Original Contribution

Is there a link between hyperbilirubinemia and elevated urine nitrite^B

Susan Watts PhD^a,*, David Bryan MD^a, Keith Marill MD^b

^aDepartment of Emergency Medicine, Texas Tech University Health Sciences Center, El Paso, TX 79905, USA

^bMassachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA

Received 5 December 2005; revised 27 June 2006; accepted 2 July 2006

Abstract

Objective: The aim of this study was to determine whether hyperbilirubinemia affects the association between a positive urine nitrite test and a positive urine culture.

Methods: We conducted an institutional review board-approved, retrospective review of 12 months of patient data, compiling information for patients having urinalysis, urine culture, and total Serum bilirubin. Patients were divided into 3 groups according to total serum bilirubin: less than 1.5 mg/dL, 1.5 to 3.0 mg/dL, and greater than 3.0 mg/dL. The point estimates and 95% confidence intervals of the sensitivity, specificity, false-positive proportion (proportion of false positive to all positive tests), and other test characteristics of urine nitrite as an indicator of urinary tract infection were calculated and tested for trend as a function of the 3 total serum bilirubin ranges.

Results: Three thousand one hundred seventy-four patients met our study criteria. Specificity of the nitrite test decreased as a function of increasing total serum bilirubin (0.974, 0.966, and 0.855 for the 3 total bilirubin levels, respectively) with a significant trend ( P b .0001). There was no significant trend in comparable sensitivity values (0.380, 0.417, and 0.241, respectively) with P = .55. The false-positive proportion also increased as a function of total serum bilirubin (17.5%, 17.3%, and 72.0%) with P b

.0001. Thus, if a patient’s total serum bilirubin was elevated to the point of jaundice (N3.0 mg/dL), it was approximately 4 times more likely that a positive urine nitrite test would be a bfalse positiveQ (ie, nitrite-positive/culture-negative) compared with those with normal serum bilirubin levels.

Conclusions: Specificity of the urine nitrite test for urinary tract infection decreases as a function of increasing serum bilirubin. Most patients with hyperbilirubinemia and a positive nitrite test in our sample did not have an associated urinary tract infection.

D 2007

Introduction

Published literature says that the specificity of urine nitrite for urinary tract infection is high, ranging from

^B This research has not been previously presented at a meeting or published in any form. There was no external funding for this project.

* Corresponding author. Tel.: +1 915 545 7333; fax: +1 915 545 7338.

E-mail address: [email protected] (S. Watts).

85% to 99% [1,2]. In spite of this, many physicians in our emergency department (ED) have independently noticed an apparent association between elevated bilirubin and false- positive urine nitrite tests. A typical example is a young man with signs and symptoms of viral hepatitis, who has a positive urine nitrite test with no evidence of a UTI (no white blood cells or bacteria in his urine and a negative urine culture). Such cases are not rare events in our border city ED because the seropositivity rate for hepatitis A in

0735-6757/$ – see front matter D 2007 doi:10.1016/j.ajem.2006.07.002

	n	TP	FP	FN	TN	FPP	SENS	SPEC	PPV	NPV
						(95% CI)	(95% CI)	(95% CI)	(95% CI)	(95% CI)
All patients	10204	964	246	1595	7399	0.203	0.377	0.968	0.797	0.823
with UA/UC						(0.181-0.227)	(0.358-0.395)	(0.964-0.972)	(0.774-0.819)	(0.815-0.831)
Study population	3174	295	79	481	2319	0.211	0.380	0.967	0.789	0.828
						(0.172-0.257)	(0.346-0.414)	(0.960-0.974)	(0.747-0.830)	(0.814-0.842)
Total serum
bilirubin
b1.5 mg/dL 2656		245	52	399	1960	0.175	0.380	0.974	0.825	0.831
						(0.135-0.224)	(0.343-0.418)	(0.967-0.981)	(0.782-0.868)	(0.816-0.846)
z1.5 mg/dL 365		43	9	60	253	0.173	0.417	0.966	0.827	0.808
and V3.0 mg/dL						(0.087-0.308)	(0.322-0.513)	(0.944-0.988)	(0.724-0.930)	(0.765-0.852)
N3.0 mg/dL	153	7	18	22	106	0.720	0.241	0.855	0.280	0.828
						(0.504-0.871)	(0.086-0.397)	(0.793-0.917)	(0.104-0.456)	(0.763-0.893)

