Drawing Conclusions from the Record, Draws Contention

By
Original story posted on: April 2, 2014

Can coders infer or deduce from the medical record? It’s elementary, my dear physician.

During a recent Monitor Monday broadcast, I postulated that coders could draw some conclusions from medical records. In particular, I asserted that if the records contain facts that demonstrate that a patient has a particular diagnosis, but the records don’t include the actual label, I believed the coder could apply the label as long as the facts in the medical record lead inexorably to that conclusion.

(Unfortunately, the actual statement given during the broadcast wasn’t nearly that clear!) Sufficed to say, reaction has been vocal. Questions and comments have poured in. The debate is wonderful; this segment started with my acknowledgement that I was testing the hypothesis. I invited people who could challenge my conclusion to so. Thank you to all of you who did. This article will explore some of the comments and questions.

As an introduction, however, I want to note that some emails contained statements like “I am sure that I could find lots of places where the instructions contradict Mr. Glaser.” One of the goals of the “Risky Business” segment is to steer the debate away from unsupported assertions and toward an analysis of relevant texts. There is certainly a possibility that my assertion is incorrect. But while many people claimed to be aware of authorities refuting the argument, to date not one has submitted anything to that effect. I don’t mind losing an argument, but it will take more than hyperbole, asserting that I am not a trained coder, or observing that “everyone knows” the truth to convince an objective observer that coders can’t infer. Absent text from the Centers for Medicaid & Medicare Services (CMS) or a CMS-authorized body reading something like “unless a physician has used a particular term, the coder can’t code it; coders can’t use logic to select a code,” the assertion that coders can infer remains valid.

Here are some of those aforementioned comments and questions, along with my thoughts. My goal here is to determine the right answers, so I welcome additional comments and emails. My only request is that you include citations to support your positions.

Q/C: Determining the diagnosis relative to the signs and symptoms and what is known is outside of the scope of practice of a coder would essentially be allowing a coder to practice medicine. It’s a scope of practice issue, not a coding guideline.

A: I completely agree that coders cannot diagnose a patient. That comment is entirely valid. I may not have been sufficiently articulate, but the issue I was discussing is different. Can a coder draw obvious deductions from the medical record? This issue arose when coders were asserting that if the physician didn’t use the name of a particular diagnosis in the medical record, the coder could not assign the diagnosis. This was counterintuitive to me. If a medical record sufficiently describes a diagnosis without the word actually appearing, I think it would be the coder’s job to label it. Let me start with a non-medical example. If the chart says “the patient reports that her husband said x,” can someone conclude that the patient is married? The chart doesn’t state that the patient is married, but the reference to a husband should allow a reasonable person to infer that the patient is married. Which brings us to the next question:

Q/C: Coders cannot assume or infer. Can Mr. Glaser give us the citation that he believes allows coders to diagnose a patient?

A: First, the request for a citation is an excellent way to frame it. The burden, however, should be on those asserting that coders are forbidden to make an inference. In America, things are generally legal unless something prohibits it.

It is important to separate this question into distinct pieces. Again, I completely agree that coders can’t diagnose a patient. But let’s explore the claim that coders can’t assume or infer. First, it is essential to define terms. “Assumption” and “inference” are almost opposites. Oxford defines “infer” as a verb meaning to “deduce or conclude (information) from evidence and reasoning rather than from explicit statements.” Oxford defines an assumption as “a thing that is accepted as true or as certain to happen, without proof.” Inference is science; assumption is guesswork. I agree that coders shouldn’t assume. The difference between an assumption and an inference (or deduction) is data. I assert that coders can rely on data in the record to select a code.

My statement that “coders should not make assumptions” may be too categorical. For example, in Chapter 11: Complications of Pregnancy, Childbirth, and the Puerperium (630-679), the “General Rules for Obstetric Cases” includes this statement: “Should the provider document that the pregnancy is incidental to the encounter, then code V22.2 should be used in place of any Chapter 11 codes. It is the provider’s responsibility to state that the condition being treated is not affecting the pregnancy.” Note that if coders could act only on the word of a physician, this instruction would be different. It would guide the coder not to assume that the pregnancy was relevant unless the physician specifically stated that it was. Note also that the instruction was not to “query the physician to verify that the pregnancy was relevant to the treatment.” The text clearly anticipates an assumption. A coder may assume that pregnancy is relevant unless there is documentation to the contrary.

Q/C: If the physician does not state the diagnosis and the coder recognizes a disease process or comorbidity, it is our practice to query the physician and include that documentation in the medical record.

A: It is always permissible to ask the physician for additional detail. From a risk management standpoint, the safest approach is always to ask. However, there are also costs to making inquiries to a physician. Most systems will want to seek additional information from the physician only when such an inquiry is necessary. My goal is to determine when the rules require an inquiry. Some organizations will wisely choose to make inquiries more freely than others.

Q/C: What was David’s Medicare manual citation?

