Acute Pelvic Inflammatory Disease
David A. Eschenbach
Table Of Contents
David A. Eschenbach, MD
PATHOGENESIS AND PATHOLOGY
The true incidence of PID is unknown because of inaccuracy of diagnosis. The frequency of recognized PID increased dramatically from 1960 to 1975, with a peak in Sweden in 1977 and in the United States in 1993, after which the frequency began to decrease.4–6 This increase was particularly dramatic for women younger than 25 years. Among teenagers, the incidence of PID is estimated to be 1.5% annually, and it is the most frequent serious infection that teenagers acquire.
The cause of the rise in the rate of PID has not been well documented, but it is in part related to an increase in sexually transmitted agents known to cause PID, to changes in sexual behavior, and to changes in contraceptive methods. For example, a parallel rise was observed in the rates of PID and gonorrhea.7 However, in European countries, the proportion of gonorrhea-associated PID has decreased while there was a dramatic increase of total PID due to diseases other than gonorrhea.3 The influence of changes in sexual behavior among women has been less well documented, but an increase in the proportion of young sexually active women may increase the PID rates. The influence of contraceptive changes is discussed later. In the 1980s, up to 15% of women of reproductive age had prior PID, and many were left with the sequelae of the infection. In 1980, and estimated 3.5% of women 30 years of age were sterile or had an ectopic pregnancy as a result of PID.3
Recurrent Pelvic Inflammatory Disease
A subsequent episode occurred in 25% of women after one episode of PID.8 The frequency of infertility and other sequelae increases after multiple infections, and physicians should attempt to reduce subsequent PID through education, treatment of male sexual partners, and close follow-up. The high subsequent infection rate has been related to the susceptibility of damaged fallopian tube mucosa to recurrent infection and perhaps to unrecognized persistent infection. Women with sexually transmitted organisms tend to reacquire these organisms, which further contribute to recurrent PID.
Infertility is one of the most common and dreaded consequences of PID. PID is a major cause of infertility, and 30% to 40% of infertile women have tubal disease from prior salpingitis. Infertility resulting from tubal occlusion occurs in 8% of women with one episode, 23% of women with two episodes, and 43% of women with three episodes of PID.9,10 The frequency of tubal occlusion also is related to the type and severity of PID and to age. Infertility is more common when PID is caused by organisms other than gonococci.9 Infertility is more common after laparoscopically documented severe tubal infection (21%) than after moderate (16%) or mild (6%) tubal infection.9,10 Older women are more likely than young women to develop tubal occlusion after PID.3 Tubal infertility is also increased among women with >3 days of pain before treatment.11 The cost of infertility resulting from PID is estimated to be $60 billion annually, ten times the cost of treating acute PID.12
The number of ectopic pregnancies reported in the United States doubled in the 1980s but has decreased since 1990.13 Between 1930 and 1980, the frequency of ectopic pregnancy after PID increased,3 perhaps related to antibiotic therapy, which reduces total tubal occlusion rates but does not reverse tubal damage sustained before therapy. Tubal damage from prior PID can be observed in about 50% of women who have an ectopic pregnancy.3 Tubal entrapment of the ovum or slowed tubal transit time from damaged cilia presumably accounts for the tubal rather than intrauterine implantation.
Seven percent of women having their first pregnancy after PID develop an ectopic pregnancy.3,14 The risk of an ectopic pregnancy is 7 to 10 times greater in women after PID than in women without prior PID. Ectopic pregnancy rates increase with the more severe the tubal damage and with the number of episodes of PID.15 The impact of ectopic pregnancy on infertility is considerable; about one half of women with an ectopic pregnancy never again conceive, and about 20% of those with an ectopic pregnancy have another ectopic pregnancy if they do conceive.15
Chronic Abdominal Pain
Chronic abdominal pain lasting longer than 6 months occurred in 17% to 18% of women after one or more episodes of PID, compared with 2% to 5% of women who have never had PID.6,16 Chronic pain is present in 12% with one episode, 30% with two, and more than 60% with three or more episodes.16 The pain is often constant, interferes with work, and results in increased physician visits, hospitalizations, and abdominal operations.17 Increased pain with menses and with intercourse is also common.17 The concept that the pain is related to residual pelvic pathology is strengthened by an increased rate of infertility and dyspareunia among women with pain after PID. Many of the hysterectomies and salpingo-oophorectomies after PID are performed because of chronic pain, although this surgery should diminish because pain can often be controlled by ovarian hormonal suppression or by laparoscopic lysis of pelvic adhesions.
|PATHOGENESIS AND PATHOLOGY|
The vagina normally contains a predominance of facultative lactobacilli and numerous but low concentrations of other facultative and anaerobic bacteria.18 It is unclear whether the endocervix normally supports a flora19 or the recovery of bacteria from the cervix represents cervical contamination. However, bacteria are not recovered by culture from normal endometrium and fallopian tubes,20 although it is possible that bacteria may transiently colonize the upper genital tract during menses. Factors important for the dissemination of organisms from the cervicovaginal flora to the uterus and fallopian tubes are not well understood, but data collected in several studies suggest that cervicitis, vaginitis, and endometritis are related to PID (Table 1).
Cervicitis is a common infection that is often the source of organisms that cause PID. A reduction in the prevalence of cervicitis probably reduces the rate of PID, and cervicitis should be actively diagnosed and treated.
Cervicitis can be diagnosed by finding mucopus (i.e. yellow or opaque appearance of endocervical mucus on a swab or in situ) or by identifying ten or more white blood cells per 400× power on a Gram stain of cervical mucus.21 The cervical surface has to be wiped free of vaginal material to properly visualize for mucopus. Infection within columnar cervical epithelial cells also often produces erythema, edema, and bleeding of the cervix. Chlamydia trachomatis and Neisseria gonorrhoeae are the most common organisms causing cervicitis,21 and together they can be isolated in more than one half of the women with the infection. The cause of infection in the remaining women is unknown.
Spontaneous PID is limited to menstruating women. Women who develop PID with N. gonorrhoeae or C. trachomatis usually experience the onset of pain within 7 days of their menstrual period.22 This observation suggests that PID caused by these bacteria is influenced by menses. The growth of N. gonorrhoeae, particularly the virulent transparent gonococcal phenotypes, may be stimulated during the secretory phase of the cycle just before menstruation.23 The bacteriostatic effect of the cervical mucus on bacteria, which is greatest during the ovulatory phase of the cycle, is at its minimum before the onset of menses.24 Cervical mucus also probably acts as a mechanical barrier, and the loss of cervical mucus during menses potentially allows bacteria to penetrate into the endometrial cavity at this time. Gonococci,25 chlamydiae,26 and other bacteria appear to attach to sperm, and it is even possible that sperm can act as vectors to carry these bacteria into the uterus and fallopian tubes.
