Hospitals are facing Medicare recoupment when auditors determine they billed for MS-DRGs based on documentation of physician diagnoses that turned out to be wrong. In a recent Medicare compliance review, the HHS Office of Inspector General (OIG) called this a coding error. It’s sort of a Catch 22: Coding is based on physician documentation, but when it turns out the diagnosis was unreliable, it means the code was wrong as well.

This tension between clinicians and coding, which perpetuates hospital billing errors, may get worse next year with the implementation of ICD-10 diagnosis and procedure codes because they require greater documentation and coding specificity.

“Coders are faced with two opposing rules: They have to code what the doctor documents but they aren’t supposed to code anything they think is unclear or contradictory to other information in the chart,” says Andrew Rothschild, M.D., a director with FTI Healthcare in Austin. If it’s obvious to the coder that something is amiss with the documentation, they can query physicians for more information. But because coders aren’t clinicians, they are cautioned not to question clinical opinions or ask leading questions, which the government may perceive as an attempt to push the physician down the path to a higher-paying MS-DRG.

“You don’t question the physicians clinically when it is clear what their opinion is. But when they contradict themselves or when discrepant information is in the chart, you are supposed to ask physicians for a clinical opinion. That is not the same thing as questioning the physician’s clinical judgment,” he says. “This is even more true today than it was five years ago because RACs and other auditors say they are looking for support in the chart to justify the diagnosis. There has to be something within the chart that makes the diagnosis understandable. It’s not good clinical practice to make diagnoses out of thin air and you shouldn’t expect to get paid if you’re doing that.”

When a documented diagnosis doesn’t seem consistent with the chart, coders are instructed to query for clarification before coding it, Rothschild says. The query can state, for example, that a patient admitted for a migraine was noted to also have sepsis at discharge, and there were negative clinical indicators and no prior mention of infectious processes and then request confirmation and support for the diagnosis. “It’s important, both clinically and to address compliance guidance,” according to Rothschild.

The coder/clinical dilemma was on display in the recent Medicare compliance review of 360-bed Palmetto General Hospital in Hialeah, Fla. OIG audited 170 inpatient claims and 17 outpatient claims that were deemed at risk of billing errors. Of them, 14 inpatient claims and one outpatient claim had billing errors, OIG said. The overpayment for these claims, which had 2009 to 2010 dates of service, was $124,566.

Two of the inpatient risk areas with errors were claims paid in excess of charges and high-severity MS-DRGs, which OIG blamed on “human error.” The hospital billed Medicare for MS-DRGs that were based on the physician’s diagnosis but not supported by clinical documentation, OIG said. Either the hospital coders relied on the physician’s written diagnosis and/or “the coder did not re-review the patient’s account when the discharge summary became available.”

Hospital Blames Documentation, Not Coding

The hospital says these are not coding errors. Instead, there are issues with physician documentation, according to a written response from Palmetto Compliance Officer Melissa Paper. Coders just coded what the physicians documented and had no grounds to question it.

In one case, the coder had support for coding a diagnosis of gastrointestinal bleed based on the physician entering “Upper GI bleed” in the progress note, the compliance officer said. Additional work-up and documentation couldn’t be completed because the patient left against medical advice. On later review, the chief medical officer for the hospital concluded “the patient likely did not have an Upper GI bleed.“ In another case, the physician documented “?COPD, ?CHF” — which means they were possible diagnoses at the time they were documented — as well as acute and chronic respiratory failure. Because coders are allowed to select any of these diagnoses if each qualifies as a principal diagnosis, the coder chose CHF. But the chief medical officer later determined that acute and chronic respiratory failure was the principal diagnosis.

The same problem happened in another risk area, inpatient claims with high severity level diagnosis-related group codes (MCC/CC). OIG said four of the 40 sampled claims were billed with the wrong DRG codes. In three cases, the physicians’ documentation of “acute renal failure” did not hold up under scrutiny. The coders based their coding selection on the information they had, the compliance officer stated.

