Confidence Intervals in the Table are Wrong
There are some errors in the Table of the article by Nassar et al, on
diagnostic accuracy of clinical examination for detection of non-cephalic
presentation in late pregnancy. A clue that most of the confidence
intervals for sensitivity are wrong is that they are narrower for almost
all of the subgroups than for the whole group of 130 noncephalic
presentations, even though the sample sizes of the subgroups are smaller.
In addition, the confidence interval for the sensitivity in Asian patients
(67% to 72%) does not include the point estimate (46%). Similarly, the
confidence intervals for specificity all have width of 2%, even though the
number of subjects per group varies more than 10-fold (from 115 to 1158).
The authors of this otherwise fine study may wish to publish a correction
to set the record straight.
Competing interests: No competing interests