Principles of Improving Diagnostic Accuracy (Part II of Diagnostic Accuracy Series)

By Robert L. Moore, MD, MPH, MBA, Chief Medical Officer

“There are three constants in life: change, choice and principles.” –Steven Covey

Steven Covey notes that actions and intent flow from principles; they are the foundation used to choose between different courses of action and to decide where to invest energy in self-improvement. The importance of principles extends to specific fields as well, including medicine.

In Part I of this series, we reviewed the extent of diagnostic inaccuracy in medicine, ranging from an error rate of 5% to 50%, depending on the nature of the patient/problem. A key contributor to this inaccuracy is our way of thinking about uncertainty: we are trained to be overconfident in the accuracy of our decisions.

Unfortunately, the other extreme, excessive concern about diagnostic uncertainty, leads less confident clinicians to order excessive laboratory and radiological tests. The Choosing Wisely campaign begins to shine a light on the scenarios where such tests are unequivocally useless but does not provide a framework for unnecessary testing when there is even slight uncertainty.

Last year, the American College of Physicians convened a group of experienced clinicians, teachers, and communications experts to address this challenge of diagnostic uncertainty. The product of this effort is called, “Ten Principles for More Conservative, Care-full Diagnosis.” Here is a brief summary of the first five principles:

  1. Promoting Enhanced Care and Listening. Perform an appropriate and thoughtful history and physical exam. When the diagnosis is unclear, continue collecting the history and evaluating changes in the physical exam at subsequent visits to determine how the patient’s clinical course is unfolding.
  2. Understand Uncertainty. Become comfortable with it, learn how to respond to it, and how to convey it to patients.
  3. Respond Carefully to Symptoms. Balance the natural history of common symptoms (75% – 80% of self, resolve within 4 to 12 weeks) with a consideration of potential psychological causes of symptoms (2/3 of patients with anxiety, depression, or somatoform disorders are undiagnosed), considering both the Social Influencers of Health and the long-term effects of Adverse Childhood Events, which also cause or accentuate symptoms.
  4. Maximize Continuity and Trust. Continuity of care by a primary care clinician is not only the single best predictor of patient satisfaction but also generates the trust needed to address the psychosocial issues mentioned above and to have patients trust the strategy of “watchful waiting” to observe the natural history of symptoms.
  5. Taming the Time Pressures around Patient Visits. Ensure the clinician has adequate time to listen, observe, discuss, and think. Adjust the system and the environment of care, as needed, to support this.

The derivation of these principles, which perhaps seems self-evident, required thought and effort of experts; trying to improve without guiding principles to guide us is disjointed at best. Like moral philosophy, it is the application of principles which is more challenging. Some approaches include taking time to think about how to apply these principles, finding small self-improvements or system changes to move towards achieving them, and telling stories to help reinforce how we approach gaps.

You, as clinical leaders in your settings, have an especially important role to play in helping your clinicians learn and apply these principles. On behalf of your patients, thanks for addressing these challenges in your setting.

Diagnostic Accuracy (Part I in a series)

By Robert L. Moore, MD, MPH, MBA, Chief Medical Officer

In 2001, the Institute of Medicine (IOM) published Crossing the Quality Chasm, which defined 6 realms of quality:

  • Safe
  • Effective
  • Patient Centered
  • Timely
  • Efficient
  • Equitable

Most quality improvement and quality assurance activities related to health care address one or more of these realms.  However there is one critical area that is missing from both the IOM realms and most QI plans.  This missing realm is at the heart of what it means to be a clinician, is infrequently measured and is uncomfortable to verbalize.

This realm is Accuracy, in particular the diagnostic accuracy of clinicians.

Diagnostic Inaccuracy:  Studies show that between 5 and 50% of diagnoses are erroneous, depending on the type of patient/problem.  The low range applies to a population of patients where most are normal; the high range applies to a population where all patients have complex abnormalities.  Autopsy studies and studies with “secret shopper” patients show rates of inaccuracy of between 10 to 20%.  (Mark Graber:  “The Incidence of Diagnostic Error in Medicine”, BMJ October 2013)

What is the source of this diagnostic inaccuracy?

At the core is an insufficient appreciation of uncertainty.  Put another way, clinicians and scientists are often overconfident in the accuracy of their decisions.  This psychological trait develops when we are trainees, as it makes us appear confident in the eyes of our patients, and helps prevent us from being paralyzed by indecision.  Fortunately, many medical conditions in primary care resolve on their own, so neither the clinician nor the patient ever become aware of the inaccuracy.

We can reduce diagnostic inaccuracy by changing the way that we think.

Daniel Kahneman (winner of the 2002 Nobel Prize in Economics) describes two ways of thinking:

  1. Fast thinking (also known as intuitive thinking or system I thinking)
  2. Slow thinking (also known as rational thinking or system II thinking)

Fast, intuitive thinking tends to be automatic, with input from emotions.  In his book, Thinking Fast and Slow, Dr. Kahneman notes 12 different classes of bias and 5 heuristics which can lead to irrational decisions, when we think intuitively.

