Journal of Clinical Endocrinology & Metabolism, Vol 46, 907-910, Copyright © 1978 by Endocrine Society

ARTICLES

Subhourly variability of circulating third trimester maternal steroid concentrations as a source of sampling error

JE Buster, PJ Meis, CJ Hobel and JR Marshall

Seven plasma samples from five normal third trimester pregnant women, drawn every 5 and 15 min at similar times on 2 days, 2 days apart, were measured in quadruplicate for estradiol (E2), estriol, progesterone, 16 alpha-hydroxyprogesterone, and 17 alpha-hydroxyprogesterone (17P). The mean of the 22 samples obtained from each subject for each steroid was used as a reference mean. Individual determinations were converted to percentages of the reference mean which was normalized to 0%. All pregnancies were uncomplicated, and all mean values were within the normal range for gestational age. Variability about the reference mean for single samples, however, range from a low of -80% to a high of 150%. The single sample, 95% confidence intervals for individual steroids pooled from the five subjects (110 determinations) range from +/- 36% (E2) to +/- 60% (17P). Mean percentage coefficients of variation between 5-min and 15-min sampling sequences were compared by analysis of variance. There is no significant difference between the mean percentage coefficients of variation of a 5-min as opposed to a 15- min sampling sequence for any of the hormones measured. The 95% confidence interval width around the reference mean is a function of the number of samples obtained. Because the 95% confidence interval width from 110 measurements decreases approximately as 1/ square root n with increasing sampling size, the decrement progressively diminishes. For E2, the least variable steroid, a one-sample 95% confidence interval width of +/- 36% decreases to approximately +/- 18% with four samples or approximately +/- 12% with nine samples. For 17P, the most variable steroid, a one-sample 95% confidence interval width of +/- 60% decreases to approximately +/- 30% with four samples or approximately +/- 20% with nine samples. Multiple sampling with plasma pooling is required for the accurate study of steroid concentrations in individual subjects in late pregnancy.