DOI: 10.1111/j.1471-0528.2007.01372.x www.blackwellpublishing.

com/bjog

General obstetrics

Randomised controlled trial of two antenatal care models in rural Zimbabwe

F Majoko,a SP Munjanja,b L Nystro m,c E Mason,d G Lindmarka
of Womens & Childrens Health, Section for International Maternal & Child Health, Uppsala University, Uppsala, Sweden of Obstetrics & Gynaecology, University of Zimbabwe Medical School, Harare, Zimbabwe c Department of Public Health & Clinical Medicine, Epidemiology & Public Health Sciences, Umea University, Umea, Sweden d World Health Organization Country Ofce, Harare, Zimbabwe Correspondence: Dr F Majoko, Department of Obstetrics and Gynaecology, Singleton Hospital, Sketty Lane, Swansea SA2 8QA, UK. Email: majokof215@doctors.org.uk
b Department a Department

Accepted 4 February 2007.

Objective To compare a ve-visit antenatal care (ANC) model

with specied goals with the standard model in a rural area in Zimbabwe.
Design Cluster randomised controlled trial with the clinic as the

randomisation unit.
Setting Primary care setting in a developing country where care

was provided by nurse-midwives.

Population Women booking for ANC in the clinics were eligible. Main outcome measures Number of antenatal visits, antepartum

and intrapartum referrals, utilization of health centre for delivery and perinatal outcomes.
Methods Twenty-three rural health centres were stratied prior to

maternal age, parity and gestational age at booking between women in the standard model and those in the new model. The median number of visits was four for both models. The proportion of women with ve or less visits was 77% in the new and 69% in the standard model (OR 1.5; 95% CI 1.082.2). The likelihood of haemoglobin testing was higher in the new model (OR 2.4; 95% CI 1.05.7) but unchanged for syphilis testing. There were fewer intrapartum transfers (5.4 versus 7.9% [OR 0.66; 95% CI 0.440.98]) in the new model but no difference in antepartum or postpartum transfers. There was no difference in rates of preterm delivery or low birthweight. The perinatal mortality was 25/1000 in standard model and 28/1000 in new model.
Conclusion In Gutu district, a focused ve-visit schedule did not

random allocation to the new (n = 11) or standard (n = 12) model of care.

Results We recruited 13 517 women (new, n = 6897 and standard,

change the number of contacts but was more effective as expressed by increased adherence to procedures and better use of institutional health care.
Keywords Antenatal care, number of visits, pregnancy outcome,

n = 6620) in the study, and 78% (10 572) of their pregnancy records were retrieved. There was no difference in median

rural Zimbabwe.

m L, Mason E, Lindmark G. Randomised controlled trial of two antenatal care models in rural Zimbabwe. Please cite this paper as: Majoko F, Munjanja S, Nystro BJOG 2007;114:802811.

Introduction
There has been little change to the schedule of antenatal visits that was recommended by the British Department of Health in a 1929 circular which set out the timing of visits as every 4 weeks from booking until 30 weeks, every 2 weeks between 30 and 36 weeks and then weekly until delivery. Increased awareness regarding the value of antenatal care (ANC) results in women initiating care early, thus an average of 14 visits if the standard model is implemented. The majority of women have an uncomplicated antenatal course and would therefore have received excessive, probably unnecessary, care. Epidemiolog-

ical studies suggest that ANC is benecial in that women with no ANC have poor pregnancy outcomes compared with women with some ANC, and those with inadequate ANC have poorer outcomes compared with women with adequate care.13 Questions about the appropriateness of the current ANC model for low-risk women have arisen in the past three decades.49 The questions have been directed at the number of visits and at whether all procedures performed at routine visits were necessary and based on evidence for effectiveness.1015 This has resulted in recommendations of modied ANC models with reduced frequency of visits and only those procedures considered effective. A large multicentre trial

802

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

Antenatal care in rural Zimbabwe

Table 1. Timing and content of antenatal visits for new model Visit number Gestation age (weeks) ,20 (ideal) 2428 3234 3638 4041 Goal Procedures*

1 2 3 4 5

Risk assessment, health education and delivery plan Exclude multiple pregnancy and check for hypertensive disorders Exclude anaemia, check fetal growth and review delivery plans Check fetal growth, exclude abnormal presentation, discuss labour Check fetal wellbeing, referral for post-term induction at 42 weeks

