Community-Based Urine Screening for Chlamydia trachomatis with a Ligase Chain Reaction Assay

Methods

Study Sample and Design

Study sites were chosen if they predominantly served clients younger than 21 years of age, but no upper age limit was stipulated. Sites included a Department of Public Health adolescent primary care clinic in a Seattle, Washington, neighborhood with high rates of sexually transmitted diseases; the King County Division of Youth Services juvenile detention facility; seven health clinics for teenagers based in secondary schools in Seattle and King County; and three community-based organizations that provide social services to homeless youth or youth who spend most of their time in urban street settings. Clients of the adolescent primary care clinic are routinely offered screening for sexually transmitted diseases. Screening for sexually transmitted diseases is available at the school-based clinics and the juvenile detention facility on request, and it is not routinely offered in the community-based organizations. The study was conducted from November 1994 to July 1996.

All persons 12 years of age and older who received services at the study sites were invited to participate in the study by their providers, who included physicians, midlevel clinical practitioners, and public health outreach workers. Participation did not require a comprehensive evaluation for sexually transmitted diseases. Each participant was given a 30-mL urine specimen container and was instructed to initiate voiding into the container and to void until the container was full. Limited demographic, behavioral, and clinical data were collected by using a standardized form. Race or ethnicity was recorded according to clients' self-descriptions. The coding of all racial or ethnic categories, including Hispanic, was mutually exclusive; in King County, most Hispanic persons are white. Genitourinary symptoms were recorded as present or absent; if they were present, the provider recorded the nature of the symptoms. Symptoms that were included in the analysis as potential indicators of chlamydial infection for male participants were urethral discharge, dysuria, and urinary frequency; for female participants, symptoms were abnormal vaginal discharge or odor, vulvar pruritus, dysuria, urinary frequency, lower abdominal pain, and dyspareunia. All participants gave verbal informed consent, and the study was approved by the University of Washington Human Subjects Review Committee. No funding source for this study participated in the collection, analysis, or interpretation of data or in the decision to publish the results.

Laboratory Methods

Urine specimens were refrigerated for up to 48 hours before testing and were processed immediately after arrival in the laboratory. The ligase chain reaction assay for C. trachomatis (Abbott Laboratories, North Chicago, Illinois) was done in a single laboratory with the procedures stipulated by the manufacturer, as described elsewhere [12]. Leukocyte esterase testing of urine specimens was performed by dipstick (Miles, Inc., Elkhart, Indiana) by a research technologist in the central laboratory, and the results were scored as negative, trace, or 1+ to 4+; for analysis, a result of 1+ or more was considered positive. Leukocyte esterase testing was performed only on the first 6889 participants; results were compared against the standard of chlamydial infection as defined by the ligase chain reaction assay.

Statistical Analysis

Data were entered into a computer database by using Microsoft Access (Microsoft Corp., Redmond, Washington) and analyzed using SPS software (SPS, Chicago, Illinois). Statistical associations for dichotomous variables were assessed by using chi-square analysis, and continuous variables were compared by using the Student t-test. Univariate and multivariate associations of chlamydial infection were analyzed separately for male and female participants. Logistic regression was used to perform multivariate analyses. All independent variables (age, race, number of sex partners in the previous 2 months, condom use during the last sexual encounter, study site, and symptoms) were entered into the model, and a P value less than 0.05 was designated as the level of significance. Study site was included as a categorical variable to account for differences in participants that could not be captured with basic demographic data or by our relatively limited risk assessment. Interactions between independent covariates were assessed and included in the final model if they were statistically significant. The performance of leukocyte esterase testing in predicting chlamydial infection as defined by ligase chain reaction assay was assessed separately in asymptomatic and symptomatic male participants and in female participants by calculating sensitivity, specificity, and positive and negative predictive values.

Results

Prevalence and Univariate Correlates of Chlamydial Infection

Urine specimens for ligase chain reaction testing for C. trachomatis were collected from 11 068 persons. Of these persons, 950 (8.5%) denied ever having had sexual intercourse. These persons were excluded from the final analysis, leaving 10 118 participants whose median age was 16 years; 9349 (92.4%) were 19 years of age or younger. Table 1 shows the sample's characteristics and prevalence of chlamydial infection. Chlamydia trachomatis was detected in 699 participants (6.9% [95% CI, 6.4% to 7.4%]); 424 were female (8.5% [CI, 7.8% to 9.3%]) and 275 were male (5.3% [CI, 4.8% to 6.0%]).

