ARTICLES
Articles
WHO antenatal care randomised trial for the evaluation of a new
model of routine antenatal care
José Villar, Hassan Ba’aqeel, Gilda Piaggio, Pisake Lumbiganon, José Miguel Belizán, Ubaldo Farnot, Yagob Al-Mazrou,
Guillermo Carroli, Alain Pinol, Allan Donner, Ana Langer, Gustavo Nigenda, Miranda Mugford, Julia Fox-Rushby, Guy Hutton,
Per Bergsjø, Leiv Bakketeig, Heinz Berendes, for the WHO Antenatal Care Trial Research Group*
Summary
Background We undertook a multicentre randomised
controlled trial that compared the standard model of
antenatal care with a new model that emphasises actions
known to be effective in improving maternal or neonatal
outcomes and has fewer clinic visits.
Methods Clinics in Argentina, Cuba, Saudi Arabia, and
Thailand were randomly allocated to provide either the new
model (27 clinics) or the standard model currently in use (26
clinics). All women presenting for antenatal care at these
clinics over an average of 18 months were enrolled. Women
enrolled in clinics offering the new model were classified on
the basis of history of obstetric and clinical conditions. Those
who did not require further specific assessment or treatment
were offered the basic component of the new model, and
those deemed at higher risk received the usual care for their
conditions; however, all were included in the new-model
group for the analyses, which were by intention to treat. The
primary outcomes were low birthweight (<2500 g), pre-
eclampsia/eclampsia, severe postpartum anaemia (<90 g/L
haemoglobin), and treated urinary-tract infection. There was
an assessment of quality of care and an economic
evaluation.
*Other members listed at end of paper
UNDP/UNFPA/WHO/World Bank Special Programme of Research,
Development, and Research Training in Human Reproduction,
Department of Reproductive Health and Research, World Health
Organisation, Geneva, Switzerland (J Villar MD, G Piaggio PhD,
A Pinol MSc); National Guard King Khalid Hospital, Jeddah, Saudi
Arabia (H Ba’aqeel MD); Khon Kaen University, Khon Kaen, Thailand
(P Lumbiganon MD); Centro Rosarino de Estudios Perinatales,
Rosario, Argentina (J M Belizón MD, G Carroli MD); Hospital Gineco-
Obstétrico “América Arias”, Havana, Cuba (U Farnot MD); Ministry
of Health, Riyadh, Saudi Arabia (Y Al-Mazrou MD); University of
Western Ontario, London, Ontario, Canada (A Donner PhD);
Population Council, Office for Latin America and the Caribbean,
Mexico (A Langer MD); Centro de Investigación en Sistemas de
Salud, Instituto Nacional de Salud Pública, Mexico (G Nigenda PhD);
School of Health Policy and Practice, University of East Anglia,
Norwich, UK (M Mugford DPhil); Health Policy Unit, London School of
Hygiene and Tropical Medicine, London, UK (J Fox-Rushby PhD,
G Hutton PhD); Department of Obstetrics and Gynaecology,
University of Bergen, Bergen, Norway (P Bergsjø MD); National
Institute of Public Health, Oslo, Norway (L Bakketeig MD); and
National Institute of Child Health and Human Development,
National Institutes of Health, Bethesda, MD, USA (H Berendes MD)
Correspondence to: Dr José Villar, Department of Reproductive
Health and Research, World Health Organization, 1211 Geneva 27,
Switzerland
(e-mail: villarj@who.int)
THE LANCET • Vol 357 • May 19, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Findings Women attending clinics assigned the new model
(n=12 568) had a median of five visits compared with eight
within the standard model (n=11 958). More women in the
new model than in the standard model were referred to
higher levels of care (13·4% vs 7·3%), but rates of hospital
admission, diagnosis, and length of stay were similar. The
groups had similar rates of low birthweight (new model
7·68% vs standard model 7·14%; stratified rate difference
0·96 [95% CI –0·01 to 1·92]), postpartum anaemia (7·59%
vs 8·67%; 0·32), and urinary-tract infection (5·95% vs 7·41%;
–0·42 [–1·65 to 0·80]). For pre-eclampsia/eclampsia the
rate was slightly higher in the new model (1·69% vs 1·38%;
0·21 [–0·25 to 0·67]). Adjustment by several confounding
variables did not modify this pattern. There were negligible
differences between groups for several secondary outcomes.
Women and providers in both groups were, in general,
satisfied with the care received, although some women
assigned the new model expressed concern about the timing
of visits. There was no cost increase, and in some settings
the new model decreased cost.
Interpretations Provision of routine antenatal care by the
new model seems not to affect maternal and perinatal
outcomes. It could be implemented without major resistance
from women and providers and may reduce cost.
Lancet 2001; 357: 1551–64
See Commentary page ??
Introduction
Antenatal care is perhaps the most common routine
medical activity. Components of antenatal care and the
timing of visits have mostly been introduced without
proper scientific evaluation,1,2 although several attempts
have been made lately to evaluate the number of antenatal
visits and the types and levels of care provision through
randomised controlled trials.3 To a large extent, less
developed countries have adhered to the antenatal
programmes of the more developed countries with only
minor adjustments. However, the care commonly consists
of irregularly spaced visits with long waiting time and poor
feedback to the women, with little communication
between antenatal-care clinics and delivery units.
Archie Cochrane wrote in 1972; “By some curious
chance, antenatal care has escaped the critical assessment
to which most screening procedures have been
subjected”. He further recommended that “the emotive
atmosphere should be removed and the subject treated
like any other medical activity and investigated by
randomised controlled trials”.4
Specifically, evidence was needed on whether a
programme of antenatal care that emphasises essential
elements of care that have been shown to improve selected
pregnancy outcomes is as effective as a traditional
1551
ARTICLES

15 small clinics in Argentina; four medium and eight

53 clinics agreed to participate
and were randomised
small clinics in Saudi Arabia; and 12 small clinics in
Cuba. A detailed description of the rationale and study
design has been published elsewhere.5
This trial was centrally coordinated by the
27 assigned new model of 26 assigned standard model UNDP/UNFPA/WHO/World Bank Special Programme
antenatal care of antenatal care of Research, Development, and Research Training in
Human Reproduction, WHO, Geneva, Switzerland. An
Women asked for consent Women not asked for consent independent external data and safety monitoring
committee reviewed monthly any cases of maternal death,
Women received standard
fetal death, or eclampsia. These events were recorded
Women Women did care as recommended in the because they are the most serious complications of
consented not consent clinic pregnancy and birth. The committee did not adopt formal
stopping rules before the beginning of the trial, but it
agreed that any increased rate in the new-model group of
Provided information for more than 20% compared with the standard-model rate
classifying form*
for any of the primary outcomes should be specifically
considered by the committee. Rules for the withdrawal of
Further study clinics owing to non-adherence to the protocol or
assessment very low recruitment rates were also adopted. However,
or referred there were no withdrawals among the 53 clinics that
for special agreed to enter the trial.
care A detailed protocol was prepared in English and
Spanish, supported by the Manual of Operations and the
Received Received Manual of Clinical Activities (also in English and Spanish),
basic standard care as which were used at the clinics and local study offices. A
component recommended checklist of clinical activities was followed during each
of new in clinic or visit of women receiving the basic component of the new
programme elsewhere model of antenatal care (figure 2). All data-collection
of antenatal forms were translated from English into Spanish and
care Arabic.
The protocol and supplementary documents and the
Figure 1: Study design funding for the trial were completed in June 1994. The
Antenatal-care clinics are the unit of randomisation (cluster preparatory phase started during November 1995, and
randomisation). randomisation and staff training in the intervention clinics
*Women who did not consent to participate in the trial were asked to
provide the information needed to complete the classifying form for were started by March 1996. The first woman was
baseline descriptive purposes only. enrolled, in Thailand, on May 1, 1996, and the last, in
Argentina, in April 1998. The completed data file was
“western” type of antenatal care in preventing maternal ready for analysis by September 1999.
and fetal morbidity. The costs of such programmes, and
care providers’ and women’s perception of the new Study population
antenatal-care model also needed assessment. We present The sites selected for the trial were the province of Khon
the results of a randomised controlled trial with these Kaen (Thailand) and the cities of Havana (Cuba), Jeddah
aims. (Saudi Arabia), and Rosario (Argentina). A fifth site was
Our primary hypothesis was that a new model of considered during the planning phase of the trial but not
antenatal care based on components shown to improve included because of the high rate of loss to follow-up
maternal, perinatal, and neonatal outcomes would be as during a pilot exercise. The eligibility criteria for clinics
effective as the traditional package in terms of specified were as follows. Each clinic had to be able to provide at
maternal and perinatal endpoints among singleton least 300 new patients in a period not longer than 24
pregnancies, cost, and acceptability to women and months. Intervention and control clinics had to be in the
providers. same geographical area, but serving distinct neigh-
bourhoods. Women from these clinics had to be traceable
Methods at delivery. The follow-up mechanism had to have access
Trial design and organisation to the hospitals to which all high-risk patients would be
The study was a multicentre randomised controlled trial referred. The clinics had to be part of a public (or semi-
in clinics in Argentina, Cuba, Saudi Arabia, and public) antenatal-care system. Military hospitals or social
Thailand. Antenatal-care clinics serving four well-defined security institutions were also eligible. The study did not
geographical areas were randomly assigned to the two include clinics at which direct fee-for-services payments
programmes of antenatal care (figure 1) within each of were required. All clinics had to have an antenatal-care
four study sites and within strata defined on the basis of system in place with norms and predefined activities that
clinic characteristics, strongly correlated with clinic size. were followed. The clinics had to be able to implement
Country-specific definitions of strata were used to new simple tests or activities as required by the protocol.
implement the randomisation. For each country, clinics Funds for these new activities had to be provided by the
were classified as small, medium, or large. These institution(s), because they were for direct care of patients
definitions were based on the number of new patients only. These few new activities replaced several presently
attending antenatal care per year, type of clinic (free- implemented. The clinics had to have an already working
standing or hospital-associated), and health-care system and economically supported minimum staff required for
to which they belonged. There were two large, six care of patients as per the protocol.
medium, and four small clinics in Thailand; two large and There were 12 antenatal-care clinics in each of three

1552 THE LANCET • Vol 357 • May 19, 2001

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 ARTICLES

BASIC ANTENATAL CARE CHECKLIST

CHECK THE ACTIVITIES CARRIED OUT WHERE APPROPRIATE (UNSHADED BOXES)

(Use the closest gestational age at the time of visit)

