Essential Information to Protect Your Facility from Auditors

EDITOR’S NOTE: This is the third in a multi-part series on the Comprehensive Error Rate (CERT) study in which senior healthcare analyst Cohen describes how to read the study and interpret the data to protect your facility from audits.

As discussed in prior articles, the Comprehensive Error Rate Testing (CERT) study is broken into three primary types: Part A, Part B, and durable medical equipment, prosthetics, orthotics, and supplies (DMEPOS).

In this article, we are going to go through the Part B tables. 

Part B claims are for what many refer to as “profee,” or professional, services. These are services provided by physician and non-physician providers (i.e. nurse practitioners, or NPs, and physician assistants, or PAs), either in the provider’s office or some other facility, but where the claims are submitted by and on behalf of the physician or non-physician practitioner (NPP). The most obvious example would be an office visit in the provider’s office. It’s also possible that, for the convenience of the patient, the office visit service is provided in some other location, like the emergency room. In this case, the location code would indicate that the patient was seen in the emergency room (POS 23); however, the provider (not the facility) would still bill for the visit.

If you were to look in the 2024 National Physician Fee Schedule Relative Value File (January release), you would find 16,323 unique procedure codes (both HCPCS levels 1, 2, and II). When you include the 2,176 procedures with modifiers (26, TC and 53), the total goes to 18,500. That’s a lot of procedures!! As one might expect, it is about impossible for the Centers for Medicare & Medicaid Services (CMS) to sample every one of these in order to produce its improper payment results. Rather, using a statistically valid stratified random sample, the idea is that the sample itself will mimic the distribution of these codes within the general Medicare database (and therefore, be mostly representative of the utilization data). Unfortunately, not having access to the raw data, this is not something I am able to validate, but assuming that the sample is truly a statistically valid random sample (SVRS), then the assumption of representativeness can be made.

For the Part B analysis, CMS pulled 12,303 claims, of which 12,001 were reviewed as part of the study. For those 12,001 claims reviewed, CMS estimated an average improper payment rate of 10 percent, with a 95-percent confidence interval of 8.6 to 11.5 percent. This equates to a precision of under 15 percent, which is not bad, considering the diversity of codes and sample size. Using extrapolation, CMS estimated that this error rate translated to a projected improper payment total of around $11 billion, which accounts for approximately one-third of the projected total overpayment of $32.2 billion. When broken out by under- and overpayments, table B4 shows that overpayments accounted for $10.7 billion, and underpayments accounted for around $1 billion.

Table D1 reports the top 20 service types with the highest improper payment amounts for Part B services, while table E1 reports the same, but by improper payment rate. Below is a snapshot of that table.

Note that these are listed in descending order by improper payment rate, not amount. For example, radiation oncology services, on average, have an improper payment rate of 36.3 percent, meaning that statistically speaking, over a third of all radiation oncology services are paid in error. If you go back to table D1, you will see that radiation oncology services are ranked 11, with a total projected improper payment amount of $331 million. Conversely, table D1 projects improper payment for established office visits at over $1.1 billion. Yet, this group of services does not even show up in the top 20 for overpayment rate. It is important when looking at these tables, that you read the title carefully, as data for similar types are often presented in different ways, each having its own individual interpretation.

Appendix F reports improper payments by type of service for each type of error. For example, table F1 reports the top 20 types of services with errors due to no documentation. Table F2 is the same, except it’s for insufficient documentation errors. Table F3 is for errors associated to medical necessity. Interestingly, table F5 reports the top 20 types of services with down-coding errors. Notated on this table is the following: “downcoding refers to billing a lower-level service or a service with a lower payment than is supported by the medical record documentation.” The top of the list here is hospital outpatient services, followed by established office visits.

You will notice that these tables are not restricted to Part B services. Those data can be found in Appendix G. For example, table G1 reports improper payment rates by service type for Part B services. 

