Perinatal mortality and severe morbidity in low and high risk term pregnancies in the Netherlands: prospective cohort study

Dear editor,

We appreciate the remarks and comments of de Jonge et al and Pop et
al on our paper "Perinatal mortality and morbidity in low and high risk
term pregnancies in the Netherlands: prospective cohort study'. We are
pleased to have the opportunity to clarify their remarks and comments.

De Jonge et al commented on the nomenclature of our study. We
performed a cohort study, including cases in a prospective manner. With
this design, important problems associated with retrospective data
collection - such as dependence of the investigator on the availability
and accuracy of medical case records or classification bias of the cases -
are prevented. The population at risk was estimated retrospectively using
aggregated -prospectively collected- data from the national perinatal
register. We are aware of the limitations of this part of the study and
emphasized this in the abstract, the sensitivity analysis and the
discussion session of our paper.
Secondly, de Jonge et al are concerned that the mortality rates in
midwifery practices may be artificially inflated since midwives in
practices at the periphery of the catchment area also care for women in
neighboring regions and these births are not included in the study.
However, why should there be a bias in referral regarding the condition of
the fetus or why would all mortality cases be referred to our NICU?
Midwife practices residing just outside our catchment area but caring
partly for women within our catchment area will probably send their
morbidity and mortality cases to another NICU than the UMC Utrecht. In
other words, the mechanism will work in both ways. Nevertheless, since
predefined postal codes were used to extract data of normal births from
the perinatal data base, underreporting cannot be completely excluded. To
show that our findings were not caused by underreporting of normal births
we artificially increased the denominator by 10% and no difference in
outcome was shown. If we increase the denominator with another 5 %,
accounting for the possibility that we have missed births from the
periphery of our catchment area, the results are still robust. So, the
perinatal mortality rate in women starting labour in primary care is not
artificially inflated.
The third comment of de Jonge et al is the discrepancy of our results
compared to previous studies (1) (2). In our study all death and NICU
admissions were prospectively evaluated by a multidisciplinary team of
experts. They used clear and beforehand determined definitions to classify
the timing of deaths in a systematic way. In all other studies data were
retrospectively extracted from large databases, in which data can be
missing or filled in incorrectly. Notwithstanding, the differences in
study design of our study and the studies mentioned by de Jonge et al, our
perinatal mortality rate among term infants (2.6 per 1000, excluding
congenital malformations and late perinatal death) is similar to a large
nationwide study of Ravelli et al (2.8 per 1000) and the numbers reported
in PERISTAT 2 (3.2 per 1000, including congenital anomalies) (3) (4). We
hypothesize that classification problems in data bases could explain the
differences with other studies.
Furthermore, de Jonge et al were surprised that most babies that died
during labour were born in primary care. They concluded that this means
that either the midwife noticed fetal distress too late to refer a woman
because the birth was imminent or a deceased baby was born so fast that
the midwife arrived too late to a woman's home. This comment is probably
based on a misunderstanding of our results. Of all intrapartum
stillbirths, 12 intrapartum stillbirths occurred under supervision of a
midwife in primary care. In 10 cases the midwife found no heartbeat during
auscultation and the woman was referred to secondary care where the
stillbirth was confirmed. In these cases, the baby was born in secondary
care, but intrapartum death occurred under supervision of the midwife in
primary care and was therefore classified as death in primary care. This
is explained in the second footnote (+) of Table 2, which should have been
printed after "delivery in primary care", instead of after the head of the
column "intrapartum stillbirth".