Patient’s name

Clinic record number

Study subject number / / /

Visit
FIRST VISIT for all women at first contact with clinics, regardless of gestational age. If first visit 1st 2nd 3rd 4th
later than recommended, carry out all activities up to the time. <12
Date: / / weeks

Classifying Form indicates eligibility for the basic component

Clinical examination
Clinically severe anaemia: haemoglobin test
Obstetric examination: gestational-age estimation, uterine height
Gynaecological examination (can be postponed until second visit)
Blood pressure
Maternal weight/height
Rapid syphilis test, detection of symptomatic sexually transmitted diseases – treatment
Urine test (multiple dipstick)
Blood type and rhesus status
Tetanus toxoid
Provide iron/folic acid supplementation
Recommendation for emergencies/hotline for emergencies
Complete antenatal card
SECOND and SUBSEQUENT VISITS Gestational age – approximate number of weeks:
Date: / / 26 32 38
Clinical examination for anaemia
Obstetric examination: gestational-age estimation, uterine height, fetal heart rate
Blood pressure
Maternal weight (only women with low weight at first visit)
Urine test for protein (only nulliparous/women with previous eclampsia)
Provide iron/folic acid supplementation
Recommendation for emergencies
Complete antenatal card
THIRD VISIT: add Date: / /
Haemoglobin test
Tetanus toxoid (second dose)
Instructions for delivery
Recommendations for lactation/contraception
FOURTH VISIT: add Date: / /
Detection of breech presentation and referral for external version
Complete ANC card, recommend it be brought to hospital
Staff responsible for antenatal care: Name

Signature

Figure 2: Antenatal-care checklist included in all medical records of women who attended clinics randomised to the new antenatal-
care model and who were considered eligible for the basic component of the new model

THE LANCET • Vol 357 • May 19, 2001 1553

For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
study sites and 17 in the fourth site that were eligible for
the study and where the health authorities agreed to let
the clinics be included in the trial.
All women attending prenatal care for the first time
after the start of the study at each of the selected clinics,
irrespective of their duration of gestation, medical or
obstetric characteristics, or previous antenatal care, were
enrolled in the trial. However, women subsequently
found not to be pregnant were excluded from all analyses.
Women enrolled for antenatal care at the control clinics
followed the standard procedure and delivered at the
usual hospitals (figure 1). Women enrolled in the clinics
assigned to the new-model group were treated by the new
basic component if they were classified as at low risk
according to a set of risk factors specified in the
classifying form. The remainder received any medically
required treatment or were referred if found to need
further assessment or treatment. These women delivered
at the usual hospitals also.
Trial procedures
The model in the control clinics was the antenatal care
currently offered, following guidelines formally
recommended by the local health authorities6–9 based on
the traditional western model. In general, the programme
was that women made visits once a month during the first
6 months, one every 2–3 weeks for the next 2 months,
and then one every week until delivery. Under ideal
circumstances, a woman booking early in pregnancy
would have about 12 visits. Clinical activities, urinary
tests, syphilis screening, haemoglobin measurement, and
blood-group typing were done routinely.
The new model assigned some women to routine
antenatal care on the basis of scientifically evaluated and
objective-oriented activities, called the basic component
of the new model. These were the women judged not to
need further assessment or special care at the time of the
first visit according to predefined risk criteria. The others
were given the care appropriate to any detected condition
or risk factor. Women who refused to receive the basic
component of the new model followed the standard
pretrial procedures in these clinics (figure 1).
At the first antenatal visit to new-model clinics, women
were classified as to whether or not they needed further
assessment or special care (eg, referral to a specialised
clinic). The classifying form contained 18 binary
responses (yes/no) covering obstetric history (previous
stillbirth or neonatal loss; history of three or more
consecutive miscarriages; last baby’s birthweight <2500 g
or >4500 g; hospital admission for hypertension, pre-
eclampsia, or eclampsia in last pregnancy; previous
surgery on the reproductive tract), present pregnancy
(multiple pregnancy; age <16 years or >40 years; Rh(-)
isoimmunisation; vaginal bleeding; pelvic mass; diastolic
blood pressure 90 mm Hg or more at booking), and
general medical conditions (insulin-dependent diabetes
mellitus; renal or cardiac diseases; known substance
abuse, including alcohol; and any other severe medical
disease or condition). Women with a positive response to
at least one of the questions were not eligible for the basic
component of the new model but remained in the
intervention group as randomised, receiving the care
corresponding to the detected condition.
The use of the classifying form is an integral part of the
new model, which had two components (the basic
component and further assessment and/or special care)
and was not considered as an entry criterion to the trial.
Therefore, no corresponding form was introduced in the
control clinics.
1554
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Activities included in the new basic programme fell
within three general areas: screening for health conditions
likely to increase the risk of specific adverse outcomes;
therapeutic interventions known to be beneficial; alerting
pregnant women to emergencies and instructing them on
appropriate responses. The activities distributed in four
visits are presented in figure 2. This checklist was
included in the medical records of all women classified
for the basic component of the new model in the
intervention clinics and followed up by their health-care
providers.5
Efforts were made to allow the clinics assigned the new
model to implement the recommended activities:
multiple dipsticks for urine tests were made available to
all these clinics in which routine urine culture was not
possible; iron and folic acid tablets were provided to those
clinics that did not have them available, to be given free of
charge to all women.10
Implementation of the intervention varied somewhat
from site to site. In Thailand, uterine height was recorded
only in the medical records, and a haemoglobin test was
done at the first visit, which was not in the protocol. In
Saudi Arabia, rhesus-negative women were not eligible
for the basic component of antenatal care, and routine
vaginal examination was not acceptable. In Argentina,
some physicians in the new-model clinics requested one
ultrasonographic assessment without medical indication,
which was not recommended. Also in Argentina, if the
woman had not delivered by week 41 of gestation, the
next visit took place at the antenatal clinic rather than in
the hospital, as required in the protocol, and some
physicians in one large clinic continued to do glucose
tolerance screening tests, although these were not
included in the new protocol.
Compliance with the basic component of the new
model was formally assessed by means of the antenatal-
care clinical checklist (figure 2). The checklist was
completed during this assessment both by the main care
provider and by direct observation by the trial’s clinic
supervisor or an external observer. Therefore, for each
woman enrolled in this exercise, two checklists were
completed for the same visit. All patients attending one
randomly selected day in each new-model clinic were
studied. Clinics were not notified in advance. A total of
481 antenatal-care visits were evaluated.
Outcome measures
The primary fetal/neonatal outcome was low birthweight
(<2500 g) among singleton births. The primary maternal
outcome was a maternal morbidity index,5 defined as the
presence of at least one of the following severe conditions
for which antenatal care is believed to be effective: pre-
eclampsia or eclampsia during pregnancy or within 24 h
of delivery (pre-eclampsia defined as hypertension and
proteinuria 2·0 g or more in 24 h or 2+ or more on
qualitative examination [dipstick]); severe postpartum
anaemia (haemoglobin <90 g/L); and treated urinary-
tract infection including pyelonephritis (defined as any
episode requiring antibiotic treatment or hospital
admission). Several other standard maternal and
perinatal events were considered as secondary
outcomes.11
Economic evaluation
An economic evaluation was undertaken alongside this
trial.12 Data were collected about costs borne by providers
and women. Information on the use of services was
derived from the trial summary forms. This information
was combined with unit costs, estimated in a separate
THE LANCET • Vol 357 • May 19, 2001

costing study, to calculate the total costs of care for each
model of care. Health-care providers’ costs per pregnancy
for each woman were calculated as those arising from
visits to outpatient antenatal care, including non-routine
visits, days of inpatient antenatal care, delivery (according
to type of delivery), days of postnatal care, and days of
neonatal specialist care. The costing studies estimated the
unit costs of each of these forms of service use within each
facility used by women in the trial in Cuba and Thailand.
Costs borne by women are those associated with
attendance at outpatient visits and were estimated from a
survey of a sample of women attending participating
clinics. All costs are expressed in US$ at prices on Jan 1,
1998, by use of purchasing power parity weights. Owing
to the differences in the economic circumstances in each
country, estimation of pooled costs for the whole trial was
not appropriate.13
Because funds were limited, costs could be estimated in
detail in all the trial clinics only in Cuba and Thailand.
Full reports of the costing exercises are available from the
economics researchers (MM, JF-R, GH).
Assessment of women’s and providers’ perception of
quality of care
This component aimed to compare the perception of the
quality of antenatal care (particularly satisfaction) and the
reasons behind it, in women attending both types of
clinic. It also explored the health-care providers’
perception of the two antenatal models.14 The assessment
was organised in two stages. In the first, an ethnographic
approach, including focus groups and in-depth interviews
with health personnel and women was used to assess the
frame of culturally related values in every country. These
findings were incorporated in the second stage
(quantitative) which used a standard, closed-ended
questionnaire to all providers of antenatal care (92 in the
new model, 82 in the standard model) in all clinics
participating in the trial and another questionnaire to a
random sample, stratified by clinic, of 790 women in the
new model and 748 in the standard model. Women were
included if they were at 32 or more weeks of gestation
and had had at least two antenatal-care visits before the
interview. All interviews took place during 1997, when
the trial was well established.
Design and randomisation
Sample-size formulae corresponding to a stratified cluster
randomisation design15,16 were applied to both the low
birthweight outcome and the maternal morbidity index,
both expected to occur in about 10% of cases. The
calculation took into account that the design involved
four sites, with clinics (clusters) within each site randomly
assigned the new or standard model of care. Further
stratification by clinic characteristics was expected to
have a conservative impact on power and was therefore
ignored for this purpose.
With the assumption of a constant number of clinics
per stratum, and clinic sizes taken as 300, 450, or 600
patients attending per year, the sample-size requirements
for the design were calculated with a two-tailed
significance level at =0·05 and power 1-=0·90 for three
values of the intervention odds ratio (1·16, 1·18, and
1·20) and for values of the intracluster correlation
coefficient () that ranged from 0 to 0·002. This range
was based on an estimate of 0·00065 from one of the
study sites. These calculations showed, for example, that
19 087 women in four study sites enrolled in 12 clinics
per site, each clinic recruiting 450 patients, would provide
power of 90% for detection of an intervention odds ratio
THE LANCET • Vol 357 • May 19, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
of 1·2 in a two-sided test with a level of significance of 5%
if  is 0·001. We chose the value 1·2 as the maximum
value of the odds ratio that would be regarded as
consistent with the conclusion that the new programme is
as effective as the standard programme, taking into
account the presumed increased costs and inconveniences
to women and families associated with the latter. With an
average outcome rate across control group sites of 0·10,
this decision implies that a rise in this rate to about 0·12 is
regarded as substantively important to detect. Sample-
size estimates with an equivalence approach were also
calculated.16 To establish the equivalence with a two-
sided confidence interval within a difference of 0·02 with
a power of 90%, and application of a design effect of 1·45
to account for clustering, about 17 000 women would be
needed.15 Both approaches showed that the target of
20 000 women for the trial provided sufficient power for it
to detect a relevant difference or demonstrate practical
equivalence, as defined above.
Nevertheless, the trial’s protocol stipulated that the
recruitment period should last at least 1 year to capture
all seasonal variations. We therefore had to continue
recruitment beyond the minimum number required.
The allocation schedule for random assignment of care
models to clinics was computer generated, including
stratification by study site and clinic characteristics, at a
central location (WHO, Geneva, Switzerland) by the
Statistical Unit of the UNDP/UNFPA/WHO/World
Bank Special Programme of Research, Development, and
Research Training in Human Reproduction. The
treatment allocation for each site was kept in Geneva
until the site had completed the basic introductory
training of study personnel (both in standard-model and
control-model clinics). When local investigators were
ready to implement the training workshops for the staff if
the clinic were assigned the new model, the study
statistician sent the treatment allocation by facsimile
directly to the principal investigator of the selected site.
The data-management system was the standard DMS2
system for microcomputers of the WHO/Special
Programme of Research, Development, and Research
Training in Human Reproduction. This system consists
of a comprehensive set of software that captures the data
collected in clinics and hospitals into electronic files
according to the WHO Good Clinical Practice guidelines.
Each country’s coordinating unit was responsible for data
entry and preparation of the country data file, following
the WHO system. Data files were sent to Geneva
regularly to consolidate the data master file and so that
overall monitoring reports could be produced.
The staff responsible for data collection after birth,
which included outcome variables, were unaware of the
group status of women in the study. In addition to
monitoring of possible bias in data routinely collected,
information on intrapartum events, partially unrelated to
antenatal care, such as emergency caesarean section and
forceps delivery, was collected.
The data-collection form was validated by the trial
supervisor, who completed a second data-collection form
for all deliveries taking place during one randomly
selected day in each of the study hospitals. This form was
additional to that completed for the study by the hospital
data clerk. Agreement between these two observers was
assessed by the percentage of agreement and the 
statistic adjusted for strata for qualitative variables and
the intraclass correlation coefficient17 for quantitative
variables. During the trial, 428 women in the new-model
group and 331 women in the standard-model group were
included in this analysis.
1555
ARTICLES

