Estimation of proteinuria as a predictor of complications of pre-eclampsia: a systematic review

The review was carried out with a prospective protocol using widely recommended methods [8‐12]. We searched MEDLINE (1951 to 2007), EMBASE (1980 to 2007), Cochrane Library (2007) and MEDION (a database of diagnostic test reviews set up by Dutch and Belgian researchers) for relevant citations. A search term combination was constructed after exhaustive planning and piloting of possible search concepts capturing the relevant population, tests and outcomes. Our search terms are shown in Additional file 1. An initial search in Medline yielded 11,711 citations. The search strategy was adapted for searching in Embase to obtain a total of 19,500 citations. From this citation set, studies were selected for inclusion in the review in a two-stage process if they studied the accuracy of proteinuria in the prediction of maternal and fetal complications in women with pre-eclampsia. The reference lists of all known primary and review articles were examined to identify cited articles not captured by electronic searches. Language restrictions were not applied. A comprehensive database of relevant articles was constructed.

Studies which evaluated the accuracy of proteinuria in women with pre-eclampsia for the prediction of maternal or fetal complications were selected in a two-stage process. We included studies that pre-specified the patients to have pre-eclampsia, used bedside (urine dipstick) or laboratory methods (24-hour protein estimation, urine protein:creatinine ratio) to measure levels of proteinuria and assessed maternal or fetal clinical complications as outcome. First, the electronic searches were scrutinised and full manuscripts of all citations that were likely to meet the predefined selection criteria were obtained. Second, final inclusion or exclusion decisions were made by the reviewers (ST and FOM) after examination of these manuscripts. Studies which met the predefined and explicit criteria regarding population, tests, outcomes and study design (Additional file 2) were selected for inclusion in the review. When disagreements occurred, they were resolved by consensus (ST, KI and FOM). In cases of duplicate publication, the most recent and complete versions were selected. There were no language restrictions.

Information was extracted from each selected article on study characteristics, quality and accuracy results. Accuracy data were used to construct 2 × 2 tables of proteinuria result (test positive if proteinuria levels were above the threshold defined in the primary study, and test negative if these were below the threshold) and maternal and fetal outcomes.

All manuscripts meeting the selection criteria were assessed for their methodological quality. Quality was defined as the confidence that the study design, conduct and analysis minimised bias in the estimation of test accuracy. Based on existing checklists [13, 14], quality assessment involved scrutinizing study design and relevant features of the population, test and outcomes of the study. A study was considered to be of good quality if it used a prospective design, consecutive enrolment, full verification of the test result with reference standard, and had adequate test description. We excluded studies with case-control design as these are known to result in substantial bias [13].

Likelihood ratios (LRs) for positive and negative test results were calculated for each study, separately for each test threshold. Summary LRs were then computed where appropriate for positive and negative test results for each individual test threshold and for each outcome of interest. This information is clinically more relevant than traditional summaries of accuracy such as sensitivity and specificity, as LRs allow the estimation of post-test probabilities of various complications for women at different risk levels [15]. The LR indicates by how much a given test result raises or lowers the probability of having the disease. The higher the LR of an abnormal test, the greater is the value of the test. Conversely, the lower the LR of a normal test, the greater is the value of the test. An LR of >10 or <0.1 is regarded as 'very useful' test accuracy, whilst an LR of 5 to 10 or 0.1 to 0.2 is regarded as 'moderately useful', and an LR of 2 to 5 or 0.2 to 0.5 is regarded as 'somewhat useful'. An LR of 1 to 2 or 0.5 to 1 is only regarded as 'little useful' and an LR of 1 as 'useless'. Although, this categorization is useful for interpretation of LRs, it should be noted that the value of a test may vary depending on the pre-test probability of the condition, and the consequences of treatment.

Heterogeneity of diagnostic odds ratio (dOR) was assessed graphically using forest plot [16] (not shown) and statistically using chi-squared test [17] to aid in decisions on how to proceed with quantitative synthesis [18]. As, for some tests and outcomes, there was either graphical or statistically significant heterogeneity, we used random effects model meta-analysis [17]. Where a quantitative approach was not appropriate due to significant clinical heterogeneity, we refrained from pooling and the results are described narratively and the LR of maternal and fetal complication estimated in each study is reported. All statistical analyses were performed using Meta Disc statistical package.

Figure 1 summarizes the process of literature identification and selection. There were 16 primary articles that met the selection criteria including a total of 6749 women (Figure 1) [1, 19‐34]. Eight articles reported estimation of proteinuria by laboratory method only [20, 21, 23, 26, 28, 29, 33, 34], five by bedside dipstick urinalysis only [22, 24, 25, 30, 32], two by either of the above methods [27, 31] and one by spot urine protein:creatinine ratio [1]. The salient features (population subgroups, test characteristics and reference standards) of each individual study can be obtained from the authors. The definition of pre-eclampsia differed widely between the studies. The test threshold in individual studies for laboratory estimation varied from 0.3 g/24 h to 10 g/24 h, or was reported as an increase in proteinuria by 2 g/24 h between two measurements. The cut-off levels for bedside urinalysis using visual reagents ranged from 1+ to 4+ of proteinuria. One study evaluated the accuracy of spot urine protein: creatinine ratio for threshold levels of 500 mg/mmol and 900 mg/mmol in the prediction of maternal and fetal complications [1]. The methodological quality of the included studies is given in Figure 2.

Three test accuracy studies evaluated the accuracy of proteinuria in predicting eclampsia for cut off levels of 5 g/24 h, 10 g/24 h and an increase by 2 g in 24 hours [23, 26]. The LR of positive test ranged from 1.7 (95% CI 0.94, 3.1) to 2.7 (95% CI 1.1, 6.2) respectively. The negative LR ranged from 0.41 (95% CI 0.04, 4.5) to 0.62 (95% CI 0.28, 1.4) (Figure 3).

Two studies estimated the accuracy of proteinuria in predicting placental abruption using a cut-off of increase in level more than 2 g/24 h [23, 29]. The pooled positive and negative LRs for the above cut-off were 0.88 (95% CI 0.42, 1.86) and 1.1 (95% CI 0.75, 1.6) respectively (Figure 3).

HELLP syndrome prediction was evaluated in four test accuracy studies [23, 26, 29]. The pooled estimate of LRs for positive and negative test for a cut-off of increase in levels more than 2 g in 24 hours were 0.86 (95% CI 0.38, 2) and 1.1 (95% CI 0.74, 1.6) respectively (Figure 3).