Definition of Diagnosis Error

Bruno, Michael mbruno at PENNSTATEHEALTH.PSU.EDU
Mon Feb 6 15:46:49 UTC 2017

Hi Tom,

Thanks for your message.  Yes, I think these are all errors, and potentially very bad ones, as you point out.  In your appy example, the 3rd year surgery resident was clearly in the wrong – and I hope that both surgery residents learned something from their attending’s behavior (how he was able to make a correct diagnosis and do the appendectomy without needing a CT first).  Over-utilization of CT scans for PE is another tough issue, and one that our group took on (unsuccessfully, as it turned out, see the attached PDF).

I don’t think it is all driven by financial reward under fee-for-service payment, but instead I think that this happens primarily because of physicians’ difficulty with the concept of uncertainty in general and of Bayesian “pretest probability” of disease in particular.  In the case you described, the pretest probability of appendicitis was so high that an imaging study wouldn’t actually have changed the diagnosis, even if it had been (surprisingly) negative.

I often say that the role of the radiologist is not actually to “make diagnoses,” but rather our role is to reduce the level of clinical uncertainty to the point where a doctor feels comfortable enough to take (or forgo) action.  When the level of uncertainty is already low enough for that, there is probably no role for imaging.

I don’t think this sort of thing will automatically stop when more of our payments are “value” based – responding to cash incentives is a rational behavior and this problem is driven by human irrationality.

We actually had a similar case here, in which an absolutely classic appy patient was sent for CT (needlessly, in my opinion) and the ER doc managing the patient mistakenly came to believe that the CT result was negative (it was, in fact, grossly positive—supporting his clinical diagnosis).  The doc in this case decided to disbelieve his own eyes, his own history and his own physical findings and discharge the patient home on the basis of the “negative CT report!”  Luckily the error was discovered in time and the patient was brought back for his appendectomy within a few hours.

Uncertainty is the reality of our lives in medicine, and imaging doesn’t erase that; to the contrary, all tests have a false-negative rate and a false-positive rate.  That’s what ROC analysis is all about, right?  Have we become like the automobile drivers who can’t find their way home from two blocks away without resorting to their GPS?


From: Tom Benzoni [mailto:benzonit at]
Sent: Monday, February 06, 2017 10:21 AM
To: Society to Improve Diagnosis in Medicine; Bruno, Michael
Subject: Re: [IMPROVEDX] Definition of Diagnosis Error

Dr. Bruno:
You're stirring a thought; coffee is likely helping.
How much overdiagnosis is cultural, how we're training the "kids?"
I come from before CT; we did a pretty good job on appy dx.
So I have a 23 y o male a few months ago with such classic signs and symptoms I thought he'd read Schwartz Textbook of Surgery.
I called the surgeon; we had a long relationship and he trusted my judgment.
I called his resident; the first year did a good job, even having appy in the top 5.
Then the 3rd year surgery resident came down and reamed him a new one for not having a CT; "Don't you know it's standard of care to have a CT?! You can't diagnose appendicitis without a CT!!" Poor kid was devastated and he learned.
(The surgeon came down and took out a bad appy without a CT.)
My colleagues order d-dimer as reflex; we have a yield of CT(+) PE (not symptomatic as in RV dysfunction or change in vitals) ~5%. These end up on NOACs, likely causing more deaths from bleeds and radiation than lives saved. Yet their efforts are rewarded financially and so flourish.
Are these errors? They're treated as diagnoses, and positively reinforced.


