Discussion
Statement of principal findings
The results of this national, population based study of PPROM before 23 weeks' gestation highlight the diverse perinatal and maternal outcomes of this condition. Maternal sepsis developed in 14% (50/364) of women and two died. Although 26% (54/207) of women with expectantly managed singleton pregnancies had a baby that survived to discharge from hospital, only 18% (38/207) of babies survived without severe morbidity. Most neonatal morbidity and mortality were related to extreme prematurity, secondary to 39% of births within a week of PPROM. Uncertainty exists for perinatal outcomes because 32% of pregnancies were terminated for medical reasons.
Comparison with previous studies
The two maternal deaths gave a rate of 549 per 100 000 maternities with PPROM at 16+0-22+6 weeks’ gestation (95% confidence interval 151 to 1981 per 100 000). This rate is considerably higher than the baseline UK maternal mortality rate of 11 per 100 000 maternities,15 and comparable with the risk of death after admission to hospital with SARS-CoV-2 infection when pregnant (406 per 100 000)16 or having a pulmonary embolism in pregnancy (3413 per 100 000).17
In the past decade, more than 700 women have been included in observational cohorts of PPROM at similar gestational ages with no maternal deaths reported.3 18–24 This finding concurs with no maternal deaths reported in the most recent review on this topic,25 and only one maternal death reported in the largest review, published in 2009, citing a publication from 1988.2 26 The key difference from previous studies is that our cohort was population based because we surveyed all 194 consultant led maternity units in the UK in contrast with previously published studies from five or fewer (often specialised) centres. In accordance with our population based approach, second trimester PPROM associated with maternal sepsis and death has featured in three of the UK's maternal mortality reports in the past decade.14 26 27 Population level data have also identified seven maternal deaths after PPROM at 14+0-24+6 weeks' gestation from 2001 to 2015 in France, giving an estimated risk of death of 45 per 100 000 maternities with this complication.27 Similar causes of death to our cohort were found: six were attributed to sepsis and one to haemorrhage secondary to placenta accreta spectrum. Hence the population based literature suggests that these deaths are not isolated incidents and might have previously been under-reported, potentially because the literature has been based on data from a small number of centres. Also, studies might have tended to preferentially report on cohorts with favourable outcomes, with possible positive publication bias, which is much less likely in population based studies.
The rate of maternal infectious morbidity in previous studies of expectantly managed singleton pregnancies was similar to our value of 12%.3 14 18 In our cohort, maternal sepsis was higher in twin pregnancies, also in agreement with previous work.28 Similar to current clinical expertise and one randomised controlled trial that was ended early, maternal sepsis was higher with cervical cerclage.29 30 Surgery for removal of the placenta is a less well recognised complication of birth after early PPROM but occurred in 20% of our cohort, including four of the five women who became severely unwell. This rate concurs with a cohort with PPROM in mid-trimester in Ireland.3 The combination of requiring surgery for removal of the placenta and sepsis should alert clinicians to the possibility of maternal deterioration.
Even if all women in our study had opted for termination of pregnancy for medical reasons without expectant management, and if termination for medical reasons had avoided all incidences of maternal sepsis after termination was performed, a reasonable expectation is that more than half of the instances of maternal sepsis would still have developed. In our cohort, 10% of women who opted for immediate termination of pregnancy for medical reasons with singleton pregnancies developed sepsis. This finding agrees with a retrospective case control study in the US of 174 pregnances with PPROM at 14+0-22+6 weeks' where expectant management was not associated with an increase in composite morbidity (including sepsis and endometritis) compared with immediate termination of pregnancy for medical reasons.31 We suggest that this finding is because sepsis can develop quickly. Infection could even have contributed to the occurrence of PPROM whereas termination of pregnancy for medical reasons takes time to arrange and perform. Among the 33 women with singleton pregnancies that developed sepsis after initial expectant management, median time from PPROM to a diagnosis of sepsis was two days (interquartile range 1-8) and among the 62 singleton pregnancies that were terminated for medical reasons without expectant management, median time from PPROM to termination of pregnancy for medical reasons was also two days (1-3.5).
