"not NCT classes or pregnancy books which are all a bit lightweight"
Actually the NCT is a great resource for evidence based information. This is an evidenced based briefing on research into homebirth (or at least as much as I could C&P):
Research
Evidence Review
20 New Digest 40 ? October 2007
Safety of planned home birth:
an NCT review of evidence
by Gill Gyte,NCT antenatal teacher and research associate for
Cochrane Pregnancy and Childbirth Group, and Miranda Dodwell,
NCT antenatal teacher and co-creator of BirthChoiceUK
This paper provides a review of the evidence
on the safety of planned birth at
home compared with planned birth in hospital,
looking in particular at the evidence
for women at low risk of complications during
labour. It examines the research papers
identified by the NICE Intrapartum Care
Guideline Development Group (IPC GDG)
for the chapter on place of birth in the NICE
Intrapartum Care Guideline.1,2 A group of
NCT staff and research networkers
reviewed the identified papers on home
birth, using the NICE methodology for
assessing the risk of bias,3 to inform the
NCT?s position on the NICE guideline and
recommendations. The NCT group reached
different conclusions from the IPC GDG.
This paper, written by two members of the
NCT review group, provides the NCT?s
assessment of the evidence.
The paper addresses the following
questions:
What evidence is available on the comparative
safety of planned home birth in
terms of babies dying during labour or
shortly after birth (perinatal mortality
(PNM))?
How reliable is the evidence?
What does the evidence tell us about
the safety of planned home birth for
women at low risk of complications?
This paper does not address important
questions about morbidity for either the
baby or the mother, nor important questions
around well-being and psychological
outcomes linked to place of birth. The
authors are planning to undertake a full
systematic review of home birth addressing
a broader range of benefits and risks.
Background
The NICE Intrapartum Care Guideline, published
in September 2007, includes a chapter
on place of birth which considers the
evidence on the comparative safety of
planned home birth and planned hospital
birth. The draft guideline was circulated for
consultation to stakeholders in June-August
20061 and the chapter on place of birth was
circulated for a second consultation in
March-May 2007.2 Staff and volunteers at
the NCT decided to review the identified
papers on home birth to inform the NCT?s
position on place of birth, in preparation for
publication of the NICE guideline, when the
organisation?s views would be sought and
NCT workers would need guidance.
Ever since the Peel report recommended
that ??sufficient facilities should be provided
to allow for 100 per cent hospital
deliveries?, claiming that, ??the greater
safety of hospital confinement for mother
and child justifies this objective?,4 there has
been controversy over the evidence on
?safety? of home birth. In the 1970s, Marjorie
Tew began work which established that
studies which assessed comparative perinatal
mortality (PNM) were misleading
when the home birth group included data
for unbooked, unintended home births.5 It is
now widely accepted that unplanned, particularly
unattended, home births have a
much higher mortality rate than planned
home births6-10 and it is, therefore, now
acknowledged in the literature on home
birth that care needs to be taken in defining
terms and distinguishing between planned
and unplanned home births.
The NICE IPC guideline
The NICE IPC GDG assessed the evidence
on the comparative safety of planned home
birth and planned hospital birth. As part of
the IPC guideline development process, a
literature search of comparative home birth
studies published in English was undertaken.
1 This initially identified two systematic
reviews11,12 and 17 papers from 16 studies.13-29
All the English language papers contained in
the two systematic reviews were included
among the 17 studies identified.
A further paper30 from one of the studies
was identified later.2 In addition, at the
request of the IPC GDG, the National
Collaborating Centre for Women?s and
Children?s Health (NCC-WCH) conducted an
analysis of data from England and Wales on
intrapartum related perinatal mortality
(IPPM).31 A ten year period of data collection
was chosen to provide sufficient power to
assess IPPM. As the paper had not been
published, it was provided as Appendix D in
the second consultation document.2
The IPC GDG reviewed these 19 papers
from 17 studies. It assessed each paper in
terms of the criteria set by NICE, which
assigns research papers an evidence level
(EL) according to the type and quality of the
study (Figure 1).1 In addition, the IPC GDG
established its own validity criteria, published
in Appendix C of the second consultation
document (Figure 2)2 These are
described in more detail below.
There are a number of methodological
challenges affecting the study of the safety
of home birth, so this grading of evidence
is very important. The following section
covers how this grading was addressed by
the IPC GDG.
NICE and IPC GDG grading of the
evidence
When considering the effectiveness of
healthcare treatments, interventions or ?packages
of care?, well-conducted randomised
controlled trials (RCTs) are the most reliable
way to compare two or more alternatives,
because they produce two groups with similar
baseline characteristics.RCTs are given a
NICE evidence level grade of 1.
For interventions where randomisation is
difficult or unethical, non-randomised comparative
studies can sometimes be used.
However, without randomisation, it is
impossible to achieve groups that are similar
in every respect except for the intervention
or package of care being studied (in
this case, planned home birth and planned
hospital birth). Where there are differences
- known or unknown - between the groups
being studied, these may affect the outcomes
in the groups as much, or more
than, the different packages of care. These
differences may be socio-demographic
(e.g. age, education, social class), or clinical
(medical or obstetric history).
Differences in any of these can introduce
?bias? or ?confounding? which may affect the
reliability of the comparison and thus the
conclusions of the research.32-5 To ensure
as much similarity as possible, steps need
to be taken to try to balance the background
risk factors of the two groups being
compared. This is usually attempted by
selecting individuals in the group of primary
interest (in this case women at low
risk of complications choosing a home
birth) and matching them with one or more
people from the comparative group (in this
case women at low risk of complications
choosing hospital birth) using a number of
key known characteristics. Alternatively,
or in addition, sophisticated statistical
adjustments can be made according to
the variations in risks identified between
the two groups.35 Neither of these methods
can be relied on to make full adjustment
of known confounders, and there is
no way of adjusting for any unknown confounders
(except through good randomisation),
so non-randomised studies, such
as these, are given a NICE evidence level
grade of 2.
A study?s considered susceptibility to
bias is recognised in the grading systems
by the use of ?++? , ?+? or ?-? in the NICE evidence
level allocation, with ?++? representing
studies with very low risk of bias, ?+?
representing studies with low risk of bias
and ?-? representing studies with high risk
of bias. NICE includes these symbols for
evidence levels 1 and 2 to further clarify
the quality of the study (Figure 1).
