Objective: To estimate and compare contamination rates of three different urine-sampling methods in pregnant women to assess bacteriuria.
Methods: In this cross-sectional study, 113 pregnant women collected three different midstream urine samples consecutively: morning (first void); midstream (void without further instructions); and clean-catch sample (void after cleaning). The following end points were considered contaminants: epithelial cells, Gram-positive rods or mixed bacteria in the Gram stain, and mixed growth or skin flora in the urine culture. Intraindividual variability in contaminants was quantified with Fleiss-Cohen's weighted κ statistic. Differences between samples were assessed using generalized estimating equations.
Results: Mainly low numbers of Gram-positive rods were more likely to be present in Gram stains of midstream samples compared with clean-catch samples (77.7% compared with 66.7%, P=.022). Morning samples showed more mixed growth compared with midstream samples (6.2% compared with 0.9%, P=.050). No consistency in quantity of contaminants was found in midstream samples compared with morning and clean-catch samples. No differences were found between the other end points in all three urine samples (P>.05). The study could detect an odds ratios of 2.0 for differences in urine-sampling methods with 80% power and 5% significance for most end points.
Conclusion: In pregnant women, the contamination rate of midstream samples is comparable with the contamination rates of morning and clean-catch samples. The quantity of contaminants varied among the three samples collected by one woman. These results show that more complex, unpractical, and time-consuming morning and clean-catch samples are not superior. Therefore, we recommend a midstream sample to assess bacteriuria in pregnant women.
Level of evidence: II.