Risk of Sequelae after Chlamydia trachomatis Genital Infection in Women

Abstract

Chlamydia trachomatis infection, the most common reportable disease in the United States, can lead to pelvic inflammatory disease (PID), infertility, ectopic pregnancy, and chronic pelvic pain. Although C. trachomatis is identified among many women who receive a diagnosis of PID, the incidence and timing of PID and longterm sequelae from an untreated chlamydial infection have not been fully determined. This article examines evidence reviewed as part of the Centers for Disease Control and Prevention Chlamydia Immunology and Control Expert Advisory Meeting; 24 reports were included.We found no prospective studies directly assessing risk of long-term reproductive sequelae, such as infertility, after untreated C. trachomatis infection. Several studies assessed PID diagnosis after untreated chlamydial infection, but rates varied widely, making it difficult to determine an overall estimate. In high-risk settings, 2%–5% of untreated women developed PID within the ∼2-week period between testing positive for C. trachomatis and returning for treatment. However, the rate of PID progression in the general, asymptomatic population followed up for longer periods appeared to be low. According to the largest studies, after symptomatic PID of any cause has occurred, up to 18% of women may develop infertility. In several studies, repeated chlamydial infection was associated with PID and other reproductive sequelae, although it was difficult to determine whether the risk per infection increased with each recurrent episode. The present review critically evaluates this body of literature and suggests future research directions. Specifically, prospective studies assessing rates of symptomatic PID, subclinical tubal damage, and long-term reproductive sequelae after C. trachomatis infection; better tools to measure PID and tubal damage; and studies on the natural history of repeated chlamydial infections are needed.

Genital infection with Chlamydia trachomatis, the most common reportable disease in the United States [1], can lead to serious sequelae among women, including pelvic inflammatory disease (PID), tubal factor infertility, ectopic pregnancy, and chronic pelvic pain [2–7]. Approximately 8% of US women and 15% of Swedish women have reported a PID diagnosis in their lifetimes [8–10]. PID is thought to occur as microorganisms ascend from the lower genital tract, infecting and causing inflammation of the uterus, fallopian tubes, and ovaries [11]. Although the microbial etiology of PID is not fully delineated, C. trachomatis, Neisseria gonorrhoeae, Mycoplasma genitalium, and microorganisms associated with bacterial vaginosis are frequently isolated from the lower and upper genital tracts of women with PID [12–17]. Although C. trachomatis is among the most frequent pathogens associated with symptomatic PID [15, 18, 19], isolated in the upper genital tract of up to a quarter of these patients [12, 18, 20], it has also been associated with a wide spectrum of upper genital tract pathology ranging from asymptomatic endometritis [21–25] to symptomatic, laparoscopically confirmed salpingitis [18]. This highlights the importance of this pathogen in the etiology of both acute PID and subclinical upper tract disease. The reproductive and gynecologic consequences of PID, including infertility [2, 7, 26, 27], ectopic pregnancy [2, 7, 26, 28], recurrent PID [26, 28], and chronic pelvic pain [26–29], can result from damage to the cilia lining the fallopian tubes, fallopian tube blockage or closure, or adhesion formation among pelvic organs.

Because of the potential for C. trachomatis infection to cause serious sequelae, chlamydia screening and treatment programs have been implemented in many countries to shorten the duration of infection, prevent tubal damage among those infected, and reduce C. trachomatis transmission. However, recent surveillance data in several countries, including the United States, suggest that chlamydia rates have not been decreasing, despite ongoing control efforts [30, 31]. This has raised several fundamental questions about the natural history of C. trachomatis infection [32]. For example, if C. trachomatis infections were not detected and treated through a control program, what proportion would result in sequelae? This influences the overall potential benefit of the program and its cost-effectiveness. An even more important consideration may be the timing of tubal inflammation and damage relative to acquisition of infection. This timing affects the likelihood that infections can be detected and treated by a control program before development of symptomatic PID or development of tubal damage that could ultimately lead to infertility or ectopic pregnancy. Contributing to observed increases in chlamydia case rates is likely an increase in repeat infections, which are common in some populations [33]. Thus, another fundamental question is how harmful repeated C. trachomatis infections are in leading to sequelae. This review was developed to address these key questions, which were raised at the April 2008 Chlamydia Immunology and Control Expert Advisory Meeting sponsored by the Centers for Disease Control and Prevention (CDC). This article critically examines evidence addressing the risk and timing of female reproductive tract sequelae after untreated C. trachomatis infection and after repeated chlamydial infection. Gaps in knowledge are identified, and future research needs are proposed.