The compliance officer attributed these errors to “physicians documenting an unreliable diagnosis.” Coders are limited in what they can do when physicians write the wrong diagnosis or fail to support it in the documentation. “Coders do not have the clinical training to question a physician’s diagnostic statement,” the compliance officer contends. They can, however, submit queries to physicians to clear up confusion about “a significant reportable condition or procedure or other reportable data element dependent on health record documentation,” according to the 2008 practice brief on queries developed by the American Health Information Management Association (AHIMA).

Based on AHIMA’s practice brief, Palmetto’s compliance officer said, coders should query physicians only when documentation is “not legible, complete, clear, consistent and/or precise.” That means “a coder cannot make a determination as to whether a physician’s diagnostic statement is reliable or not.”

In fact, the compliance officer used CMS’s own words in its response to OIG. CMS noted in the 2011 RAC statement of work that “clinical validation is beyond the scope of DRG (coding) validation, and the skills of a certified coder. This type of review can only be performed by a clinician or may be performed by a clinician with approved coding credentials.”

Hospitals May Be Too Strict With Queries

But Rothschild says that hospitals are doing themselves a disservice if they forbid coders to query physicians when documentation is confusing or potentially unreliable based on other documentation. If the physician documents the removal of a toenail on a lifelong bilateral amputee, “the coder can’t just code it without asking where the physician got the toenail,” he says. “It’s like putting your fingers in your ears and your head in the sand and ignoring the other stuff in the chart.” Pointing out an apparent or potential inconsistency and asking the clinician to interpret it is different from questioning the reliability of the diagnosis, Rothschild says.

Joyce Leppo, a certified inpatient coder at Gettysburg Hospital in Gettysburg, Pa., says she walks a fine line every day to get the information needed to code charts accurately. “I am limited in what I can do” within the constraints of her profession. For example, in a chart she coded last week, the diagnosis of acute myocardial infarction “was plastered all over” by the attending physician, even though he hadn’t documented elevated troponin levels or chest pain, and the cardiologist disagreed. “I can’t query the doctor and say ‘did you really mean it was an acute MI’ because the rules for coders say you can’t question physician documentation,” Leppo says. “Ideally, I could say, ‘I am uncomfortable with the diagnosis on this chart. The clinical evidence and documentation does not support it.’” But that wouldn’t go over well. “Even trying to form the query in my mind — I can’t do that. Unless AHIMA tells me I can query like that, that we are allowed to query on the basis of that,” she won’t do it. At best, Leppo can ask if there is a more definitive diagnosis that would indicate the severity of illness. If her hospital had a physician documentation review committee, she would refer charts like this for peer review.

There’s a good chance, however, that the physician was diagnosing acute MI despite the negative clinical indicators, Rothschild says. The patient may have had ischemic chest pain radiating down the left arm that was relieved after nitroglycerine. Physicians are allowed to differ on how to diagnose conditions, and that goes beyond the purview of coding and clinical documentation improvement, he says. But if there are extreme cases of unsupported diagnoses or trends along those lines, hospitals should have a plan of action. It may include: (1) involving the CDI physician adviser, (2) incorporating the issue into clinical education, (3) referring it to the clinical quality team, (4) encouraging the department chief to help reinforce the importance of inter-physician diagnostic consistency, and (5) peer review in severe cases.

But it’s also true that coders have more room to maneuver than they think when the query is designed to clarify a diagnosis, Rothschild says. There are issues in the chart that are pretty obvious and logical and only require basic clinical knowledge, he says. The coder can write a query that goes something like this: “The patient was admitted for ‘black stools, suspected GI bleed.’ Per gastroenterology: ‘Pepto Bismol caused black stools. Endoscope normal. All Guiac tests have been negative for occult blood.’ The discharge summary specifies that initial impression was a suspected GiB, but I’m uncertain if that diagnosis was upheld or revised during the admission. In order for me to best represent the clinical picture, please document your final impression as to the final diagnosis that caused the patient’s presenting symptoms and resulted in the admission.”

Rothschild explains that the coder is just restating relevant facts and asking for a summary diagnosis or more specific diagnosis and there’s nothing wrong with that, “as long as queries are written compliantly.”

© 2012 by Atlantic Information Services, Inc. All Rights Reserved.

Get instant compliance news! Follow AISCompliance on Twitter or “Like” AISHealth on Facebook.