Pop et al wonder how many women who start labour with the community
midwife were already in hospital before they were referred to secondary
care. Our research question did not include the comparison of home
delivery versus hospital delivery of low-risk women in primary care. We
focused on the comparison between low-risk women in primary care versus
high-risk women in secondary care. We get the impression that our remark
"the Dutch obstetrical care system may contribute to the high perinatal
mortality" was explained as if we suggested 'home births' instead of
'obstetrical care system'. With 'Dutch obstetrical care system', however,
we meant the system as a whole, since events and processes at all stages
may have contributed to our results. We regret if, against our intention,
the idea has developed that we suggest that home delivery is the cause of
the higher perinatal mortality rate among low-risk women. This is
explicitly not what we want to imply.
Furthermore, Pop et al wonder whether the cohort is representative for the
whole country. Although we cannot be certain, there are indications in
favour of representativeness, both from a theoretical point of view (e.g.
national guidelines, close cooperation and communication between hospitals
and midwife practices over regions) and from the results. For example, we
described in our paper that 65 % of the perinatal mortality cases, in
which labour started in primary care actually started at home and the
remaining 35 % in the hospital. This is in agreement with the national
ratio of home and hospital deliveries starting in primary care (1) (2).
The fact that the percentage of low-risk women who starts labour in
primary care is higher than the percentage mentioned in the national
registry can be explained by our selection criteria (proportionally more
low-risk women, because we selected only term women, carrying babies
without congenital anomalies).
Secondly, we understand the concern regarding the univariate analysis of
the data and as a consequence the lack of adjustment for potential
confounders. We agree that there is no comparability of baseline
characteristics between the two groups (low-risk and high risk). However,
that is inherent to the Dutch obstetrical care system. To get more
insight, we extracted data from the national perinatal registry including
several important variables and compared these risk factors between low-
risk women in primary care and high-risk women in secondary care. As
mentioned in the results section, these data show that age ? 35, non-
western ethnicity and low socio-economic class are more prevalent among
high-risk women. Therefore we believe that, if it would be possible to
adjust for these risk factors, the effect would probably be even larger.
However, we agree that this is a weakness of the study.
Last, Pop et al have concerns regarding potential cross-overs between low-
risk and high-risk women. We agree that there will be cross-overs. It is
known that there are low-risk women under care of a gynaecologist and
probably also vice versa. Our rates of 49.5 % (overall) en 46.7 % (for
nulliparous women) women starting labour in primary care, are not
different from the national rates reported in 2007 (5).

Finally, we would like to add a general comment. To our knowledge,
this is the only study in the Netherlands showing a higher risk of
delivery related perinatal mortality among women with the intention to
deliver in primary care compared with women who start delivery in
secondary care. Given the limitations of the study, we must interpret the
results carefully and we agree that with our study design we are not able
to demonstrate a causal association between the results and (specific
parts of) the obstetric care system. We would like to emphasize that in
the discussion section of our paper we only philosophized about possible
explanations and solutions, without the intent to make statements or
strong conclusions. However, our findings are unexpected and the obstetric
care system of the Netherlands as a whole needs careful further
evaluation. We hope our paper and the discussion presented here will
stimulate gynaecologists, midwives and paediatricians to think further
about these remarkable findings and to cooperatively explore possibilities
that may lead to improvement in outcome of newborns.

(1) Amelink-Verburg MP, Verloove-Vanhorick SP, Hakkenberg RM,
Veldhuijzen IM, Bennebroek GJ, Buitendijk SE. Evaluation of 280,000 cases
in Dutch midwifery practices: a descriptive study. BJOG 2008; 115(5):570-
578.

(2) De Jonge A, Van der Goes BY, Ravelli AC, Amelink-Verburg MP, Mol
BW, Nijhuis JG et al. Perinatal mortality and morbidity in a nationwide
cohort of 529,688 low-risk planned home and hospital births. BJOG 2009;
116(9):1177-1184.

(3) Ravelli ACJ, Eskes M, Tromp M, Van Huis AM, Steegers EAP,
Tamminga P, et al. Perinatal mortality in the Netherlands 2000-2006: risk
factors and risk selection. Ned Tijdschr Geneesk2008;152:2728-33.

(4) Zeitlin J, Mohangoo A, Cuttini M, EUROPERISTAT Report Writing
Committee, Alexander S, Barros H, et al. The European perinatal health
report: comparing the health and care of pregnant women and newborn babies
in Europe. J Epidemiol Community Health2009;63:681-2.

(5) Stichting Perinatale Registratie Nederland. Perinatale Zorg in
Nederland 2007. Utrecht: Stichting Perinatale Registratie Nederland, 2009

We read the article from Evers et al with great interest (1). Since
the poor ranking of the Netherlands in the Peristat report (2), the
typical Dutch obstetric system has been continuously in the spotlights,
nationally as well as internationally. The crucial question is whether the
relatively high rate of perinatal mortality can be (partly) explained by
the unique obstetric system with independent community midwifes being
responsible for roughly 40% of all births, be it at home or in the
hospital. The article by Evers et al adds interesting new, as they call
unexpected, insights to the debate. But unfortunately, the article
generates questions rather than answers.

The fact that the results are unexpected, unexplained and
contradictory to earlier nationwide research (3,4), warrants very cautious
interpretation of these new findings. It is the same uniqueness of the
Dutch obstetric system that makes it also very difficult to study
properly. Characteristics of women cared for by community midwifes appear
to be clearly different from those of women cared for by obstetricians.
Even between women with an uneventful pregnancy that choose to deliver at
home or in hospital, there appear to be background differences
complicating direct comparison of groups. Criteria for low or high risk
pregnancy are less clear than they seem and the criteria are not applied
rigorously.
Clearly, the risk of perinatal mortality is expected to be higher in women
with a high risk pregnancy. But obviously, these women receive different
antenatal care, hopefully resulting in their risk of perinatal mortality
to be reduced by that. Would a case of antenatal mortality have been
prevented if antenatal care was given by the obstetrician? Can peripartum
mortality after referral from midwife to obstetrician be attributed to the
fact that the woman was at home when delivery commenced? What about the
25% of low risk women who delivered in the hospital with their midwife?

