Use of prophylactic antibiotics in women with previable prelabor rupture of membranes

Abstract

OBJECTIVE:

To measure the effect of prophylactic antibiotics given at time of previable prelabor rupture of membranes (PROM) on latency.

METHODS:

Single center, retrospective cohort study of singleton pregnancies with previable (<23 0/7weeks) PROM. Antibiotics were given at clinician discretion. The primary outcome was latency, defined as duration of time between previable PROM and delivery. Secondary outcomes included delivery at ≥ 23weeks, infant survival, and maternal morbidity. Bivariate analysis compared maternal covariates between women who did and did not receive antibiotics. Antibiotic effect on latency was modeled using a Cox proportional hazards ratio.

RESULTS:

213 women with previable PROM were identified; 77 (36%) remained pregnant and thus were included in this analysis. Forty (52%) of 77 received antibiotics. Compared to women who did not receive antibiotics, those who did had PROM at a later median (IQR) estimated gestational age, EGA, (22.2weeks [20.7, 22.5] vs. 19.3weeks [18, 20.7], p < 0.01). Median (IQR) latency was not different between women who did and did not receive antibiotics (2.2 [0.7, 3.9] vs. 1.5 [0.5, 4.6] weeks, p = 0.49). More infants survived to discharge among women who received antibiotics compared to those who did not [17(43%) vs. 3(8%), p < 0.01]. When adjusted for EGA at PROM, antibiotics were associated with longer latency (HR 0.57 [95% CI 0.33, 0.97], p = 0.01). Antibiotic use was not associated with differences in maternal morbidity.

CONCLUSION:

After adjusting for EGA at PROM, antibiotic receipt was associated with longer latency. Larger prospective studies are needed to define the utility of prophylactic antibiotics in previable PROM.

Keywords

Previable PROM preterm premature rupture of membranes latency latency antibiotics

1 Introduction

Preterm prelabor rupture of membranes (PROM) occurs in 3–4.5% of pregnancies and precedes one-third of all preterm births. PROM prior to 24 weeks, or previable PROM, complicates less than 1% of pregnancies [1, 2]. Although rare, previable PROM raises many challenging decisions for the provider and pregnant women. In cases of very early PROM, increased duration of time from rupture to delivery, latency, is associated with a reduction in the risk of severe prematurity complications. Management considerations have become further muddied as neonatal interventions have improved, and the lower limit of viability has decreased to 22-23 weeks gestation [3]. Recent national guidelines regarding the optimal timing of obstetrical intervention to improve neonatal survival including corticosteroids for lung maturity and magnesium for neuroprotection suggest administration a as early as 23-24 weeks gestational age [3, 4].

Data regarding the timing of antibiotics for latency in the previable PROM population remains unclear. Since Mercer et al. noted that antibiotics at the time of PROM between 24 and 32 weeks extended latency, this practice has been almost universally instituted [5]. However, data regarding the effects of antibiotics for women who have PROM at less than 23 weeks is lacking. No randomized trials exist to address this question. Despite lack of evidence, the usage of latency antibiotics in this scenario has increased in the past two decades [4] Retrospective data suggests that antibiotic administration among women with previable PROM is associated with increased neonatal survival and decreased maternal chorioamnionitis [6, 7]. To date, the few studies that assess outcomes or latency in women with previable PROM have described antenatal complications, maternal morbidities, and gestational age at delivery; however there is a dearth of information regarding outcomes based on the initiation and timing of interventions at the time of rupture of membranes [7 –9].

Our objective was to measure the effect of prophylactic antibiotics given at the time of previable PROM (<23 0/7weeks) on latency and infant survival. Our hypothesis was that antibiotic administration at time previable PROM extends latency and increases neonatal survival.

