Diagnostic accuracy of urine dipstick tests for proteinuria in pregnant women suspected of preeclampsia: A systematic review and meta-analysis

Objectives: Dipstick tests are frequently used as bedside proteinuria tests to evaluate women suspected of preeclampsia and may inform diagnosis in low resource settings lacking laboratory facilities. This systematic review and meta-analysis aimed to (1) estimate the diagnostic accuracy of urine dipsticks in diagnosing proteinuria, (2) compare performance of different dipstick types and (3) estimate their related costs. Methods: MEDLINE and EMBASE were searched up to August 1, 2020 for primary studies with cross-sectional diagnostic accuracy data on dipstick test(s) compared to a laboratory reference standard (24-hour protein > 300 mg or protein-creatinine ratio > 30 mg/mmol) in pregnant women > 20 weeks of gestation suspected of preeclampsia. Risk of bias and applicability was assessed with QUADAS-2. Data were analysed using a bivariate model with hierarchical addition of covariates for subgroups. Results: Nineteen studies were included. Protein-only dipsticks at 1 + threshold had a pooled sensitivity of 0.68 [95%CI: 0.57-0.77] and specificity of 0.85 [95% CI: 0.73-0.93] (n = 3700 urine samples, 18 studies). Higher specificity was found with automatedly (0.93 [95% CI: 0.82-0.98]) compared to visually (0.81 [95% CI: 0.65-0.91]) read dipsticks, whereas sensitivity was similar and costs were higher. The use of albumin-creatinine ratio (ACR) dipsticks was only reported in two studies and did not improve accuracy. Heterogeneity in study design and prevalence of preeclampsia amongst studies complicated interpretation of pooled estimates. Conclusion: Urine dipsticks performed poorly at excluding preeclampsia in hypertensive pregnant women. Further development of accurate and low-cost bedside proteinuria tests is warranted.