Methods

A 3-member committee was composed to systematically identify the literature for review. A search of the literature from 1950 through 2008 was conducted with the Medline computerized database of the US National Library of Medicine. The term "Chlamydia trachomatis" was combined with “pelvic inflammatory disease,” “salpingitis,” “endometritis,” “infertility,” or “ectopic pregnancy.” A separate search was conducted as follows: “pelvic inflammatory disease,” “salpingitis,” or “endometritis” and “infertility” or “ectopic pregnancy.” This yielded a total of 3308 citations. Citations were then limited to human studies involving nonpregnant women, and postabortion and transcervical instrumentation studies were excluded. Additional articles were identified by cross-listing bibliographies of reviewed articles. The selected literature was examined for content, and 24 articles deemed to be most relevant to the key questions were selected for critical review. Six articles examined the prospective risk of PID after untreated chlamydial infection [34–39], and 12 examined risk of long-term reproductive sequelae after PID, including either PID of any cause [7, 15, 27–29, 40] or C. trachomatis-associated PID [2, 4, 20, 26, 41, 42]. Two articles prospectively explored the risk of PID after detected and treated chlamydial infection [43, 44], and 6 provided information on the risk of sequelae associated with repeated infection [4, 7, 19, 45–47]. These studies are discussed narratively in the text, and information on study design, population, methods, exposure and outcome measurement, results, strengths, and limitations were tabulated (Tables 1–5).

Results

What is the Risk of Sequelae over Time after an Untreated C. trachomatis Infection?

The ultimate objective of chlamydia control programs is to prevent the most serious long-term reproductive consequences of C. trachomatis infection-mainly, infertility [48]. However, this outcome may not be recognized for several years after a chlamydial infection has caused tubal damage, because the affected woman may not have tried to become pregnant. In addition, there are ethical and technical limitations in following the natural course of infection, because an infection should be treated promptly once it is detected. Thus, although a number of case-control studies have demonstrated associations between serologic evidence of past chlamydial infection and either tubal factor infertility [5, 49–52] or ectopic pregnancy [52, 53, 54], there are no prospective studies directly evaluating risk of longterm reproductive tract morbidity after untreated C. trachomatis infection. PID can serve as a surrogate or intermediary outcome, because its temporal relationship to both chlamydial infection and long-term outcomes is more conducive to study and because it has substantial morbidity and costs [8, 11]. Several studies have attempted to assess the proportion of untreated C. trachomatis infections leading to PID [34–39], and another set of studies evaluated the proportion of PID cases leading to infertility and ectopic pregnancy [2, 4, 7, 15, 20, 26–29, 40–42]. Synthesizing these data can offer some insight into the risk of long-term sequelae after untreated chlamydial infection.

PID after untreated chlamydial infection. It is challenging to assess the true incidence of PID among women with untreated C. trachomatis infection. Despite this, several studies have described aspects of the natural history of untreated chlamydial infection (Table 1). In 3 investigations involving populations at high risk, occurrence of clinically diagnosed PID in women with untreated chlamydial infection was assessed during the ∼14-day interval between testing and treatment. PID occurrence in this interval ranged from 2% to 4.5% among women returning for a follow-up visit [34–36]. In a study in 2 Baltimore sexually transmitted diseases (STD) clinics, 3 (3%) of 93 women who tested positive for C. trachomatis by culture developed PID within a median of 2 weeks between testing and treatment [36]. Similarly, in a prospective STD clinic study involving 129 adults who tested positive for C. trachomatis by culture and nucleic acid amplification tests (NAATs), 2 women (2%) received a diagnosis of PID at a treatment visit that occurred a median of 13 days later [35]. Both of these women had ongoing chlamydial infection, and one had acquired a new gonococcal infection [35]. In a retrospective chart review study, Bachmann et al [34] also investigated the occurrence of PID during the period between testing and treatment in 67 mostly symptomatic women who tested positive for C. trachomatis in an emergency department or other high-risk clinical setting and reported that 3 (4.5%) of 67 women who did not receive therapy at the time of initial evaluation received a diagnosis of PID when they returned for treatment.