Haemoglobin, rapid plasma reagin, tetanus vaccination and urinalysis Urinalysis** Haemoglobin and urinalysis** Urinalysis** Urinalysis**

*Blood pressure and symphysis fundus height were measured at each visit. **If blood pressure 140/90.

coordinated by World Health Organization (WHO) in four middle-income countries conrmed the safety of a reduced visit programme.16 However, this trial was in urban or periurban centres and did not include sub-Saharan Africa where reproductive morbidity is high. A randomised controlled trial conducted in Harare, Zimbabwe, conrmed that a reduction in number of visits and change of routines were not associated with adverse maternal or perinatal outcomes.17 However, in Zimbabwe, as in most African countries, the majority of the population live in a rural setting, and there are major differences between urban and rural communities in availability, accessibility and utilisation of health facilities.18 The results from the study in an urban area in Zimbabwe therefore needed to be conrmed in a rural setting prior to implementation of the recommendations at national level. Furthermore, the urban study had looked at a low-risk population as only uncomplicated pregnancies were booked in the clinics, whereas in rural areas, all women attend the same health centres, making risk assessment and appropriate referral crucial. The primary objective of this study was to compare two ANC models in a rural population in Zimbabwe. Our null hypothesis was that in a rural unselected population, an ANC model with ve planned visits and goal-oriented routines was as effective as the standard model, where effectiveness was measured by utilisation of health facility for ANC and childbirth, referrals to the district hospital and the fetal outcomes of preterm birth and low birthweight.

Materials and methods

This was a cluster randomised controlled trial (RCT) where the health facility was the unit of randomisation. The cluster design was chosen for practical reasons, as effective individual randomisation was not possible in this setting. It would not have been feasible for nurse-midwives to give alternative models of care to randomly allocated women in individual health centres. The control arm of the study used the traditional ANC model that was the standard for the country. The

experimental arm introduced a ve-visit model with dened goals for each visit (Table 1). Gutu district was chosen as the study area because the utilisation of maternity services19 and reproductive health status of the community had been previously studied.20 The district had 25 health facilities, comprising a district hospital and 24 rural health centres (RHCs) serving a population of 195 000. The normal practice was for women to register for care at the nearest health facility. Utilisation of the RHCs for ANC was high, with 9497% of women attending at least once during the pregnancy, but the use of the health facilities for delivery was low at 7785%.1921 A survey conducted prior to the trial revealed a median number of seven antenatal visits in the district.19 Of the 25 health facilities in the district, two were excluded from the RCT because of their function as referral centres and geographical location at the commercial centre of the district (Gutu Mission Hospital and Gutu Rural Hospital). The remaining 23 health facilities were stratied by availability of radio telecommunication facilities and/or maternity waiting shelter and then allocated by simple randomisation within strata to the new (n = 11) or standard (n = 12) model (Figure 1). The health facilities were stratied according to the availability of radio telecommunication as this had an impact on ease of communicating with the district hospital in the event of an intrapartum referral. The planning of the study was performed in collaboration with the Masvingo Provincial Medical Director and discussed with the doctors working in the district hospital. Preparatory meetings were held with community and opinion leaders in the district to inform them about the study. It was not possible to offer choice and an alternative to individual women as randomisation was at health facility level. However, women in RHCs that implemented the new model were informed about the study and the schedule of visits at the booking visit. Women were also informed that additional visits would be arranged if they felt the need to attend more frequently than the suggested schedule or if pregnancy complications developed. Women were asked for verbal consent to use data from their pregnancy records. The Medical Research Council of

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

803

Majoko et al.

Figure 1. Stratication and randomisation of health facilities.

