Incidence and outcomes of vasa praevia in the United Kingdom

Introduction

Vasa praevia is a rare obstetric complication which is defined as fetal vessels coursing through the membranes over the internal cervical os and below the fetal presenting part, unprotected by placental tissue or the umbilical cord¹. This can be secondary to a velamentous cord insertion (vasa praevia type 1) or to fetal vessels running between lobes of a placenta with one or more accessory lobes (vasa praevia type 2). Rupture of the vessels during labour can lead to rapid fetal exsanguination and fetal death.

Vasa praevia can be associated with high perinatal mortality if it is not diagnosed antenatally^2,3. Although antenatal screening is possible, a suspicion of vasa praevia usually leads to caesarean section before labour, avoiding the chance of fetal vessel rupture during birth. However, the rarity of the condition; the possibly high false positive rate of antenatal diagnosis; the potential harms of iatrogenic late preterm delivery, such as neurodevelopmental issues⁴; and the inability of antenatal diagnosis to predict which cases of vasa praevia will rupture and bleed during labour means that antenatal screening has the potential for greater harm than benefit. Consequently, routine screening for vasa praevia is not advised by Royal College of Obstetricians and Gynaecologists guidelines⁵ or the National Screening Committee (NSC)^6,7, although both organisations indicate a paucity of evidence on which to base a recommendation. Conversely, a recent international expert consensus statement supported routine screening because of the potential reduction in preventable perinatal mortality⁸.

A recent systematic review of 24 studies suggested a mean incidence of vasa praevia of 7.9 (95% CI 5.9–10.1) per 10,000 pregnancies⁹. The majority of the included studies were retrospective, single-centre investigations and ultrasound protocols for antenatal detection of vasa praevia varied between studies. Two included prospective studies were small (4 and 11 cases of vasa praevia) and each based in two institutions^10,11. Two large population-based studies were identified: a retrospective study of the California birth cohort from 2007 to 2012 found an incidence of 2 per 10,000 live singleton births¹², while a prospective study using the Australasian Maternity Outcomes Surveillance System (AMOSS) showed a similar incidence of 2.1 per 10,000 women giving birth¹.

Data from the UK are limited; a retrospective study of data from a prospective vasa praevia screening programme in a single UK fetal medicine unit estimated an incidence of 8 per 10,000 singleton pregnancies, in a study population of 26,830¹³, while a 5-year historical cohort study at a single UK hospital with a screening program estimated a similar incidence of 7.7 per 10,000 births¹⁴. However, no study has investigated the population incidence of vasa praevia in the UK prospectively or using nationwide data. We used the United Kingdom Obstetric Surveillance System (UKOSS) to investigate prospectively the incidence and outcomes of vasa praevia in the UK. We also surveyed all UKOSS reporting centres about vasa praevia antenatal screening practices.

Methods

Results

Following exclusion of duplicates, 73 cases were reported to UKOSS. However, only 51 met the case definition (details of excluded cases are in Figure 1). Most excluded cases did not meet the placental confirmation criteria of our case definition: 10 cases had planned caesarean section for suspected vasa praevia but there was no documentation of the placenta being examined or sent to histopathology after birth, so there was no confirmation of vasa praevia; 7 cases did not have vasa praevia (2 had cord prolapse, 2 had placenta praevia, 3 others had suspected vasa praevia but no evidence of vasa praevia on postnatal examination of the placenta and membranes); and for 4 other cases vasa praevia was not confirmed on histopathology despite suspicious delivery events or outcomes.

Figure 1. Reported and confirmed cases of vasa praevia during the study period.

The total number of maternities for 2015 was 768,161. Therefore, we calculated the incidence of vasa praevia in the UK as 6.64 per 100,000 maternities (95% CI 5.05–8.73). This equates to approximately 1 case for every 15,062 maternities.

Of the 51 cases included, 28 cases (55%) were antenatal (diagnosed or suspected in the antenatal period). The remaining 23 cases (45%) were peripartum (not suspected/diagnosed in the antenatal period). All but one of these 23 cases were associated with bleeding during labour or during induction of labour.

Demographic data and pregnancy characteristics are summarised in Table 2. A low placenta or a succenturiate lobe were more likely to have been identified in the antenatally diagnosed cases. Otherwise, there were no differences between the two groups. It is notable that all cases had at least one risk factor for vasa praevia.