Table 1

Studies Assessing the Risk of Pelvic Inflammatory Disease (PID) after Untreated Chlamydia trachomatis Infection

If the mean rate observed in this 2-week period (∼3%) is assumed to be constant and to apply to all women with chlamydia, it would be expected that close to 18% of women would develop PID in 12 weeks, and >50% would develop PID in 1 year. However, 2 studies with longer follow-up periods did not report such high rates of PID (Table 1). The first, conducted in Sweden before the need to treat chlamydia was universally accepted, comprised 109 asymptomatic adolescent girls with untreated, culture-proven C. trachomatis infection, and 4 (3.7%) reported being hospitalized for salpingitis or seen in the emergency department for lower abdominal pain and vaginal discharge in the 3-month observation period [38]. In a more recent long-term follow-up study involving 30 healthy adult women who screened positive for C. trachomatis by NAAT, no women developed symptoms of chlamydial infection, none received C. trachomatis-specific antibiotic treatments, and none received a diagnosis of PID from her general practitioner or gynecologist within 1 year [37].

The highest estimate of PID after untreated chlamydial infection comes from a randomized trial in which 20 women coinfected with C. trachomatis and N. gonorrhoeae received adequate therapy for gonococcal but not chlamydial infection and were followed up for up to 7 weeks [39]. Six women (30%) received a diagnosis of PID from clinicians who were masked to the patients' chlamydial culture results.

Infertility and ectopic pregnancy after PID. Several studies have shown increased risks of reproductive and gynecologic sequelae after PID of any cause (Table 2) [7, 15, 27–29, 40]. Most notably, a landmark prospective study of 2501 Swedish women that was conducted by Weström et al [7] from the 1960s through the 1980s found that 16% of women with laparoscopically confirmed salpingitis developed infertility, compared with 2.7% of control women with clinically suspected PID who did not have salpingitis determined by laparoscopic examination. Infertility was defined by inability to conceive after 1 year of attempting to become pregnant. Tubal factor infertility was confirmed in 11.1% of cases and in none of the control women. In addition, among women with salpingitis, 9.1% of first pregnancies were ectopic pregnancies, compared with 1.4% of first pregnancies among control women. The severity of PID on laparoscopic examination affected long-term outcomes. Overall, 26% of women with clinically suspected PID had normal-appearing tubes on laparascopy; none of these women developed proven tubal factor infertility. Among women with a mild episode of salpingitis, only 0.6% developed tubal factor infertility, but 21% of those with a single episode of severe salpingitis had tubal factor infertility in ensuing years [7].

Table 2

Studies Assessing the Risk of Reproductive and Gynecologic Sequelae after Pelvic Inflammatory Disease (PID) of Any Cause

As part of a randomized controlled trial of treatment regimens for PID (the PID Evaluation and Clinical Health [PEACH] study), Ness et al [15] observed 831 women with mild to moderate clinically suspected PID for adverse outcomes during 1996–1999. Over a mean of 35 months of follow-up, 18% of the women reported infertility, 0.6% had an ectopic pregnancy, and 29% had some degree of chronic pelvic pain, with no differences by treatment arm. Among the 42% who became pregnant, the mean time to pregnancy was 21 months [15]. Laparoscopic verification of PID diagnoses, as done in the study by Weström et al [7], was not feasible in the PEACH study; however, endometrial biopsy was performed for a subset of 614 women [20, 41]. Rates of pregnancy, infertility, and chronic pelvic pain were not significantly different between women with and those without evidence of histologic endometritis [20, 41].

Some studies have also assessed the risk of infertility after PID that is specifically associated with C. trachomatis infection (Table 3) [2, 4, 20, 26, 41, 42]. In a retrospective cohort study involving 51 women hospitalized for PID in the 1980s, among women suffering their first episode of PID, those who were culture positive for C. trachomatis were more likely to experience involuntary infertility than were those who tested negative (relative risk, 2.5; 95% confidence interval [CI], 1.0–6.2) [26]. Furthermore, in a study involving women hospitalized during 1983–1987 for clinically suspected PID, 0 of 10 women with gonococcal PID experienced an adverse reproductive outcome, compared with 7 of 13 women with nongonococcal infection [2]. However, in a more recent study involving 614 women with clinically suspected PID, women with endometrial C. trachomatis infection had rates of subsequent infertility that were similar to those among women who did not have C. trachomatis detected in the endometrium (19% vs 16.8%) [20, 41]. In this study, endometritis and/or endometrial infection with C. trachomatis or N. gonorrhoeae was not associated with reduced pregnancy, elevated infertility, or increased chronic pel vic pain [20, 41]. The reasons for this are unclear. Antichlamydial and gonococcal PID treatment might have reduced the degree of damage preferentially in women with these infections (compared with other causes of clinically suspected PID), and endometritis does not always correlate with salpingitis [55–57]. In addition, women in all groups of this high-risk cohort may have had prior or subsequent C. trachomatis infection that resulted in tubal damage before or after the baseline PID episode, biasing results to the null. Indeed, a separate serologic investigation in this cohort revealed an association between C. trachomatis elementary body antibodies measured during the final year of follow-up and lower pregnancy rates [4].