Zimbabwe and the Medical Research Ethics Committee of Uppsala University approved the study. The control arm followed the standard schedule with a visit every 4 weeks from booking until 28 weeks, every 2 weeks between 28 and 36 weeks and weekly after 36 weeks until delivery. Risk assessment was performed at the booking and subsequent visits, and referral for hospital delivery was made using a list of risk markers recommended by the Zimbabwe Ministry of Health and Child Welfare. Blood pressure, body weight and urinalysis were measured at each visit, while haemoglobin and syphilis test (rapid plasma reagin) were performed at the rst visit. The use of a rapid test meant that women who tested positive for syphilis had treatment initiated at the booking visit. Oral iron supplementation was provided to all women in both models. The experimental arm implemented a modied programme (Table 1) with a new visit schedule, revised procedures with clear goals and symphysiofundal height measurement in screening for multiple pregnancy and abnormal fetal growth. Before the trial, nurse-midwives from all RHCs participated in workshops to upgrade their knowledge and skills about the ANC model they would implement. All RHCs were supported and supervised by the same team during the trial. To ensure that differences in outcomes were not related to differences in resources, RHCs were supplied with equipment to perform all procedures as in the protocol. There were no additional personnel introduced into the RHCs for the purpose of the trial. The women and care givers were aware of their allocated care model. Primary outcomes were chosen for their ability to assess effectiveness and quality of service and included number of visits, referrals from RHC for antenatal, intrapartum or postpartum problems, place of delivery and low birthweight infant (<2500 g). The secondary outcomes were antenatal

diagnosis of hypertension and twin pregnancy, perinatal mortality, operative delivery, preterm delivery (<37 weeks) and the proportion of visits at which fundal height measurement was recorded and plotted on the antenatal record in the new model.

Data generation
All women booking for ANC in the 23 RHCs between January 1995 and October 1997 were eligible for recruitment into the trial. Womens baseline characteristics were recorded in the maternity record and in the clinic register at the booking visit. At subsequent visits, examination ndings were recorded in the record held by the mother. The maternity record was retained at the health facility after delivery. Some women who delivered at home attended the health centre for immunisation of the baby, and the maternity record was then retrieved. Women who had not reported back to the RHC 3 months after the expected date of delivery were followed up in the community through the village community workers, and the maternity record was retrieved or outcome of the pregnancy was recorded. Follow up in the community to establish pregnancy outcomes continued till December 1998. The information from the maternity record was transcribed into a data entry form by specially trained midwives and then entered into an Epi Info data le.

Sample size and statistical analysis

A survey of health facilities conducted prior to the trial provided information used for the stratication and sample size estimation.19 We estimated a mean cluster size of 500 women. With an intracluster coefcient of 0.05 and an a of 0.05, we need to recruit 6900 women in 14 clusters in each model to have a power of 80% (1-b) of detecting a 10% decrease in the proportion of women making ve or fewer visits. Statistical

804

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

Antenatal care in rural Zimbabwe

analyses were by intention to treat and accounted for the within-cluster correlation. As this was an equivalence trial, efcacy analyses were conducted comparing the two models using the standard model as a reference. Rate difference (RD) and odds ratio with 95% condence intervals were adjusted for the cluster randomisation. The analysis was performed using ACLUSTER for a stratied design, with comparison of proportions according to the method described by Donner and Klar.22

Results
There were 13 517 women recruited from January 1995 to October 1997. Pregnancy records were retrieved from 10 572 women (78%), and maternal and neonatal outcomes were known in a further 2651 women (20%), but in 2% of women, there was no information on maternal and fetal outcomes (Figure 2). There was no difference in age, parity and gestational age at booking by cluster and by model (Table 2). There was no difference by maternal risk factors (young/old age, late booking [>28 weeks], nulliparity/multiparity and previous pregnancy complications) (Table 3). The mean haemoglobin at booking was 11.7 g/dl and 12.1 g/dl in the standard and new models, respectively. A box plot of number of visits by RHC (cluster) and model is presented in Figure 3 illustrating a homogeneous pattern. In the new model, 9 of 11 and in the standard model, 8 of 12

clusters had a median of four visits, resulting in a median of four visits in both models. The likelihood of making ve or fewer visits was signicantly increased in the new model, RD 84 (95% CI 1.4169). In the standard model, only 11% (236/2111) of women booking before 20 weeks had the specified minimum of nine antenatal visits. In the new model, however, 42% (945/2236) of women booking before 20 weeks had the specified minimum of five visits. The likelihood of having their haemoglobin checked at booking was increased in the new model (RD 118; 95% CI 3.8233), but there was no effect on the rate of syphilis testing (RD 71; 95% CI 2.5 to 145) (Table 4). Almost all women (97.3%) who tested positive for syphilis received treatment at their rst visit, with the remainder treated later in the pregnancy. In both models, the blood pressure was checked in 99% of women at the booking visit, and this high level of performance was consistent at subsequent visits. Fundal height measurement was included only in the new model and was performed in 84% of the women during 77% of visits. The prevalence of fundal height measurement was around 85% in visits 13 but decreased to 76, 61 and 45% in visits 4, 5 and 6, respectively. Six percent (273/4488) of fundal height measurements at the first visit were large for dates. Fifty-seven of these women (21%) were referred as suspected multiple pregnancies, out of which 32 (56%) were confirmed. The proportion of women diagnosed with

