Coding: The Under-coding Epidemic

April 1, 2006

Expert Bill Dacey on the undercoding epidemic.

Physicians really need to take a look at the aggregate pattern of their coding practices; physician office service coding is ailing.

The symptoms of the coding problem are clear enough if you look at the patterns and percentages of office visit (E&M) codes assigned, and you also know what those codes are intended to represent. It helps to have coding data specific to discreet segments of the population, too.

We will look at Medicare data here because it is the easiest to come by, and also because we can make some generalizations about the Medicare population and the likely intensity of office visits.

The focus of this discussion will be the under-coding of services - that is, selecting a code that does not capture the true intensity or amount of work actually performed. It's true that over-coding exists as well, and also that services performed are under-documented. But here we'll focus on plain old under-coding - be it done out of fear, lack of education, or one of the many other reasons that physicians just don't seem to get this right.

Refer to Tables A and B. Table A represents the new patient office visit codes assigned to Medicare patients during calendar year 2004 by providers of the various specialties shown. Table B illustrates the same for established patient office visits. We have shown the data for both 2003 and 2004 for internal medicine to illustrate how little the data changes from year to year. There are tens of millions of encounters represented in these charts - and tens of billions of dollars. We can call this method of observing coding "pattern monitoring."

Although the data for new patients is meaningful, there are many more visits for the established patients, so let's look there first. Observe the full range of E&M codes - 99211 through 99215. The middle code of this range, 99213, has long been the most widely reported E&M code. It also represents close to one relative value unit (RVU), the basic unit of physician work.

In 2004, internal medicine physicians reported this code 31 million times, representing more than $1.5 billion in allowed charges. Family medicine physicians reported it 25 million times (they have fewer aggregate encounters). What is this code that has been reported for these two primary-care specialties to the tune of nearly $3 billion?

Well, per both AMA and federal coding guidelines, 99213 represents an office visit involving some history- taking and an exam, but most characteristically involving low medical decision-making. This is further defined in federal tables as one stable chronic illness or injury, an acute uncomplicated illness or injury, or a presenting problem of low-to-moderate severity requiring a review of limited data and having limited management options.

Does this sound like the average visit for your average Medicare patient? Probably not.

Remember, all the codes assigned in these tables are for Medicare patients - patients likely requiring a higher average acuity or breadth of medical management than the general population.

Look at the 99214 codes. Per AMA and federal guidelines, these represent those patients with two or three stable chronic illnesses, or those with one chronic illness with mild to moderate exacerbation, progression, or treatment side effects. Shouldn't there be more of these than the 99213 visits for Medicare patients?

Physicians should know what these codes mean. They have been used in their present form since 1992. They have been the subject of endless debates, seminars, and audits. Physicians should by now be competent in their use - but yet again, year after year, the data tells us something that our senses and daily observations reject. These specialties treat Medicare patients at a roughly 5:3 ratio of single-system, low-complexity problems versus multi-system or exacerbated/progressing problems. But the reality is that most Medicare patients have multiple complex and progressing problems.

This data is even more disturbing when compared to data on general population, non-Medicare patients. Such patient profiles are not all that different in some specialties than data on Medicare patients. This suggests that the codes reported should be identical for the general population as for the Medicare population - which is just not true if the codes are used correctly.

These profiles also seem to show that if coding by medical decision-making is any indicator, the reported codes reflect a significant under-coding of services. One state Medicare medical director has said that he "expects most established patient visits for Medicare patients to be of the 99214 variety." If this is so, why are between 50 percent and almost 70 percent of these encounters for Medicare patients coded at the 99212 and 99213 levels?

The distribution of codes across this spectrum tells us that many physicians under-code services when compared to the work they actually perform for this population. Perhaps we are seeing the cumulative result that occurs when physicians are either afraid to code for what they do or don't understand what the codes mean.

When are physicians going to step up to the plate and accurately code for their services? It is true that these profiles have steadily improved over the past 10 years, and by improved I mean have shifted more to the right as appropriate. But for now, or at least as 2004 data indicates, most providers still have difficulty accurately reporting these codes. In my day-to-day experience of auditing and coding, I see that physicians know how to treat these patients. There is no shirking of the work and resource expenditures, but when it comes time to ask for the money, or to just make it clear what they did - physician efforts fall short.

So physicians, if you know you aren't coding for your services as you should be, ask for help, get some education, or solicit a friendly audit or chart review - just take your medicine.

Bill Dacey is principal in The Dacey Group, a consulting firm dedicated to coding, billing, documentation, and compliance concerns. Dacey is a PMCC-certified instructor and has been active in physician training for over 10 years. He can be reached at or via

This article originally appeared in the April 2006 issue of Physicians Practice.