Skip to content Skip to navigation

Part 2: Blind Spots Can Kill Patients

January 12, 2011
by Joe Bormel
| Reprints
Cognitive Errors–The Conclusions that Lead to Unintended Consequences


In part one of this series , I introduced the widely accepted premise that delivery of complete, accurate, up-to-date, and relevant information to care providers, surpassing today’s norms by using information technology, would lead to better care. Care is complex, of course. It’s not simply the challenge of getting a patient’s prior medical information to the clinician, although this is clearly a challenge. It’s the absence, for instance, of good, up-to-date problem lists and many disconnects in today’s medication reconciliation systems that create dangerous blind spots. Beyond the patient’s specific information, medical knowledge is both incomplete and incompletely disseminated. So, how much should we expect HCIT to improve the care we actually deliver?

Fortunately for all of us, several researchers, including Dr. Mark Graber, have studied exactly this issue in the domain of Internal Medicine. Writing with collaborators in 2005 for the “Archives of Internal Medicine, Diagnostic Error in Internal Medicine,” (full text here) he stated, “The goal of this study was to determine the relative contribution of system-related and cognitive components to diagnostic error, and to develop a comprehensive working taxonomy.” In 93 percent of cases reviewed, fault was identified in system-related and cognitive categories. Over 500 factors in these cases were noted to contribute. System-related factors contributed in 65 percent of the cases; cognitive factors in 74 percent. “The most common cognitive problems involved faulty synthesis.”

Premature closure, i.e. the failure to continue considering reasonable alternatives after an initial diagnosis was reached, was the single most common cause.

The punch line for me was, “Faulty or inadequate knowledge was uncommon.”

Graber has gone on since 2005 to describe the errors that contribute to the majority of cognitive errors, and elaborate the implications for medical education, as well as the structure of healthcare delivery systems. Patient safety and medical diagnostics error literature have conveyed that more than 30 biases and fallacies lead all of us, including doctors, to take correct information and come to incorrect conclusions. (Complete list with more background in Dennis Boyle's 2010 article here) As we roll-out HCIT with the goal of improving care, we need to evolve to not just delivering the capacity to address Meaningful Use; we need to factor into the designs of our care delivery environments the practices that will have the most impact on improving care. HCIT plays an important role, and cannot be blind to the cognitive realities highlighted by Graber, Berner, Gladwell, Boyle, and many others.

Cognitive Error Type

Description

Improves with HCIT?

Aggregate bias:

 

The tendency for physicians to believe that aggregated data, such as those used to develop clinical practice guidelines, do not apply to their own individual patients.

 

Not directly.

 

Anchoring:

 

The tendency to rely too heavily on one trait or piece of information when making decisions.

 

Not directly.

 

Ascertainment bias:

 

Occurs when a physician’s thinking is shaped by prior expectations, stereotypes, and biases.

 

Not directly.

 

Availability:

 

The tendency to assign a probability to a disease according to vividness of memory.

 

Not directly.

 

Base rate neglect:

 

The tendency to base judgments on specifics, ignoring general statistical information.

 

Not directly.

 

Commission bias:

 

Pages

Topics

Comments

Joe,
Thank you for this FASCINATING post. I guess one very basic question I would like to ask you is, if it's true that healthcare IT doesn't DIRECTLY improve physicians' approaches to their tendencies toward certain cognitive error types, does that mean that some of the foundational thought behind some core clinical information technology is off-base? Or does it just highlight some of the challenges inherent in using clinical IT to facilitate better care processes? I'd love your thoughts.

Doc Benjamin,

Thanks for your kind words and encouragement.

Yes, there are many, many more recent references that are in subsequent parts of this series. I'd like to point out that the Dennis Boyle article I cited above and provided the link for was published in November of 2010.

I have been corresponding with Dr Graber and other internationally recognized experts in a group email conversation. I don't say that to be boastful, but to call out the critical importance of conversations that are virtual. This blog is another such example. In any event, it has been Graber's evolution in thinking about dimensions of the solution to the cognitive predicament that you'll want to stay tuned for.

Regarding the trust dimension and the role of incentives like P4P, that's a topic that I've blogged about before. Look for "Fast Learning (link)", Posted on: 3.12.2010 and the discussion of Amy Edmondson's perspective on learning in organizations.  That one looks at trust between providers and healthcare delivery organizations.  On P4P, take a look at "What do Wall Street Bonuses and HCIT Incentives Have in common? (link) --- Does "Pay-for-Performance" send the same wrong signals?" Posted on: 2.4.2009

You were very astute to catch that brief reference to Trust. Good for you!

Thanks again for your kind comments. Please keep them coming!