In addition, the IPC GDG devised its
own scoring system for overall validity
involving a combination of external and
internal validity (Figure 2) where ?internal
validity? measured the risk of bias in a way
similar to the NICE criteria and ?external
validity? was related to the relevance of the
research to women in the UK. External
validity was graded highest with studies
conducted in the UK since 1980, and older
studies or those conducted outside highincome
countries were ranked lowest. (In
the final version of the IPC guideline, published
on 26 September 2007,36 this combined
internal/external assessment scoring
system was replaced with assessment
of internal validity only).
Another factor which can affect the reliability
of research findings is the size of
the study. A large number of ?events? are
needed (?an event? being PNM in this
case) if any difference in outcomes is to be
attributed to one of the packages of care
with any degree of certainty in a statistical
analysis. Since adverse outcomes are
rare for healthy women with a straightforward
pregnancy (the group for whom
home births are considered most suitable),
37 studies have to be sufficiently powered
(i.e. large enough) to answer questions
about the comparative safety of
home birth.38,39 In addition, studies may be
undertaken retrospectively (collecting data
from the past) or prospectively (setting out
to collect future data). Prospective studies
can more easily gather data on potential
confounding factors and so more easily
balance the two groups being compared.
Only one very small RCT has ever been
published comparing the outcomes for
planned home birth with planned hospital
birth. This was a small pilot study of 11
women to see if women at low risk of
complications would be willing to be randomised
in a trial on home birth.22 This
concluded that it was unlikely that sufficient
numbers of women would be willing
to be randomised to assess the comparative
PNM adequately. Therefore, most
studies comparing the safety of planned
home birth with planned hospital
birth have been done using non-randomised
studies.
Consequently, there are considerable
difficulties inherent in conducting and interpreting
research on the safety of place of
birth. The grading of evidence itself can be
subjective and hence the NCT decided to
undertake its own review of the evidence
relating to the safety of planned home birth
for women at low risk of complications.
IPC GDG information on the safety
of home birth
The IPC GDG assessed the outcomes from
their included studies. For the first consultation,
the main outcome measure to
assess safety was PNM. In the UK, PNM is
usually defined as the number of stillbirths
(after 24 weeks gestation) and early neonatal
deaths (those occurring less than seven
days after birth) per 1,000 live births and
stillbirths.10 The World Health Organisation
(WHO) has a different criteria for registering
stillbirths, namely loss occurring after 22
weeks gestation. Thus PNM measured with
the WHO definition is greater than that
measured with the UK definition.10
Individual research papers may have used
their own definitions.
In the second consultation, a new outcome
measure was introduced, the intrapartum-
related perinatal mortality (IPPM).
This was defined in Appendix C as
??deaths from intrapartum ?asphyxia?,
?anoxia? or ?trauma? derived from the extended
Wigglesworth classification3?This
includes deaths and stillbirths in the first
week of life.?2 This measure is subjective
and classification may vary.
The IPC GDG reported in the second
consultation document2 that six of the 17
studies met the inclusion criteria.19,20,23,27,28,31
(there were seven studies in the final published
guideline with the paper by Dowswell
also included.22) However, only four of these
studies were reported to provide data on
PNM and IPPM19,23,27,31 although two of the
other studies did report the number of
babies who died.20,28 The inclusion and
exclusion decisions in the second consultation
document are summarised in Table 1.
Methodology
In February 2007, a group of six NCT staff
and research networkers (members of the
NCT with an interest in research) agreed
to review the research papers on home
birth, as published in the first consultation
draft of the NICE IPC guideline.1 Two people
assessed each paper independently
using the NICE methodology (Figure 1).3
The six people then attended a meeting to
discuss the papers and agree the assignment
of the type of study and the level of
evidence, using the NICE guideline
methodology (Figure 1).3 In April 2007, a
smaller group met to assess the additional
studies identified in the second IPC stakeholder
consultation draft.2 Overall, the NCT
group reviewed and graded the same 19
papers from 17 studies as the IPC GDG.
Research
Evidence Review
22 New Digest 40 ? October 2007
Outcomes assessed
After grading the 19 papers from 17 studies,
the NCT group looked at reported PNM
rates in the groups who had planned home
births compared with the groups who had
planned hospital births. Where available,
the IPPM rates were also compared. The
group looked, in particular, at the evidence
for women at low risk of complications during
labour.
Results
Assessment and inclusion
of studies
The NCT group assessed the 17 studies
differently from the IPC GDG (Table 1). The
NCT excluded: two case series studies
because they had no comparative data;16,21
one study that looked at actual place of
birth14 and also the one small RCT because
it did not report PNM.22 In addition, the
study by Tew looked at out-of-hospital birth
(home births plus GP unit births) compared
with in-hospital birth and was, therefore,
addressing a slightly different question and
so was omitted.26 This left the NCT group
with 12 studies for detailed reporting.
15,17,19,20,23,24,25,27,28,29,30,31 The NCT group further
divided these studies based on:
a) the assessed level of risk of confounding
bias (2++, 2+ or 2-)
b) whether they considered women at low
risk of complications
c) whether they were conducted in the
UK.
The four studies considered to have a low
risk of bias (but still with some risk of bias,
EL 2+) are detailed in Table 2.19,24,25,28 Three
of these considered women at low risk of
complications24,25,28 with one study being
conducted in the UK.25 The eight studies
considered to have a high risk of bias (EL
2-) are detailed in Table 315,17,20,23,27,29,30,31 of
which only two considered women at low
risk of complications.15,20 The studies in
Table 3 were included in the NCT review to
show the judgements made by the NCT
group about the quality of the evidence.
However, these eight studies were considered
to have too high a risk of bias for the
PNM data to be compared.
Findings
Reasonable quality evidence for
non-randomised studies - women
with low risk of complications
There were three studies of reasonable
methodological quality for non-randomised
studies (EL: 2+) addressing safety for
women at low risk of complications,24,25,28
one of which was undertaken in the UK.25
All these studies were underpowered for
assessing comparative PNM and still had
some risk of bias. They are described in
Table 2a.
