A regional audit system for stillbirth: a way to better understand the phenomenon

Despite the large number of stillbirths (SB) occurring every year [1, 2], global attention regarding this issue is still insufficient. SB rates have declined more slowly since 2000 than both maternal mortality and mortality in children younger than 5 years [3]. Evidence suggests that this gap can be narrowed by improving SB registrations, data collection, investigation into the cause of death, reducing social disadvantages and preventing modified risk factors [4]. The Every Newborn Action Plan (ENAP), endorsed by the United Nations, aims to reduce the SB rate to 12 or fewer per 1000 births in every country by 2030 [5]. Moreover, ENAP suggests all countries to implement high quality national audits for perinatal mortality, which translates into improvements in quality of care, and the registration of all perinatal deaths together with the identification of the cause of death. Achieving these goals requires optimal diagnostic testing and multidisciplinary review as part of a high-quality perinatal mortality audit. According to a recent report [6], countries that have implemented national perinatal audit programs have achieved a reduction in SB rates. In addition, they have found an unacceptably high proportion of cases with elements of sub-optimal care. Despite the benefits of perinatal audit, still today few Nations have implemented this scheme [7‐9].

The existence of a large number of classification systems does not always lead to the identification of the cause of death, leaving many cases unexplained [10]. Establishing an accurate cause of death is necessary for parents to understand why their baby died, help them to cope with the death and reduce the risk of recurrence [11, 12].

A national perinatal audit program is currently lacking in Italy. In Emilia-Romagna (ER), a region in the North of the country, a Regional SB Audit System, managed by a multidisciplinary panel, has been implemented since 2014 when the Regional Council approved a resolution [13] to set up multidisciplinary working groups, to perform local and regional perinatal audits.

This study describes the process of the SB audit programme and presents the results after the first 3 years of implementation.

From 2014 to 2016, 332 SB occurred out of a total of 107,528 births, with a SB rate of 3.1 per 1000. Of these SB, 81 (24.4%) occurred before 28 weeks and 251 thereafter (75.6%), yielding a late SB rate of 2.3 per 1000. The SB rate constantly declined between 2014 (3.2 per 1000) and 2016 (3.0 per 1000). Seventeen cases (5.1%) occurred after the onset of labour and were considered intrapartum. Double-check led to the identification of 21 early SB which had not been recorded in the current Birth Register.

Fifteen cases (4.5%) originated from multiple pregnancies. Our population was heterogeneous in term of ethnicity: there were 190 (57.6%) Italian women, the others were from North Africa (11.5%), East Europe (10.3%), Sub-Saharan Africa (8.2%), Indian Subcontinent (7.6%) and other countries (4.8%).

Complete diagnostic protocol was applied in the majority of cases. Placental examination and autopsy were performed in 298 (90.3%) and 290 cases (87.8%), respectively.

Clinical records were available for all cases but two, seized by the judicial authority. For this reason, analyses were conducted on a population of 330 SB.

Risk factors for SB are reported in Table 1. After 41 weeks, the risk of fetal demise was lower than in preterm and full-term pregnancies (OR 0.3, CI95% 0.1–0.8). A progressively increasing risk was identified in overweight (OR 1.42, CI95% 1.07–1.86) and obese women (OR 1.96, CI95% 1.40–2.74). Women from Indian Subcontinent, North and Sub-Saharan Africa presented a higher risk of SB than Caucasian and above all Sub-Saharan Africa women had an almost tripled risk (OR 2.9, CI95% 1.94–4.35). Finally, women who had a previous SB carried a greater risk of recurrence (OR 2.62, CI95% 1.34–5.14). The risk was not significantly increased for smoking in pregnancy, multiple gestations, maternal age and education.

Table 2 shows the distribution of causes of death in our cohort. The Regional Audit Program reviewed all cases previously evaluated by local groups and established a different cause of death in 54 (16.4%) cases. Disagreement on the causes of death between local and central committees were more frequent during the first year of the project and progressively reduced in the following years, when local committees improved their ability in interpreting the results of the investigations and classifying the cause of deaths, thanks to discussions during the meetings (data not shown). Placental disorders were the most frequent relevant condition (33.3%), followed by the fetus (17.6%), cord (14.2%) and maternal disorders (7.6%). Forty-seven (14.2%) cases remained unexplained even though the diagnostic protocol had been carried out. However, it was not possible to assign a cause of death in 14 cases (4.2%) because the diagnostic work-up was incomplete.