Acute infection also is enhanced by the presence of virulent bacteria and diminished by the presence of specific antibody. Some bacteria are more virulent than others. N. gonorrhoeae and C. trachomatis consistently cause tubal infection in experimental animals and in a high proportion of women with these bacteria in the cervix. In contrast, genital mycoplasmas and bacteria of the normal cervicovaginal flora are less virulent than gonococci or chlamydiae, and they cause PID in only a small proportion of those with these microorganisms in the cervix. Most women with cervical gonorrhea for longer than 1 month have a gonococcal bactericidal antibody that may protect against the development of PID.27 Women without the bactericidal antibody are susceptible to the development of PID.27 It is likely that most women destined to develop salpingitis do so during the first few menses after the acquisition of gonorrhea and before the development of immunity conferred by gonococcal antibodies. C. trachomatis infection may be governed by similar humoral immune mechanisms.
Intrauterine devices (IUDs) annually cause acute salpingitis among 1% to 2% of IUD users. An IUD's tail, particularly a multifilament tail, is capable of assisting movement from the cervicovaginal area to the uterus.28 The tail acts as a reservoir for bacteria through colonization on the surface.29 IUDs may also increase infection rate by promoting anaerobic bacterial growth, by producing microulcers, or by producing a chronic inflammatory reaction of the endometrium30 or the fallopian tube.31
The usual infection ascends along the mucosa from the cervix into the endometrium and to the fallopian tubes. After bacteria reach the uterus, they may be transported to the fallopian tubes by contiguous spread or carried by ciliary motion. Endometritis appears to be an intermediate level of infection between cervicitis and salpingitis. Endometritis diagnosed by plasma cells within the endometrium32 is present in approximately 50% of women with cervicitis.33 Women with endometritis have an increased rate of abnormal vaginal bleeding and uterine tenderness. It is possible for women to have symptoms of endometritis for considerable time before the development of overt salpingitis, and women with these findings should be scrutinized for cervicitis and endometritis.
Once in the fallopian tube, organisms attach to the mucosa and initiate endosalpingitis. Experimental gonococcal infection has provided a model for study of salpingitis. Gonococci attach to nonciliated mucosal epithelial cells and are phagocytized into the cell. Mucosal cells become markedly distorted. A reaction between IgG antibody and gonococci fixes complement and generates an intense inflammatory reaction that includes white blood cells, edema, vasodilatation, and loss of cilia. Liberation of a gonococcal endotoxin causes tubal cilial destruction on the adjacent uninfected ciliated cells.34N. gonorrhoeae lipopolysaccharide also effects ciliary activity. The intense inflammatory reaction causes the short duration and marked symptoms characteristic of gonococcal PID.
C. trachomatis is engulfed within epithelial cells, where it replicates. The takeover of the epithelial cell metabolism and the rupture of mature chlamydial particles back into the tubal lumen causes death of the epithelial cell during chlamydial infection. The intracellular location of C. trachomatis offers limited protection of the bacteria to immune recognition. In human fallopian tube culture, C. trachomatis invaded ciliated and nonciliated tubal cells,35 and in experimental monkey infection, both cell types were lost. Chlamydial infection also induced an inflammatory response characterized by cytokines, including interferon-γ (INF-γ).36 Expression of major histocompatibility complex class II molecule-bound exogenous antigens against epithelial cells activates an immune response directed against the tubal cell itself, resulting in tubal destruction. Cell-mediated mechanisms play the dominant role in the resolution of chlamydial infection. C. trachomatis DNA in the tube has been identified by immunofluorescence among infertile women, even after antibiotic therapy,37 raising the possibility that chlamydiae can remain in tubal mucosa even after treatment.
In experimental infection in monkeys, repeated C. trachomatis inoculation results in an accelerated tubal inflammatory reaction and increased tubal scarring.38 The first exposure of the tube to C. trachomatis produced a self-limited infection and no tubal scarring, but repeated inoculation produced tubal occlusion and peritubal adhesions.38,39 A cyto-kine response of INF-γ and various interleukins occur with experimental infection.36 With acute infection, 60% of the lymphocytes were cytotoxic lymphocytes, which together with finding porforin indicates activation of cytotoxic lymphocytes.36 Repetitive infection may produce the severe permanent tubal damage by the immunopathogenic lymphocyte response. The importance of cytotoxic lymphocytes is furthered by the observation that chlamydial salpingitis occurs more frequently among humans with human leukocyte antigen-31 (HLA-31).40 HLA-31 is an MHC class I molecule that presents cytoplasmically processed antigen to cytotoxic lymphocytes.
Helper T lymphocytes may also play a role in the immunopathology of chlamydial salpingitis. Helper T lymphocytes participate in a delayed-type hypersensitivity (DTH) response. DTH is produced by placing chlamydial heat shock protein-60 (CHSP-60) in the conjunctiva of animals previously infected with Chlamydia.41 Antibody to CHSP-60 increases the risk of developing chlamydial PID fivefold after chlamydial cervical infection in humans40 and is associated with severe tubal disease observed with acute infection42 and with tubal infertility43 and ectopic pregnancy.44
Local and serum IgA, IgM, and IgG class antibodies are made in response to chlamydial infection. Partial protection against reinfection has been observed, but this protection appears short and serotype specific.45 Antibody appears to play a small role compared with cell-mediated immune responses in the resolution of infection.
N. gonorrhoeae and C. trachomatis do not produce abscesses in animals or humans.46 Abscesses form from a combination of facultative and anaerobic bacteria. Anaerobes dominate the local environment of low oxygen tension.47
Gonococcal infection is usually limited to the mucosa of the tube. In contrast, aerobic and anaerobic bacteria and perhaps chlamydiae often cause inflammation below the basement membrane,28 which can result in fibrosis and permanent tubal damage. Inflammation can also spread to the muscularis and serosal layers. The tubal lumen remains open early in the infection, and infected material may be exuded through the fimbriated end, producing a localized or generalized peritonitis and contamination of ovarian, bowel, and peritoneal surfaces. Later in infection, the fimbriae can be obstructed. Intraabdominal abscess formation occurs in 5% to 15% of patients with acute salpingitis as a result of a mixed aerobic and anaerobic infection. Abscess can occur within the tube, within the ovary, or between genital and gastrointestinal structures.
Genital Mycoplasmas may not cause primary tubal infection. Little is known about Mycoplasma infection in humans, but in studies of grivet monkeys, Mycoplasma hominis cause inflammation along the vascular and lymphatic channels of the broad ligament.48 Tubal serosal swelling occurs, but tubal mucosa is neither inflamed nor infected. Lymphatic or hematogenous spread of organisms other than Mycoplasma from the uterus to the broad ligament and ovaries is uncommon among nonpregnant women because of the endometrial-myometrial barrier.