UK studies
The one UK study of reasonable methodological
quality for a non-randomised study
(EL: 2+) (which was excluded from the IPCGDG
review) was a prospective cohort
study involving just under 8,000 women. It
collected data on 61% of all the planned
home births in the UK during 1990, and
compared outcomes with a matched group
of women at low risk of complications planning
birth in hospital. Midwives identified
women for both groups at 37 weeks gestation,
but they often found it hard to find a
suitable hospital control for each home birth
they booked.25 Matching was considered
reasonable for some risk factors but the
home birth group had more women of higher
social class and more years in full-time
education. The NCT group considered this
study to have relative methodological
strengths in that it was prospective, focused
on low-risk women, was carried out in the
UK, and had some control for confounders,
but it also had some limitations including
some imbalances of background risk factors
and incomplete data collection. The
study found no statistically significant difference
in PNM, with five out of 4,665 babies
dying in the planned home birth group (1.07
per 1,000) and five out of 3,319 babies
dying in the planned hospital birth group
(1.51 per 1,000). However, the authors stated:
?We had recognised from the outset that
this study did not have the power to detect
any differences in perinatal death between
women intending home or hospital birth?It
is therefore essential that no conclusions
are drawn from the figures relating to perinatal
death.?
Non-UK studies
The non-UK studies of reasonable methodological
quality for non-randomised studies
(EL: 2+) were small studies but neither
identified any increased risk in terms of
PNM or IPPM in women at low risk of complications
choosing home birth.24,28 The
study in the Netherlands (excluded from the
IPC-GDG review) found four out of 1,140
babies died in the home birth group (3.5
per 1,000) and two out of 696 in the
planned hospital birth group (2.9 per
1,000).24 In the study from British Columbia,
Canada (included in the IPC-GDG review
but not used in their assessment of PNM),
again there was no significant difference in
PNM with three out of 862 babies dying in
the planned home birth group (3.5 per
1,000), one out of 743 in the matched
planned hospital birth group attended by a
physician (1.3 per 1,000) and no babies
dying out of 571 in the unmatched planned
hospital group cared for by midwives.28
Women with low and increased risk
of complications
There was one study of reasonable
methodological quality for a non-randomised
study (EL: 2+) addressing safety
for a combination of women at low and
increased risk of complications. This was a
small non-UK study carried out in Western
Australia, and found no difference in PNM
between the two groups though, like the
studies above, it was underpowered for
assessing this outcome (Table 2b).19
Unreliable evidence for assessing
comparative safety
The remaining eight studies were considered
by the NCT group to have a high risk
of confounding bias (EL: 2-), mainly
because they studied populations of
women and did not balance, or adjust, for
confounding factors like socio-demographic
differences, or medical or obstetric risk factors.
27,17,15,30,29,23,31,20 Five of these studies were
also carried out retrospectively, which often
means that less is known about the individual
characteristics of the woman involved.
In particular, these studies were unable to
address the question of the comparative
safety of planned home birth for women at
low risk of complications compared with a
similar low-risk population of women who
choose hospital birth, as the risk factors for
the women in the studies are not known in
any detail. Therefore, these studies are only
described briefly, and their PNM and IPPM
data is not reported (Table 3).
UK studies
Three of these studies were undertaken in
the UK.15,23,31 One small study (387 women)
23
Research
Evidence Review
New Digest 40 ? October 2007
provided no information on how the groups
were matched and the study had no baby
deaths.15 The other two studies had no balancing
for background risk factors (social,
medical or obstetric) and they addressed a
mixed risk group of women who had
planned a home birth.23,31 More detail is
given for these two larger, more recent,
studies as they were included in the IPC
GDG review but were not considered of
good enough quality by the NCT review
group to contribute to the evidence on comparative
safety of planned home birth.
The most recent UK study was a retrospective
case control study (EL: 2-) (included
in the IPC GDG review but unpublished
at the time of the second ?place of birth?
consultation). This study concluded that the
incidence of PNM and IPPM in planned
home births is very low, but reported that
IPPM appeared to be significantly higher
than rates for hospital birth for the period
1999-2003.31 This study attempted to estimate
the number of planned home births at
booking by taking the known number of
actual home births and adjusting for an
estimate of both unplanned home births
and transfers of care. It also relied on data
from two different, unlinked, sources
(CEMACH and the Office of National
Statistics) to calculate the IPPM rate for
births in different settings. The NCT review
group considered these assumptions and
extrapolations to be unreliable. For example,
the NCC-WCH study used the
assumption that unintended home births
are correlated with the number of home
births and used the figure of 50% calculated
from studies in the Northern Region of
the UK.23,21 However, Murphy showed that
unintended home birth rates are correlated
with total birth rates rather than total home
birth rates, and are in the region of about
0.3% of all births.6 As the home birth rate in
the Northern Region was about 0.6%, this
alternative assumption is compatible with
both studies. Using this new assumption,
together with an alternative calculation of
transfer rates, IPPM rates in planned home
births for 1999-2003 could be shown to be
no different from those in planned hospital
birth. Given the sensitivity of the analysis to
the range of unplanned home births and
transfer rates, this study was considered to
provide unreliable comparative data.
The UK Northern Regional Health
Authority study (EL: 2-) (included in the IPC
GDG review), was also considered to provide
unreliable comparative data because
of the type of study and the assumptions
made.23 This was a retrospective case control
study which took the number of babies
who had died in out-of-hospital births, both
planned and unplanned, and tried to estimate
how many of these had been planned
home births and how many planned home
births resulted in transfer to hospital. This
involved a number of calculations and
assumptions all of which carried a considerable
range of uncertainty. The authors
reported that the number of women planning
a home birth was hard to assess and
the transfer data even more difficult to
assemble. The study concluded: ?All we can
say with certainty is that of the 1,890
women who were estimated to have
booked for home delivery in this region in
the last ten years of the study period, only
five lost a baby and intrapartum events
were implicated in only one of those
deaths.? In addition, because half the
women who gave birth outside hospital in
this study were not booked for home birth,
the authors concluded that: ?A service
geared to cope with these unplanned
events ought to be able to deal with a proportion
of planned low risk deliveries.?
Non-UK studies
Five of the studies of poor methodological
quality for comparative assessments of
safety (EL: 2-) came from high-income
countries other than the UK. There was a
high risk of bias because there was no balancing
for background risk factors (social,
medical or obstetric) and they involved a
mixed risk group of women planning home
birth, including women with breech babies
and twins.30,17,27,29,20
The most recent study was a large
prospective cohort study (EL: 2-) (excluded
from the IPC GDG review) assessing outcomes
for 5,418 planned homebirth in
North America in 2000.29 This prospective
cohort study reported on a number of characteristics
of the women included but not
on their medical or obstetric risk factors.