Associated relevant conditions at death were found in 329 cases and the most frequent one was FGR which was ascertained in 49 cases (14.9%).

For multivariate analysis, relevant conditions at death have been categorized in seven groups: placental disorders (110 cases), unexplained (61 cases), cord accidents (45 cases), infections (including proven fetal infection, histologic funisitis and chorioamnionitis −39 cases-), fetal disorders (38 cases), maternal disorders (25 cases) and others (12 cases).

Infections were associated with early intrauterine fetal death (OR 3.38, CI95% 1.62–7.03) while placental disorders were related to preterm SB (OR 1.86, CI95% 1.11–3.15) and FGR (OR 1.89, CI95% 1.06–3.36). Furthermore, maternal disorders were associated with overweight (OR 3.38, CI95% 1.33–8.6) and LGA (OR 4.26, CI95% 1.07–12.87).

No significant association was found between causes of death and smoking (although abruptio was more frequent in this group), level of education, multiple pregnancies and maternal country of birth, even though women from Indian Subcontinent had a high proportion of SB due to placental disease (48%).

The quality of care provided during pregnancy and labour has been assessed as shown in Table 3. Elements of substandard care were present in 48 (14.5%) cases. A different management might have made or would have reasonably made a difference to outcome in 36 (10.9%) cases. Elements of sub-optimal care were identified only in antepartum cases. The first cause of death among such cases was placental insufficiency (30.5%), followed by maternal disorders -diabetes, hypertension disorders, antiphospholipid syndrome- (25%). At multivariate analysis the group of maternal disorders was associated with substandard care of grade 2 and 3 (OR 6.64, CI95% 2.61–17.02). No significant association has been found for the remaining causes of death. Eleven women (30.5%) had a delayed access to antenatal care or missed the antenatal appointments because ignored their pregnancy or either refused hospitalisation. In 9 cases (25%) clinicians failed to diagnose or manage diabetes or hypertensive disorders. Six cases (16.6%) affected by fetal growth restriction were either not detected or managed inappropriately. In 9 cases different suboptimal care factors were identified, e.g. entering the pregnancy with an elevated BMI or continuing smoking during pregnancy.

Among the 9 cases evaluated of grade 3, seven occurred in racial minorities and only 2 in Italian women. Figure 1 shows the distribution of women’s country of birth in relation to quality of care grading. The majority of cases without substandard care occurred in Italian women, while most preventable stillbirths (grade 3) occurred in migrant women born in other countries.

The leading cause of death in intrapartum SB was infection (41.1%), followed by asphyxia (11.8%), uterine rupture (11.8%) and others (17.7%). Unexplained cases were 17.6%. Among the seven intrauterine fetal deaths caused by infection, 5 occurred before 28 weeks. At logistic regression, intrapartum SB were associated with infections (OR 7.15, CI95% 2.35–21.77), while there was no association with inadequate care (grade 2 and 3).

The audit process allowed a more realistic evaluation of stillbirth phenomenon, highlighting the underestimation of early SB, due to the present Italian law which defines miscarriage as products of conception prior to 180 days of development (25 weeks and 5 days). Thus, we explained the discrepancy of 21 SB recorded in the audit process, and not in the Birth Register, with the fact that stillbirths occurred between 22 weeks and 25 weeks and 5 days were registered as miscarriage, according to the Italian law, instead of using WHO’s definition for stillbirths.

Late SB rate in ER was one of the lowest rates in Europe after Iceland, Denmark, Finland, the Netherlands, Croatia, Norway and Portugal [4].

Most of the risk factors for SB known in literature are confirmed by our study - overweight/obesity [22], racial disparity [2, 4] and previous SB - [23‐25]. On the other hand, while post-term pregnancy is widely recognized as a condition of higher risk for SB [26], this is not confirmed in our population where the risk decreased after 41 weeks compared to 39–40 weeks and no fetal death occurred after 42 weeks. We hypothesized that such finding could be related to the strict surveillance of pregnancies above term in our Region. Indeed, there is a weekly monitoring and scheduled induction for pregnancies with medical indications, like gestational diabetes, hypertensive disorders and late term pregnancy (> 41 weeks) [27‐29].