Though the outcomes of the current study can be questioned, the topic
is so crucial and fundamental to the Dutch obstetric system that further
analysis of these results is mandatory. The only answer to this
uncertainty clearly is audit. Structured audit will reveal whether the
perinatal death would have been preventable, and if so, at what point in
the chain of obstetric care. Fortunately, our ministry of health has now
facilitated extensive audit of every single case of term perinatal
mortality in the Netherlands, including all cases described in the study
of Evers et al.1 Based on the outcomes, the Dutch obstetric system should
be modernized, thereby realizing that informed choice of place of delivery
is a precious property of the system.

What needs to be done now is wait for the audit conclusions. In the
meantime all Dutch obstetric professionals need to care for the thousands
of worried pregnant women in the Netherlands, carefully explaining them
that obstetric care is among the best in the world. Let's not jump to
conclusions...

Joost J. Zwart, MD, PhD. resident gynaecology, department of
obstetrics, Leiden University Medical Center, Leiden the Netherlands

Jeroen van Dillen, MD, PhD. resident gynaecology, department of
obstetrics, Radboud University Medical University, Nijmegen the
Netherlands

1. Annemieke C C Evers, Hens A A Brouwers, Chantal W P M Hukkelhoven,
Peter G J Nikkels, Janine Boon, Anneke van Egmond-Linden, Jacqueline
Hillegersberg, Yvette S Snuif, Sietske Sterken-Hooisma, Hein W Bruinse,
and Anneke Kwee. Perinatal mortality and severe morbidity in low and high
risk term pregnancies in the Netherlands: prospective cohort study. BMJ
2010 341:c5639; doi:10.1136/bmj.c5639

2. Zeitlin J, Mohangoo A, Cuttini M, EUROPERISTAT Report Writing
Committee, Alexander S, Barros H, et al. The European perinatal health
report: comparing the health and care of pregnant women and newborn babies
in Europe. J Epidemiol Community Health2009;63:681-2.

3. Amelink-Verburg MP, Verloove-Vanhorick SP, Hakkenberg RM,
Veldhuijzen IM, Bennebroek GJ, Buitendijk SE. Evaluation of 280,000 cases
in Dutch midwifery practices: a descriptive study. BJOG 2008; 115(5):570-
578.

4. De Jonge A, Van der Goes BY, Ravelli AC, Amelink-Verburg MP, Mol
BW, Nijhuis JG et al. Perinatal mortality and morbidity in a nationwide
cohort of 529,688 low-risk planned home and hospital births. BJOG 2009;
116(9):1177-1184.

Home or hospital delivery, who will sing the blues?

Victor J Pop, MD, PhD, professor of Primary Care; Hennie Wijnen, RM,
PhD.

To the editor

Evers et al conclude in their paper: '' the Dutch obstetric care
system may contribute to the high perinatal mortality'' 1.
The selective way of inclusion of perinatal death cases in the paper has
already been seriously questioned 2.
There are, however, several other important comments on the study.

Primary obstetric care in The Netherlands refers to care by a community
midwife for low-risk women and secondary care to care for high-risk women
by an obstetrician in hospital, based on strict criteria. However, crucial
with regard to Evers conclusions, low-risk women in primary care can
deliver at home but also in hospital with a community midwife. In both
cases, this is regarded as primary obstetric care. This means that in low-
risk women there can be a transition from primary to secondary care during
labour when she is already in hospital. Strikingly, the authors do not
mention any data on the number of women who deliver at home, who deliver
in hospital with a community midwife and who deliver after being referred
in hospital from primary to secondary care. But in the discussion section
all kinds of possible mechanisms to explain their findings are mentioned
without being assessed. One explanation they suggest: ''the community
midwife is not present during the first hours after labour started at
home''. However, the Dutch Perinatal Registry3 shows that by far the
majority of referrals to secondary care occur after the first hours of
labour (most often because of delayed dilatation time). ''There might be a
time-delay because of transport to the hospital in case of emergency''.

But how many women who start labour with the community midwife were
already in hospital before they were referred to secondary care?