Randomised (N = 13 517)

Standard model (n = 6620)

Pregnancy loss <20 weeks (n = 19) Outcomes not traceable (n = 118)

New model (n = 6897)

Pregnancy loss <20 weeks (n = 25) Outcomes not traceable (n = 176)

Maternal and perinatal outcome only (n = 6384) Maternal outcome only (n = 99)

Maternal and perinatal outcome only (n = 6614) Maternal outcome only (n = 82)

Full record not retrieved (n = 1279)

Full record not retrieved (n = 1372)

Full data analysed (n = 5204)

Full data analysed (n = 5324)

Figure 2. Flow chart of participants in the trial.

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

805

Majoko et al.

Table 2. Median age (years), parity, gestational age (weeks) at booking in the standard and new models by cluster Cluster number Number of women 1 2 3 4 5 6 7 8 9 10 11 12 Total 453 491 1282 238 1084 547 932 170 346 593 170 314 6620 Standard model Age 24 23 25 24 24 24 25 23 24 25 24 23 24 Parity 1 1 1 1 1 1 1 1 1 2 1 1 1 Gestational age 23 23 22 21 21 21 22 20 22 24 21 21 22 Number of women 251 413 1291 421 1074 1046 712 218 419 505 547 6897 New model Age 24 24 24 24 24 24 24 24 24 24 24 24 Parity 1 1 1 2 1 1 1 1 1 1 1 1 Gestational age 20 21 24 22 21 21 20 23 21 21 22 22

a breech presentation (54 versus 63%) or twin pregnancy (41 versus 43%) was low in both the standard and new models, respectively, and was not significantly different. There was no difference in the prevalence of antepartum and postpartum referrals (Table 4). However, the risk of intrapartum referral was signicantly reduced in the new model (RD 25; 95% CI 49 to 1.4). There were fewer women from the new model who delivered at home (RD 60; 95% CI 153 to 34) or at the district hospital

Table 3. Maternal characteristics at booking in the standard and new models for women with retrieved pregnancy records Characteristic Standard model (n 5 5223) Ratio % New model (n 5 5349) Ratio %

Age (years) 19 815/5193 15.7 8714/5311 16.9 35 742/5193 14.3 714/5311 13.4 Parity 0 1618/5222 31.0 1734/5346 32.4 6 389/5222 7.4 388/5346 7.3 Gestational age at booking (weeks) 20 2122/5028 42.2 2235/5229 42.7 29 843/5028 16.8 931/5229 17.8 Haemoglobin at booking (g/dl) 11 3267/4052 80.6 4270/4782 89.3 Previous pregnancy complications (multiparous women only) Stillbirth 123/3604 3.4 125/3613 3.5 Preterm birth 115/3604 3.2 130/3613 3.6 Neonatal death 119/3604 3.3 123/3613 3.4 Caesarean section 174/3604 4.8 153/3613 4.2 Any complication 531/3604 14.7 531/3613 14.7

(RD 62; 95% CI 154 to 30), but these differences were not statistically signicant. There was no signicant difference in rates of preterm delivery, low birthweight and perinatal death between the two models (Table 5). The mean gestational age at delivery was 38 weeks for both the models. There was no difference in mean birthweight, 3134 g (SD 478) and 3169 g (SD 506) for standard and new models, respectively. The perinatal mortality rate was similar, 25/1000 and 28/1000 for standard and new models, respectively. There was no difference in the detection of hypertensive disorders of pregnancy between the models, RD 7.9 (95% CI 3.6 to 19.4). The likelihood of operative interventions was reduced in the new model, RD 9.2 (95% CI 1.4 to 17.0), but the difference was not signicant when adjusted for cluster (Table 5). There were six maternal deaths (two in the standard model and four in the new model), thus giving a maternal mortality rate of 45/100 000 pregnancies. In two deaths (one in each model), there was insufcient information to attribute a cause, as the maternal record was not available. The second woman in the standard model died from postpartum haemorrhage after a home birth at 36 weeks. In the new model, one woman died from puerperal sepsis associated with HIV infection after a preterm delivery in a RHC. Two further women in the new model died after home births, one from postpartum haemorrhage and the other from sepsis. The maternal deaths could not be attributed to any factors related to the ANC received.