This study showed planned home birth in
North America to be associated with low
PNM (2.0 per 1,000 excluding congenital
birth defects and 1.7 per 1,000, excluding
congenital birth defects, twins and breech
births). The study also reported on timings
and reasons for transfer during labour. The
authors stated that the main limitation of
their study was ??the inability to develop a
workable design from which to collect a
national prospective low risk group of hospital
births to compare morbidity and mortality
directly.?
A retrospective cohort study from the
early 1990s, (EL: 2-) (included in the IPC
GDG review), assessed the outcomes for
7,002 planned home births in Australia.27
This study had no information on the background
risk status of the women participating,
made no adjustments for differing risk
status in the outcome assessments but did
report on the causes of death for the 50
babies who died in planned home births.
The study reported a higher PNM for
planned home births in Australia (6.4 per
1,000 and IPPM 2.7 deaths per 1,000)
compared with planned home births in
other high-income countries. The study
attributed some of the higher PNM rate to
women at increased risk of complications
(e.g. twins and breech births) choosing home
birth. The study concluded: ?While home
birth for low risk women can compare
favourably with hospital birth, high risk home
birth is inadvisable and experimental.?
One study from Switzerland (excluded
from the IPC GDG review) created matched
pairs based on background characteristics,
but did not report baby deaths by these
matched pairs.20 The other two studies30,17
(both excluded from the IPC GDG review)
were from the USA both looking at mixed risk
status with The Farm study by Durand17 being
considered a unique setting not comparable
with current UK maternity care systems.
Discussion
Reliability of the evidence
The available evidence on the safety of
planned birth compared with planned hospital
birth is limited.The lack of rigorous evidence
is particularly marked when considering
data for women at low risk of complications.
In general, the studies are non-randomised
observational studies, all of which
have risk of bias and are too small to detect
differences in PNM and IPPM. Many were
undertaken outside the UK where maternity
care systems differ considerably from the UK.
Differences in assessment of evidence
between the IPC GDG and the NCT
review group
There are some significant differences
Research
Evidence Review
24 New Digest 40 ? October 2007
between the assessment of the evidence
by the IPC GDG and the NCT group. The
IPC GDG second consultation draft included
four studies in its assessment of
IPPM,19,23,27,31 three of which are identified in
this paper as having a high risk of confounding
bias because they did not balance
for risk factors.23,27,31 These included the
NCC-WCH analysis of the CEMACH data
specifically undertaken for the guideline.31
Balancing for background risk factors was
one of the criteria for internal validity set by
the IPC GDG, so it is unclear why these
studies were included in their assessment
of the evidence. By contrast, some studies
included in the NCT review of the evidence
were excluded by the IPC GDG assessment,
most notably the large UK study of
1997.25 It could be argued that there was
sufficient bias in this study to exclude it, but
those criteria would then also exclude three
of the studies in the IPC GDG review which
had greater risk of bias.23,27,31 This would
leave only one small study of sufficient
quality but assessing a mixed risk population
of women, and showing no evidence of
greater safety in hospital births.19
Overall, the NCT review group considered
that none of the studies were of high
enough quality to be considered as good
evidence of the comparative safety of
planned home birth and planned hospital
birth. However, there were four studies that
could be considered of reasonable quality
to provide some information about
PNM.24,25,28,19 of which three looked at
women at low risk of complications.24,25,28
These studies indicated that PNM was low
in both planned home birth and planned
hospital birth in women at low risk of complications.
None of these studies identified
any significant difference in the PNM
between women planning a home birth and
women planning a hospital birth, though all
were underpowered to assess this outcome.
Several of the authors themselves
stated that their studies were too small to
detect any differences.
Therefore, we concluded that there is no
evidence to suggest that hospital birth is
safer than home birth for low risk women,
that is, healthy women with a straightforward
pregnancy, when considering PNM or
IPPM. If the comparative safety of home
birth for women at low risk of complications
in the UK is to be properly assessed, more
rigorous, good quality prospective data are
needed. The Department of Health has
commissioned research to produce better
evidence on the safety of out-of-hospital
birth (see www.npeu.ox.ac.uk/birthplace).
Until good quality evidence about comparative
safety is available, the choice a
woman makes about where to give birth will
have to rely on other factors. However, the
likelihood of a baby dying is very low for
women at low risk of complications wherever
they choose to give birth.
Key points
The incidence of perinatal mortality
(PNM) and intrapartum related perinatal
mortality (IPPM) in the UK is very low,
with PNM around 8/100040 and IPPM
less than 1/1000 births.2
The quality of the comparative evidence
on safety of home birth is poor; however,
our assessment suggests that there
is no evidence that the risk of a baby
dying during or shortly after labour is
any higher in women at low risk of complications
choosing home birth compared
with women at low risk of complications
choosing hospital birth.
Women should be offered a choice of
place of birth and should be provided
with unbiased, evidence-based information
to help them decide what is right
for them.
Women should be supported by a high
quality maternity service that meets
their needs, irrespective of where they
choose to give birth, including good
transfer arrangements with the ambulance
service and the medical and midwifery
staff receiving women at the hospital.
Unintended home births carry a high
risk of mortality for the baby, and hospital
trusts and PCTs should provide sufficient
community-base midwives with
experience in home birth to provide
support and care promptly when
unplanned home birth occurs.
Glossary
Case control
A study that compares people with a specific
disease or outcome of interest (cases) to
people from the same population without
that disease or outcome (controls).
Case series
A study reporting observations on a series
of individuals, usually all receiving the
same intervention, with no control group.
Cohort
An observational study in which a defined
group of people (the cohort) is followed
over time. Because subjects are not allocated
by the investigator to different interventions
or other exposures, adjusted analysis
is usually required to minimise the influence
of other factors (confounders).
Cross-sectional
A study measuring the distribution of some
characteristic(s) in a population at a particular
point in time.
RCT
An experiment in which two or more interventions,
possibly including a control intervention
or no intervention, are compared by
being randomly allocated to participants.