Here, we can see that established office visits report an improper rate of 6.4 percent for just over $1 billion. Both initial and subsequent hospital visits are on the list as numbers 6 and 7 in projected overpayment amounts, at $661 million and $474 million, respectively. With respect to overpayment rates, chiropractic services continue to top the list in many categories, including this one.

Appendix H reports improper payments by referring provider. The header text for this appendix states the following: “This series of tables is sorted in descending order by projected improper payments. All estimates in these tables are based on a minimum of 30 lines in the sample. Appendix H shows the referring providers or provider types for the top three service types for Part B and DMEPOS”.

For example, table H1 reports improper payment rates for established office visits by referring provider. 

As far as improper payment rates go, you can see that internal medicine is on the top of the list, followed by family medicine and nurse practitioners. One might opine that these primary care specialties are atop the list because they provide the most referrals, which is likely quite true. And because their referrals account for a greater proportion of all referrals, it stands to reason that the rate of improper payment as a percentage of all improper payments would also be higher.

Appendix I reports projected improper payment by provider type for each claim type. For example, in table I1, we can see those improper payment rates by Part B provider type.

Note that, when it comes to amounts, internal medicine is again at the top of the list. And again, this is because internal medicine providers report more procedures as a group than other specialties. You can also see that cardiology and family practice are in the top five, and again, this is because they are primary care specialties that report lots of procedures.

Appendix J reports projected improper payment, including type of error, by provider type (specialty) for each claim type. In essence, this is the same as G1, except that it includes the root cause. Let’s take a look at the root causes for some of those same provider types.

Let’s start with row 5: pulmonary disease. Here, we see that the improper payment rate for pulmonary disease physicians is 32.4 percent. We can also see that insufficient documentation is responsible for 83.3 percent of the improper payment errors. If I were a pulmonary disease provider, my audits would focus on whether I have sufficient documentation to meet the criteria for each billed service. This is a great tool for performing value-based audits!

Appendix K looks at specific codes. Table K1, a very popular table among auditors, looks at improper payments for evaluation and management (E&M)-specific codes. 

On row 1, you can see that this is reporting 99214 only. Here, it shows the improper payment rate at 6.7 percent. But look at row 3, the highest level for initial hospital visits (99223). The improper payment rate is 25.6 percent. Again, if I were an outside auditor, I would look for organizations that had high billing amounts for these codes, because statistically speaking, some quarter of them would be adjudicated as improperly paid (and mostly overpaid).

The CERT report offers huge incentives for these contractors, and we know from speaking with them that they do rely upon these results to assist with targeting physicians for audit.

Table K2 reports the impact of 1-level E&M errors. The footnote for this table states “table K2 shows the improper payment rate estimate for claims that were found in error due to 1-Level E&M coding difference.” This is a very controversial issue since many studies have shown that the level of disagreement among professional coders for 1 level is around 42 percent. For example, based on these studies, if I were to throw a vignette up on a screen in front of a room of 100 professional and experienced coders, 42 of them would disagree with the other 58 on which code this should be, within 1 level. That’s pretty significant. And table K3 reports E&M service categories with upcoding errors. The footnote for this table states, “Upcoding refers to billing a higher level service or a service with a higher payment than is supported by the medical record documentation.”

Here, we can see, for example, that 13.7 percent of the improper payments can be attributed to upcoding initial hospital visits, and this aligns with what we saw in the previous table.

Appendix L reports overpayments only, and the data are very similar to what we have seen prior, only excluding projected underpayments. 

Appendix M, on the other hand, reports the same improper payment data, only exclusively for underpayments, which we have already seen are few and far between (0.2 percent).

Finally, Appendix N contains the summary of the SSOE (statistical sampling and overpayment methodology) utilized by CERT in both creating their samples and projecting overpayments.

In summary, for those providers that bill under Part B, the CERT study contains a plethora of very important and useful information. Since we already know that outside contractors use the results of the CERT study to target physicians, it shouldn’t be a giant leap for those provider organizations to use the CERT study to both predict what those audits might scrutinize, as well as proactively review those targets to help mitigate future audit risk.