A Randomized Controlled Trial of Ceftriaxone and Doxycycline, With or Without Metronidazole, for the Treatment of Acute Pelvic Inflammatory Disease

Abstract

Background

Anaerobic organisms are important pathogens in acute pelvic inflammatory disease (PID). The currently recommended PID regimen of a single dose of ceftriaxone and doxycycline for 14 days has limited anaerobic activity. The need for broader anaerobic coverage is unknown and concerns have been raised about metronidazole tolerability.

Methods

We conducted a randomized, double-blind, placebo-controlled trial comparing ceftriaxone 250 mg intramuscular single dose and doxycycline for 14 days, with or without 14 days of metronidazole in women with acute PID. The primary outcome was clinical improvement at 3 days following enrollment. Additional outcomes at 30 days following treatment were the presence of anaerobic organisms in the endometrium, clinical cure (absence of fever and reduction in tenderness), adherence, and tolerability.

Results

We enrolled 233 women (116 to metronidazole and 117 to placebo). Clinical improvement at 3 days was similar between the 2 groups. At 30 days following treatment, anaerobic organisms were less frequently recovered from the endometrium in women treated with metronidazole than placebo (8% vs 21%, P < .05) and cervical Mycoplasma genitalium was reduced (4% vs 14%, P < .05). Pelvic tenderness was also less common among women receiving metronidazole (9% vs 20%, P < .05). Adverse events and adherence were similar in each treatment group.

Conclusions

In women treated for acute PID, the addition of metronidazole to ceftriaxone and doxycycline was well tolerated and resulted in reduced endometrial anaerobes, decreased M. genitalium, and reduced pelvic tenderness compared to ceftriaxone and doxycycline. Metronidazole should be routinely added to ceftriaxone and doxycycline for the treatment of women with acute PID.

Clinical Trials Registration

NCT01160640.

pelvic inflammatory disease, anaerobes, metronidazole

(See the Editorial Commentary by Mitchell on pages 1190–91.)

Pelvic inflammatory disease (PID) results from ascension of microorganisms from the vagina or endocervix to the endometrium and fallopian tubes. Organisms recognized to cause PID and its sequelae include Chlamydia trachomatis and Neisseria gonorrhoeae. Mycoplasma genitalium has been associated with endometritis but its association with infertility is less certain [1]. Facultative and anaerobic microbes associated with vaginal dysbiosis have been associated with endometrial and tubal infections and are recovered from the upper genital tract at higher frequency than C. trachomatis or N. gonorrhoeae [2–4].

The Centers for Disease Control and Prevention (CDC) recommends a single intramuscular dose of a cephalosporin in combination with oral doxycycline for the outpatient treatment of PID [5]. This regimen is effective against N. gonorrhoeae and C. trachomatis, but has limited activity against anaerobic organisms. Despite the frequent recovery of anaerobic organisms in women with acute PID, the need for antimicrobial therapy with broader anaerobic coverage is unknown. This uncertainty is reflected in the CDC guidelines that list metronidazole as an optional addition to ceftriaxone and doxycycline, while the European guidelines recommend the addition of metronidazole [6]. However, a recent systematic review and meta-analysis of acute PID treatment did not show clear evidence for the use of nitroimidazoles, although none of the studies in the analysis involved the current CDC-recommended regimen of ceftriaxone and doxycycline [7]. In addition to the uncertainty whether the addition of metronidazole improves treatment outcomes, there are concerns about the tolerability of metronidazole as part of a 3-drug combination. To determine whether metronidazole should be incorporated in the first-line recommended PID treatment regimen, we compared a single dose of ceftriaxone and 14 days of doxycycline to the same regimen with the addition of 14 days of metronidazole in women diagnosed with acute PID.

