THE INTRODUCTION of nucleic acid amplification tests (NAAT) for the diagnosis of Chlamydia trachomatis in urine specimens has led to the development of widespread chlamydia screening programs in clinical 1,2 and nonclinical 3–8 settings. In most sexually transmitted disease (STD) clinics in the United States and Europe, testing of both symptomatic and asymptomatic men and women for chlamydia by ligase chain reaction (LCR), polymerase chain reaction (PCR), or other NAAT is increasingly considered to be a standard of care. 9 However, the effectiveness of these testing approaches in reducing the prevalence of chlamydia and its sequelae may be limited if there is a high chlamydia incidence in the target population, and if persons treated for chlamydia are rapidly reinfected. Insights into factors associated with chlamydia incidence and repeat infection may assist in the development of targeted interventions, including client-centered counseling, rescreening, partner services, and interventions aimed at sexual networks with high chlamydia prevalence.
We conducted a retrospective cohort study of patients visiting the Denver Metro Health Clinic who were tested for chlamydia by PCR on more than one occasion during a 30-month interval after universal chlamydia testing was instituted among clinic patients in 1996. The specific aims of this study were to assess chlamydia incidence in this population, to investigate factors associated with chlamydia incidence, and to evaluate how these factors differ between new incident infections and repeat infections.
The Denver Metro Health Clinic (DMHC) is the largest STD clinic in the Rocky Mountain region, logging approximately 12,000 patient visits for a new problem each year. Universal PCR screening for chlamydia by Amplicor (Roche, Diagnostic Systems, Branchburg, NJ) among women (using cervical samples) and men (using urine samples) was established in the DMHC by the end of 1996. Medical records at this clinic have been computerized since 1987. A scannable form is generated for each patient and completed by the clinician. Information collected on this form includes demographics, sexual and STD history, risk behaviors, clinical findings, laboratory results, and diagnosis.
The forms are scanned into a computer database by the clinicians, and missing or incorrect information flagged by the computer during the scanning process is corrected by the clinician when this information is still available to them. Test results are added to the database when they become available, including chlamydia PCR results. In the interpretation of our study results, it is important to note that limited disease intervention outreach staff has thus far precluded formal partner notification activities for chlamydial infections. Instead, infected patients are encouraged to refer sex partners for care, and are provided with partner referral cards.
For this study, we created an analysis data set composed of clinic visits by persons screened for chlamydia more than once, with visits at least 30 days apart, during a 30-month study period between January 1, 1997 and June 30, 1999. For each person in this database, we defined a study interval, for which the first date was the date of the first chlamydia screening during the study period, further referred to as the baseline visit, and for which the last date was either the first visit at which an incident chlamydia infection was diagnosed or the last visit of the study if no incident infection developed. We then calculated and summed intervals between these two visits to arrive at the denominator for the calculation of incidence density rates, which were expressed as numbers of incident infections per 100 person-years of follow-up evaluation. Finally, we condensed all the visits into one patient record, in which demographic and risk behavior data from both visits were retained. For persons with positive test results for chlamydia at their baseline visit, we performed a visual chart review to evaluate whether they had received appropriate treatment. For 18 persons with baseline infections, receipt of treatment could not be verified, so these subjects were excluded from the analysis.
For analyses in which we explored the association of risk factors with incident infections, we used data from the second visit, primarily because this was likely to be closest to the date of incident infections. However, for certain analyses, we also included data from the first visit. Factors considered in these analyses included demographic variables (age, race or ethnicity, and gender), risk behaviors (number of sex partners and new sex partners in the past 30 days and the past 4 months as well as condom use in the past 4 months), clinical diagnoses associated with chlamydia infections (nongonoccoccal urethritis and epididymitis in men, mucopurulent cervicitis and pelvic inflammatory disease in women, and gonorrhea in both men and women), and whether a person was a partner of a patient with known or presumed chlamydia infection (e.g., a partner with confirmed chlamydia or chlamydia-associated conditions). Condom use was recorded as the approximate percentage of times that condoms had been used over the past 4 months (0%, 25%, 50%, 75%, 100%), with no distinction being made between condom use with main or occasional partners. For our analyses, we classified condom use as “never” (0%), “consistent” (100%), or “inconsistent” (25–75%).
