Original Contribution

Can Urine dipstick predict an elevated serum creatinine?

Kaushal Shah MD?, Barbara Kilian MD, Wei-Jen Hsieh, Emily Kyrillou, Vishal Hedge, David H. Newman MD

St. Luke’s-Roosevelt Hospital Center, Columbia University, New York, NY 10025, USA

Received 29 January 2009; revised 3 February 2009; accepted 4 February 2009

Abstract

Objective: Chart review studies have suggested that point-of-care urine dipstick testing may accurately predict an elevation in serum creatinine (Cr). We aimed to prospectively evaluate the test characteristics of proteinuria/hematuria in predicting elevated serum Cr.

Methods: A prospective, observational study was conducted between March 2007 and June 2008 at 2 affiliated, urban hospitals with an annual emergency department census of 150 000. Patients undergoing laboratory urinalysis, point-of-care urine dipstick, and a serum chemistry panel were enrolled. Trained research assistants collected data on consecutive patients 18 hours per day using preformatted data forms and entry into an anonymized Access (Microsoft, Seattle, Wash) database. Demographic baseline variables including age, sex, chief complaint, vital signs, and source of sample (catheter vs “clean catch“) were also collected. An elevated Cr level was defined as greater than 1.3 based on the laboratory reference range. Standard statistical methods were used to calculate diagnostic test operating characteristics of proteinuria or hematuria as a predictor of elevated serum Cr.

Results: Five thousand four hundred sixteen subjects were enrolled with 28.3% male and a mean age of

50.2 years. Elevated serum Cr greater than 1.3 mg/dL was found in 13.9% (755/5416) of subjects. The sensitivity of either proteinuria or hematuria for elevated Cr was 82.5% (95% confidence interval [CI], 80%-85%) and specificity was 34.4% (95% CI, 33%-36%). Positive predictive value was 16.9% (95% CI, 16%-18%) and negative predictive value was 92.4% (95% CI, 91-94%). The likelihood ratio for a positive test was 1.3 (95% CI, 1.1-1.5), and the likelihood ratio for a negative test was 0.5 (95% CI, 0.3-0.8). Conclusions: Although negative predictive value was high, the presence of proteinuria/hematuria was only moderately predictive of elevated serum Cr level.

(C) 2010

Introduction

Identification of patients with renal insufficiency has important implications for emergency management, particu- larly in the setting of potentially life-threatening conditions requiring iodinated intravenous contrast media for rapid diagnosis. Despite advances in radiographic imaging that have led to faster, more detailed, and more accurate anatomic

* Corresponding author. Tel.: +1 646 369 2747; fax: +1 212 523 2186.

E-mail address: [email protected] (K. Shah).

diagnosis, serum creatinine (Cr) levels often continue to delay imaging because of a higher risk of Contrast-induced nephropathy in those with renal insufficiency [1]. This is a particular problem with many life-threatening, time-sensitive diagnoses (eg, Mesenteric ischemia, aortic dissection, and bowel perforation) because delay in performing contrast- enhanced computed tomography (CT) imaging may lead to a Delay in diagnosis and treatment resulting in increased morbidity and mortality.

Although extensive research has been done on prophy- lactic pretreatment methods, contrast types, and the impact of

0735-6757/$ – see front matter (C) 2010 doi:10.1016/j.ajem.2009.02.020

intravenous hydration, there has been little published research on the diagnostic utility of urinalysis results (specifically the presence of hematuria and/or proteinuria) as indicators of renal insufficiency. With rising health care costs to patients, and emergency department (ED) over- crowding, an efficient and effective screening test could potentially reduce both time in the ED, and perhaps most importantly, time to accurate diagnosis and management.

In this study, we sought to prospectively test the diagnostic performance of abnormalities found on urine dipstick in identifying elevated serum Cr levels. We hypothesized that urine dipstick test results could indicate elevated serum Cr level with reasonable and useful sensitivity, specificity, and likelihood ratios.

Methods

Study design

This is a prospective, observational study that was approved by the institutional board review with a waiver of informed consent.

