REVIEWS OF INFECTIOUS DISEASES • VOL. 12, SUPPLEMENT 6 • JULY-AUGUST 1990
Pelvic Inflammatory Disease: Review of Treatment Options Herbert B. Peterson, Edward I. Galaid, and Jonathan M. Zenilman
From the U. S. Department of Health and Human Services, Public Health Services, Division of Reproductive Health; and the Division of STDIHIV Prevention, Centers for Disease Control, Atlanta, Georgia
Pelvic inflammatory disease (PID) is an oftenserious consequence of gonococcal and chlamydial infection. An estimated 1 million women in the United States are treated each year for this condition, and at least one-fourth of these women will suffer one or more serious sequelae, including infertility, ectopic pregnancy, and major abdominal/pelvic surgery. It has been estimated that the cost of these sequelae will be $3.5 billion per year [1] by 1990.Since the early 1970s, the Centers for Disease Control (CDC) has recommended therapeutic regimens for PID as part of the treatment guidelines for sexually transmitted diseases (STDs). In this report we will review the current understanding of PID as a clinical syndrome, the 1985 CDC recommendations and their rationale, and the advances that have been made in the understanding of the diagnosis and treatment of PID since 1985.We will also make further recommendations for PID treatment regimens. Clinical Background PID is an entity that has escaped consistent definition. In the 1985 STD treatment guidelines published by the CDC [2], PID was defined as "the acute clinical syndrome (unrelated to pregnancy or surgery) attributed to the ascent of microorganisms from the Please address requests for reprints to Technical Information Services (E06), Center for Prevention Services, Centers for Disease Control, Atlanta, Georgia 30333.
S656
vagina and endocervix to the endometrium, fallopian tubes, and/or contiguous structures." Important gaps in our understanding of the pathophysiology of PID still exist. PID is a syndrome that may include endometritis, salpingitis, oophoritis, tubo-ovarian abscess, and pelvic peritonitis. Infection of the fallopian tubes is the hallmark of PID. In some women, infection of the fallopian tubes and contiguous structures leads to the formation of tubo-ovarian abscesses. In contrast to what is known or suspected regarding ab-scesses elsewhere within the body, tubo-ovarian abscesses often respond to antibiotic therapy without surgical intervention. The exact mechanism by which microorganisms infect the upper genital tract is unknown. One unifying hypothesis regarding the pathogenesis of acute PID suggests that it usually starts as a cervical infection with Neisseria gonorrhoeae and/or Chlamydia trachomatis, with subsequent alteration in the cervicovaginal microenvironment as a result of the consumption of nutrients and the production of metabolic waste products. These changes cause others, including alterations in vaginal pH and the availability of oxygen, which result in overgrowth of endogenous and anaerobic flora. The original cervical pathogens and/or endogenous organisms then ascend into the upper genital tract [3]. This hypothesis would explain the polymicrobial nature of PID. Alternative theories assume that polymicrobial PID can be initiated without N. gonorrhoeae [4] or
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
Decisions regarding appropriate antibiotic therapy for treatment of pelvic inflammatory disease (PID) are complicated by an incomplete understanding of the syndrome. Further, the lack of data on laparoscopic diagnosis and cure of PID severely limits our ability to interpret data on therapy outcome. Validation studies of the treatment regimens recommended by the Centers for Disease Control in the 1985treatment guidelines for sexually transmitted diseases (STDs) suggest that both the cefoxitin/doxycycline and clindamycin/aminoglycoside combination regimens result in consistently high rates of clinical evidence of cure. The 1989 STD treatment guidelines were based on these studies and on available data regarding newer treatment regimens. Empiric, broad-spectrum therapy remains the treatment of choice. The two regimens recommended for inpatient therapy in 1989 are similar to those recommended in 1985. The recommendation for management of ambulatory patients has been substantially changed, however, because of increasing resistance of Neisseria gonorrhoeae to penicillin.
