Skip to main content

Reducing Diagnostic Errors

News  |  By Sandra Gittlen  
   October 01, 2016

A look at how healthcare organizations are working to improve diagnostic accuracy.

This article first appeared in the October 2016 issue of HealthLeaders magazine.

Children's Hospital Colorado, University of Colorado Hospital, and the University of Colorado School of Medicine are collaborating on an initiative to better understand and address the problem of diagnostic errors across their organizations. In addition to addressing the first-stage remedies of better communication of results and better handoffs in transition of care, the three institutions are now focusing on the issue of cognitive bias.

"We know that as providers, when our brains are thinking fast, we are prone to bias. Our hypothesis is that we can teach ourselves to also think slowly, and try to avoid anchoring and other cognitive biases that can lead to diagnostic errors and delays," says Daniel Hyman, MD, MMM, chief medical and patient safety officer at Children's Hospital Colorado, a private, nonprofit pediatric health network with over 3,000 pediatric specialists.

Diagnostic errors are under the microscope at many healthcare organizations following a report from the National Academies of Sciences, Engineering, and Medicine—Health and Medicine Division that found that 5% of U.S. adults who seek outpatient care each year experience a diagnostic error.

The Health and Medicine Division also found that "postmortem examination research spanning decades has shown that diagnostic errors contribute to approximately 10 percent of patient deaths, and medical record reviews suggest that they account for 6% to 17% of adverse events in hospitals."

Diagnostic errors, no matter their origin, are costly to health systems. In a news release in 2013, Johns Hopkins researchers reported that after reviewing 25 years of U.S. malpractice claim payouts "diagnostic errors—not surgical mistakes or medication overdoses—accounted for the largest fraction of claims, the most severe patient harm, and the highest total of penalty payouts." Diagnosis-related payments, the researchers found, amounted to $38.8 billion between 1986 and 2010.

To reengineer clinicians' approaches to diagnosis and to encourage them to slow down and avoid bias, Hyman is working with a group of adult and child health faculty at the three Colorado organizations to develop and introduce a curriculum that will be used at their hospitals and medical school starting in the fall. Ahead of that rollout, they will be surveying the faculty about their knowledge and attitudes about cognitive bias and incorporating the concepts into case review processes similar to other patient safety efforts.

"Physicians have the opportunity to expand the diagnostic assessment or treatment plan to include things that are less common," Hyman says. "In fact, anybody on the care team, including the patient and family, can speak up if they have a concern."

For example, a patient may present with complaints including abdominal pain, vomiting, and fever, and clinicians are likely to initially focus on gastrointestinal causes of the symptoms. They may even become "anchored" in an abdominally focused list of possible causes. "Until the caregivers expand their thinking, they may not consider the diagnosis of pneumonia," he says.

Slowing down and being more efficient in diagnosing patients also aligns well with the new world of value-based payments, Hyman says. "In a capitation model, the sooner we make the right diagnosis and provide appropriate treatment, the greater the cost efficiency," he says, contrasting it to fee-for-service where physicians are reimbursed for every visit and every test.

Understanding diagnostic error rates can be tricky, though. "You can't run a report on how many patients had a diagnostic error," he says. "But you can run a report on patients whose diagnoses changed from one visit to another and analyze those." What providers might uncover, for example, are patients who were treated for a viral illness but actually had leukemia, or were treated for a headache when they had a brain tumor.

Expanding the traditional care team
Through his work in ambulatory care clinics, Mikael Jones, PharmD, BCPS, clinical associate professor at the University of Kentucky College of Pharmacy, says he has realized that an efficient way to decrease diagnostic error rates is to form cohesive care teams.

He points to the case of an elderly patient he consulted on five years ago. The woman presented to the clinic with severe diarrhea and generally wasn't feeling well. Her history showed a recent course of antibiotics, and the nurse practitioner was worried about Clostridium difficile colitis or C. diff, which would be catastrophic in a patient her age.

Before concluding that diagnosis, Jones asked about other medications she was taking. The woman had been consuming a dietary herbal supplement, which Jones found to have a high likelihood of causing diarrhea.

"I suggested taking a step back and seeing if stopping the supplement would make a difference, and it did," he says, adding the woman did not have to undergo taxing C. diff treatment.

Jones learned from that experience to make the diagnostic process a team sport and to be more specific in patient questioning. "Don't just ask about medications; ask them about prescription, non-prescription, and supplements," he says.

Information flow has to improve, too, he says. For instance, while e-prescribing has made it easier to get prescriptions to pharmacies, the information flow back to the prescriber about whether a patient has filled the prescription is lacking. "In the clinic, we have to play detective to see if they filled the prescription or not," he says.

Jones adds that knowing how soon the prescription was picked up also is important because some medications, to be effective, have to be taken in a certain time frame.

He also believes provider notes should be looked at as a way to reduce diagnostic errors. "The notes are getting longer and longer and a lot of information is being imported. One incorrect fact can be easily propagated," he says. "I make sure to look at all medication notes and reconcile them with what the patient is saying and what was intended by the healthcare provider."