Study setting and population

This study was conducted between March 2007 and June 2008 at 2 affiliated, urban hospitals with a combined ED census of 150 000 patients per year. Adult patients under- going laboratory urinalysis, point-of-care urine dipstick, and a serum chemistry panel (regardless of chief complaint or final diagnosis) were enrolled.

Study protocol

Trained research assistants collected data on consecutive patients 18 hours per day. Research assistants reviewed all digital medical records on patients in the ED. Laboratory tests and point-of-care urine dipsticks are routinely entered into the digital medical record. Patients with a urine dipstick, Urine analysis, and a Cr level entered into the medical record were enrolled in the study. Demographic baseline variables including age, sex, chief complaint, vital signs, and source of sample (catheter vs “clean catch”) were also collected. Information was immediately entered into an anonymized Access (Microsoft, Seattle, Wash) database.

For the purposes of enrolling all patients, including those unable to give consent or authorization, “minimal risk” criteria for waiver of informed consent were met and medical record information was formally de-identified. Although a consecutive sample was sought, formal confirmation of the consecutiveness of the sample or cross-checking methods was not feasible because of formal de-identification procedures (specific discarding of unique identifiers) under-

taken during data acquisition and database entry. In the absence of evidence to the contrary, we therefore consider the sample to be one of convenience. However, investigators conducted frequent (multiple times per day on average) on- site departmental checks and conducted direct supervision of research assistants and enrolling physicians. Eligible patients who were not enrolled were infrequently noted, although no formal cross-checking procedures were undertaken.

A standardized data form was used. Demographic data, medical history, and serum laboratory values were obtained from both the ED chart-tracking and electronic charting systems. Training on completion of data forms, digital medical record data extraction, and database entry were completed during multiple 1-hour training sessions with all research assistants. In addition, digital and paper documents explaining each step of study enrollment were provided and available to all research assistants. Ongoing review of data sheets was completed approximately biweekly throughout the study period, and individual training issues were remediated as needed. When Cr values were noted to be greater than 1.3 (out of the institutional laboratory’s reference range), research assistants brought data forms to the treating physician for formal confirmation of the abnormal laboratory finding. This step served the dual purpose of ensuring a correctly transcribed value (physicians were asked to check and reconfirm the Cr result documented in the laboratory) and notification of the treating team of this abnormality.

All data were recorded on a numbered data form in the ED. After completion of the form, all data were entered into an Access 2000 database in the ED by the research assistant or physician completing the form. Forms were then placed in a locked research box and were collected weekly by investigators who matched form numbers to database entries. Any non-entered forms were entered into the database by the investigators.

Definitions

An elevated Cr level was defined as greater than 1.3 based on the hospital laboratory reference range. The presence of any amount of proteinuria or hematuria was considered positive.

Data analysis

Standard statistical methods were used to calculate diagnostic test operating characteristics of proteinuria or hematuria as a predictor of elevated serum Cr. Sensitivity, specificity, and likelihood ratios were calculated with 95% confidence intervals (CIs). Logistic regression analysis was performed on an additional, predetermined, prospectively collected sample (n = 424) for statistical association between Cr elevation and demographic and medical history variables among the population.

Results

Five thousand four hundred sixteen subjects were enrolled with males accounting for 28.3% of the population. The mean age was 50.2, and most Urine samples (84.9%; 95% CI, 84%-86%) were obtained as clean-catch specimens as opposed to via straight or Foley catheter.

Elevated serum Cr greater than 1.3 mg/dL was found in 13.9% (755/5416) of subjects. The sensitivity of either proteinuria/hematuria for elevated Cr was 82.5% (95% CI, 80%-85%) and specificity was 34.4% (95% CI, 33%-36%).

Positive predictive value was 16.9% (95% CI, 16%-18%)

and negative predictive value was 92.4% (95% CI, 91%- 94%). The likelihood ratio for a positive test was 1.3 (95% CI, 1.1-1.5), and the likelihood ratio for a negative test was

0.5 (95% CI, 0.3-0.8).

Elevated serum Cr greater than 1.5 was found in 10.2% (553/5416) of subjects and greater than 2.0 was found in 6.4% (344/5416) of subjects. The test characteristics of urine dipstick for the various Cr cutoffs are in Table 1. Included in Table 1 are the test characteristics of the urine dipstick for a Cr more than 1.3 for subjects older than 50 years as well.