counties along the Texas-Mexico border is 2 to 3 times greater than the rest of the country [3]. In addition, the prevalence of chronic Hepatitis C in our county has been estimated to be greater than 2% (N10000 cases), whereas most Texas counties have fewer than 1000 cases and a prevalence less than 1.75% [4].

Table 1 Test characteristics of urine nitrite for urinary tract infection as a function of 3 serum bilirubin ranges

TP, true positive; FP, false positive; FN, false negative; TN, true negative; FPP, false positive proportion (FP/FP + TP); SENS, sensitivity; SPEC, specificity.

We undertook this study to determine if, in fact, the anecdotal observations correlating hyperbilirubinemia with false-positive urine nitrite tests were correct. We tested this hypothesis by determining whether the specificity and positive predictive value of urine nitrite for UTI differed for patients with normal and elevated total serum bilirubin levels.

Methods

Study design

This project was an institutional review board-approved, retrospective review of patient records. Patient consent was waived.

Study setting and population

This study was conducted at a 327-bed urban teaching hospital and its associated outpatient clinics. In 2004, there were more than 60000 visits to the 33-bed ED, about 81000 clinic visits, and approximately 19 000 hospital admissions.

Study protocol

We reviewed 12 months of electronic records (Jan 2004 to Dec 2004) and compiled laboratory results into an Excel 2003 (Microsoft Corp, Redmond, Wash) spreadsheet for those patients having a workup that included urinalysis (UA), urine culture (UC), and total serum bilirubin (from complete metabolic panel or liver function tests). Specimens from UA and UC were collected on the same day, and serum tests were measured within 3 days of the urine studies. All urines were tested using the automated Clinitek Atlas (Bayer

Healthcare LLC, Tarrytown, NY) or semiautomated Clinitek 200+ (Bayer).

Positive urine cultures were those that had more than 100000 colony-forming units per milliliter. Cultures grow- ing out multiple organisms resembling contamination were considered negative cultures (based on the report from the microbiology laboratory).

Data analysis

Using standard formulas, we calculated the specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) of urine nitrite for UTI for patients whose total serum bilirubin was less than 1.5 mg/dL,

1.5 mg/dL or greater but 3.0 mg/dL or less, or greater than

3.0 mg/dL. Similarly, we determined the proportion of

Fig. 1 Receiver operating characteristic curve for the accuracy of the urine nitrite test for urinary tract infection as a function of total serum bilirubin.

	T bili b1.5 mg/dL	T bili z1.5 and V3.0 mg/dL	T bili N3.0 mg/dL
Escherichia coli	410 (65%)	66 (59%)	18 (58%)	494 (64%)
Klebsiella sp	78 (12%)	6 (5%)	2 (6%)	86 (11%)
Proteus sp	15 (2%)	6 (5%)	2 (6%)	23 (3%)
Other Enterobacteriaceaea	27 (4%)	4 (4%)	2 (6%)	33 (4%)
Pseudomonas sp	17 (3%)	2 (2%)	0 (0%)	19 (2%)
Enterococcus fecalis	38 (6%)	11 (10%)	1 (3%)	50 (6%)
Staphylococcus sp	19 (3%)	4 (4%)	3 (10%)	26 (3%)
Streptococcus sp	18 (3%)	4 (4%)	0 (0%)	22 (3%)
Candida sp and other yeasts	5 (1%)	4 (4%)	1 (3%)	10 (1%)
No identificationb	6 (1%)	5 (4%)	2 (6%)	13 (2%)
Totalc	633 (100%)	112 (101%)	31 (98%)	776 (99%)