METHODS

Trial Design and Oversight

In this randomized, double-blind, placebo-controlled trial, we compared ceftriaxone and doxycycline to ceftriaxone, doxycycline, and metronidazole for the treatment of acute PID. The investigators initiated this National Institutes of Health–funded study, which was conducted in accordance with the Consolidated Standard of Reporting Trials guidelines and is registered at ClinicalTrials.gov (NCT01160640). The study was approved by the University of Pittsburgh’s Institutional Review Board and was monitored by an independent data and safety monitoring committee. Written informed consent was obtained from each participant.

Participants

Enrollment occurred at the emergency departments at Magee-Womens Hospital and University of Pittsburgh Medical Center Mercy Hospital, and at the Allegheny County Health Department Sexually Transmitted Diseases (STD) Clinic, all in Pittsburgh, Pennsylvania. Clinicians referred women diagnosed with acute PID for possible participation. Women 15–40 years of age were eligible for enrollment if they met the following criteria for acute PID: pelvic or lower abdominal pain, and the presence of cervical motion tenderness, uterine tenderness, or adnexal tenderness on pelvic examination [5]. Women requiring inpatient treatment (tuboovarian abscess, severe illness, vomiting, inability to follow oral treatment) and women who were pregnant were not eligible. Other exclusion criteria included use of systemic or vaginal antibiotics within the preceding 7 days, allergy to cephalosporins, doxycycline, or metronidazole, uterine procedure or miscarriage within the prior 6 weeks, hysterectomy, and menopause.

Study Procedures

At enrollment, a history and physical examination were performed by trained study clinicians. Participants rated their current pain on a 100-mm visual analogue scale. A clinical tenderness score was determined by assessing pelvic tenderness using a modification of a previously established grading system [8, 9]. Cervical motion tenderness, uterine tenderness, and tenderness of the right and left adnexae were each graded from 0 (no tenderness) to 3 (tenderness causing marked distress), with a maximal tenderness score of 12. Vaginal fluid samples were obtained by swabbing the lateral vaginal walls, and endocervical swabs were collected. The cervix and endocervix were then cleansed with povidone-iodine and excess disinfectant was removed with a sterile swab. An endometrial aspirate was obtained by passing a sterile suction catheter (Pipelle, Cooper Surgical, Trumbull, Connecticut) through the endocervical canal into the endometrial cavity. After removal, the external catheter surface was wiped with alcohol to reduce surface contamination. The endometrial sample was expulsed into a sterile dish for microbiologic processing. Women in whom endometrial sampling was unsuccessful were discontinued from further participation and treated per routine clinical care.

Treatment Regimens

All women received combination antibiotic therapy of a single intramuscular 250-mg dose of ceftriaxone and doxycycline 100 mg orally twice daily for 14 days [5]. All participants were randomized, in 1:1 fashion, to also receive metronidazole 500 mg orally twice daily or matching placebo for 14 days, with equal frequency to the 2 treatment arms using a permuted block design with random block sizes of 2, 4, and 6. The research staff was blinded to treatment allocation. Doxycycline, metronidazole, and placebo were distributed in blister cards. Women were instructed to inform their partner(s) to seek evaluation and presumptive STD treatment.

Follow-up

All women were asked to return at 3 days (range, 2–7 days) after enrollment, at which time an interim history and pelvic examination were performed. A final evaluation, performed at 30 days (range, 25–45 days) after enrollment, included symptom assessment and a pelvic examination. Vaginal and cervical swabs were obtained as on enrollment, and endometrial sampling was repeated.

Women without clinical improvement or who did not tolerate study medications were offered hospitalization for parenteral antimicrobial therapy. Those women were followed for safety only and did not undergo further study procedures. Women who were pregnant at the follow-up visits were withdrawn from the study.

Microbiology

A Gram stain was prepared from a vaginal swab sample and interpreted for bacterial vaginosis [10]. A score of 7 or greater indicated bacterial vaginosis. Another vaginal swab was assayed for Trichomonas vaginalis by transcription-mediated assay (TMA, Hologic, San Diego, California). Endocervical swabs and endometrial tissue were analyzed for C. trachomatis and N. gonorrhoeae by TMA (Aptima Combo 2 Assay for CT/NG, Hologic) and for Mycoplasma genitalium by TMA using analyte-specific reagents supplied by Hologic, Inc. The endometrial aspirate was transported in anaerobic transport medium and culture cultivation was performed for facultative and anaerobic microorganisms as previously described [11].