PID Treatment Options
rent STD treatment guidelines [12-14]. The importance of optimal therapy was addressed indirectly by studies of laparoscopically verified PID in Sweden; these investigations suggested that women with an early diagnosis and effective treatment have fewer long-term sequelae [5]. Decisions regarding optimal antibiotic therapy for PID are further complicated by an incomplete understanding of the relationship between in vitro observations and clinical efficacy. For example, an aminoglycoside may be highly effective in the laboratory but less effective clinically, perhaps because of conditions associated with abscess cavities or anaerobic environments. Paradoxically, treatment with antibiotics that are theoretically inadequate has sometimes resulted in clinical evidence of cure of PID. However, studies reporting such favorable clinical outcomes do not account for either the natural history of the infection, which may resolve spontaneously, or the relationship between clinical diagnosis and true resolution of pathology in the upper genital tract. Only recently has C trachomatis been appreciated as a significant sexually transmitted pathogen. The important role that this organism plays in the etiology of PID complicates therapeutic considerations. Because C trachomatis infection (of both the cervix and the upper genital tract) is often asymptomatic, diagnosis is difficult, especially when laparoscopy is not performed. C trachomatis often causes substantial tissue destruction, even in asymptomatic cases, and such destruction is a major factor in subsequent infertility [15, 16]. Clinical evidence of improvement, therefore, may not correlate well with resolution of primary pathology. For example, resolution of symptoms may not be an accurate reflection of cessation of tubal inflammation. It is also difficult to assess whether drugs with in vitro activity against C trachomatis are clinically effective, including whether they prevent serious tissue destruction. Because symptoms may improve during antibiotic treatment despite continuing mucosal and soft tissue infection with C trachomatis, clinical cure may be a poor measure of outcome. Thus, any decision regarding treatment regimens must include a consideration of the serious sequelae of C trachomatis infections and of what little is known about the relationship between in vitro and in vivo antibiotic efficacy in these infections. Data regarding treatment of PID are based primarily on hospitalized patients. Few data exist on
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
C trachomatis. Organisms isolated from women with polymicrobial PID have generally fallen into four broad categories: N. gonorrhoeae alone, C trachomatis alone, N. gonorrhoeae and C trachomatis together, and anaerobic and/or facultative bacteria [4-6]. From the standpoint of diagnosis, the presence of a pathogen in the endocervix may not always indicate that this organism is the agent causing infection of the endometrium, the fallopian tubes, or the peritoneal cavity. Further, infectious agents can be recovered from endometrial cultures even when cervical cultures for the same agents are negative [7]. In their classic studies, Jacobson and Westrom [8] found that a clinical diagnosis of PID was erroneous in one-third of cases when laparoscopy was used as the diagnostic gold standard. At least 11 other studies have reported that laparoscopy is required for establishing the diagnosis of PID with a high degree of specificity [6, 9, 10]. Hadgu et al. [11] developed a mathematical model that correctly predicted 87070 of cases of acute PID jn a series of laparoscopically confirmed diagnoses in Lund, Sweden. Variables that were good predictors in that model were purulent vaginal discharge, an erythrocyte sedimentation rate of ~ 15 mm/h, a culture positive for N. gonorrhoeae, adnexal swelling on bimanual examination, and a rectal temperature of ~ 38°C. Unfortunately, few studies evaluating treatment have included any laparoscopically verified cases. Studies of therapeutic efficacy have used clinical parameters for diagnosis that cannot be evaluated relative to the model described above or relative to any other useful standard for establishing commonality of diagnoses among the various studies. The lack of data on laparoscopic diagnosis and cure of PID severely limits our ability to interpret data on the outcome of therapy. Few studies include laparoscopic verification 0'[ cure, and most studies fail to report the basis on which cure was determined. In almost all cases, clinical cure was determined without laparoscopic or hysterosalpingographic follow-up. Available data on antibiotic efficacy do not address the prevention of tubal destruction or the related serious (and often life-threatening) long-term sequelae of PID. To date, no studies in the United States have compared different treatment regimens with fertility as an outcome measure. The limited available data on efficacy focus on antibiotic regimens not recommended in either the 1985 or the cur-
5657
Peterson, Galaid, and Zeni/man
5658
which to base guidelines for outpatient treatment. Clearly, the polymicrobial nature of PID necessitates broad-spectrum antibiotic coverage, even if hospitalization is not required. Clinical Validation of the 1985 Treatment Guidelines Inpatient Regimens
1989
1985
Inpatients" Regimen A Cefoxitin (2 g iv every 6 h) Cefoxitin (2 g iv every 6 h) PLUS OR Doxycycline (100 mg iv Cefotetan t (2 g iv every 12 h) every 12 h) PLUS Doxycycline (100 mg po* or iv every 12 h) After discharge: Doxycycline (100 mg po twice daily to complete a total of 10-14 d of therapy)
After discharge: Doxycycline (100 mg po twice daily to complete a total of 10-14 d of therapy)
Regimen B Clindamycin (600 mg iv Clindamycin (900 mg iv every 8 h) every 6 h) PLUS PLUS Gentamicin (2.0 mg/kg as iv Gentamicin (2.0 mg/kg as iv loading dose followed by loading dose followed by 1.5 mg/kg iv every 8 h in 1.5 mg/kg iv every 8 h in patients with normal renal patients with normal renal function) function) After discharge: Clindamycin (450 mg po 4 times daily to complete a total of 10-14 d of therapy)
Cefoxitin] (2 g im)
After discharge: Doxycycline (100 mg po twice daily to complete a total of 10-14 d of therapy)§
Outpatients Cefoxitin] (2 g im)
OR
OR
Amoxicillin (3.0 g po)
Ceftriaxone (250 mg im) PLUS Doxycycline (100 mg po twice daily for 10-14 d)
OR Ampicillin (3.5 gpo)
OR Aqueous procaine penicillin G (4.8 mU im)
OR Ceftriaxone (250 mg im) PLUS Doxycycline (100 mg po twice daily for 10-14 d)
OR Tetracycline (500 mg po 4 times daily) • All regimens are given for at least 48 hours after the patient improves clinically. t Other cephalosporins (such as ceftizoxime, cefotaxime, and ceftriaxone), which provide adequate coverage against gonococci and other facultative gram-negative aerobic and anaerobic bacteria, may be utilized in appropriate doses. See text and table 2 for further discussion. po = per os. § Continuation of clindamycin (450 mg po, 4 times daily, for a total of 10-14 d) may be considered as an alternative. See comments regarding oral clindamycin in text. II Each of these regimens except ceftriaxone is accompanied by probenecid (1 gpo).