Glossary entries modified from: Glossary,
Cochrane handbook for systematic reviews
of interventions, Version 4.2.5, Updated
May 2005. Available at: www.cochrane.
org/resources/handbook/index.htm
Notes
i Gill Gyte was one of the three ?women?s
representatives? on the IPC GDG, but
resigned in June 2007 as she could not
support the methodology used in the
systematic review on home birth.
ii In June 2007, NCT made a formal complaint
to NICE regarding the methodology
in this systematic review on home
birth in the IPC guidelines. The complaint
was partly upheld though the two
non-executive directors at NICE asked
to investigate the complaint were only
able to investigate process and not content.
References
- National Institute for Clinical Excellence. Intrapartum care:
consultation. 23 June 2006 - 29 August 2006. Available from:
guidance.nice.org.uk/page.aspx?o=333766
- National Institute for Health and Clinical Excellence.
Intrapartum care: second consultation (chapter 3 only).
Consultation dates 22 March - 3 May 2007. Available from:
guidance.nice.org.uk/page.aspx?o=PlaceOfBirth
Consultation
- National Institute for Health and Clinical Excellence.
Guideline development methods: information for national
collaborating centres and guideline developers.
Available from:
www.nice.org.uk/page.aspx?o=249088
- Standing Maternity and Midwifery Advisory Committee.
Domiciliary midwifery and maternity bed needs. Report of
the sub-committee. Chairman: Sir John Peel. London:
HMSO; 1970.
- Tew M. Safer childbirth: a critical history of maternity care.
New Digest 40 ? October 2007 25
Research
Evidence Review
London: Chapman and Hall; 1990.
- Murphy JF, Dauncey M, Gray OP, et al. Planned and
unplanned deliveries at home: implications of a changing
ratio.BMJ 1984; 288 (6428): 1429-32.
- Campbell R, Macfarlane A. Place of delivery: a review. Br J
Obstet Gynaecol 1986; 93 (7): 675-83.
- Campbell R, Macfarlane A. Where to be born? 2nd edition
Oxford: National Perinatal Epidemiology Unit; 1994.
- Rodie VA, Thomson AJ, Norman JE. Accidental out-of-hospital
deliveries: an obstetric and neonatal case control study.
Acta Obstet Gynecol Scand. 2002; 81 (1): 50-4.
10. Confidential Enquiry into Maternal and Child Health.
Perinatal mortality 2005: England, Wales and Northern
Ireland. London: CEMACH; 2007.
Available from:
www.cemach.org.uk/
11. Olsen O. Meta-analysis of the safety of home birth. Birth
1997; 24 (1): 4-13.
12. Olsen O, Jewell MD. Home versus hospital birth. Cochrane
Database of Systematic Reviews 1998, Issue 3. Date of most
recent amendment 19 May 2006.
Available from:
www.library.nhs.uk/Default.aspx
13. Mehl LE. Research on alternatives in childbirth: what can it
tell us about hospital practice? In: Stewart L, Stewart D, editors.
21st century obstetrics now. Marble Hill, MO: NAPSAC;
1977. pp. 171-207
14. CaplanW, Madeley RJ. Home deliveries in Nottingham 1980-
81. Public Health 1985; 99 (5): 307-13.
15. Shearer JM. Five year prospective survey of risk of booking
for a home birth in Essex.BMJ 1985; 291: 1478-80.
16. Ford C, Iliffe S, Franklin O. Outcome of planned home births
in an inner city practice.BMJ 1991; 303 (6816): 1517-9.
17. Durand AM. The safety of home birth: the farm study.
American Journal of Public Health 1992; 82 (3): 450-2.
18. Woodcock HC, Read AW, Moore DJ, et al. Planned homebirths
in Western Australia 1981-1987: a descriptive study.
Medical Journal of Australia 1990; 153: 672-8.
19. Woodcock HC, Read AW, Bower C, et al. A matched cohort
study of planned home and hospital births in Western
Australia 1981-1987.Midwifery 1994; 10 (3): 125-35.
20. Ackermann-Liebrich U, Voegeli T, Gunter-Witt K, et al. Home
versus hospital deliveries: follow up study of matched pairs
for procedures and outcome. BMJ 1996;313(7068):1313-8.
21. Davies J, Hey E, Reid W, et al. Prospective regional study of
planned home births.BMJ 1996; 313 (7068): 1302-6.
22. Dowswell T, Thornton JG, Hewison J, et al. Should there be
a trial of home versus hospital delivery in the United
Kingdom? BMJ 1996; 312 (7033): 753-7.
23. Northern Region Perinatal Mortality Survey Coordinating
Group. Collaborative survey of perinatal loss in planned and
unplanned home births.BMJ 1996; 313 (7068): 1306-9.
24. Wiegers TA, Keirse MJ, van der Zee J, et al. Outcome of
planned home and planned hospital births in low risk pregnancies:
prospective study in midwifery practices in the
Netherlands.BMJ 1996; 313 (7068): 1309-13.
25. Chamberlain G, Wraight A, Crowley P editors. Home births:
the report of the 1994 confidential enquiry by the National
Birthday Trust Fund.Carnforth, Lancs: Parthenon Publishing;
1997.
26. Tew M. Safer childbirth? A critical history of maternity care.
3rd edition London: Free Association Books; 1998.
27. Bastian H, Keirse MJ, Lancaster PA. Perinatal death associated
with planned home birth in Australia: population based
study.BMJ 1998; 317 (7155): 384-8.
28. Janssen PA, Lee SK, Ryan EM, et al. Outcomes of planned
home births versus planned hospital births after regulation of
midwifery in British Columbia.CMAJ 2002; 166 (3): 315-23.
29. Johnson KC, Daviss BA. Outcomes of planned home births
with certified professional midwives: large prospective study
in North America. BMJ 2005; 330 (7505): 1416-9.
30. Mehl LE, Peterson GH, Whitt M, et al. Outcomes of elective
home births: a series of 1,146 cases. J Reprod Med 1977; 19
(5): 281-90.
31. National Collaborating Centre for Women's and Children's
Health. NCC-WCH analysis to obtain the best estimate of
intrapartum-related perinatal mortality in England and Wales
Appendix D. In: Second consultation on chapter 3, Planning
place of birth, Intrapartum care. Available from:
guidance.nice.org.uk/page.aspx?o=417943
32. Reeves BC, van Binsbergen J, Van Weel C. Systematic
reviews incorporating evidence from nonrandomized study
designs: reasons for caution when estimating health effects.
Eur J Clin Nutr 2005; 59 Suppl 1: S155-S161.
33. Rochon PA, Gurwitz JH, Sykora K, et al. Reader?s guide to
critical appraisal of cohort studies: 1. Role and design. BMJ
2005; 330 (7496): 895-7.
