Materials and Methods
Pregnant women undergoing routine screening for asymptomatic bacteriuria at Duke University Medical Center between April 1998 and June 1999 comprised the study population. Though informed consent was not required, our institutional review board approved this study. Urine cultures were obtained in all pregnant women undergoing screening for asymptomatic bacteriuria and sent to the laboratory for urine culture. Women were given instructions for collecting clean-catch midstream urine samples. Urine samples from these women were randomly selected by one of the laboratory personnel involved with the study (KK). Selection was made independent of clinical knowledge of the patient. After collection, each sample was divided into three portions, one of which was submitted for immediate culture, another tested for the presence of nitrites and leukocyte esterase using dipstick analysis (Multistix 10 SG, Bayer, Elkhart, IN), and the third frozen at −80C.
Urine was cultured by plating samples on blood and McConkey agar and incubated for 24 hours at 37C. Culture was used to determine the presence of asymptomatic bacteriuria with one culture of at least 100,000 colony-forming units of a single organism per mL used to define asymptomatic bacteriuria. Cultures were considered contaminated if multiple gram-positive organisms were identified after incubation. Isolates of gram-negative rods at 10,000–100,000 colony-forming units per mL were reported.
Dipstick analysis for nitrites was reported as positive or negative. Leukocyte esterase activity was reported as negative, trace, 1+, 2+, or 3+, with a “positive” test defined as any finding of trace or greater. Interleukin-8 levels were assayed in batch using frozen urine aliquots. This was accomplished with a chemiluminescent immunoassay (Immulite IL-8, Diagnostics Products Corp.), which uses monoclonal antibodies specific for IL-8. This assay has a sensitivity (detection limit) of 2 pg/mL and a coefficient of variation of 3.6–3.8% (intraassay) and 5.2–7.4% (interassay). A receiver operating characteristic curve was generated to determine the optimal cutoff point for IL-8.
For our sample size calculation, we estimated that IL-8 would detect 99% of patients with asymptomatic bacteriuria,11 and urine dipstick analysis would predict 50%.5 To detect a difference in sensitivities from 99% to 50% using an alpha of 0.05, a beta of 0.2, and a two-tailed test, 15 patients with asymptomatic bacteriuria were needed. Because approximately 7–8% of our pregnant population demonstrated asymptomatic bacteriuria during routine screening, we estimated that screening of 200 patients would be required. Statistical analyses were performed using GraphPad InStat (San Diego, CA). Dichotomous variables were analyzed using χ2 test. Comparisons of nonparametric continuous variables were analyzed using Mann-Whitney U test.
Twenty women were identified with asymptomatic bacteriuria. The most common pathogen identified was E coli (n = 15), followed by Proteus mirabilis (n = 2), Enterococcus faecalis (n = 1), Citrobacter koserii (n = 1), and Klebsiella pneumoniae (n = 1). No growth was seen in 36 samples, whereas 144 returned as “mixed flora,” consistent with contamination from vaginal or skin flora. These “mixed flora” cultures contained mixed gram-positive organisms, and none contained gram-negative organisms at levels greater than 10,000 colony-forming units per mL. Indications for screening were as follows: initial screen (n = 97), prior urinary tract infection (n = 56), sickle cell trait (n = 25), abnormal urine dipstick during visit (n = 12), diabetes (n = 7), and history of renal disease (n = 3). None of the women developed a clinical infection (ie, pyelonephritis, chorioamnionitis, pneumonia, etc) within 1 week of testing.
The median IL-8 concentration was 356 pg/mL (22–5620 pg/mL) for women with and 125 pg/mL (6–5365 pg/mL) for women without asymptomatic bacteriuria (P < .01, Mann-Whitney U test). The optimal cutoff point for IL-8, determined by a receiver operating characteristic curve, was 264 pg/mL. Using this value, sensitivity, specificity, positive and negative predictive values were calculated for urinary IL-8 and compared with the presence of urinary nitrites and/or leukocyte esterase (Table 1). Although IL-8 had a better sensitivity (70%) for detecting asymptomatic bacteriuria than did having either leukocyte esterase or nitrite present (45%), the difference was not significant (P = .20, χ2).
Urinary IL-8 identified 70% of pregnant women with asymptomatic bacteriuria, but this corresponded to missing 30% of women with infection. Although the impact of asymptomatic bacteriuria treatment on pregnancy outcomes such as preterm labor, hypertension, low birth weight, or anemia is debatable,12–15 the subsequent development of pyelonephritis in 20–40% of untreated pregnant patients with asymptomatic bacteriuria is well known.1 Therefore, a screening method that misses 30% of asymptomatic bacteriuria is unacceptable.
Our reported sensitivity for urine dipstick testing, 45%, was consistent with that of Bachman et al.5 However, urinary IL-8 did not perform as well in our population as it did in that reported by Ko et al,11 who found a sensitivity of 99% for identifying urinary tract infections. This is despite the fact that post hoc selection of a cutoff point by a receiver operating characteristic curve has the potential for overestimating sensitivity and specificity. Because IL-8 did not perform as well as we anticipated, we lacked sufficient power to detect a difference in sensitivities between urinary IL-8 and urine dipstick analysis. With our sensitivities of 70% for IL-8 and 45% for dipstick analysis (presence of nitrites and/or leukocyte esterase), we would need to evaluate 68 patients with asymptomatic bacteriuria to have 80% power to detect this difference using a two-tailed test with an alpha of 0.05.