Among participants for whom symptom status was known, 98.6% of male participants and 76.6% of female participants were asymptomatic. Those for whom symptom status was recorded as “unknown” (9.9% of all female and 21.3% of all male participants) did not differ significantly from asymptomatic participants in the prevalence of chlamydial infection or in any demographic or behavioral characteristic (data not shown). Chlamydial infection was documented in 14 (26%) of 55 male participants with genitourinary symptoms compared with 212 (5.3%) of 3982 asymptomatic male participants (P < 0.001). Infection was documented in 112 (10.7%) of 1045 female participants with symptoms and 260 (7.6%) of 3423 asymptomatic female participants (P = 0.001). Male participants were almost three times as likely as female participants to report having had two or more sex partners in the previous 2 months (33.6% compared with 12.3%; P < 0.001), and almost twice as many male participants as female participants reported having used a condom during their last sexual encounter (69.4% compared with 37.4%; P < 0.001).

Table 2 shows the univariate associations of chlamydial infection with demographic characteristics, sexual behavior, and symptoms. Significant associations were found for younger age in female participants, for older age and condom use in male participants, and for number of sex partners and symptoms in both male and female participants. All of the nonwhite racial and ethnic categories were significantly associated with C. trachomatis infection among male participants, but none was significantly associated with infection among female participants. Infection was diagnosed in 127 (4.2%) of 2995 male participants who said that they had used condoms during their most recent sexual encounter compared with 113 (8.6%) of 1320 male non-condom users (P < 0.001). In contrast, the prevalence of infection did not differ between female participants who did and did not use condoms during the most recent sexual encounter (8.7% and 8.3%, respectively; P > 0.2). With regard to the reported number of sex partners, the largest incremental change in prevalence occurred between zero partners and one partner for men (2.3% compared with 5.4%) and between one partner and two partners for women (7.9% compared with 13.2%).

The Figure 1 shows the association between chlamydial infection and age. Among female participants, prevalence declined with increasing age, from 11.7% for those 14 years of age and younger to 5.3% for those 21 years of age and older. The opposite trend was seen among male participants, in whom prevalence increased from 2.2% for those 14 years of age and younger to a peak of 16.4% at 20 years of age. Both trends were statistically significant (P < 0.001, chi-square test for trend).

View larger version:

• In this window
• In a new window

Figure 1. < 0.001 in both groups; chi-square test for trend. White bars indicate female participants; striped bars indicate male participants. Age-specific prevalence of Chlamydia trachomatis infection by sex of participants. P

Multivariate Analysis

The relation between chlamydial infection and age, race, genitourinary symptoms, condom use during the most recent sexual encounter, number of sex partners, and study site was analyzed in a logistic regression model. The results are shown in Table 3. Among female participants, independent associations with chlamydial infection were found for age 17 years or less, report of two or more sex partners in the previous 2 months, and genitourinary symptoms; no association was found for condom use or race or ethnicity. No significant interaction effects among covariates were noted. Among male participants, a significant interaction effect was seen between study site and age (data not shown); the final model included this interaction term. Significant independent associations with chlamydial infection were found in male participants for all racial and ethnic categories relative to white persons and for two or more sex partners in the previous 2 months. The risk for infection was significantly reduced in relation to condom use during the most recent sexual encounter. Contrary to the age-related trend among female participants, being 17 years of age or younger was not an independent risk factor for infection among male participants. In addition, young age was not independently associated with infection after stratification by site (data not shown). In male participants, reporting the number of sex partners as “unknown” was associated with an increased risk for infection, whereas “unknown” condom use was associated with a reduced likelihood of infection.

Performance of Leukocyte Esterase Testing

In asymptomatic male participants, the sensitivity of leukocyte esterase testing in predicting chlamydial infection was 58.9% (CI, 50.5% to 67.1%), the specificity was 94.9% (CI, 93.9% to 95.7%), the positive predictive value was 38.4% (CI, 31.9% to 45.1%), and the negative predictive value was 97.7% (CI, 97.0% to 98.3%). Among 43 symptomatic male participants in whom the leukocyte esterase test was done, the sensitivity of the test was 66.7% (CI, 37.7% to 88.4%), the specificity was 67.7% (CI, 50.0% to 82.3%), the positive predictive value was 44.4% (CI, 23.2% to 67.3%), and the negative predictive value was 84.0% (CI, 65.8% to 94.7%). In female participants, the sensitivity was 45.6% (CI, 40.1% to 51.2%), the specificity was 76.4% (CI, 74.9% to 77.9%), the positive predictive value was 16.5% (CI, 14.2% to 19.2%), and the negative predictive value was 93.2% (CI, 92.1% to 94.2%). The presence of genitourinary symptoms in female participants did not significantly affect the performance of leukocyte esterase testing (data not shown).