The clinical presentation and management are summarised in Table 3. Most peripartum-diagnosed cases presented with bleeding and this bleeding was at membrane rupture for 13 (57%) of these cases. Of the 23 women diagnosed peripartum, 22 (96%) delivered by caesarean section, of which 83% were category 1 surgeries; 15 women (65%) had general anaesthesia. There was one vaginal birth of a stillborn baby in this group. Of the 23 peripartum-diagnosed cases, 20 (87%) were born after 37 weeks and all after 36 weeks.

Only 14% (n=4) of antenatally diagnosed cases were not admitted to hospital prior to delivery. The cervical length was measured in 10 (36%) of the antenatally diagnosed cases but it was used as a factor in the decision to admit for only 3 (11%) cases. Fetal fibronectin was measured in only one case and it was not a factor in the decision to admit. The median gestation at diagnosis was 29 weeks, with the earliest at 17 and the latest at 36 weeks. The earliest admission for antenatally diagnosed vasa praevia was at 21 weeks. This woman had 4 admissions for a total of 73 days. Only 3 other women with an antenatal diagnosis of vasa praevia were admitted to hospital prior to 30 weeks (two of them because of vaginal bleeding). The median gestation at admission was 32.6 weeks with the latest admission at 36.5 weeks. The median duration of admission was 6 days (range 1–73). Seventeen (61%) women with antenatal diagnosis stayed in hospital for more than 10 days. The 4 women who were not admitted had planned caesarean sections at 38–39 weeks.

All antenatally diagnosed cases were scheduled for a caesarean section, with the majority scheduled between 36 and 37 weeks. The earliest scheduled caesarean section was at 34 weeks (n=4) and the latest at 39 weeks (n=2). However, 9 (32%) of these women had a category 2 or 3 caesarean section earlier than planned. The earliest caesarean birth was at 31 weeks (n=1) because of recurrent antepartum bleeding. Only one other woman was delivered before 34 weeks, because of threatened preterm labour. Of the prenatally diagnosed cases, 25 (89%) had regional anaesthesia.

The perinatal outcomes are summarised in Table 4. There was a multiple pregnancy in the peripartum diagnosis group, so the outcomes refer to 52 fetuses. There were 4 stillbirths and 5 neonatal deaths in the peripartum group, making the perinatal mortality rate 37.5% for this group. The overall perinatal mortality rate was 17%. Almost 50% (7 of 15) of the surviving babies in the peripartum group had hypoxic ischaemic encephalopathy (HIE). In the group with no antenatal diagnosis, 11 (46%) of babies had a blood transfusion for anaemia. There were no stillbirths, neonatal deaths or HIE cases in the antenatally diagnosed group but 2 babies in this group developed respiratory distress syndrome after delivery at 35–36 weeks.

Of the 198 reporting centres, 174 (88%) replied to our question: “Between the 1st of December 2014 and the 30th of November 2015 was there a formal screening programme for vasa praevia at your hospital/centre?”. Only 10 hospitals (6% of respondents) declared they had a formal screening programme for vasa praevia during the time of the study. Fifteen replies (9%) were “unknown”. One of the 10 hospitals with screening programmes reported 7 cases, all antenatally diagnosed (14% of all cases and 25% of all antenatally diagnosed cases). The other 9 hospitals did not report any cases. This difference in antenatally diagnosed cases between screening and non-screening hospitals was statistically significant (Table 5).

Discussion

This prospective population-based study showed a lower than anticipated incidence of vasa praevia in the UK: 6.64 per 100,000 maternities, or 1 case for every 15,062 maternities. There was high perinatal mortality and morbidity with peripartum diagnosis. The incidence of antenatally identified cases in the few centres that actively screened for this condition was much higher, and the perinatal outcomes were better. However, this group were all delivered by caesarean section and some infants had respiratory morbidity as a consequence of elective preterm delivery.