Table 3

Studies Assessing the Risk of Reproductive and Gynecologic Sequelae after Chlamydia trachomatis-Associated Pelvic Inflammatory Disease (PID)

The symptoms of PID may be less severe with C. trachomatis infection than with N. gonorrhoeae infection [58], which, in turn, may cause women to delay care for PID. In a nested casecontrol study in the cohort observed by Weström et al [7], among 76 case women who experienced infertility or ectopic pregnancy and 337 control women with a subsequent intrauterine pregnancy after PID, C. trachomatis was not associated with impaired fertility overall, compared with other causes of PID (odds ratio [OR], 0.9; 95% CI, 0.5–1.7) [42]. Although C. trachomatis infection was associated with delayed care (OR, 2.1; 95% CI, 1.0–4.1), which, in turn, was strongly associated with impaired fertility (adjusted OR, 2.8; 95% CI, 1.3–6.1), prompt treatment of chlamydia-associated PID dramatically lowered risk of sequelae much more so than did prompt treatment of gonococcal infection [42].

Summary. We found no prospective studies directly assessing risk of infertility after untreated C. trachomatis infection, and precise rates of progression are unknown. However, some data are available on risk of PID after untreated chlamydial infections and risk of infertility and other long-term outcomes after PID. The rate of PID after untreated C. trachomatis genital infection is challenging to determine accurately, because estimates vary widely across studies. In STD clinic or other highrisk populations in which untreated, detected chlamydial infection was followed up for ∼2 weeks, rates of short-term PID diagnosis ranged from 2% to 4.5% [34–36]. If these rates were extrapolated to longer periods, we would expect a greater proportion of patients to develop PID. However, in a population of asymptomatic, untreated C. trachomatis-positive adolescent girls seeking birth control in Sweden, PID occurred in 3.7% over 12 weeks [38]. In the lowest-risk population evaluated thus far, Morré et al [37] observed no PID developing in 30 healthy adult women followed up for 1 year. All of these studies were relatively small and had limitations that could affect the accuracy of risk estimates. Nonetheless, differences in these results may be explained by several possible factors. First, PID rates may not be constant over time for several reasons. A disproportionate amount of PID might occur early in the course of chlamydial infection, when care-seeking in STD clinics or emergency departments is more likely because of recent high-risk behavior or symptoms. Host factors may contribute, with susceptible individuals developing tubal pathology early. Higher organism load may also play a role. In addition, immune responses developing over time could limit progression to the upper genital tract even when the infection is not resolved at the level of the cervix. Second, symptomatic infection prompting care-seeking may result in higher rates of PID than asymptomatic infection (eg, because of differences in host response). Third, there may be a lower threshold for PID diagnosis in high-risk settings or with a known untreated infection. Finally, PID rates may be higher in populations considered to be at high risk of sexually transmitted infections, because they may be more likely to have coinfections or bacterial vaginosis, have a history of PID, or experience recurrent infection [44]. Another factor that may influence differences among rates is the use of highly sensitive NAATs in some studies that may detect infections with a lower C. trachomatis burden and, perhaps, a lower likelihood for progression. The highest rates of PID were seen in the small (n=20) but widely cited study by Stamm et al [39]. Coinfection with N. gonorrhoeae and a greater likelihood for recurrent chlamydial infection in this particularly high-risk population may explain the higher observed rate of sequelae.

After symptomatic PID has occurred, even with treatment, it is associated with significant reproductive and gynecologic morbidity, including infertility, ectopic pregnancy, and chronic pelvic pain [7, 15, 27–29, 40]. In the largest study of its kind, from the 1960s through the 1980s in Sweden, Weström et al [7] found that 16% of women with laparoscopically verified salpingitis developed infertility in the ensuing years, compared with 2.7% of control women with clinically suspected PID but no laparoscopic evidence of salpingitis. Ness et al [15] found that 18% of women developed infertility after clinically diagnosed PID during the 1990s in the United States, and the rate did not differ by presence or absence of histologic endometritis in a subsequent analysis by Haggerty et al [20]. In the study by Weström et al [7], severity of PID, as judged by laparoscopic examination, was associated with infertility, suggesting that tubal damage sustained at the time of acute PID may lead to sequelae [7, 29]. Among women with clinically suspected PID, none of those with normal-appearing tubes developed tubal factor infertility, whereas 21.4% of women with an episode of severe salpingitis did [7]. Although PID of any cause is strongly linked to sequelae [2, 4, 20, 26, 41, 42], data from the largest studies suggest that chlamydial PID is no more or less likely to lead to sequelae than other causes of PID [20, 41, 42].