Statistical analysis Secretariat Committee for Research into Human Subjects,

Baseline tables compared the two study groups with
respect to cluster-level and individual-level risk factors.
Statistical analyses were by intention to treat and
accounted for the within-clinic correlation. However,
given the equivalence nature of the trial, efficacy analyses
were also done.
The primary unit of inference in this trial is the
individual patient, with the unit of randomisation (the
clinic) adopted only for practical reasons to facilitate the
implementation of the intervention. 95% CI (two-sided)
were constructed on the intervention effect odds ratios18
and also on the intervention effect rate difference.
Practical equivalence was defined a priori to exist for the
two primary outcomes if the upper limit of the 95% CI
was 1·2 or less. We controlled for prognostically important
baseline variables by the generalised estimating equations
approach.19 These variables were age (20 years, 21 to 30
years, 31 years), previous low birthweight, and nulli-
parity, as well as any variable showing substantial baseline
imbalance between the study groups.
Analyses were also done to explore the possibility of
effect modification on the primary outcomes involving the
strata (country and clinic size) and type of baseline
antenatal care, although the power to detect interaction
effects was low. Formal inferences (significance tests and
95% CI) were made only for the main analyses of primary
outcomes.
Ethics
This trial was approved by the Scientific and Ethical
Review Group of the UNDP/UNFPA/WHO/World Bank
Special Programme on Research, Development, and
Research Training in Human Reproduction, the WHO
and the Institutional Review Boards of the individual
participating centres and corresponding health authorities
of the regions where the trial was implemented.
The informed consent procedure was based on the
single-consent design proposed by Zelen.20 Thus,
informed consent was requested only from women
attending the antenatal clinics assigned to the new model;
women who refused were cared for according to the
standard practice in their clinic. However, such women
were counted in the intention-to-treat analysis as being
assigned to the new-model group. Women attending the
standard-model clinics received the protocols recom-
mended in each country, in the best format offered in
these clinics, which were updated when necessary to allow
satisfactory care. Staff of the standard-group clinics were
strongly encouraged to follow the recommended
antenatal-care guidelines of their country.6–9
Results
Characteristics of clinics and women
Most of the participating clinics were urban, and all
belonged to public-health systems. 39 were part of a
polyclinic and 14 were within a hospital. Most clinics were
of medium size with 33–38 pregnant women booking per
month.10 Overall, resources available were sufficient for the
implementation of adequate, basic, routine, western type
of antenatal care. Most of the clinics were able to refer
women to other levels of care within the same building.
Some were able to admit women in the same facility,
although most women in the study sites, except in
Thailand, delivered in large hospitals.
Physicians (specialists in obstetrics and gynaecology or
general practitioners) were available in all clinics for the

24 678 women enrolled in 53 antenatal-care clinics during study period

152 not pregnant

24 526 pregnant women

12 568 in 27 new 11 958 in 26 standard

model clinics model clinics

253 lost to follow-up 290 lost to follow-up

537 abortions 474 abortions

11 778 births 11 194 births

106 multiple births 73 multiple births

excluded excluded

11 672 single births 11 121 single births

Outcome
Low birthweight 138 missing 11 534 for analysis 81 missing 11 040 for analysis

Pre-eclampsia, urinary-tract infection 0 missing 11 672 for analysis 0 missing 11 121 for analysis

Postpartum anaemia 952 missing 10 720 for analysis 1071 missing 10 050 for analysis

Figure 3: Trial profile

1556 THE LANCET • Vol 357 • May 19, 2001

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 ARTICLES

provision of antenatal care. In Thailand, clinics had
midwives providing basic case and physicians were used
for referrals. Half of the clinics had special staff to obtain
blood or urine samples, and the other clinics were
routinely visited by staff from the central laboratories for
samples to be drawn or collected.
Data were obtained from a survey by interviewing staff
in all clinics and a random sample of women attending
antenatal care in these clinics shortly before
randomisation.10 The distribution of clinic characteristics,
location, number of new patients enrolled per month, and
resources available were very similar between clinics that
were subsequently randomised to the two models in the
trial.
27 clinics were assigned the new antenatal-care model
and 26 the standard model. All 53 clinics completed the
trial. There were 24 678 women in the four countries
attending the antenatal-care clinics for the first time
during the study period; all of them were enrolled in the
trial. 100 women in the new model (0·8%) and 52 (0·4%)
women in the standard model were found after their first
visits not to be pregnant and were excluded from all
analyses. Of the remaining women, 12 568 started care in
the clinics providing the new model and 11 958 women in
clinics assigned the standard model (figure 3). 2·0% of
women enrolled in the new model and 2·4% in the
standard model were lost to follow-up. The proportion of
abortions (spontaneous or induced) was similar in the two
study groups. The proportion of women who were not lost
to follow-up and had single births but completed the trial
without birthweight information was 1·1% in the new
model and 0·7% in the standard model (figure 3). The
proportions without data for postpartum anaemia were
7·6% and 9·0%, respectively. 164 (1·3%) women
attending clinics assigned the new model refused to
participate. These women followed the standard care in
the same clinics but were included in the new-model
group of the trial for the analysis. Multiple births were
excluded from the main analysis.
The professional level of the primary-care provider was
similar between the two trial groups within countries and
overall. For example, 61·7% of women had antenatal care
with an obstetrician/gynaecologist in the new-model group
compared with 57·1% in the standard model. The
respective proportions for general practitioners were
18·9% and 19·9%, and for midwives 19·1% and 18·8%.
However, the primary provider varied by country:
specialists in obstetrics and gynaecology provided most of
the routine antenatal care in Argentina and Cuba, general
practitioners provided most of the care in Saudi Arabia,
and professional midwives cared for almost all women in
Thailand.
The mean duration of gestation at entry to the trial was
16·5 weeks (SD 8·4) in the new model and 16·0 weeks
(8·0) in the standard model (table 1). Women enrolled in
the two trial groups were similar in terms of demographic
and obstetric characteristics. Nevertheless, there were
some differences in baseline variables of potential
prognostic importance, overall and within single countries.
These variables were: smoking during pregnancy,
education less than primary, hospital admission in the
preceding pregnancy for hypertension or pre-
eclampsia/eclampsia, any previous surgery on the
reproductive tract, and late booking (>28 weeks of
gestation) for antenatal care.
The proportion of women in the new model who did
not require further assessment or treatment after the first
visit as determined by the classifying form was 76·8%
(ranging from 69·3% to 88·5% in the four countries).
History of low birthweight, previous stillbirth, pre-
eclampsia in the preceding pregnancy, surgery of the
reproductive tract, vaginal bleeding, and severe medical
conditions in the present pregnancy were the most
common causes for further action.

