Two new psychology papers on diagnosis

Neal Dawson ndawson at METROHEALTH.ORG
Sat Dec 17 17:21:18 UTC 2016

Three important aspects of CPCs are not discussed by Dr. Jain.
1) CPCs are selected, often by a pathologist or chief resident to be
challenging - not to be representative of what clinicians are likely to
see in practice. Every one involved in CPCs is well aware that this is
how the game is played. (As chief resident, I once gave an ordinary case
of lung cancer to a visiting professor (nationally known pulmonologist)
for a CPC. His discussion included lung cancer early on but then went on
to discuss much more esoteric possibilities, one of which was his final
diagnosis - an anecdote with the expected course given how and why CPCs
are selected.)
2) CPCs are prone to the adverse effects of hindsight bias (See Dawson
et al, Med Decis Making 1988;8:259) and thus may produce thought
processes that may be misleading when applied to daily diagnostic
3) As the citations in Dr. Rottman's papers document, ordinary practice
often does not provide a method for accurate feedback about many cases
seen in daily practice. CPCs, as usually performed, do not provide
adequate feedback regarding thought processes appropriate to the context
of usual practice.
In addition, the prior probabilities for any case quickly become
conditional probabilities (i.e. they are context dependent) once the
history, physical exam and routine data become available.
Neal V Dawson, MD

>>> "Jain, Bimal P.,M.D." <BJAIN at PARTNERS.ORG> 12/16/16 10:12 AM >>>
Ben Rottman's two papers raise interesting and important issues about
role of probability in diagnosis. We briefly discuss findings in our two
recent papers on method of diagnosis.
We investigated the method of diagnosis employed by experienced
clinicians for diagnosis in real patients in 50 CPCs and found it to 
consist of the following steps:
1.A number of diseases are suspected from the presentation.
2.Each disease is formulated as a diagnostic hypothesis that is as an
assumption which may or may not be correct.
3. Each suspected disease is evaluated for its presence in the given
patient in terms of its likelihood given findings in the patient.
4. The disease with the greatest likelihood is diagnosed to be present.
With this method, a correct diagnosis is made in 49 out of 50 CPCs.
We find that prior probability is not even mentioned in any CPC let
alone employed as prior evidence for a disease in any CPC.
In a subsequent paper, we discuss why probability does not play any
significant role in diagnosis in a CPC.
A probability, being a frequency in a population represents evidence
only in a large group and therefore it plays a key role in inference in
life insurance business for example in which the aim is accuracy in the
long run in a large group. Errors in some individual persons are
expected and tolerated in this business.
In diagnosis, on the other hand, our aim is correct determination of a
disease in every individual patient in whom a probability represents not
evidence but chance. The only role of prior probability in diagnosis, we
believe is in setting the order in which various suspected diseases are
The probabilistic or Bayesian method with its notion of prior
probability as prior evidence may encourage failure to suspect a disease
with an atypical presentation leading to diagnostic error.
We believe the prescription of the probabilistic method for diagnosis
should be re-evaluated as this has important implications for teaching
diagnosis to novice physicians and for developing computer systems which
aid physicians in diagnosis.

Bimal P Jain MD
Northshore Medical Center
Lynn MA 01904.

-----Original Message-----
From: Ben Rottman [mailto:rottman at PITT.EDU] 
Sent: Thursday, December 15, 2016 11:14 AM
Subject: [IMPROVEDX] Two new psychology papers on diagnosis

Dear ImproveDX Readers,

I recently published two new psychology papers on diagnosis that might
interest some of you. One paper investigates the use of base rate
(disease prevalence) knowledge when forming a preliother investigates how well physicians make post-test judgments in
reference to their own pre-test judgments and beliefs about the
sensitivity and specificity of the test. I believe that these papers
provide some of the strongest evidence that physicians actually can be
fairly rational (in terms of Bayes’ rule) when forming diagnoses.

The main limitation of these papers is that they both involve reasoning
about vignette cases, not real patients in-person. The main strength is
that they investigate physicians’ use of their own beliefs about disease
prevalence, which has rarely been done in the past, and probably
explains why physicians often look so irrational in prior studies.

Paper 1 on Preliminary Diagnoses: Open Access Link
Paper 2 on pre-test post-test diagnostic updating:
or if you have trouble downloading it you can get it here:

Feel free to share your thoughts on the listserv or to me via email.

Ben Rottman
Assistant Professor Psychology, University of Pittsburgh
rottman at

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

To unsubscribe from the IMPROVEDX list, click the following link:<br> <a

The information in this e-mail is intended only for the person to whom
it is
addressed. If you believe this e-mail was sent to you in error and the
contains patient information, please contact the Partners Compliance
HelpLine at . If the e-mail was sent to you
in error
but does not contain patient information, please contact the sender and
dispose of the e-mail.

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

To unsubscribe from the IMPROVEDX list, click the following link:<br>

 MetroHealth’s Mission: Leading the way to a healthier you and a
healthier community through service, teaching, discovery and teamwork.  
This email and all attachments that may have been included are intended
only for the use of the party to whom/which the email is addressed and
may contain information that is privileged, confidential, or exempt from
disclosure under applicable law. If you are not the addressee or the
employee or agent of the intended recipient, you are hereby notified
that you are strictly prohibited from printing, storing, disseminating,
distributing, or copying this communication. If you have received this
notification in error, please contact 216-778-7751. For a copy of our Notice of Privacy Practices, please

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

HTML Version:
URL: <../attachments/20161217/28e4e030/attachment.html> ATTACHMENT:
Name: IMAGE.jpg Type: image/jpg Size: 555 bytes Desc: JPEG image URL: <../attachments/20161217/28e4e030/attachment.jpg> ATTACHMENT:
Name: IMAGE.jpg Type: image/jpg Size: 565 bytes Desc: JPEG image URL: <../attachments/20161217/28e4e030/attachment-0001.jpg> ATTACHMENT:
Name: IMAGE.jpg Type: image/jpg Size: 592 bytes Desc: JPEG image URL: <../attachments/20161217/28e4e030/attachment-0002.jpg> ATTACHMENT:
Name: IMAGE.jpg Type: image/jpg Size: 578 bytes Desc: JPEG image URL: <../attachments/20161217/28e4e030/attachment-0003.jpg> ATTACHMENT:
Name: IMAGE.png Type: image/png Size: 499 bytes Desc: Portable Network Graphics Format URL: <../attachments/20161217/28e4e030/attachment.png>

More information about the Test mailing list