A: During the broadcast I mentioned the Medicare Claims Processing Manual, 100-04, Ch. 23, § 10. However, there are also relevant regulations. According to 42 C.F.R. § 424.32, use of the International Classification of Diseases, Ninth Revision, Clinical Modification, Volumes I, II, and III is a condition of Medicare payment to providers and suppliers. The regulations do not indicate that use of The Official ICD-9-CM Guidelines for Coding and Reporting is a condition of payment.

Q/C: David used the example of hypertension. One blood pressure (reading) does not mean the patient has hypertension. This is the doctor’s call.

A: I completely agree that no one blood pressure reading would allow a coder to conclude that hypertension was present. In fact, as a non-medical professional, I still have doubts that one reading would allow a physician to make that determination. But I believe that there could be a point at which the record contains the information to permit this determination. Again, one reading wouldn’t do it. One year’s worth of consistent readings might, however. (I have to confess to not knowing the definition of hypertension. If there is an agreed-upon standard then I believe that the presence of information in the record meeting that standard would allow the coder to act, even if the physician hadn’t used the word “hypertension.”)

Q/C: I was listening to Monday Monitor this morning, and I heard David Glaser’s part of the program. It is clear he does not understand or know about coding, or medicine for that matter.

All coding comes from physician documentation – not nursing, not therapy, not dietary, not pathology, not lab reports, not an administrator, not a secretary, and not coder assumption or inference. This is Coding 101.

The coding guidelines are developed by the federal government, including CMS, NCHS, DHHS, AHA, and AHIMA. I could probably find plenty of places where coders are instructed or regulated not to assume or infer a diagnosis.

A: It is true that I am not a coder and I have absolutely no medical training. The statement indicating that I don’t understand coding or medicine is a fair one. (Though is not entirely accurate to say that NCHS, AHA and/or AHIMA are part of the federal government, even if it is true that they are part of the cooperating

parties for ICD-9-CM, and CMS has given them authority over ICD coding.) Remember that in a dispute with CMS about coding, a final decision would be made by a judge, not a coder or a physician. A judge is going to look at the regulatory framework and try to determine what it requires. The regulatory framework is the key to getting to the bottom of this issue. Based on our research, my conclusion is that the available guidance contains examples where it explicitly allows inference, and it doesn’t have language specifically forbidding it. (It is impossible to cite a negative. That is why the burden must be on the party trying to prove that something is impermissible.)

Q/C: The ethical standards of AHIMA make me think that coders cannot code unless the physician or appropriate provider indicates as such in documentation, but the coder is encouraged to query the provider of things to more accurately code the issues the patient is facing. I have copied/pasted below the two elements to which I refer. This both helps the coder from accidently misinterpreting and possibly leaving the coder and facility open to fraud (allegations). After all, the coder is not the provider!

1. Assign and report only the codes and data that are clearly and consistently supported by health record documentation, in accordance with applicable code set and abstraction conventions, rules, and guidelines.

2. Query the provider (a physician or other qualified healthcare practitioner) for clarification and additional documentation prior to code assignment when there is conflicting, incomplete, or ambiguous information in the health record regarding a significant reportable condition, procedure, or other reportable data element dependent on health record documentation (e.g. a present-on-admission indicator).

A: I don’t think anything in those two bullet points is inconsistent with the idea that a coder can draw proper deductions. Asking the practitioner for additional clarification is always permissible. When the record is unclear or inconsistent, the coder absolutely should seek clarification. But if the record is sufficient to allow the coder to reach a conclusion, he or she can.

In closing, I think the difference between “assumption” and “inference” is important. I need to develop a few real-world examples to help illustrate how I see those two terms as different. If you have a good case example, I would appreciate it if you would share it.

The truth is well-served by vigorous debate.

About the Author

David Glaser is a shareholder in Fredrikson & Byron's Health Law Group and helped establish its Health Care Fraud & Compliance Group. David helps healthcare entities negotiate the maze of healthcare regulations, providing advice about risk management, reimbursement and business planning issues. He has considerable experience in healthcare regulation and litigation, including compliance, criminal and civil fraud investigations, and reimbursement disputes.

Contact the Author

dglaser@fredlaw.com

To comment on this article go to editor@racmonitor.com

David M. Glaser, Esq.

David M. Glaser, Esq., is a shareholder in Fredrikson & Byron’s Health Law Group. David helps clinics, hospitals, and other healthcare entities negotiate the maze of healthcare regulations, providing advice about risk management, reimbursement, and business planning issues. He has considerable experience in healthcare regulation and litigation, including compliance, criminal and civil fraud investigations, and reimbursement disputes. David’s goal is to explain the government’s enforcement position and to analyze whether the law supports this position. David is a popular panelist on Monitor Mondays and a member of the RACmonitor editorial board.

 

This email address is being protected from spambots. You need JavaScript enabled to view it.