The intervention
Characteristic New model Standard model
(n=12 568) (n=11 958)
Overall, women attending clinics assigned the new model
of antenatal care had a median of five (IQR three to six)
Demographic
Married or stable union 11 609 (92·4%) 10 841 (90·7%)
antenatal-care visits compared with a median of eight (five
Education less than primary 2204 (17·5%) 1882 (15·7%) to 11) in the standard model. In all countries, the number
Smoking during pregnancy 1308 (10·4%) 1495 (12·5%) of visits was lower in the new-model than in the standard-
Substance abuse 64 (0·5%) 41 (0·3%) model group (figure 4). 47·6% of women had fewer than
Ratio persons per room in the house* 2·4 (1·4) 2·4 (1·2)
Recruitment in 1997 8010 (63·7%) 7588 (63·5%)
five visits in the new model compared with 19·6% in the
Maternal age in years* 25·5 (6·0) 25·2 (6·0) New model
Outcome of previous pregnancy Standard model
15 =18
Abortion 2681 (21·3%) 2554 (21·4%)
Stillbirth 92 (0·7%) 74 (0·6%)
14
*

Low birthweight (<2500 g) 374 (3·0%) 384 (3·2%) 13

Neonatal death 66 (0·5%) 47 (0·4%) 12
Hospital admission for hypertension 215 (1·7%) 132 (1·1%) 11
Number of visits

or pre-eclampsia/eclampsia 10
9
Reproductive history
*

Any previous low-birthweight baby 715 (5·7%) 702 (5·9%)

8
*

Any previous surgery reproductive tract 227 (1·8%) 351 (2·9%) 7

Any previous abortion 4037 (32·1%) 3809 (31·9%) 6
*

Any previous stillbirth or neonatal loss 563 (4·5%) 464 (3·9%) 5

*

4
Present pregnancy
Isoimmunisation Rh() 143 (1·1%) 76 (0·6%)
3
Vaginal bleeding first trimester 445 (3·5%) 333 (2·8%) 2
Date of LMP unknown 641 (5·1%) 538 (4·5%) 1
Nulliparous 4661 (37·1%) 4665 (39·0%) 0
Primigravida 3310 (26·3%) 3353 (28·0%) All sites Argentina Cuba Saudi Arabia Thailand
Number of 4 5
Maternal height in cm* 156·8 (6·6) 156·4 (6·5)
66 10 16 593 54 721 42 17 52 74
Maternal weight at first visit in kg* 59·4 (12·6) 59·0 (12·2) women
11 11
2
3 3 8
2 2 23 17 32 30
Duration of gestation at first visit, in weeks* 16·5 (8·4) 16·0 (8·0)
Late booking for antenatal care† 1653 (13·2%) 1346 (11·3%) Figure 4: Total number of antenatal-care visits according to
LMP=last mentrual period. Data are numbers of women except: *Mean (SD). model overall and by study site
†>28 weeks of gestation at first visit. Boxes show interquartile range, with medians shown as heavy bars. Error
Table 1: Baseline characteristics of women at trial entry bars indicate 5th and 95th percentiles. *Indicates mean.

THE LANCET • Vol 357 • May 19, 2001 1557

For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES

New model
(n=11 672)
Referral to higher level of antenatal care
Any referral 1563 (13·4%)
First trimester 304 (2·6%)
Second trimester 641 (5·5%)
Third trimester 595 (5·1%)
Late referral ( 35 weeks) 236 (2·0%)
Hospital admissions for women referred 524/1563 (33·5%) 420/811 (51·8%)
Undiagnosed disorders at admission in labour
Fetal heartbeat absent 218 (1·9%)
Any condition 342 (2·9%)
Untreated syphilis 31 (0·3%)
Hospital admission
Any admission 1043 (8·9%)
More than one admission 182 (1·6%)
Emergency/unplanned 652 (5·6%)
Late hospital admission ( 35 weeks) 454 (3·9%)
Day-care (< 1 day) 77 (0·7%)
Stay 3–7 days 475 (4·1%)
Stay >7 days 306 (2·6%)
Hospital admission for clinical conditions 90/1043 (8·6%)
Data are numbers of women.
Table 2: Use of referral facilities and hospital admissions
standard model. Women who were classified as eligible for
the basic component in the new model had, as planned, a
median of four visits, and those identified at the first visit of
the new model to require further assessment or special care
had a median of six. 50·6% of women in the new model
eligible for the basic component had fewer than five visits
compared with 37·1% of those who required further
assessment.
Compliance with the activities included in the basic
component of the new model (figure 2) was assessed in a
random sample of 481 women. Overall, agreement between
the care provider and the external observer about each
recommended activity was above 83% in 11 of the 14
activities recommended for the first visit (n=204). During
the second visit (n=104), the corresponding figures were six
of the eight activities, during the third visit (n=107), eight
of 12 activities, and during the fourth visit, eight of the 12
activities (n=66). Activities for which there was lower
agreement were clinical examination for anaemia,
gynaecological examination, and recommendations for
emergencies, delivery, lactation, and contraception.
Selected practices were specifically promoted as part of
the new antenatal-care model. Overall, a higher proportion
of women received iron supplements in the new model than
in the standard model (85·6% vs 63·8%). This difference
was mostly due to the study site in Argentina, where iron
supplementation was provided to 85·0% of women in the
new model but only 20·6% in the standard model. In the
other study sites, the percentages were similarly high for
both models. The proportion with complete tetanus
immunisation was similar in the two groups (84% vs 85%).
Syphilis was treated more frequently in the new model than
in the standard model (1·1% vs 0·7%). Other sexually
transmitted infections were treated in 3·1% of women in
Standard model the new model and 2·8% in the standard model. The
(n=11 121) proportion with external cephalic version was similarly low
in the two groups.
811 (7·3%) To explore the possibilities that fewer antenatal-care
89 (0·8%) visits could lower the rate of detection or treatment of
241 (2·2%)
477 (4·3%)
relevant disorders, or that patients would move from
235 (2·1%) regular antenatal-care clinics to other outpatient
departments, emergency visits, or hospital admissions
(planned or otherwise), we studied the patterns of disease
170 (1·5%)
diagnosis and referral to other levels of care (including
471 (4·2%) hospitals). Higher proportions of women in the new model
47 (0·4%) than in the standard model were referred to a higher level of
antenatal care particularly during the first and second
860 (7·7%) trimesters of pregnancy (table 2). Women in the new model
156 (1·4%) were less likely than those in the standard model to be
577 (5·2%)
353 (3·2%)
referred with a diagnosis of pregnancy-induced
74 (0·7%) hypertension, severe anaemia, urinary-tract infection,
367 (3·3%) diabetes mellitus, or low uterine height measurement, but
254 (2·3%) they were more likely to be referred with a diagnosis of
80/860 (9·3%)
vaginal bleeding or rhesus isoimmunisation (data not
shown). Among those referred, the rate of admission was
lower in the new model than in the standard model (33·5%
vs 51·8%). Hospital admission during pregnancy overall,
late, emergency, and unplanned admissions, and length of
hospital stay were similar between the two groups (table 2).
The mean gestation at admission was (31·7 [SD 8·3] vs
31·5 [8·0] weeks). 13·3% of the women referred to a higher
level in the new model were referred back to the original
clinic, compared with 23·0% of women in the standard
model.
Agreement among staff responsible for data abstraction
and the trial supervisor for primary outcomes obtained
from 759 randomly selected records was more than 98%;
 statistics adjusted for strata were 0·93 or higher for
low birthweight and the maternal morbidity index. The
percentage of agreement for baseline characteristics and
secondary outcomes was more than 99% in most cases and
more than 94% in all cases.  statistics, calculated only for
those variables with frequencies between 5% and 95%,
were 0·79 or higher.
To investigate the possibility of bias in data extraction we
compared the rates of forceps and vacuum delivery,
intrapartum caesarean section due to failure to progress,
and caesarean section due to cephalopelvic disproportion
between models. Antenatal care should have had very
limited influence on these variables. The rates were almost
identical in the new-model and standard-model groups
(forceps or vacuum delivery 3·2% vs 3·4%, p=0·57;
intrapartum caesarean section 7·8% vs 7·6%, p=0·60;
intrapartum caesarean section due to failure to progress
2·3% vs 2·6%, p=0·24; and due to cephalopelvic
disproportion 1·5% vs 1·4%, p=0·85).
Outcome measures
The rate of low birthweight was very similar in the new-
model and standard-model groups. The stratified rate
difference was 0·96% (95% CI –0·01 to 1·92; p=0·06). The

Outcome New model Standard model Stratified rate difference Stratified summary Adjusted odds ratio
in % (95% CI)* odds ratio (95% CI)† (95% CI)‡
Low birthweight (<2500 g) 886/11 534 (7·68%) 788/11 040 (7·14%) 0·96 (0·01 to 1·92) 1·10 (0·95 to 1·27) 1·06 (0·97 to 1·15)
Pre-eclampsia/eclampsia 197/11 672 (1·69%) 153/11 121 (1·38%) 0·21 (0·25 to 0·67) 1·22 (0·92 to 1·60) 1·26 (1·02 to 1·56)
Postpartum anaemia 814/10 720 (7·59%) 871/10 050 (8·67%) 0·32§ 1·02§ 1·01§
Treated urinary-tract infection 695/11 672 (5·95%) 824/11 121 (7·41%) 0·42 (1·65 to 0·80) 0·90 (0·56–1·45) 0·93 (0·79 to 1·10)
*Adjusted for stratum effects by two-way ANOVA of cluster proportions. †Clustered Woolf method.18 ‡Generalised estimating equations: adjustment for strata and the baseline variables:
smoking during pregnancy, education less than primary, hospital admission in preceding pregnancy for hypertension or pre-eclampsia/eclampsia, gynaecological surgery before the index
pregnancy, previous low-birthweight baby, first-trimester vaginal bleeding, late booking (>28 weeks), maternal age (20 years or less, 21 to 30 years, 31 years or more), nulliparity. §Effect
was heterogeneous across sites and strata, therefore pooled estimates may hide site-specific effects. Confidence interval not shown because computational methods assume
homogeneity.
Table 3: Primary outcomes