When using PID as an intermediary outcome to estimate risk of long-term reproductive sequelae resulting from untreated C. trachomatis infection, it is important to understand the extent to which chlamydial infection may lead to these sequelae outside the pathway involving symptomatic PID.Most women with tubal factor infertility and ectopic pregnancy have no history of diagnosed PID, including women in case-control studies showing strong associations between these outcomes and serologic evidence of past chlamydial infection [3, 5, 49]. However, in one study, further questioning of infertile women with no history of diagnosed PID revealed that 60% of those with tubal infertility reported health care visits for abdominal pain, compared with only 19% of those without tubal disease [59]. Nonetheless, it is known that chlamydial infection can cause asymptomatic or mildly symptomatic upper tract infection and inflammation [23, 25]. In addition, pathologic damage in fallopian tube biopsy specimens from women with tubal infertility is similar whether or not there is a history of overt PID [60]. Subclinical tubal infection and inflammation likely lead to some degree of infertility and other complications, but the full extent to which this occurs remains unclear.

Research needs and future directions. Quantifying the risks of PID, infertility, and ectopic pregnancy after untreated C. trachomatis infection would provide vital data for chlamydia control programs and for clinicians to share with patients on the importance of screening to prevent sequelae. To better understand the risk and timing of sequelae after untreated C. trachomatis infection, improvements must first be made in measuring the short-term complications of chlamydial infection. All of the studies reviewed in Table 1 followed up women for the development of clinically suspected PID and were therefore limited by the imprecise measurement of this outcome. The studies were also unable to capture cases of asymptomatic tubal inflammation and damage. As diagnostic misclassification compromises not only the estimation of PID after an untreated chlamydial infection but also sequelae after PID, it is of critical importance to develop standardized and innovative methods to ascertain both acute PID and subclinical tubal involvement associated with chlamydial infection. To increase sensitivity, the CDC recommends the minimum criteria for the diagnosis of clinically suspected PID as either uterine or adnexal tenderness or cervical motion tenderness [61]. However, this clinical approach, used by many studies to identify cases of PID, suffers from extremely poor specificity [17]. Laparoscopic examination or endometrial biopsies have been used by some studies to confirm PID, with laparoscopic examination considered to be the gold standard. However, neither of these confirmatory methods is very precise. Compared with laparoscopically diagnosed salpingitis, histologically confirmed endometritis has a sensitivity of 70%–89% and a specificity of 67%–92% [55– 57]. Even laparoscopic examination has been found to have an extremely low sensitivity for the diagnosis of PID (25%–50%), when compared with fimbrial minibiopsy showing histopathologic evidence of PID [62, 63]. Furthermore, laparoscopic examination, which lacks standardization and relies on subjective interpretation of pelvic structure photographs, has only a fair intraobserver reproducibility for the diagnosis of PID (κ=0.58) and a poor to fair interobserver reproducibility (κ=0.43) [62].

In addition to concerns about its sensitivity and standardization, laparoscopic examination is an invasive procedure and is not routinely used in clinical practice. Noninvasive measures of PID are needed not only to be more clinically feasible but also to capture cases of subclinical tubal involvement in clinical studies. Magnetic resonance imaging (MRI) has been investigated as an alternative diagnostic procedure, although MRI facilities are not widely available at settings where patients with PID are typically seen. Data are limited, but those from at least 1 study (n=30) suggest that MRI is sensitive (95%) and specific (89%) for the diagnosis of PID, compared with laparoscopic examination [64]. Transvaginal ultrasound is another minimally invasive procedure, but it has a much lower sensitivity for laparoscopically diagnosed PID (32%–81%) [64, 65]. Power Doppler, a recent innovation that allows improved detection of blood flow and inflammation-induced hyperemia, has been found in a study to have both high sensitivity (100%) and high specificity (80%), compared with laparoscopic examination [66]. Lastly, vaginal white blood cell count was found to be a sensitive marker of upper genital tract infection in a study involving 121 women meeting the CDC's minimal criteria for PID [67]. More work is needed to verify the diagnostic accuracy of these newer measures and additional inflammatory markers, such as interferon and other cytokines.

