No
Yes
Yes
No
No
Yes
Yes
No
Yes
No
Yes
Yes
No
No
Yes
6. Teaching Points
1. Generalizability to practice: only apply these findings to women
who are similar to the women who were studied
2. Cutoff points change what you tell a patient
This systematic review and meta-analysis demonstrate the need to identify
the population that is being studied when reading a research article. In this
case, applying the results wrong could lead you to draw an incorrect
conclusion or assumption, causing more worry and anxiety to a patient.
This article is very specific about the women who are included in the studies
that are combined in this meta-analysis. The women had to be less than 14
weeks pregnant, and all of them were experiencing some kind of symptom,
pain or bleeding or both. Since only these women were included, the results
can only be applied to similar women in practice.
This article asked two different questions:
1. what are the diagnostic characteristics of a single progesterone
measurement in women who have symptoms and have an inconclusive
ultrasound
2. what are the diagnostic characteristics of a single progesterone
measurement in women who have symptoms but no ultrasound?
The systematic review of the literature turned up 26 studies that the authors
planned to include. In the end, they were able to work with fourteen of them
(5 of 7 for the with-ultrasound group and 9 of 19 in the no-ultrasound group).
The 12 that were omitted were excluded because they did not use a cutoff
that was similar enough to the included studies cutoffs to allow them to be
combined in the meta-analysis.
Its important to note, when applying this to practice, that these two
questions were answered using different cutoff measurements for normal
progesterone. The authors chose the most common cutoffs from the included
studies to highlight in the text and to produce sensitivity and specificity
numbers and likelihood ratios. (For the interested or advanced crowds, figure
6 on page 10 shows the sensitivities and specificities generated by three
different cutoff points for normal and not normal progesterone levels in
women with symptoms but no ultrasound. The graphs neatly show that as
you increase the cutoff point, the sensitivity rises but the specificity falls.
More true negatives are identified, but at the cost of more false positives.