1558 THE LANCET • Vol 357 • May 19, 2001

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 ARTICLES

Outcome Number of events Number of visits New model Standard model

New model Standard model Number Events Number Events
of women of women
Maternal (n=11 672) (n=11 121)
Pregnancy-induced hypertension 402 (3·4%) 554 (5·0%) Clinics at recommended western standards*
Pre-eclampsia 189 (1·6%) 144 (1·3%) Low birthweight 2852 204 (7·2%) 2721 183 (6·7%)
Hospital admission for pre-eclampsia 50 (0·4%) 32 (0·3%) Pre-eclampsia/eclampsia 2854 56 (2·0%) 2721 44 (1·6%)
Eclampsia 8 (0·07%) 9 (0·08%) Severe postpartum anaemia 2850 268 (9·4%) 2719 279 (10·3%)
Severe anaemia of pregnancy 514 (4·4%) 435 (3·9%) Treated urinary-tract infection 2854 204 (7·2%) 2721 253 (9·3%)
Hypertension with referral/treatment 272 (2·3%) 438 (3·9%)
Clinics below recommended western standards†
Hypertension without referral/treatment 131 (1·1%) 116 (1·0%)
Low birthweight 8682 682 (7·9%) 8319 605 (7·3%)
Vaginal bleeding second trimester 98 (0·8%) 54 (0·5%)
Pre-eclampsia/eclampsia 8818 141 (1·6%) 8400 109 (1·3%)
Vaginal bleeding third trimester 82 (0·7%) 68 (0·6%)
Severe postpartum anaemia 7870 554 (7·0%) 7331 597 (8·1%)
Any vaginal bleeding 378 (3·2%) 241 (2·2%)
Treated urinary-tract infection 8818 491 (5·6%) 8400 571 (6·8%)
Syphilis postpartum 31 (0·3%) 47 (0·4%)
Postpartum hospital stay 7 days 388 (3·3%) 382 (3·4%) *12 or more antenatal-care visits before trial: six clinics in the new model (median six
Caesarean section 1640 (14·1%) 1569 (14·1%) visits for study women); six clinics in the standard model (median 13 visits for study
Assisted vaginal delivery 428 (3·7%) 423 (3·8%) women). †Six to eight antenatal-care visits before trial: 21 clinics in the new model
Any breech presentation 406 (3·5%) 335 (3·0%) (median four visits for study women); 20 clinics in standard model (median seven visits
Vaginal breech delivery 52 (0·5%) 49 (0·4%) for study women).
Maternal death 7 (0·06%) 6 (0·05%) Table 5: Subset analysis for primary outcomes according to
Fetal-neonatal (n=11 534) (n=11 040) baseline number of antenatal-care visits
Small for dates* 1743 (15·2%) 1657 (15·1%)
Preterm delivery (<37 weeks) 910 (7·9%) 852 (7·7%) Adjustments for smoking during pregnancy, education
Very low birthweight (<1500 g) 121 (1·1%) 108 (1·0%) less than primary, hospital admission in the preceding preg-
Medically indicated preterm delivery 75 (0·7%) 74 (0·7%)
(<35 weeks)
nancy for hypertension or pre-eclampsia/eclampsia,
Medically indicated preterm delivery 74 (0·6%) 79 (0·7%) gynaecological surgery before the index pregnancy,
(35–36 weeks) previous low-birthweight baby, first-trimester vaginal
PROM (<35 weeks) 84 (0·7%) 68 (0·6%) bleeding, late booking (>28 weeks), maternal age, and
PROM (35–36 weeks) 67 (0·6%) 86 (0·8%)
Apgar score 1 min <7† 396 (3·5%) 353 (3·2%)
nulliparity did not change the patterns but produced
Apgar score 5 min <5† 21 (0·2%) 23 (0·2%) narrow confidence intervals (table 3). Thus, with account
Admission to neonatal intensive care >2 days† 617 (54%) 700 (6·4%) taken of the cluster-randomised design and adjustment for
Fetal mortality (n=11 672) (n=11 121) selected variables, we can say with 95% confidence that any
Fetal death 161 (1·4%) 119 (1·1%) increase in the odds (or roughly in the risk) of low
Fresh stillborn 99 (0·9%) 80 (0·7%) birthweight associated with the new model will not be
Macerated stillborn 62 (0·5%) 39 (0·4%)
Fetal death
36 weeks 122 (1·0%) 77 (0·7%)
higher than 15% (upper limit of the 95% CI for the odds
Fetal death >36 weeks 37 (0·3%) 42 (0·4%) ratio 1·15) and that for urinary-tract infection the risk
Fetal death with no congenital malformations 117 (1·0%) 85 (0·8%) increase will not be higher than 10% (upper limit of the
Fetal death with congenital malformations 19 (0·2%) 22 (0·2%) 95% CI for adjusted odds ratio 1·10).
Neonatal mortality The comparison between groups did not show any
Neonatal mortality day 1 only 31 (0·3%) 32 (0·3%) pattern in maternal and fetal or neonatal secondary
Neonatal mortality from day 1 until discharge 42 (0·4%) 39 (0·4%)
Perinatal mortality 234 (2·0%) 190 (1·7%)
outcomes in favour of either group (table 4). There were
Clinically detectable congenital malformations 101 (0·9%) 125 (1·1%) eight cases of eclampsia in the new-model group and nine
PROM=prelabour rupture of membranes. *n=11 440 for new model, 10 974 for in the standard-model group, and seven versus six maternal
standard model. †n=11 397 for new model, 10 934 for standard model. deaths. Overall, the rate of severe anaemia during
Table 4: Secondary outcomes pregnancy was similar in the two groups (stratified
summary odds ratio 0·96 [95% CI 0·48 to 1·90]). In
stratified summary odds ratio was 1·10 (95% CI 0·95 to the study site with the largest difference in iron
1·27); (table 3). The mean birthweight was 3120 g in both supplementation, the rate of anaemia during pregnancy was
groups. lower in the new model than in the standard model (1·7%
The maternal morbidity index occurred in 14·5% of vs 2·9%). Rates of fetal, neonatal, and perinatal mortality
women for the new model and 16·5% for the standard were similar in the two groups (all p values >0·05, table 4).
model (stratified summary odds ratio 0·98). During the There was, however, a higher rate of fetal death at 36
analysis of the trial, statistically significant qualitative weeks of gestation or earlier in the new model than in the
heterogeneity (p=0·012) among strata was found, mostly standard model (1·0% vs 0·7%) though it did not reach
related to severe postpartum anaemia. Although this index statistical significance (p=0·08). The rates were similar in
was one of the two primary outcomes selected in the the two models for fetal death after 36 weeks of gestation.
protocol, we judged that further analysis on an overall basis There were more small-for-gestational-age fetuses among
was inappropriate. The three components of the index were those who died at or before 36 weeks in the new model than
therefore analysed separately. in the standard model (40·2% vs 28·6%), but fewer among
The rates of severe postpartum anaemia (no p value those who died after 36 weeks (37·8% vs 47·6%). The
owing to heterogeneity) and urinary-tract infections mean gestational age at delivery for fetuses who died before
(p=0·95) were similar in the new model and the standard 36 weeks was 28 weeks in both models; among those who
model (table 3). However, the rate of severe postpartum died after 36 weeks, the mean gestational age was also
anaemia was substantially lower in the new model (8·8%) identical at 38 weeks.
than in the standard model (13·3%) for Argentina, where To explore a prestated secondary hypothesis,5 a subset
the largest difference in iron supplementation was achieved. analysis was done according to the characteristics of
Sensitivity analysis excluding data from Argentina antenatal care measured by the median number of visits in
suggested that the observed heterogeneity in postpartum the study clinics before initiation of the trial.10 The study
anaemia was mostly related to the large protective effect of site offering antenatal care in a recommended western
the new model in this study site. For pre- pattern was considered as a first subset and the other study
eclampsia/eclampsia, the rate was slightly higher in the new sites grouped as the second subset. There was no indication
model (p=0·63). of a differential effect associated with the new model among

THE LANCET • Vol 357 • May 19, 2001 1559

For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES

Type of cost New model Standard model Mean difference

(95% CI)
Number of women Mean (SD) Number of women Mean (SD)
Providers’ cost*
Cuba 2870 885·4 (1632·0) 2734 956·8 (1294·2) –71·4 (–148·8 to 2·5)
Thailand 3278 167·2 (144·7) 3091 206·1 (172·9) –38·9 (–46·3 to –30·9)
Women’s out-of-pocket costs*
Cuba 170 174·4 (470·2) 170 242·4 (174·4) –68·0 (–144·0 to 7·7)
Thailand 205 11·9 (16·7) 226 18·4 (27·1) –6·5 (–10·8 to –2·19)
Women’s time in access to care†
Cuba 170 15·9 (17·3) 170 25·0 (23·5) –9·1 (–13·5 to –4·7)
Thailand 205 14·8 (12·1) 226 29·7 (19·2) –14·9 (–18·0 to –11·8)
*Average cost per pregnancy 1998 US$ purchasing power conversion. The official exchange rates on Jan 1, 1998, were: $US1=1·00 Cuban Peso=52·3 Thai Bhat. The purchasing power
parity rate on Jan 1, 1998 was US$1=0·42 Cuban Peso=26·9 Thai Baht. These were calculated on the basis of costs of a basket of selected food items in each country. This allowed a
common purchasing power parity conversion method to be used.
†Average time per pregnancy in h.
Table 6: Costs to providers and women in Cuba and Thailand

women enrolled in the study site that had the largest services and the unit costs of each type of service. Costs
reduction in the number of antenatal-care visits (table 5). per pregnancy were higher in the standard-model clinics
We considered an important part of this equivalence than in the clinics offering the new model of antenatal
trial to be an analysis comparing subpopulations that care. The differences between the models in the mean
received the basic component of the new model, providers’ cost and in the women’s out-of-pocket costs
representing the most innovative part of the model, with were significant at p<0·05 in Thailand and at p<0·1 in
the women who would have been eligible for this basic Cuba. Typically, less than 25% of average costs to
component but who attended the control clinics. We providers are variable in the short term,21 and these costs
compared all women in the trial with the same variables can be taken as a proxy for marginal costs.22 These
included in the classifying form used in the new model, variable costs were lower for the new-model than for
but obtaining the data for both groups from the trial’s standard-model clinics in all countries, with reductions in
data collection form because there was no classifying form median costs of between 6% and 15%. The mean
in the standard-model clinics. 9314 (79·8%) women in difference in variable costs was significant at p=0·1 in
the new model and 8984 (80·8%) in the standard model Thailand but was not significant in Cuba (p>0·1).
would have been eligible for the basic component of the The costs per pregnancy in each country were
new model. Baseline characteristics were similar for these distributed differently between outpatient and inpatient
two subpopulations. No major change in the pattern of antenatal, intrapartum, postpartum, and neonatal care.
results was found. There were differences between countries, especially in
the costs of neonatal care. In Cuba, costs for neonatal care
Costs of health-care providers and women were high, accounting for over 20% of the total. In
Table 6 shows costs for the two countries in which Thailand, fewer babies were referred for neonatal
detailed information was available. Average providers’ intensive care, for a shorter time, and at lower average
costs per pregnancy reflect both the number of uses of the cost per day. Neonatal-care costs were less than 3% of the
total. The costs of outpatient antenatal care formed the
New model Standard model Stratified rate highest proportion of cost, but this proportion was less in
790 women, 748 women, (%) difference the new model. The amount of time women in the new-
92 providers* 82 providers* (95% CI)
model clinics spent during their pregnancies in accessing
Women’s survey care was significantly lower in the new model than in the
Number of visits right† 612 (77·6%) 649 (87·2%) –7·9 (–16 to 0·2)
Happy with spacing between 572 (73·2%) 625 (84·0%) –8·3 (–16·8 to 0·3)
standard model in the two countries (table 6, p<0·05).
visits†
Happy with waiting 639 (81·9%) 610 (82·1%) 0·7 (–7·4 to 8·8) Women’s and providers’ perception of antenatal care
time† In the qualitative study, variations in women’s general
Time with doctor/midwives 684 (86·7%) 598 (80·1%) 6·6 (–0·5 to 13·7)
about right†
ideas about pregnancy and their views and previous
Very satisfied with antenatal 318 (40·5%) 301 (40·7%) 0·4 (–8·6 to 9·3) experiences of care underlay the opinions they expressed.
care in this clinic† For health services and providers, the most important axis
Satisfied with antenatal care 460 (58·5%) 426 (57·6%) –0·1 (–9·1 to 8·8) of categorisation was modern versus traditional. Antenatal
in this clinic†
Would come back next 758 (96·7%) 703 (94·7%) 1·4 (–2·2 to 4·9)
services were generally seen as very satisfactory but some
pregnancy† concerns were raised—for example, anxieties originated in
Would recommend this clinic 760 (97·4%) 706 (95·0%) 1·6 (–1·4 to 4·7) changes of the standard patterns of care, especially the
to a relative or friend† number and spacing of visits, the need to improve the way
Information received 4·4 4·0 0·40 (0·13 to 0·67)
(score 0–6)‡
staff treat women, and the need to obtain better
Information on recognition/ 3·4 2·9 0·56 (0·06 to 1·06) information about issues such as nutrition and personal
procedure with pregnancy health.
problems (score 0–6)‡ Providers, on the other hand, showed strong country-
Providers’ census specific views, both positive and negative. In general,
Number of visits right 63 (68·5%) 53 (64·6%) . .§ although doctors were not opposed to the changes, they
Time spent with women right 79 (85·9%) 57 (69·5%) . .§
Information provided 5·6 (0·9) 5·2 (1·3) . .§
were keen to make sure that modifications in the model
did not limit their clinical control. In the country with the
*Owing to missing values, denominator ranges from 790 to 780 women in the new
model and from 748 to 740 women in the standard model.
largest reduction in the number of visits between the
†Number of positive responses. standard model and the new model, doctors expressed
‡Mean score 0–6, with 6 maximum information. satisfaction because they believe that the change will
§Not given because all providers were surveyed. reduce paperwork and allow them to spend more time
Table 7: Women’s and providers’ perception and satisfaction with their patients.

