Twenty studies (2,978 participants analysed; range 15 to 927) were included in the review. Thirteen were cohort studies, four were cross sectional studies, one a case-control and two were diagnostic accuracy studies. Ten studies used prospective design (four of which recruited patients consecutively), one was a retrospective study and the design was unclear in nine studies. Only three studies reported blinding of assessors. Verification bias was minimised in 18 studies. Five studies had greater than 20% drop-out rates.
Protein to creatinine ratio versus 24 hour urine collection (15 studies): None of the studies assessed adverse pregnancy outcomes. Sensitivity estimates for the detection of proteinuria ranged from 0.65 to 0.89 depending on the threshold assessed; sensitivity decreased as thresholds increased. Specificity ranged from 0.63 to 0.87; specificity increased as thresholds increased. Positive likelihood ratios ranged from 2.38 to 4.90 and negative likelihood ratios ranged from 0.17 to 0.40. The area under the curve was 0.69 which suggested that the protein to creatinine ratio performed well in detecting proteinuria. The optimum threshold (maximises sensitivity and specificity jointly) appeared to be between 0.30 and 0.35 inclusive. There was evidence of significant heterogeneity across thresholds for which two or more studies were pooled. Other findings were reported in the review.
Albumin to creatinine ratios versus 24 hour urine collection (five studies): Four studies assessed accuracy of albumin to creatinine ratio for detecting proteinuria, but results could not be pooled due to different study characteristics and thresholds assessed. One study reported adverse pregnancy outcome results.