Statistical analysis
A power calculation was performed. The calculation was based on the
occurrence of OASIs being 6% in the first examination (clinical
diagnosis of OASIs by midwives/doctors). As the data is ‘paired’ (i.e.
two measurements, clinical examination and TPUS, for each subject), in
addition to the OASIs rate for each method, it is also necessary to
estimate the proportion of subjects were there will be a discordant
response. This is where the outcome will be different from the two
methods. With increased surveillance with TPUS and a detailed
examination by a trained doctor, we predicted that the detection rate of
OASIs would be rise from 6% to 12%. Assuming that 10% of the subjects
would have discordant responses, and also assuming a 5% significance
level and 80% power, it was calculated that 216 women would be required
for the study. An extra 12 patients were recruited as we anticipated
that some of the ultrasound images would be in suboptimal quality to be
included in the study. Sensitivity of a test was calculated by the
number of true positives / the number or true positives plus number of
false negatives. Specificity of a test was calculated by the number of
true negatives / number of true negatives plus number of false
positives. The positive predictive value of a test was calculated by the
number positives of true positives / number of true positives plus
number of false.
Data was entered onto a Microsoft Excel database and analysed by SPSS
version 26 (Chicago, Illinois, USA). This study was approved by the
South East Coast - Surrey Research Ethics Committee (REC 16/LO/2140).