Discussion
In this cluster randomised trial, a modied antenatal programme was introduced in a rural African setting where women booked late, staff had a number of other

806

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

Antenatal care in rural Zimbabwe

12 10 8 6 4 2 0

Standard model

12 10 8 6 4 2 0

New model

Number of visits

1 2 3 4 5 6 7 8 9 10 11 12

Number of visits

9 10 11

Clinic number
Figure 3. Box plots for number of visits in each clinic in the standard and new models.

Clinic number

responsibilities, transport both to the primary clinic and to higher level of care was often difcult and resources to care for complications at the primary level were limited. The new model aimed to provide similar or more effective care than the standard model through clear goals in a limited number of planned visits. To our knowledge, this is the only reported RCT of an ANC model implemented in a rural setting from sub-Saharan Africa. Minimising loss to follow up is a major challenge to conducting a large trial in a rural African setting. The population of pregnant women was mobile, and therefore, there was a high rate of failure to retrieve the pregnancy records because of migration and home births. Women sometimes received

care in more than one geographical area during a pregnancy. Some women joined their husbands in the urban centres and gave birth in an urban health facility. We consider a complete follow up of 78% of women and information on crude outcomes in a further 20% under the circumstances in which the study was conducted as acceptable. The baseline booking data for the 22% of women whose records were not retrieved were similar to the data of those with complete information, and there is no reason to assume that their outcomes would have been signicantly different. Problems with retrieval of pregnancy records have been experienced with similar trials in Zimbabwe where only 49% of records were retrieved.23 Nurse-midwives providing ANC had other primary care

Table 4. Stratied cluster analysis of antenatal procedures and utilisation of health facilities Characteristic New model Ratio Number of visits 5 Syphilis testing Haemoglobin check Place of delivery Home or in transit Health centre District hospital Referrals Antepartum Intrapartum Postpartum 4106/5327 4697/5349 4782/5349 964/5261 2660/5261 1499/5261 1531/5349 283/5261 45/5257 Rate/1000 771 878 894 183 506 285 286 54 8.6 Standard model Ratio 3561/5182 4214/5223 4052/5223 1248/5137 1986/5137 1782/5137 1558/5223 406/5136 29/5129 Rate/1000 687 807 776 243 387 349 298 79 5.7 0.041 0.047 0.112 0.059 0.103 0.044 0.036 0.0091 0.0068 84 71 118 260 119 262 212 225 22.9 1.4 to 169 22.5 to 145 3.8 to 233 2153 to 34 230 to 268 2154 to 30 2101 to 64 249 to 21.4 24.3 to 10 1.5 1.7 2.4 0.70 1.7 0.75 0.94 0.66 1.5 1.082.2 0.973.1 1.005.7 0.401.2 0.883.0 0.491.2 0.621.4 0.440.98 0.534.3 ICC RD Adjusted 95% CI OR Adjusted 95% CI

CI, condence interval; ICC, intracluster coefcient; OR, odds ratio. CIs are adjusted for cluster randomisation.

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

807

Majoko et al.

Table 5. Maternal and neonatal outcomes in new and standard models with ratio, rate, RD and OR with 95% CI Characteristic New model Ratio Maternal complications Hypertensive disorders 492/5324 Eclampsia 3/5238 Antepartum bleeding 9/5239 Operative delivery 190/5232 Postpartum haemorrhage 34/5238 Maternal death 4/6696 Fetalneonatal Preterm delivery 599/5058 (,37 weeks) Birthweight ,2500 g 267/4280 Birthweight ,1500 g 25/4280 Fetal and neonatal mortality Stillbirth 63/5242 Stillbirth 36 weeks 36/5242 Stillbirth .36 weeks 24/5242 Early neonatal death 19/5242 Late neonatal death 49/4173 Perinatal death 185/6614 CI, condence interval; OR, odds ratio. Rate/1000 Standard model Ratio Rate/1000 RD 95% CI OR 95% CI