*
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
The antibiotic regimens recommended in the 1985 guidelines (table 1)were empirically designed to provide broad coverage for mixed, aerobic, and anaerobic infections. In regimen A (cefoxitin/doxycycline), cefoxitin was chosen for coverage of N. gonorrhoeae, the Enterobacteriaceae, and anaerobes. The inclusion of doxycycline, which also gives some broad-spectrum coverage, was based on recognition of the role of C trachomatis in PID. Neither regimen A nor regimen B provided coverage for enterococci. In regimen B (clindamycin/gentamicin), clindamycin was chosen for coverage of anaerobes and grampositive aerobes. The efficacy of clindamycinin treating these organisms (especially Bacteroides), even when they occur in a tubo-ovarian abscess, is considered a particular advantage in clinically advanced PID. Although clindamycin has only moderate in vitro activity against C trachomatis and although the ability of oral clindamycin to eradicate C trachomatis from the cervix or urethra has been variable [17, 18], treatment with high-dose intravenous clindamycin has resulted in clinical evidence of cure and eradication of C trachomatis from the endometrium in a relatively small series of cases [19-27]. In one report, women with severe salpingitis at laparoscopy responded slowly to therapy; the three women (of 19 treated) in whom treatment failed clinically all were initially infected with C trachomatis. The authors of that report speculated that the eradication of C trachomatiswas likely attributable to the high doses of intravenous and oral clindamycin administered [19]. Aminoglycosides cover gram-negative aerobes, including N. gonorrhoeae; when administered in combination with clindamycin, they are highly effective in treating abscesses. Brunham has reviewed studies of antimicrobial treatment for PID reported from 1974 to 1984 [28]. Since the 1985 guidelines were issued, studies of the two inpatient regimens indicate that these antibiotic combinations produce clinical evidence of cure in most cases (table 2). The superiority of one regimen
Table 1. Summary of 1985 and 1989 guidelines for treatment of PID.
PID Treatment Options
S659
Table 2. Results with selected regimens for inpatient treatment of PID. Type of regimen, reference 1985 inpatient treatment schedules Cm-based 19
20 21 23 24
25 26 27 Cfox-based 23 24
25 27
29 30
31
Cm + Tm Cm + Gm Cm + Gm Cm + Gm Cm + Gm Cm + Tm Cm + Amik Cm + Gm Cm + Gm Cfox Cfox + Dox Cfox + Dox Cfox + Dox Cfox + Dox Cfox + Dox Cfox
22 28
Clinical cure (%)
19 37
84 100 94 83 84* 92 90 90* 92
23
90*
18
6 30
39 31
40
95
31 40
97
32 36
13
100 92 88
9
100
98
Treatment schedules using new antibiotics Crn-based
33 34 Ceph3-based
29 29 35 35 Ceph2-based 30
31 32 Pen-based 21 26 36 37
38
Cm Cm
+ +
Azt Azt
Czox + Dox Czox Czoxt
Ctax l
+
42
98
30 59 18 19
100 97
88 94
29 46
94 92* 93*
Amp + Sulb Pip Amp + Sulb Pen + Mtz BPen + Dox
21 18
97
Cpfx Cpfx Mtz + Gm Mtz + Dox TMP-SMZ
28
89
8
100 84*
Cfot Cfot Cfot
Dox
32
20
86
17
95 47
32
97:1:
Outpatient Regimens
The use of intramuscular cefoxitin plus oral probenecid followed by doxycycline for the outpatient treatment of PIO was evaluated in one report [39]. N. gonorrhoeae and C trachomatis were isolated in initial endocervical or endometrial specimens in more than half of the cases identified as probable or possible PIO. In these two groups, 31 of 34 infections were cured or mitigated.
Other
20 22 36 37
38
19 16 32
81
97:1:
NOTE. Cm = clindamycin; Tm = tobramycin; Gm = gentamicin; Amik = amikacin; Cfox = cefoxitin; Dox = doxycycline; Azt = aztreonam; Ceph3 = third-generation cephalosporin; Czox = ceftizoxime; Ctax = cefotaxime; Ceph2 = second-generation cephalosporin; Cfot = cefotetan; Pen = penicillin; Amp = ampicillin; Sulb = sulbactam; Pip = piperacillin; Mtz = metronidazole; BPen = benzylpenicillin; Cpfx = ciprofloxacin; TMP-SMZ = trimethoprim-sulfamethoxazole. * Other pelvic infections included in addition to PID; cure rate based on all cases treated. t Tetracycline administered orally to patients with endocervical cultures positive for C. trachomatis. :I: Based on posttreatment laparoscopy indicating at least one patent oviduct.
New Antibiotic Choices Inpatient Clinical Trials
Since 1985, clinicians have had new options for the treatment of PIO; these alternatives include new second- and third-generation cephalosporins, combinations of penicillins with 13-lactamaseinhibitors, a monobactam, a carbapenem, and new oral quinolones. Selected clinical trials of regimens including these antibiotics are summarized in table 2. These studies have the same methodologic limitations as the validation studies of the 1985recommendations. Some of the new antibiotics appear to be clinically effective; some cause fewer adverse reactions and of-
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
22
Antibiotics
No. treated
to the other cannot be discerned from available data [19-27, 29-38]. Interpretation of these studies is complicated by a lack of uniformity in study design; the investigations reviewed used various diagnostic criteria, culture techniques, methods of determining clinical or microbiologic cure, and follow-up periods. Few studies used laparoscopy for direct visualization and culture of the adnexae for confirmation of the diagnosis of PIO; likewise, few used endometrial biopsy for histologic confirmation of endometritis and isolation of the etiologic microbes. Treatment failures occurred infrequently when the 1985 guidelines were used. In such instances, most failures appeared to be due to the presence of tuboovarian abscess. In a series of 68 patients treated with cefotetan plus doxycycline or with cefoxitin plus doxycycline, four patients in whom treatment failed all responded clinically to clindamycin and gentamicin [30]. This response perhaps reflects better coverage of Bacteroides by clindamycin. No antibiotic regimen was proved to preserve tubal patency more effectively than another. However, long-term follow-up was generally not conducted in the treatment efficacy trials.
