Sudden Outbreak of Kwashiorkor (Severe Malnutrition) in Maryland, or Computer-Facilitated Upcoding?

At a Feb. 2011 post "Does EHR-Incited Upcoding Need Investigation by CMS, And Could it Explain HIT Irrational Exuberance?" I wrote about how EMR's can facilitate medically unjustified upcoding, a form of fraud.

Here's a peculiar upcoding story out of Baltimore. I am cited:

Malnutrition diagnoses at Kernan were fraud, feds say
Baltimore Sun
Jay Hancock
7:44 p.m. EST, November 7, 2011

It looked like a public health emergency. Hundreds of patients checking into Kernan Hospital were getting diagnosed with a severe form of malnutrition called kwashiorkor.

Taking its name from a Ghanian word, kwashiorkor (link) is typically seen in children and is marked by swollen feet, a swollen stomach and skin ulcers. It's common in Africa and developing nations elsewhere but is hardly heard of in the United States.

But by 2008, according to records obtained from state regulators, the West Baltimore orthopedic hospital was diagnosing one in every eight patients with the disease that helps define famines in Somalia or Bangladesh. A few years earlier no Kernan patients were diagnosed with kwashiorkor.

"That's very odd," said Dr. Scot Silverstein, an expert in electronic patient records and an adjunct professor at Drexel University in Philadelphia. "For an orthopedic hospital? In Maryland? That's a disease that's typically a disease of foreign countries."

I've never seen a case personally, despite having trained at places like Boston City Hospital in the 1970's, an extremely poor area at the time.

Perhaps the deteriorating economy was harming Marylanders a lot more than anybody suspected. Or perhaps the lucrative medical payment system was inducing Kernan employees and clinicians to report a disease that wasn't really there.

I tend to think the latter is more likely.

Last month the Justice Department accused Kernan of fraud, alleging in a civil complaint that it falsely labeled patients as kwashiorkor victims to inflate bills paid by government health programs and insurance companies.

Co-morbidities increase reimbursement.

"There are a variety of reasons why patients at Kernan may be classified with malnutrition," system spokeswoman Mary Lynn Carver said in the statement, including low weight and nutritional problems indicated by blood tests.

Another reason, government attorneys said, was to rip off Medicare, Medicaid and health insurers. Kernan's patients typically need new hips, ligament repairs and so forth. But labeling patients as simultaneously suffering from kwashiorkor or other malnutrition increased the amount Kernan was able to bill.

Malnutrition is one thing. Kwashiorkor is another entirely. It is a syndrome with multiple symptoms, signs and features.

Kwashiorkor is so rare I had trouble finding American experts to discuss it. The Centers for Disease Control and Prevention in Atlanta don't track it.

"One of our nutritionists is aware of a handful of cases in children whose parents are vegan and whose diets don't include animal protein," said CDC spokeswoman Kathryn Harben.

I wanted to interview Dr. Jaime Sepulveda, head of global health sciences at the University of California, San Francisco, and former head of Mexico's National Institutes of Health, about kwashiorkor.

"He said he didn't feel comfortable serving as an expert on it, since he's never seen a case in his life," a UCSF spokeswoman said.

But somehow Kernan experienced an apparent kwashiorkor outbreak starting in 2006, the only Maryland hospital to do so. Its cases of kwashiorkor as a secondary diagnosis grew a hundredfold, from three in 2005 to 358 in 2008, according to data from the state Health Services Cost Review Commission.

That was more than a third of all the diagnosed kwashiorkor cases statewide. At Kernan, 12.7 percent of all patients were diagnosed in 2008 with the rare ailment. At all Maryland hospitals, on the other hand, only 0.13 percent of that year's patients had kwashiorkor on their charts, according to the HSCRC data.

Stunning. Here's the catch:

Thanks to inappropriate diagnosis of kwashiorkor and other malnutrition, Kernan fraudulently collected $1.6 million from federal health programs, according to the complaint filed Oct. 17 by Rosenstein's office. The government seeks more than $8 million in damages and recompense.

Upcoding has been a problem before computers. As in the aforementioned post, computers can pose a "moral hazard" by making "suggestions" that can facilitate upcoding.

