Het is algemeen bekend dat PPROM een groot risico geeft op perinatale complicaties. In het geval van intra-uteriene en/of neonatale infectie (bij (subklinische) chorioamnionitis) neemt het risico op nadelige maternale en perinatale uitkomsten verder toe. Het is daarom van groot belang om een strategie te zoeken die deze risico’s reduceert.

Het beleid in Nederland wisselt; lokaal (per ziekenhuis) is afgesproken of er wel of niet standaard antibiotica worden toegediend bij PPROM. Er is onduidelijkheid over het gunstige effect, als ook over eventuele nadelige bijeffecten. Mogelijk zijn deze effecten ook nog afhankelijk van de zwangerschapsduur ten tijde van PPROM. De NVOG richtlijn Dreigende vroeggeboorte uit 2012 ziet geen rol voor profylactische toediening van antibiotica bij PPROM.

In de NICE guideline (2015) wordt geadviseerd om zwangeren met PPROM wel profylactisch te behandelen met antibiotica (erytromycine)(NICE, 2015).

Critical outcome measures

Neonatal outcomes

(1) Neonatal morbidity

(1.a) Major cerebral abnormality on ultrasound before discharge

(2) Perinatal mortality

(3) Long term effects on the child

(3.1) Cerebral Palsy, as reported by parents of children at age 7 years

(3) BPD

Important outcome measures

Neonatal outcomes

(4) Perinatal infections

(5) Apgar score < 7 after 5 minutes

(6) Neonatal Intensive Care Unit admission

(7) Prolonged latency/pregnancy duration until birth

(7.1) Birth within 48 hours of randomisation

(7.2) Birth within 7 days of randomisation

(7.3) Birth before 37 weeks of gestation

Maternal outcomes

(8, 9, 10) Allergy, anaphylaxis and sepsis

(11) Chorioamnionitis

(12) Endometritis

(13) Delivery mode

(14) Maternal death

(15) Postpartum haemorrhage

Description of study

The SR by Kenyon (2013) investigated the effect of immediate and long-term effects of administering antibiotics to women with PPROM < 37 weeks. The review included 14 RCTs comparing antibiotics with placebo, and 6 non-placebo controlled RCTs, which contributed data to the outcome perinatal death alone. In two included RCTs a comparison was made between various lengths of antibiotics prophylaxis. This summary is limited to the 20 trials comparing antibiotics with placebo or no treatment, in which 6740 women and their babies were included. Women were recruited between 20 and 37 weeks of gestation. Ten trials tested broad spectrum antibiotics, five tested macrolide antibiotics (erythromycin) and one tested clindamycin and gentamicin. The duration of treatment varied between two doses and 10 days, or until delivery. Both intravenous and oral antibiotics or a combination was prescribed. The primary outcomes were maternal death, serious maternal morbidity (composite of septicemia, need for intensive care, organ failure, need for hysterectomy), perinatal death and perinatal morbidity (defined as neonatal infection including pneumonia, enterocolitis, oxygen treatment until more than 36 weeks post conceptual age and major cerebral abnormality on ultrasound prior to discharge).

Results

Critical outcome measures

N.B. For the Critical outcome measures information was also extracted from the original RCTs.

(1) neonatal morbidity

Defined by the working group as a composite of (1.1) bilateral periventricular leukomalacia (PVL), (1.2) intraventricular haemorrhage (IVH) gr III/IV, (1.3) neonatal insult, (1.4) neonatal encephalopathy. None of the included studies reported this composite outcome. However, the outcome (1.a) Major cerebral abnormality on ultrasound before discharge was reported in 12 included RCTs. The pooled RR (95% CI) was 0.81 (0.68; 0.98) (Figure 1.1 12 trials, 6289 babies).

Figure 1.1 Outcome (1.a) Major cerebral abnormality on ultrasound before discharge; comparison any antibiotic versus placebo

Source: Kenyon (2013): Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(1.1) bilateral PVL

This outcome was not reported.

(1.2) IVH gr III/IV

This outcome was not reported.

(1.3) Neonatal insult

This outcome was not reported.

(1.4) Neonatal encephalopathy

This outcome was only reported by Garcia-Burguillo (1995) reporting 0/30 events in de antibiotic group and 0/30 in the placebo group. The risk ratio could therefore not be estimated.