A popular but unproved hypothesis suggests that aerobic and anaerobic bacteria frequently invade the fallopian tubes after an initial gonococcal or chlamydial infection. The common finding of concomitant tubal infection with these bacteria supports this hypothesis. Virulent bacteria are present in the vagina. Aerobic and anaerobic bacteria are also capable of producing primary salpingitis after uterine instrumentation, in the presence of an IUD, and occasionally without any known risk factor. The presence of particularly virulent organisms such as streptococci or Haemophilus or the presence of high concentrations of organisms, as in the case of bacterial vaginosis, may be prerequisites for a primary infection with those organisms.
After bacteria are eliminated by antibiotics, residual inflammatory cells remain in the tubal submucosa for prolonged periods. However, ordinary bacteria were uncommonly recovered from the tube, ovaries, or adjacent structures after antibiotic therapy.2 An exception may occur with C. trachomatis, which has been recovered from the fallopian tubes and peritubal tissue of women undergoing tuboplasty months or years after an episode of PID.37C. trachomatis has been identified in 35% of women with gross tubal inflammation and in 20% without gross tubal inflammation,49 which raises the possibility that a prolonged infection and an inflammatory response can result from chlamydiae.
The highest rate of PID occurs among the youngest sexually active women. The PID rate of 15- to 19-year-old women is approximately three times that of 20- to 24-year old women and is approximately twice the PID rate of 25- to 29-year old women.3 PID is clearly the most common serious infectious disease of young women. However, sexually transmitted disease (STD) rates are highest in young women and age is no longer an important risk factor for PID when N. gonorrhoeae and C. trachomatis are controlled.
Women with an IUD are two to three times more likely to develop PID than nonusers.50,51 IUD use has been a risk factor for PID even in recent reports.51 A foreign body increases infection rates, and human52 and primate53 IUD users frequently have intrauterine bacterial colonization, while nonusers remain without bacteria. Despite controversy about the selection of control groups and the criteria used to diagnose PID in some studies, the combined weight of epidemiologic and in vitro and in vivo data represents overwhelming evidence that IUDs can cause PID.
Although most patients with PID are in a low socioeconomic group, IUD use constitutes a higher relative risk of PID for patients of high than low socioeconomic status. Patients who present with an IUD-associated infection are therefore often nulliparous women who do not have gonorrhea and are in a middle to high socioeconomic group.54 The severity of PID among IUD users is similar to nonusers. The impact of IUD-caused salpingitis is further emphasized by reports of threefold increased rates of primary infertility among copper IUD users55 and ectopic pregnancy among IUD users compared with women who never used an IUD.
Multiple mechanisms are probably responsible for the increased rate of PID among IUD users. Early studies suggested that bacteria that were introduced into the uterus during the insertion of an IUD were cleared within 48 hours20 and that the endometrial cavity of an IUD user was sterile. However, later data indicate endometrial bacteria in all women with tailed IUDs months after insertion, although not in women who used tailless IUDs or no IUD.52
Anaerobic vaginitis is more common among IUD users that IUD nonusers, and anaerobic Actinomyces species have been identified in large concentrations more frequently among IUD users than nonusers.56 These observations raise the possibility that an IUD promotes the growth of anaerobic bacteria, perhaps to a critical level, which contributes to infection. Although Actinomyces israelii has been isolated from only a few IUD-associated PID infections, until the mechanism and PID attack rate are better described, IUD-using women with Actinomyces identified on Papanicolaou smear should be told of the organism's presence and asked to immediately report abnormal discharge, bleeding, and pain. Symptomatic women with Actinomyces should have the IUD removed and receive antibiotic treatment. However, asymptomatic women with Actinomyces should not be treated or necessarily have to IUD removed, particularly if plans include reinsertion of another IUD. It is estimated that the net rate of PID would be higher if the IUD was removed and reinserted than if it were left in situ because of the low risk of PID from Actinomyces.
Douching has been related to salpingitis and ectopic pregnancy.57 Douching three or more times a month with each commercial product was associated with a threefold increased rate of PID when demographic variables, gonorrhea, and Chlamydia infection were controlled in a multivariate analysis.58 Douching was usually undertaken as a habit and not in response to symptoms of infection. Because there is no known benefit to douching, patients should be discouraged from the practice.
Barrier contraception reduces the acquisition of STDs and appears to confer modest reduction in the rate of PID. However, barrier methods need to be used consistently to protect against STDs, and male condom use is usually inconsistent.
Oral contraceptive (OC) use, perhaps because of the progesterone content, appears to reduce the rate of PID.3,59 OCs may markedly reduce the rate of PID (up to 80%) among patients with C. trachomatis but have little effect among patients with N. gonorrhoeae or neither organism.59 OCs may downregulate the immune response to C. trachomatis salpingitis. There is evidence that mild salpingitis is significantly more likely to result when a woman is on OCs than if she uses an IUD or other contraceptives.60
Previous Pelvic Inflammatory Disease
Women with prior PID are twice as likely to develop the infection as those who have never had PID. From 20% to 25% of patients with one bout of PID suffer subsequent episodes of PID.8,12 Although these subsequent episodes of PID could represent recrudescences of a chronic latent bacterial infection, many appear to represent reinfection. Fortunately, recurrent bouts of PID are not as common when adequate antibiotic therapy is administered. Untreated male sexual patterns remain an important cause of reinfection. It is also likely, although unproved, that many recurrences result because tubal epithelium, previously damaged by infection, is susceptible to bacterial colonization because of ciliary destruction or depressed local defense mechanisms.
Untreated Male Sexual Contacts
Untreated males with urethral N. gonorrhoeae and C. trachomatis infection are an important source of the initial and subsequent episodes of PID. More than 80% of male sexual partners have not been treated for gonorrhea when the female presents with PID.61 About one half of the male contacts with gonorrhea were asymptomatic, and 40% of asymptomatic men had urethral N. gonorrhoeae. The males remained asymptomatic for as long as 180 days after the females were treated. Symptomatic nongonococcal urethritis and asymptomatic chlamydial urethritis are also common among male contacts of patients with PID. Examination and treatment of male contacts are crucial to decrease the recurrence of PID.