false-positive urine nitrite tests (ie, the proportion of nitrite- positive and culture-negative samples divided by all nitrite- positive samples) for patients with normal and elevated serum bilirubin levels. 95% Confidence intervals (CIs) were calculated using the Bayesian Calculator on the Web site of the University of British Columbia, Department of Healthcare and Epidemiology (http://www.healthcare. ubc.ca/calc/bayes.html) and the Clinical Calculator on the VassarStats Web site of Vassar College (http://faculty. vassar.edu/lowry/clin1.html). Hypothesis testing was per- formed using the Cochran-Armitage trend test with exact computations (Statxact 3, Version 3.0.2, Cytel Software Corporation, Cambridge, Mass).

Table 2 Distribution of organisms isolated from urine cultures stratified by three total serum bilirubin ranges

Organism Frequency of isolation (%)

Total

T bili, total serum bilirubin.

^a Includes Burkholderia, Citrobacter, Enterobacter, Morganella, Serratia, and Kluyvera.

^b Includes patients whose laboratory report had a blank field for organism ID or which stated bno workup done.Q

^c May not equal 100% because of rounding.

The frequency of isolation for each identified organism was compiled for each range of serum bilirubin and percent- ages calculated. A difference in distribution of organisms between patients with the 3 different serum bilirubin ranges was tested with the asymptotic Kruskal-Wallis test.

Results

Our study population consisted of 3174 patients. In this group, the prevalence of UTI was 24.4%, and the specificity of the urine nitrite test for a positive culture was 0.967, similar to that of the total population of patients with only UA and UC results (Table 1).

When we stratified the study population by serum bilirubin level, we found that the proportion of false-positive urine nitrite results was almost 4 times higher (72% vs 17%) in patients with serum bilirubin greater than 3.0 mg/dL compared with those with normal or moderately elevated bilirubin (Table 1). Similarly, specificity and PPV of nitrite for a positive urine culture decreased as serum bilirubin increased. If total serum bilirubin was within normal limits (b1.5 mg/dL) or moderately elevated (z1.5 but V3.0 mg/dL), the specificity of the urine nitrite test was 0.974 and 0.966,

respectively; PPV was 0.825 and 0.827, respectively; and the proportion of false positives was 17.5% and 17.3%, respectively (Table 1). However, when patients’ bilirubin reached the level of jaundice [5] (N3.0 mg/dL), specificity dropped to 0.855, PPV decreased to 0.280, and the proportion of false positives increased to 72% (Table 1). For patients with bilirubin greater than 3.0 mg/dL, there was only an approximately 1-in-4 chance that a positive urine nitrite test was associated with a positive urine culture. Testing for trend as a function of the 3 serum bilirubin ranges demonstrated highly significant relationships ( P b.0001) for the proportion of false-positive, specificity, and PPV test characteristics. Sensitivity did not vary significantly as a function of serum bilirubin with trend test ( P = .55). Because of a decrease in specificity, the accuracy of the urine nitrite test for urinary tract infection decreased as a function of serum bilirubin (Fig. 1).

The distribution of isolated organisms was similar among patients with the 3 serum bilirubin ranges ( P = .08) (Table 2). Cultures with no growth or with multiple isolates suggestive of contamination were classified as negative cultures by the laboratory, and no additional information about isolates was available.

Discussion

Nitrite is not found in bnormalQ Urine samples, and its presence is presumed to indicate a urinary tract infection caused by Gram-negativE coliform bacteria that reduce nitrate (NO₃) to nitrite (NO₂) as part of their normal metabolic activity [6]. In our study population, about 21% of positive urine nitrite tests appeared to be false positives because they were not associated with UTI. We also found that this false-positive percentage was markedly increased if serum bilirubin was greater than 3.0 mg/dL, the level at which jaundice is likely to occur.

