Chorioamnionitis, or intraamniotic infection, occurs in 1% to 5% of term pregnancies and may complicate up to 25% of cases of preterm labor.1,2 The most extensively tested antibiotic regimen for treating chorioamnionitis is the combination of ampicillin, 2 g intravenously every 6 hours, or penicillin, 5 million units intravenously every 6 hours, plus gentamicin, 1.5 mg/kg every 8 hours.3 If a patient requires a cesarean delivery, a drug to provide enhanced anaerobic coverage, such as clindamycin or metronidazole, should be added.3,4
For the patient diagnosed with chorioamnionitis during labor, many authors have advocated continuing parenteral antibiotic therapy until the patient has been afebrile and asymptomatic for 24 to 48 hours.3,5 However, this recommendation is based primarily on expert opinion, and whether this postpartum antibiotic therapy is necessary recently has been challenged.6,7
Chapman and Owen6 performed a randomized trial, in women delivered vaginally, of a single dose of cefotetan for postpartum treatment of women diagnosed with chorioamnionitis intrapartum. In their study, the single-dose group had a shorter median interval to discharge. In addition, the rate of treatment failure was statistically similar between groups, despite being almost three times as high in the single-dose group (3.7% versus 11%; P = .27). At least one clinical trial has been published that evaluated giving no postpartum antibiotics to women with chorioamnionitis delivered via cesarean. Turnquest et al7 reported a rate of “endometritis” (treatment failure) of 21.8% in the group receiving no postoperative antibiotics, compared with 14.8% in the group receiving at least 24 hours of postoperative clindamycin and gentamicin (P = .32).
Neither of these clinical trials was sufficiently powered to address treatment failure as the primary outcome variable. The purpose of our study was to evaluate, in a sufficiently powered randomized clinical trial, whether one additional dose of a broad-spectrum combination of antibiotics is sufficient postpartum therapy for women with chorioamnionitis.
MATERIALS AND METHODS
Subject enrollment occurred from December 26, 1999, to March 18, 2003. Women were eligible for inclusion in the study if diagnosed with clinical chorioamnionitis (defined by a temperature of 38.0C or more and one or more of the following findings: maternal heart rate >100 beats per minute, baseline fetal heart rate >160 beats per minute, uterine tenderness, or foul-smelling amniotic fluid), if they were receiving care in the Labor and Delivery Unit at Shands Hospital at the University of Florida, and if the plan of treatment was intravenous antibiotics and delivery. Women were excluded from consideration for the trial if they were allergic to β-lactam antibiotics, were immunocompromised, were at risk for bacterial endocarditis, had received β-mimetic drugs in the preceding 8 hours, or had a concurrent febrile illness (eg, pyelonephritis). The study was conducted in accordance with guidelines established by the University of Florida Health Center Institutional Review Board.
When women were diagnosed with chorioamnionitis, they received intravenous ampicillin, 2 g every 6 hours, and gentamicin, 1.5 mg/kg every 8 hours. This combination of antibiotics is the standard regimen for treating chorioamnionitis at our center. Those women who met the criteria for inclusion in this clinical trial were approached for entry. After providing written informed consent, women were randomized, by selection of the next sealed, opaque envelope containing order sheets, to one of two groups. All women continued the above regimen of antibiotics until delivery. After delivery, women randomized to the control group (long arm) continued to receive ampicillin and gentamicin according to the above schedule until they were afebrile and asymptomatic for 24 hours. Women randomized to the study group (short arm) received only the next scheduled dose of each drug. If delivered via cesarean, all patients received clindamycin, 900 mg intravenously, at the time of umbilical cord clamping. The control group continued receiving clindamycin every 8 hours until afebrile and asymptomatic for 24 hours. Like women delivering vaginally, this group of subjects also received ampicillin and gentamicin until afebrile and asymptomatic for 24 hours. Women in the study group who were delivered via cesarean received only the initial dose of clindamycin and the first postpartum dose each of ampicillin and gentamicin. A random number generating software program (Research Randomizer; Social Psychology Network, Middleton, CT) was used to assign subjects to groups.
The primary outcome was treatment failure, defined as a single temperature after the first postpartum dose of antibiotics of 39.0C or more or two temperatures 38.4C or more at least 4 hours apart. Patients identified as treatment failures immediately received intravenous ampicillin, 2 g every 6 hours, gentamicin, 7 mg/kg ideal body weight every 24 hours, and metronidazole, 500 mg every 12 hours. This regimen was continued until the woman had been afebrile and asymptomatic for at least 24 hours. Blood cultures and other diagnostic tests were performed only as deemed clinically indicated by the treating physician. Secondary outcome variables included number of doses of antibiotics, duration of hospital stay, and infection-related complications, such as wound infection, pelvic abscess, septic pelvic vein thrombophlebitis, and Clostridium difficile colitis.