Mark,

Thanks for your kind words.

I'm pretty sure that the foundational thought behind clinical HCIT is correct and absolutely on point. In fact, the evolutions in thinking, such as better semantic interoperability support through HL7v3, it's presence throughout the Meaningful Use criteria, and it's evolution to support business process management are broadening that foundation.

It's more akin though, to delivering clean water to the kitchen. It's a necessary precondition to a fine restaurant, a fast food outlet or ones home kitchen. In and of itself, it does not assure a safe meal, much less a tasty one. And, contaminated water can ensure an unsafe meal.

Once we evolve to safe, healthy kitchens, the practices of chefs become critical. In healthcare and healthcare IT, that means reaching the best possible diagnostic and therapeutic processes. Dr Graber's seminal highlights some areas that have clearly received inadequate attention. And, as we will see in subsequent posts, human cognitive tendencies, both conscious and semi-conscious, can be improved in a variety of specific ways, beyond the required first step of cleaning up the information and delivery process environment.

A lot of people have expressed interest in this topic.  The original editors of the American Journal of Medicine supplement, Mark Graber, MD and Eta S. Berner, Ed.D reached out as well.  They wanted to make sure that readers know that there is a very good summary of the state of the art and the commentaries are particularly insightful that is freely available at http://www.amjmed.com/issues?issue_key=S0002-9343%2808%29X0007-5
 
The other article is an opinion piece that Jackie Moss and Eta S. Berner wrote in 2005 that  is still relevant.  It is also freely available on pubmed central and relevant to your discussion.
http://jamia.bmj.com/content/12/6/614.abstract

Thanks for all of the wonderful email, kind words and support.  As before, stay tuned, there's more to come!

Dr. Bormel,
Your response to my comment is most appreciated. Note that you're being far too humble. I don't think any of your readers, me included, would find your e-mail group conversation with Dr. Graber et al as being boastful. Quite the contrary. I believe such adds significantly to your credibility, and feel you should never apologize for rising to the level of being a member of one of our profession's quietly (and well-earned) elite groups. That said, let's move on.

I do not believe that P4P sends the wrong signals, and further, I believe that comparing this health care incentive to Wall Street bonuses is comparing apples to oranges.

Wall Street often uses bonuses to circumvent the wishes of corporate stockholders, to sidetrack the media, and to avoid government investigation. This is not the case with P4P.

In my opinion, P4P is more than a "not so gentile nudge" into the 21st century for clinicians who ignore the "facts of life" when it comes to modern medicine. With P4P, your income is what it should be at any given point in time non-performance lowers your income and jeopardizes your employment and career. The latter, P4P is fair. It's right. The former is a national embarrassment that should lead to criminal prosecutions. Of course, P4P may lead to the same if certain docs continue to provide care based upon nothing more than personal opinion and experience. That's probably a good thing.

Doc Benjamin

Dr. Bormel,
This is a very solid post. I like the way your series is coming together. As it progresses, will you continue to support your statements with third party research, and if so, have you found any useful information and/or data that's more recent than 2005?

In response to your question, I do believe "that earning and building the trust of clinicians will be a critical component to informing better decisions." But human nature being what it is, we cannot expect full cooperation across the spectrum of clinicians, or full cooperation at every level from each clinician. Such are the egos involved. By definition, cognition is the process of both "knowing" and "perceiving." You must admit that combining those two processes in a single definition leaves considerable latitude for problematic results. Oops, blind spot, the patient died!

Although it appears you'll be dealing at a very high level throughout this series of posts, would you not agree, at least within a reply to my comment, that pay-for-performance will likely prove to be a primary catalyst to improve patient safety and quality of care? After all, impact on one's income, and even job security, can be a splash of cold reality in the face of unacceptable performance. Sort of cognitive reality. What are your thoughts?

Doc Benjamin

Thanks again for your comment. I agree that P4P sends an exactly correct signal to those who perform. I am under an arrangement with a P4P component and many of my colleagues are; we welcome P4P.

The challenge, as brought up by the AAFP and others in my referenced post, is that a singular focus on a set of objectives can cause neglect of professional standards. As was the case with Wall Street, the combined effects of individual players optimizing for themselves can and has caused systems problems.

Accountable Care Organization models, of course, seek to counter the Pay-for-Performance-of-Procedures (let's call that P4P4P for now) that currently directly rewards delivery of any and all services.

So, we are in violent agreement. P4P is important, especially to the degree that it promotes good cognitive practices of application of knowledge, and mitigates against inappropriate P4P4P.

Pages

Joe Bormel

Healthcare IT Consutant

Joe Bormel

@jbormel

...