In the analyses of incident infections, we distinguished new infections (i.e., incident infections among persons whose baseline test results were negative) from repeat infections (i.e., incident infections among persons whose baseline test results were positive). A history of chlamydia infection before the initial visit in our study was not considered in the definition of repeat infection. All analyses were conducted using the SAS statistical package (SAS Institute, Cary, NC). The t test and chi-square were used in the univariate analysis, whereas multiple logistic regression with backward stepwise elimination was used in the multivariate analysis.
A total of 3568 individuals met the study criteria. Of these, 41.2% were women and 58.8% were men. Of the study sample, 37.5% were black, 31.9% white, and 27.1% Hispanic, whereas 3.5% were of another race or ethnicity. The mean age of the entire sample was 29.6 years (range, 12–79 years). The mean follow-up interval was 335 days (range, 31–893 days). There were 136 men (6.5%) and 16 women (1.1%) who reported that during the 12 months before baseline their sexual contacts had been with partners of the same sex exclusively.
A flow diagram of the study population with regard to baseline and incident infections, including new and repeat infections, is shown in Figure 1. Overall, 491 (13.8%) of the 3568 persons had a positive chlamydia infection at baseline, and 385 (10.8%) had incident infections, including 286 new infections and 99 repeat infections. The overall incidence rate was 11.7 per 100 person-years (95% CI, 10.6–12.9). The demographic characteristics of the persons with baseline and incident infections are summarized in Table 1. Because an unknown proportion of baseline infections were incident infections (the testing history before baseline was either not known or not taken into account), we did not perform any formal comparisons between persons with baseline infections and those with incident infections.
Age was significantly associated with chlamydia incidence. Of the 385 incident cases, 247 (64.2%) occurred among persons younger than 25 years of age, giving an incidence rate of 19.7 per 100 person-years (95% CI, 17.3–22.2), as compared with 6.8 per 100 person-years (95% CI, 5.7–7.9) among persons 25 years or older (relative hazard, 3.0; 95% CI, 2.5–3.7). Adolescents (younger than 20 years) had the highest rates (27.7/100 person-years; 95% CI, 21.9–33.5), and in this group, the rates were still higher among black males and females. Of the adolescent black males, 18 in 63 (28.5%) had incident infections, giving a rate of 38.7 per 100 person-years (95% CI, 20.8–55.6). Of the adolescent black females, 15 in 68 (22%) had incident infections, giving a rate of 29.5 per 100 person-years (95% CI, 14.6–44.4).
Overall, 337 (68.6%) of 491 baseline infections and 254 (65.9%) of 385 incident infections were empirically treated at the time the patient first came to the clinic, either because of a clinical diagnosis associated with chlamydia infection, or because of contact with a partner known or presumed to have chlamydia infection (Table 2). Empirical treatment did not vary by ethnicity, but was less common among females than among males, and also less common among adolescents than among older subjects for both baseline and incident infections. Whereas the proportion of patients treated because of clinical diagnoses associated with chlamydia infections did not differ significantly between persons with baseline infections (54.4%) and those with incident infections (57.9%), the proportion treated because of contact with a partner known or presumed to have chlamydia infection was significantly higher among persons with baseline infections (22.6%) than among those with incident infections (12.9%;P < 0.001). Because of concern that the strong correlation between empirical treatment at the follow-up visit and incident infections detected at this visit would obscure potential associations in further incidence analyses, we excluded this variable from further analysis.
Among persons without a baseline infection, the incidence of new infections was 10.0/100 person-years (95% CI, 8.9–11.2), whereas the incidence of repeat infections was 23.6/100 person-years (95% CI, 18.9–28.2) among persons who had presented with a baseline infection (relative hazard, 2.4; 95% CI, 1.9–3). Given the potential difference between the factors associated with new infections and those associated with repeat infections, we considered these two groups separately. Table 3 summarizes and compares the univariate analyses of the two groups. The multivariate analyses are presented in Table 4.
