Dipstick proteinuria: caveat emptor
Apart from the measurement of blood pressure, analysis of urine by dipstick for proteinuria is the most commonly performed antenatal screening test. The presence of proteinuria is central to the diagnosis of pre-eclamp-sia in a hypertensive pregnancy; is an important marker of the seventy or progression of the disease; and is a component of classification systems for hypertensive pregnancy.
The classification of hypertensive disorders during pregnancy is an obstacle to research in this life threatening condition1–3. Integral to the classification is the definition of proteinuria and hypertension. Although the definition of hypertension within the context of pregnancy has long been the subject of controversy, for many obstetricians the detection of significant proteinuria in a hypertensive pregnancy is regarded as a sinister finding. In pregnant women with mild chronic hypertension but no proteinuria, the outcome of pregnancy is similar to non-hypertensive pregnant women4. However, hypertension with proteinuria is associated with poor fetal outcome, particularly when they occur remote from term. It is associated with an increased rate of small for gestational age pregnancies, perinatal mortality and poorer maternal prognosis5–10. Of women with chronic hypertension who also developed proteinuria, 10%-20% had a poor pregnancy outcome, with a 10% incidence of abruptio placentae, a 33% incidence of intrauterine growth restriction, and a perinatal mortality rate of 24%4. When severe hypertension is associated with significant proteinuria (>5 g in 24 h) delivery is usually indicated within 2–3 weeks6.
Dipstick urinalysis and assessment of proteinuria
Despite the prognostic role of proteinuria, it could be argued that the assessment of proteinuria in present routine obstetrical practice has taken a backward step. Although it is accepted that proteinuria is assessed most appropriately by the biochemical measurement of total protein excretion over a 24-hour period, and that this assessment is the standard test in hospital laboratories, the most commonly employed front-line method to screen for proteinuria antenatally is semi-quantitative dipstick urinalysis. This test has not been validated clinically and is performed in a largely unsupervised fashion by uninstructed medical staff, midwives, nursing auxiliaries or medical students. For most maternity units, a negative dipstick for proteinuria is synonymous with the absence of proteinuria. This is reflected by a recommendation to obstetricians that ‘in the absence of proteinuria on reagent strip testing, there is no need to carry out 24-hour estimation of proteinuria’11.
The primary reason for the dependence upon dipstick urinalysis, performed upon randomly voided urines, is that results of these tests are considered to agree with the results of total protein measurements performed at 24-hour urine collections. Thus, proteinuria may be defined with respect to a score on the dipstick test [e.g. ≥+1 ≥ 0.3 mg/mL)], and a level of total protein concentration in 24-hour urine collections [e.g. 2 0.3 g total proteid 24 hours]. Based on dipstick urinalysis, women are classified as nonproteinuric or proteinuric, and are grouped together irrespective of the test employed. The ease of the methodology involved in the dipstick analysis of a randomly voided urine sample, compared with the more laborious 24-hour urine collections for measurement of total protein, inevitably has led to the routine Introduction of and dependence upon the dipstick to assess proteinuria.
Dipstick urinalysis versus 24-hour urinary protein excretion
Recent reports indicate that these different definitions of significant proteinuria, based upon differing methodologies and types of urine specimens, do not agree12–14. Disagreement exists when the diagnosis of proteinuria in hypertensive pregnant women using dipstick analysis was compared with the diagnosis based on the standard total protein excretion in 24 hours.
Kuo et al.12 compared the dipstick diagnosis of significant proteinuria in 24-hour urine collections with dipstick urinalysis in a group of hypertensive pregnant women and found poor agreement: there was a wide range of total protein values for urine score at+1 on dipstick analysis13. In 1994 Meyer et al.13 conducted a retrospective study of 300 urine samples from hypertensive pregnant women and reported that 66% of the women with a negative or trace dipstick result had significant proteinuria, defined as 20–3g total proteid 24 hours13. They also reported a significant false positive rate of 26% with a +1 dipstick result. We have recorded14 a similar high false negative rate in 500 urines from hypertensive pregnant women of whom 66% with negative or trace dipstick results were actually proteinuric. In a prospective study Brown and Buddle15 found that dipstick testing produced false negative results of 8%-18% and a very high false positive rate of 67% in urines with +I scores. These studies all question the value of random semi-quantitative dipstick analysis in the diagnosis of proteinuria in pregnancy.
False positive results may result in over-investigation and possible over-intervention, while false negative results place women and their infants at risk. There are several reasons for such disagreements, including observer error, the characteAstics of the semi-quantitative dipstick tests, the units of protein estimation, the urine specimens employed and the biochemical methods upon which the tests rely.
Dipstick urinalysis and the observer error
Observer error appears to be a significant source of the inaccuracy of dipstick analysis. The type of staff employed to perform urinalysis, their training and the environment in which they read the test can affect the outcome of the reading. In our study, published in this issue of the Journal on pages 1177–1180, nursing auxiliaries recorded a higher number of false positives than midwives (47%versus 17%, respectively) when performing dipstick analysis for proteinuria on standard human serum albumin solutions at ‘nonproteinuric’ concentrations. We investigated whether this inaccuracy may be due to lack of training. In a sequential study, untrained laboratory staff initially achieved a false positive rate of 35%. Following minimal education in the methodology and interpretation of the results of dipstick urinalysis, the laboratory staff achieved substantial improvement in their accuracy levels (a false positive rate of 5%). A similar improvement may be achieved by the Introduction of automated devices. Saudan et al.16 recently reported reducing a high false positive rate for visual dipstick urinalysis using an automated device. The persistent false positive rates in this later study were accounted for by the authors' interpretation that the dipstick was oversensitive at the +1 threshold16. Although in our study we decreased our false positive rate, we could not account for the persistent high false negative rates of 40%45%, and we interpreted this as our dipsticks being under-sensitive at the +1 threshold. These inaccuracies will not be influenced by training or by automated devices, but by a re-examination of the threshold sensitivities of the dipsticks currently available.