34. Mamdani M, Sykora K, Li P, et al. Reader's guide to critical
appraisal of cohort studies: 2. Assessing potential for confounding.
BMJ 2005; 330 (7497): 960-2.
35. Normand SL, Sykora K, Li P, et al. Readers guide to critical
appraisal of cohort studies: 3.Analytical strategies to reduce
confounding. BMJ 2005; 330 (7498): 1021-3.
36. National Collaborating Centre for Women's and Children's
Health. Intrapartum care: care of healthy women and their
babies during childbirth. Clinical guideline. London: RCOG
Press; 2007. Available from:
guidance.nice.org.uk/
CG55/niceguidance/pdf/English
37. Vedam S, Kolodji Y. Guidelines for client selection in the
home birth midwifery practice. J Nurse Midwifery 1995; 40
(6): 508-21.
38. Chalmers I. Evaluating the effects of care in pregnancy and
childbirth. In: Chalmers I, Enkin M, Keirse MJ, editors.
Effective care in pregnancy and childbirth. Oxford: Oxford
University Press; 1989. pp. 3-38
39. Moore RA, Gavaghan D, Tramer MR, et al. Size is everything--
large amounts of information are needed to overcome
random effects in estimating direction and magnitude of
treatment effects. Pain 1998; 78 (3): 209-16.
40. Office for National Statistics. Key population and vital statistics:
local and health authority areas 2005. Series VS no 32,
PP1 no 28. Palgrave Macmillan; 2007. Available from:
www.statistics.gov.uk/statbase/Product.asp?vlnk=539&
More=N
Figure 1: Levels of evidence for intervention studies
From NICE Guideline Methodology 1 (reproduced with permission from the
Scottish Intercollegiate Guidelines Network [SIGN 2002])
Level of
evidence
Type of evidence
1++ High-quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk
of bias
1+ Well conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk
of bias
1- Meta-analyses, systematic reviews of RCTs, or RCTs with a high risk of bias*
2++ High-quality systematic reviews of case control or cohort studies
High quality case-control or cohort studies with a very low risk of confounding, bias or
chance, and a high probability that the relationship is causal
2+ Well-conducted case-control or cohort studies with a low risk of confounding, bias or
chance and a moderate probability that the relationship is causal
2- Case-control or cohort studies with a high risk of confounding, bias or chance and a
significant risk that the relationship is not causal*
3 Non-analytical studies (for example, case reports, case series)
4 Expert opinion, formal consensus
*Studies with a level of evidence ?-? should not be used as a basis for making a recomendation
(see section 7.4)
Figure 2: Criteria for strength of validity and inclusion/exclusion of studies for
the systematic review comparing planned home and hospital birth
From NICE IPC GDG second consultation 2
Strength of validity
++ + -
External validity Conducted in the UK
since 1980
Conducted in high-income countries
other than the UK since 1980
Any other studies
Internal validity Adequate randomised
controlled design
Any observational study with
planned places of birth with additional
adequate study design to
control background medical and/or
obstetric risks of women between
places of birth and/or relevant outcome
measures
Any other study
Total validity Any study in which
both internal and
external validity were
++
Any study in which either internal
and external validity were + but
neither were -
Any study in which
either internal or
external validity were
-
Research
Evidence Review
26 New Digest 40 ? October 2007
No. Study IPC GDG assessment NCT assessment
Type of study Validity
and evidence
level (EL) if
included
Included/
excluded for safety of
home birth
Type of study Evidence
level
EL
Detailed assessment
carried out
Yes/No
Included/
excluded in
safety of home birth
1 Mehl 197730
California, USA
1970-1973
Cross-sectional External ?-?
Internal ?-?
Total ?-?
Excluded
??these were comparing
actual home birth with actual
hospital birth with significantly
different backgrounds.? Too old,
not applicable to the UK.
Retrospective
cohort
EL: 2- Yes
NCT felt this study
did assess planned
place of birth
Excluded
No balancing nor adjustment
for differing risk factors.
Compared with all
California births for 1973.
Lay midwives involved.
2 Caplan 198514
UK 1980-1981
Cross-sectional External ?-?
Internal ?-?
Total ?-?
Excluded
?Use of actual place of birth
populations, rather than
planned birth ones, and the
lack of any controlling for background
risk of these two
groups.?
Retrospective
cohort
EL: 2- No
Compared actual
place of birth, not
planned place of
birth.
Excluded
3 Shearer 198515
UK 1978-1983
Cross-sectional External ?-?
Internal ?-?
Total ?-?
Excluded
?No control of background
obstetric risks was attempted.?
Prospective
cohort
EL: 2- Yes Excluded
High risk of confounding
bias as no information on
how the groups were
matched.
4 Ford 199116 Case series External ?-?
Internal ?-?
Total ?-?
Excluded
No control group
Case series EL: 3 No
EL: 3 (no control
group)
Excluded
5 Durand 199217
Tenessee, USA
1971-1989
Cross-sectional External ?-?
Internal ?-?
Total ?-?
Excluded
??these were comparing
actual home birth with actual
hospital birth with completely
different backgrounds.? In control
group ??low birthweights
and fetal deaths were deliberately
over sampled.?
Retrospective
cohort
EL: 2- Yes Excluded
High risk of confounding
bias due the way the control
group was obtained.
6 Woodcock 199419
Western Australia
1981-1987
Cross-sectional EL: 3
External ?-?
Internal ?-?
Total ?-?
Included Retrospective
cohort
EL: 2+ Yes Included
Some balancing of baseline
characteristics and
adjustment.
7 Ackermann-Liebrich
199620
Switzerland
1898-1992
Cohort EL: 2+
External ?-?
Internal ?-?
Total ?-?
Included
But PNM not discussed
because not reported in
matched pairs.
Prospective
cohort
EL: 2+ Yes Excluded
Some balancing of baseline
characteristics but
PNM not assessed by
matched pairs.
8 Davies 199621 Case series External ?-?
Internal ?-?
Total ?-?
Excluded
No control group
Case series EL: 3 No
EL: 3 (no control
group)
Excluded
9 Dowswell 199422
UK 1994
RCT External ?-?
Internal ?-?
Total ?-?
Excluded
No relevant outcomes
RCT EL: 1+ No
Did not assess
PNM
Excluded
10 NRPMSCG 199623
UK
1981-1994
Cross-sectional
population based
EL: 3
External ?-?