92.4 0.6 1.7 36.3 6.5 0.6 118.4 62.4 5.8 12.0 6.9 4.6 3.6 11.7 28.0

522/5204 12/5126 12/5126 233/5118 34/5123 2/6483 588/4930 227/3834 22/3834 69/5105 27/5105 37/5105 15/5105 35/3941 161/6384

100.3 2.3 2.3 45.5 6.6 0.3 119.3 59.2 5.7 13.5 5.3 7.2 2.9 8.9 25.2

27.9 21.7 20.6 29.2 0.1 0.3 20.9 3.2 0.1 21.5 1.6 22.6 0.7 2.8 2.8

23.6 to 19.4 20.1 to 23.4 22.6 to 1.3 21.4 to 217.0 23.2 to 3.5 212 to 6.0 213.8 to 12.0 213.9 to 7.5 23.7 to 3.4 26.0 to 3.0 21.6 to 4.8 25.8 to 0.5 23.1 to 1.7 21.8 to 7.4 22.9 to 8.5

0.91 0.24 0.73 0.79 0.98 1.94 0.99 1.06 1.02 0.89 1.30 0.63 1.23 1.33 1.11

0.801.04 0.050.93 0.291.86 0.650.97 0.591.62 0.3115.2 0.881.12 0.881.27 0.551.88 0.621.27 0.772.21 0.361.08 0.602.56 0.842.10 0.891.39

responsibilities for which there were performance targets that were directly supervised. They had to prioritise between several competing programmes and were not always inclined to adhere to a programme with a message towards changing established practices. Trial contamination is always a danger when different programmes are implemented in the same area. Nurse-midwives in the district had monthly meetings, and it is likely that aspects of the study were discussed among staff implementing the different models. Another factor was the resistance from the mothers to changes in a programme that was well established and accepted.24,25 Some women, especially those who were nulliparous or of low parity, were not supportive of changing the spacing of visits. They felt that fewer contacts with nurse-midwives could reduce chances of detecting pregnancy complications and that their needs for reassurance could not be met. Although individual randomisation would have been ideal, it was not considered feasible in this rural setting. There would have been problems associated with randomisation of women to receive different types of care from the same nurse-midwives. Cluster design has been used in similar ANC trials.5,16,17 The number of women who had equal to or less than the recommended visits in the new model was increased, which indicates that women who booked late were seen according to the protocol. However, in the standard model, the median number of visits decreased during the trial compared with

what was reported before.19 The modest decrease in the proportion of women with more than six visits in the new model conrms that even in a setting where the number of visits could already have been considered low, it is possible to make further reduction through a focused programme. In this setting with late initiation of ANC, a large reduction in number of antenatal visits was unlikely, and emphasis was on the value of a focused programme in which the number of visits was determined by the womans clinical needs. Previous unsuccessful efforts to reduce the number of antenatal visits have been reported in several trials and are ascribed to resistance to change by women and their carers.5,2530 However, the WHOcoordinated multicentre and Harare trials were successful in reducing visits among low-risk women.16,17 The increased emphasis on clear goals for each contact with the pregnant woman resulted in a better adherence to performance of well-known standard procedures such as haemoglobin measurement or syphilis testing. It has been demonstrated that poor quality in routine ANC is common in low-resource settings.18,31 The poor quality of care is partly attributed to high volume of work, limited resources, insufcient manpower and lack of resources for management of complications. Health personnel lacking understanding of the significance of the procedures, poor supervision and feedback also contribute to poor quality of care. Compliance with new components such as fundal height measurement was low, probably because of staff perception. The intended benets