Outcomes

The primary study outcome was clinical improvement at the 3-day follow-up visit, defined by a reduction in the clinical tenderness score. Secondary outcomes included the presence of anaerobic organisms in the endometrium at 30 days following treatment, and clinical cure at 30 days (> 70% improvement in the clinical tenderness score and the absence of fever [oral temperature < 38°C]).

Adverse Events and Medication Adherence

Adverse events were assessed at each follow-up visit. Blister cards were examined at the 30-day visit, and participants were considered to be adherent with study medications if they returned their blister card and took at least 75% of both the metronidazole and doxycycline doses.

Statistical Analysis

Data analyses were conducted using SPSS statistical software, release 26.0 (IBM Corporation, Armonk, New York) and statistical tests were evaluated at the 2-sided .05 significance level. Differences in participant characteristics and study outcomes were compared between the 2 treatment arms using Fisher exact test, Student t test, and Mann-Whitney U test, where appropriate. The intent-to-treat analyses on the primary endpoint, clinical improvement at 3 days, included all participants as randomized and missing data were treated as failure to improve. The secondary outcomes were analyzed using participants who were evaluated at the 30-day visit and who were not withdrawn because of failure to improve on study medications or intolerance of study medication.

RESULTS

From November 2010 through January 2015, we enrolled 233 women with acute PID. Of these women, 116 were randomly assigned to receive metronidazole and 117 received placebo. The median age of the participants was 23 years, 59% were black, and 28% were white (Table 1). The median age of women randomized to metronidazole was slightly higher than the median age of women in the placebo regimen (24 vs 22 years, P = .02). A history of chlamydia, gonorrhea, or PID was reported by 140 (60%), 63 (27%), and 68 (29%) participants, respectively, and was similar between the 2 study groups. At enrollment, the clinical presentations were similar between the 2 treatment groups.

Table 1.

Participant Characteristics and Clinical Presentation at Enrollment

Characteristic	Ceftriaxone and Doxycycline Plus Metronidazole (n = 116)	Ceftriaxone and Doxycycline Plus Placebo (n = 117)	P Valuea
Age, y, median (IQR)	24 (21–28)	22 (20–26)	.02b
Race			.29
Black, non-Hispanic	58 (50.0)	66 (56.4)
Black, Hispanic	9 (7.8)	5 (4.3)
White, non-Hispanic	37 (31.9)	28 (23.9)
White, Hispanic	0	1 (0.9)
Other	12 (10.3)	17 (14.5)
Current smoker	77 (66.4)	67 (57.3)	.18
Ever pregnant	87 (75.0)	87 (74.4)	> .99
History of chlamydia	65 (56.0)	75 (64.1)	.23
History of gonorrhea	35 (30.2)	28 (23.9)	.30
History of PID	37 (31.9)	31 (26.5)	.39
Currently using intrauterine device	14 (12.1)	19 (16.2)	.45
New male sexual partner in last 30 d	19 (16.4)	19 (16.2)	> .99
Pelvic pain duration, d			.43
1–2	23 (19.8)	29 (24.8)
≥ 3	93 (80.2)	88 (75.2)
Pelvic pain severityc, mm, mean (SD)	62.4 (24.4)	59.6 (23.0)	.36d
Clinical tenderness score, mean (SD) [8]	9.9 (4.1)	9.9 (4.8)	.94d

Data are presented as no. (%) unless otherwise indicated.

Abbreviations: IQR, interquartile range; PID, pelvic inflammatory disease; SD, standard deviation.

^aP value from Fisher exact test unless otherwise noted.

^bP value from Mann-Whitney U test.