A significant strength of our study is its prospective population-based design, which does not rely on routinely coded data to ascertain cases. This is a national study, conducted at all obstetric units in the UK and therefore covering the whole of the pregnant population. UKOSS is a well-established and validated system for identifying cases of uncommon pregnancy complications and, because it covers all obstetric units, is not subject to the usual biases of single-centre studies. While we cannot exclude the possibility that there may have been a few cases that were diagnosed but not reported, we have no reason to suspect substantial under-ascertainment or bias in reporting. This study therefore represents a true snapshot of the current number of diagnosed cases in the UK pregnant population, with the caveat that only a small number of centres are actively screening for vasa praevia.

We used a robust case definition in order to accurately capture clinically diagnosed and confirmed cases. However, this strict definition was a potential weakness, as demonstrated by the 10 antenatally suspected cases who could not be included as there was no evidence of examination (histopathological or not) of the placenta after birth. Although we had multiple reporters in each hospital, we cannot be certain all cases were identified. In particular, the diagnosis may never have been considered as there is a possibility that ruptured vasa praevia mimics other conditions (e.g., abruption) and there is no “gold” standard to confirm the diagnosis after birth if the vessels are not ruptured.

There are guidelines about the diagnosis and management of vasa praevia in the UK, USA, Canada and Australia highlighting the importance of the condition. These were compared in a recent publication¹⁹. All guidelines agree that colour and pulsed Doppler transvaginal ultrasound should be used in mid-trimester to diagnose the condition, but third trimester confirmation is also needed. Universal screening is not recommended but rather a targeted screening for women with risk factors, such as low placenta, is advocated. Antenatal steroids are to be considered from 28–32 weeks and hospitalisation from 30–32 weeks. All four guidelines agree the optimal timing of birth is unknown, but they recommend birth by caesarean section at 34–37 weeks. A recent expert consensus statement recommended caesarean birth at 35–37 weeks, but advocated second-trimester screening as routine for all pregnancies, via transabdominal ultrasound with colour Doppler sweep over the cervix, followed by third-trimester confirmation of the diagnosis by transvaginal ultrasound⁸.

Our study demonstrates that vasa praevia diagnosis in the UK is less common than anticipated, with a lower incidence of vasa praevia than found in previous studies^1,9,12. Even if the 10 planned caesarean sections with no postnatal confirmation were included, the estimated incidence would still be relatively low at 7.94 per 100,000 maternities (1 case for every 12,593 maternities). Although our strict case definition may be partially responsible, the AMOSS study in Australia demonstrated about 2–3 times higher incidence (21 per 100,000) using an almost identical case definition. However, in the Australian study, 92% of cases were identified antenatally with only 5 cases being identified peripartum. Peripartum cases might be considered to represent the true incidence of ruptured vasa praevia; the incidence of peripartum vasa praevia in Australia was 1.7 per 100,000, and in the UK it was 3 per 100,000. The differing thresholds and screening protocols for detecting vasa praevia antenatally are likely to explain these apparent differences in the overall incidence between the two countries.

Other possible reasons include cases not being reported or not recognised during the study period; women with undiagnosed vasa praevia having pre-labour caesarean section for other indications (especially low lying placenta); and some cases with poor outcome could have been attributed to other causes, such as abruption, or ruptured fetal vessels on placental histopathology being attributed to “snapped” cord during continuous cord traction at delivery, which is more likely to occur with velamentous cord insertion.

Our study found very different outcomes in those cases that were identified antenatally compared with those diagnosed intrapartum or postnatally. In cases not diagnosed antenatally there was an almost 40% perinatal mortality. This high perinatal mortality is remarkably similar to that observed in a retrospective study (44%)² and in the prospective AMOSS study (40%)¹ and confirms the high fatality rate of vasa praevia when it is not suspected or diagnosed before labour. Our study also found that almost half of the survivors in cases with no antenatal diagnosis had hypoxic ischaemic encephalopathy, although the grade and consequences of this are unknown; in contrast, none of the infants in cases with antenatal diagnosis had hypoxic ischaemic encephalopathy. This is in keeping with the findings of a recent systematic review, which found that 58% of surviving neonates in cases of vasa praevia without antenatal diagnosis displayed hypoxic morbidity, compared with only 2.7% in cases with antenatal diagnosis³.

Most previous studies²⁰ demonstrated the presence of risk factors in most vasa praevia cases; in our study all cases had at least one risk factor. This might suggest that targeted screening for cases with risk factors, particularly low placenta and velamentous cord insertion, might be useful as one or both of these were present in almost half of the cases in our study.