808

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

Antenatal care in rural Zimbabwe

of this method can be difcult to bring across, and there are no studies that show improved perinatal outcome from its use in settings similar to that of Gutu, where follow-up investigations for conrming a diagnosis of abnormal fetal growth or for monitoring the fetal condition are not available. Previous studies in Africa have mainly discussed fundal height measurement as cheap and easy to perform but not the necessary resources for follow up of abnormal ndings, even though in some afuent settings, more elaborate technologies will be available.3235 Fundal height measurement has low sensitivity and specicity, necessitating further investigation such as ultrasound which is not available in many lowresource settings. The diagnostic ability for twin pregnancy, which this method should increase, did not change in either of these trials. This is partly because of the low referral rate of those found to have abnormal measurements. Better education of the midwifery staff is clearly important, although change may be difcult to achieve in midwives who have been practising the same way for many years. It seems that the best place to introduce new activity such as fundal height measurement is into the initial training of midwives and other staff. There is evidence that the clear recommendation to discuss and counsel women on place of delivery had an effect. Health facility delivery was increased in the new model, and there was more appropriate use of RHC for delivery. The trial used revised referral criteria for the new model, but there was no difference in antepartum referrals. The signicant reduction in intrapartum referrals in the new model was not associated with an increase in either postnatal referrals or neonatal morbidity. There was also a reduction in home births for women in the new model. This conrms that discussing about delivery several times during antenatal visits and encouraging women to make use of the health centre for delivery can successfully reduce unsupervised home deliveries. This can be expected to have a positive impact on both maternal and perinatal outcomes. A major concern with reduced-visits programmes might be failure to detect or late detection of complications such as hypertensive disorders of pregnancy because of decreased contacts. Failure to detect severe pre-eclampsia could be reected in an increased number of women experiencing complications of severe hypertension such as eclampsia. Reduced detection of hypertension has been reported in previous reduced-visits programmes,16,17,36 but other trials have found no difference in detection of complications.26,27 In this study, there was no difference seen in the rate of observed hypertensive disease. The eclampsia rate was lower in the new model, but the difference was not signicant when adjusted for cluster design. Ideally, this trial would have been powered to assess equivalence in fetal and maternal morbidities between the groups, but this was not feasible with the resources available to us. Hence, we had to rely on process indicators for our outcome

measures. The problem with this is that signicant differences may be detected in process indicators when in fact there is no difference in important outcomes.

Conclusion
This was a pragmatic trial assessing ANC in practice and the feasibility of introducing change through a modied schedule of visits and procedures. Process outcomes were improved in the new model with less emergency referrals during labour and with more women delivering at health institutions with trained staff without an increase in operative delivery. The new model appears suitable for implementation in other settings as it is associated with efcient use of a limited number of visits. It remains to be seen whether this reduced-visit schedule has any effect on maternal or fetal morbidity.

Acknowledgements
The ANC study in Gutu was funded by the Swedish International Development Cooperation Agency (Sida/SAREC) through the SidaUniversity of Zimbabwe Reproductive Health Research Programme. We thank all the women who participated in the trial for their willingness to cooperate and the health personnel in the district for their cooperation, despite the increased demands on their limited time. The support received from the Provincial Medical Director, District Medical Officer and District Nursing Officers was important for the successful conduct of the study. j

References
1 Gissler M, Hemminki E. Amount of antenatal care and infant outcome. Eur J Obstet Gynecol Reprod Biol 1994;56:914. 2 Delvaux T, Buekens P; the Study group on barriers and incentives to prenatal care in Europe. Disparity in prenatal care in Europe. Eur J Obstet Gynecol Reprod Biol 1999;83:18590. 3 Fink A, Yano EM, Goya D. Prenatal programs: what the literature reveals. Obstet Gynecol 1992;80:86772. 4 Hall MH, Chng PK, MacGillivray I. Is routine antenatal care worth while? Lancet 1980;ii:7880. 5 Hall M, MacIntyre S, Porter M. Antenatal Care Assessed. Aberdeen, UK: Aberdeen University Press, 1985. 6 Rooney C. Antenatal Care and Maternal Health: How Effective is It. A Review of the Evidence. WHO/MSM/92.4. Maternal Health and Safe Motherhood Programme, Division of Family Health. Geneva, Switzerland: World Health Organization, 1992. 7 Villar J, Garcia P, Walker G. Routine antenatal care. Curr Opin Obstet Gynecol 1993;5:68893. 8 Tucker J, Florey C du V, Howie P, McIlwaine G, Hall M. Is antenatal care apportioned according to obstetric need? The Scottish antenatal care study. J Public Health Med 1994;16:6070. 9 McDufe RS, Beck A, Bischoff K, Cross J, Orleans M. Effect of frequency of prenatal care visits on perinatal outcome among low risk women. J Am Med Assoc 1996;275:84751.