^cVisual Analogue Scale (0–100 mm).

^dP value from Student t test.

Clinical Outcomes

Two hundred eight (89.3%) women returned for the primary outcome visit to assess clinical improvement at 3 days after enrollment (median, 3; interquartile range [IQR], 3–5) (Figure 1). The proportion of women reporting pelvic pain and the severity of pain were similar between the 2 study groups (Table 2). Overall, 91.3% of the participants who returned for evaluation had clinical improvement. Clinical improvement was similar in the women receiving metronidazole and those randomized to placebo, both in the per protocol and intent-to-treat analyses.

Table 2.

Clinical Outcomes at 3-Day Follow-up Visit

Outcome	Standard Plus Metronidazole	Standard Plus Placebo	P Valuea
	(n = 116)	(n = 117)
Clinical improvement, intent-to-treat	96 (82.8)	94 (80.3)	.74
	n = 104	n = 104
Clinical improvement, per protocol	96 (92.3)	94 (90.4)	.81
Pelvic pain	40 (38.5)	39 (37.5)	> .99
Pelvic pain severity, mm, mean (SD)b	21.9 (24.7)	17.3 (23.0)	.17c
Difference from enrollment, mean (SD)	−39.8 (30.9)	−42.4 (28.2)	.53c
Clinical tenderness score, mean (SD) [8]	3.4 (3.3)	3.1 (3.4)	.62c
Difference from enrollment, mean (SD)	−6.6 (4.1)	−6.8 (5.2)	.78c

Data are presented as no. (%) unless otherwise indicated.

Abbreviation: SD, standard deviation.

^aP value from Fisher exact test unless otherwise noted.

^bVisual Analogue Scale (0–100 mm).

^cP value from Student t test.

Figure 1.

Study screening, randomization, and follow-up. *Participants did return for 30-day evaluation. Abbreviations: ITT, intent-to-treat; PID, pelvic inflammatory disease.

Thirteen women (8 receiving metronidazole and 5 randomized to placebo) failed to respond to their assigned treatment and were either hospitalized for parenteral antibiotics or elected alternate antimicrobial therapy. One woman in each arm was determined to have another cause of pelvic pain after enrollment (ovarian cyst, endometriosis).

One hundred eighty four (79%) women continued on study medication and returned for the final assessment at a median 31 days from enrollment (IQR, 29–33) (Table 3). Improvement in pelvic pain was similar between the 2 treatment groups. Pelvic organ tenderness, however, was still present in 20% of women who did not receive metronidazole, compared to 9% of women randomized to metronidazole (P < .05). The clinical cure rates were similar in women randomized to metronidazole compared to women in the placebo-containing treatment arm (97% vs 90%, P = .38).

Table 3.

Clinical Outcomes at 30 Days

Outcome	Standard Plus Metronidazole (n = 90)	Standard Plus Placebo (n = 94)	P Valuea
Clinical cure, No. (%)	87 (96.7)	85 (90.4)	.13
Pelvic pain, No. (%)	8 (8.9)	11 (11.7)	.63
Pelvic pain severityb, mm, mean (SD)	4.0 (9.3)	4.2 (8.8)	.83c
Difference from enrollment, mean (SD)	−58.5 (24.5)	−56.6 (22.7)	.60c
Pelvic tenderness, No. (%)	8 (8.9)	19 (20.2)	.037

Abbreviation: SD, standard deviation.

^aP value from Fisher exact test unless otherwise noted.

^bVisual Analogue Scale (0–100 mm).

^cP value from Student t test.