Ten maternity units declared they had a formal screening programme for vasa praevia. Of these, only one centre reported any cases of antenatally detected vasa praevia during the study period. Table 5 illustrates that the number of reported cases of vasa praevia was substantially higher if there was antenatal screening. Large, well-designed prospective screening studies may elucidate this observation further; our study was not designed to evaluate screening efficiency or accuracy. A historical cohort study of a screening program in a single UK institution reported 100% sensitivity and 99.78% specificity, but a perinatal mortality of 5% (one death in 19 confirmed cases of vasa praevia)¹⁴.

Evaluation of accuracy of screening methods for vasa praevia is difficult because of the absence of a diagnostic “gold standard”. Once the placenta is delivered, velamentous cord insertion or bilobed placenta can be confirmed but it is impossible to know what the proximity of the vasa praevia to the cervix was. Inspection during the caesarean section is often difficult because of bleeding from the uterine incision and the placental bed. In addition, there is no clear definition of vasa praevia as there is no evidence about a “safe” distance from the internal os. A distance of 2 cm between the fetal vessels and the internal os has been proposed^21,22 but it is not based on robust evidence. It is likely that many clinicians would find it difficult to recommend expectant management and normal birth to a woman with fetal vessels at e.g. 2.3 cm from the cervix. Indeed, a recent expert consensus process failed to reach agreement on the fetal vessel-internal os distance that should be used to define vasa praevia, but concluded that it should not be limited to 2 cm⁸, and a recently registered clinical trial of fetoscopic laser photocoagulation in management of vasa praevia (https://clinicaltrials.gov/study/NCT06290232) will use a 5-cm threshold for diagnosis. A UK single-institution study of a two-stage screening strategy for vasa praevia used a diagnostic threshold of 5 cm and estimated an incidence of 8 per 10,000 singleton pregnancies¹³. If reporting centres in our study were using more restrictive criteria, this may go some way to explaining the much lower incidence that we observed.

The Royal College of Obstetricians and Gynaecologists⁵ recommends consideration of prophylactic hospitalisation for confirmed vasa praevia after 30–32 weeks. Our study showed a wide variation in practice with most women with an antenatal diagnosis of vasa praevia being admitted after 30 weeks, at a median gestation of 32–33 weeks. Although most women spent up to 18 days in hospital in the antenatal period, there were 2 women who were admitted for over 2 months. The 4 women who did not have antenatal hospital admission had later planned caesarean sections at 38–39 weeks, suggesting the absence of any episodes of antepartum bleeding, no risk factors for early labour or even perhaps maternal preference.

Our data showed that most caesarean sections for the antenatally diagnosed cases were scheduled for 36–37 weeks and some up to 39 weeks. In about one third of these cases the caesarean section had to be brought forward, usually because of episodes of bleeding or threatened preterm labour. Only 2 caesarean sections (7%) were performed before 34 weeks. Within the antenatally diagnosed group there were only 2 cases of respiratory distress syndrome but 10 (36%) babies were admitted to a neonatal unit. Given the recently recognised additional risks of late preterm birth^4,23, it would seem sensible to aim for delivery at 36–37 weeks and perform earlier caesarean sections for cases with episodes of vaginal bleeding or other risk factors. Clinicians can also consider using cervical length and/or fetal fibronectin or other cervicovaginal biochemical swab tests to stratify further the risk of preterm labour before 35 weeks of gestation although there is no current evidence to support this approach²⁴. Of course, maternal anxiety cannot be underestimated in these circumstances²⁵ and can be a significant factor in birth timing.

Conclusion

This prospective national study showed a lower than anticipated incidence of vasa praevia, and a very low population incidence of poor perinatal outcome after vasa praevia. However, peripartum diagnosis was associated with an almost 40% perinatal mortality and about 50% risk of HIE in the surviving neonates. Antenatal diagnosis led to planned caesarean section for the majority of infants at 36–37 weeks with no observed deaths or cases of HIE, but some infants had respiratory morbidity. Our study could not address screening efficacy or the optimal screening method, but showed that a screening programme was associated with an increased number of vasa praevia diagnoses and no perinatal deaths, although the difference on the latter metric between maternity units with and without a screening program did not reach statistical significance.