Factors Associated With New Infections
Among the 3077 participants who did not have a baseline infection, the men were at a higher risk for incident infections than the women (adjusted odds ratio [AOR], 1.5; 95% CI, 1.1–1.9;P < 0.01;Table 4). However, separate logistic regression models for the men and women indicated no differences between the genders in risk factors for new infections (data not shown). Blacks were at higher risk than whites (AOR, 1.8; 95% CI, 1.3–2.5;P < 0.0001), and there was a higher incidence with younger age: participants younger than 20 years were at higher risk than those 30 years of age or older (AOR, 6.8; 95% CI, 4.5–10.2;P < 0.0001) (Table 4). A history of STD was also associated with higher chlamydia incidence rates (AOR, 1.8; 95% CI, 1.2–2.7;P < 0.01;Table 4).
We considered five behavioral variables in our exploratory analyses: multiple sex partners in the past 4 months and the past 30 days, new sex partners in the past 4 months and the past 30 days, and the use of condoms in the past 4 months. In the univariate analysis, higher incidence rates were noted among those who had multiple sex partners in the past 4 months or past 30 days and among those who had a new sex partner in the previous 4 months or previous 30 days (Table 3), but these variables were not retained in the multivariate model.
Identical chlamydia incidence rates were observed for persons who used condoms consistently (7.2%) and those who never used them (7.2%), whereas the rate was almost double (13.9%) among those who used condoms inconsistently (AOR, 2.2; 95% CI, 1.6–3;P < 0.0001, as compared with “always users,”Table 4). We speculated that the lower incidence rate among persons who never used condoms, as compared with those who used them inconsistently, reflected the fact that this group was more likely to have sex with a single, presumably uninfected, partner, and to forego condom use because they felt themselves to be at lower risk for STD acquisition. Indeed, only 18.4% of the persons who never used condoms reported having multiple partners in the past 30 days, as compared with 39.1% of the persons who used condoms inconsistently and 30.2% of persons who used condoms consistently (P < 0.0001). Likewise, only 26% of the persons who never used condoms reported having a new partner in the past 30 days, as compared with 48.2% of the persons who used condoms inconsistently and 48.3% of persons who used them consistently (P < 0.0001). Interestingly, the chlamydia incidence rate among the persons who never used condoms was significantly higher among those who reported having a new partner in the past 30 days (9.6%) than among those who did not have a new partner (6.1%;P < 0.05). In the multivariate analysis, the interaction term “absence of condom use × presence of a new partner in previous 30 days” was retained in the logistic regression model, indicating that those who had a new partner and never used condoms had a higher risk for incident infection (AOR, 1.9; 95% CI, 1.2–3.1) than either those who had a new partner but used condoms or those who did not have a new partner regardless of condom use (Table 4).
Factors Associated With Repeat Infections
The univariate analysis of factors associated with repeat infection among the 491 participants who had a baseline infection is summarized in Table 3. In the multivariate analyses, repeat infection was significantly associated with younger age. As compared with persons 30 years of age or older, those 20 to 24 years of age (AOR, 2.2; 95% CI, 1.1–4.5;P < 0.05) and those younger than 20 years (AOR, 2.1; 95% CI, 1–4.8;P < 0.05) were at significantly higher risk (Table 4). In contrast to incident infections among persons with no infection at baseline, repeat infections were not associated with inconsistent condom use, as measured at the follow-up visit. However, repeat infections were associated with the absence of any condom use, as measured at the initial visit (AOR, 1.8; 95% CI, 1.1–2.9;P < 0.05). Finally, having received treatment because of contact with a person suspected at the initial visit to have chlamydia infection was associated with a lower rate of repeat infection (AOR, 0.5; 95% CI, 0.3–0.9;P < 0.05).
