24 September 2012

Upcoding is in the eye of the beholder

There was an interesting and important article in the NY Times the other day about the gradual increase in the average E/M coding levels used by doctors over the last few years. For the non-docs, med students and ER trainees out there, here is a brief summary of the way physician billing works in the ER:

During and after the patient encounter, the physician creates a medical record. This is a pretty standardized document including the history the doctor got from the patient, the exam, any tests and the medical decision-making. And, of course, the diagnosis. This gets reviewed, typically, by a professional coder who then applies Medicare's rules to determine what level of complexity the service was, and the associated cost of that service. (In some cases, the doctors code their own charts, and some are automatically coded by a computer, but both of these are exceptions to the rule.) ER visits are coded on a 5-level scale, with a Level 1 (99281) being the simplest and a Level 5 being the most intense. These are called E/M codes because they refer to the "Evaluation and Management" of the case, as opposed to, say, a surgical procedure.

The amount Medicare pays the doctor for an E/M code in the ER ranges from about $20 to $175. (If you've ever received a multi-thousand-dollar ER bill, chances are good that was the hospital bill. The physician's fees are much more modest in most cases.

Medicare, of course, tracks utilization and recently published a report (PDF Link) about the recent trends in the doctors' use of the codes. What they found is that over the last decade, we have shifted from using the lower codes and towards the higher ones:

This is where it gets interesting. The feds and some consumer watchdogs view this trend as clear evidence of fraud and abuse, that the physicians are "upcoding" the visit levels to increase their income. On the other hand, I've spent the last decade trying to educate physicians on how to document the patient encounter so that you can accurately capture the legitimate value of the service provided. So I look at that trend and think to myself "Job well done."

From the perspective of a practicing physician, the rules that govern the documentation required to capture a service level are deliberately onerous and designed to produce downcodes. They require the doc to collect far more data than is actually necessary based on the actual condition of the patient. You forget to check one box, you leave out one required element, and despite the complexity, gravity and risk of a patient's condition, you will not be paid for the service. For years, we have been losing money to these archaic rules (they date from 1995), and we have been struggling to stem the leakage of revenue from our practices.

The industry responded to these rules by developing tools to comply with them. The first thing was to have professional coders. Prior to 2000, a substantial majority of ER charts were hand-coded by the physician; now that is quite rare. Then we got templated paper records which prompted the docs to get all the required data points. Now we have EMRs which do the same thing more efficiently. It's no surprise that as an industry we have gotten better at meeting the guidelines.

And then there is the fact that the ER is a different place than it was in 2000. Our patients are older and sicker. We do more in the ER than was true in the past. Patients are rarely directly admitted any more, but rather get the majority of their admitting workup done in the ER. I don't know how much of the skew in the above graph is due to these factors, but they shouldn't be disregarded.

But I don't like where this is going. The government is desperate, understandably, to save money on healthcare expenditures. They seem to have assumed their conclusion that the increased coding levels is fraudulent and unjustified, and there seem to be few voices disagreeing with them. Furthermore, there is some inappropriate upcoding, and it's very easy for a patient with an egregious bill or a certain physician (or group) who pushed the envelope too far to be held up as anecdotal proof that doctors are all a bunch of thieves.

The article quotes some insurers as saying that nearly half of the charges being submitted are "upcodes," without noting that the insurers have a vested interest in not paying and may not be presenting an honest picture of the claims. We had a run-in with a certain national insurer who analyzed a few hundred claims, found that 80% of them were upcoded and demanded a couple of million dollars in restitution. We fought back, of course, defended each chart on the merits of the care, and established that in fact 97% of the time the correct, higher, code was appropriately used. That the insurers are not necessarily honest players here passes unremarked.

I think changes are coming, and it worries me. We already have one insurer saying that they simply won't pay for cloned records — but how will they tell if a chart has been cloned? Do charts have DNA? I think what we are going to see is the carriers becoming very aggressive in arbitrarily denying payment or demanding restitution. I would love to see revised and streamlined coding rules to replace the old ones, but I suspect that if that were to happen, the new rules will be designed to be more restrictive and more burdensome.


  1. This comment has been removed by the author.

  2. My understanding is that an E/M 3 code is supposed to be a routine visit of a known patient at a primary care physician's office. In that case, I'd say that anything that qualifies as an emergency should be at least a level 4 or level 5.

    Tangentially, it's good to see you posting again.

  3. Nice summary of a complicated subject, and there is good cause to worry, especially about health plans and aggressive auditing and down coding. Unfortunately, upcoding by some groups is a reality, and perhaps we should do some of our own housecleaning. Also, I have on more than one occasion suggested that the AMA's examples in CPT need to be greatly expanded, and also need to highlight and help differentiate the thresholds between E&M levels, using progressively more complicated iterations of illness and injury presentations.

  4. Good to see you back posting. Please stay with your loyal fans.

    Interesting that insurance companies are presented as having no bias, when the reverse is so obvious.

  5. I'm a little ashamed to admit it took me a couple minutes to read that chart right. The bars are grouped by coding level. The sum of all the bars for a single year is 100 percentage points. The percentages above the tops of the groups are: (100% * billed visits at this level in 2010 / all billed visits in 2010) - (100% * billed visits at this level in 2001 / all billed visits in 2001 * 100%). It's just a comparative bar graph.

  6. Ignore that last "* 100%". Sorry. Always miss one typo ...

  7. Just a random question - how much of your time would you say you spend worrying about how to code your work? I was thinking about how US healthcare is supposed to be so much more expensive, and wondering how much of that is that we waste doctors time doing administrative tasks solely for the purpose of determining who pays who how much.

  8. The other thing that has been happening is that we've all been trying to educate patients on what really constitutes an emergency. Whether we've actually accomplished that is debatable, but if the percentage of patients who are there for non-emergent and non-urgent complaints is dropping, it will at least appear that the providers are upcoding, even when they aren't.

  9. I tend to be with you, that this represents the effects of EHR/EMR/PM software that generates an E/M code for clinicians who can't be bothered, and/or would otherwise undercode to stay on the safe side of RAC and denials.

  10. Yay, you're posting again! I agree that this is ominous. I would wonder even further about the trend towards critical care billing as well; we've been undercoding those for years. As a new grad, we never thought about this in residency. I feel betrayed in some senses--the feds set the rules, then asked us to get EMRs or else, and are now crying fraud when we use their rules and the EMRs they forced us to get make us better at billing. As for the 'cloned chart' issue--what did they expect from computers? Can't win.