig their heels in. In these cases, it is often necessary to take stronger action such as suspending payments or even removing a provider from network entirely.

A bigger challenge is where an issue affects an entire industry or a substantial part of it. This sort of problem can still be tackled but it may need to be at the infrastructure or industry level. It is often helpful to start by seeing if the reimbursement policy of the insurer is driving the dysfunctional behaviour as that may be amenable to change with a policy or rule amendment. For example, one of the best solutions I have seen to the problem of providers using expensive drugs where a cheaper one would generally suffice was the agreement of a fixed price which allowed the hospital to optimise its profits by using the most cost-effective version of the drug.

The UK National Health Service used such a strategy called "best practice purchasing" to influence behaviour. For example, the NHS successfully encouraged day case gall bladder surgery by temporarily offering a higher reimbursement for surgery done as a day case compared with the rate for inpatient treatment. This encouraged hospitals to invest in and adopt day case pathways which then became business as usual.

Sometimes more radical solutions are needed, for example. One of the worst areas for fraud is dental treatment and following an audit which showed nearly 50% of claims to have a fraudulent element, an insurer took the decision to carve out some maxillofacial procedures from standard reimbursement and only pay for these in a small network of units which agreed to follow good practices and bill ethically. I have seen this done in other areas and it works well.

Dealing with borderline cases in medical claims requires a change the mindset. We must go from managing individual claims to managing Providers and Practitioners as a whole. One of the ways to embrace this mindset is to adopt a suitable software solution to assist the inhouse experts in their work. To tackle endemic or large-scale behavioural cases, we recently added the concept of "Advisory Alerts" to the Kirontech Health Insurance Platform.

One of the recent upgrades to the Kirontech software was the creation of Advisory Alerts or simply "Advisories". Advisories highlight features in claims that are not in themselves worthy of creating an alert but which if repeated merit further investigation. They were very much created for this situation. Advisories are, as far as I know, unique to our software.

The types of issue we find vary enormously, and they can affect a single provider or sometimes be much more widespread. They can be fraudulent in nature, but more often they represent poor or wasteful practice or low-value care.

Here are a few examples we have seen over the last 12 months.

During an analysis of outpatient claims, we came across a medical clinic which used or claimed to use intravenous carbapenems (like meropenem) as a first line treatment for conditions like simple urinary tract infections. As often is the case, challenged on its own, any individual claim would tick all the appropriate boxes to justify the treatment. However, meropenem is fairly heavy artillery as far as antibiotics are concerned: it is a type of medication that should be reserved for serious infections and is not intended as the first line of defence for simple community-acquired infections. It is inappropriate and wasteful and leads to the development of antibiotic resistance, which is already a global threat to healthcare.

Another finding was a hospital where all patients were billed doses of chemotherapy which would only be appropriate for large or obese patients. Again, insurers don't generally know the size of policyholders, so no one would query these doses in a single case. But it's extremely unlikely all patients would be large and heavy, requiring such a dose.

We often see hospitals which use or claim to use expensive branded drugs when cheaper and equally effective generics are available. Sometimes this is simply an example of wasteful behaviour; in other cases, it is fraud where the hospital uses the cheaper version and bills for the most expensive. We know of one insurer that has made multi-million dollar savings in this arena simply by checking pharmacy dispensing records against bills.

In addition to the solutions above, there are of course many other potential approaches to tackling the problems of claims that are not clear cut. The scope is limited only by the imagination and creativity of the purchaser - the real message here is that these issues can generally be successfully addressed if the purchaser is aware of them and is prepared to think outside of the usual systems for managing claims.

As always, we would be interested in hearing views from other organizations about their experiences and how they address such issues.

Dr Simon Peck, Chief medical Advisor, Kirontech UK Ltd