The best way to answer the question where to confine is to randomize
low-risk women to primary or secondary care delivery. It is obvious that
no low-risk pregnant woman will accept randomization regarding place of
delivery for the benefit of research. An alternative is an ''open''
prospective study in which, as much as possible, important known
confounders of peri-natal death are included in the design of the study.

It is unacceptable that the authors - drawing conclusions which have such
an enormous impact in our society - just state in their limitations of the
study section that no confounders at all were included (because they did
not assess them, so it was not a prospective study as they stated).

But there is more. The number of low-risk women who starts labour in
primary care is 5% higher than the number mentioned in the Dutch Perinatal
Registry of 2007. So, were more high-risk women included in the low-risk
group? Also, the national registry has shown that during the last decade
up to one third of the 25% of all pregnant women who never see a midwife
during pregnancy (defined as high-risk) and who deliver in secondary care,
in fact are - according to the strict criteria - low-risk women. This
means that when applying the national data to the current cohort there
were 5% more low-risk women starting labour in primary care as expected
while in the high-risk group 8% women in fact were low-risk. We wonder
what the differences of peri-natal death figures would have been if only
the real low- and high-risk women were included in the analysis. Or is the
cohort not representative for the whole country which does not allow the
authors to generalize the conclusions?

1. Evers AC, Brouwers HA, Hukkelhoven CW, Nikkels PG, Boon J, van
Egmond-Linden A, Hillegersberg J, Snuif YS, Sterken-Hooisma S, Bruinse HW,
Kwee A. Perinatal mortality and severe morbidity in low and high risk term
pregnancies in the Netherlands: prospective cohort study. BMJ.
2010;341:c5639.

2. De Jonge A, Mol BW, Goes v/d B, Nijhuis J, Post v/d J, Buitendijk S.:
Too early to question effectiveness of Dutch maternity care system. BMJ
Rapid response November 2010.

3. Netherlands Perinatal Registry. Perinatal Care in the Netherlands 2007.

This remarkable study of Annemieke C C Evers et al. (1) demonstrated
that deliveries at home or in primary care of pregnant women at "low risk"
supervised by midwifes had a higher risk of perinatal death compared to
deliveries of women at high risk supervised by obstetricians in hospital
centres.

However, the low risk for pregnant women does not exist. Indeed, many
events like abnormalities of foetal heart rate, prolapse of umbilical cord
occurring during labour are totally unexpected in most cases and need
emergent cesarean deliveries (2). Unfortunately, the time of transfer of
the mother and her baby from home to hospital (secondary care) is often
too long to avoid a foetal ischemic encephalopathy or to save the life of
the newborn. Indeed, in many cases the prolonged diagnosis-delivery time
will severely affect the cerebral prognosis of the newborn, e.g. in
prolapse of umbilical cord (3) or in foetal bradycardia (4, 5).

Delivery at home is a veritable flashback of obstetrics as the conditions
of quick delivery in emergent situations at risk of foetal asphyxia on one
side and medical environment to manage them on the other are not met.
Although some installations should be improved in maternity hospitals to
make the deliveries more intimate and cordial, deliveries at home should
be reconsidered as they could compromise the life of newborns.

References

1. Evers AC, Brouwers HA, Hukkelhoven CW, Nikkels PG, Boon J, van
Egmond-Linden A, Hillegersberg J, Snuif YS, Sterken-Hooisma S, Bruinse HW,
Kwee A. Perinatal mortality and severe morbidity in low and high risk term
pregnancies in the Netherlands: prospective cohort study. BMJ. 2010 Nov
2;341:c5639. doi: 10.1136/bmj.c5639.

2. Lagrew DC, Bush MC, McKeown AM, Lagrew NG. Emergent (crash)
cesarean delivery: indications and outcomes. Am J Obstet Gynecol. 2006
;194:1638-43; discussion 1643.

3. Alouini S, Mesnard L, Megier P, Lemaire B, Coly S, Desroches A.
Management of umbilical cord prolapse and neonatal outcomes. J Gynecol
Obstet Biol Reprod. 2010 ;39:471-7.

4. Leung TY, Chung PW, Rogers MS, Sahota DS, Lao TT, Hung Chung TK.
Urgent cesarean delivery for fetal bradycardia. Obstet Gynecol. 2009
;114:1023-8.

5. Kamoshita E, Amano K, Kanai Y, Mochizuki J, Ikeda Y, Kikuchi S,
Tani A, Shoda T, Okutomi T, Nowatari M, Unno N. Effect of the interval
between onset of sustained fetal bradycardia and cesarean delivery on long
-term neonatal neurologic prognosis. Int J Gynaecol Obstet. 2010 ;111:23-
7.