In this study, we found an overall chlamydia incidence of 11.7 per 100 person-years, indicating that in this population of persons making repeat visits to an STD clinic, approximately 12% will have an incident chlamydial infection within a year after initial screening. Among adolescents, this percentage rises to almost 30%, of whom nearly 50% will not receive empirical treatment on the basis of clinical findings or contact with a person presumed to have chlamydia infection. A number of studies have investigated chlamydia incidence, largely in women, among whom screening in STD clinics and other settings has been practiced for many years. Our data are comparable to the findings from these studies.
Burstein et al 10 found an incidence of 28 cases per 1000 months of follow-up evaluation (33.6/100 person-years) among predominantly black inner-city adolescent females in Baltimore, which is similar to the rate of 29.5 per 100 person-years among female black adolescents in our study. Richey et al, 11 in a study from Birmingham, found a crude incidence rate of 11.5 cases per 1000 months of follow-up evaluation (13.8/100 person-years) among persons with initial negative chlamydia test results and a rate of 28 per 1000 months of follow-up evaluation (33.6/100 person-years) among persons with initial positive chlamydia test results. These rates appear to be somewhat higher than those in our study (respectively, 10.0 and 23.6 cases per 100 person-years), and could even be a low estimate in that chlamydia detection in the latter study was based on culture rather than NAAT. Nonetheless, the Richey et al 11 study population was largely composed of black females (77%), which could account for some of the differences between this study and ours.
Besides gender, age, and race or ethnicity, we also found inconsistent condom use to be associated with the occurrence of new infections. Interestingly, we found that the incidence of new infections was the same among persons who either always or never used condoms. This paradoxical finding may be explained by the fact that persons who never used condoms were significantly more likely to report only a single partner than persons who used condoms consistently or inconsistently. In addition, the rate of new incident infections was lowest among the persons who reported no new partners and no condom use. We speculate that these persons were more likely to be in a mutually monogamous relationship. If the partners in such relationships are uninfected, then the result is a protective effect from the lack of condom use for the acquisition of an incident chlamydial infection. However, this protective effect disappears when the partner in the relationship is infected and has remained untreated, in which case the absence of condom use would constitute a risk factor for repeat infection, as discussed later. Regardless of their interpretation, these findings point to the importance of formulating appropriate questions regarding condom use in relation to incident STD.
In a recently published study from San Diego, California, Gunn et al 12 found that among STD clinic clients enrolled over a 6-month period, 7.9% had a subsequent STD during a 1-year follow-up period. They also found that a history of gonorrhea or chlamydia was the strongest predictor of a subsequent gonorrhea or chlamydia infection, and that unsafe sexual behavior had little effect on further increasing this risk. These data were interpreted as supporting the core transmitter hypothesis, 12,13 in that these persons may be part of a sexual network with high prevalence of STDs, and therefore may easily be reexposed to these pathogens through other members of this network. In turn, such persons may contribute to the high endemicity of STD in these networks by transmitting them to other network members.
Our study can certainly be interpreted as supporting this hypothesis because the sexual behavior variables that we could measure were less strongly associated with incident infections than were demographic variables that may have been proxies of sexual networks. However, our data also support recent findings of Whittington et al, 14 suggesting that a potentially large proportion of repeat infections may be the result of reexposure to untreated previous partners rather than exposure to new partners from the network.
First, clients with repeat infections in our study were significantly less likely to have been treated at the time of their initial visit because they had a partner with known or presumptive chlamydia infection than individuals with prevalent infections who did not have repeat infections. This suggests that clients with repeat infections were more likely to have been exposed to partners who had not been diagnosed or treated, and who therefore may have been the source of repeat infection.
Second, in contrast to the finding for persons with new infections, we did not find any associations between repeat infection and number of partners or new partners, once again suggesting that the source of the initial infection may also have been the source of the repeat infection. The fact that the absence of condom use, as measured at the initial visit rather than the follow-up visit, was associated with repeat infection further suggests a pattern of unprotected intercourse with the source partner continuing after the first infection and thus increasing the risk for repeat infection.