Cervical vaginal microorganisms are pushed into the endometrium by surgical procedures, including IUD insertion, dilation and curettage, hysterosalpingography, and induced abortion. Such procedures occurred before the onset of PID in 12% of cases.62 PID has been reported in 1.5% of women within the first 3 months of IUD insertion.63 Induced abortion led to PID in up to 20% of women with untreated C. trachomatis64 and bacterial vaginosis.65 Treatment of these infections before induced abortions reduces the rate of subsequent PID to baseline.66
Human Immunodeficiency Virus Infection
Women with human immunodeficiency virus (HIV) infection may have increased rates of PID. The seroprevalence of HIV was higher in U.S. studies67,68 and one Kenya study69 among those with compared with controls without PID. Patients with PID appear to have more abscesses69,70 and more frequently require surgery71 than PID patients without HIV. Early hospitalization and intravenous treatment is recommended for HIV-positive women with PID.72
Sexually transmitted bacteria, most notably N. gonorrhoeae and C. trachomatis, are recovered from a large number of women with spontaneous PID. A small number of women have genital Mycoplasma recovered from the peritoneum. The rate with which STD microorganisms are isolated from patients with PID differs widely in various studies, probably because of different baseline rates of these STD bacteria between the populations and the severity of infection.
Cervical isolation is not a reliable indicator that it is the only bacteria in the tube or even that it is in the tube. A sizable proportion of patients have N. gonorrhoeae and C. trachomatis cervical infections. A wide variety of bacteria that can be recovered from the fallopian tubes in the presence of all types of cervical infection. Women thought to have gonococcal PID have been found to have fallopian tube infection caused by a combination of N. gonorrhoeae and C. trachomatis, by N. gonorrhoeae and endogenous aerobic and anaerobic bacteria, by endogenous bacteria alone, and by mycoplasmas. Similarly, women without gonorrhea in the cervix can have tubal infection caused by the same variety of microorganisms. Although a large number of possible combinations exist, it is useful to discuss each group separately.
Between 10% and 19% of women with N. gonorrhoeae in the cervix have signs of PID.73–75 Gonococcal virulence factors may explain why only some women develop PID. Transparent colonies of N. gonorrhoeae attach more readily to epithelial cells than opaque-appearing colonies, and transparent but not opaque colony-forming gonococci have been recovered from the abdomen of patients with gonococcal salpingitis.16 Bactericidal antibody develops in women with cervical gonorrhea present for more than 1 month, and specific gonococcal antibody appears to protect against PID.21
A wide range of N. gonorrhoeae recovery is present in PID. Cervical N. gonorrhoeae has been recovered from 80% of some urban American populations with PID73 and from less than 10% in European studies (75–85) (Table 2). In the 17 studies cited in Table 2, the mean N. gonorrhoeae recovery rate was 26%. These studies were selected because C. trachomatis cultures were also performed. Among studies that found high N. gonorrhoeae rates, there tended to be a lower C. trachomatis rate.
*Isolation of N. gonorrhoeae from the peritoneum of the total number of women studied.
N. gonorrhoeae is the most common tubal isolate from women with acute PID and N. gonorrhoeae recovered from the cervix. N. gonorrhoeae is usually recovered from the fallopian tube or abdomen of 20% to 30% of women with the bacteria in the cervix and from 5% to 30% of all women with PID. Because of the exceptional virulence of N. gonorrhoeae, it is easy to assume that gonococci initiate or play a major part in the infection when it is present in the cervix. However, a significant proportion of women with gonococcal PID have no bacteria or bacteria other than gonococci isolated from the fallopian tube. A decrease in the proportion of patients from whom gonococci are recovered is noted when symptoms are present more than 7 days.74 Gonococcal tubal infection could have an initial stage when the gonococci produce the infection and a later stage of infections occurs with other organisms. The decreased rate of recovery in the later stage may be a consequence of gonococcal inhibition by leukocytes, or it may reflect their intracellular location. Women in whom N. gonorrhoeae is the sole bacteria isolated from the tubes tend to respond to therapy more rapidly than do women from whom other bacteria are cultured, suggesting that polymicrobial infection reflects a distinct stage of infection.
The proportion of PID attributed to Chlamydia is higher than that attributed to the gonococcus. C. trachomatis has been isolated from the cervix of 5% to 56% of women with PID. The mean recovery rate of C. trachomatis is 30%, with widely variable rates73–87 (Table 3). All of these studies used culture to identify C. trachomatis, and because DNA detection of Chlamydia by ligase or polymerase chain reaction (PCR) is more sensitive than culture,88 additional women with PID were probably infected with Chlamydia.C. trachomatis was isolated from the fallopian tubes of about one third of all women with the organism in the cervix and from 10% of all women with PID. A complicating finding is that chlamydiae, in contrast to gonococci, can remain in the fallopian tube area for months and years after an initial infection. Women with chlamydial salpingitis may have progressive tubal damage caused by persistent infection if it is not effectively treated.
†Criteria of antibody response not given.
‡Additional women had C. trachomatis in the peritoneum but not in the cervix.
No characteristic gross appearance of the fallopian tubes distinguishes chlamydial from gonococcal infection. C. trachomatis has been recovered in the presence of mild and severe tubal inflammation. However, women with C. trachomatis often have milder clinical manifestations than women with other forms of tubal infection.89 Despite the more benign clinical findings, women with C. trachomatis have tubal damage at least as severe as that of women with N. gonorrhoeae and more severe than that of women with neither. Between 20% and 50% of women with PID have serologic evidence of chlamydial infection that approximates the cervical isolation rate (Table 3), but between 10% and 30% of culture-negative women have antibody evidence of chlamydial infection.
The presence of C. trachomatis does not exclude a concomitant N. gonorrhoeae infection. A sizable proportion of women with C. trachomatis in the cervix also have N. gonorrhoeae isolated. N. gonorrhoeae has been recovered from the same proportion of women with (23%) as without (22%) C. trachomatis. Similarly, C. trachomatis was isolated from the cervix of 33% of women with N. gonorrhoeae and of 31% of women without N. gonorrhoeae (Table 4). Concomitant infection with both organisms is common. These data underscore the rationale of treating women with PID for both organisms.
The importance of C. trachomatis is further emphasized in the study of infertile women and ectopic pregnancy. Although N. gonorrhoeae tends to produce mucosal infection, C. trachomatis and other aerobic and anaerobic bacteria may produce infection below the basement membrane, resulting in permanent tubal damage. Chlamydia antibody was present in 35% to 90% of infertile women with tubal abnormalities detected by hysterosalpingography or direct tubal visualization or both, compared with 0% to 50% of infertile women with no tubal abnormality90–98 (Table 5). These data suggest that a significant amount of tubal infertility is caused by chlamydial salpingitis. The relative importance of N. gonorrhoeae in these women usually has not been simultaneously determined, but in one report infertility rates after chlamydial salpingitis (23%), gonococcal salpingitis (24%), and nonchlamydial nongonococcal salpingitis (22%) were similar.99 Higher Chlamydia antibody rates have also been detected among patients with ectopic pregnancy than controls.100C. trachomatis has been isolated from the tubes of some patients with an ectopic pregnancy, suggesting that an ongoing infection may further contribute to tubal pregnancy.