On Mon, Feb 6, 2017 at 8:37 AM, Bruno, Michael <mbruno at<mailto:mbruno at>> wrote:
Good points, Michael.
Our profession is impacted by several drivers leading to overdiagnosis, which I believe is a much bigger problem in terms of patient harm than is generally acknowledged (even in this forum).  I applaud you, Bob and Tom Benzoni, Albert Yu and others for bringing it up.
Certainly one of those drivers is defensive medicine, which is itself greatly underestimated in terms of its importance.  Dr. Lenny Berlin has pointed out that defensive medicine, in which an excess of diagnostic tests are ordered to reduce physician anxiety rather than for the benefit of the patient, may have had its genesis in the malpractice crisis of the 1970’s but now has taken on a life of its own, passed along in the GME process to successive generations of physicians.  All of us who struggle at our home institutions – as I do – to get clinicians to move more toward evidence-based utilization of imaging has run into this, and it is a formidable problem.
Another driver of overdiagnosis is, of course, the emphasis on screening of healthy populations – which is done in the hope that early detection will lead to better patient outcomes.  The two biggest examples in Radiology, of course, are mammography and lung cancer screening.  While the evidence to support the benefits of mass screening is relatively weak, the false-positive rate is fairly high.  Since pathology is not an exact science by any means, this becomes a perfect formula for overdiagnosis leading to overtreatment.  Even the biopsy itself can be classified as “patient harm” when the false-positive rate is very high relative to the true-positive rate of screening.  A very nice, recent book (2016) on the subject (in case anyone is interested) is Steven Hatch’s Snowball in a Blizzard, although if you have read Gilbert Welch’s book, Overdiagnosed, you have pretty much covered the same material.
Overutilization of imaging for whatever reason (including for valid reasons, such as patient demand, or to speed discharge from the ED so that more patients can be served, or even because of lack of physician confidence in their own physical diagnosis, etc) is another major driver of overdiagnosis.  This is simply because many disease entities cannot reliably be discriminated from each other by imaging and because a larger volume of imaging studies being performed will produce a proportionately larger volume of false positives—all of which must be worked up in one way or another.  Unnecessary imaging is truly a Pandora’s box for patients, as studies will often raise more questions for a patient than they answer.
Moving physicians toward evidence-based utilization of medical imaging, based on objective clinical criteria whenever possible – and perhaps having a more clear-eyed (skeptical?) approach to mass population screening with imaging would carry us a long way toward reducing the currently severe problem of overdiagnosis, in my opinion.
One approach to this problem is to provide electronic “decision support” to clinicians that can be accessed seamlessly at the time a physician encounters a patient, in order to help them weigh the evidence and decide whether an imaging study would be appropriate.  The nonprofit American College of Radiology (ACR) has invested quite a bit of volunteer physician effort and other resources to develop some guidelines, the ACR Appropriateness Criteria, and lately to develop deployable decision-support tools, such as the ACR’s R-Scan tool.  There is a need for more research in this area, both to develop better evidence and more guidelines where patient outcome is considered, and also to understand how better to deploy decision support so that it is accepted by physicians and is ultimately effective.  We are hoping to convene a research development and consensus conference this Fall in Hershey and hope that many SIDM members will attend (if we can get the funding to hold the meeting).
All the best,
[cid:image004.png at 01D112FF.F77F98B0]
Michael A. Bruno, M.S., M.D., F.A.C.R.
Professor of Radiology & Medicine
Vice Chair for Quality & Patient Safety
Chief, Division of Emergency Radiology
Penn State Milton S. Hershey Medical Center
• (717) 531-8703<tel:(717)%20531-8703>  |   6 (717) 531-5737<tel:(717)%20531-5737>
• mbruno at<mailto:mbruno at>
From: Kohn, Michael [mailto:Michael.Kohn at UCSF.EDU<mailto:Michael.Kohn at UCSF.EDU>]
Sent: Sunday, February 05, 2017 1:30 PM

Subject: Re: [IMPROVEDX] Definition of Diagnosis Error
Dear Colleagues,

As per my prior post, I see two types of diagnostic error: 1) failing to provide effective treatment for something the patient has, and 2) treating unnecessarily for something the patient doesn't have.  If you believe tPA is effective treatment for acute stroke, then failing to provide it to a patient with a stroke is the first type of error, and providing it to a patient with a stroke mimic, such as a complicated migraine, is the second type of error.  Overdiagnosis is the second type of error.  Giving the wrong treatment combines (1) failing to treat with (2) treating unnecessarily.

One correspondent correctly pointed out that these are actually treatment errors.  I am focused on incorrect decisions, not incorrect naming of a patient's illness.  With my narrower view, failing to apply the correct name to a benign, self-limited condition is not an error so long as you don't provide unnecessary and harmful treatment.  Also, failing to distinguish between two illnesses with the same effective treatment is not an error unless you fail to provide that treatment.  I am not one of those physicians who tries to please patients by applying a Greek name to their benign problem (cephalgia for headache), but this isn't necessarily an error.  Failing to identify an aneurysmal subarachnoid hemorrhage is an error because prompt referral for coiling can save the patient's life.