Whereas in our study we were not able to make a distinction between repeat infections resulting from reexposure to untreated previous partners and those resulting from exposure to new partners belonging to the same network, such a distinction may be important from a prevention perspective. Whereas in the former situation, efforts should be focused on partner referral and treatment, in the latter situation, a more network-oriented approach may be necessary, including outreach testing and treatment as well as focus on the role of the index person as a potential core transmitter. 12
Our study had a number of limitations. First, the sample studied was from an STD clinic population. Therefore, the results cannot be generalized outside this population. Because chlamydial infection is largely asymptomatic, determination of incidence is dependent on test frequency. By definition, studied persons made at least two visits during the interval. Because these visits were likely related to STD symptomatology, this group may have been at higher risk for STD than persons who did not make repeat visits. The incidence of new and repeat infections found in this study may therefore be high estimates. Also, persons who had chlamydia infection at baseline may have been primed to recognize symptoms of chlamydial infection or potential reexposure to untreated partners, prompting a return visit, whereas persons with negative chlamydia test results at baseline may have returned for other reasons less likely to be associated with chlamydial infection. This could have caused repeat infection rates to seem higher than new infection rates.
Nonetheless, despite these possible biases, the fact remains that in our clinic population, repeat infection appears to be a substantial problem. Although chlamydia screening programs have been proved effective in some settings, 15 high repeat infection rates may substantially limit their effectiveness. Additionally, given the possible role of repeat infection in the pathogenesis of upper genital tract damage caused by chlamydial infection, 16,17 clearance of infection through detection by screening and treatment followed by repeat infection could theoretically enhance the risk of tubal damage. Therefore, interventions to prevent repeat infection are a high priority.
A practical approach would be to conduct active retesting of persons in whom chlamydia has been diagnosed. Our data suggest that such an approach may be most effective when conducted among persons younger than 25 years with chlamydia infections because they experienced repeat infection rates of 24% and accounted for 71% of all repeat infections. Because the incidence of new infections was also high among participants younger than 25 years (15%; 61% of all new infections), a case could be made for retesting regardless of chlamydial infection at the initial visit. Burstein et al, 18 for example, have recommended screening of all sexually active women younger than 25 years every 6 months. Furthermore, because of the concurrent epidemiology of multiple STDs in this population, it may be prudent not to limit such retesting to chlamydia alone, but to include screening for gonorrhea, trichomoniasis, and other STDs also. 19 However, although there is a good rationale for retesting, the response to such a recommendation by patients may be limited, especially among those who remain asymptomatic.
Our findings suggesting that untreated partners play an important role in repeat infection point to the need for the establishment and evaluation of partner services for persons with identified chlamydial infections. Because many health departments, like our own, lack resources that can be allocated to partner notification for diseases other than HIV and syphilis, they often rely on partner referral. For example, the infected persons are asked to distribute “contact” cards among their partners advising them to come to the clinic for diagnosis and treatment. The results of this strategy are often disappointing, and alternatives such as patient-delivered therapy (i.e., supplying patients with extra medication for their most recent partners) may be useful for some patients. 20 Finally, persons with repeat infection may maintain high prevalence rates of chlamydia and other STDs in their sexual networks, and intensive individual prevention case management, including risk-reduction counseling, periodic screening, and partner management must also be considered as STD control options. 12
1. Chernesky M, Jang D, Lee H, et al. Diagnosis of Chlamydia trachomatis
infections in men and women by testing first-void urine by ligase chain reaction. J Clin Microbiol 1994; 32: 2682–2685.
2. Schachter J, Moncada J, Whidden R, et al. Noninvasive tests for diagnosis of Chlamydia trachomatis
infection: application of ligase chain reaction to first-catch urine specimens of women. J Infect Dis 1995; 172: 1411–1414.
3. Rietmeijer C, Yamaguchi K, Ortiz C, et al. Feasibility and yield of screening urine for Chlamydia trachomatis
by polymerase chain reaction among high-risk male youth in field-based and other nonclinic settings: a new strategy for STD control. Sex Transm Dis 1997; 24: 429–435.