H, hysterosalpingography; L, Laparoscopy; H/L, laparoscopy or hysterosalpingography; L/L, laparoscopy or laparotomy.
*Calculated p value using Yates correction of χ2.
Culture, serologic, and experimental lines of evidence suggest some role for genital Mycoplasma. Patients with PID frequently have M. hominis (55% to 70%) isolated from the cervix. However, M. hominis was recovered from the tubal area in only 4% to 17% of women with PID (Table 6).
Serologic evidence of acute mycoplasmal infection is more common than recovery of the organism from the tubal area. Increases in IgG antibody to M. hominis have been identified in 10% to 30% of patients with PID; most of these women also had significant serum elevations of M. hominis IgM antibody, which is further evidence of acute infection. The fact that more patients have serologic evidence than fallopian tube culture evidence of acute mycoplasmal infection may be attributable to the propensity of mycoplasmas to cause parametritis rather than salpingitis, as discussed in the Pathogenesis and Pathology section. Mycoplasma genitalium has produced tubal infection in monkeys.103M. genitalium is present by PCR in up to 25% of women attending STD clinics104 and serum antibody to M. genitalium is present in one third of women with acute PID, but no gonorrhea or chlamydia.105
U. urealyticum is commonly isolated from the cervix and can be isolated from the tubes. Although U. urealyticum serum antibody titers changed significantly in up to 18% of PID cases, its isolation into primate and human tubes fails to produce evidence of infection. There appears to be a minimal or no role for U. urealyticum in PID.
Aerobic and Anaerobic Bacteria
Aerobic and anaerobic bacteria are isolated from the cul-de-sac and tube of up to 60% of women with acute salpingitis.86 The frequency of isolation from direct tubal cultures obtained through the laparoscope has been lower in Sweden. Although the frequency of these bacteria may be overestimated as a result of contamination of the culdocentesis samples by normal vaginal flora, aerobic and anaerobic bacteria are clearly important pathogens in acute salpingitis. The disparity between the findings of investigators in the United States and Sweden may be partially explained by population differences; women with PID in U.S. studies tend to report for care later in the course of infection than do Swedish women. Finding other aerobic and anaerobic bacteria among women with gonococcal PID is probably attributable to the ability of nongonococcal bacteria to invade areas of fallopian tubes with inflammation and reduced redox potential after an initial gonococcal infection. An analogous situation probably occurs after primary C. trachomatis infection. Aerobes and anaerobes are often primary pathogens among women with instrumentations, and IUDs, and perhaps those with recurrent salpingitis and bacterial vaginosis.
The most common aerobic isolates are streptococci, Escherichia coli, and Haemophilus influenzae. The most common anaerobic organisms consist of peptostreptococci and Bacteroides species. Anaerobic bacteria, alone or more commonly as part of a polymicrobial infection, are present in 15% to 60% of women with PID. Patients with the most severe clinical findings, such as peritonitis, often have aerobic and anaerobic bacterial infections. Anaerobic bacteria are virtually always present in intraabdominal abscesses caused by PID. The important role of anaerobes is further illustrated by the finding that patients with anaerobic pelvic infection respond poorly to therapy and tend to develop sequelae. Fallopian tube occlusion was more frequent among patients with C. trachomatis or anaerobic salpingitis than those with N. gonorrhoeae salpingitis.106
Abdominal pain is the most common symptom of PID, although the pain may be mild or, in at least 5% of patients with laparoscopically verified PID, absent.107 It has been well documented that 30% to 75% of women with occluded tubes have never had recognized PID or abdominal pain severe enough to receive antibiotic therapy.92 Physicians must be aware that mild or atypical symptoms may represent PID. When pain is present, it has usually been present for less than 10 days.89 About 75% of patients with gonococcal or chlamydial PID complain of vaginal discharge of recent onset. Intermenstrual or excessive menstrual bleeding, probably because of endometritis, occurs in 40% of patients.107 Dysuria or urinary frequency occurs in 15% of women. Fever is present in only about 40% of patients,107 most commonly in those with severe clinical disease. Nausea and vomiting occur later in the course of PID than in appendicitis and are assumed to result from bowel inflammation. In patients harboring gonococci or chlamydiae, the onset of pain tends to occur with menses. In general, gonococcal PID usually has a rapid onset; the pain is usually of short duration, the temperature is often higher than 100.4°F (38°C), and patients have a cervical exudate. Gonococcal PID usually is not associated with a history of previous PID or with IUD use. In contrast, chlamydial PID is characterized by a significantly longer duration of usually less severe pain, a higher sedimentation rate, and less fever than gonococcal PID.89 Women with chlamydial PID more often have undergone recent genital tract instrumentation than patients with gonococcal PID.
A syndrome of perihepatitis associated with gonococcal salpingitis was described in the 1920s. Chlamydial salpingitis also causes perihepatitis. Perihepatitis is usually diagnosed when severe, pleuritic, most frequently right, upper quadrant pain is present. The onset of pleuritic pain is most likely to occur after or simultaneously with the onset of lower abdominal pain, although it can occur occasionally without any recognized lower abdominal pain. Symptomatic perihepatitis occurs in 5% to 10% of women with salpingitis, but it is frequently not considered in the differential diagnosis of right upper quadrant pain, which can be misdiagnosed as cholecystitis or pneumonia. To limit diagnostic errors, a careful pelvic examination should be performed in any sexually active woman who presents with pleuritic or right upper quadrant abdominal pain. The pleuritic pain is caused in the early stage by inflammation of the liver capsule. During the later stage, “violin string” adhesions form between the liver and the anterior abdominal wall.
The most widely accepted mechanism of infection has been the spread of intraabdominal bacteria from the fallopian tubes to the liver surface. However, clinical and experimental evidence suggests that lymphatic or hematogenous bacterial spread from the pelvis to the liver may be a more likely route than direct intraabdominal extension. An immune component is also possible, with increased levels of CHSP-60 found in cases of perihepatitis.108
The physical findings consist of moderate to severe liver tenderness without hepatomegaly. A transient friction rub is occasionally present. Signs of salpingitis are usually demonstrable by a pelvic examination. Results of serum bilirubin and liver function tests are normal or only mildly elevated. The gallbladder may not be visualized with an oral cholecystogram during the acute phase. Laparoscopy provides the most precise diagnosis; a purulent or fibrinous exudate over the liver capsule in visualized early in the infection, followed by adhesions later in the course. The pain usually resolves rapidly after antibiotic therapy is administered.