Peterson, Galaid, and Zeni/man
S660
Outpatient Clinical Trials
Oral preparations of ampicillin or amoxicillin plus ~-lactamase inhibitors such as clavulanic acid or sulbactam are theoretically promising for the outpatient management of PID and have been clinically effective for those patients able to complete therapy. However, adverse reactions, usually gastrointestinal, were reported for 28% of the patients in one study [42]. Oral doxycycline should be included in such regimens for adequate coverage of C. trachomatis.
Development of Treatment Guidelines for 1989 Practical Considerations
The validation studies of the 1985 treatment regimens and clinical trials of the new antibiotics suggest that a number of regimens may be useful for the treatment of PID. Until data are uniformly available to place these observations in perspective with microbiologic data, laparoscopic evidence of cure, and follow-up of long-term sequelae (for example, tubal pregnancy, infertility, and chronic pelvic pain), the optimal treatment of PID cannot be singled out from the therapeutic trials that have been reported. Clinical judgment still plays a major role in the formulation of these recommendations. Clinicians, administrators, and lawyershave sometimes interpreted a given list of specific antibiotics as representing the only agents acceptable for the treatment of PID. We emphasize that these lists are treatment recommendations and that the choice of a regimen may be influenced by many factors, including institutional cost-control efforts, acceptability to the patient, and regional differences in bacterial resistance.
In attempting to control costs, hospital pharmacies have often restricted formularies to a limited number of drugs within each class. Price schedules for commonly used antibiotics vary widely (table 3). The use of relatively simple drug regimens and oral medications reduces administration costs. Depending upon the direct cost of the drug and related indirect costs, some newer cephalosporins offer savings. Cefotetan (2 g iv every 12 hours) offers the advantage of less frequent administration than is needed with cefoxitin (2 g iv every 6 hours). Treatment with ceftriaxone (1 g iv daily) reduces administration costs further. Aztreonam, a new monobactam, offers the coverageprovided byaminoglycosides without ototoxicity or nephrotoxicity, albeit at much higher cost. The proportion of gonococcal infections caused by antibiotic-resistant strains of N. gonorrhoeae has risen substantially over the past 5 years. In 1988, 13010 of gonococcal isolates had cefoxitin MICs of ~1 mg/mL, and 4010 had MICs of ~2 mg/mL (CDC, unpublished data). PID caused by penicillinaseproducing N. gonorrhoeae is not adequately treated by the combination of doxycycline with amoxicillin, ampicillin, or aqueous procaine penicillin. Single doses of penicillin or cephalosporin followed by oral tetracycline may not provide sustained activity against many strains of N. gonorrhoeae with chromosomally mediated resistance or against the facultative or anaerobic organisms involved in PID. These circumstances have prompted us to discontinue our previous recommendation of therapy with penicillin or ampicillin for PID. All gonococci isolated from patients with PID should be evaluated for susceptibility to tetracycline, penicillin, cefoxitin, and ceftriaxone. Cases in which cefoxitin failure is suspected should be managed in consultation with an expert. No new drugs are available at this time for the treatment of infection with C. trachomatis. Doxycycline remains the mainstay of therapy. Although clindamycin has moderate in vitro activity against C. trachomatis, clinical studies of its efficacy are still limited. Hospitalization
The criteria for hospitalization and inpatient treatment that appeared in the 1985 guidelines are as follows: Hospitalization of patients with acute PID is indicated when (1) the diagnosis is uncertain; (2) sur-
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
fer potential cost-savings. The latter advantages may influence the choice of antibiotics for the inpatient management of PID. This issue willbe discussed further in the section on practical considerations. The search for a single agent effective in the treatment of PID has led to several comparative studies of ciprofloxacin, a new quinolone, vs. clindamycin plus an aminoglycoside. Although clinical cure rates with ciprofloxacin have been high, only small numbers of patients have been treated, and the results of treatment of mucosal C. trachomatis infections appear inadequate [20,22, 40]. Further, ciprofloxacin has limited activity against anaerobes [41].
PID Treatment Options
8661
Table 3. Cost of selected antibiotics for the treatment of PID. Type of regimen, drug(s)
Drug cost in dollars/d*
4 2 3 1 4 2 2
52 36 62 30 0.24 0.22 8
3
52t
.. .+ .. .+ 4 2
18 10 2.40§ 2.20§
NOTE. The source of these data is the list of average wholesale prices in [43]. Costs of treatment may vary because of discounts given to volume purchasers. Use of generic formulations may also lower costs. * Cost per day excludes administration costs and pharmacist's professional fee. t Cost of determining therapeutic levels of gentamicin is excluded. :j: One dose is given im stat. § Daily cost is calculated on the basis of a lO-day course.