Microbiologic Outcomes

At enrollment, cervical C. trachomatis infection was identified in 34 (15%) women, N. gonorrhoeae was detected in 17 (7%), and 41 women (18%) tested positive for M. genitalium (Table 4). Bacterial vaginosis was present in 127 (55%) women, and 20 women (9%) were infected with T. vaginalis. The proportions of women at enrollment with C. trachomatis, N. gonorrhoeae, M. genitalium, T. vaginalis, or bacterial vaginosis were similar between the 2 treatment groups. At the 1-month follow-up visit, 6 of 182 women (3%) were positive for cervical N. gonorrhoeae (3 in each treatment group) and 1 woman in each group (1%) had chlamydia. Even though metronidazole has no activity against M. genitalium, cervical infection following PID treatment was less frequent in women receiving metronidazole than in women in the placebo-containing group (4% vs 14%, P < .05). Bacterial vaginosis (20% vs 54%, P < .001) and T. vaginalis (5% vs 12%, P = .10) were also less prevalent in women who were randomized to metronidazole.

Table 4.

Lower Genital Tract and Endometrial Microorganisms Before and After Treatment

Organism	Standard Plus Metronidazole (n = 116)	Standard Plus Placebo (n = 117)	P Valuea
Lower genital tract
Cervical Chlamydia trachomatis
Enrollment	18 (15.5)	16 (13.7)	.71
30 days	1/90 (1.1)	1/92 (1.0)	> .99
Cervical Neisseria gonorrhoeae
Enrollment	6 (5.2)	11 (9.4)	.31
30 days	3/90 (3.3)	3/92 (3.3)	> .99
Cervical Mycoplasma genitalium
Enrollment	17 (14.7)	24 (20.5)	.30
30 days	4/90 (4.4)	13/92 (14.1)	.039
Vaginal Trichomonas vaginalis
Enrollment	9 (7.8)	11 (9.4)	.82
30 days	4/88 (4.5)	11/92 (12.0)	.10
Bacterial vaginosis (Nugent score ≥ 7)
Enrollment	65 (56.0)	62 (53.0)	.69
30 days	18/88 (20.5)	51/92 (54.4)	< .001
Endometrium
C. trachomatis
Enrollment	12 (10.3)	12 (10.3)	> .99
30 days	0/84 (0)	1/90 (1.1)	> .99
N. gonorrhoeae
Enrollment	6 (5.2)	6 (5.1)	> .99
30 days	0/84 (0)	0/90
M. genitalium
Enrollment	7 (6.0)	11 (9.4)	.46
30 days	1/84 (1.2)	4/90 (4.4)	.37
Haemophilus influenzae
Enrollment	1 (0.9)	5 (4.3)	.21
30 days	0/84 (0)	2/90 (2.2)	.50
Gardnerella vaginalis
Enrollment	24 (20.7)	31 (26.5)	.36
30 days	10/84 (11.9)	30/90 (33.3)	.001
Atopobium vaginae
Enrollment	13 (11.2)	13 (11.1)	> .99
30 days	2/84 (2.4)	13/90 (14.4)	.006
Anaerobic gram-negative rods
Enrollment	12 (10.3)b	10 (8.5)c	.66
30 days	4/84 (4.8)	11/90 (12.2)	.11
Anaerobic gram-positive cocci
Enrollment	11 (9.5)d	12 (10.3)e	> .99
30 days	6/84 (7.1)	10/90 (11.1)	.44
A. vaginae and/or anaerobic gram-negative rods and/or anaerobic gram positive cocci
Enrollment	20 (17.2)	22 (18.8)	.86
30 days	7/84 (8.3)	19/90 (21.1)	.020

Data are presented as no. (%). Endometrial samples were not obtained at the 30 day visit from 10/184 women (received non-study antibiotics for another infection following enrollment, refused endometrial sampling, inadequate sampling at either visit)

^aP value from Fisher exact test.

^bIncludes Dialister microaerophilus, Porphyromonas asaccharlytia (n = 2), Porphyromonas uenonis, Prevotella amnii, Provotella bivia, Prevotella denticola, Prevotella disiens, Prevotella massiliensis, Prevotella melaninogenica, Prevotella timonensis, Prevotella novel species group 1.

