Problems for clinicians and patients in interpreting diagnostic test results

A study published in today's British Medical Journal by Bramwell et al (BMJ 2006; 333:284), Health professionals' and service users' interpretation of screening test results: experimental study, examines accuracy of interpretation of the results of fetal testing for Down's syndrome in pregnant patients, obstetricians, and midwives.

Participants were provided with either frequencies (e.g. "Roughly 100 out of 10,000 babies have Down's syndrome...") or percentages ("Roughly 1% of babies have Down's syndrome...") to illustrate the risk of Down's syndrome actually being present when a test result is positive; 86% of responses were incorrect, though clinicians had a higher rate of correct answers (43%) than midwives (0%) or patients (9%). The groups did not differ significantly in overestimating or underestimating risk.

The data presented as frequencies were more often interpreted correctly than percentage data, though overall rates of correct answers remained low. Researchers also queried participants regarding their confidence in their answers; health professionals indicated that they were confident in their answers, though many turned out to be incorrect.

Boxes 1 and 2 of the article provide examples of the information provided to participants and an explanation of the calculations to accurately interpret risk based on a positive test result.

Given these difficulties in interpreting diagnostic testing information and the apparent misplaced confidence exhibited by clinicians in the study, the authors recommend increased awareness of difficulties in interpreting such information and consideration of different ways of displaying information for increased success.

Related link: A training case, "Screening for Antibody to the Human Immunodeficiency Virus," from the Centers for Disease Control and Prevention uses HIV testing examples to illustrate how population characteristics (pre-test probability) and diagnostic test properties (sensitivity and specificity) affect the interpretation of diagnostic test results.