Critically Appraised Topic

Michael Steiner, MD

November 2, 2001

Case Description: Female patient who initially presented with LE edema, gross hematuria and an active urinary sediment. After initial presumptive diagnosis, disease progression led to pancytopenia, decreased creatinine clearance, and a spot urine protein/creatinine ratio of 8. During discussion of the case, residents ask how accurate spot urine evaluation was in comparison to 24-hour collection.

Clinical Question: In patients with proteinuria, does a spot urine protein/creatinine ratio accurately reflect the amount of protein excreted in a 24-hour period?

Search: The English language literature was searched for human studies via Pubmed. Search terms used were protein and urine and creatinine and spot and protein and urine and creatinine and single. These searches revealed 63 and 273 articles respectively. Eight of these articles were directly related to the clinical question. Seven articles all referred back to an older 8^th article. This paper was then chosen for further review.

Ginsberg, JM et al. Use of single voided urine samples to estimate quantitative proteinuria. N Engl

J Med 1983; 309: 1543-6.

Methods: 46 patients being followed at Rhode Island Hospital Renal Clinic were asked to collect 24 hr urine samples as outpatients. Each void during that day was collected in a separate container. A spot urine pr/cr ratio was calculated for each sample and then the samples were combined and a 24-hour urine protein was determined. When patients went to clinic with their samples a final urine was also collected for analysis. Limited information is presented about the selection and characteristics of patients. Mean age was 43 with a range of 13 to 76 years. Renal function was proclaimed ‘stable’ based on serial creatinine measurements. 57% had serum creatinines of <1.5, 28% had creatinines between 1.5 and 4.0, and 15% had serum creatinine stable at >4.0. No other baseline information was presented.

A second group of urine samples was also collected from a normal control group of 30 people who had no known renal disease. Their ages ranged from 26 to 57 and they were 66% women.

Results: Various analyses were carried out on the data. First of all, the urine sample obtained at the clinic was compared to the sum of the protein in the 24-hour collection. Though not clearly stated, it appears the data was analyzed by linear regression analysis generating a slope, Y-axis crossing point, and correlation coefficient (Figure 1). This revealed two highly correlated tests (.97) with a slope of the line of nearly 1 (1.13).

Next the authors took the individual samples obtained during the 24 hr. collection and ran spot Pr/Cr ratios on them. These were then categorized by the time of day collected and regression analysis was performed to compare 24hr collection to spot ratios at various points during the day. Figure 2 attempts to graphically represent the slopes of the spot urine pr/cr ratio through different times of the day. I believe this figure is mislabeled, the letters at the bottom should read A (first morning urine), B, C, D, and E last (urine voided after midnight but before A). This would agree with what is described in text regarding the close correlation of spot samples obtained during the day with the corresponding 24 hr samples. Samples obtained during times of predominant recumbency had significantly lower slopes in the regression analysis. This suggests that spot urines in those conditions underestimate the total 24 hr urinary protein excretion (i.e. less proteinuria when supine or prone vs. standing and active).

Lastly, the urine samples from the control group were also analyzed. All of the controls had protein/creatinine ratios within the normal range, 0.2 or less. 24 hr urine was not collected from this sample of patients.

Reanalysis of the Data:

Using the data points presented in Figure 1. I attempted to find a protein/creatinine ratio that could identify the presence or absence of proteinuria. This was somewhat limited since the actual data values were not presented and therefore I solely used their graphic position.

Urine Pr/Cr ratio of 2 in predicting >2g per 24 hours of proteinuria

Sensitivity: 0.92, Specificity: 0.89, + Likelihood Ratio: 8.4, Negative LR: 0 .09

Urine Pr/Cr ratio of 3.5 in predicting >3.5g/24 hr of proteinuria

Sensitivity: 0.92, Specificity: 0.83, + LR: 5.4, Neg. LR: 0.10

Evidence Based Evaluation (based on criteria from Sackett book):

Validity assessment for studies of diagnostic tests:

1. Was there an independent, blind comparison with a reference standard?

The spot urine pr/cr ratio was compared to the gold standard, however the presence of blinding during calculations can only be assumed.

2. Was the diagnostic test evaluated in an appropriate spectrum of patients (similar to

those on whom it would be used in practice)?

Information on inclusion and exclusion criteria was sparse. Our patient

previously had a renal disorder identified and probably could have been included in the

study group. Note that the patients did not have newly identified renal problems,

therefore it would be hard to extrapolate using spot Ur Pr/Cr as part of new renal

work-up based on this article.

3. Was the reference standard applied regardless of the test result? Yes

4. Was the test validated in a second, independent group of patients?

No. The control group didn’t have the gold standard performed though it is unlikely any had significant proteinuria.

Application of this evidence:

1. Is the diagnostic test available, affordable, accurate, and precise in our setting? Yes

2. Can we generate an accurate pretest probability for our patients?

The problem here is that this patient group is potentially different than a group of primary care patients. Also, the disease possibilities and probabilities may have changed slightly since this data.

3. Will resulting post-test probabilities affect our management? Yes, if proteinuria were

present it would often affect management.