Editorial: Too much of a good thing?

pdf path

In 2002, I wrote an editorial titled “Garbage in, garbage out,” emphasizing that the provision of appropriate clinical indications on study requests could improve the accuracy of radiologists’ interpretations.1 In 2005, I wrote an editorial called “R/O disease.” Basically, my point in that piece was that computed tomography (CT) studies in the emergency department were being increasingly performed, but quite often not justified by adequate clinical indications, at least in what was being provided to the interpreting radiologist. Though a phone call to the requesting physician might retrieve an appropriate indication, often the CT was a “witch hunt” for pathology.2

I wrote another editorial in 2009 titled, “Clinical indication: Patient tripped over wagon walking to Walmart.” My point here was to emphasize what nonsense such information was versus having a valid clinical sign or symptom provided as the study indication. It was ludicrous to get all this irrelevant stuff.3 Well, it appears my dream of having appropriate signs and symptoms to justify imaging studies will finally be realized, but the possible resolution looks more like a nightmare than a dream come true.

For most of my 30-year radiology career, I have begged referring physicians to provide some information in the indication part of their imaging requests to justify performing the examination; that is, a clinical sign or symptom. These pleas went largely ignored. While the problem is not limited to radiologists working in the emergency care arena, that is the subspecialty I am most familiar with, and which, at my institution, is documented as the weakest area for obtaining appropriate indications for examinations.4

Now, tracking on a quick time horizon, we have the ICD-10 coding system, which will become the new standard for healthcare electronic transactions, including of course, reimbursement for services. The current ICD-9 system uses 3 to 5 alphanumeric items to create about18,000 codes, while ICD-10 has 7 alphanumeric characters creating about 72,000 codes.5 The codes are quite specific, allowing very detailed designation of imaging pathology. Unfortunately, the detail that needs to be provided for a complete indication borders on the absurd.

The excruciating level of specifics required for some codes is well described in a wonderful article by Dr. Ronald Bucci.5 I highly recommend this article to any practicing radiologist or anyone considering becoming a radiologist. Dr. Bucci mentions the example of someone falling while ice skating. In the current ICD-9 system, the imaging study request would say “fell.” Occasionally you might also get scalp swelling or altered mental status—an actual clinical sign and symptom. If you were going to apply the expectations of ICD-10 to this scenario, the indication would include multiple codes, such as concussion (the occurrence), ice (the external cause), skating (the activity), location of the ice-skating rink, and acute versus chronic injury. How does all this really differ from “concussion”? Actually, I believe having some specific clinical signs like diplopia or disoriented to time and place might actually be worth the extra effort. Requiring all this stuff just “blows my mind” (I guess there is a new ICD-10 code for this condition as well). Some very specific code examples given by Dr. Bucci include T71.233 “asphyxiation due to being trapped in a discarded refrigerator, assault” or V04.09 “pedestrian on snow skis injured in collision with heavy transport vehicle or bus in non-traffic accident.” My favorite one mentioned is W61.42 “struck by turkey” versus W61.43“pecked by turkey.”1 Isn’t a peck a form of strike? What about a sharp-beaked versus a dull-beaked turkey? What if it’s Thanksgiving and the turkey has all the rights in the world to defend itself? Who’s really assaulting whom?

For non-emergency in-patients, there is at least something of a clinical track record established from previous admissions or initial information that is required for the current admission, which can be searched for relatively quickly using an electronic medical record (EMR) for specific, clinically relevant information. That is, assuming your institution has an EMR. In the emergency care setting, there is essentially no EMR created until after the patient encounter. If there is any EMR information available (almost never present in my institution when the imaging work-up is performed), it is extremely unlikely to provide the data needed to get anywhere near the type and quantity of information required to fully code by ICD-10. Thus, in the ED setting, the imaging study request is the primary or only contemporaneous source of this information. Given that currently, one cannot often even get a single clinical sign or symptom, how can we ever hope to be able to obtain the level of detail expected for complete ICD-10 coding?

No doubt a lot of problems with ICD-9 will be addressed by this update. The expressed intent of this new coding system is to improve just about every measurement/assessment of any sort a hospital could want to make, ranging from healthcare policy to research to resource utilization ad infinitum. I assume there is absolutely no incentive in this new methodology to decrease reimbursements by requiring lots of information that, in many cases, is irrelevant to the patient’s clinical condition. I speculate most radiologists working in the ED will spend 20minutes per case calling the referring physician or interviewing the patient and 1 minute interpreting the study. That is not a great prescription for providing better healthcare.

Now, using ICD-10, radiologists will be heavily dependent on referring clinicians to provide this meticulous level of information. The less detail provided, the less specific the coding, and the less the reimbursement provided, down to zero. What is abundantly clear is that without the staunch support of our referring physicians in providing all this data and hospital administrators demanding it, reimbursements, at least from ED radiology, are heading down. I hope this area is not a large part of your practice. I realize that other radiology specialties, as well the other diagnostic and direct patient care clinical services, will also have their respective sources of difficulty in coping with this system, but given my experience in emergency radiology, the coming difficulties in coding for emergency imaging services in this new system clearly stand out as problematic.

I really hope that the time and effort that will be needed to put into adapting to this coding system will provide something useful in the end. It may require a long time to get there with no small amount of pain along the road.


  1. Mirvis SE. Garbage in, garbage out. Applied Radiol. 2002;31:5.
  2. Mirvis SE. R/O disease. Applied Radiol. 2005;34:6.
  3. Mirvis, SE. Clinical indication: Patient tripped over red wagon walking to Walmart. Applied Radiol. 2009;38;6.
  4. Unpublished data, Mirvis, SE, 2012.
  5. Bucci RV. ICD-10 and its impact on radiology. Radiology Today. 2011;12:10.
Back To Top

Editorial: Too much of a good thing?.  Appl Radiol. 

By Stuart E. Mirvis, MD, FACR| February 06, 2014

About the Author

Stuart E. Mirvis, MD, FACR

Stuart E. Mirvis, MD, FACR

Dr. Mirvis is the Editor-in-Chief of this journal and a Professor of Radiology, Diagnostic Imaging Department, University of Maryland School of Medicine, Baltimore, MD.

Copyright © Anderson Publishing 2020