2 Methods

Women were identified by searching our tertiary care center’s perinatal database (which catalogues all deliveries after 20 weeks) as well as institutional discharge data using the ICD-9 codes 634 and 637 (to identify subjects with previable PROM who delivered at less than 20 weeks gestational age, and were coded as spontaneous abortions). In our initial search parameters, we included all women who presented with previable non-iatrogenic PROM between January 1st, 2000 and July 1st, 2015. Women who presented with previable PROM of a singleton gestation defined as rupture of membranes at <23 0/7weeks were eligible for inclusion. Those who elected for immediate termination, delivered within 24 hours of presentation, had a fetus with major anomalies, or who presented with a contraindication to expectant management, such as active intrauterine infection, hemorrhage, or labor, were excluded. Non-iatrogenic PROM was defined as membrane rupture that did not occur 24 hours of chorionic villus sampling or genetic amniocentesis, or at the time of cervical cerclage. In this data, all patients with cerclage had cerclage placed >2 weeks prior to previable PROM for history indicated or ultrasound indicated indications, none for asymptomatic cervical dilatation. Removal of the cerclage at the time of PPROM was done at the discretion of the provider. Patients with incomplete delivery or neonatal data were also excluded. This study was reviewed and approved by the Institutional Review Board at the University of North Carolina at Chapel Hill.

Demographic data, pregnancy information, clinical characteristics, laboratory values at the time PPROM, and delivery data were collected. Hospital course, neonatal outcomes, and morbidities were abstracted from the infants’ charts. Gestational age at rupture of membranes was determined from the woman’s last menstrual period if this was concordant with an ultrasound based on contemporary guidelines [10]. Otherwise, the gestational age was based on ultrasound examination. Prenatal infection was defined as evidence for treatment for one or more of the following at any time during pregnancy: urinary tract infection, trichomonas, bacterial vaginosis, gonorrhea, and chlamydia. Women who elected expectant management were admitted for inpatient observation for 24–48 hours and monitored for evidence of labor or infection.

At our institution, most women with previable PROM were then discharged with close outpatient follow up and readmitted between 23-24 weeks for glucocorticoids and inpatient monitoring. The timing of admission was based on the patient and clinician’s decision as to the gestational age at which neonatal intensive interventions were deemed appropriate. Prophylactic antibiotic administration was at the discretion of the clinician; the majority of antibiotics utilized were combination of ampicillin and azithromycin intravenously for 48 hours, then orally for five days, as is the standard at our institution. There were four women who did not receive this standard dosing; they received azithromycin only (n = 1), only the oral aspect of the above regimen (n = 2), unasyn for 3 days (n = 1). Women were considered to have received prophylactic antibiotics if they received any combination of antibiotics within 48 hours of rupture, which were described as being for the specific purpose of latency or pregnancy prolongation. When cervical lengths and/or amniotic fluid index, AFI, were recorded on admission, this was abstracted. When described as “anhydramnios”, the AFI was defined as zero. If other AFI values were recorded in the medical record, these were also abstracted. However, serial assessment of amniotic fluid volume is not used in the management of previable PROM at our institution, and thus not consistently assessed. Indications for delivery after initial stabilization included placental abruption (defined as vaginal bleeding out of proportion to what would be caused by cervical change or exam), active labor, chorioamnionitis (defined as maternal temperature >38.0°C, fundal tenderness, purulent vaginal discharge, or fetal tachycardia) or reaching 34 weeks gestational age.

The primary outcome was defined as weeks of latency from diagnosis of PROM until delivery. Secondary outcomes included delivery at >23 weeks, survival to hospital discharge, admission to the neonatal intensive care unit, and maternal morbidity. A composite for maternal morbidity included maternal sepsis, postpartum endometritis, maternal intensive care unit (ICU) admission, acute renal insufficiency (serum creatinine >1.2), need for dilation and curettage or hysterectomy after delivery, uterine rupture, deep vein thrombosis, pulmonary embolus, need for blood transfusion, and/or need for readmission. There were no cases of maternal death.