In such cases of hyperbilirubinemia, we had assumed that urine bilirubin causes a false-positive nitrite test via chemical interference or color-masking. However, a review of laboratory literature does not support this assumption. The chemical reactions on Urine dipsticks are highly specific, with little to no likelihood of cross-reactivity, and quality control materials include specific testing to detect interferences between analytes [7]. The test for nitrite on urine dipsticks is a modified Griess reaction, which is highly specific for nitrite and has been a standard method for quantitation of nitrite for more than 100 years. There are reports of false positives that mimic the pink color of a positive nitrite test on a urine dipstick, caused by medications that color urine red or which become red under acidic conditions (eg, phenozopyridine) [8]. It is unlikely that the yellow-brown color of bilirubin would cause this type of reaction.

An alternate explanation for the presence of nitrite in apparently sterile urine is that urine nitrite may be derived from a source other than microbial metabolism, namely, the renal clearance of elevated plasma nitrite, a stable metab- olite of nitric oxide produced in response to disease and infection. Plasma nitrite and nitrate levels are used as a marker for plasma nitric oxide because it is a very short- lived molecule, quickly oxidized to nitrite in the presence of oxygen. Nitric oxide is involved in immune responses mediated by macrophages and neutrophils along with Inflammatory processes initiated by Endothelial cells and the nervous system in response to disease and infection.

Nitric oxide’s role in inflammatory processes is well documented [9,10], and previously published studies [9-27] indicate that it is reasonable to expect elevated serum and urine nitrites in patients with various systemic diseases. Elevated levels of nitrite or nitrate in the blood have been reported for Helicobacter pylori infection [11,12], sepsis [13,14], gastrointestinal infection [15], and for various types of liver disease [16], including hepatitis [17], cirrhosis [18-23], and Hepatocellular carcinoma [24-26]. Additional studies have documented elevated nitrite levels in urine for Disease states, such as acute pancreatitis [27], Multiple sclerosis [28], and inflammatory bowel disease [29].

There are several published studies that provide measured levels of serum nitrite for various types of liver disease. The levels vary widely from study to study for both patients and controls depending on whether the patients were on nitrogen-restricted diets or not. However, the reported average or median concentration of NO₂ in patients with liver disease was at least 1.7 times, up to 2.9 times, greater than their respective controls in all studies [17,21-23,25,26]. Similarly, a couple of studies have examined the relation- ship between blood and urine nitrite levels and have reported correlations between them [30], although excretion of nitrite and the final urine concentration are affected by renal function and renal disease [31-34]. There are no studies that have reported simultaneous blood and urine

nitrite levels for people with liver disease, but one study

[35] reported that patients with cirrhosis had a nitrite clearance rate that was 4 times greater than that of the controls (18 vs 4.2 mL/min, respectively). Thus, it is reasonable to expect that a person with elevated plasma nitrite could have elevated urine nitrite levels, even if there is Renal impairment, as is often seen with cirrhosis.

Limitations

The major limitation of the current study is that our conclusions are based on composite data that were not separated by patient treatment location (inpatient vs outpatient), disease severity, comorbidities, age, or sex. Any one of these demographic features could affect prevalence, so generalizing the results of this study to specific populations should be done with care. Similarly, because of the higher prevalence of hepatitis in our region, these results may not be generalizable to other geographic areas. In particular, false-positive proportion, PPV, and NPV will be affected by the prevalence of UTI overall and within each serum bilirubin stratum, whereas sensitivity and specificity will generally not be affected. In addition, the possibility of confounding by altered renal function, or current or recent antibiotic treatment was not measured or assessed.

Conclusions

Our data strongly suggest that in patients with hyper- bilirubinemia, the presence of nitrite in urine is much less likely to predict a UTI, compared with patients whose bilirubin is within normal limits. The results of our study also suggest a need for research that would assess the relationship between plasma nitrite levels and urine nitrite levels in patients with different types of systemic disease.