gical emergencies such as appendicitis and ectopic pregnancy cannot be excluded; (3) a pelvic abscess is suspected; (4) the patient is pregnant; (5) the patient is a prepubertal child; (6) severe illness precludes outpatient management; (7) the patient is unable to follow or tolerate an outpatient regimen; (8) the infection has failed to respond to outpatient therapy; or (9) clinical follow-up within 72 hours of the start of antibiotic treatment cannot be arranged. Many experts recommend that all patients with PID be hospitalized for treatment. Special consideration with regard to hospitalization should be given to adolescents because their compliance with therapy is unpredictable and the long-term sequelae of PID are particularly severe in this group. In addition, it would be ideal to hospitalize nulliparous women
Pelvic Inflammatory Disease: Review of Treatment Options Herbert B. Peterson, Edward I. Galaid, and Jonathan M. Zenilman
From the U. S. Department of Health and Human Services, Public Health Services, Division of Reproductive Health; and the Division of STDIHIV Prevention, Centers for Disease Control, Atlanta, Georgia
Pelvic inflammatory disease (PID) is an oftenserious consequence of gonococcal and chlamydial infection. An estimated 1 million women in the United States are treated each year for this condition, and at least one-fourth of these women will suffer one or more serious sequelae, including infertility, ectopic pregnancy, and major abdominal/pelvic surgery. It has been estimated that the cost of these sequelae will be $3.5 billion per year [1] by 1990.Since the early 1970s, the Centers for Disease Control (CDC) has recommended therapeutic regimens for PID as part of the treatment guidelines for sexually transmitted diseases (STDs). In this report we will review the current understanding of PID as a clinical syndrome, the 1985 CDC recommendations and their rationale, and the advances that have been made in the understanding of the diagnosis and treatment of PID since 1985.We will also make further recommendations for PID treatment regimens. Clinical Background PID is an entity that has escaped consistent definition. In the 1985 STD treatment guidelines published by the CDC [2], PID was defined as "the acute clinical syndrome (unrelated to pregnancy or surgery) attributed to the ascent of microorganisms from the Please address requests for reprints to Technical Information Services (E06), Center for Prevention Services, Centers for Disease Control, Atlanta, Georgia 30333.
S656
vagina and endocervix to the endometrium, fallopian tubes, and/or contiguous structures." Important gaps in our understanding of the pathophysiology of PID still exist. PID is a syndrome that may include endometritis, salpingitis, oophoritis, tubo-ovarian abscess, and pelvic peritonitis. Infection of the fallopian tubes is the hallmark of PID. In some women, infection of the fallopian tubes and contiguous structures leads to the formation of tubo-ovarian abscesses. In contrast to what is known or suspected regarding ab-scesses elsewhere within the body, tubo-ovarian abscesses often respond to antibiotic therapy without surgical intervention. The exact mechanism by which microorganisms infect the upper genital tract is unknown. One unifying hypothesis regarding the pathogenesis of acute PID suggests that it usually starts as a cervical infection with Neisseria gonorrhoeae and/or Chlamydia trachomatis, with subsequent alteration in the cervicovaginal microenvironment as a result of the consumption of nutrients and the production of metabolic waste products. These changes cause others, including alterations in vaginal pH and the availability of oxygen, which result in overgrowth of endogenous and anaerobic flora. The original cervical pathogens and/or endogenous organisms then ascend into the upper genital tract [3]. This hypothesis would explain the polymicrobial nature of PID. Alternative theories assume that polymicrobial PID can be initiated without N. gonorrhoeae [4] or
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
Decisions regarding appropriate antibiotic therapy for treatment of pelvic inflammatory disease (PID) are complicated by an incomplete understanding of the syndrome. Further, the lack of data on laparoscopic diagnosis and cure of PID severely limits our ability to interpret data on therapy outcome. Validation studies of the treatment regimens recommended by the Centers for Disease Control in the 1985treatment guidelines for sexually transmitted diseases (STDs) suggest that both the cefoxitin/doxycycline and clindamycin/aminoglycoside combination regimens result in consistently high rates of clinical evidence of cure. The 1989 STD treatment guidelines were based on these studies and on available data regarding newer treatment regimens. Empiric, broad-spectrum therapy remains the treatment of choice. The two regimens recommended for inpatient therapy in 1989 are similar to those recommended in 1985. The recommendation for management of ambulatory patients has been substantially changed, however, because of increasing resistance of Neisseria gonorrhoeae to penicillin.
PID Treatment Options
rent STD treatment guidelines [12-14]. The importance of optimal therapy was addressed indirectly by studies of laparoscopically verified PID in Sweden; these investigations suggested that women with an early diagnosis and effective treatment have fewer long-term sequelae [5]. Decisions regarding optimal antibiotic therapy for PID are further complicated by an incomplete understanding of the relationship between in vitro observations and clinical efficacy. For example, an aminoglycoside may be highly effective in the laboratory but less effective clinically, perhaps because of conditions associated with abscess cavities or anaerobic environments. Paradoxically, treatment with antibiotics that are theoretically inadequate has sometimes resulted in clinical evidence of cure of PID. However, studies reporting such favorable clinical outcomes do not account for either the natural history of the infection, which may resolve spontaneously, or the relationship between clinical diagnosis and true resolution of