Regression analysis was performed on variables including history of diabetes, history of hypertension, age (cutoffs above and below 50, above and below 65, and age as a continuous variable), and history of renal insufficiency. Only a reported history of renal insufficiency demonstrated statistically significant association (df = 7, -2 log-likelihood ?2 = 83.97, P b .0001) with Cr elevation.

Discussion/clinical relevance

Many radiology departments are unwilling to perform CT imaging with intravenous contrast without an evaluation of Cr (surrogate marker for Cr clearance) because elevated Cr levels are associated with an increase chance of contrast- induced nephropathy, which is the third leading cause of acute renal failure among hospitalized patients [1,2]. Emergency physicians use contrast-enhanced CT imaging

Table 1 Test characteristics of urine dipstick for elevation in Cr

Cr N1.3 Cr N1.5 Cr N2.0 Cr N1.3 and age N50

Subjects 755 (13.9%) 553 (10.2%) 344 (6.4%) 579 (24.7%)

Sensitivity 82.5 85.4 91.3 81.9

Specificity 34.4 34.0 33.6 33.4

PPV 16.9 12.8 8.5 28.7

NPV 92.4 95.3 98.3 84.9

LR+ 1.26 1.29 1.38 1.23

LR- 0.51 0.43 0.26 0.54

PPV indicates positive predictive value; NPV, negative predictive value; LR, likelihood ratio.

on a regular basis. In most cases, the delay in waiting for a Cr level is not critical. However, when life-threatening, time- sensitive diagnoses are possible, the delay may affect patient outcomes.

The urine dipstick as a surrogate marker for Cr has been evaluated retrospectively. A 2-year retrospective study with more than 2400 enrollments by Firestone et al (2007) determined the sensitivity and negative predictive value for abnormal urine dipstick in detecting elevated Cr were 85.5% and 96.2%; subjects with a negative urine dipstick but elevated Cr all reported significant medical history: prior renal disease, hypertension, diabetes, congestive heart fail- ure, or age of more than 60 years [3]. In another study of 310 patients who had no protein in their urine and no history of disease that potentially could cause renal insufficiency (hypertension, diabetes mellitus, multiple myeloma, or systemic lupus erythematosus), none had a serum Cr concentration greater than 2.0 mg/dL [4].

The most important test characteristics of proteinuria/ hematuria for predicting an elevated Cr are sensitivity and negative likelihood ratio. Ideally, a patient with an elevated Cr would always have protein or blood on urine dipstick testing. Emergency physicians and radiologists might then feel comfortable administering intravenous contrast based on a urine dipstick without waiting for a confirmatory serum Cr level. We, however, determined a sensitivity of 82.5%. This means that approximately 1 in 6 patients with an elevated Cr in our cohort had no protein or blood in their urine. The negative likelihood ratio for Cr of greater than 2.0 was 0.26; although not less than 0.1 (the commonly accepted ideal for the test), it still provides evidence against disease and may be clinically useful.

An alternative perspective may be considered in inter- preting these data. The negative predictive value was 92.4%, suggesting that if the urine dipstick was negative, then the chance of administering intravenous contrast to a patient with an elevated Cr was low in our cohort, less than 8%. The likelihood of clinically important harm (ie, temporary or permanent renal failure requiring treatment or mortality) secondary to radiographic contrast is low, estimated to be as low as 1% and as high as 20% based predominantly on intravenous contrast given for percutaneous coronary inter- ventions [5-8]. If we aggregate the likelihood of harm into a very conservative estimate of approximately 10%, it would mean that fewer than 1% of those exposed to contrast dye with a urine dipstick test negative for blood and protein would be expected to experience a harmful outcome due to contrast dye exposure (10% of 7.5% = 0.75%). This risk calculation allows one to compare the possibility of harm due to a delay in diagnosis to the possibility of harm due to contrast dye and to undertake a discussion of risks with radiology consultants and patients.