References

1. Rahn DD, Boreham MK, Allen KE, Nihira MA, Schaffer JI. Predicting bacteriuria in urogynecology patients. Am J Obstet Gynecol 2005;192(5):1376 – 8.
2. Deville WL, Yzermans JC, van Duijn NP, Bezemer PD, van der Windt DA, Bouter LM. The urine dipstick test useful to rule out infections. A meta-analysis of the accuracy. BMC Urol 2004;4(4):1 – 14.
3. Doyle TJ, Bryan RT. Infectious disease morbidity in the US region bordering Mexico, 1990-1998. J Infect Dis 2000;182(5):1503 – 10.
4. Yalamanchili K, Saadeh S, Lepe R, Davis GL. The prevalence of Hepatitis C virus infection in Texas: implications for future health care. Proc (Bayl Univ Med Cent) 2005;18(1):3 – 6.
5. Roche SP, Kobos R. Jaundice in the adult patient. Am Fam Physician 2004;69(2):299 – 304.
6. Talaro KP, Talaro A. Foundations in microbiology. 4th ed. New York7 McGraw-Hill; 2002.
7. Peele Jr J, Gadsden R, Crews R. Evaluation of Ames’ bClini-TekQ. Clin Chem 1977;23(12):2238 – 41.
8. Henry JB, editor. Basic examination of urine. Clinical diagnosis and management by laboratory methods. Philadelphia, WB7 Saunders; 2001. p. 367 – 401.
9. Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P. Molecular biology of the cell. 4th ed. New York: Garland Science; 2002.
10. Lamas S, Michel T, Endothelium-derived nitric oxide and control of vascular tone. In: Jameson JL, editor. Principles of Molecular medicine. Totowa (NJ): Humana Press; 1998.
11. Kodama K, Sumii K, Kawano M, Kido T, Nojima K, Sumii M, et al. Helicobacter pylori infection increases serum nitrate and nitrite more prominently than serum pepsinogens. Helicobacter 2002;7(1):9 – 13.
12. Tari A, Kodama K, Kurihara K, Fujihara M, Sumii K, Kajiyama G. Does serum nitrite concentration reflect gastric carcinogenesis in Japanese Helicobacter pylori-infected patients? Dig Dis Sci 2002; 47(1):100 – 6.
13. Wong HR, Carcillo JA, Burckart G, Shah N, Janosky JE. Increased serum nitrite and nitrate concentrations in children with the sepsis syndrome. Crit Care Med 1995;23(5):835 – 42.
14. Nakae H, Endo S, Kikuchi M, Yamada Y, Shibata M, Ishikura H, et al. Nitrite/nitrate (NOx) levels and hemodynamics during septic shock. Surg Today 2000;30(8):683 – 8.
15. Charmandari E, Meadows N, Patel M, Johnston A, Benjamin N. Plasma nitrate concentrations in children with infectious and noninfectious diarrhea. J Pediatr Gastroenterol Nutr 2001;32(4): 423 – 7.
16. Geller DA, Di Silvio M, Nussler AK, Wang SC, Shapiro RA, Simmons RL, et al. Nitric oxide synthase expression is induced in hepatocytes in vivo during hepatic inflammation. J Surg Res 1993; 55(4):427 – 32.
17. Tankurt E, Kirkali G, Ozcan MA, Mersin N, Ellidokuz E, Akpinar HA. Increased serum nitrite and nitrate concentrations in chronic hepatitis. J Hepatol 1998;29(3):512 – 3.
18. Heller J, Kristeleit H, Brensing KA, Woitas RP, Spengler U, Sauerbruch T. Nitrite and nitrate levels in patients with cirrhosis of the liver: influence of kidney function and fasting state. Scand J Gastroenterol 1999;34(3):297 – 302.
19. Hori N, Okanoue T, Mori T, Kashima K, Nishimura M, Nanbu A, et al. Endogenous nitric oxide production is augmented as the severity advances in patients with liver cirrhosis. Clin Exp Pharmacol Physiol 1996;23(1):30 – 5.