pathology in the upper genital tract. Only recently has C trachomatis been appreciated as a significant sexually transmitted pathogen. The important role that this organism plays in the etiology of PID complicates therapeutic considerations. Because C trachomatis infection (of both the cervix and the upper genital tract) is often asymptomatic, diagnosis is difficult, especially when laparoscopy is not performed. C trachomatis often causes substantial tissue destruction, even in asymptomatic cases, and such destruction is a major factor in subsequent infertility [15, 16]. Clinical evidence of improvement, therefore, may not correlate well with resolution of primary pathology. For example, resolution of symptoms may not be an accurate reflection of cessation of tubal inflammation. It is also difficult to assess whether drugs with in vitro activity against C trachomatis are clinically effective, including whether they prevent serious tissue destruction. Because symptoms may improve during antibiotic treatment despite continuing mucosal and soft tissue infection with C trachomatis, clinical cure may be a poor measure of outcome. Thus, any decision regarding treatment regimens must include a consideration of the serious sequelae of C trachomatis infections and of what little is known about the relationship between in vitro and in vivo antibiotic efficacy in these infections. Data regarding treatment of PID are based primarily on hospitalized patients. Few data exist on
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
C trachomatis. Organisms isolated from women with polymicrobial PID have generally fallen into four broad categories: N. gonorrhoeae alone, C trachomatis alone, N. gonorrhoeae and C trachomatis together, and anaerobic and/or facultative bacteria [4-6]. From the standpoint of diagnosis, the presence of a pathogen in the endocervix may not always indicate that this organism is the agent causing infection of the endometrium, the fallopian tubes, or the peritoneal cavity. Further, infectious agents can be recovered from endometrial cultures even when cervical cultures for the same agents are negative [7]. In their classic studies, Jacobson and Westrom [8] found that a clinical diagnosis of PID was erroneous in one-third of cases when laparoscopy was used as the diagnostic gold standard. At least 11 other studies have reported that laparoscopy is required for establishing the diagnosis of PID with a high degree of specificity [6, 9, 10]. Hadgu et al. [11] developed a mathematical model that correctly predicted 87070 of cases of acute PID jn a series of laparoscopically confirmed diagnoses in Lund, Sweden. Variables that were good predictors in that model were purulent vaginal discharge, an erythrocyte sedimentation rate of ~ 15 mm/h, a culture positive for N. gonorrhoeae, adnexal swelling on bimanual examination, and a rectal temperature of ~ 38°C. Unfortunately, few studies evaluating treatment have included any laparoscopically verified cases. Studies of therapeutic efficacy have used clinical parameters for diagnosis that cannot be evaluated relative to the model described above or relative to any other useful standard for establishing commonality of diagnoses among the various studies. The lack of data on laparoscopic diagnosis and cure of PID severely limits our ability to interpret data on the outcome of therapy. Few studies include laparoscopic verification 0'[ cure, and most studies fail to report the basis on which cure was determined. In almost all cases, clinical cure was determined without laparoscopic or hysterosalpingographic follow-up. Available data on antibiotic efficacy do not address the prevention of tubal destruction or the related serious (and often life-threatening) long-term sequelae of PID. To date, no studies in the United States have compared different treatment regimens with fertility as an outcome measure. The limited available data on efficacy focus on antibiotic regimens not recommended in either the 1985 or the cur-
5657
Peterson, Galaid, and Zeni/man
5658
which to base guidelines for outpatient treatment. Clearly, the polymicrobial nature of PID necessitates broad-spectrum antibiotic coverage, even if hospitalization is not required. Clinical Validation of the 1985 Treatment Guidelines Inpatient Regimens
1989
1985
Inpatients" Regimen A Cefoxitin (2 g iv every 6 h) Cefoxitin (2 g iv every 6 h) PLUS OR Doxycycline (100 mg iv Cefotetan t (2 g iv every 12 h) every 12 h) PLUS Doxycycline (100 mg po* or iv every 12 h) After discharge: Doxycycline (100 mg po twice daily to complete a total of 10-14 d of therapy)
After discharge: Doxycycline (100 mg po twice daily to complete a total of 10-14 d of therapy)
Regimen B Clindamycin (600 mg iv Clindamycin (900 mg iv every 8 h) every 6 h) PLUS PLUS Gentamicin (2.0 mg/kg as iv Gentamicin (2.0 mg/kg as iv loading dose followed by loading dose followed by 1.5 mg/kg iv every 8 h in 1.5 mg/kg iv every 8 h in patients with normal renal patients with normal renal function) function) After discharge: Clindamycin (450 mg po 4 times daily to complete a total of 10-14 d of therapy)
Cefoxitin] (2 g im)
After discharge: Doxycycline (100 mg po twice daily to complete a total of 10-14 d of therapy)§
Outpatients Cefoxitin] (2 g im)
OR
OR
Amoxicillin (3.0 g po)
Ceftriaxone (250 mg im) PLUS Doxycycline (100 mg po twice daily for 10-14 d)
OR Ampicillin (3.5 gpo)
OR Aqueous procaine penicillin G (4.8 mU im)
OR Ceftriaxone (250 mg im) PLUS Doxycycline (100 mg po twice daily for 10-14 d)
OR Tetracycline (500 mg po 4 times daily) • All regimens are given for at least 48 hours after the patient improves clinically. t Other cephalosporins (such as ceftizoxime, cefotaxime, and ceftriaxone), which provide adequate coverage against gonococci and other facultative gram-negative aerobic and anaerobic bacteria, may be utilized in appropriate doses. See text and table 2 for further discussion. po = per os. § Continuation of clindamycin (450 mg po, 4 times daily, for a total of 10-14 d) may be considered as an alternative. See comments regarding oral clindamycin in text. II Each of these regimens except ceftriaxone is accompanied by probenecid (1 gpo).