Next, to fully understand the natural history and sequelae of untreated chlamydial infection, we need additional prospective studies assessing rates of both clinically suspected PID and asymptomatic tubal inflammation after C. trachomatis infection in diverse populations encompassing the full spectrum of symptomatology and risk of sexually transmitted infection. Additional information on the 12-month incidence of PID after untreated C. trachomatis infection among asymptomatic college-aged women was recently collected as part of a randomized trial of chlamydia screening in the United Kingdom that was conducted before such screening was nationally recommended there [68, 69]. Although final results of the study were published too late for inclusion in this review, the natural history analysis revealed that 9.5% of 74 women with untreated chlamydial infection developed PID in 12 months [70]. Studying the timing of PID occurrence is also critical. The picture emerging from the studies listed in Table 1 suggests higher short-term rates of PID, with risk of PID decreasing after the first few weeks, and low rates within a year after asymptomatic infection. Understanding the timing of PID development is critical in optimizing the frequency and structure of chlamydial screening and other control strategies. Natural history studies are limited by the fact that it would be unethical to withhold treatment for diagnosed chlamydial infection, and it is unclear how long a woman has already had infection at the time it is detected through testing. Nonetheless, creative strategies to develop prospective studies of chlamydia natural history are vital. Innovative use of stored genital specimens from existing or ongoing prospective studies of other infections (eg, human papillomavirus vaccine trials and human immunodeficiency virus prevention trials) in which specimens are collected beyond those used to diagnose and treat chlamydial infection as part of standard medical practice might also provide opportunities for better understanding of chlamydia natural history. Any study of C. trachomatis natural history would have to be carefully designed to ensure protection of human subjects. Finally, because of the challenges facing accurate diagnosis of PID and the occurrence of asymptomatic chlamydial upper tract involvement, as well as the difficulties in obtaining better natural history data, primary and secondary prevention strategies for C. trachomatis infection and its sequelae should be a focus of future studies, as discussed by Gottlieb et al in this supplement [71].

A primary necessity for research on sequelae after PID is identification of better, more proximal markers of tubal damage that are predictive of long-term sequelae. This would not only allow the outcomes of chlamydial infection to be more accurately classified but would also make prospective research on chlamydia and long-term outcomes more feasible. The landmark study by Westro"m et al [7] provided excellent data on risk of sequelae among women who were hospitalized with PID, compared with a control group of women with abdominal pain who did not have laparoscopically verified PID. However, these data were obtained in Sweden 20–40 years ago in a potentially much different microbiological and clinical milieu (eg, higher prevalence of N. gonorrhoeae and older PID treatment regimens). The PEACH study provided modern-day estimates of adverse outcomes after mild to moderate PID in the United States and stratified participants according to endometrial biopsy results but did not include a control group of women without clinically suspected PID [15]. Additional studies evaluating reproductive and gynecologic morbidity among women with PID, compared with an appropriate control group, in a modern-day setting would be valuable. In addition, prospective studies evaluating the risk of reproductive sequelae after subclinical upper genital tract infection and inflammation are needed. Preliminary data from a prospective study showed that 17% of 58 women with subclinical endometritis at a baseline visit had evidence of fallopian tube damage demonstrated by hysterosalpingogram 3 months later, whereas only 8% of 362 women without endometritis had such evidence [72]. Final results from this study have not yet been published. Finally, current evidence suggests that the vast majority of women infected with C. trachomatis do not develop PID, and not all women with chlamydial PID become infertile. Host factors and immunologic predictors explaining differences in morbidity risk should be explored in future studies, as discussed by Darville et al in this supplement [73]. Differences in morbidity after C. trachomatis infection may also be explained by simultaneous infection with other pathogens, such as N. gonorrhoeae [15] and M. genitalium [74], and the impact of such coinfection should be explored in future studies of PID and its sequelae.

Is the Risk of Sequelae Increased during a Repeat Chlamydial Infection?

PID after ⩾1 detected and treated chlamydial infection. A prospective assessment of PID after detected and treated C. trachomatis infection comes from a study of 1170 women from 5 US sites; all of the women were at high risk of chlamydia based on their demographic risk scores (Table 4) [44]. The women were tested for C. trachomatis and N. gonorrhoeae at baseline and were retested at follow-up visits every 6–12 months for a median of 3 years. Of these women, 122 tested positive for C. trachomatis at baseline and received antibiotic therapy. Twenty-three C. trachomatis-positive women (18.8%) received a diagnosis of PID (primarily mild to moderate) during followup. This rate of PID was substantially higher than that among women who did not have gonococcal or chlamydial cervicitis at baseline (7.0%). The etiology of subsequent PID episodes was unknown. The incidence of severe PID from any cause, stratified by C. trachomatis test history, was also assessed in a retrospective cohort study involving 43,715 Swedish women followed up from 1985 through 1999 [43]. Low et al [43] found that, by 15 years of follow-up, 6% of women had tested positive for C. trachomatis (and were assumed to have been treated), 4% of those who were screened and tested negative, and 3% of those never screened were subsequently treated for PID. Although some outpatient data were captured, most of the registry data were from inpatient records and, therefore, primarily measured the overall rate of severe PID. Women who tested positive for C. trachomatis were 50% more likely to be subsequently treated for PID than were women who tested negative (hazard ratio [HR], 1.5; 95% confidence interval [CI], 1.2–1.8), although this relationship was attenuated when adjusted for demographic and socioeconomic factors (HR, 1.3; 95% CI, 1.0–1.6) [43].