20. Arkenau HT, Stichtenoth DO, Frolich JC, Manns MP, Boker KH. Elevated nitric oxide levels in patients with Chronic liver disease and cirrhosis correlate with disease stage and parameters of hyperdynamic circulation. Z Gastroenterol 2002;40(11):907 – 13.
21. Hokari A, Zeniya M, Esumi H, Kawabe T, Gershwin ME, Toda G. Detection of serum nitrite and nitrate in primary biliary cirrhosis: possible role of nitric oxide in bile duct injury. J Gastroenterol Hepatol 2002;17(3):308 – 15.
22. Campillo B, Bories PN, Benvenuti C, Dupeyron C. Serum and urinary nitrate levels in liver cirrhosis: endotoxemia, renal function and hyperdynamic circulation. J Hepatol 1996;25(5):707 – 14.
23. Oekonomaki E, Notas G, Mouzas IA, Valatas V, Skordilis P, Xidakis C, et al. binge drinking and nitric oxide metabolites in chronic liver disease. Alcohol Alcohol 2004;39(2):106 – 9.
24. Moriyama A, Masumoto A, Nanri H, Tabaru A, Unoki H, Imoto I, et al. High plasma concentrations of nitrite/nitrate in patients with hepatocellular carcinoma. Am J Gastroenterol 1997;92(9):1520 – 3.
25. Notas G, Xidakis C, Valatas V, Kouroumalis A, Kouroumalis E. Levels of circulating endothelin-1 and nitrates/nitrites in patients with virus-related hepatocellular carcinoma. J Viral Hepat 2001;8(1):63 – 9.
26. Moriyama A, Tabaru A, Unoki H, Abe S, Masumoto A, Otsuki M. Plasma nitrite/nitrate concentrations as a tumor marker for hepatocel- lular carcinoma. Clin Chim Acta 2000;296(1-2):181 – 91.
27. Rahman SH, Ammori BJ, Larvin M, McMahon MJ. Increased nitric oxide excretion in patients with Severe acute pancreatitis: evidence of an endotoxin mediated inflammatory response? Gut 2003;52(2):270 – 4.
28. Giovannoni G, Silver NC, O’Riordan J, Miller RF, Heales SJ, Land JM, et al. Increased urinary nitric oxide metabolites in patients with multiple sclerosis correlates with early and relapsing disease. Mult Scler 1999;5(5):335 – 41.
29. Goggins MG, Shah SA, Goh J, Cherukuri A, Weir DG, Kelleher D, et al. Increased urinary nitrite, a marker of nitric oxide, in active inflammatory bowel disease. Mediators Inflamm 2001;10(2):69 – 73.
30. Rejdak K, Petzold A, Sharpe MA, Smith M, Keir G, Stelmasiak Z, et al. Serum and urine nitrate and nitrite are not reliable indicators of intrathecal nitric oxide production in acute brain injury. J Neurol Sci 2003;208(1-2):1 – 7.
31. Kurioka S, Koshimura K, Sugitani M, Murakami Y, Nishiki M, Kato

Y. Analysis of urinary nitric oxide metabolites in healthy subjects. Endocr J 1999;46(3):421 – 8.

1. Baylis C, Vallance P. Measurement of nitrite and nitrate levels in plasma and urine–what does this measure tell us about the activity of the endogenous nitric oxide system? Curr Opin Nephrol Hypertens 1998;7(1):59 – 62.
2. Rabbani GH, Islam S, Chowdhury AK, Mitra AK, Miller MJ, Fuchs

G. Increased nitrite and nitrate concentrations in sera and urine of patients with cholera or shigellosis. Am J Gastroenterol 2001;96(2): 467 – 72.

1. Dubey NK, Dey PK, Saxena S, Batra S, Khapekar P, Gupta A. Serum nitrite and urinary nitrite excretion in Nephrotic syndrome. Indian Pediatr 2001;38(9):1025 – 9.
2. Campillo B, Bories PN, Benvenuti C, Dupeyron C. Serum and urinary nitrate levels in liver cirrhosis: endotoxemia, renal function and hyperdynamic circulation. J Hepatol 1996;25(5):707 – 14.