1560 THE LANCET • Vol 357 • May 19, 2001

For personal use. Only reproduce with permission from The Lancet Publishing Group.

The subsample of women in the quantitative study was
representative of the total population of enrolled women.
However, the former had on average slight advantages:
the duration of gestation at the first visit was lower (12·7
vs 16·2 weeks) and a smaller proportion had had a
previous low-birthweight baby (4·0% vs 5·8%). The
requirements of two previous visits and 32 weeks or more
of gestation as inclusion criteria for the survey could
account for these differences. Women surveyed were
similar between the new-model and standard-model
groups in terms of baseline variables.
Women in both groups of the trial expressed high
satisfaction with their care, as assessed by a direct
question about their satisfaction and two indirect
questions (“Would you come back to this clinic?” “Would
you recommend it?”). However, women in the new model
were less satisfied with the number and spacing of visits
than women in the standard model. On the other hand,
more women in the new model considered that the time
spent with doctors or midwives was about right. Women
were more satisfied with the information received from
providers (table 7).
About two thirds of providers were satisfied with the
number of visits they offered, in both models. More of
those in the new-model clinics than in the standard-model
clinics said that the time they spent with women at each
visit was right and were more satisfied with the
information received from providers (table 7).
Discussion
For both low birthweight and urinary-tract infections,
there was a strong indication that the two models are
equivalent; the upper limit of the 95% CI for the adjusted
odds ratios was below the equivalence limit (1·20)
prespecified in the protocol. For pre-eclampsia, the rates
were clinically similar, but an increase in risk of up to 56%
cannot be ruled out. For severe postpartum anaemia,
there was a large protective effect of the new model in the
country with the largest increment in the provision of iron
supplementation. Secondary outcomes suggest no
clinically important differences between the care models.
We used a cluster-randomised design, with a sample
size providing sufficient power to demonstrate practical
equivalence for the primary outcomes. Cluster
randomisation has been used in perinatal and HIV/AIDS
trials in less and more developed countries.23–25 All
analyses accounted for the within-clinic correlation that
would invalidate the application of standard statistical
methods. We used Zelen’s single-consent design as
adapted to cluster-randomisation trials. Few such trials
have been published,26,27 but this design is appealing for
trials that assess interventions at clinic level rather than for
therapeutic trials.
The four trial centres differed socioculturally and
economically. The baseline antenatal care offered in the
clinics ranged from a pragmatic adaptation of
recommended western routine visits, common in
developing countries’ urban health facilities and in public
health services in more developed countries, to almost
ideal antenatal care of the standard in the more developed
countries. These diversities provide good external validity
to the trial.
We did not use mortality as primary outcome. This
endpoint requires an unrealistically large number of
participants, but there is pathophysiological support for
the endpoints pre-eclampsia and eclampsia, infection, and
severe anaemia to be considered as in the causal pathway
to the need for intensive care, near-misses, and maternal
death.28
THE LANCET • Vol 357 • May 19, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
We disentangled the morbidity index, for which we had
a formal hypothesis of equivalence, because we found
heterogeneity of the intervention effect across strata.
Combined indices have been suggested for use in trials
when modest improvement or equivalence in rare
outcomes is expected; in our trial, the different
mechanisms and effectiveness of treatments made the use
of the maternal morbidity index unsatisfactory. After
disentangling the index, we could detect equivalence only
for urinary-tract infection. For pre-eclampsia/eclampsia,
of the three statistical methods used to compare rates
between models, the differences were not significant by
two; the adjusted odds ratio, which provides the narrowest
95% CI, was only marginally significant. The increase in
the risk of pre-eclampsia in the new model could be in
absolute terms, with 95% confidence, at most 0·67%
(table 3).
Although the new model may have missed more cases
of hypertension that subsequently developed into pre-
eclampsia than the standard model, we believe that this
possibility is unlikely. Urine protein tests were done on all
women, at all visits, in the new model, whereas in the
standard model they were done only at the first visit and
on women with hypertension. Therefore, there may have
been more likelihood of detection in the new model.
Furthermore, if there were more cases of early
undiagnosed pre-eclampsia in the new model, a higher
rate of related complications, such as eclampsia,
hypertension with referral and treatment, and hospital
admissions for pre-eclampsia could have been expected,
but this was not the case.
The definition of urinary-tract infection was intended to
capture only severe cases, but the high rates observed,
particularly in two study sites, could be related not only to
the high risk of these populations but also to detection and
treatment of less severe cases.
The intervention was largely implemented as planned,
including the initial selection of women for the basic
component of the new model. This simple tool identified
20% of women requiring a higher level of care on the basis
of history or present complications. The objective of this
selection was to call attention to conditions or diseases
that may need further assessment or follow-up at a
different level of care. The classifying procedure was not a
risk score, because each disease or condition was assessed
independently, and we did not produce a global risk
estimate. Furthermore, it includes elements of history,
medical conditions, and obstetric characteristics, most of
them present at the time of the first examination. Women
with a diagnosis of any of these disorders qualified
automatically for further assessment or treatment for
that particular condition, not necessarily related to
the frequency or content of routine antenatal
care (eg, nutritional interventions). The prediction of
intrapartum events was not considered because the
objective in this evaluation was to assess the type of
antepartum care.
The reduction in the number of visits in the new model
compared with the standard model was large and
clinically relevant, with one study site halving the number
of visits. The basic component of the new model is the set
of effective, goal-oriented activities implemented on a
four-visit schedule. There were, however, recommended
activities of the new model that were not fully
implemented. An example is external cephalic version, an
effective intervention to reduce the rates of breech
presentation and caesarean section.29 Provision of
information on effectiveness clearly is not sufficient to
change clinical practice. Furthermore, recommendations
1561
ARTICLES
to women for emergency situations were reported by
women less often than the protocol dictated, although
providers reported that they were implemented. Clinicians
and midwives may require special training and motivation
to include these activities in their routine, or perhaps they
should be given by other staff. More women in the new
model were referred for higher levels of care, perhaps
reflecting concern of clinicians about the longer time
between visits. This over-referral did not result in an
increase in hospital admissions, and the rate of
undiagnosed disorders at hospital or at labour admission
was not increased in the new model.
Routine iron supplementation was provided before the
trial in three study sites to most women. In the other site,
the supplementation, free of charge to women in the new
model, accorded with the well-known protective effect of
this intervention against anaemia. Perhaps more
importantly, it had long-lasting action on haemoglobin
concentrations up to the postpartum period.
We did not have any formal mechanism to assess
compliance of the standard programme with national
guidelines. The assumption was that it was carried out in
the best way services allowed, and that availability of
services was not a major limitation for a pragmatic
implementation of a western type of antenatal care. This
is an important assumption in an equivalence trial, if the
standard model was not implemented as required.
We could not pool data for the economic analysis owing
to the differences in the economic circumstances in each
country, and we investigated costs in detail in only two
countries. In both, there was a trend to lower costs with
the new model. Results of simpler studies in the other two
countries do not contradict these findings. Although 10%
is too high a level of statistical significance for clinical
decisions, two further factors are relevant. First, large
sample sizes are known to be needed to detect differences
in economic evaluations.30 Second, Briggs and Fenn31 have
argued that the appropriate level of acceptable error may
vary between therapeutic area, type of intervention, and
decision-makers’ thresholds in health economics.
There was, overall, high satisfaction among women in
both groups. However, women in the new-model clinics
showed some concern about the number of visits being
too few and the spacing between them too long. In the
opposite direction, but consistent with the components of
the new model, women were more satisfied with the
information received than those in the control clinics.
The assessment of providers’ views indicated no
opposition to the new model of care. Similar studies have
shown how they perceive the balance of advantages and
disadvantages of new schedules, both to women and to
themselves.32,33 There was an interesting mismatch
between the perception of women and providers in
relation to information given during visits on key topics.
For example, providers gave themselves higher scores
than their patients in relation to the information they
provided.
There have been four published randomised trials
comparing the effectiveness of routine antenatal-care
models with reduced number of visits including some
degree of goal-oriented activities.3 Three of these trials34–36
were of good methodological quality with small or
moderate risk of bias. As in our trial, masking of care
providers and women was not possible, and the problem
of treatment “contamination” in all trials is likely. This
effect could explain the small reduction in the number of
visits (mean difference two visits) in two trials from more
developed countries.34,36 Overall, evidence from published
trials in settings with well-established obstetrics
1562
For personal use. Only reproduce with permission from The Lancet Publishing Group.
services34,36,37 shows that small reductions in the number of
visits are compatible with good perinatal outcomes; this
conclusion is supported by the results from both the trial
from Zimbabwe35 and our trial. The subset analyses from
our trial including only the study site with baseline
antenatal care similar to that of more developed countries,
and with the largest reduction in the number of visits, is
reassuring that harm is unlikely.
The rate of pre-eclampsia or pregnancy-induced
hypertension in all previous trials was consistently lower in
the new-model group than in the standard group.
Although we did not observe such a protective effect for
pre-eclampsia, the rate of pregnancy-induced hyper-
tension was lower in the new than in the standard model.
This effect on hypertension alone may reflect a lower
diagnosis rate because of fewer visits, rather than a
preventive effect of the new model. The similar rates of
pre-eclampsia and other related outcomes in our trial
suggest no detrimental effect of these “missed” cases of
pregnancy-induced hypertension. Our finding on the lack
of effect of the intervention on the rate of low birthweight
was similar to those of previous trials, showing that
changes in antenatal-care schedule have little effect on this
outcome.
There was evidence of heterogeneity of results among
the previously published trials for preterm delivery. In the
three trials done in more developed countries, there was a
tendency towards an increased risk with the new model,
whereas the trial in a developing country35 showed a
significantly lower rate of preterm delivery in the new
model. This promising result is not supported by our trial.
Only one of the previous trials36 reported the economic
impact of the reduction of two antenatal-care visits.
Unfortunately, that evaluation did not include the cost to
women, such as travel, child care, or working time loss.
There was a small reduction in the antenatal cost to the
providers in the new model. However, because there was a
non-significant increase in the rate of admission of
neonates to intensive-care units, the saving was offset.38
Most of the women in both groups of our trial would
recommend the clinic that they attended to friends or
would use it themselves during another pregnancy. The
overall high satisfaction in both groups can be related to
the expected-response bias commonly observed when
these types of questionnaires are administered to women
in health premises.39 As in the previous trials34,36,37 more
women in the new model said that the number of visits
was too small and the spacing between them too long,
which confirms that under different cultural situations,
some women are concerned about these reduced models.
This concern could be related to women’s expectations
and their negative attitude to change,40 as well as to a
desire to be reassured by health professionals that the
pregnancy is progressing well. Women responded
positively when asked whether they would recommend the
intervention clinic to friends or would use it again, which
suggests that any dissatisfaction would not persist after
they had experienced the new model or if the new model
became a norm. A follow-up study of women enrolled in a
previous trial36 reassuringly did not show any negative
long-term effect.41 Nevertheless, if the new model is to be
adopted, women need to be well informed about its
safety.
Our trial showed that for women without previous or
current complications, a reduction in the number of visits
including goal-oriented, effective activities is not
associated with increased risk for them or their infants.
Efforts should be concentrated on implementing only
activities with scientifically demonstrated efficacy.
THE LANCET • Vol 357 • May 19, 2001