... the incidence of upcoding may be increasing.

"You might call it an institutionalized practice that is especially facilitated by electronic health records," which are used by more and more hospitals, said Drexel's Silverstein. The computer might prompt a clinician, he said: "If you just do this one little thing, you might generate a higher level of acuity" and more revenue.

When the upgrade is justified by patient symptoms, it's fine. But the government says Kernan crossed the line.

The Complaint, United States of America (on behalf of HHS and DoD) vs. Kernan Hospital can be found here (PDF).

It is quite a read, especially what appears to be, if true, an internal push for rendering the malnutrition and kwashiorkor diagnoses through 'creative' computing and internal pressures on clinicians. See, for example, pgs. 10-11.

I note that I could have easily advised the hospital not to use the diagnosis of kwashiorkor or do these things, had I been on staff there. I would have opined that it would stick out like a sore thumb via the algorithms payers use to detect fraud, e.g., via sudden, strange changes in claims. One wonders why nobody else gave that advice ... (fear?)

-- SS

Dr. Ross Koppel: Another Angle on EMR-Incited Upcoding

At my Feb. 15, 2011 post "Does EHR-Incited Upcoding (Also Known as "Fraud") Need Investigation by CMS?" I explained how, based on data I received from an active ED physician, EHR's and their algorithmic "billing advice" tools were apparently inciting massive upcoding.

The upcoding was causing a huge spike in "level 4" and "level 5" determinations for ED billing when comparing 2002 data to 2008 data, with resultant billions of dollars of increased Medicare expenditure for a few extra mouse clicks. See the above post for details, as shown on the graphs below:

(E/M level increases, 2002 compared to 2008. Click to enlarge)

(Est. increased Medicare expenditures from EHR upcoding, comparing 2002 to 2008. Click to enlarge)

Another angle on EHR-incited upcoding comes from University of Pennsylvania sociologist and EHR expert Ross Koppel, PhD. His words are reproduced with his permission below.

First, Dr. Koppel was quoted in today's Wall Street Journal in an article entitled "Many Medical Bills Contain Errors That Could End Up Wrecking Your Credit Score, Here's What You Need to Know" by Jessica Silver-Greenberg:

... What's more, some health-care experts say the number of errors could jump in coming years. That's because the 2009 economic-stimulus legislation included $19 billion in spending to encourage the use of electronic health records—a major source of billing mistakes, says Ross Koppel, a sociology professor at University of Pennsylvania's Center for Clinical Epidemiology and Biostatistics who has studied electronic records extensively.

Dr. Koppel adds the following commentary for HC Renewal:

With several EHRs, there's "auto-fill" for the EHR that reports on parts of the body the physician does not examine. So, for example, if the physician is a hand specialist, she/he will probably not examine the legs, head, chest, back, etc.

On many EHRs this is listed [e.g., entered by the user through clicking a form or checkbox - ed.] as "unremarkable" which in medical-speak means "I've examined these organs/body parts and found no problems."

In billing terms, that means "I've examined these parts and should be paid for a more comprehensive visit....what's called "up-coding."

It's actually dangerous for physicians because if the patient next week walks in with a major tumor that the doc should obviously have seen (had she/he examined that body part) then the physician is on the line and can get sued.

On other other hand, if the unexamined parts don't turn up in a medical case, it means the doc will make more money...with the only danger to the doc that she/he can be charged with fraud by the government and or by a payer.

So, in effect, today's EHR's, not just through algorithmic "recommendations" for upcoding but also through their template-checkbox nature, incite upcoding.

The resultant increased expenditures do not reflect more medical care provided or cognitive processes utilized, but merely a small amount of elbow, carpal and index finger activity:

Click...click...an easy way to generate billions of dollars of additional billing?

-- SS

Does EHR-Incited Upcoding (Also Known as "Fraud") Need Investigation by CMS, And Could it Explain HIT Irrational Exuberance?