*
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
The antibiotic regimens recommended in the 1985 guidelines (table 1)were empirically designed to provide broad coverage for mixed, aerobic, and anaerobic infections. In regimen A (cefoxitin/doxycycline), cefoxitin was chosen for coverage of N. gonorrhoeae, the Enterobacteriaceae, and anaerobes. The inclusion of doxycycline, which also gives some broad-spectrum coverage, was based on recognition of the role of C trachomatis in PID. Neither regimen A nor regimen B provided coverage for enterococci. In regimen B (clindamycin/gentamicin), clindamycin was chosen for coverage of anaerobes and grampositive aerobes. The efficacy of clindamycinin treating these organisms (especially Bacteroides), even when they occur in a tubo-ovarian abscess, is considered a particular advantage in clinically advanced PID. Although clindamycin has only moderate in vitro activity against C trachomatis and although the ability of oral clindamycin to eradicate C trachomatis from the cervix or urethra has been variable [17, 18], treatment with high-dose intravenous clindamycin has resulted in clinical evidence of cure and eradication of C trachomatis from the endometrium in a relatively small series of cases [19-27]. In one report, women with severe salpingitis at laparoscopy responded slowly to therapy; the three women (of 19 treated) in whom treatment failed clinically all were initially infected with C trachomatis. The authors of that report speculated that the eradication of C trachomatiswas likely attributable to the high doses of intravenous and oral clindamycin administered [19]. Aminoglycosides cover gram-negative aerobes, including N. gonorrhoeae; when administered in combination with clindamycin, they are highly effective in treating abscesses. Brunham has reviewed studies of antimicrobial treatment for PID reported from 1974 to 1984 [28]. Since the 1985 guidelines were issued, studies of the two inpatient regimens indicate that these antibiotic combinations produce clinical evidence of cure in most cases (table 2). The superiority of one regimen
Table 1. Summary of 1985 and 1989 guidelines for treatment of PID.
PID Treatment Options
S659
Table 2. Results with selected regimens for inpatient treatment of PID. Type of regimen, reference 1985 inpatient treatment schedules Cm-based 19
20 21 23 24
25 26 27 Cfox-based 23 24
25 27
29 30
31
Cm + Tm Cm + Gm Cm + Gm Cm + Gm Cm + Gm Cm + Tm Cm + Amik Cm + Gm Cm + Gm Cfox Cfox + Dox Cfox + Dox Cfox + Dox Cfox + Dox Cfox + Dox Cfox
22 28
Clinical cure (%)
19 37
84 100 94 83 84* 92 90 90* 92
23
90*
18
6 30
39 31
40
95
31 40
97
32 36
13
100 92 88
9
100
98
Treatment schedules using new antibiotics Crn-based
33 34 Ceph3-based
29 29 35 35 Ceph2-based 30
31 32 Pen-based 21 26 36 37
38
Cm Cm
+ +
Azt Azt
Czox + Dox Czox Czoxt
Ctax l
+
42
98
30 59 18 19
100 97
88 94
29 46
94 92* 93*
Amp + Sulb Pip Amp + Sulb Pen + Mtz BPen + Dox
21 18
97
Cpfx Cpfx Mtz + Gm Mtz + Dox TMP-SMZ
28
89
8
100 84*
Cfot Cfot Cfot
Dox
32
20
86
17
95 47
32
97:1:
Outpatient Regimens
The use of intramuscular cefoxitin plus oral probenecid followed by doxycycline for the outpatient treatment of PIO was evaluated in one report [39]. N. gonorrhoeae and C trachomatis were isolated in initial endocervical or endometrial specimens in more than half of the cases identified as probable or possible PIO. In these two groups, 31 of 34 infections were cured or mitigated.
Other
20 22 36 37
38
19 16 32
81
97:1:
NOTE. Cm = clindamycin; Tm = tobramycin; Gm = gentamicin; Amik = amikacin; Cfox = cefoxitin; Dox = doxycycline; Azt = aztreonam; Ceph3 = third-generation cephalosporin; Czox = ceftizoxime; Ctax = cefotaxime; Ceph2 = second-generation cephalosporin; Cfot = cefotetan; Pen = penicillin; Amp = ampicillin; Sulb = sulbactam; Pip = piperacillin; Mtz = metronidazole; BPen = benzylpenicillin; Cpfx = ciprofloxacin; TMP-SMZ = trimethoprim-sulfamethoxazole. * Other pelvic infections included in addition to PID; cure rate based on all cases treated. t Tetracycline administered orally to patients with endocervical cultures positive for C. trachomatis. :I: Based on posttreatment laparoscopy indicating at least one patent oviduct.
New Antibiotic Choices Inpatient Clinical Trials
Since 1985, clinicians have had new options for the treatment of PIO; these alternatives include new second- and third-generation cephalosporins, combinations of penicillins with 13-lactamaseinhibitors, a monobactam, a carbapenem, and new oral quinolones. Selected clinical trials of regimens including these antibiotics are summarized in table 2. These studies have the same methodologic limitations as the validation studies of the 1985recommendations. Some of the new antibiotics appear to be clinically effective; some cause fewer adverse reactions and of-
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
22
Antibiotics
No. treated
to the other cannot be discerned from available data [19-27, 29-38]. Interpretation of these studies is complicated by a lack of uniformity in study design; the investigations reviewed used various diagnostic criteria, culture techniques, methods of determining clinical or microbiologic cure, and follow-up periods. Few studies used laparoscopy for direct visualization and culture of the adnexae for confirmation of the diagnosis of PIO; likewise, few used endometrial biopsy for histologic confirmation of endometritis and isolation of the etiologic microbes. Treatment failures occurred infrequently when the 1985 guidelines were used. In such instances, most failures appeared to be due to the presence of tuboovarian abscess. In a series of 68 patients treated with cefotetan plus doxycycline or with cefoxitin plus doxycycline, four patients in whom treatment failed all responded clinically to clindamycin and gentamicin [30]. This response perhaps reflects better coverage of Bacteroides by clindamycin. No antibiotic regimen was proved to preserve tubal patency more effectively than another. However, long-term follow-up was generally not conducted in the treatment efficacy trials.