Subjects' medical records were reviewed no less than 30 days after hospital discharge. Demographic data, specific information regarding the course of labor, and outcome variables were entered into a relational database (Access; Microsoft Corporation, Redmond, WA). All data management and analysis were performed using SAS 8.0 (SAS Institute, Cary, NC). All tests of significance were 2-tailed and used an α level of .05. Categoric data were analyzed by the uncorrected χ2 and Fisher exact test, as appropriate. Continuous data were analyzed by the unpaired Student t test. For our power analysis, we anticipated a treatment failure rate in the control group of 3.3% for women delivered vaginally and 15% for women delivered via cesarean. If 60% of women delivered vaginally, the treatment failure rate in the control group would be 8%. To detect a 150% increase in rate of treatment failures from 8% in the control group to 20% in the study group, 292 subjects were needed (α = .05; 1 −β = .80).
Analysis of outcome variables was performed by intent to treat. We also performed a per-protocol analysis, excluding subjects who were enrolled in the study despite not having met all of the enrollment criteria or whose antibiotic therapy deviated from that prescribed by randomization.
During the study period, 292 women were enrolled in this clinical trial; 141 of them were randomized to the long arm, and 151 were randomized to the short arm. According to a search of the database maintained by the University of Florida Division of Maternal-Fetal Medicine, 382 women were diagnosed with chorioamnionitis during the study period. Therefore, our sample size represents 76% of the women diagnosed with chorioamnionitis during the study period. No women who met the enrollment criteria were intentionally omitted. We did not formally account for the reasons that these other women were not included in the study. However, these reasons would have included β-lactam allergy or subclinical infection (diagnosed via amniocentesis); the women could have declined to participate; or the women were not approached for consent.
Maternal demographic data are shown in Table 1. The groups were well matched, both for risk factors for chorioamnionitis and for severity of illness. Women admitted to the hospital for expectant management of preterm premature rupture of membranes were eligible for the study if they later developed clinical chorioamnionitis. There were 13 such women (9.2%) in the long arm and eight (5.3%) in the short arm (P = .19).
The results of the intent-to-treat analysis for the primary outcome variable, treatment failure, are listed in Table 2. The proportion of women who were classified as having treatment failures did not differ between groups overall, or within the groups of women who had either vaginal or cesarean delivery.
Table 3 displays the results of the intent-to-treat analysis for secondary outcomes. As anticipated, there were no differences between groups for number of doses of antibiotics received intrapartum, but subjects in the short arm received significantly fewer doses of antibiotics in the postpartum period and overall. In addition, subjects randomized to the short arm of the study had shorter mean hospital stays. Although the number of postpartum days did not differ significantly between groups, subjects in the long arm of the study were discharged more often after postpartum day 2 if delivered vaginally (14.8% versus 6.9%; P = .049) and more often after postoperative day 3 if delivered via cesarean (20.8% versus 10.9%; P = .143). In the entire cohort, there were no cases of urinary tract infection, septic pelvic vein thrombophlebitis, or C difficile colitis. Wound infections occurred in one woman in the long arm and two women in the short arm of the study (P = .99 for both total group and cesarean deliveries only proportion comparisons). There was one case of pelvic abscess; this woman was randomized to the short arm of the study.
Eighteen subjects (6.2%) experienced protocol violations, five in the long arm and 13 in the short arm. Reasons for these protocol violations included: subject had a history of β-lactam allergy but was enrolled in the study and treated with gentamicin and clindamycin (n = 5); failure to receive clindamycin at cord clamping during cesarean delivery (n = 7); enrolled despite only subclinical chorioamnionitis (n = 4); treated with antibiotics intrapartum in addition to ampicillin and gentamicin (n = 1); and concurrent febrile illness (pyelonephritis) (n = 1).