It has long puzzled me why an experimental technology that costs tens or hundreds of billions of dollars to roll out organizationally and nationally, that individual medical centers are spending upwards of $100 million or more upon, and that has a track record as poor as suggested in posts such as "An Updated Reading List on Health IT" and "MAUDE and HIT Risks: Mother Mary, What in God's Name is Going on Here?" is so popular.

At a Feb. 2010 post "A Lawsuit Over Healthcare IT Whistleblowing and Wrongful Discharge" I summed up other critical HIT issues succinctly, with abundant hyperlinks to other materials as below:

... I have written about health IT problems extensively on this blog and at my academic website on HIT failure. These include but are not limited to: medical informatics specialists ignored by nonclinician IT personnel [the link is to another case involving critical care]; inverted and paradoxical organizational structures where IT facilitators become HIT project leaders and clinical leaders become HIT project facilitators; ill-conceived and poorly implemented mission hostile health IT; perverse and in fact clinically cavalier IT "politics"; failure to obtain patient informed consent as if health IT is an elite world not subject to the same ethical obligations as medicine; probable violations of Joint Commission safety standards and hospital executive fiduciary obligations, and numerous others.

At a Jan. 2011 post "Healthcare IT Delirium" wrote that:

... On top of an irrational exuberance (see this blog query) largely unsupported by the literature (e.g. here), the technology is experimental, its rollout is a grand national experiment in social re-engineering of medicine, there is no patient informed consent, nobody is in control, and nobody is taking responsibility for regulating the domain despite known risks. The results will very likely reflect the Wild West free-for-all that is now extant.

Finally, at a Feb. 2011 post "An Updated Reading List on Health IT" I aggregated recent literature that sheds significant doubts on health IT beneficence.

Is the popularity of health IT as it exists in 2011 a paradox, or not?

When I was dealing with public employees in a transit authority several decades ago, I learned that mysterious circumstances and paradoxes were not paradoxes at all. The assumptions were simply wrong. Often, mysterious circumstances proved to be related to money (as in worker's comp fraud) and/or money related to [illegal] drugs.

Below are some charts and data from a HC Renewal reader, an ED physician, that may answer the mystery of the "irrational exuberance" exhibited by healthcare organizations towards clearly disruptive, medically risky and liability-provoking experimental technology. This is especially so in high risk areas such as an ED.

It should be noted that my recommendations as a CMIO a number of years ago in a very large regional medical center were that an ED would benefit best clinically from paper charts supplemented by document imaging, so that images of past charts could be instantly retrieved. ED's are busy places where clinicians have little time for distractions (such as computer data entry), but where access to past charts can be important. ED charts are also not generally very long or complex documents. Further, document imaging is a mature technology (e.g., Documentum).

Those ED recommendations still stand.

(Incidentally, a relative, injured in 2010 via an EHR issue, would certainly be in far better shape had this been the case. The error would not have occurred. I've worked in settings where document imaging systems were used.)

Note that the data and charts below are estimates only, pointing to a potentially huge problem -- with computer-assisted Medicare fraud implications -- if the projections are reasonable.

First, from the ED physician reader, a graph showing an increase in overall ED billing levels of 2002 versus 2008:

(E/M level increases, 2002 compared to 2008. Via CMS data; see below. Click to enlarge)

This appears to indicate a great increase in upper level (level 4 and 5) billings from 2002 to 2008 in ED's with increasing advent of EHR (part of this is likely due to "paper template" charting).

A very terse summary of these billing codes, for laypeople. Claims submitted to managed care organizations and States include the emergency levels of screening and treatment. They range from CPT 99281 ("straightforward medical decision making") to CPT 99285 ("medical decision making of high complexity"). These codes reflect not only the complexity of the treatment but also the time and difficulty of making a diagnosis:

• 99281 - level 1 - very short visit - low level service, minimal charge
• 99282 - level 2 - e.g. ankle sprain
• 99283 - level 3 - limited focus history & physical exam performed, e.g. physisian evaluation with additional tests such as an x-ray.
• 99284 - level 4 - significant history & physical exam performed, significant or major injury, often admission
• 99825 - level 5 - significant time, e.g., 45 minutes to an hour spent by physician. (Note that considering that a high % of ED visits are using that code, it might be physically impossible for ED physicians to be spending this much time seeing each patient.)