The perinatal survival to hospital discharge rate of 26% for expectantly managed singleton pregnancies, and the range of possible survival (17-53%) when pregnancies with termination of pregnancy for medical reasons and unknown outcomes were accounted for, were similar to observational studies over the past decade that reported rates of 17-40% with similar gestational ages of PPROM.18 20 24 32–36 Among the babies that survived to discharge from hospital, 70% (38/54) avoided severe morbidity by our definition, in keeping with recently published observational studies.14 18 24 34 35
The definition of severe neonatal morbidity was chosen for consistency with Kibel et al14 who related neonatal outcomes at hospital discharge after pregnancies complicated by PPROM at 20-24 weeks' gestation to outcomes at a corrected age of 18-21 months. Of 24 babies with grade 3 or 4 intraventricular haemorrhage, requiring oxygen at 36 weeks' postmenstrual age, or with ≥grade 3 retinopathy of prematurity, or a combination of these conditions, eight (30%) had moderate-to-severe morbidity at age 18-21 months. Among 27 babies who were born after pregnancies complicated by PPROM at 20-24 weeks' gestation but who did not have any intraventricular haemorrhage, retinopathy of prematurity, or require oxygen at 36 weeks' postmenstrual age, none had moderate-to-severe morbidity at age 18-21 months.14 Our results could indicate long term moderate-to-severe morbidity in a similar proportion of children, although we did not collect data about retinopathy of prematurity so the burden of severe morbidity might be slightly underestimated.
Women who had expectant management after PPROM at 16+0-17+6 weeks’ gestation and gave birth after 23 weeks' gestation had comparable live births and perinatal morbidity rates with those who had PPROM at 22+0-22+6 weeks’ gestation. This finding could be a result of lack of statistical power because only seven babies survived to discharge after PPROM at 16+0- 17+6 weeks' gestation. Also, when PPROM occurred earlier in pregnancy, possibly more pregnancies with less favourable characteristics had a termination of pregnancy for medical reasons, negating any negative effect of gestational age of PPROM on perinatal morbidity. A study from the Netherlands, however, with a rate of termination of only 2%, also found no difference in perinatal morbidity after PPROM at ≥13-<20 weeks' gestation (n=21) compared with ≥20-<24 weeks' gestation (n=41).34 We suggest that the effect of gestation on perinatal morbidity after PPROM needs further evaluation.
Median gestational age at birth of babies surviving to discharge from hospital was 29+4 weeks, and median length of hospital stay after birth for surviving babies was 59 days (interquartile range 17-100). This length of stay in hospital is shorter than in recent studies from Australia20 (median 76 days, interquartile range 44-111) and Japan37 (mean 155 days, standard deviation 53), but is still a substantial amount of time and likely to have a major effect on the whole family in the medium term. We suggest this information should be included in patient counselling.
As gestational age at birth advanced, the possibility of live birth and perinatal survival to discharge from hospital improved, as expected.2 Among those born after 34 weeks' gestation, however, two babies (2/15, 13%) had severe morbidity, highlighting the complexities of these cases and the need for ongoing multidisciplinary team care, including neonatologists, even at relatively advanced gestations.
Babies of dichorionic-diamniotic multiple pregnancies with PPROM seemed to have comparable pregnancy outcomes with singletons, similar to previous studies.38 Monochorionic-diamniotic twin pregnancies had lower perinatal survival, but 60% of these pregnancies also had pathologies of monochorionicity, such as twin-to-twin transfusion syndrome and selective growth restriction. Therefore, the pathologies unique to monochorionic pregnancies are likely to be important contributors to mortality in these instances.
Strengths and limitations
The UKOSS infrastructure allowed us to conduct a large population based study of PPROM before 23 weeks' gestation. Unavoidable uncertainty exists, however, for perinatal outcomes because 31% (113/364) of the population opted for termination of pregnancy for medical reasons and we could not follow-up 16 liveborn babies (5% of all singleton births), particularly those that moved to another hospital as part of their care. The UKOSS infrastructure does not facilitate long term follow-up of babies into childhood, and therefore the information most wanted by parents, on the longer term outcomes of offspring, is not available with this methodology. Also, we defined neonatal morbidity as intraventricular haemorrhage grade 3 or 4, or requiring supplemental oxygen at 36 weeks' postmenstrual age, or both, which does not rule out other morbidities.
The incidence of PPROM between 16+0 and 22+6 weeks’ gestation of 0.04% was likely underestimated because of under-reporting, particularly during the covid-19 pandemic. We have no evidence of biased reporting, however, which might influence the generalisability of the results. A recent analysis of hospital episode statistics for the whole of France found a prevalence of 0.2% of PPROM at 14+0-24+6 weeks' gestation, which was considered to be likely underestimated because to fully validate the diagnosis with coded data is not possible.27 Within our study the incidence of PPROM was lowest between July and December 2020, possibly because of staffing pressures or potentially a lower rate of PPROM secondary to public health control measures, such as lockdowns.