^cIncludes Dialister microaerophilus, Porphyromonas asaccharlytia, Porphyromonas bennonis, Prevotella amnii, Provotella bivia, Prevotella disiens, Prevotella massiliensis, Prevotella melaninogenica, Prevotella timonensis, Bacteroides ureolyticus.

^dIncludes Anaerococcus species, Anaerococcus tetradius, Anaerococcus vaginalis, Finegoldia magna, Gemella haemolysans, Gemella species, Peptoniphilus harei, Peptoniphilus ivorii, Peptoniphilus lacrimalis, Peptoniphilus species, Petostreptococcus anaerobius.

^eIncludes Anaerococcus tetradius, Anaerococcus vaginalis, Anaerococcus hyrdogenalis, Facklamia hominis, Finegoldia magna, Peptoniphilus harei, Peptoniphilus lacrimalis, Peptoniphilus species, Peptococcus niger, novel anaerobic gram positive coccus (n = 3).

Culture cultivation of the endometrial aspirate was available from 174 women at the 30-day follow-up visit (Table 4). Women who were randomized to PID treatment with metronidazole were less likely to have recovery of Atopobium vaginae, anaerobic gram-negative rods or anaerobic gram-positive cocci from the endometrium following treatment compared to women who received placebo (8% vs 21%, P < .05). Similarly, recovery of Gardnerella vaginalis from the endometrium was also less frequent in women receiving metronidazole (12% vs 33%, P = .001). Each case of persistent anaerobes in the endometrial samples from women receiving only ceftriaxone and doxycycline was reviewed. In 1 case, an isolate of Prevotella timonensis recovered at enrollment persisted after treatment even though the isolate had intermediate susceptibility to ceftriaxone (minimum inhibitory concentration [MIC] >16) and doxycycline (MIC = 8). In 4 additional cases, Anaerococcus tetradius, Peptoniphilus harei, and Prevotella amnii persisted despite susceptibility to both ceftriaxone and doxycycline at baseline and follow-up, suggesting poor clearance of these anaerobic microorganisms without metronidazole. In 1 additional patient, A. vaginae was recovered at follow-up, whereas no persistent A. vaginae was recovered from those randomized to metronidazole.

Following treatment, none of the participants randomized to metronidazole tested positive for C. trachomatis or N. gonorrhoeae in the endometrium. Only 1 woman randomized to the placebo arm was positive for C. trachomatis in the endometrium following treatment. Mycoplasma genitalium was identified in the endometrium at the 30-day visit in only 1 woman (1%) receiving metronidazole and in 4 (4%) women in the placebo arm (P = .37).

Adherence and Tolerability

Adherence with medication, defined as using 75% of all tablets, was similar between the 2 treatment groups (84% in the metronidazole arm and 82% in the placebo arm, P = .85). Adverse events were reported by 104 women (89.7%) assigned to metronidazole and 94 women (80.3%) in the placebo group (P = .07; Table 5). Gastrointestinal symptoms (eg, nausea, vomiting, diarrhea) were frequently reported but were similar in the 2 groups. Vulvovaginal candidiasis was more common among women receiving metronidazole (15.5% vs 6.0%, P = .02). Serious adverse events occurred in 7 women (6.0%) in the metronidazole group and 5 women (4.3%) receiving placebo. Eleven of the 12 serious adverse events were hospitalizations for inpatient treatment. One patient in each group required a change in antimicrobial therapy due to intolerance of study medications. Five women in the metronidazole arm and 4 women receiving placebo were withdrawn for other reasons: 5 women were newly pregnant following study entry, 2 women stopped their medications because of adverse events and were withdrawn (1 for vomiting and 1 for diarrhea and vertigo), 1 participant declined therapy after enrollment, and 1 was withdrawn because she was incarcerated.

Table 5.