Internal ?-?
Total ?-?
Included Retrospective
case control
EL: 2- Yes Excluded
No balancing nor adjustment
for differing risk factors
social, medical or
obstetric risk factors.
11 Wiegers 199624
Netherlands
1990-1993
Cross-sectional External ?-?
Internal ?-?
Total ?-?
Excluded
Outcome reported was 'perinatal
outcome index' defined
by authors, and each relevant
clinical outcome was not
obtained.
Prospective
cohort
EL: 2+ Yes Included
Though main outcome was
'perinatal outcome index'
did report the numbers of
babies who died.
Table 1: Studies on home birth
New Digest 40 ? October 2007 27
Research
Evidence Review
No. Study IPC GDG assessment NCT assessment
Type of study Validity
and evidence
level (EL) if
included
Included/
excluded for safety of
home birth
Type of study Evidence
level
EL
Detailed assessment
carried out
Yes/No
Included/
excluded in
safety of home birth
12 Chamberlain 199725
UK 1994
Populationbased
cohort
External ?-?
Internal ?-?
Total ?-?
Excluded
?There were over 1,000
unmatched planned home
birth women, but these
women were included in the
analysis. Moreover, social
economic status and obstetric
backgrounds of these two
groups were reported as statistically
significantly different;?
No regression analysis
was used.?
Prospective
cohort
EL: 2+ Yes Included
A prospective matched
study of women at low
risk of complications.
Although the demographics
showed some differences,
some factors were
balanced between the
groups.
13 Tew 199826 Cross-sectional External ?-?
Internal ?-?
Total ?-?
Excluded No
Compared ?home
birth? + GP units
with hospital births
Excluded
14 Bastian 199627
Australia
1985-1990
Cross-sectional
population-based
EL: 3
External ?-?
Internal ?-?
Total ?-?
Included Retrospective
cohort
EL: 2- Yes Excluded
No balancing nor adjustment
for differing social,
medical or obstetric risk
factors.
15 Janssen 200228
Canada
1989-1999
Cross-sectional EL: 3
External ?-?
Internal ?-?
Total ?-?
Included Prospective
cohort
EL: 2+ Yes Included
Some balancing of baseline
characteristics
16 Johnson 200529
North America
2000
Case series External ?-?
Internal ?-?
Total ?-?
Excluded
No control group
Prospective
cohort
EL: 2- Yes
Included PNM rates
from other birth settings
Excluded
No balancing nor adjustment
for differing medical
or obstetric risk factors.
17 NCC-WCH 200731
UK 1994-2003
Cross-sectional
population based
EL: 3
External ?-?
Internal ?-?
Total ?-?
Included Retrospective
case control
EL: 2- Yes Excluded
No balancing nor adjustment
for differing social,
medical or obstetric risk
factors.
IPC GDG inclusion criteria:
Total validity: any study in which both internal and external validity were the most valid was regarded as the most valid, and any study in which either internal and external
validity were [+] and neither were [-] was also considered.
Internal validity: [++] = good RCT;
[+] = any study with adequate design to control background medical and/or obstetric risks of women reporting relevant outcomes;
[-] = study not reporting relevant outcomes and not meeting other validity criteria.
External validity: [++] = any study conducted in UK since 1980;
[+] = any study in high-income country since 1980 where no UK study available;
[-] = any other study
NCT inclusion criteria: all comparative studies (evidence levels 1 & 2) addressing the question of the safety of homebirth in terms of PNM or IPPM but only those graded
as [+] were used to assess the comparative safety of planned home and planned hospital birth.
Table 1 (continued):
Table 2: Studies of reasonable quality for non-randomised studies, but still with a risk of confounding bias
a)Women at low risk of complications
Study Description Quality of evidence PNM or IPPM outcome
Chamberlain
199725
UK 1994
Prospective
cohort study
Women at low risk of complications planning a home
birth at 37 weeks were compared with a matched
group of women planning a hospital birth, identified by
the midwife during the same time period.
Matching was on age, parity and obstetric history. Data
on unplanned home births was also collected.
There were 4,665 women in the planned home birth
group and 3,319 women in the planned hospital group,
with data also collected on 1600 unplanned home
births.
Graded as evidence level 2+.
It was analysed on both intention to
treat and by actual place of birth.
Sometimes midwives could not find
a matched control but were encouraged
to collect the data on the
home birth anyway. Thus there is a
discrepancy between the numbers
of planned home births and planned
hospital births. Matching was reasonable
except for social class
where the home birth group included
more women of higher social
class.
The perinatal mortality rate was no different between the two
groups.
Home: Five babies died out of 4,665 = 1.07 per 1000.
Hospital: Five babies died out of 3,319 = 1.51 per 1000
The authors concluded: ?It is clear that the size of the potential bias
arising from incomplete data collection is small where frequently
occurring outcomes such as transfers to hospital, induction of
labour, CS are concerned. Rare outcomes such as perinatal death
are invalidated?We had recognised from the outset that this study
did not have the power to detect any differences in perinatal death
between women intending home or hospital birth? It is therefore
essential that no conclusions are drawn from the figures relating to
perinatal death.?
Research
Evidence Review
28 New Digest 40 ? October 2007
Table 2a (continued):
Study Description Quality of evidence PNM or IPPM outcome
Wiegers 199624
Netherlands
1990-1993
Prospective
cohort study
Women at low risk of complications booking home birth
with a midwife at the time of booking compared with
women choosing hospital birth, controlling for parity
and social, medical and obstetric background.
In the planned home birth group, there were 471
women having their first baby and 669 having their second
or subsequent baby. In the planned hospital group,
there were 369 women having their first baby and 327
having their second or subsequent baby.
Main outcome was perinatal outcome index consisting
of 36 items.
Graded as evidence level 2+.
It was analysed by intention to treat.
Women were matched using a 'perinatal
background index'.
The perinatal mortality rate was no different between the two
groups.
Home: Four babies died out of 1,140 = 3.5 per 1000.
Hospital: Two babies died out of 696 = 2.9 per 1000
The study was small and underpowered to assess perinatal mortality
outcomes.
The authors concluded: ?The outcome of planned home births is at
least as good as that of planned hospital births in women at low
risk receiving midwifery care in the Netherlands.?
Janssen 200228
British Columbia,
Canada, 1989-
1999.