4. Cohen D, Nsuami M, Etame R, et al. A school-based chlamydia control program using DNA amplification technology. Pediatrics 1998; 101: E1.
5. Marrazzo J, White C, Krekeler B, et al. Community-based urine screening for Chlamydia trachomatis
with a ligase chain reaction assay. Ann Intern Med 1997; 127: 796–803.
6. Gunn R, Podschun G, Fitzgerald S, et al. Screening high-risk adolescent males for Chlamydia trachomatis
infection: obtaining urine specimens in the field. Sex Transm Dis 1998; 25: 49–52.
7. Bull S, Jones C, Granberry-Owens D, Stoner B, Rietmeijer C. Acceptability and feasibility of urine screening for Chlamydia trachomatis
and Neisseria gonorrhoeae
in community-based organizations: perspectives from Denver and St. Louis. Am J Public Health 1999; 90: 285–286.
8. Jones C, Knaup R, Hayes M, Stoner B. Urine screening for gonococcal and chlamydial infections at community-based organizations in a high-risk morbidity area. Sex Transm Dis 2000; 27: 146–151.
9. Beck-Sague C, Cordts J, Brown K, et al. Laboratory diagnosis of sexually transmitted diseases in facilities within the United States: results of a national survey. Sex Transm Dis 1996; 23: 342–349.
10. Burstein G, Gaydos C, Diener-West M, Howell R, Zenilman J, Quinn T. Incident Chlamydia trachomatis
infections among inner-city adolescent females. JAMA 1998; 280: 521–526.
11. Richey C, Macaluso M, Hook E III. Determinants of reinfection with Chlamydia trachomatis
. Sex Transm Dis 1999; 26: 4–11.
12. Gunn R, Fitzgerald S, Aral S. Sexually transmitted disease clinic clients at risk for subsequent gonorrhea and chlamydia infections: possible “core” transmitters. Sex Transm Dis 2000; 27: 343–349.
13. Rothenberg R. Recidivism redux. Sex Transm Dis 2000; 27: 350–352.
14. Whittington WL, Kent C, Kissinger P, et al. Determinants of persistent and recurrent Chlamydia trachomatis
infection in young women: results of a multicenter cohort study. Sex Transm Dis 2001; 28: 117–123.
15. Scholes D, Stergachis A, Heidrich F, Andrilla H, Holmes K, Stamm W. Prevention of pelvic inflammatory disease by screening for cervical chlamydial infection. N Engl J Med 1996; 334: 1362–1366.
16. Van Voorhis WC, Barrett LK, Sweeney YT, Kuo CC, Patton DL. Repeated Chlamydia trachomatis
infection of Macaca nemestrina fallopian tubes produces a Th1-like cytokine response associated with fibrosis and scarring. Infect Immun 1997; 65: 2175–2182.
17. Hillis SD, Owens LM, Marchbanks PA, Amsterdam LF, MacKenzie WR. Recurrent chlamydial infections increase the risks of hospitalization for ectopic pregnancy and pelvic inflammatory disease. Am J Obstet Gynecol 1997; 176: 103–107.
18. Burstein G, Zenilman J, Gaydos C, et al. Predictors of repeat Chlamydia trachomatis
infections diagnosed by DNA amplification testing among inner-city females. Sex Transm Infect 2001; 77: 26–32.
19. Fortenberry J, Brizendine E, Katz B, Wools K, Blythe M, Orr D. Subsequent sexually transmitted infections among adolescent women with genital infection due to Chlamydia trachomatis,Neisseria gonorrhoeae,
or Trichomonas vaginalis.
Sex Transm Dis 1999; 26: 26–32.
20. Schillinger J, Kissinger P, Berman S, et al. A Randomized Trial of Patient-Delivered Therapy to Prevent Recurrent Chlamydial Infection Among Adolescent and Young Women. Paper presented at: the National STD Prevention Conference; 2000; Milwaukee, Wisconsin.