Salpingitis has a broad spectrum of clinical manifestations, varying from mild to severe. Insistence on rigid diagnostic criteria such as fever, abdominal tenderness, leukocytosis, and an elevated sedimentation rate can uncover only the most overt cases of PID while excluding most cases. In patients with severe, clinically overt PID, it is possible to establish the diagnosis with reasonable certainty by a combination of history, physical examination, Gram stain of cervical secretions, and examination of the male sexual partner. Laparoscopy is of limited benefit for the 20% of women with clinically severe salpingitis.
Several studies have addressed the difficulties of accurately establishing a clinical diagnosis of PID (107–112) (Table 7). Among patients with a clinical diagnosis of acute PID, salpingitis could be confirmed by laparoscopy in only about 60%, whereas 18% of patients had other diseases, most commonly benign bleeding ovarian cysts, ectopic pregnancy, appendicitis, and endometriosis. Of the women clinically diagnosed as having PID, 4% have appendicitis, and 3% have an ectopic pregnancy (Table 7). About 20% of patients had no pathologic pelvic findings visualized by laparoscopy.
Up to 65% of salpingitis is not diagnosed or treated,92 and the aim should be to diagnose all cases of PID by increasing the sensitivity while sacrificing specificity. To increase the number of patients treated for PID, those with mild or atypical signs and symptoms need to be detected. Patients with mucopurulent cervicitis and mild pelvic pain or tenderness and patients with abnormal bleeding with menses, mild pelvic pain, and uterine tenderness (representing endometritis) need to be treated for PID. However, the already low 60% specificity of PID is even lower in this group with more mild clinical findings, but the advantage of this approach is treatment of a larger number of patients for PID, which may lower the rate of sequelae.
Although it is not practical to perform laparoscopy on every woman suspected to have salpingitis, laparoscopy should be used in certain situations. Laparoscopy remains the most accurate method to diagnose salpingitis, and it is relatively safe. It should be used for women with severe abdominal rigidity to exclude appendicitis, intraperitoneal bleeding, or a ruptured abscess. Laparoscopy is cost effective for women with persistent pain after one or more courses of antibiotics for presumptive PID. These women rarely have active infection and have adhesions from prior salpingitis, other diseases (most commonly endometriosis), or normal pelvic structures. Laparoscopy is also useful among women with an unclear cause of mild abdominal pain who would otherwise not be treated with antibiotics because of a clinical diagnosis other than salpingitis.
Clinical symptoms do not accurately predict salpingitis observed at laparoscopy. Lower abdominal pain is the most frequent symptom among patients with acute salpingitis visualized at laparoscopy, although the pain may be mild or even absent. The abdominal pain is usually continuous, bilateral, and most severe in the lower quadrants. It frequently increases with movement, a Valsalva maneuver, and intercourse. Intermittent sharp pain that lasts for a few seconds or minutes is unlikely to be a manifestation of PID. Pain resulting from salpingitis had been present for less than 15 days in 85% of patients with visually confirmed salpingitis.113 Chronic pain, particularly intermittent pain of more than 3 weeks' duration, is unlikely to be caused by PID. Women with chronic pain should usually undergo laparoscopy rather than treatment with antibiotics.
The presence or absence of other symptoms is of limited use in establishing the diagnosis of PID. Women with visually confirmed salpingitis had the same rate of increased vaginal discharge, irregular vaginal bleeding, urinary symptoms, and vomiting as women without salpingitis.107 A history of fever or chills may be helpful to distinguish patients with salpingitis from patients without infection, but most women do not have these symptoms. Salpingitis is particularly common among women with a recent episode of PID and among sexual contacts of men with gonococcal or chlamydial urethritis.
Patients with salpingitis commonly have tenderness of the lower abdomen, cervix, uterus, and both adnexa. Cervical motion tenderness is particularly common in PID. Unilateral salpingitis occurs in fewer than 10% of women with PID, and most women with unilateral adnexal tenderness on phy-sical examination were found to have bilateral sal-pingitis on laparoscopy.107 When unilateral adnexal tenderness is present, other diagnoses should be considered. Few physical findings are specific for salpingitis, and patients with other disease or no disease may have physical findings similar to those of patients with PID. Grossly abnormal cervical discharge is found is less than one half of these patients, although cervical and vaginal leukocytosis is common.114 A temperature over 100.4°F (38°C) is detected in only about one third of patients with visually confirmed salpingitis.107
An almost universal belief is that women with the greatest abdominal and adnexal tenderness have the largest amount of tubal damage. However, clinical signs do not predict the severity of tubal disease observed at laparoscopy.115 Women with mild clinical manifestations require the same treatment as women with more severe findings.
Laboratory tests are not often helpful for establishing the diagnosis of salpingitis, because false-negative values are common. The hematocrit, white blood cell count, and sedimentation rate are of benefit only when abnormal. For example, only 45% of patients with laparoscopically confirmed salpingitis had a white blood cell count of 7900 or greater, and only 75% had a sedimentation rate greater than 15 mm/hour.107 An elevated C-reactive protein value may be more predictive of PID than these traditional tests.
A Gram stain of cervical secretions is one of the most helpful laboratory procedures to the diagnosis of cervicitis. The findings of ten or more polymorphonuclear leukocytes (PMNs) establishes a diagnosis of cervicitis.21 Most women with PID have an increased number of PMNs in the Gram-stained cervical or vaginal discharge. To exclude a possible ectopic pregnancy, serum pregnancy tests should be performed for any woman who may be pregnant.
Endometrial Biopsy and Culdocentesis
Endometrial biopsy has been used to study the pathogenesis of PID.32,33 However, the presence of plasma cells within endometrium,32 indicative of endometritis, may also be used to diagnose PID. In our experience, 80% of women with laparoscopically verified PID have plasma cell endometritis. About 20% of patients with plasma cell endometritis do not have visible salpingitis. Finding plasma cells in the endometrium can help to increase the number of patients identified with PID. Aspiration of purulent culdocentesis fluid can be useful to establish the diagnosis of salpingitis in those with signs of peritonitis. Culdocentesis is of benefit when the abdominal fluid contains white blood cells or nonclotting blood. Appendicitis or other causes of peritonitis that could also be associated with a purulent exudate should be excluded. Abdominal exploration is necessary when nonclotting blood is present to diagnose ectopic pregnancy or a bleeding ruptured ovarian cyst. Culdocentesis is of no benefit for diagnosis when no fluid or fluid devoid of while blood cells is obtained.