Incidence of Nonserious Adverse Events, by Study Arm

Adverse Event	Standard Plus Metronidazole (n = 116)	Standard Plus Placebo (n = 117)	P Valuea
Participants with at least 1 AE	104 (89.7)	94 (80.3)	.07
Constitutional
Headache	12 (10.3)	10 (8.5)	.66
Gastrointestinal complaint, any	93 (80.2)	86 (73.5)	.28
Nausea	58 (50.0)	55 (47.0)	.69
Vomiting	40 (34.5)	37 (31.6)	.68
Diarrhea	42 (36.2)	35 (29.9)	.33
Abdominal pain	12 (10.3)	9 (7.7)	.50
Altered taste	8 (6.9)	5 (4.3)	.41
Genitourinary complaint, any	55 (47.4)	48 (41.0)	.36
Vulvovaginal candidiasis	18 (15.5)	7 (6.0)	.02
Bleeding abnormality	11 (9.5)	10 (8.5)	.82
Genital irritation	13 (11.2)	5 (4.3)	.053
Vaginal itching	11 (9.5)	7 (6.0)	.34
Pelvic pain	4 (3.4)	11 (9.4)	.11
Vaginal discharge, abnormal	10 (8.6)	4 (3.4)	.11
Bacterial vaginosis, symptomatic	4 (3.4)	8 (6.8)	.38
Nervous system
Paresthesia/numbness of extremities	6 (5.2)	5 (4.3)	.77
Respiratory
Upper respiratory infection	10 (8.6)	4 (3.4)	.11
AE severity (any body system)
Grade 1	96 (82.8)	87 (74.4)	.15
Grade 2	53 (45.7)	45 (38.5)	.29
Grade 3	6 (5.2)	3 (2.6)	.33
AE relatedness to study product (any body system)
Related	95 (81.9)	84 (71.8)	.09
Not related	63 (54.3)	65 (55.6)	.90

Data are presented as no. (%). Each participant contributes only 1 observation per category.

Abbreviation: AE, adverse event.

^aP value from Fisher exact test.

Discussion

In this randomized, double-blind, placebo-controlled trial of PID treatment, women receiving oral metronidazole in addition to ceftriaxone and doxycycline had fewer endometrial anaerobic organisms and reduced cervical M. genitalium at 1 month after treatment than women treated with ceftriaxone and doxycycline alone. Moreover, while early clinical response was similar with either PID treatment regimen, fewer women treated with metronidazole had pelvic tenderness at 1 month. Despite the concern for side effects associated with metronidazole use, medication adherence and tolerability were similar between the 2 regimens.

In addition to C. trachomatis and N. gonorrhoeae, the microbiologic etiology of PID includes endogenous anaerobic and facultative microorganisms that ascend from the lower genital tract into the endometrium and associated pelvic structures. Targeted qualitative polymerase chain reaction of endometrial samples has identified an association between bacterial vaginosis–associated bacteria, including A. vaginae, and acute PID [4]. Beyond the pathogenicity of anaerobic microorganisms in PID, the presence of anaerobic organisms following treatment of pelvic infections has been associated with subsequent infectious morbidity when anti-anaerobic therapy has not been administered. For example, the failure to administer antibiotics with activity against anaerobes to women with postpartum endometritis has been associated with abscess formation [12].

Our study demonstrated that enhancing the anaerobic coverage of the CDC-recommended outpatient PID regimen was associated with significantly fewer anaerobic organisms in the endometrium after treatment, including organisms previously associated with acute PID. While our study was not designed to determine whether the presence of anaerobic organisms following treatment is associated with longer-term reproductive complications such as infertility and ectopic pregnancy, the pathogenicity of anaerobic organisms in PID suggests that eradication of anaerobic organisms from the upper genital tract is of importance. It was notable that some Prevotella, Peptoniphilus, and Anaerococcus species persisted in the endometrium in 5 women treated with ceftriaxone and doxycycline alone even though in vitro susceptibility testing suggested that the isolates were susceptible to these drugs. These data suggest that the addition of metronidazole may be needed to effectively eradicate anaerobic pathogens from the upper genital tract of women with PID.

