Sensitivity and Specificity of Urinary Nitrite for UTIs
to the editor: I would like to comment on the article on acute pyelonephritis by Drs. Ramakrishnan and Scheid1 in the March 1, 2005, issue of American Family Physician. Table 3 of this article1 details the laboratory diagnosis of urinary tract infections (UTIs). According to data in this table, the urinary nitrite test has a sensitivity of 92 to 100 percent and a specificity of 35 to 85 percent for diagnosing UTI. The reference cited for these estimates is a 1983 review article,2 not a primary data source.
Multiple references support different sensitivity and specificity numbers. For example, Campbell’s Urology states that “the specificity of the nitrite dipstick for detecting bacteriuria is over 90 [percent]. The sensitivity of the test, however, is considerably less, varying from 35 [percent] to 85 [percent].”3
A meta-analysis4 that included sensitivity and specificity of nitrite and leukocyte esterase for UTI found that the sensitivity of nitrite clustered around 50 percent, and the specificity averaged around 95 percent. For leukocyte esterase, sensitivity appeared highly variable depending on study and setting but reasonably can be inferred to be in the 50 to 60 percent range. Its specificity also was highly variable but, on average, was lower than that of nitrite.4
A study5 of symptomatic men with UTI in primary care found urinary nitrite to have a sensitivity of 38 percent and a specificity of 84 percent; leukocyte esterase had a sensitivity of 54 percent and a specificity of 55 percent.
The reference cited in the table1 to support their data states: “These sticks [nitrite test sticks] are 90.7 percent sensitive and 99.1 percent specific in detecting gram-negative bacteriuria. Although false-positive tests are uncommon [leading to high specificity], false-negative results [imperfect sensitivity] may be caused by (1) lack of dietary nitrate, (2) reduction of nitrate due to diuresis, or (3) infections due to enterococci and Acinetobacter, that do not reduce nitrate, as well as some pseudomonads that reduce nitrate to nitrogen gas.”2
Therefore, the clinical impression that leukocyte esterase is the more sensitive test (more likely to be the sole indicator of UTI) and that the presence of nitrite is highly specific (i.e., highly predictive of UTI) appears to still hold. In usual clinical scenarios that involve obtaining a urinalysis (reasonable pretest probability), a positive urinary nitrite indicates a high likelihood of UTI, although it is likely to be present in a minority of patients with UTI. However, even the absence of leukocyte esterase and nitrite cannot be used to exclude UTI.
in reply: We thank Dr. Fox for his critique and agree with some of his comments. The sensitivities and specificities mentioned for the nitrite test in Table 3 of our article1 are transposed and should read sensitivity (35 to 85 percent) and specificity (92 to 100 percent). The meta-analysis2 of the accuracy of the urine dipstick test cited by Dr. Fox reported that the sensitivity of the urine dipstick test for nitrites ranged from 45 to 60 percent and the specificity from 85 to 98 percent (with children removed). While this reference was not available when our article was submitted, it is reassuring that the specificities are similar to the corrected table values.
The leukocyte esterase urine dipstick test, which screens for pyuria, was reported in the same meta-analysis2 to have a sensitivity of 48 to 86 percent and a specificity of 17 to 93 percent. Perhaps this slight increase in sensitivity supports the clinical impressions of Dr. Fox that the leukocyte esterase test is more sensitive because the sensitivity was found to be the highest in primary care populations (83 to 92 percent).2
Regarding the clinical use of the urine dipstick nitrite test, we are not sure what Dr. Fox means by the term “reasonable” pretest probability. However, in the worst-case scenario (sensitivity: 45 percent and specificity: 85 percent), a patient with a pretest probability of 50 percent and a positive nitrite test would have a 75 percent chance of a urinary tract infection. Although we could agree with Dr. Fox that the absence of leukocyte esterase and nitrite cannot exclude a urinary tract infection in a patient with a high pretest probability, the meta-analysis2 he cited does not support his more general statement.