Clarity in reporting
Poor documentation can also flummox radiologists, according to Joseph Glaser, MD, at Middletown, New York–based Radiologic Associates, PC, who says more attention has to be paid to basics such as systematic reporting.

"In communicating results, you not only have to report what's urgent and important but also secondary findings," says Glaser, a nuclear medicine physician.

For instance, if a patient comes in for a chest x-ray with a presumed diagnosis of pneumonia, the report must not only describe whether there are clear lungs, but also the observation of a broken rib. "We may find a surprise that can change patient management," he says.

And how these findings are relayed to the doctor is equally important. While new technology has helped tremendously, he says, there are still times when a simple phone call is the most effective way to ensure timely and clear communication. This can also help on the incoming end as well.

"While some incoming documentation is concise and clearly states what is being evaluated, there are also occasions when incoming documentation contains a great deal of additional material—often automatically generated—that may make it difficult for the physician to find what they need," he says.

He encourages imaging physicians to learn how referring doctors receive information and incorporate those preferences into their reporting. "Doctors on both ends can suffer from information overload, so it's better to know what findings they want about certain conditions," he says.

Slow and steady
Don Goldmann, MD, chief medical and scientific officer at the Institute for Healthcare Improvement, agrees that clinicians are moving too fast these days, leaving the opportunity for diagnostic errors.

"Physicians used to take the time for careful observation, but now we think everything has to be solved right away," he says, adding "not all diagnoses are evident right away."

This hurry extends to direct patient interaction. "Physicians don't tend to wait for patients to answer questions. If there's any pause, they fill it with the next question or their opinion," he says.

He encourages doctors to really listen to patients and explore possible diagnoses with simple and open-ended questions, such as, "How long have you felt like that?" and "What do you think caused this?"

By slowing down, he says, physicians can gain tremendous insight on issues such as family problems that are causing stress. Knowing this might avoid needless tests looking for very unlikely chronic conditions.

"The worst thing to do is to order a bunch of tests to cover yourself because you're worried you'll miss something. Every test has a margin of error, so they will likely lead to more tests," he says, which, in turn, "will do more harm than good."

He says physicians who do order tests must get better about closing the loop on delivering results to patients. "Misdiagnoses can be made simply because the referral loop isn't closed and assumptions are made about who knows what," he says. "We keep holding out the dream of an integrated medical records system" where results are well communicated. The reality, though, requires physicians to reach out to referring groups to obtain results and then directly communicate those results to patients.

Reaching for the closed loop
Closing the loop is a mission for Johns Hopkins Community Physicians, a network that includes internal medicine, family practice and pediatrics, and specialty services with locations in Virginia, Maryland, and the District of Columbia.

Johns Hopkins Community Physicians has created redundant processes in the EMR system to ensure that physicians are alerted in a reasonable time when tests they've ordered have been completed, says President Steven Kravet, MD, MBA, FACP. The system also alerts physicians when they haven't been received in a reasonable amount of time so they can be checked on.

"I don't believe no news is good news," Kravet says. The goal is to avoid results falling off the map like an abnormal Pap smear that gets sent to the wrong doctor and the patient is never called.

The network now labels certain tests "critical" so physicians can run reports and quickly track down missing results. Even noncritical tests are reported on so that all results make their way into the system.

Kravet also would like to see certain testing systems rise to the level of mammography where women are told immediately whether there is cause for concern.

"This does not exist for other imaging tests such as chest x-rays," he says. If a radiologist finds a 5 mm nodule on the lung, they'll note it in the record but may not necessarily call the doctor. However, that still requires a follow-up, he says, so it's critical that the information is directed to the provider. "There's more information nowadays, and there is a higher tendency for things to get lost in a pile of paper."

Team time
Richard Friedberg, MD, PhD, FCAP, president of the College of American Pathologists, hopes new payment models will help in the fight to reduce diagnostic errors. "In a capitated system, every time you order a test, you lose a buck," he says, and, therefore, pathologists are being encouraged to set standards for appropriate usage.

Friedberg says most laboratories have become highly controlled and regulated and, therefore, highly efficient, so most of the diagnostic errors are occurring before the specimen hits the lab and after it is tested.

In pathology—as other diagnostics—it's possible to be 99% correct and 100% wrong if an error occurs, such as putting the wrong name on a tube. And as care teams grow, he says he believes so does the potential for mistakes.

"We're getting more and more people at the bedside who have less and less experience with diagnostics," he says, pointing to the influx of nurse practitioners and physician assistants to address gaps in primary care but who may not be properly trained in lab diagnostics.

Pathologists must partner with providers and other members of the care team for better training in ordering and more awareness of how to construct and communicate reports. For example, a best practice could be to put critical words such as "malignant" in bold at the top of a report.

As healthcare organizations across the country address their diagnostic error issues, Children's Hospital Colorado's Hyman is optimistic that like other challenges, healthcare will emerge victorious in solving this problem through changes such as how providers think about diagnosing patients. "I'm confident that 20 years from now people will say we used to have diagnostic errors, but now they are much less common because we implemented changes to the way we communicate and think about diagnoses," Hyman says.

Get the latest on healthcare leadership in your inbox.