“Our Reasons are not prophets When oft our fancies are”
– William Shakespeare, The Two Noble Kinsmen
One area in which I specialize is the field of predictive analytics, so it would seem that I should be pretty good at predicting what to expect in the coming year.
The problem, however, is that predicting the future normally requires at least a modicum of accurate information about the past, and in our industry, there is so much chaos that it is difficult to rely upon the past as an indicator of the future. It was Niels Bohr, the famous Danish physicist, who was quoted as saying “prediction is very difficult, especially about the future.” Of course, that would depend on whether you are a fox or a hedgehog, but alas, I digress.
There are several interesting and important analytics topics I wanted to discuss with regard to what I see as changes that likely will occur throughout this year.
The Fraud Prevention System: Since I already began with a mention of predictive analytics, let’s jump right into a discussion of the Fraud Prevention System, or FPS. FPS was created in 2010 as a joint effort between Northrop Grumman’s IT division, Verizon, and Wellpoint’s National Government Services division. The FPS is a sophisticated predictive analytics tool that the Centers for Medicare & Medicaid Services (CMS) uses to identify claims in real time that may meet their criteria for fraud and abuse; the system then uses those results to review additional claims for providers that have been identified by the system as presenting the potential for high risk. This is similar to how FICO (Fair Isaac Corporation) works to prevent credit card fraud.
For example, a few weeks ago I went to a gas station and the pump was so slow that after about a gallon, I turned it off and went to another gas station up the street. This time, when I tried to use the same credit card, it was rejected. The FICO system knew that this was an unusual pattern of purchasing for me, so it held the account until I could call and verify that I did, indeed, have the card and that the first purchase and second attempt both were made by me.
In the case of the FPS, 100 percent of all Medicare fee-for-service claims are passed through these algorithms, amounting to millions of claims a day, and according to CMS, in the first three years FPS prevented the payment of $820 million in potentially fraudulent claims.
One problem that CMS has experienced, possibly as a result of the “pay-and-chase” audits that have been initiated because of the FPS system, is a reduction in recovery coinciding with a very significant reversal rate on appeal. Because of that, CMS, through the FPS, has been looking at ways to minimize false positives, which would be a benefit to both the program and to providers. As such, beginning this year I expect that the FPS algorithms will be expanded to identify lower levels of non-compliant healthcare providers, possibly through the inclusion of taxonomy codes, in order to identify overpayment issues that fall below the threshold of fraud.
Improvements in statistical analyses: This is always a difficult area for me to predict because it tends to be somewhat unstable. Actually, it’s a bit like the stock market in that over a long period of time, the market tends to climb, but in the short term fluctuations resemble that of the Brownian motion, or what is referred to as a “drunkard’s walk.”
I like to believe that CMS auditors are working to improve their statistical capabilities because we have done such a great job at successfully challenging their extrapolations. In any case, I have seen an increase in the level of competency surrounding CMS extrapolation audits. Part of the problem still lies within the program integrity manual (PIM), which conflicts with what I consider reasonable and acceptable statistical practices. For example, the PIM allows huge degrees of elasticity when it comes to things such as sample size and methodology. For example, in Chapter 8, Section 4.2, the PIM reads as follows:
“If a particular probability sample design is properly executed, i.e., defining the universe, the frame, the sampling units, using proper randomization, accurately measuring the variables of interest, and using the correct formulas for estimation, then assertions that the sample and its resulting estimates are ‘not statistically valid’ cannot legitimately be made. In other words, a probability sample and its results are always ‘valid.’”
I would vehemently disagree with many parts of the above statement. In fact, I have been before administrative law judges (ALJs) in circumstances during which the auditor would argue that even though my findings supported that the sample was not statistically representative of the universe of data because it fit the definition above, it is still acceptable. No sample is “always valid” until it passes tests for representativeness, and in all of the audits in which I have participated, I have never seen an auditor conduct a post-hoc test on the sample to ensure that this was true. Another problem I am hoping to see addressed is cooperation from the auditor when it comes to providing data. I actually have won several cases because the auditor simply failed to provide the information required by the PIM. But in other cases, even when this failure occurred, the ALJ would remand the case back to a lower appeal level or give the auditor more time to produce the documents or the data. The PIM is pretty clear that the auditor must provide the practice with everything necessary to replicate and validate the study.
In summary, while I continue to see an increase in the sophistication of auditors with regard to statistical sampling, extrapolation, and cooperation in providing information, there always remains the possibility of what looks like a random walk, depending on the case and the contractor.
Appeals and accountability: On April 28, 2015, the U.S. Senate Finance Committee held a hearing titled “Creating a More Efficient and Level Playing Field: Audit and Appeals Issues in Medicare.” As noted by Chairman Hatch in his opening statement, Medicare’s hiring of contractors to conduct audits of claims submitted to Medicare “has led to a seemingly insurmountable increase in appeals, with a current backlog of over 500,000 cases … (which) has resulted in long delays for beneficiaries and providers alike.” Chairman Hatch also noted that “large portions of the initial payment determinations are reversed on appeal” and “such a high rate of reversals raises questions about how the initial decisions are being made and whether providers and beneficiaries are facing undue burdens on the front end.”
The report stated that, back in the 1990s, some 30 to 40 percent of appeals were overturned at the first level. This figure has declined precipitously over the years, and for 2014, the committee found that only 1.2 percent of appeals were overturned at redetermination. Unfortunately, there are strong incentives in play that conflict with what should be an honest appraisal of the appeal facts at that level. On May 15, 2015, the CMS indicated that Recovery Auditors (RAs) are required to maintain an overturn rate of less than 10 percent at the first level of appeal. Failure to do so will result in CMS placing the RA on a corrective action plan. This seems to present quite a conflict of interest for the RAs, and in my opinion, it supports the need for a complete restructuring of the RA policies and guidelines. How about this: for every claim over the 10 percent threshold that is reversed on appeal, the auditor has to pay the provider $100 to help cover the expense of what is revealed as clearly a frivolous finding. Unfortunately, I don’t expect to see much in the way of increased accountability on the part of the government or its contractors this year.
Communication and coordination: From the standpoint of compliance analytics, I believe that we are going to see more communication and coordination between Medicare carriers, private payors, and law enforcement. I remember a case I worked on a few years ago in which a private payer, concerned over how much a provider was charging for some personal injury services, was able to convince the government to start a fraud investigation, which then lasted nearly two years before it was summarily dismissed. Defending against what turned out to be a tactic to force the practice to lower their fees cost them a quarter of a million dollars in fees for attorneys and experts. In 2015, I saw many private payor audits that seemed to mirror Zone Program Integrity Contractor (ZPIC) audits that had occurred only months prior. Payors definitely have the upper hand when it comes to analytics, as providers are slow to embrace new technologies and to spend the money necessary to engage in advanced compliance analytics. I think that will change this year as more and more pressure is put on the provider side through an increase in extrapolation recovery audits and pre-payment withholds. It’s going to be buy or die for many practices this year.
Overall, I predict that we are once again in for a tough ride, as it appears that the Patient Protection and Affordable Care Act will continue to be funded off the backs of hard-working providers. According to Eric Hoffer, “the only way to predict the future is to have power to shape the future.” So maybe it’s time for a revolution.
And that’s the future according to Frank.
About the Author
Frank Cohen is the director of analytics and business intelligence for DoctorsManagement, a Knoxville, Tenn.-based consulting firm. Mr. Cohen specializes in data mining, applied statistics, practice analytics, decision support, and process improvement.
Contact the Author
Comment on this Article