Table 4

Studies Assessing the Risk of Sequelae after at Least 1 Detected and Treated Chlamydia trachomatis Infection

Repeat infections with C. trachomatis are common [33, 75, 76] and may contribute to the higher incidence of PID among women at high risk [39, 44], compared with women in the general population [37]. Similarly, the higher risk of repeat chlamydial infection among women with ⩾1 detected infection may help explain the higher rates of PID associated with longer follow-up of these women [43, 44].

PID after repeat chlamydial infection. The association between repeat infection and PID sequelae was assessed by a retrospective cohort study involving 11,000 women and girls aged 10–44 years who tested positive for C. trachomatis in Wisconsin during 1985–1992 (Table 5) [46].Women who tested positive twice were 4 times as likely (OR, 4.0; 95% CI, 1.6–9.9) and women who tested positive ⩾3 times were 16 times as likely (OR, 6.4; 95% CI, 2.2–18.4) to receive a diagnosis of PID [46]. It is difficult to determine the true impact of repeat chlamydial infection on PID from this study, however, because clinicians may be more likely to diagnose PID in women with a history of repeated chlamydial infection. Similarly, a prospective cohort study involving 302 urban female sex workers in Nairobi, Kenya, reported a significant relationship between repeated C. trachomatis isolation and the cumulative risk of chlamydial PID over ∼18 months (adjusted OR, 1.8; 95% CI, 1.3–2.4) [47]. However, the risk of PID with each individual chlamydial infection appeared to be similar among those with one infection and repeated infections [47], although the power to detect a difference may have been limited. Thus, this study suggests that, although cumulative risk increases, the risk of PID per chlamydial infection may not be any greater with each recurrent episode. Although these studies were unable to distinguish between persistent and new repeat infection, they suggest that the risk of PID increases in parallel with the number of detected C. trachomatis infections.

Table 5

Studies Assessing the Risk of Sequelae after Repeat Chlamydia trachomatis Infections

Further demonstration of the relationship between recurrent chlamydial infection and risk of PID was evident in a prospective study involving 443 women with clinically suspected mild to moderate PID who were followed up for a mean of 84 months with repeated chlamydial serologic testing [4]. Although baseline antibodies to C. trachomatis elementary bodies were not associated with reproductive morbidity, rates of PID recurrence were higher among women whose anti-chlamydial antibodies were in the highest tertile during the final year of follow-up (adjusted HR, 2.5; 95% CI, 1.0–6.3). Later serologic testing, reflecting both baseline and subsequent chlamydial infections, was associated with PID recurrence, suggesting that additional exposures to C. trachomatis may increase the risk of subsequent PID [4]. Supporting this was the finding that consistent condom use in the same population was associated with a marked decrease in the incidence of recurrent PID [19].

Long-term reproductive sequelae after repeat chlamydial infection. In the same way that repeated chlamydial infection may increase the risk of PID, recurrent infection with C. trachomatis may also increase the risks of infertility and ectopic pregnancy. There is good evidence to suggest that recurrent PID increases sequelae risk, as first evidenced in the landmark Scandinavian cohort study involving 1844 women with laparoscopically confirmed salpingitis that was conducted by Westro "m et al [7]. In this study, each episode of salpingitis roughly doubled the risk of tubal factor infertility (8% after 1 episode, 20% after 2 episodes, and 40% after 3 episodes) [7]. Similarly, the studies by Ness et al [4, 19], in which higher titers of antichlamydial antibodies at follow-up and less consistent condom use were linked with recurrent PID, also showed that these factors were associated with longer times to pregnancy. These findings suggest that additional exposures to C. trachomatis after an episode of PID can lead to an increased risk of longterm complications.