Peterson, Galaid, and Zeni/man
S660
Outpatient Clinical Trials
Oral preparations of ampicillin or amoxicillin plus ~-lactamase inhibitors such as clavulanic acid or sulbactam are theoretically promising for the outpatient management of PID and have been clinically effective for those patients able to complete therapy. However, adverse reactions, usually gastrointestinal, were reported for 28% of the patients in one study [42]. Oral doxycycline should be included in such regimens for adequate coverage of C. trachomatis.
Development of Treatment Guidelines for 1989 Practical Considerations
The validation studies of the 1985 treatment regimens and clinical trials of the new antibiotics suggest that a number of regimens may be useful for the treatment of PID. Until data are uniformly available to place these observations in perspective with microbiologic data, laparoscopic evidence of cure, and follow-up of long-term sequelae (for example, tubal pregnancy, infertility, and chronic pelvic pain), the optimal treatment of PID cannot be singled out from the therapeutic trials that have been reported. Clinical judgment still plays a major role in the formulation of these recommendations. Clinicians, administrators, and lawyershave sometimes interpreted a given list of specific antibiotics as representing the only agents acceptable for the treatment of PID. We emphasize that these lists are treatment recommendations and that the choice of a regimen may be influenced by many factors, including institutional cost-control efforts, acceptability to the patient, and regional differences in bacterial resistance.
In attempting to control costs, hospital pharmacies have often restricted formularies to a limited number of drugs within each class. Price schedules for commonly used antibiotics vary widely (table 3). The use of relatively simple drug regimens and oral medications reduces administration costs. Depending upon the direct cost of the drug and related indirect costs, some newer cephalosporins offer savings. Cefotetan (2 g iv every 12 hours) offers the advantage of less frequent administration than is needed with cefoxitin (2 g iv every 6 hours). Treatment with ceftriaxone (1 g iv daily) reduces administration costs further. Aztreonam, a new monobactam, offers the coverageprovided byaminoglycosides without ototoxicity or nephrotoxicity, albeit at much higher cost. The proportion of gonococcal infections caused by antibiotic-resistant strains of N. gonorrhoeae has risen substantially over the past 5 years. In 1988, 13010 of gonococcal isolates had cefoxitin MICs of ~1 mg/mL, and 4010 had MICs of ~2 mg/mL (CDC, unpublished data). PID caused by penicillinaseproducing N. gonorrhoeae is not adequately treated by the combination of doxycycline with amoxicillin, ampicillin, or aqueous procaine penicillin. Single doses of penicillin or cephalosporin followed by oral tetracycline may not provide sustained activity against many strains of N. gonorrhoeae with chromosomally mediated resistance or against the facultative or anaerobic organisms involved in PID. These circumstances have prompted us to discontinue our previous recommendation of therapy with penicillin or ampicillin for PID. All gonococci isolated from patients with PID should be evaluated for susceptibility to tetracycline, penicillin, cefoxitin, and ceftriaxone. Cases in which cefoxitin failure is suspected should be managed in consultation with an expert. No new drugs are available at this time for the treatment of infection with C. trachomatis. Doxycycline remains the mainstay of therapy. Although clindamycin has moderate in vitro activity against C. trachomatis, clinical studies of its efficacy are still limited. Hospitalization
The criteria for hospitalization and inpatient treatment that appeared in the 1985 guidelines are as follows: Hospitalization of patients with acute PID is indicated when (1) the diagnosis is uncertain; (2) sur-
Downloaded from http://cid.oxfordjournals.org/ at University of New South Wales on August 16, 2015
fer potential cost-savings. The latter advantages may influence the choice of antibiotics for the inpatient management of PID. This issue willbe discussed further in the section on practical considerations. The search for a single agent effective in the treatment of PID has led to several comparative studies of ciprofloxacin, a new quinolone, vs. clindamycin plus an aminoglycoside. Although clinical cure rates with ciprofloxacin have been high, only small numbers of patients have been treated, and the results of treatment of mucosal C. trachomatis infections appear inadequate [20,22, 40]. Further, ciprofloxacin has limited activity against anaerobes [41].
PID Treatment Options
8661
Table 3. Cost of selected antibiotics for the treatment of PID. Type of regimen, drug(s)
Drug cost in dollars/d*
4 2 3 1 4 2 2
52 36 62 30 0.24 0.22 8
3
52t
.. .+ .. .+ 4 2
18 10 2.40§ 2.20§
NOTE. The source of these data is the list of average wholesale prices in [43]. Costs of treatment may vary because of discounts given to volume purchasers. Use of generic formulations may also lower costs. * Cost per day excludes administration costs and pharmacist's professional fee. t Cost of determining therapeutic levels of gentamicin is excluded. :j: One dose is given im stat. § Daily cost is calculated on the basis of a lO-day course.
gical emergencies such as appendicitis and ectopic pregnancy cannot be excluded; (3) a pelvic abscess is suspected; (4) the patient is pregnant; (5) the patient is a prepubertal child; (6) severe illness precludes outpatient management; (7) the patient is unable to follow or tolerate an outpatient regimen; (8) the infection has failed to respond to outpatient therapy; or (9) clinical follow-up within 72 hours of the start of antibiotic treatment cannot be arranged. Many experts recommend that all patients with PID be hospitalized for treatment. Special consideration with regard to hospitalization should be given to adolescents because their compliance with therapy is unpredictable and the long-term sequelae of PID are particularly severe in this group. In addition, it would be ideal to hospitalize nulliparous women