For less developed countries, the goal should be to
extend coverage to all pregnant women with the packages
shown to be effective and to avoid setting unrealistic
goals. All the activities of the basic component should
always be available to all women, as well as the required
special care for women with complications or
emergencies, if comparable results are expected.
Furthermore, the new model should be supplemented
with other activities also known to be effective that may
be relevant only to some populations (eg, malaria
programmes). For more developed countries, each
activity included in routine antenatal care should be
scrutinised or tested for evidence of its effectiveness before
being retained in the standard model. If this strategy is
systematically applied, a simpler model with a reduced
number of visits will be identified. There is ample
evidence now that the risk of harm to women and their
pregnancies is unlikely.42
Finally, the new model is in general accepted by users
and providers, does not increase cost, and in some settings
decreases cost. Although providers are unlikely to achieve
actual cost savings, resources such as staff and buildings,
and the time of women and families, will be freed for
extension of the service into more effective care provision
or other activities.
WHO Antenatal Care Trial research group
Trial Coordinating Unit—J Villar, D Khan, O Meirik, R Guidotti,
A Donner.
Data Coordinating Unit—G Piaggio, A Pinol, M Vucurevic, C Hazelden.
Steering Committee—Y Al-Mazrou, H Ba’aqeel, L Bakketeig, J M Belizan
H Berendes, G Carroli, U Farnot, A Langer, G Lindmark,
P Lumbiganon, M Mugford, V Wong.
Health Economics Group—M Mugford, G Hutton, J Fox-Rushby.
Quality of Care Group—A Langer, G Nigenda, M Romero, G Rojas,
C Kuchaisit.
Data and Safety Monitoring Committee—P Bergsjø, G Bréart, A Morabia,
H Berendes, G Piaggio, J Villar.
Country Data Coordinators—E Bergel (Argentina), L Campodónico
(Argentina), E Diaz (Cuba), M Gandeh (Saudi Arabia), Y Singuakool
(Thailand).
Field Coordinators—A del Pino (Argentina), J Vazquez (Cuba), A Helal
(Saudi Arabia), K Chaisiri (Thailand).
Participating Institutions and Staff
Centro Rosarino de Estudios Perinatales (CREP), Rosario, Argentina—
J M Belizan, G Carroli, S Difulvio, M Ferronato, S Galliano, J Chacra,
P Leto, E Hails, A Paris, H Martinez Maussión, M Baraldi, H Martinez,
A Yunes, H Rodriguez, G Renzi, G Galacho, A Amavet, N Caporaletti,
E Di Orio, G Graciano, G Englander, M Cabral, M A Perotti, J C Tresso,
E Mesa, R Martino, H Sandiano, C Maggi, O López, R Feldman,
G Guerschanik, E Delgado, G Baccifava, M Nasazzi, N Tisera,
M E Ventura, M Raffagini, M Osta, D Cremer, C Ceballos, A Goldberg,
R Quiroga, C Carobini, E Reviglio, M Quacesi, M E Conochiari,
M M Lo Valvo, D Garcia, F Baro Graf, O Cafarell, E Ludmer,
A Armando, G Diaz, C Roasio, R Cipolla, N Maino, W Barbato,
H Delprato, G Strada Saenz, J C Nardin, I Pentimalli, J Malamud,
M L Carradori, G Sinopoli, D Villeco, G U Paz, R A Gorina,
G Lombarte, J L Rivas, H Costanti, A Leroux, C Zoffora, S Carbognani,
E F Guzman, D Crosta, E Abalos, C Vigetti, G Covian, M Meneghini,
S Parfait, A Carrizo, I Alcacer, C Zimmerman, O Pace, G Gottardi,
E Pelozzi, D Moyano, A Treidel, M Alesandrelli, A Montesanto,
R Muller, G Gioia, M Villata, I Blanco, M Barreiro, S Vera, I Córdoba,
A Martinez, N Sosa, E Rolón, S Pirani, B Moreto, M Goistoma,
R Leguizamón, N Band, G Deffés, N Arine, F Burgueño, S Clemente,
A Pietrella, A Marquez, S Aguilar, S Abarno, T R Paz, R A Rainome,
G Ameriso, M Carré, M Lopez, M E Galfano, L Badin, O Sampieri,
C Llompart, S Seco.
America Arias Hospital, Havana, Cuba—U Farnot, J Vazquez, J C Vazques,
I Rivero, G Curra, J Quintero, A Abelenda, E de Armas, J L Garcia Mesa,
M De los Angeles Golpe, L Mercedes Valdés, I Cabrera, R Páez,
Z Riverón, D Mustelier, L Echemendia, M Aquino, C Hernández Bango,
N Triana, P Venereo, G Espinosa, S Niela, D Masó, O Scull, I Pla,
S Pérez Hernánez, A la Rosa, M Alvaro Diaz, R Calvo, A Martin,
J Delgado, O Santo Domingo, R Vanegas, C Katric, B Trillo,
D Casagrandi, L Mavero.
King Abdulaziz University and Ministry of Health, Jeddah, Saudi Arabia—
Y Al-Mazrou, H Ba’aqeel, M Baldo, T Khoja, T Ikram, K Abbas,
THE LANCET • Vol 357 • May 19, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Z Emam, K Marzoc, A Savier, O Salem, K Mostafa, S Baraka, F Akbar,
K ElAhmady, B Mashhour, G El-Tayeb, S El-Sodany, S Al-Sheikh,
S Baryan, A Hemaid, N Maniyar, R El-Birmawi, H Hassan, W Soliman,
F Ebada, F Sarawak, M Hafez, A Kamal, S Hatata, F Ashkan.
Faculty of Medicine, Khon Kaen University Provincial Health Office and
Health Promotion Centre, Region Kohn Kaen, Thailand—P Lumbiganon,
N Winiyakul, K Chaisiri, C Chongsomchai, P Sanchaisuriya,
P Madsthan, J Pratniwat, P Sutthiwong, U Ukhotsanakarn, P Kowit,
S Rattanaparinya; S Wattayingcharoenchai, K Phermchitrphong,
K Phatharithigul, C Woratharakul; C Nilvarangkul, T Saybuathong,
W Ussavaphark, P Siripaopradist, W Saliddeechaikool,
K Thjungsadaothong, A Puchoksiri, J Kitlerpornpairoj,
A Jesadamethakajorn, P Phupan, S Srisanarat, W Ratanachailit,
S Sriputta, U Kaseburt, L Changthom, K Mahem, S Chaipromma,
J Puwasunti, J Suvannatat, P Laoprasert, N Chuaksuchinda,
T Kammaneejan, P Kongngaen, K Chansamran, S Asadamongkol,
P Salee, R Kongsap, B Chanchumni, S Sonhai.
Administrative and Secretarial Support—J Starks, C Gray, C Peters.
Members of the Economic Evaluation research group—S Bastus,
A M Galvez, M Alvarez, G Sanabria, M Morales, J Thinkamrop,
B Thinkamrop, C Kuchaisit, C Leela Kraiwan, M Mugford,
J Fox-Rushby, G Hutton, J Borghi, N Lord.
Contributors
José Villar, Leiv Bakketeig, Per Bergsjø, Heinz Berendes, and José Belizán
were responsible for the idea and conception of the trial. José Villar and
Per Bergsjø prepared the protocol. José Villar conducted external
fundraising and supervised the trial’s overall execution. Guillermo Carroli,
Ubaldo Farnot, Hassan Ba’aqeel, Pisake Lumbiganon, José Belizán, and
Yagob Al-Mazrou collaborated in the preparation of the protocol and the
trial and implemented it in their respective countries. They actively
contributed to the overall execution of the trial. Alain Pinol and Gilda
Piaggio were responsible for data management, wrote the plan of analysis,
in collaboration with Allan Donner and José Villar, and did the statistical
analysis. José Villar and Gilda Piaggio wrote the paper with inputs from all
the investigators, especially Allan Donner and Per Bergsjø. All the
investigators read the report and made substantive suggestions on its
content. Miranda Mugford, Julia Fox-Rushby, and Guy Hutton were
responsible for designing, analysing, and interpreting the evaluation of the
cost-effectiveness alongside the trial, in collaboration with local
investigators who were responsible for the data collection, analyses, and
interpretation of unit cost. Ana Langer and Gustavo Nigenda were
responsible for the design and coordinated the implementation of the
satisfaction component of the trial in close collaboration with local
investigators. Alain Pinol and Gilda Piaggio were responsible for data
management of this component, and wrote the plan of analysis with Ana
Langer and Gustavo Nigenda, and did the analyses.
Acknowledgments
This trial was supported by the UNDP/UNFPA/WHO/World Bank
Special Programme of Research, Development and Research Training in
Human Reproduction of WHO. Additional support was provided by:
Municipal Government, City of Rosario, Argentina; Ministry of Health,
Cuba; National Institute of Public Health, Mexico; The Population
Council—Regional Office for Latin America and the Caribbean; Ministry
of Health, Saudi Arabia; Swedish Agency for Research Cooperation with
Developing Countries; Ministry of Public Health and Faculty of
Medicine, Khon Kaen University, Thailand; Department for International
Development, UK; Mother Care—John Snow Inc; National Institute for
Child Health and Human Development, National Institutes of Health,
