Objective: To determine what threshold for proteinuria could best predict clinical outcome and whether this threshold could be applied universally to any biochemical assay.
Design: A prospective observational study of hypertensive pregnancies referred for further assessment after in a UK University hospital (n=197). Twenty-four hour urine protein was measured by two different assays [benzethonium chloride assay (BCA) and Bradford assay]. The differences between the two assays were calculated from Receiver Operating Characteristic (ROC) curves. Commonly used thresholds for defining preeclampsia (0.3 and 0.5 g/24 hours) were explored for both assays for the prediction of adverse clinical outcomes (severe hypertension, Birthweight<10th percentile, preterm delivery, and a composite biochemical/haematological derangement).
Results: The two assays are not equivalent. The prevalence of>300 mg/24 hour proteinuria and, hence, the prevalence of preeclampsia differed between the two assays. ROC curve analysis demonstrates that the two assays are similar in terms of overall performance as predictive tests. However the threshold of 300 mg/24 hours performs poorly as a predictor of clinical risk. Likelihood ratios (LR) for the BCA at the 300 mg/L threshold for each clinical outcome do not achieve statistical significance. At the 500 mg/L threshold, the LR+for the BCA assay does achieve statistical significance for severe hypertension (LR+:1.51 95% CI 0.99-2.28) and for birthweight<10th percentile (LR+:1.72 95% CI 1.11-2.66). For the Bradford assay at the 300 mg/24 hour threshold, the LR+does achieve statistical significance for birthweight<10th percentile (LR+:1.71 95% CI 1.41-4.31). However, at the 500 mg/24 hour threshold, the LR+is significant for severe hypertension (LR+:2.15 95% CI 1.07-4.34), birthweight<10th percentile (LR+:2.79 95% CI 1.4-5.54) and biochemical disease (LR+:2.47 95% CI 1.22-5.01).
Conclusions: This study suggests that thresholds for proteinuria need to be higher (possibly>or=0.5 g/24 hours) and there is the need for a "gold standard" proteinuria assay against which all other measures of quantification can be assessed.