Discussion

Clinic-based screening for C. trachomatis infection has been associated with declining prevalences in some populations [5-9, 18], and selective screening of women for chlamydial infection in a large health maintenance organization was associated with a 40% reduction in the incidence of acute pelvic inflammatory disease relative to unscreened controls [19]. However, implementation of widespread screening has been limited by the need for endocervical or urethral specimens. We used an alternative approach: urine-based screening for C. trachomatis with the ligase chain reaction assay among adolescents and young adults seen in community-based settings, many of which do not routinely offer testing for sexually transmitted diseases. Despite marked decreases in the prevalence of chlamydial infection in sexually transmitted disease and family planning clinics over the past decade in the Pacific Northwest [5, 18], we found prevalences of 8.6% among female participants and 5.4% among male participants. These prevalences are higher than prevalences seen in the Seattle-King County sexually transmitted disease clinics in 1995, where 5.0% of 3450 female persons and 5.9% of 5040 male persons had C. trachomatis detected by culture, and approximate prevalences seen in areas without widespread chlamydial screening [3]. In other cities, the prevalence of chlamydial infection in female adolescents in juvenile detention facilities [20] or adolescent clinics [9, 21] was 10% to 16% using nonamplified methods and 17.4% with ligase chain reaction or PCR [22]. Fewer comparative data are available for males, but prevalence as assessed by urethral culture has ranged from 4.7% to 8.6% [23-25].

Many of the factors independently associated with chlamydial infection in this study are concordant with those found in other studies [26-30]. However, we found that the prevalence of C. trachomatis increased in a linear manner with increasing age in male participants. This was the opposite of our expectations and the opposite of the trend in female participants, but it is consistent with observations for national data on reported cases of gonorrhea, in which rates are highest in males at 20 to 24 years of age and in females at 15 to 19 years of age [31]. The association of sexually transmitted infection with increasing age in males is also consistent with data on the age of fathers in adolescent births, which indicate that many sexually active teenage girls engage in sex with partners older than themselves [32].

A surprising 69% of male participants, but only 37% of female participants, reported having used condoms during their most recent sexual encounter. Condom use was associated with a reduced prevalence of infection among male but not female participants. Newly acquired (incident) chlamydial infection may be more likely to cause symptoms among males than among females; hence, the most recent exposure to C. trachomatis and the consequent symptoms may be more strongly associated with concomitant condom use in males than in females. This hypothesis is also supported by the data showing that males were more likely to acquire C. trachomatis from their most recent sex partner: The difference in the prevalence of chlamydial infection between persons reporting zero sex partners and persons reporting one sex partner in the previous 2 months was markedly greater for male than for female participants.

Among the 950 persons who denied ever having had sexual intercourse, 6 had positive results on ligase chain reaction testing. Review of primary records revealed that 5 of these 6 persons had in fact been sexually active in the recent past. Thus, the specificity of the ligase chain reaction assay on urine specimens for C. trachomatis, using report of sexual activity as an admittedly imperfect standard, is no lower than 99.4% (944 of 950) and more likely approximates 99.9% (949 of 950). Although the validity of “discrepant analysis” in determining the performance of the chlamydial ligase chain reaction assay has been debated [33, 34], our results support the performance characteristics that were calculated with this technique [16, 17].

The sensitivity of leukocyte esterase testing in asymptomatic male participants was only 59% in our study, and 55 (41%) of 134 infections in asymptomatic male participants would have been missed if ligase chain reaction testing had been limited to those with positive results on leukocyte esterase testing. Leukocyte esterase test performance in symptomatic male participants was not notably better, and the test performed poorly in female participants. Other investigators have found sensitivities of 41% to 85% among asymptomatic males, with specificities of 75% to 95%, prompting the conclusion that leukocyte esterase testing is useful in deciding whether to test young men for C. trachomatis[25, 35-38]. In one study [38], the sensitivity of the leukocyte esterase test was favorably modified by participant age, performing better in younger male persons. However, none of these studies used amplified DNA tests for C. trachomatis, the sensitivity of which is considerably greater than that of the other nonculture assays. The ligase chain reaction assay can detect DNA quantities equivalent to 10 or fewer C. trachomatis elementary bodies; 10⁴ to 10⁵ elementary bodies are required for detection by enzyme immunoassay [39]. Because organism load is probably correlated with leukocytic response, ligase chain reaction may present a more rigorous challenge than other tests to the utility of leukocyte esterase testing in selecting persons for screening. In a recent small study [40], leukocyte esterase test results were positive in 8 of 10 men with asymptomatic chlamydial infection detected by ligase chain reaction; this result is not statistically different from ours. Until formal cost-effectiveness analyses based on universally screened populations are available, it is uncertain whether a negative result on leukocyte esterase testing is a valid basis on which to exclude asymptomatic young men from screening. However, in some settings, the positive predictive value of 38% that we found may be considered sufficient to warrant empirical therapy for asymptomatic teenage males in some circumstances (for example, if assuring subsequent treatment on the basis of a positive ligase chain reaction test result is unlikely). Further evaluation that combines leukocyte esterase testing with risk stratification may elucidate a role for leukocyte esterase testing in decisions about empirical therapy.