Mycoplasma genitalium is associated with male urethritis, but its role in STD syndromes in women is less well understood [1]. Doxycycline has limited activity against M. genitalium, curing about one-third of infections [13]. In our study, M. genitalium was detected in only 4% of women following treatment with metronidazole, significantly less often than in women not receiving metronidazole. The difference in M. genitalium infection following the 2 treatment regimens, along with the infrequent identification of M. genitalium following treatment with metronidazole, was not anticipated. However, these results are similar to those observed in a retrospective study demonstrating microbiologic cure in 19 of 20 women with M. genitalium–associated PID who were treated with metronidazole and either azithromycin or doxycycline, which was not different than the cure rate of women receiving moxifloxacin [14]. These findings raise important questions about the role of this organism in PID and whether there is a need to address M. genitalium when treating women with current PID treatment that includes metronidazole. Additional studies are needed to understand the role of M. genitalium in the pathogenesis of PID and subsequent reproductive sequelae.

Despite concerns for tolerability of metronidazole and its impact on adherence with a 2-week oral antibiotic regimen, adverse effects and medication adherence were similar in each treatment group. Gastrointestinal side effects were commonly experienced by women whether or not they received metronidazole. While vulvovaginal candidiasis was more common in women receiving metronidazole than in women in the placebo group, bacterial vaginosis at study completion was significantly less common in women receiving metronidazole. Importantly, adherence to oral doxycycline was unaffected by the addition of metronidazole.

A limitation of our study was that the diagnosis of PID was based on minimal CDC criteria and was not confirmed by laparoscopy. However, the clinical criteria are widely accepted and would trigger treatment for PID in clinical practice [5]. A limitation of this and most other PID studies is that we were unable to evaluate long-term reproductive outcomes including recurrent PID, infertility, and ectopic pregnancy. Cultivation-based techniques were used for the identification of anaerobic and facultative organisms in the endometrium rather than nucleic acid–based amplification. This approach was taken because of the limited availability of endometrial tissue and the relative sensitivity of cultivation from normally sterile sites when appropriate anaerobic techniques are employed. It was notable that some novel microorganisms were recovered including Megasphaera phylotype I, microorganisms that had been considered to be noncultivable but had been associated with bacterial vaginosis [15]. Since a goal of the study was to perform antibiotic susceptibility testing of endometrial isolates, the use of cultivation was necessary [11].

Although the microbiologic etiology of PID commonly involves anaerobic organisms, a single dose of a cephalosporin with oral doxycycline provides limited anaerobic activity [5]. Outcomes of PID treatment have historically centered on treating N. gonorrhoeae and C. trachomatis and the resolution of clinical signs and symptoms. Our study is consistent with these goals of PID treatment and demonstrates that the addition of metronidazole to ceftriaxone and doxycycline is associated with significantly less frequent recovery of anaerobic organisms from the upper genital tract, decreased M. genitalium, and improved clinical response compared to ceftriaxone and doxycycline. This study supports the routine use of metronidazole with ceftriaxone and doxycycline for the treatment of women with acute PID and should inform the formulation of future treatment guidelines.

Notes

Acknowledgments. The authors are indebted to the medical staff at the emergency departments at Magee-Womens Hospital and University of Pittsburgh Medical Center Mercy, and to the staff at the Allegheny County Health Department’s Sexually Transmitted Diseases Clinic. The authors thank the staff of the Reproductive Infectious Diseases Clinical Research Unit at Magee-Womens Research Institute, including Tracy Campbell, Jamie Haggerty, Carol Priest, Julie Mckechnie, Abi Jett, and Sarah Thase for their invaluable assistance.

Financial support. This work was supported by the National Institutes of Health (grant number AI084024).

Potential conflicts of interest. H. W. reports grants from SpeeDx Pty Ltd. S. H. reports that testing reagents were supplied by Hologic during the conduct of the study; she also reports grants from Becton-Dickinson Life Sciences and personal fees from Hologic, outside the submitted work. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.