One other point: all of this refers to patients who have symptoms, i.e., they feel sick.  Screening of patients with no known symptoms of disease is a different matter.  That is primarily though not exclusively what H. Gilbert Welch was referring to in "Overdiagnosed: Making People Sick in the Pursuit of Health".



Michael A. Kohn, MD, MPP

Chairman, Emergency Department

Mills-Peninsula Medical Center

Associate Professor

UCSF Epidemiology and Biostatistics

(Email created using voice recognition.  Please excuse transcription errors.)

From: Joe Graedon [jgraedon at GMAIL.COM<mailto:jgraedon at GMAIL.COM>]
Sent: Sunday, February 05, 2017 5:32 AM
Subject: Re: [IMPROVEDX] Definition of Diagnosis Error
I think the IOM definition is comprehensive and helpful.

Sent from my iPad

On Jan 18, 2017, at 8:54 AM, Bob Latino <blatino at RELIABILITY.COM<mailto:blatino at RELIABILITY.COM>> wrote:
Is over-diagnosis considered a diagnostic error?
Robert J. Latino, CEO
Reliability Center, Inc.
blatino at<mailto:blatino at><>

From: Bob Latino [mailto:blatino at RELIABILITY.COM]
Sent: Wednesday, January 18, 2017 6:23 AM
Subject: [IMPROVEDX] Definition of Diagnosis Error

Is this IOM Definition of Diagnosis Error an accepted definition by SIDM?
What is Diagnostic Error?
The Institute of Medicine recently defined diagnostic error as the failure to (a) establish an accurate and timely explanation of the patient’s health problem(s) or (b) communicate that explanation to the patient. Simply put, these are diagnoses that are missed altogether, wrong, or should have been made much earlier.

These categories overlap, but examples help illustrate some differences:
A missed diagnosis refers to a patient whose medical complaints are never explained. Many patients with chronic fatigue, or chronic pain fall into this category, as well as patients with more specific complaints that are never accurately diagnosed.

A wrong diagnosis occurs, for example, if a patient truly having a heart attack is told their pain is from acid indigestion. The original diagnosis is found to be incorrect because the true cause is discovered later.

A delayed diagnosis refers to a case where the diagnosis should have been made earlier. Delayed diagnosis of cancer is by far the leading entity in this category. A major problem in this regard is that there are very few good guidelines on making a timely diagnosis, and many illnesses aren’t suspected until symptoms persist, or worsen.

 Robert J. Latino, CEO
Reliability Center, Inc.
blatino at<mailto:blatino at><>



To unsubscribe from IMPROVEDX: click the following link:

Visit the searchable archives or adjust your subscription at:

Moderator:David Meyers, Board Member, Society for Improving Diagnosis in Medicine

To learn more about SIDM visit:

To unsubscribe from the IMPROVEDX:
or click the following link: IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG


For additional information and subscription commands, visit:

http://LIST.IMPROVEDIAGNOSIS.ORG/ (with your password)

Visit the searchable archives or adjust your subscription at:

Moderator: David Meyers, Board Member, Society to Improve Diagnosis in Medicine

To unsubscribe from the IMPROVEDX list, click the following link:<br>
<a href="" target="_blank"></a>

HTML Version:
URL: <../attachments/20170206/d9aa26fc/attachment.html> ATTACHMENT:
Name: image001.png Type: image/png Size: 2281 bytes Desc: image001.png URL: <../attachments/20170206/d9aa26fc/attachment.png> ATTACHMENT:
Name: image002.png Type: image/png Size: 14098 bytes Desc: image002.png URL: <../attachments/20170206/d9aa26fc/attachment-0001.png> ATTACHMENT:
Name: Geeting et al AJR.pdf Type: application/pdf Size: 939599 bytes Desc: Geeting et al AJR.pdf URL: <../attachments/20170206/d9aa26fc/attachment.pdf>

More information about the Test mailing list