A prospective study of the impact of automated dipstick urinalysis on the diagnosis of preeclampsia.

Objective: To determine prospectively in hypertensive pregnant women 1) the accuracy of dipstick testing for proteinuria using automated urinalysis, 2) factors that might affect such accuracy, and 3) the potential impact of automated dipstick testing on the accuracy of diagnosis of preeclampsia according to acceptance of proteinuria at either 1 + or 2 + level.

Design: Prospective study.

Setting: Antenatal day assessment unit and antenatal ward of St George Hospital, a teaching hospital in Sydney, Australia.

Population: 170 hypertensive pregnant women attending as outpatients or inpatients.

Methods: 503 midstream urine samples were collected prospectively on separate occasions from 170 women. Full urinalysis was recorded using the Bayer Clinitek 50 automated urinalysis device and Multistix 10SG urinalysis strips (Bayer Diagnostics, Victoria, Australia). Each MSU was analysed for spot protein/creatinine ratio and also for culture and sensitivity if symptoms of a urinary tract infection were present or dipstick included positive nitrites. Urinalysis protein results were compared with spot urinary protein/creatinine ratio (previously shown to correlate with 24-hr urine protein excretion) to determine the accuracy of urinalysis. True proteinuria was defined as a ratio >/= 30 mg protein/mmol creatinine.

Results: False positive dipstick tests ranged from 7% at 3 + level to 71% at 1 + proteinuria level while false negative rates were 7% for "nil" and 14% for "trace" proteinuria, 9% overall. Accepting the dipstick proteinuria result at face value led to an incorrect diagnosis of preeclampsia or gestational hypertension in 85 (50%) women. Dipstick proteinuria was significantly more likely to be correct (true positive/true negative) if diastolic blood pressure was elevated > 90 mmHg (p = 0.032) and in the absence of ketonuria (p = 0.001). Accepting a diagnosis of preeclampsia on the basis of de novo hypertension and dipstick testing alone was accurate less often (70%) when > 1 + was used as a discriminant value than at the 82% of presentations when > 2 + was used (p = 0.001).

Conclusion: Accepting "nil" or "trace" proteinuria as a true negative dipstick results fails to identify approximately 1 in 11 hypertensive pregnant women with true proteinuria, a false negative rate that may be acceptable provided these women are subject to ongoing vigilant clinical review. Even with automated urinalysis the false positive rate for dipstick levels >/= 1 + is very high, particularly in the presence of ketonuria and relying on this alone to diagnose preeclampsia leads to significant errors in diagnosis. Accepting >/= 2 + dipstick proteinuria improves overall diagnostic accuracy for preeclampsia at the expense of a higher false negative rate. This study emphasizes the need to confirm dipstick proteinuria with a further test such as a spot urine protein/creatinine ratio in all hypertensive pregnant women, particularly in research studies.