Examination of Male Sexual Partners
An opportunity to help establish a diagnosis is provided by examining the male sexual partners of patients with suspected PID. Even male partners without an overt discharge should have urethral material obtained. The sample should be Gram stained and cultured for N. gonorrhoeae and C. trachomatis. Finding gonococci within the urethral by Gram stain is strong evidence that a symptomatic female sexual contact has salpingitis. Chlamydial PID should be suspected when male sexual contacts have a urethral discharge or a large number of white blood cells in the absence of gonococci.
Ultrasound provides a noninvasive test to help diagnose PID and follow the course of severe PID.117 An adnexal mass that represents an abscess appears as a cystic mass containing variable amounts of debris. An abscess can often be differentiated from an inflammatory pelvic mass, which consists of swollen tubes, ovary, and bowel. However, the ultrasonographic appearance of an inflammatory mass is extremely variable, and abscesses may be cystic or solid or have a fluid level. Ultrasonography should be performed to determine whether a large adnexal mass is an inflammatory mass or an abscess. Ultrasonography has been used to guide needles to drain abscesses when they are close to the vaginal or abdominal surface.118 However, ultrasound is less useful to diagnose PID when a mass is not present. Ultrasound correctly diagnosed PID confirmed by laparoscopy in 94% of patients with severe PID but only 80% with moderate and 64% with mild PID.117
A cervical culture for N. gonorrhoeae is mandatory. C. trachomatis should also be detected by culture or DNA testing. However, with rare exception, there is no evidence that the recovery of other bacteria from the cervix is necessary in diagnosing PID.
Culdocentesis fluid collected in a syringe should have the air expelled and be injected into an anaerobic transport system or immediately placed onto an anaerobic culture medium. Solid Thayer-Martin medium or chocolate agar should be used for the recovery of N. gonorrhoeae, because liquid medium usually fails to support growth of the organisms.
During laparoscopy or laparotomy, purulent material can be obtained from the pelvic cavity. Culture material also can be obtained by cannulating the tube with a small catheter, bronchoscopy brush, or a calcium alginate swab. If an abscess is present, pus can be aspirated with a syringe and needle. If the abscess is removed, a section of the abscess wall should be sent for culture. N. gonorrhoeae, anaerobic bacteria, and C. trachomatis are more likely to be recovered from injected tissue or epithelial cells brushed from the tubes than from pus.
Adequate treatment of salpingitis includes an assessment of the severity of the infection, administration of antibiotics to eradicate a wide variety of possible organisms, use of other health measures, close patient follow-up, and treatment of male sexual partners. Treatment within 3 days of the onset of symptoms reduces infertility.11
Because of costs, most patients with salpingitis have not been hospitalized in the United States, although hospitalization of all patients has been advocated and is standard practice in many European countries. A high failure rate among women who receive oral antibiotics on an outpatient basis has been reported,119 and hospitalization should be used for patients judged likely to be noncompliant. Patients with the following features should be hospitalized: severe peritonitis (severe nausea or a temperature 38°C or higher), suspected pelvic abscess, pregnancy, failure to respond to outpatient antibiotics, an unclear diagnosis of salpingitis, and perhaps HIV infection.
Antibiotic therapy should ideally be directed toward the organism recovered from the fallopian tube, although fallopian tube cultures are not easy to obtain. Antibiotics must be administered empirically before identification of the organisms in most cases. There has understandably been much interest in the appropriate antibiotic treatment of women with PID and probably an unrealistic hope that antibiotic therapy can markedly reduce sequelae. This hope is based on the belief that the newer broad-spectrum regimens that are effective against chlamydiae, anaerobes, and gonorrhea can improve the outcome. However, data from Sweden indicate identical infertility rates among women who received antibiotics with widely different antimicrobial spectrums.10 The data suggest that most permanent tubal damage has occurred before antibiotic therapy is begun. This possibility is reinforced by the finding that the most predictive parameter of subsequent infertility was the degree of tubal damage observed laparoscopically before therapy.7,10 The cornerstone of treatment should be prevention of salpingitis by the diagnosis and treatment of cervicitis and endometritis. Physicians tend to use two regiments of intravenous antibiotics to treat women with severe clinical signs but a single oral antibiotic to treat women with mild manifestations. Increasing evidence suggests that women with mild manifestations are just as likely as women with severe clinical manifestations to have severe tubal damage.15
Antibiotic regimens recommended by the Centers for Disease Control are designed to be effective against N. gonorrhoeae, C. trachomatis, and the most commonly isolated aerobic and anaerobic bacteria.72 Each of these regimens has its strong and weak points (Table 8). Alternative parenteral regimens include ofloxacin plus metronidazole, ampicillin/sulbactam plus doxycycline or ciprofloxacin, phosdoxycycline, and metronidazole. Two reviews were published about the treatment of PID.120,121 The clinical response to the administration of each of the various antibiotic regimens was usually favorable, and no particular regimen has been shown to provide a superior clinical response over another. Subsequent fertility would be expected to provide a more sensitive index of therapeutic success than the initial clinical response, but subsequent fertility among women receiving several different antibiotics has been similar10 or has not been reported.122
Compiled from 1998 Guidelines for treatment of sexually transmitted disease. MMWR Morb Mortal Wkly Rep 47:79–86, 1998.
Each antibiotic regimen should be selected for the situation in which it is most potentially useful. For example, the cefoxitin or cefotetan-doxycycline combination is an effective overall regimen that can inhibit N. gonorrhoeae (including penicillinase-producing strains), C. trachomatis, most aerobic, and more than 90% of (but not all) anaerobic bacteria. Clindamycin and an aminoglycoside can inhibit anaerobes and most aerobic bacteria, but clindamycin is less active than tetracycline against Chlamydia and high intravenous clindamycin doses (2400 mg daily) are required to inhibit Chlamydia. This regimen, however, is well tested for the treatment of abscesses. Cephalosporins should not be given as a single agent, because although patients usually respond clinically, C. trachomatis persists despite therapy,123 and there is concern that subclinical persistent Chlamydia may continue to increase the tubal damage.
Women with an IUD should have it removed, usually after some interval of antibiotic therapy. All patients should be reexamined at 2 to 3, 7, and 21 days after the beginning of therapy to document a satisfactory clinical response and to obtain repeat cervical cultures. Patients with a slow response should be examined more frequently. All recent male sexual contacts of women with acute salpingitis, including those who are asymptomatic, should be examined, have samples cultured, and be treated, if necessary.