(2) Perinatal mortality

Perinatal death or death before discharge was reported in 12 individual trials with data for 6301 babies. Perinatal death was comparable between the antibiotics (any type and dose) and placebo arms (Figure 1.2, 12 trials, 6301 babies, RR 0.93, 95% CI 0.76 to 1.14).

Figure 1.2 Outcome (2) Perinatal mortality: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(3) Long term effects in the child

Long term effects in the child were defined by the working group as (3.1) cerebral palsy (CP) and (3.2) bronchopulmonary dysplasia (BPD). The only long term child outcome reported by Kenyon (2013) was Serious childhood disability at 7 years, reported by Kenyon (2008) (n=3171) in a follow-up study of the trial included in the Cochrane SR (Kenyon, 2013) as Kenyon (2001). The reported RR (95% CI) was 1.01 (0.91; 1.12) (Kenyon, 2013).

(3.1) CP

The outcome CP, as reported by parents of children at age 7 years was only reported by Kenyon (2008). The OR (95% CI) was 1.18 (0.77; 1.81) and 0.81 (0.53; 1.24) for erythromycin and co-amoxiclav, respectively.

(3.2) Bronchopulmonary dysplasia (BPD)

This outcome was not reported. However, Kenyon, 2013 did report the outcomes (3.2.1) Oxygen treatment > 36 weeks’ gestational age and (3.2.2) Neonatal oxygenation > 28 days. The first outcome was reported in one of the included studies (Kenyon, 2001), including 4809 babies. The RR (95% CI) was 0.91 (0.70;1.17). Neonatal oxygenation was reported in 3 trials (n=5487). The RR (95% CI) was 0.79 (0.61; 1.03).

Figure 1.3 Outcome (3.2.2) Neonatal oxygenation > 28 days: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

Important outcome measures

Neonatal outcomes

(4) Perinatal infections

Kenyon (2013) reported ‘neonatal infections including pneumonia’. A total of 12 trials, with 1680 babies, were included. The authors reported a significant reduction in children with a neonatal infection whose mothers had received antibiotics (Figure 1.4, 12 trials, 1680 children, RR 0.67, 95%CI 0.52 to 0.85). Heterogeneity was low (I² = 0%).

Figure 2.4 Outcome (4) Perinatal infections: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(5) Apgar score < 7 after 5 minutes

This outcome was not reported.

(6) Neonatal Intensive Care Unit (NICU) admission

4 trials, including 5023 children, reported on NICU admission. NICU admission was comparable between children whose mothers had or had not received antibiotics (Figure 1.5, 4 trials, 5023 children, RR 0.98, 95% CI 0.84 to 1.13). The heterogeneity among studies was high (I² = 78%).

Figure 1.5 Outcome (6) NICU admission: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(7) Prolonged latency/pregnancy duration until birth

The number of days from PPROM to birth was not compared, but the review did report on (7.1) ‘birth within 48 hours of randomisation’ and (7.2) ‘birth within 7 days of randomisation’ and (7.3) ‘birth before 37 weeks’ of gestation’.

(7.1) Birth within 48 hours of randomisation,

In 7 trials, with 5927 children, data were reported on birth within 48 hours of randomisation. A significant difference was found between children born within 48 hours favouring children whose mothers had received antibiotics (Figure 1.6, 7 trials, 5927 children, RR 0.71, 95% CI 0.58 to 0.87).

Figure 1.6 Outcome (7.1) birth within 48 hours of randomisation: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(7.2) Birth within 7 days of randomisation

There were 7 trials (of which 6 also reported birth within 48 hours after randomisation), 5965 children, reporting on birth within 7 days (Figure 1.7, 7 trials, 5965 children, RR 0.79, 95% CI 0.71 to 0.89).

Figure 1.7 Outcome (7.2) birth within 7 days of randomisation: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I2: statistical heterogeneity, CI: confidence interval

(7.3) Birth < 37 weeks of gestation

Trials, including 4931 children were found that reported birth < 37 weeks of gestation. The number of babies born before 37 weeks of gestation was comparable among groups (antibiotics versus placebo) (Figure 1.8, 3 trials, 4931 children, RR 1.00, 95%CI 0.98 to 1.03, I²=0%)

Figure 1.8 Outcome (7.3) birth < 37 weeks’ of gestation: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

Maternal outcomes

(8) Allergy

This outcome was not reported.

