Infectious Disease

First responders overestimate the likelihood of illness before, after certain tests

April 09, 2021

2 min read

Source / information

Disclosure:
Manrai does not report any relevant financial information. Morgan reports that while conducting the study, he received grants from the NIH and grants from the Agency for Research and Quality in Health Care, the CDC, and the U.S. Department of Veterans Affairs outside of the study. In the study you will find all relevant financial information from all other authors.

ADD SUBJECT TO EMAIL ALARMS

Receive an email when new articles are published

Please enter your email address to receive an email when new articles are published . “data-action =” subscribe “> subscribe

We could not process your request. Please try again later. If you continue to have this problem, please contact customerservice@slackinc.com.

Back to Healio

Primary care providers often overestimate the likelihood of pre- and post-test diagnosis, suggesting that many “are not used to using the likelihood in diagnosis and clinical practice,” the researchers said.

“This research grew out of a clear separation between the way we taught medical students tests and the way we practice clinically.” David J. Morgan, MD, MS, Professor of Epidemiology and Public Health at the University of Maryland School of Medicine, said Healio Primary Care. “We teach tests as math equations and 2×2 tables, but that doesn’t lead to patient care.”

Morgan and colleagues surveyed 553 practitioners – including general practitioners (n = 290), treating physicians (n ​​= 202), and nurses (n = 61) – and asked them to estimate the likelihood of illness in four clinical scenarios that are common in the primary setting Care: pneumonia, cardiac ischemia, breast cancer, and urinary tract infections.

“Each scenario was created for a general situation, but provided essential details for calculating the actual risk to patients (e.g. age and lack of risk factors for breast cancer in questions about mammogram screening),” the researchers wrote. “The primary outcome of test questions was to accurately identify the likelihood that a patient had an illness after positive or negative results.”

The responses to the survey were compared with evidence-based estimates of the disease by a panel of experts, according to the researchers.

They found that practitioners overestimated the likelihood of illness before testing all scenarios.

Based on the survey, the probability of pneumonia after positive radiological results was 95% (evidence range = 46% -65%; P <0.001); the probability of breast cancer after positive mammography was 50% (evidence range = 3% -9%; P <0.001); The probability of cardiac ischemia after positive stress test results was 70% (evidence range = 2% -11%; P <0.001). and the probability of urinary tract infection after positive urine culture was 80% (evidence range = 0% -8.3%; P <0.001).

After negative test results, the likelihood of developing pneumonia decreased to 50% (evidence range = 10% -19%; P <0.001); 5% for breast cancer (evidence range = <0.05%; P <0.001); 5% for cardiac ischemia (evidence range = 0.43% -2.5%; P <0.001); and 5% for UTI (evidence range = 0% -0.11%; P <0.001) - but it was still significantly higher than the panel's estimates.

Morgan said in an interview that a nationwide study would produce results “pretty close” to those they reported.

“We describe a problem at first sight that has been largely ignored but is critical to everyday patient decisions. This could explain a lot of the over-treatment and overuse of drugs, ”he said.

To address the problem, doctors need to recognize that they are likely to overestimate the likelihood of disease and think about the “real value” of testing.

“This process can be difficult without obvious clues to guide you,” he said. “We’re developing a website to make this easy: Testingwisely.com.”

In an invited comment Arjun K. Manrai, PhD, An assistant professor in the Computational Health Informatics Program at Harvard Medical School wrote that Morgan and colleagues’ study “suggests new goals for medical education and research pathways on how to better integrate probabilistic information into care.”

References:

Manrai AK et al. JAMA Intern Med. 2021; doi: 10.1001 / jamainternmed.2021.0240.

Morgan DJ et al. JAMA Intern Med. 2021; doi: 10.1001 / jamainternmed.2021.0269.

ADD SUBJECT TO EMAIL ALARMS

Receive an email when new articles are published

Please enter your email address to receive an email when new articles are published . “data-action =” subscribe “> subscribe

We could not process your request. Please try again later. If you continue to have this problem, please contact customerservice@slackinc.com.

Back to Healio

Related Articles