Stata software (version 14.0; Stata Corporation, College Station, TX) was used to perform statistical analysis. Bivariate analysis using χ² test, Fischer’s exact test, or Mann Whitney-U as appropriate compared maternal characteristics between women who did and did not receive antibiotics. Statistical significance was defined as p < 0.05. The effect of antibiotics on latency was modeled using Cox regression model to calculate a hazards ratio for delivery following PROM. A multivariable logistic regression model measured the association between maternal receipt of antibiotics and infant survival. In both models, we adjusted for gestational age at membrane rupture a priori. Maternal characteristics that differed between groups in bivariate analysis at p < 0.1 (including prior preterm birth, mode of delivery, and cerclage) were entered into a multivariable regression model and removed in a stepwise fashion to include only those who remained significant at p < 0.20.

3 Results

From 2000–2015, 213 women with PROM at less than 23 weeks were identified, and 77 (36%) met inclusion criteria (Fig. 1). Of those excluded, 64 (24%) elected for immediate termination, 10 (3.7%) had a fetus with major anomalies and 47 (18%) presented with a contraindication to expectant management (e.g. fever, significant vaginal bleeding, or progressive labor). Of the 77 women included, 40 (52%) received prophylactic antibiotics and 37 (48%) did not.

Fig.1

Study population.

Table 1 shows maternal characteristics of women who did and did not receive prophylactic antibiotics. Only history of prior PTB significantly differed between groups. Cervical dilation at presentation and white blood cell count at the time of membrane rupture were also not significantly different between the two groups (Table 2). In contrast, women who received latency antibiotics were more likely to have had a cerclage placed earlier in the pregnancy and less likely to deliver by dilation and evacuation, (Table 2). The median gestational age of membrane rupture in women who received antibiotics was 22.2 weeks (interquartile range (IQR) 20.7, 22.5) compared to 19.3 weeks (IQR 18, 20.7) in the group that did not receive antibiotics, p < 0.01).

Table 1

Maternal demographic characteristics

	Prophylactic antibiotics given n = 40 (%)	No antibiotics given n = 37 (%)	P-value
Primigravid	7 (18)	10 (27)	0.31
Black	24 (62)	18 (49)	0.32
Maternal age, (IQR)	29.5 (24.5, 33.5)	27 (22, 33)	0.42
Insurance: Medicaid/uninsured	23 (59)	16 (44)	0.21
Smoking (current)	9 (24)	7 (20)	0.70
Prior PTB	18 (69)	8 (31)	0.05
Prenatal infection^a	20 (50)	18 (49)	>0.99
Asthma	4 (10)	8 (23)	0.13
cHTN	5 (13)	6 (17)	0.60
Other drugs	4 (10)	5 (14)	0.73

Abbreviations: IQR interquartile range; PTB preterm birth; HPV human papilloma virus; cHTN chronic hypertension. P value is chi-squared, Fisher’s exact or Mann-U Whitney, as appropriate. ^aPrenatal infection includes one or more of the following: urinary tract infection, trichomonas, bacterical vaginosis, gonorrhea, chlamydia.

Table 2

Admission and delivery details

	Prophylactic antibiotics given n = 40 (%)	No antibiotics given n = 37 (%)	P-value
Positive GBS on admission	4 (10)	7 (22)	0.20
Gestational age at ROM, wks (IQR)	22.2 (20.7, 22.5)	19.3 (18, 20.7)	<0.001
Dilatation, cm, (IQR)	0.75 (0, 2.75)	0.5 (0, 2.5)	0.81
Amniotic fluid index, cm, (IQR)^a	1 (0, 3.2)	0 (0,0)	0.013
White blood cell count at ROM, (IQR)	10.7 (9.5, 12.8)	13.9 (9.5, 15.4)	0.26
White blood cell count at delivery, (IQR)	15.0 (11.8, 18.1)	14.3 (9.8, 18)	0.44
Cerclage placed previously in pregnancy	9 (23)	1 (2.7)	0.02
Betamethasone given^b	22/28 (79)	8/9 (89)	0.66
Infant sex (male)	25 (64)	15 (54)	0.45
Chorioamnionitis	17 (43)	17 (46)	0.76
Delivery route			0.01
Cesarean	13 (32)	6 (16)
Vaginal	27 (68)	25 (68)
Dilation and evacuation	0	6 (16)

Abbreviations: GBS Group B Strep; ROM Rupture of membranes; IQR Interquartile range; P value is chi-squared, Fisher’s exact, or Mann-U Whitney, as appropriate. ^aData represents 12/40 (32%) women who did not receive latency antibitoics, and 25/37 (63%) women who did. ^bAt the time was readmitted when considered viable. Note, this was not at the time of membrane rupture.

