Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Carter and colleagues have done a superb job in summarising the theory and practice of tools used in assessing children for autism.1 Unfortunately, their paper is marred by a mathematical error which, if left unchallenged, could undermine trust in the use of one such tool, the Autism Diagnostic Observation Schedule (ADOS).
According to the paper, the positive predictive value (PPV, ie, the proportion of those who test positive who actually have the condition) of the ADOS is 80.4% in a neurodevelopmental clinic population, while the negative predictive value (NPV, proportion of those who test negative who do not have the condition) in the same population is only 11.9%. If this were true, it would mean that the proportion of those in the test negative group who have the condition would actually be higher than in the test positive group: 88.1% against 80.4%. Fortunately it is not true.
Using the population prevalence of 53% quoted, and the figures of 91% sensitivity and 75% specificity, the true figure for PPV is as given, but the NPV is actually 87.5%. Carter et al's conclusion, therefore, that tests should be interpreted in the light of history and information from other sources stands, but the ADOS is not as hopeless as their paper might have suggested.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.