There are several reasons to explain why IL-8 may not have performed as well in our patient population as we expected. First, we had many urine samples that were apparently contaminated with skin or vaginal flora. In the study by Ko et al,11 controls were chosen from patients known to have normal urinalyses. Such patients are notably different from our “control” patients, of whom only 20% had cultures with no bacterial growth. Contaminated urine samples may contain vaginal or cervical secretions, which are known to have increased levels of IL-8.16 Therefore, urinary IL-8 levels in these women may be falsely elevated by cervicovaginal IL-8.
Another reason that IL-8 may not have performed well in our patients is that we studied asymptomatic patients, whereas all patients evaluated by Ko et al11 were symptomatic. Women with symptomatic urinary tract infections may have more mucosal inflammation, which could lead to higher levels of urinary IL-8, because IL-8 is produced at the mucosal surface.10 Finally, we studied only pregnant women, whereas Ko et al11 evaluated male and female patients of all ages. It is possible that urinary IL-8 production may be exaggerated with urinary tract infections in pregnancy.
We cannot recommend the use of urinary IL-8 or urine dipstick analysis as methods for detecting asymptomatic bacteriuria in pregnant women. At our institution, the costs for urine culture, IL-8, and dipstick analysis are $35, $12.85, and $2.68, respectively. Based on the poor performance of urine dipstick and IL-8 testing, the costs of the tests and the potential impact of missing asymptomatic bacteriuria in pregnancy, urine culture remains the best screening method for detecting asymptomatic bacteriuria in our patients. Although Rouse et al17 reported that either dipstick analysis (without confirmatory urine culture) or urine culture is a cost-effective method for asymptomatic bacteriuria screening, they used a sensitivity of 72% for dipstick analysis in their model, as opposed to 45% noted in our patients. Additionally, under their model, patients with a positive nitrite or leukocyte esterase would receive empiric treatment for an infection, and urine culture would not be performed. Because the positive predictive value for dipstick analysis varies from 16–62%,5–7 treating women based on the presence of urinary nitrites or leukocyte esterase would expose approximately 40–85% of these mothers and their fetuses to unnecessary antibiotics.
1. Duff P. Pyelonephritis in pregnancy. Clin Obstet Gynecol 1984;27:17–31.
2. Hauth JC, Merenstein GB, eds. Guidelines for perinatal care. 4th ed. Washington, DC: Library of Congress, American Academy of Pediatrics, American College of Obstetricians and Gynecologists, 1997.
3. Soisson AP, Watson WJ, Benson WL, Read JA. Value of a screening urinalysis in pregnancy. J Reprod Med 1985;30:588–90.
4. Lenke RR, Van Dorsten JP. The efficacy of the nitrite test and microscopic urinalysis in predicting urine culture results. Am J Obstet Gynecol 1981;140:427–9.
5. Bachman JW, Heise RH, Naessens JM, Timmerman MG. A study of various tests to detect asymptomatic urinary tract infections in an obstetric population. JAMA 1993;270:1971–4.
6. Robertson AW, Duff P. The nitrite and leukocyte esterase tests for the evaluation of asymptomatic bacteriuria in obstetric patients. Obstet Gynecol 1988;71:878–81.
7. Etherington IJ, James DK. Reagent strip testing of antenatal urine specimens for infection. Br J Obstet Gynaecol 1993;100:806–7.
8. Van Dorsten JP, Bannister ER. Office diagnosis of asymptomatic bacteriuria in pregnant women. Am J Obstet Gynecol 1986;155:777–80.
9. Smith WB, Gamle JR, Clark-Lesis I, Vadas MA. Interleukin-8 induces neutrophil transendothelial migration. Immunology 1991; 72:65–72.
10. Agace WW, Hedges SR, Ceska M, Svanborg C. Interleukin-8 and the neutrophil response to mucosal gram-negative infection. J Clin Invest 1993;92:780–5.
11. Ko YC, Mukaida N, Ishiyama S, Tokue A, Kawai T, Matsushima K, et al. Elevated interleukin-8 levels in the urine of patients with urinary tract infections. Infect Immun 1993;61:1307–14.
12. Whalley P. Bacteriuria of pregnancy. Am J Obstet Gynecol 1967; 97:723–38.
13. Gilstrap LC, Leveno KJ, Cunningham FG, Whalley PJ, Roark ML. Renal infection and pregnancy outcome. Am J Obstet Gynecol 1981;141:709–16.
14. Andriole VT, Patterson TF. Epidemiology, natural history, and management of urinary tract infections in pregnancy. Med Clin of N Amer 1991;75:359–73.
15. MacDonald P. Summary of a workshop on maternal genitourinary infections and the outcome of pregnancy. J Infect Dis 1993;147:596–605.
16. Tanaka Y, Narahara H, Takai N, Yoshimatsu J, Anai T, Miyakawa I. Interleukin-1 β and interleukin-8 in cervicovaginal fluid during pregnancy. Am J Obstet Gynecol 1998;179:644–9.
17. Rouse DJ, Andrews WW, Goldenberg RL, Owen J. Screening and treatment of asymptomatic bacteriuria of pregnancy to prevent pyelonephritis: A cost-effectiveness and cost-benefit analysis. Obstet Gynecol 1995;86:119–23.