A per-protocol analysis was performed that excluded data from these 18 subjects. In this analysis, treatment failure rates were similar for the long arm (n = 136) and short arm (n = 138), respectively (2.9% versus 4.3%; P = .749). Two of the subjects with infection-related complications in the short arm represented protocol violations. Both of these subjects delivered via cesarean and did not receive a dose of clindamycin at cord clamping. One subject experienced a treatment failure complicated by a pelvic abscess that required a 10-day postoperative hospital stay. With antibiotic therapy (ampicillin, gentamicin, and metronidazole), her infection resolved, and she did not require laparotomy. The other subject sought care at the emergency department on postoperative day 5 with a wound abscess. Her wound was opened and packed, and she was treated as an outpatient with a 7-day course of cephalexin; and her wound healed over 2 to 3 weeks by secondary intention. Both of the wound infections that occurred in subjects who did not represent protocol violations (one in each arm of the study) were cases of wound cellulitis. Neither wound required opening of the skin and subcutaneous tissue; both responded to a 7-day course of cephalexin. The above-described woman with a wound abscess was the only subject to seek care at our clinics or emergency department with a postpartum complication. No subjects sought care from our system in the postpartum period with signs and symptoms consistent with endometritis.
Although routine pathologic examination of the placenta was not part of the study protocol, we recorded the results of placental pathology when this was evaluated for subjects included in this study. Thirty (75%) of 40 subjects whose placentas were analyzed showed histologic chorioamnionitis. It is of note that the clinical history of more than half of these 40 pathology submissions failed to include the fact that the patient had been diagnosed clinically with chorioamnionitis.
There are no universally accepted criteria for diagnosing chorioamnionitis. The term is a histologic one; the clinical diagnosis is usually based on a combination of fever and tachycardia, either of the mother or fetus; uterine tenderness; foul-smelling amniotic fluid; or leukocytosis. We chose to use a temperature of 38.0C or more to define fever in this study. Some authors have advocated a lower threshold of 37.8C.2
Beginning intravenous antibiotic therapy intrapartum, at the time chorioamnionitis is diagnosed, consistently has been shown to reduce rates of neonatal infection.8–10 However, clinical decision making regarding duration of postpartum antibiotic therapy for women with chorioamnionitis has been based on level 3 evidence (ie, expert opinion). We report here an appropriately powered clinical trial that shows that a single postpartum dose of a broad-spectrum combination of antibiotics is sufficient postpartum therapy to prevent treatment failure for women diagnosed with chorioamnionitis intrapartum.
We chose treatment failure as the primary outcome variable for this study. Because we defined this variable on the basis of exceeding a threshold temperature during the postpartum period, the outcome is somewhat arbitrary. However, we defined this outcome in an attempt to give rescue therapy to subjects before, rather than after, development of serious infections or infection-related complications. Even with this rather conservative definition of treatment failure, there was a low proportion of this outcome in both arms of the study.
Recently, Lieberman et al11 reported that intrapartum fever is more common in women receiving epidural analgesia. Those authors postulate that many women diagnosed as having clinical chorioamnionitis instead have “epidural fever.” However, Dashe and colleagues12 evaluated the relationship between epidural analgesia and inflammation of the placenta. They concluded that epidural analgesia is associated with fever during the intrapartum period, but only in the presence of placental inflammation. Although histologic examinations of the placentas of only a minority of the subjects in our study were performed, the fact that the majority of these placentas displayed histologic chorioamnionitis supports the conclusion of Dashe et al that fever reported with epidural analgesia is most often the result of infection and not the analgesia itself.
The two subjects in the short arm of our study who developed complications (one a pelvic abscess and the other a wound abscess) after being delivered by cesarean but failing to receive clindamycin illustrate the importance of anaerobic coverage in patients with chorioamnionitis who are undergoing cesarean delivery. Neither of these women received clindamycin, and therefore, their cases represent study protocol violations. In the per-protocol analysis, the only complications were wound cellulitis in one patient in each group. Although the study was not powered to detect differences in rates of infection-related complications, these results attest to the safety of the short arm regimen in our study, provided that a dose of an anti-anaerobic antibiotic is provided intraoperatively.
This study provides evidence that a single postpartum dose of a combination of broad-spectrum antibiotics is equivalent to a regimen extended until patients are asymptomatic and afebrile for at least 24 hours after delivery. In addition, our data suggest that such an approach might be cost saving. The study was not designed to address cost comparisons. However, women in the short arm of this study received fewer doses of antibiotics, with the difference between groups being solely due to the postpartum period. In addition, although the duration of postpartum hospitalization did not differ between groups, women in the long arm of the study were more often discharged home later than usual (postpartum day 2 if delivered vaginally or postpartum day 3 if delivered via cesarean). We plan to adopt this single postpartum dose regimen as our conventional treatment and to evaluate its clinical effectiveness (as opposed to clinical trial efficacy) in a cohort study. If its effectiveness is confirmed, this regimen should become the standard of care for the postpartum treatment of women with chorioamnionitis.