USA; and The World Bank. For the preparatory phase: University of
Western Ontario, Department of Epidemiology and Biostatistics, Canada;
National Institute of Public Health, Norway; United Nations
Development Programme, and the University of Uppsala, Department of
Obstetrics and Gynaecology, Sweden.
We thank the women and their babies who participated in this trial; the
many doctors, nurses, and other staff of the clinics and hospitals who
made the implementation of this project possible; G Lindmark and
V Wong for their active participation as members of the Steering
Committee and their continuous support during the trial; M Koblinsky
for personal interest and support for this project; O Meirik for his
continued encouragement and support; Dr Khan for editing the trial’s
newsletter; and Carol Peters for her help in the preparation of the report.
References
1 Villar J, Bergsjø P. Scientific basis for the content of routine antenatal
care: I, philosophy, recent studies, and power to eliminate or alleviate
adverse maternal outcomes. Acta Obstet Gynaecol Scand 1997; 76:
1–14.
2 Bergsjø P, Villar J. Scientific basis for the content of routine antenatal
care: II, power to eliminate or alleviate adverse newborn outcomes;
some special conditions and examinations. Acta Obstet Gynaecol Scand
1997; 76: 15–25.
3 Villar J, Khan-Neelofur D. Patterns of routine antenatal care for low-
1563
ARTICLES
risk pregnancy. In: Cochrane Library, Issue 4. Oxford: Update
Software, 2000.
4 Cochrane AL. Effectiveness and Efficiency. Random Reflections on
Health Services. London: The Nuffield Provincial Hospitals Trust,
1972.
5 Villar J, Bakketeig L, Donner A, et al. The WHO antenatal care
randomized controlled trial: rationale and study design. Paediatr
Perinat Epidemiol 1998; 12 (supp): 27–58.
6 Ministerio de Salud y Acción Social de la Nación. Republica
Argentina. Propuesta normativa perinatal. Buenos Aires, Argentina:
1996; 1: 29–69.
7 Ministerio de Salud. Manual de diagnostico y tratamiento en
Obstetricia y Perinatologia. La Havana, Cuba, 1999.
8 Al-Mazrou Y, Al-Shehri S, Rao M. Principles and practice of primary
health care, 2nd edn. Riyadh, Saudi Arabia: Ministry of Health, 1990.
9 Department of Health, Ministry of Public Health, Thailand. Maternal
and child health record. Bangkok, Thailand, 1999.
10 Piaggio G, Ba’aqeel H, Bergsjø P, et al. The practice of antenatal care:
comparing four study sites in different parts of the world participating
in the WHO Antenatal Care Randomised Controlled Trial. Paediatr
Perinat Epidemiol 1998; 12 (suppl 2): 116–41.
11 Bakketieg LS. Methodological problems and possible endpoints in the
evaluation of antenatal care. Int J Technol Assess Health Care 1992;
1 (suppl): 33–39.
12 Mugford M, Hutton G, Fox-Rushby J. Methods for economic
evaluation alongside a multi-centre trial in developing countries: a case
study from the WHO Antenatal Care Randomized Controlled Trial.
Paediatr Perinat Epidemiol 1998; 12 (suppl): 75–97.
13 Drummond MF, O’Brien B, Stoddart GL, Torrance GW. Methods
for the economic evaluation of health care programmes. Oxford:
Oxford University Press, 1997: 278–84.
14 Langer A, Nigenda G, Romero M, Rojas G, Kuchaisit C, Al-Osimi M,
Orozco E. The evaluation of women’s and providers’ perception of
quality of antenatal care in the WHO Antenatal Care Randomized
Controlled Trial. Paediatr Perinat Epidemiol 1998; 12 (suppl): 98–112.
15 Donner A, Piaggio G, Villar J, et al. Methodological considerations in
the design of the WHO Antenatal Care Randomized Controlled Trial.
Paediatr Perinat Epidemiol 1998; 12 (suppl): 59–74.
16 Donner A. Sample size requirements for stratified cluster
randomization designs. Stat Med 1992; 1: 743–50.
17 Villar J, Dorgan J, Menendez R, Bolaños L, Pareja GP, Kestler E.
Perinatal data reliability in a large teaching obstetric unit. Br J Obstet
Gynaecol 1988; 95: 142–55.
18 Donner A, Klar N. Confidence interval construction for effect
measures arising from cluster randomization trials. J Clin Epidemiol
1993; 46: 123–31.
19 Liang KY, Zeger SL. Longitudinal date analysis using generalized
linear models. Biometrika 1986; 73: 13–22.
20 Zelen M. A new design for randomized clinical trials. N Engl J Med
1979; 300: 1242–45.
21 Broomberg J, Rees J. Delivering at the right price—the costs of
primary maternity care at the Diepkloof community health centre.
South Afr Med J 1993; 83: 272–75.
22 Luce B, Manning W, Siegel J, Lipscomb J. Estimating costs in cost-
effectiveness analysis. In: Gold M, Siegel J, Russell L, Weinstein M,
eds. Cost effectiveness in health and medicine. Oxford: Oxford
University Press, 1994: 194.
1564
For personal use. Only reproduce with permission from The Lancet Publishing Group.
23 Bullough C, Msuku R, Karonde L. Early suckling and postpartum
haemorrhage; controlled trial in deliveries by traditional birth
attendants. Lancet 1989; 2: 522–25.
24 Kendrick JS, Zahniser SC, Miller N, et al. Integrating smoking
cessation into routine public perinatal care: The Smoking Cessation in
Pregnancy Project. Am J Public Health 1995; 85: 217–22.
25 Waver M, Sewankambo N, Serwadda D, et al. Control of sexually
transmitted diseases for AIDS prevention in Uganda: a randomised
community trial. Lancet 1999; 353: 525–35.
26 Hardcastle JD, Chamberlain JO, Robinson MHE, et al. Randomised
controlled trial of faecal-occult-blood screening for colorectal cancer.
Lancet 1996; 348: 1472–77.
27 Kronberg O, Fenger C, Olsen J, et al. Randomised study of screening
for colorectal cancer with faecal-occult-blood test. Lancet 1996; 348:
1467–71.
28 Baskett TF, Sternadel J. Maternal intensive care and near-miss
mortality in obstetrics. Br J Obstet Gynaecol 1998; 105: 981–84.
29 Hofmeyr GJ, Kulier R. External cephalic version at term. In:
Cochrane Library, Issue 4. Oxford: Update Software, 2000.
30 Briggs A. Economic evaluation and clinical trials: size matters. BMJ
2000; 321: 1362–63.
31 Briggs A, Fenn P. Confidence intervals or surfaces? Uncertainty on
the cost-effectiveness plane. Health Econ 1998; 7: 723–40.
32 Sanders J, Somerset M, Jewell D. To see or not to see? Midwives’
perceptions of reduced antenatal attendances for “low-risk” women.
Midwifery 1999; 15 (4): 257–63.
33 Sikorski J, Clement S, Wilson J, et al. A survey of health professionals’
views on possible changes in the provision and organisation of
antenatal care. Midwifery 1995; 11: 61–68.
34 McDuffie RS, Beck R, Bischoff K, Cross J, Orleans M. Effect of
frequency of prenatal care visits on perinatal outcome among low-risk
women. JAMA 1996; 275: 847–51.
35 Munjanja SP, Lindmark G, Nystrom L. Randomised controlled trial
of a reduced-visits programme of antenatal care in Harare, Zimbabwe.
Lancet 1996; 348: 364–69.
36 Sikorski J, Wilson J, Clement S, Das S, Smeeton N. A randomised
controlled trial comparing two schedules of antenatal visits: the
antenatal care project. BMJ 1996; 312: 546–33.
37 Binstock MA, Wolde-Tsadik G. Alternative prenatal care: impact of
reduced visit frequency, focused visits and continuity of care. J Reprod
Med 1995; 157: 158–61.
38 Henderson J, Roberts T, Sikorski J, Wilson J, Clement S. An
economic evaluation comparing two schedules of antenatal visits.
J Health Ser Res Policy 2000; 5: 69–75.
39 Langer A, Farnot U, Garcia C, et al. The Latin American trial of
psychosocial support during pregnancy: effects on mother’s well-being
and satisfaction. Soc Sci Med 1996; 42: 1589–97.
40 Porter M, Macintyre S. What is, must be best: a research note on
conservative or deferential responses to antenatal care provision. Soc
Sci Med 1984; 19: 1197–200.
41 Clements S, Condy B, Sikorski J, Wilson J, Smeton N. Does reducing
the frequency of routine antenatal visits have long term effects?
Follow-up of participants in a randomised controlled trial. Br J Obstet
Gynaecol 1999; 106: 367–70.
42 Carroli G, Villar H, Piaggio G, et al. WHO systematic review of
randomised controlled trials of routine antenatal care. Lancet 2001;
357: 1565–70.
THE LANCET • Vol 357 • May 19, 2001