With the use of antibiotics that inhibit anaerobic bacteria, particularly Bacteroides fragilis, hysterectomy during the acute phase of PID has decreased. However, immediate surgery is necessary to prevent septic shock in patients with a ruptured pelvic abscess. Early operative transvaginal drainage is the preferred method of treating a fluctuant pelvic abscess that becomes attached to the vagina in the midline of the cul-de-sac. More difficulty arises in the treatment of an inflammatory mass that is not attached to the vagina but requires transperitoneal drainage. Because even large masses do not always represent abscess formation, the current recommendation is that antibiotic therapy aimed particularly at anaerobes should be instituted and continued if the mass size declines. Surgery can then be delayed unless a definite abscess is identified by the clinical course (i.e. the patient fails to improve clinically or the mass remains unchanged in size) or if an abscess greater than 6 cm in diameter can be demonstrated by scans. With this approach, few patients require transabdominal drainage.
Open laparoscopy has been used to guide percutaneous placement of catheters into abscesses.124 The catheter is used to drain pus and irrigate the abscess cavity. The catheter is left to drain for 24 to 48 hours or until drainage decreases and the patient's condition improves. Care must be taken to avoid bowel injury and to identify multiloculated abscesses. About 90% of large abscesses can be cured with percutaneous drainage,118 which has the advantage of obviating extensive surgery. If abdominal exploration is necessary, the least extensive amount of surgery likely to be effective should be performed, particularly among women who wish to remain fertile. For women with a unilateral adnexal abscess or an abscess within loops of bowel, total abdominal hysterectomy and bilateral salpingooophorectomy represent surgical overkill. Simple drainage or even removal of a unilateral abscess without removal of the opposite tube and ovary and the uterus is usually successful when a patient is receiving an adequate antibiotic regimen, which includes antibiotics that inhibits B. fragilis. This conservative approach is mandatory for women who still desire children. Advances in microsurgery and in vitro fertilization continue, and destructive surgery for acute infection is passé.
Despite prompt diagnosis and treatment, sequelae are common. A limited number of patients with salpingitis suffer persistent abdominal pain without evidence of an abscess or adnexal pathology except for mild tenderness. If the pain continues for several weeks, laparoscopy should be performed to exclude other pelvic pathology or pelvic adhesions, which can then be lysed. In many instances, there is no discernible pelvic pathology. Laparoscopy should be performed before repeated antibiotic courses are given to patients with persistent pelvic pain and no objective evidence of salpingitis.
16. Weström L, Svensson L. Chronic pain after acute pelvic inflammatory disease. In Belfort P et al (eds): Advances in Gynecology and Obstetrics. Proceedings of the XIIth World Congress of Gynecology and Obstetrics, Rio de Janeiro 6: 265, 1988
23. Draper DL, James JF, Brooks GF et al: Comparison of virulence markers of peritoneal and fallopian tube isolates with endocervical N. gonorrhoeae isolates from women with acute salpingitis. Infect Immun 27: 882, 1980
27. Buchanan TM, Eschenbach DA, Knapp JS et al: Gonococcal salpingitis is less likely to recur with Neisseria gonorrhoeae of the same principal outer membrane protein antigenic type. Am J Obstet Gynecol 138: 978, 1980
32. Kiviat NB, Wolner-Hanssen P, Eschenbach DA et al: Endometrial histopathology in patients with culture-proved upper genital tract infections and laparoscopically diagnosed acute salpingitis. Am J Surg Pathol 14: 167, 1990
34. McGee ZA, Melly MA, Gregg CR et al: Virulence factors of gonococci: Studies using human fallopian tube organ culture. In Brooks GF, Gotschlich EL, Holmes KK et al (eds): Immunobiology of N. gonorrhoeae, p 258. Washington, DC: American Society for Microbiology, 1978
36. Van Voorhis WC, Barrett LK, Cosgrove Sweeney C et al: Analysis of lymphocyte phenotype and cytokine activity in the inflammatory infiltrates of the upper genital tract of female macaques infected with Chlamydia trachomatis. J Infect Dis 174: 647, 1996
39. Patton DL et al: The effects of Chlamydia trachomatis on the female reproductive tract of Macaca nemestrina after a single challenge following repeated cervical inoculations. Obstet Gynecol 76: 1271, 1990
45. Grayston JT et al: Trachoma vaccine studies in volunteer students of the National Defense Medical Care Center II. Response to challenge eye inoculation of egg grown trachoma virus. Clin Med J 8: 312, 1961
49. Henry-Suchet J, Catalan F, Leffredo V et al: Microbiology of specimens obtained by laparoscopy from controls and from patients with pelvic inflammatory disease or infertility with tubal obstruction: Chlamydia trachomatis and Ureaplasma urealyticum. Am J Obstet Gynecol 138: 1022, 1980
66. Larson PG, Platz-Chrostensen JJ, Thejla H et al: Incidence of pelvic inflammatory disease after first trimester legal abortion in women with bacterial vaginosis after treatment with metronidazole: a double-blind, randomized study. Am J Obstet Gynecol 166: 100, 1992
70. Irwin KI: Potential for bias in studies of the influence of human immunodeficiency virus infection on the recognition, incidence, clinical course, and microbiology of pelvic inflammatory disease. Obstet Gynecol 84: 463, 1994
75. Moller BR, Mardh P-A, Akrons S et al: Infection with Chlamydia trachomatis, Mycoplasma hominis, and Neisseria gonorrhoeae in patients with acute pelvic inflammatory disease. Am Vener Dis Assoc 8: 198, 1981
87. Skaug K, Vik ISS, Wvigstad E et al: Chlamydial serum IgG antibodies in patients with acute salpingitis measured by an enzyme-linked immunoabsorbent assay. Acta Pathol Microbiol Immunol Scand 90: 67, 1982
88. Pasternack P, Vuorinen P, Kuukankorpi A, et al: Determination of Chlamydia trachomatis infections in women by Amplicor PCR: comparisons of diagnostic performance with urine and cervical specimens. J Clin Microbiol 34: 995, 1996
91. Cevenini R, Possati G, LaPlacva M: Chlamydia trachomatis infection in infertile women. In Mardh P-A, Holmes KK, Oriel JD et al (eds): Chlamydial Infection, pp 189–192. New York: Elsevier Biomed Press, 1982
94. Gump DW, Gibson M, Ashikaga T: Evidence of prior pelvic inflammatory disease and its relationship to C. trachomatis antibody and intrauterine contraceptive device use in infertile women. Am J Obstet Gynecol 146: 143, 1983
113. Westrom L, Mardh P-A: Epidemiology, etiology and prognosis of acute salpingitis: a study of 1457 laparoscopically verified cases. In Hobson D, Holmes KK (eds): Nongonococcal Urethritis and Related Diseases. Washington, DC: American Society for Microbiology, 1977
119. Thompson SE, Brooks C, Eschenbach DA et al: High failure rates in outpatient treatment of salpingitis with either tetracycline alone or penicillin/ampicillin in combination. Am J Obstet Gynecol 152: 635, 1985