In the retrospective cohort study assessing diagnosed chlamydia and sequelae risk that was conducted by Hillis et al [46], women who were identified as C. trachomatis positive 2 times in Wisconsin county databases from 1985 through 1992 were twice as likely (OR, 2.1; 95% CI, 1.3–3.4) and those with ⩾3 diagnosed infections were >4 times as likely (OR, 4.5; 95% CI, 1.8–5.3) to be hospitalized with an ectopic pregnancy. Another registry study involving 20,762 Norwegian women using the health care system from 1990 through 2003 reported a similar dose-response relationship between detected C. trachomatis infection and ectopic pregnancy [45]. Compared with women who tested negative for C. trachomatis, women with a history of a diagnosed chlamydial infection had almost a 2-fold increased risk of ectopic pregnancy (adjusted HR, 1.8; 95% CI, 1.1–3.0) and those with ⩾2 diagnosed chlamydial infections had a 3-fold increased risk of ectopic pregnancy (adjusted HR, 3.0; 95% CI, 1.6–5.6) [45].

Summary. Long-term follow-up studies of the period after treated chlamydial infection show that women with ⩾1 detected C. trachomatis infection have higher rates of PID in the ensuing years than do women without a detected infection, with PID rates near 20% over 3 years in a high-risk population [44]. Although a detected chlamydial infection may simply be a marker for high-risk sexual behavior and exposure to other sexually transmitted infections, one possible explanation for these findings is an increased risk of PID related to repeated C. trachomatis infections, which are common [75, 76].

Several studies have shown that the cumulative risk of PID [4, 46, 47] and long-term reproductive sequelae [4, 45, 46] increases with repeated chlamydial infections. However, it remains unclear from these epidemiologic studies whether the risk of sequelae from a given chlamydial infection is higher with each additional repeat infection. Furthermore, methodological problems make it difficult to sort out how much of the association between recurrent chlamydia and PID is attributable to biologically plausible mechanisms and how much is attributable to diagnostic ascertainment bias. Certainly, physicians' knowledge about prior positive chlamydial results may influence their differential diagnosis of lower abdominal pain. Because of the asymptomatic nature of chlamydial infection, it is also difficult to determine how many chlamydial infections a woman has actually had, if she did not seek medical care. Furthermore, it is difficult to determine whether a first diagnosed infection is truly primary and how many past infections have occurred when there is evidence of past infection. In all of these studies, infections were detected and therefore treated; however, perhaps the most enhanced pathologic memory immune response may occur after an initial infection that has resolved on its own. Lastly, the inability to distinguish between persistent and repeat infection limits interpretation.

Research needs and future directions. Studies on the natural history of repeated chlamydial infections are needed. Such studies should determine how the risk of PID in a given period after a repeat infection compares with the risk of PID in an equivalent period after an initial infection. In addition, because it is difficult to determine whether a woman's first diagnosed chlamydial infection is truly primary, natural history studies that conduct frequent C. trachomatis screenings and PID evaluations among a group of young, seronegative women are desirable. To conduct these natural history studies, a better understanding of chlamydia-associated antibodies would be valuable, in terms of the proportion of infections that result in seroconversion, the time course of seroconversion, duration of seroreactivity, and titers with initial and repeat infection. Furthermore, better markers of repeat infection and immunologic and host factors that predict worse tubal damage with repeat infection should also be explored [73]. Because of the high rates of PID from any cause in the years after a detected chlamydial infection in some populations [44], we also need studies of prevention strategies focused on women who have already received a diagnosis of at least one chlamydial infection. Recently, Ness et al [19] reported that consistent condom use was associated with a 30%–60% reduction in recurrent PID in a subgroup of 684 sexually active women followed up after an initial episode of PID. However, additional studies are needed to confirm these data and to determine optimal prevention strategies after diagnosed chlamydial infection in addition to those after PID.

Conclusion

Although the evidence linking C. trachomatis with tubal pathology is strong, there remains a great deal of uncertainty about the progression rates of both PID and reproductive sequelae among women acquiring C. trachomatis infection. Furthermore, the ability to link a specific chlamydial infection with later reproductive and gynecologic morbidity is limited. Prospective studies assessing the rates of symptomatic PID, asymptomatic tubal damage, and reproductive sequelae after C. trachomatis infection; better tools to measure PID and tubal damage; and studies on the natural history of repeated chlamydial infections are needed to better understand the longterm risks of chlamydial infection.

Potential conflicts of interest: none reported.

Financial support: None reported.

Supplement sponsorship: This article is part of a supplement entitled “Chlamydia trachomatis Genital Infection: Natural History, Immunobiology, and Implications for Control Programs,” which was sponsored by the Centers for Disease Control and Prevention.