Prospective
cohort study
Women at low risk of complications planning home
birth with a midwife at 36 weeks were compared with
women with similar obstetric risk planning to give birth
in hospital with a physician or midwife. Women were
matched by age, lone parent status, parity, hospital
where the midwife had admitting privileges.
There were 862 women in the planned home birth
group, 743 women in the planned birth in hospital with
a physician group (matched) and 571 women in the
planned birth in hospital with a midwife group
(unmatched).
The study took place during the first two years of
implementation of midwifery in British Columbia and
authors reported: ?The rugged geography and mixed
weather conditions in Canada potentially present
unique challenges for home birth.?
Graded as evidence level 2+.
It was analysed on intention to treat
and on intended place of birth at the
onset of labour.
There were some variations in the
socio-demographic and pregnancy
related characteristics, and results
were adjusted for confounders. The
authors report problems in accessing
all the relevant data. The study
was small and underpowered to
assess PNM.
The perinatal mortality rate was no different between the two
groups.
Home: Three babies died out of 862 = 3.5 per 1000
Hospital with physician [matched]:
One baby died out of 743 = 1.3 per 1000
Hospital with midwife [unmatched]:
No babies died out of 571
The authors concluded: ?There was no increased maternal or
neonatal risk associated with planned home birth under the care of
a regulated midwife. The rates of some adverse outcomes were too
low for us to draw statistical comparisons, and on-going evaluation
of home birth is warranted.?
b) Populations of women with mixed risk of complications
Study Description Quality of evidence PNM or IPPM outcome
Woodcock
199419
Western
Australia 1981-
1987.
Retrospective
cohort study
All women who had planned birth at home, thus mixed
risk population, were traced from multiple sources and
compared with a matched group of women planning
birth in hospital.
There were 976 women in the planned home birth
group and 2,928 women in the planned hospital birth
group.
Women were matched by year of birth, parity, previous
stillbirth/death of a liveborn child, age, height and marital
status. Matching was not possible on socio-economic
status, smoking, obstetric and medial history.
Graded as evidence level 2+.
It was analysed by intention to treat.
However, women were not matched
on medical and obstetric risk factors
and analyses were subjected to
crude adjustment using logistic
regression.
The perinatal mortality rate was no different between the two
groups
Home: Five babies died out of 976 = 5.1 per 1000
Hospital: Twelve babies died out of 2928 = 4.1 per 1000
Crude odds ratios had a wide confidence interval showing the large
uncertainty in the results due to small numbers of events in both
groups. The study was small and underpowered to assess perinatal
mortality.
The authors concluded: ?Planned home births in WA appear to be
associated with less overall maternal and neonatal morbidity and
less intervention than hospital births.?
Table 3: Studies with high risk of confounding bias for a comparative estimate of PNM or IPPM
Study Description Quality of evidence PNM or IPPM outcome
NCC-WCH
200731
UK. 1994 - 2003.
Retrospective
case control
study
For a mixed risk population of women, IPPM for planned
home birth at booking was compared with planned hospital
birth.
There was no assessment of the risk status of women
included, and no assessment of the factors involved in any of
the baby deaths.
There were 96 IPPM events in the planned home birth group
and 4,991 IPPM events in the UK across this ten year period.
Outcomes were also reported for two five-year periods:
1994-1998 and 1999-2003.
The number of women who planned home births was estimated
by taking the number of actual home births, subtracting
an estimated number of unplanned home births and
adding an estimated number of transfers during pregnancy
and during labour. The percentage of unplanned home
births was estimated from three UK studies all carried out in
the Northern region between 1983 and 1993. Transfers were
averaged from four UK studies conducted between 1977 to
1994. Weighted means were calculated and sensitivity
analysis gave upper and lower ranges.
Provided data on IPPM over a ten year period
in the UK. Considered a poor quality
study for assessing comparative safety
because of high risk of confounding bias.
Graded as evidence level 2-.
There was no balancing for risk status or
confounding factors. The data used to estimate
the % of unplanned home births were
taken from small studies, all in one region
and over a different time period from the
study itself. Also, more importantly, the
authors estimated the unplanned home birth
as a % of all home births. A more accurate
estimate would have been to estimate the %
of unplanned home births as a % of all
births. This was shown to remain reasonably
steady over a 10 year period at around
0.3% of all births regardless of the changing
planned home birth rate.6
Across the ten year period, there was no statistically
significant difference in the IPPM between the planned
home births and all births.
The latter five year period did show a statistically significant
higher IPPM for home birth with the authors'
estimates. Using what NCT considered as more
appropriate estimates for unplanned home births and
transfers, calculations on the same data show no statistically
significant difference in IPPM, illustrating the
very poor quality of this data to address the safety of
planned home birth.
Authors concluded: ?Although those women who had
intended to give birth at home and did so had a generally
good outcome, those requiring transfer of care
appeared to do significantly worse?The potential for
confounding means the results of the present study
must be interpreted with caution.?
New Digest 40 ? October 2007 29
Research
Evidence Review
Table 3 (continued):
Study Description Quality of evidence PNM or IPPM outcome
NRPMSCG
199623
UK, Northern
Region. 1981-
1994
Retrospective
case control
study
For a mixed risk population of women, stillbirth and
neonatal deaths in planned and unplanned out-of-hospital
births in the Northern Regional health authority
compared with all births in the same health authority.
There was no assessment of the risk status of women
included.
There were 134 baby deaths outside hospital, and the
authors reported that data on the number of women
who had planned birth at home was hard to assemble,
and the estimates of the number of women who had
planned home birth but transferred during labour was
even more difficult to obtain.
Provided data on IPPM for home and hospital
births over a 14 year period in Northern England.
Considered poor quality study for assessing comparative
safety because of high risk of confounding
bias.
Graded as evidence level 2-.
There was no matching for risk factors and no
adjustment for confounders. Assumptions were
made and some estimates seemed imprecise.
Over the whole 14 years, the risk of death during delivery
or in the first four weeks of life in a baby of normal birth
weight and without a lethal malformation was higher in
those born to the small group of women who had booked
for home delivery.
However, during the last ten years of that period, when
the midwife was always the community lead professional,
mortality in this subgroup was lower in those booking for
home delivery. Neither difference was statistically significant.
Authors concluded: ?The perinatal hazard associated with
planned home birth in the few women who exercised this
option (