Of note, although Cr clearance is a better marker for renal insufficiency, the most commonly used Cr cutoff for administration of intravenous contrast is 1.5 [9]. For this reason, we calculated test characteristics for the urine

dipstick to predict elevated Cr at various cutoffs. Using a higher Cr cutoff only improves the sensitivity and negative predictive value of the urine dipstick.

Limitations

Proteinuria/hematuria on dipstick testing was used as a surrogate to predict serum Cr, which is a surrogate for Cr clearance or glomerular filtration rate (GFR). Although Cr clearance and GFR calculations are not typically done to determine the risk for contrast-induced nephropathy, it is the true measure that is important. We did not calculate either of these measures in our preliminary study. Ideally, future studies will attempt to correlate proteinuria/hematuria with either Cr clearance or GFR.

We collected limited data on medical history. It is possible that a combination of urine dipstick and elements of the medical history other than those that we examined (renal insufficiency, hypertension, and diabetes) could yield higher sensitivity and negative predictive values for prediction of an elevated Cr. Our analysis suggests, however, that these variables, other than a history of renal insufficiency, will be of limited value when added to urine dipstick results.

Finally, neither the accuracy of interpretation by our ED nursing and support staff nor the accuracy of urine dipstick tests in general for the detection of protein and blood were measured or considered in this analysis. Our urine reagent strip testing and reporting is monitored and our equipment maintained and tested regularly in accordance with common laboratory regulation standards for hospital-based point-of- care testing, and our personnel are trained and retrained on a regular basis. There remains however a potential for error that may be difficult to quantify in these processes.

Conclusion

The presence of proteinuria/hematuria was only moder- ately predictive of serum Cr level, although negative

predictive value was high enough to suggest that risk/benefit estimations can, in some cases, favor the use of urine dipstick as a rapid and useful screen for elevated Cr. The sensitivity and negative predictive value of urine dipstick increase when using a higher Cr cutoff.

References

1. Sinert R, Doty C. Prevention of contrast-induced nephropathy in the emergency department. Ann Emerg Med 2007;50:335-45.
2. Tumlin J, Stacul F, Adam A, Becker CR, Davidson C, Lameire N, et al, CIN Consensus Working Panel. Pathophysiology of contrast-induced nephropathy. Am J Cardiol 2006;98(6A):14K-20K [Electronic publica- tion 2006 Feb 17. Review].
3. Firestone D, Wos A, Killeen JP, Chan TC, Guluma K, Davis DP, et al. Can urine dipstick be used as a surrogate for serum creatinine in emergency department patients who undergo contrast studies? J Emerg Med 2007;33(2):119-22 [Electronic publication 2007 Jun 13].
4. Glauser J, Montgomery A. Urine protein as a rapid screen for renal function in the ED: can it replace serum creatinine in selected patients? Emerg Radiol 2004;10(6):319-22 [Electronic publication 2004 May 5].
5. Marenzi G, Lauri G, Assanelli E, Campodonico J, De Metrio M, Marana I, et al. Contrast-induced nephropathy in patients undergoing primary angioplasty for acute myocardial infarction. J Am Coll Cardiol 2004;44 (9):1780-5.
6. McCullough PA, Wolyn R, Rocher LL, Levin RN, O’Neill WW. Acute renal failure after coronary intervention: incidence, risk factors, and relationship to mortality. Am J Med 1997;103(5):368-75.
7. Rihal CS, Textor SC, Grill DE, Berger PB, Ting HH, Best PJ, et al. Incidence and Prognostic importance of acute renal failure after percutaneous coronary intervention. Circulation 2002;105(19):2259-64.
8. Roghi A, Savonitto S, Cavallini C, Arraiz G, Angoli L, Castriota F, et al, Atherosclerosis, Thrombosis and Vascular Biology Study Group and the Italian Society for Invasive Cardiology Investigators. Impact of acute renal failure following percutaneous coronary intervention on long-term mortality. J Cardiovasc Med (Hagerstown) 2008;9(4):375-81.
9. Band RA, Gaieski DF, Mills AM, Sease KL, Shofer FS, Robey JL, et al. Discordance between serum creatinine and creatinine clearance for identification of ED patients with abdominal pain at risk for contrast- induced nephropathy. Am J Emerg Med 2007;25(3):268-72.