Slow, rational thinking is more deliberative, systematic, and logical, with an evaluation of consequences of a decision.

As we go through our everyday lives and routine practice of medicine, we use fast thinking for most decisions, so we can get through our days without being paralyzed by indecision over minor decisions.  When the stakes are high, or when we notice a diagnostic pattern that doesn’t quite fit, we need to transition to slow, rational thinking.  For a clinician to be efficient and accurate, we need to know when to toggle back and forth between slow and fast thinking.

When slow thinking is associated with a retrospective analysis of a serious diagnostic error, as happens in morbidity and mortality rounds, or when a clinician becomes aware of a diagnostic error that occurred, it is good to explicitly think about which biases or heuristics contributed to the error, to help prompt us to move to slow thinking when needed.

This process is sometimes called “cognitive debiasing,” which is a fancy way of saying “learning from our mistakes”.

2022: Endemic COVID-19 and Preparing for COVID-2x

By Robert L. Moore, MD, MPH, MBA, Chief Medical Officer

“The value of experience is not in seeing much, but in seeing wisely.”

-William Olser

What does our future look like with COVID-19 becoming endemic? While vaccines and “natural” infections give some temporary immune protection from infection and re-infection, our collective immune system memories may render these infections less severe as time goes on, but this protection will be less for those who are older or immune compromised. Like influenza, regular COVID vaccine (possibly annual) will be recommended to slow spread of the infection (for healthier people), and make infections less deadly (for those with medical conditions rendering them more susceptible).

This has happened before. Paleo epidemiologists suspect that the “Russian flu” pandemic of 1889 to 1891 was actually caused by a coronavirus, possibly the grandfather of one of the subtypes that currently causes seasonal cold symptoms. Because modern DNA sequencing did not exist in the 19th century, we cannot be sure.

For this current Omicron outbreak, our system is strained, so there is a risk that too many people will let down their guards prematurely, increasing hospitalizations to a degree that forces dangerous staffing patterns, and before new powerful treatments like Paxlovid can become more widely available.

COVID-19 will become endemic, which will not make it benign. Omicron is more infectious than influenza and cold viruses, and immunity to coronaviruses if proving very transient. In the H1N1 outbreak of 2009 killed over 12,000 Americans, the majority under age 65. Hospitals were very full, but not as strained as by COVID in 2021. A massive immunization campaign resulted in just 20% of the population being vaccinated, enough vaccination with natural immunity to keep H1N1 at lower levels in subsequent years. However, the R0 of H1N1 was just 1.5, compared to about 4 for Omicron in the UK where vaccination was relatively high. Seasonal vaccination rates in the United States will not be enough to prevent annual waves of seasonally mutated COVID-19 in the years to come. We may get more used to them, but they will be worse than annual influenza seasons, at least for 2022, and maybe for years to come.

In the end, our annual winter “flu” season, which is actually caused by a collection of flu-like respiratory viruses, will be a more severe each winter, as COVID-19 joins influenza, RSV, parainfluenza, rhinovirus, the other four coronaviruses that cause milder respiratory infections, and others in causing illness and hospitalization each winter. The summer waves we saw the last two years are likely to continue, to some degree, as well.

Quarantines for COVID-19 will be phased out, but some amount of voluntary isolation of those that are ill will persist, and there will be a high usage of masks to prevent spread of respiratory pathogens in the next several years.  Virtual visits will continue to be a significant part of our health care delivery.

There will be other new coronavirus infections in the future. Now is a good time to prepare for the next coronavirus that jumps species.

In the 21st century, COVID-19 is the third coronavirus to appear that causes severe respiratory symptoms (after SARS-1 and MERS). Given the huge reservoir of viruses in other species, we are quite likely to have a Coronavirus infection caused by a different strain sometime in the next decade, a COVID-2x epidemic that could lead to another pandemic. Looking at the shortages we had early in this pandemic, we can prepare for COVID-2x, as well as more severe influenza outbreaks. There are lessons for industry, government, public health authorities, and the general public. Here are some that might be most helpful for clinical leaders in primary care:

  • Ensure local community stockpiles of highly effective masks such as N-95s are ready –enough to last for a couple of months. Encourage patients to keep some on hand as well, for future winter “flu” seasons and future pandemics. Don’t let them all expire at once, set up a system for regular purchase and replenishment.
  • Gather together and organize policies, procedures and leadership lessons, and find a way to go through these systematically each year, so the knowledge is refreshed as staff change. Include not just best practices, but also any mistakes made and options for responding to the most challenging parts of a pandemic.

Transition community connections that were forged in the COVID-19 pandemic to work on other important health issues. Set up regular meetings with your local health officer to work together on shared priorities. Relationships will be key to rapidly and effectively responding to future public health needs.