Protein/creatinine ratio in preeclampsia: a systematic review.

OBJECTIVE To estimate the accuracy of the protein/creatinine ratio in predicting 300 mg of protein in 24-hour urine collection in pregnant patients with suspected preeclampsia. DATA SOURCES Articles were identified through electronic databases (MEDLINE, CINHAL, and Cochrane) using the terms "preeclampsia," "protein/creatinine ratio," and "diagnosis," during the period January 1966 to October 2007. The relevant citations were hand searched. METHODS OF STUDY SELECTION Included studies evaluated patients for suspected preeclampsia with a 24-hour urine sample and a protein/creatinine ratio. Only English-language articles were included. Studies including patients with only chronic illness such as chronic hypertension, diabetes mellitus, or renal impairment were excluded. Using the Quality Assessment of Diagnostic Accuracy Studies questionnaire, we created group 1 satisfying all the required criteria and group 2 not satisfying all of it. Two researchers independently extracted the accuracy data. A graph comparing six receiver operating characteristic curves was plotted. TABULATION, INTEGRATION, AND RESULTS Twenty-one studies were identified, but only seven met our inclusion criteria (1,717 total patients). Group 1, with three studies, had 510 patients. The studies evaluated different cut points for positivity of protein/creatinine ratio from 130 mg/g to 700 mg/g. For protein/creatinine ratio 130-150 mg/g, sensitivity ranged from 90-99%, and specificity ranged from 33-65%; for protein/creatinine ratio 300 mg/g, sensitivity ranged from 81-98% and specificity ranged from 52-99%; for protein/creatinine ratio 600-700 mg/g, sensitivity ranged from 85-87%, and specificity ranged from 96-97%. CONCLUSION Random protein/creatinine ratio determinations are helpful primarily when they are below 130-150 mg/g, in that 300 mg or more proteinuria is unlikely below this threshold. Midrange protein/creatinine ratio (300 mg/g) has poor sensitivity and specificity, requiring a full 24-hour urine for accurate results. Higher thresholds have not been adequately studied.