Second, a spreadsheet with some estimated numbers:

(Click to enlarge)

There appears to be an absolute increase in Medicare/Medicaid payments to ER doctors due to this billing-upcoding phenomenon from 2002 to 2008 (roughly $2 billion in payments, which at 80% allowed Medicare rate represents appx. a $2.6 Billion increase in billing)

The 2006-2008 billing levels for each specialty are public info on the CMS website. The hard part was obtaining earlier numbers. These were found in 2 papers from 2004 and 2006:

Variation in Coding of Evaluation and Management (E&M) Services by Hospital Emergency Departments
Paul Shoemaker, FACHE, President and CEO
American Hospital Directory, Inc.
Jan. 2004
http://www.ahd.com/EMstudy040108.pdf


Variations and Trends in the Coding of Evaluation and Management (E&M) Services by Hospital Emergency Departments
Paul Shoemaker, FACHE
Leatrice Ford RN, BSN, CCS, Founder, ConsultCare Partners, LLC
May 2006
http://www.ahd.com/EMstudy060530.pdf

Apparently Shoemaker and Ford were able to get CMS data for 2002 and 2004 and draw the conclusions that while some hospitals and ER groups were undercoding, there was a significant level of upcoding occurring.

Note that the above projections assume the same Medicare physician allowables for 2002 and 2008; lesser values in 2002 (compared to 2008) would push the 2008 payment increases upward.

Finally, some actual ED screens from a major HIT vendor, McKesson, that "encourage" physicians to upcode.

The "coding engine" of this ED EHR system produces "recommendations" for "additional documentation" to support a higher level of coding. Such incentives themselves might be legitimate, but can create moral hazard - and the means, motive and opportunity for "playing with the coding a little bit" - if physicians' coding levels are monitored by their organizations.

(Click to enlarge. An abscess. ED EHR advising physician to "provide more documentation" in order to upcode from level 2 based on the data entered, to level 5 or maximum, via unknown algorithms. A "Correct Deficiencies" button brings up screens with selections a physician can click on to "support" the cybernetic upcoding.)

(Click to enlarge. In this example an ED physician is begin cybernetically cajoled - uh, "reminded" - to enter data to support upcoding a sprained ankle evaluation from a level 3 to a level 5.)


It should be remembered that increased ED coding levels by ED groups, often contractors to hospitals, may permit hospitals to increase their own billing charges for services rendered to the ED patients.

In summary:

• With little solid evidence of major clinical benefit, ED physicians seem to be paradoxically "asked" to utilize distracting, often mission hostile, even hateful ED EHR systems that many despise, even in face of the critical nature of an ED environment and increased risks and liabilities that can occur with distractions such as data entry. This is a seeming paradox.

• The ED physicians, after automated coding algorithms analyze the clinical data entered, may be "reminded" that if they fill in a few more details, the visit might be significantly upcoded.

• Payers such as CMS have little or no way to ascertain the veracity of the additional entered information that may result.

• The ED physicians' coding levels may be monitored by their employer(s) i.e., group or hospital compared to the "possible" coding levels suggested by the vendor software algorithms. ED EHR software makes this easy. This would create the means, motive and opportunity for upcoding, so as to prevent negative repercussions on a "low coding" doctor. (The ED physician who submitted this data claims that this is so in his organization, and in others.)
  

While the above is not a rigorous analysis, if it overall reflects the current situation with ED and other EHR's in the field in 2011, we may be seeing:

• A significant explanation for "irrational exuberance" around health IT - which would actually not be so irrational after all (just unethical) - with potentially deadly results;

• A highly systematic, major fraud participated in by major stakeholders, from software vendor to purchaser to clinical user to biller. This would represent a cybernetically-facilitated swindle of major (and growing) proportions; and

• Claims of purported savings from health IT that we hear about from pundits and government officials being a rather pathetic joke.

Perhaps CMS (and other payers as well) could perform the needed rigorous investigation using their data and resources, to either confirm or disprove these findings. That's what we, the taxpayer, pay the government for.

-- SS