Placental disorders were the leading cause of death in our cohort and explained one-third of stillbirths in line with other authors which reported placental dysfunction as a major contributor [30‐32]. Moreover, fetal disorders were responsible for about one case out of six and FGR was the principal relevant condition at death in only 3% of cases. Such frequency is much lower than the one reported by Vergani et al. (17%) [16] with the same classification system. When we attributed the relevant condition at death, we considered placental disorders as a logic antecedent of FGR and then SB, believing that growth restriction could be the cause of death only in the absence of other conditions. Indeed, considering also the associated factors, the whole rate of FGR (14.9%) was similar to the one in Vergani’s group. Our results are in general agreement with those of Ego et al. [33] who previously adopted the same logic to classify cause of stillbirths with ReCoDe, i.e. the incidence of placental causes doubled becoming almost 25% when FGR was considered as a cause only in the absence of other conditions.

Infections are another significant cause of death, counting for almost 12%. This is particularly true for early SB as previously reported by other authors [34, 35].

On the other hand, the Audit System could not explain a significant number of cases (about one case out of seven). This rate was close to those previously reported by English [15], Italian [16] and Dutch [31] studies. Therefore, despite the existence of different classification systems [10], a consistent number of SB remains unexplained.

It is important to underline that the diagnostic workup was performed in every hospital, limiting unexplained stillbirths due to lack of investigation to < 5%. At the same time, placental pathology and fetal autopsy were extensively carried out.

Based on current guidelines, substandard care factors were identified by the multidisciplinary panel in a significant number of cases where a different management might have made a difference to outcome or would have reasonably been expected to have made a difference to outcome. In high-income countries, few perinatal audits evaluated the quality of care in relation to SB. A national audit in the Netherlands identified such inappropriateness in 8% of more than 700 term perinatal deaths [9]. Also a review in New Zealand identified that up to 15% of deaths are potentially avoidable [8]. The Euronatal study [36], which included quality of care evaluation in 10 European regions between 1993 and 1998, found an even higher percentage of sub-standard care factors than previously reported. It seems likely that the above differences could be partly explained by quality of care improvement during the past 20 years.

In our study, the main issues recognised as areas for improvement in antenatal care provision included the access to antenatal care for disadvantages minorities and the identification and management of maternal disorders. Indeed, almost all cases with quality of care graded 3 occurred in women with poor access to antenatal care, also because of unintended pregnancies. This confirms that negative outcome did not always depend on care offered but pertained to very disadvantaged minorities. Therefore, it is mandatory to increase early access to the antenatal care system in the effort to minimize racial and ethnic disparities.

Furthermore, the association between the presence of maternal disorders and inadequate care should stimulate actions looking at the improvement of the management of high-risk pregnancies. Improving detection and management of metabolic and hypertensive disorders, seems of paramount importance also in reviewing the correlation between SB-related maternal disorders and overweight/obesity.

As expected in western countries [2], intrapartum cases were limited to 5% and were associated with appropriate care in the vast majority of the cases. At logistic regression, such deaths were associated with infections, namely before 28 weeks, as also reported by others [32]. Indeed, it is possible that infections may have induced preterm labour and determined fetal death because of extremely premature labour.

A limitation of the audit system is that it is time consuming. Another one is that despite the presence of multidisciplinary professionals, there is no external validation. Furthermore, histological exams were performed locally by different pathologists whose expertise in perinatal pathology was heterogenous. The importance of placental pathology, confirmed also by our work, supports the importance of having pathologists trained and specialized in perinatal pathology. Moreover, deaths occurring in the first neonatal week were not included. Finally, we do not have sufficient data to assess the impact of the validity of each investigation.

This study has several strengths. Comparison between cases detected by the audit project and current Birth Register led to a precise counting of each and every SB. Moreover, a homogeneous work-up was performed and information was prospectively collected, ensuring a high quality of data which is difficult to reach in large database. Moreover, causes of death were assigned by consensus among multidisciplinary panel of experts in maternal-fetal medicine and this guaranteed a high accuracy results.

These data demonstrate that it is possible to implement a Regional Audit System of Stillbirth as recommended by international institutions [37]. Overall results are useful to understand local reality, in order to plan interventions towards specific populations.