Table 3 shows maternal and infant outcomes. Median latency was not significantly different between women who did and did not receive antibiotics. However, compared to women who did not receive antibiotics, those who did delivered at a later gestational age, delivered heavier infants, were more likely to deliver at >23 weeks, and were more likely to have their infant survive to hospital discharge. No significant differences were identified in maternal morbidity.

Table 3

Neonatal and maternal outcomes

	Prophylactic antibiotics given n = 40 (%)	No antibiotics given n = 37 (%)	P-value
Latency, wks, (IQR)	2.2 (0.7, 3.9)	1.5 (0.5, 4.6)	0.49
Maternal morbidity	4 (10)	5 (14)	0.73
Delivery gestational age, wks, (IQR)	23.9 (22.9, 25.5)	21.6 (22.9, 25.5))	<0.001
Birthweight, g, (IQR)	666 (529, 830)	457 (320, 691)	0.0024
Delivery >23/0 weeks	29 (73)	9 (24)	<0.001
Survival to hospital discharge	17 (43)	3 (8.1)	0.001
NICU admission	25 (63)	9 (24)	0.001

Abbreviations: IQR Interquartile range, NICU Neonatal intensive care unit.

Figure 2 shows the Cox proportional regression analysis and hazard ratio for time from ROM to delivery, adjusting for gestational age at ROM (Fig. 2). Antibiotic exposure was associated with longer latency (p = 0.01). The hazard ratio was 0.57 (95% confidence interval 0.33, 0.97). Thus, after adjusting for gestational age at ROM, at any point in time after ROM, women who received antibiotics were 43% less likely to deliver compared to women who did not receive antibiotics. In the multivariable regression model assessing factors associated with survival of the infant to hospital discharge, infants of women who received latency antibiotics were 5 times more likely to survive when adjusting for gestational age at membrane rupture and prior preterm birth (Table 4).

Fig.2

Cox proportional hazards regression (controlling for gestational age at membrane rupture). Legend: Latency hazard ratio: 0.57 (0.33, 0.97), Gestational age at membrane rupture (wks) hazard ratio: 1.2 (1.1, 1.4), p = 0.01.

Table 4

Multivariate logistic regression model for infant survival to hospital discharge among women with previable PPROM, 2000–2015^a

	aOR	95% CI	p-value
Prophylactic antibiotics	5.00	1.14, 22.1	0.033
Gestational age at time of membrane rupture	1.30	0.91, 1.88	0.152

^aOther variables considered in model include: cerclage, prior preterm delivery, and mode of delivery.

4 Conclusion

In this single center retrospective cohort study, we found that receiving antibiotics within 48 hours of previable PROM was associated with increased latency when adjusted for gestational age at the time of ROM. The finding in this study is consistent with the association of PROM antibiotics in populations with PROM after 24 weeks and is a potentially significant finding if an increase in latency is associated with improved neonatal outcomes [5]. Unfortunately, our study is not powered to detect statistically differences in maternal or infant morbidity. Because this is a retrospective analysis, there were important differences between women who did and did not receive antibiotics. Compared to women who did not receive antibiotics, women who did receive antibiotics were at a later gestational age at the time of rupture, and gestational age is an important predictor of neonatal outcomes. After controlling for gestational age at rupture and prior preterm delivery, we found that receipt of latency antibiotics was associated with a five-fold increased odds of survival at hospital discharge. We found no difference maternal morbidity among women who received antibiotics compared to those who did not.

