Probability of Disease Overestimated Significantly by Primary Care Providers

Researchers compared the estimation of the presence of disease by primary care clinicians and an expert panel.

Primary care clinicians overestimate the probability of disease before and after diagnostic testing, which likely leads to overutilization of treatment that could harm patients, a recent research article says.

With 14 billion laboratory tests performed annually in the United States, effective ordering and interpretation of tests is essential to avoid waste and overutilization of treatment such as medications and procedures. Diagnostic errors account for a significant proportion of serious harm to patients.

The recent research article, which was published by JAMA Internal Medicine, is based on data collected from more than 550 primary care clinicians. The clinicians were asked to estimate the probability of the presence of disease in clinical scenarios before and after diagnostic tests for four conditions. The probability estimates were compared to the estimates of an expert panel that determined the probability of disease based on a literature review that included diagnosis text books.

The researchers found that the primary care clinicians overestimated the probability of disease for all clinical scenarios before testing:

• For pneumonia, the median estimate of pretest probability of disease by the primary care clinicians was 80%, compared to a range from 25% to 42% for the expert panel
   
• For breast cancer, the estimate of pretest probability by the primary care clinicians was 5%, compared to a range from 0.2% to 0.3% for the expert panel
   
• For cardiac ischemia, the estimate of pretest probability by the primary care clinicians was 10%, compared to a range from 1.0% to 4.4% for the expert panel
   
• For urinary tract infection, the estimate of pretest probability by the primary care clinicians was 20%, compared to a range from 0% to 1% for the expert panel

The researchers found that the primary care clinicians also overestimated the probability of disease after positive test results:

• For pneumonia, the estimated probability of the presence of disease by the primary care clinicians after positive radiology results was 95%, compared to a range from 46% to 65% for the expert panel
   
• For breast cancer, the estimated probability of the presence of disease by the primary care clinicians after positive mammography results was 50%, compared to a range from 3% to 9% for the expert panel
   
• For cardiac ischemia, the estimated probability of the presence of disease by the primary care clinicians after positive stress test results was 70%, compared to a range from 2% to 11% for the expert panel
   
• For urinary tract infection, the estimated probability of the presence of disease by the primary care clinicians after positive urine culture results was 80%, compared to a range from 0% to 8.3% for the expert panel

"This survey study suggests that for common diseases and tests, practitioners overestimate the probability of disease before and after testing. Pretest probability was overestimated in all scenarios, whereas adjustment in probability after a positive or negative result varied by test. Widespread overestimates of the probability of disease likely contribute to overdiagnosis and overuse," the researchers wrote.

Interpreting disease overestimation

Overestimation of the probability of disease can have negative consequences, the lead author of the research article told HealthLeaders.

"If doctors overestimate chance of disease, they will often diagnose patients with diseases they do not have. There is no final test that is 100% accurate. Probability is central to making a diagnosis, and we need to consider probability both before and after testing, knowing that there are often overestimates that can lead to harms for patients. Overdiagnosis of disease leads to unnecessary and sometimes harmful treatments and procedures," said Daniel Morgan, MD, MS, professor of epidemiology, public health, and medicine at University of Maryland School of Medicine.

Most doctors likely overestimate the probability of disease, not just primary care providers, he said. "Humans are not naturally good at probability and medicine is extra difficult as we often do not get feedback. We often do not follow patients for long and many diagnoses are uncertain. Doctors like other people have biases such as neglecting how rare a disease is in a population—they remember recent or rare cases, and we are generally rewarded for making diagnoses even if they are incorrect."

Probability is often not applied effectively by clinicians, Morgan said. "Probability is the scientific basis for how we teach evidence-based diagnosis and how we conduct and interpret trials for the benefits of treatments. So, probability is central to diagnosis and treatment. However, medical practice often ignores this. A shaky understanding of probability likely leads to significant medical overuse."

Related: New Research Targets Diseases for Diagnostic Improvement Initiatives