Our study has several limitations. First, although we wanted providers to screen all clients in the cooperating sites and we instructed them to do so, we could not ensure universal compliance, nor were we able to collect data on clients who refused testing. Preferential testing of clients viewed by providers as having a higher risk for chlamydial infection was especially likely to have affected the prevalence seen at the adolescent health clinic. According to providers' anecdotal reports, however, acceptance of testing by clients who were approached was nearly universal, and we attempted to partly control for undefined factors specific to study sites by including site in our logistic regression models.

Second, because we sought to determine the practical utility of urine ligase chain reaction screening in community-based settings, we were able to collect only limited clinical and behavioral information; data were incomplete for 22% of our participants. However, our findings were not affected by including or excluding these participants in the analyses, and we included “unknown” as an outcome in the final multivariate models.

A third limitation is that the prevalence of chlamydial infection is decreasing in Seattle and elsewhere in the Pacific Northwest, probably largely because of screening of sexually active women in family planning and sexually transmitted disease clinics [3, 18, 41]. Some of our findings might not apply in settings where prevention programs have not yet been implemented or where the prevalence of chlamydial infection is higher. Finally, we did not test for infection with Neisseria gonorrhoeae; in populations with a substantial prevalence of gonorrhea, this information might be important in interpreting a positive leukocyte esterase test result, particularly if empirical therapy is considered for a positive leukocyte esterase test result. However, the prevalence of gonococcal infection is less than 1% in male and female teenagers in Seattle outreach settings similar to those in our study.

The cost of DNA amplification tests has important implications for widespread screening. The ligase chain reaction and PCR assays now on the market are more expensive than the commonly used tests for C. trachomatis. Our unit cost of performing the ligase chain reaction assay is about $13.00, roughly twice that of the heretofore standard nonculture tests. However, in a separate cost-effectiveness analysis of screening strategies for women, we concluded that the improved sensitivity of ligase chain reaction compared with direct fluorescent antibody testing would offset the assay's higher cost [5]. Moreover, the cost of amplified DNA assays may decrease as large numbers of tests are done in a competitive marketplace. Nevertheless, in the immediate future, test cost is likely to limit community-based screening in many settings.

In conclusion, our study and others [22, 42, 43] indicate that community-based screening efforts targeted at adolescents are likely to detect considerable prevalences of chlamydial infection. The urine ligase chain reaction assay for C. trachomatis is a practical, readily accepted test that will permit wide-spread screening of sexually active adolescents and young adults, facilitating expansion of screening to men at risk and to women in settings where genital examinations are impractical. Such approaches can and should augment ongoing prevention programs in clinic-based populations. Finally, we found the leukocyte esterase test to be insensitive in selecting young persons for chlamydial screening, but further study is required before definitive recommendations can be made for the use of this test. The ability to test urine with the ligase chain reaction assay is an important addition to efforts aimed at preventing chlamydial infection.

From the University of Washington School of Medicine, the University of Washington School of Public Health and Community Medicine, Harborview Medical Center, and the Seattle-King County Department of Public Health, Seattle, Washington.

Ms. Krekeler and Dr. Celum: Sexually Transmitted Disease Clinic, Harborview Medical Center, Box #359777, 325 9th Avenue, Seattle, WA 98104.

Dr. Lafferty: School of Public Health and Community Medicine, Department of Health Services, #357660, University of Washington, 1959 NE Pacific Street, Seattle, WA 98195.

Dr. Stamm: Department of Medicine, Division of Allergy and Infectious Diseases, Box #356523, University of Washington, 1959 NE Pacific Street, Seattle, WA 98195.

Dr. Handsfield: Division of Infectious Diseases, Harborview Medical Center, Box #359779, 325 9th Avenue, Seattle, WA 98104.