The UKOSS methodology requires reporting clinicians to identify eligible cases. This approach might favour identification of pregnancies with a longer time between PPROM and birth, likely requiring a longer stay in hospital, which in turn could bias the results towards pregnancies with a more favourable neonatal outcome. In our study, median time between PPROM and birth for expectantly managed singleton pregnancies was 13 days (interquartile range 3-50, n=220). This value is in the higher range of other studies focusing on PPROM before 24 weeks' gestation that reported times of seven days (interquartile range 3-29, n=74),18 11.5 days (3-29, n=130),20 13 days (1-85, n=37),3 and 14 days (3-32, n=46).22 Nevertheless, in 27% of expectantly managed pregnancies, babies were born in the 72 hours after PPROM, indicating that the methodology also identified shorter times between PPROM and birth.
This study was prompted by the lack of national or international guidelines to standardise care after PPROM before 23 weeks' gestation. Women in this study likely received a variety of models of care based on their own and their clinician's preference (eg, provision of antibiotics, monitoring, and steroids). These practices could have influenced pregnancy outcomes, and analysis of these data will form a separate paper. We have presented general outcome categories to give clinicians, patients, and their families dealing with PPROM before 23 weeks' gestation a broad framework to guide counselling.
To simplify case reporting and capture the maximum number of incidences of PPROM, the inclusion criteria were brief, and questions about fetal anomalies, whether identified at the time of PPROM or not, were not included. The survey reporters were not asked whether women who opted for a termination of pregnancy for medical reasons, or had a baby loss, also had a fetus with a life threatening anomaly. Fetal anomalies might have been present in some instances, which could have influenced the decision to terminate the pregnancy. We collected a limited range of neonatal outcomes, and no outcomes after hospital discharge because of resource constraints.
Implications for clinical practice
In this study, we showed that pregnancies with PPROM before 23 weeks' gestation had a high risk of perinatal and maternal morbidities. Individual obstetricians are likely to see only one or two instances of early PPROM a year, and therefore cannot gain substantial experience relevant to the condition. Similar to other studies, we found that not all women opting for termination of pregnancy for medical reasons avoided severe morbidity.28 39
In current UK practice, pregnancies before 20 weeks' gestation are often cared for on gynaecology rather than obstetric wards. In our study, perinatal mortality was largely attributable to birth before viability and extreme prematurity because 39% of births occurred in the week after PPROM, and a further 22% in the following week. This pattern was consistent across the gestational ages of PPROM, and was similar to previous studies.18 34 We suggest that pregnancies with PPROM before 23 weeks' gestation are likely to need the support of neonatal teams, maternal critical care, and bereavement teams. Obstetricians with an interest in this condition, together with dedicated midwives, are possibly best placed to coordinate this care immediately after PPROM and for the remainder of the pregnancy. About half of the babies who were admitted to neonatal units after PPROM between 16+0 and 22+6 weeks’ survived. Most survivors did not have the morbidities we reported. Among survivors, the likelihood of the morbidities we identified was lower with each week of completed pregnancy. We suggest that teams with expertise in the management of very early PPROM, along with patient representatives, work together to develop guidelines for care.
Implications for research
This study provides baseline data on perinatal and maternal outcomes of pregnancies with PPROM before 23 weeks' gestation. Studies attempting to improve outcomes can build on our data. Interventions could be new treatments aimed at treating the pathology, or care bundles, including training, to optimise care and the timing of delivery with interventions already available. Materials for families with very early PPROM need to consider the inherent uncertainty within the data secondary to 32% of women with singleton pregnancies opting for termination of pregnancy for medical reasons. The optimal ways to communicate these uncertainties within the data to a wider audience, and support families with such complex pregnancies, have yet to be determined.
We can only comment on severe morbidity in babies at discharge from hospital, and previous work suggests that 70% of these babies will not have major morbidity when aged 18-21 months.14 The information most wanted by prospective parents is the rate of long term disability in offspring. Future research needs to incorporate ways of obtaining this information. Substantial unmeasured psychological morbidity related to this condition is likely. The optimal way of managing very early PPROM to support the psychological wellbeing of families requires further consideration.
Conclusions
Pregnancies with PPROM before 23 weeks' gestation can have favourable maternal and pregnancy outcomes, but a substantial proportion of these pregnancies are complicated by maternal morbidity and perinatal mortality and morbidity. All clinicians who care for these families need to be aware of the risk of maternal sepsis and death. These data will be helpful in counselling families with PPROM before 23 weeks' gestation and should be incorporated into updates of clinical guidelines.