Strengths and weaknesses
The present study is a population-based registry study that includes a large number of symphysis-fundus measurements. The study population from Västra Götaland county can be expected to be sufficiently representative of Norwegian conditions, and in a previous study we have shown that of the available variables, only the mother’s height and weight had an effect on the symphysis-fundus measurement. Smoking, parity and age had a minimal effect, which indicates that moderate differences in the variable distribution between Norway and Sweden will have little effect on the results (13).
The measurements are performed by a number of different midwives and doctors, and different individuals will measure in different ways. The percentile curves nevertheless represent a population of healthcare personnel performing measurements (13). This will smooth out systematic biases and make them representative in a population context. Individual errors of measurement will reduce predictive value. Since the data are retrieved from a clinical population database, measurement errors will already be present. However, this also implies that our calculated predictive value will be realistic in relation to what may be anticipated in clinical practice.
The clinician’s knowledge of any general risk profile the women may have had with regard to intrauterine growth restriction or a small infant may have affected the symphysis-fundus measurement. We had no information on the times when possible diagnoses were made, nor on whether the clinician had knowledge of these when the symphysis-fundus measurement was performed.
Another element of uncertainty in the study is that pregnancy with a risk of intrauterine growth restriction or a small infant is excluded at some point in the pregnancy from symphysis-fundus screening and further growth is estimated exclusively with the use of ultrasound. This may produce an apparently weaker predictive value of the symphysis-fundus measurement. We found, however, that the number of measurements in low-risk pregnant women was on average only 0.2 more than in the rest of the population, which does not indicate a significant degree of selection (13).