(9) Anaphylaxis

This outcome was not reported.

(10) Sepsis

This outcome was not reported.

(11) Chorioamnionitis (clinical and histological)

Chorioamnionitis was defined as ‘infection of the womb’. According to the study descriptions it was defined by most studies as ‘having clinical signs of chorioamnionitis’. Eleven trials, 1559 women, reported on chorioamnionitis. A statistically significant reduction in chorioamnionitis was seen with the use of antibiotics (Figure 1.9, 11 trials, 1559 women, RR 0.66, 95%CI 0.46 to 0.96, I²=45%).

Figure 1.9 Outcome (11) Chorioamnionitis: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(12) Endometritis

This outcome was not reported.

(13) Delivery mode (spontaneous vaginal, instrumental or caesarean section)

The 11 studies (6317 women) only reported on the outcome ‘caesarean section’. The rate of caesarean section was comparable among groups (antibiotics versus placebo) (Figure 1.10, 11 trials, 6317 women, RR 0.96, 95% CI 0.88 to 1.05, I²=0%).

Figure 1.10 Outcome (13) Delivery mode: antibiotics versus placebo in women after PPROM

Source: Kenyon (2013); Z: p-value of the pooled effect, df; degrees of freedom, I²: statistical heterogeneity, CI: confidence interval

(14) Maternal mortality

The outcome ‘maternal death’ did not occur in both groups in 3 trials including 763 women in total.

(15) Postpartum haemorrhage

This outcome was not reported.

Level of evidence of the literature

The level of evidence regarding the outcome measure (1) neonatal morbidity, consisting of 1.1 bilateral PVL, 1.2 IVH gr III/IV, 1.3 neonatal insult and 1.4 neonatal encephalopathy, could not be assessed because outcomes 1.1, 1.2 and 1.3 were not reported by Kenyon (2013) and there were no events for outcome 1.4. For the outcome (1.a) Major cerebral abnormality on ultrasound before discharge the level of evidence started high and was downgraded with one level to a moderate GRADE because of imprecision, as the 95% CI overlapped with the minimal clinically important difference (MCID).

The level of evidence regarding the outcome measure (2) perinatal mortality started high and was not downgraded .

The level of evidence regarding the outcome measure (3) long term effects in the child, consisted of (3.1) cerebral palsy (CP) and (3.2) bronchopulmonary dysplasia (BPD). The level of evidence for outcome (3.1) cerebral palsy started high and was downgraded with two levels to a low GRADE because of imprecision (the lower and upper limits of clinical relevance RR<0.8 and RR> 1.25 were exceeded by the 95% CI), and because of possible attrition bias. The level of evidence for outcome (3.2) BPD started high and was downgraded with one level to a moderate GRADE because of imprecision, as the 95% CI overlapped with the MCID.

The level of evidence regarding the outcome measure (4) perinatal infections started high and was downgraded by one level to a moderate GRADE because of imprecision (the lower limit of clinical relevance RR< 0.8 was exceeded).

Outcome (5) 5 minutes’ Apgar score < 7 was not reported.

The level of evidence regarding the outcome measure (6) Neonatal Intensive Care Unit (NICU) admission started high and was downgraded by one level to a moderate GRADE because of inconsistency.

The level of evidence regarding the outcome measure (7) Prolonged latency/pregnancy duration until birth is divided in (7.1) birth within 48 hours of randomisation, (7.2) birth within 7 days of randomisation and (7.3) birth before 37 weeks of gestation. The level of evidence for outcome (7.1) birth within 48 hours of randomisation started high and was downgraded with one level to a moderate GRADE because of imprecision (the lower limit of clinical relevance RR< 0.8 was exceeded). The results of one small study deviated from the other studies, but because of the limited size of this study, it was decided not to downgrade for inconsistency. The level of evidence for outcome (7.2) birth within 7 days of randomisation started high and was downgraded with one level to a moderate GRADE as well because of imprecision (the lower limit of clinical relevance RR<0.8 was exceeded). The level of evidence for outcome (7.3) birth before 37 weeks of gestation started high and was not downgraded.

Outcome (8) Allergy was not reported.

Outcome (9) Anaphylaxis was not reported.

Outcome (10) Maternal sepsis was not reported.