Although no previous studies have assessed the effects of latency antibiotics at the time of rupture specifically in the previable PROM population, the practice of giving antibiotics at the time of membrane rupture in previable gestational ages has been described. Dinsmoor et al. noted that 27 of 43 (63%) women with previable PROM received antibiotics at the time of rupture; though this study did not assess the impact of antibiotics on latency [8]. In a recent review, Waters et al. found that the percentage of women with previable PROM who received “antibiotics” had increased from 0% in 1984 to 97% in 2008. Some data suggest that antibiotic therapy in the setting of previable PROM may be associated with improved outcomes, though the regimens (dosing and duration) are highly variable. Xiao et al. report antibiotics to be positively associated with neonatal survival in after PROM before 25 weeks (p = 0.05) [7]. Of note, Xiao et al. did not specify the timing, dosage, or duration of antibiotics used [7]. Grisaru-Granovsky et al., compared neonates who survived after previable PROM to those who did not, and while not statistically significant, 100% of those who survived received antibiotics, compared to 82.5% of non-survivors [6]. However, in contrast to the above studies, Hunter et al., in a large retrospective cohort study with 106 women with previable PROM, antibiotics (dose and regimen not specified) were also not associated with improved survival [11]. Though these data suggest that antibiotics may improve neonatal outcomes, varying dose and duration regimens and incomplete information on regimens used make drawing any conclusions difficult.

This study has many strengths. To our knowledge, this is the only study to specifically assess effects of latency antibiotics at the time of midtrimester PROM, not only on fetal and neonatal outcomes, but also on maternal outcomes. Another strength is the size of this cohort. Other authors have published studies on outcomes in previable PPROM ranging in size from 23–57 women [7 , 13]. Additionally, the two groups were similar in factors that could potentially effect latency duration and bias our results including race, maternal age, and insurance status. GBS carriage, tobacco use, cervical dilation at time of rupture, and white blood cell count at time of rupture.

A clear limitation was selection bias, as women with rupture at a later gestational were more likely to receive antibiotics. The differences in gestational age at delivery, birth weight and neonatal outcomes as also likely confounded by this as well. In order to address this, we controlled for gestational age at rupture in the statistical analyses. Due to the retrospective nature of this study, we were limited by sample size, outcome severity, and available data in the medical record. However, even with these adjustments, the wide confidence intervals indicated that there is some imprecision in our data. Also, given the high rates of perinatal mortality, we may have been underpowered to show a difference in some of our morbidity variables. Additionally, data analysis with the results stratified by gestational age at the time of rupture could provide valuable insight. Data on the amniotic fluid levels at the time of rupture was not available in over half of patients in this study, nor was there data regarding AFI during the latency period or at delivery for the majority of women in the study. A finding of anhydramnios might have affected the clinician’s decision to administer antibiotics. In addition, initial oligohydramnios following PPROM has been associated with shorter latency and thus AFV may be an un-measurable confounder in this study [14]. However, with data on such a limited number of our participants it is difficult to draw conclusions regarding a link between antibiotic administration and AFI. Similarly, with such as large amount of missing data, we could not include this data point in our survival models, but recognize that AFI is an important element in neonatal survival and may have impacted our findings [15]. Finally, our labor and delivery is staffed by resident physicians, maternal fetal medicine fellows and attending physicians who are general obstetricians and perinatologists, which may have limited provider bias by level of training and sub-specialty. Given the variety of providers a woman may interact in our labor & delivery, we could not control for provider type or level of training. Reluctance to administer antibiotics for previable PROM may lie in the potential for alternation in normal microbiological flora or select resistant micro-organisms that could result in a reduced latency or higher rates of neonatal sepsis. While this study is limited in the ability to provide microbiological data and is not powered to determine differences in neonatal sepsis, there is no evidence that the administration of PROM antibiotics in this setting was associated with shorter latency or increased perinatal morbidity and again reflects the experience of the use of PROM antibiotics in the setting PROM after 24 weeks of gestation.

