‘Despite electronic health records, follow-up of abnormal findings is not fail-safe,’ Dr. Singh says. Hardeep Singh investigates why things go wrong and what can be done to avoid problems Physician and researcher Hardeep Singh probes one of the most vexing issues in medicine: diagnostic errors. As chief of health policy, quality and informatics at Houston’s Michael E. DeBakey VA Medical Center and an associate professor of medicine at Baylor College of Medicine, Dr. Singh measures the human toll of such mistakes and investigates how technology can help clinicians avoid making them. He recently served as an expert reviewer on a new report from the Institute of Medicine which warns that most Americans will experience at least one diagnostic error in their lifetime. He answered questions from The Wall Street Journal about what doctors and patients can do to make sure they get to the bottom of the diagnostic puzzle. WSJ: The Institute of Medicine report drew on your research which has shown that an estimated 5% of patients or 12 million American adults are misdiagnosed every year. How did you arrive at that figure? DR. SINGH: We built estimates of misdiagnosis by analyzing three previous studies on missed, delayed or wrong diagnosis in outpatient care. While studying high-risk situations such as when patients returned to the doctor’s office unexpectedly or when certain tests did not receive timely follow up, we looked for any evidence of a missed opportunity to make a correct or timely diagnosis in hindsight. Missed opportunities included several types of overlapping situations, such as when the diagnosis was completely missed (cancer diagnosis missed despite the presence of alarming symptoms), the diagnosis was wrong (patients told they have one diagnosis but actually have evidence of another) or the diagnosis was delayed (patient had an abnormal test or finding suggestive of cancer but no one has looked at it or told the patient). This allowed us to estimate that 1 in 20 U.S. adults will experience a misdiagnosis every year. Common threads WSJ: The most commonly missed diagnoses, according to a 2013 study you led, included pneumonia, heart failure, kidney failure, cancer and urinary-tract infections. Is there a common thread that makes these more likely to be missed? DR. SINGH: Yes, this study illustrates how many types of relatively common conditions seen in primary-care clinics are at risk for misdiagnoses. Problems in gathering or interpreting information related to history taking and physical examination during the doctor visit [at primary-care clinics] played a big role. Outpatient clinics are busy and time-pressured, and uncertainty is quite the norm when doctors first see patients with new or multiple symptoms. Despite detailed progress notes that doctors write to meet billing and regulatory requirements, we found little record of expanded thinking, or what we call differential diagnosis, in these errors. We think lack of time interacting with patients, listening to their stories, and thinking about how to put it together are some of the common issues here. I might add that diagnosis is not always a black-and-white phenomenon. Arriving at a definitive diagnosis, even of a common condition, is also not easy when patients’ symptoms and associated disease conditions are unfolding or evolving over time. Doctors deal with a lot of uncertainty and probabilities on how to pursue a problem and a diagnosis. Thus, we look for unequivocal evidence that some red-flag or critical finding or abnormality was missed or not investigated when it should have been. WSJ: How can doctors avoid overdiagnosing and incurring unnecessary costs for overtesting? DR. SINGH: Doctors usually need to balance between ordering additional tests or procedures that often come with their own risks versus risking “underdiagnoses” by not investigating. There is so much national conversation now on overdiagnosis, overtesting, overtreatment and health-care costs. The midpoint of the pendulum is what we need to strive for, and that’s not going to be easy. Using computer case-based simulated scenarios, we examined the alignment between doctors’ diagnostic accuracy and their confidence in that accuracy. The study found that the two are not aligned; doctors were too confident even when their diagnostic accuracy was poor and often did not request additional helpful resources such as tests and referrals in difficult diagnostic situations that involved uncertainty. Anecdotally, we find many doctors not comfortable expressing uncertainty to patients as they fear it might affect their reputation. We need to find a better way to align confidence and accuracy. But as of now we don’t have well-developed mechanisms for doctors to get feedback on the accuracy of their diagnoses, and information about the ones they get wrong. Many doctors are afraid of missing something and getting sued. And the way the current billing system works, it mostly pays for tests and procedures rather than time to think about the correct diagnosis. There is also no way to code or bill for “uncertain” diagnosis. It is critical that we do more research on how to measure problems with diagnosis so we can fix them. Electronic tools WSJ: What about tools like electronic decision-support websites that help doctors pinpoint the right diagnosis when it isn’t obvious or easy to find? DR. SINGH: Electronic tools to support diagnosis can often provide a thorough differential diagnosis and might make doctors think of a diagnosis they had not previously considered. Some systems are better in getting the job done. But to get doctors to use these tools is hard for several reasons. Most conditions they see are pretty straightforward. When they do need help with complex patient scenarios, they either don’t realize they need help or choose not to use the tool because they are confident that the diagnosis path