The level of evidence regarding the outcome measure (11) chorioamnionitis started high and was downgraded by 2 levels to a low GRADE because of inconsistency, possibly due to an unclear definition of chorioamnionitis among studies (e.g. clinical signs or histological result) and because of imprecision (the lower limit of clinical relevance of RR<0.8 was exceeded).

Outcome (12) Endometritis was not reported.

The level of evidence regarding the outcome measure (13) delivery mode (caesarean section) started high and was not downgraded.

The level of evidence regarding the outcome measure (14) maternal mortality started high and was downgraded to a very low GRADE because no maternal deaths occurred in any of the three studies that reported maternal death as an outcome.

Outcome (15) Postpartum haemorrhage was not reported.

A systematic review of the literature was performed to answer the following question:

What are the beneficial/non beneficial effects of antibiotic prophylaxis compared to placebo or no treatment in pregnant women with preterm prelabour rupture of membranes (PPROM) between 20 and 37 weeks on neonatal and maternal outcomes?

P: Pregnant woman with PPROM between 20 and 37 weeks gestational age (without signs of a clinical infection).

I: Antibiotics prophylaxis (any regime).

C: No antibiotics/placebo.

O: Neonatal morbidity: respiratory distress syndrome, bronchopulmonary dysplasia (BPD), periventricular leukomalacia (PVL), IVH gr III/IV, neurologic problems (neonatal insult, cerebral palsy (CP), neonatal encephalopathy), perinatal mortality, long-term outcomes, perinatal infections, Apgar score < 7 after 5 minutes, Neonatal Intensive Care Unit (NICU) admission, prolonged latency/pregnancy duration until birth, maternal allergy, anaphylaxis, maternal sepsis, chorioamnionitis (clinical and histological), endometritis, delivery mode (spontaneous vaginal, instrumental or caesarean section), maternal mortality, postpartum haemorrhage.

Relevant outcome measures

The guideline development group considered the neonatal outcomes: (1) neonatal morbidity (respiratory distress syndrome (RDS), (1.1) bilateral periventricular leukomalacia (PVL), (1.2) IVH gr III/IV, (1.3) neonatal insult, (1.4) neonatal encephalopathy), (2) perinatal mortality, (3) long term outcomes (composite of (3.1) cerebral palsy (CP) and (3.2) bronchopulmonary dysplasia (BPD)) as critical outcome measures for decision making; and (4) perinatal infections, (5) Apgar score < 7 after 5 minutes, (6) Neonatal Intensive Care Unit (NICU) admission, and (7) prolonged latency/pregnancy duration until birth as important outcomes. The maternal outcomes: (8) allergy, (9) anaphylaxis, (10), sepsis, (11) chorioamnionitis (clinical and histological), (12) endometritis, (13) delivery mode (spontaneous vaginal, instrumental or caesarean section), (14) maternal mortality, (15) postpartum haemorrhage were all considered as important outcomes for decision making.

A priori, the working group did not define the outcome measures listed above but used the definitions used in the studies.

For the outcome measures perinatal and maternal mortality any clinical difference was considered as a clinically important difference between groups. For delivery mode and postpartum haemorrhage 10% clinical difference was considered clinically important. For all other outcome measures, the GRADE default - RR< 0.80 or RR>1.25 for dichotomous outcomes and 0.5 standard deviation for continuous outcomes - was taken as a minimal clinically important effect.

Search and select (Methods)

The databases Medline (via OVID) and Embase (via Embase.com) were searched using relevant search terms until July 28th, 2020. The detailed search strategy is depicted under the tab Methods. The systematic literature search resulted in 602 hits. Studies were selected based on the following criteria: (design) RCTs, (P) women with PPROM, (I) antibiotics (any type/dose), (C) placebo or no treatment. 36 studies were initially selected based on title and abstract screening. After reading the full text, one Cochrane systematic review from 2013 was included, which included 22 trials. No studies published after the systematic review were found that could be included.

Results

One systematic review (SR) by Kenyon (2013), including 22 trials, was included in the analysis of the literature. Only one of the included RCTs reported long term outcomes at age 7 years; since not all of the critical outcomes were reported in the SR, the relevant information was extracted from the original RCT (Kenyon, 2008). Important study characteristics and results are summarized in the evidence table. The assessment of the risk of bias is summarized in the risk of bias table.