Urine specimen and concentration versus content
Apart from the practical aspects of dipstick measurement, however, the dipstick method is performed upon random antenatal urine specimens and yields a measure of the concentration of protein in a given specimen and is, therefore, affected by a number of variables. The concentration of protein in an individual urine specimen may be highly variable and is influenced by several factors, including contamination (false positive result), exercise (increased excretion), posture, osmolality, and urinary pH. The majority of reported studies do not comment on either urine specific gravity or pH when dipstick methods are used. In our experience, the concentration of protein in individual urine specimens from hypertensive women is very variable over a 24-hour period, changing from negative or trace to +1 to +3, (i.e. nonproteinuric to proteinuric). This variation does not affect the standard method of 24-hour estimation, where the total amount of protein excreted over a relatively long standard time is measured.
A second problem is the differing threshold in the dipstick test (i.e. the normally employed negative to +1 boundary with its indicator of significant proteinuria of 0.3 mg/mL), whereas in the standard assay the accepted threshold for proteinuria is 0.3 g total protein in 24-hour collections of urine. The assumption that the dipstick test and the standard assay are the same is based upon the equivalence of these thresholds. However, these only equate if the average 24-hour urine collection is 1000 mL (i.e. 0.3 mg/mL × 1000 mL (24 h = 0.3 g/24 h). In fact, in our recently completed study of 500 hypertensive pregnancies this value is closer to 1.5/L24 h. This may contribute to the false negative rate observed with dipstick tests.
Specificity of the protein assay and urinary protein composition
That there is no perfect method to measure ‘total’ protein concentration is often overlooked. There are a variety of total protein assay methods used to measure protein concentration, and these methods exploit specific properties of proteins. However, individual proteins may possess these properties to differing degrees, and therefore assays may exhbit differential protein specificity when applied to a heterogeneous mixture of proteins. For example, the benzethonium chloride protein assay, which is widely employed to assay the total protein content of 24-hour urine specimens, detects IgG whereas the dipstick assay does not measure IgG. Depending on the varying methodology employed, different estimates of the level of proteinuria may be generated. It is therefore necessary to standardise the type of protein assay employed in hospital laboratories. The problem is compounded if there are alterations in urinary protein composition associated with subgroups of hypertensive pregnancy, because a particular protein assay method may not detect a rise in the concentration of proteins which may be characteristic of a particular subgroup of women with hypertension in pregnancy.
It is known that the pattern of urinary protein composition, as revealed by differing ‘fingerprints’ on electrophoretic gels, in patients with diseases affecting kidney function can reflect the location of the damage (i.e. glomerular or tubular damage). Such patterns have been reported in pregnant women, and those with pure glomerular or mixed glomerular and tubular pro-teinuria have been reported to be at a higher risk of developing severe pre-eclampsia17. Studies of the urinary protein patterns associated with pre-eclampsia have identified a significant reduction, down to unde-tectable levels, in the concentration of a Tamm-Hors-fall glycoprotein and a reduction in the number of low molecular protein bands18. These changes in the urinary electrophoretic pattern have been shown to be associated with the severity of the disease19. These findings demonstrate the importance of the type of assay employed, because ‘proteinuric’ hypertensive women may not all excrete similar types of proteins, and various assays may either under- or over-diagnose the degree and type of proteinuria. As yet no specific pattern of proteinuria has been identified which corresponds to the characteristic pre-eclamptic pathology of glomerulo-endotheliosis. The identification of such a specific pattern may be of great importance, since the type of proteinuria rather than the amount of proteinuria may be of greater prognostic significance. In severe pre-eclampsia it has been reported that increasing or decreasing proteinuria irrelevant to the outcome20.
Several sources of error militate against accurate dipstick urinalysis. Although the importance of the finding of proteinuria in antenatal care is acknowledged, the process for its detection in the average antenatal clinic is not robust and does not withstand close scientific scrutiny. The widely accepted assumption that proteinuria is present only when dipstick analysis is positive ignores the substantial evidence to the contrary and so puts women with hypertension at risk. The advent of dipstick technology has by-passed accurate quantitative measurement of urinary protein, based on the false perception of few false negative and positive results. An educational programme to familiarise clinical personnel in its use, an awareness of the deficiencies associated with dipstick urine analysis and of automated devices can only improve current practice. However, it is possible that a reappraisal of the definition of clinically significant proteinuria in hypertensive pregnancy, involving reassessment of the most appropriate urine specimen to be tested, identification of a suitable threshold, and consideration of protein specificity, may require wholesale re-evaluation of the currently available dipsticks. In this context, current optimal assessment of hypertensive pregnancies requires estimation of total protein in a 24-hour collection of urine.