Although the effect size of latency antibiotics in this study was modest and the sample size in this study small, we did see a statistically significant difference in latency with antibiotic administration and odds of survival to hospital discharge. There was no difference in maternal morbidity between the groups. Consistent with effects of prophylactic antibiotics seen in women with PROM after 24 weeks, exposure to antibiotics at the time of previable PROM is associated with increased latency when adjusting for EGA at PROM. Based on our work and that of others, equipoise exists that justifies a larger, multi-center prospective trial with stratification for gestational age at PROM to ultimately determine if PROM antibiotics prior to 23 weeks gestational age improves neonatal outcomes.

Disclosure statement

The authors have nothing to discloserior presentation: This data was presented as an oral presentation at the Infectious Disease Society of Obstetrics and Gynecology in August 2015, Portland Oregon.

Source

This study was performed at the University of North Carolina at Chapel Hill.

Footnotes

Acknowledgments

None.

References

Kilbride

, Thibeault

. Neonatal complications of preterm premature rupture of membranes. Pathophysiology and management. Clin Perinatol 2001;28(4):761–85.

Yeast

. Preterm premature rupture of the membranes before viability. Clin Perinatol 2001;28(4):849–60.

American College of O, Gynecologists, Society for Maternal-Fetal M. ACOG Obstetric Care Consensus No. 3: Periviable Birth. Obstet Gynecol 2015;126(5):e82–94.

Waters

, Mercer

. The management of preterm premature rupture of the membranes near the limit of fetal viability. Am J Obstet Gynecol 2009;201(3):230–40.

Mercer

, Miodovnik

, Thurnau

, et al. Antibiotic therapy for reduction of infant morbidity after preterm premature rupture of the membranes. A randomized controlled trial. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. JAMA 1997;278(12):989–95.

Grisaru-Granovsky

, Eitan

, Kaplan

, Samueloff

. Expectant management of midtrimester premature rupture of membranes: A plea for limits. J Perinatol 2003;23(3):235–9.

Xiao

, Andre

, Lacaze-Masmonteil

, Audibert

, Zupan

, Dehan

. Outcome of premature infants delivered after prolonged premature rupture of membranes before 25 weeks of gestation. Eur J Obstet Gynecol Reprod Biol 2000;90(1):67–71.

Dinsmoor

, Bachman

, Haney

, Goldstein

, Mackendrick

. Outcomes after expectant management of extremely preterm premature rupture of the membranes. Am J Obstet Gynecol 2004;190(1):183–7.

Manuck

, Eller

, Esplin

, Stoddard

, Varner

, Silver

. Outcomes of expectantly managed preterm premature rupture of membranes occurring before 24 weeks of gestation. Obstet Gynecol 2009;114(1):29–37.

10.

Committee opinion no 611: Method for estimating due date. Obstet Gynecol 2014;124(4):863–6.

11.

Hunter

, Byrnes

, Nathan

, Gill

, Pennell

. Factors influencing survival in pre-viable preterm premature rupture of membranes. J Matern Fetal Neonatal Med 2012;25(9):1755–61.

12.

Falk

, Campbell

, Lee-Parritz

, et al. Expectant management in spontaneous preterm premature rupture of membranes between 14 and 24 weeks’ gestation. J Perinatol 2004;24(10):611–6.

13.

Muris

, Girard

, Creveuil

, Durin

, Herlicoviez

, Dreyfus

. Management of premature rupture of membranes before 25 weeks. Eur J Obstet Gynecol Reprod Biol 2007;131(2):163–8.

14.

Ekin

, Gezer

, Taner

, Ozeren

. Perinatal outcomes in pregnancies with oligohydramnios after preterm premature rupture of membranes. J Matern Fetal Neonatal Med 2015;28(16):1918–22.

15.

Cavoretto

. Prediction of pulmonary hypoplasia in mid-trimester preterm prelabor rupture of membranes: Research or clinical practice? Ultrasound Obstet Gynecol 2012;39(5):489–94.