they are on is correct. Moreover, the tools only work when patient data is gathered correctly and completely, and given problems in this area, entering an incomplete patient story could make these tools less useful. There are ongoing efforts to combine these tools with electronic health records to automatically make sense of digitally collected patient data. WSJ: Courses that teach clinical thinking and reasoning skills seem to be on the rise in medical education. How can they help? DR. SINGH: Traditional teaching focuses on diagnosis but not so much on errors of diagnosis. So it’s the error-prevention part we need to strengthen. The Society to Improve Diagnosis in Medicine received a grant recently to develop cases to teach about diagnostic error. These programs will help a lot if they can consider thinking and reasoning within the constraints of the environment in which it occurs. And what I mean is, even when I learn the right stuff in a classroom setting, I might not be able to put it together in a chaotic emergency room. Simulation could help optimize skills for the real world, and so could novel techniques to teach the future generation to make diagnosis in a technology-enabled health-care system. Programs will also need to teach how to best make diagnosis for the “online” patient who isn’t sitting face to face with you. If doctors and patients cannot spend adequate time and communicate effectively with each other to formulate a correct diagnosis, training programs won’t work. In some cases of misdiagnosis, patients had reinforced their concerns to clinicians several times, but the clinician did not change course. Training programs need to emphasize concepts related to responsibility, accountability and communication and will work best in a health-care system more favorable for improving diagnosis. Looking for ‘triggers’ WSJ: Recently, you published a study in the Journal of Clinical Oncology on the use of certain “triggers” in an electronic medical record to identify patients whose abnormal findings were not followed up on. How did it show that electronic records are used to help reduce delays in diagnosis? DR. SINGH: Despite electronic health records, follow-up of abnormal findings is not fail-safe. We found that nearly 8% of abnormal test results are lost to follow-up. This could lead to delays in cancer diagnosis and malpractice claims. So it’s essential to have backup systems to catch abnormalities that might have been missed. To do that, we developed “trigger” queries—a set of rules to electronically identify medical records of patients with potential delays in diagnostic evaluation for cancer, such as charts with no documented follow-up for abnormal findings suspicious for cancer after a certain period. There are patients with “suspicious” or abnormal chest X-rays or CT scans, abnormal PSA tests and positive fecal occult blood tests who had not been followed-up on a timely basis. We then communicated information about potential delays to clinicians taking care of these patients. We found that these electronic “trigger-based” interventions are effective in reducing time to diagnostic evaluation of colorectal and prostate cancer, as well as in improving the proportion of patients who receive follow-up as compared with usual care. WSJ: What can patients do to improve their chances of a correct diagnosis? Can giving them better access to their lab and test information help? And when should they ask for a second opinion? DR. SINGH: They could be strong partners with their doctors. It might not be easy to get to a complicated diagnosis right up front, and doctors could be wrong. All patients need to know this. A good partnership between the two would enable correct and timely diagnosis as the story becomes clearer and more information emerges. Patients should always check on their test results, regardless of whether they have heard back from their doctor or not. So no news is not necessarily good news. Better access to their own test information such as that offered via online patient portals could be of benefit. But not enough patients are using them, and when they see results, they might not always understand the meaning unless the doctor added a note to the result. So we need additional strategies that support patients beyond just access. Second opinions are often suggested as potential interventions to reduce misdiagnosis. In one of our studies, second opinions changed diagnosis around 15% of the time and treatment about 37%. That’s a pretty big impact. What we don’t know yet is if all these second opinions ultimately lead to better diagnosis and better treatment—mostly because we don’t know if the opinion ended up being correct in the long term. One should consider seeking a second opinion when symptoms are not resolving despite treatment; when diagnosis is unclear; when a patient appears to have a serious, debilitating, life-threatening, or rare condition; or when treatment options are risky or harmful. WSJ: What are the most important takeaways from the Institute of Medicine’s report on diagnostic error? DR. SINGH: The report recommends focusing not only on improving diagnostic reasoning of individual clinicians but also on engaging patients, care teams including diagnostic services such as lab and radiology, and delivery systems to support reliable and accurate diagnosis. Policy-level fixes include time and reimbursement for diagnostic “thinking,” malpractice reform, getting technology to be more supportive rather than a barrier between patients and clinicians, and improving the currently scant research funding to understand and reduce the problem. It will be a huge leap forward if the report could help foster a safety culture where clinicians are encouraged to discuss and learn from errors more openly and work with patients, policy makers and researchers to start solving this complex problem. Source