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

809

Majoko et al.

10 Lindmark G, Cnattingius S. The scientic basis of antenatal care. Report from a state of the art conference. Acta Obstet Gynecol Scand 1991;70:1059. 11 Villar J, Bergsjo P. Scientic basis for the content of routine antenatal care. I. Philosophy, recent studies and power to eliminate or alleviate adverse maternal outcomes. Acta Obstet Gynecol Scand 1997;76: 114. 12 Villar J, Carroli G, Gulmezoglu AM. The gap between evidence and practice in maternal healthcare. Int J Gynecol Obstet 2001;75:S4754 13 Carroli G, Villar J, Piaggio G, Khan-Neelofur D, Gulmezoglu M, Mugford M, et al. WHO systematic review of randomised controlled trials of routine antenatal care. Lancet 2001;357:156570. 14 Carroli G, Rooney C, Villar J. WHO programme to map the best reproductive health practices: how effective is antenatal care in preventing maternal mortality and serious morbidity? Paediatr Perinat Epidemiol 2001;15(Suppl 1):142. 15 Villar J, Carroli G, Khan-Neelofur D, Piaggio G, Gulmezoglu M. Patterns of routine antenatal care for low-risk pregnancy. Cochrane Database Syst Rev 2001;CD000934. 16 Villar J, Baaqeel H, Piaggio G, Lumbiganon P, Belizan JM, Farnot U, et al. WHO antenatal care randomised trial for the evaluation of a new model of routine antenatal care. Lancet 2001;357:155164. 17 Munjanja SP, Lindmark G, Nystrom L. Randomised controlled trial of a reduced-visits programme of antenatal care in Harare, Zimbabwe. Lancet 1996;348:3649. 18 Myer L, Harrison A. Why do women seek antenatal care late? Perspectives from rural South Africa. J Midwifery Womens Health 2003; 48:26872. 19 Nhindiri P, Munjanja S, Zhanda I, Lindmark G, Nystrom L. A communitybased study on utilisation of maternity services in rural Zimbabwe. Afr J Health Sci 1996;3:1205. 20 Nilses C, Nystrom L, Munjanja SP, Lindmark G. Self reported reproductive outcome and implications in relation to use of care in women in rural Zimbabwe. Acta Obstet Gynecol Scand 2002;81:50815. 21 Van den Heuvel OA, De Mey WG, Buddingh H, Bots ML. Use of maternal care in a rural area of Zimbabwe: a population-based study. Acta Obstet Gynecol Scand 1999;78:83846. 22 Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. London: Arnold, 2000.

23 Mahomed K, Mason E, Warndorf T. Home-based mothers record: operational feasibility, understanding and usage in a rural community in Zimbabwe. Trop Doc 2000;30:1559. 24 Mathole T, Lindmark G, Ahlberg BM. Dilemmas and paradoxes in providing and changing antenatal care: a qualitative study of nurses and midwives in rural Zimbabwe. Health Policy Plan 2006;21:38593. 25 Mathole T, Lindmark G, Majoko F, Ahlberg BM. A qualitative study of womens perspectives of antenatal care in a rural area of Zimbabwe. Midwifery 2004;20:12232. 26 Sikorski J, Wilson J, Clement S, Das S, Smeeton N. A randomised controlled trial comparing two schedules of antenatal care visit: the antenatal care project. BMJ 1996;312:54653. 27 Jewell D, Sharp D, Sanders J, Peters TJ. A randomised controlled trial of exibility in routine antenatal care. BJOG 2000;107:12417. 28 Hemminki E, Gissler M. Quantity and targeting of antenatal care in Finland. Acta Obstet Gynecol Scand 1993;72:2430. 29 Backe B. Overutilization of antenatal care in Norway. Scand J Public Health 2001;29:12932. 30 Walker DS, Day S, Diroff C, Lirette H, McCully L, Mooney-Hescott C, et al. Reduced frequency prenatal visits in midwifery practice: attitudes and use. J Midwifery Womens Health 2002;47:26977. 31 Urassa DP, Carlstedt A, Nystrom L, Massawe SN, Lindmark G. Quality assessment of the antenatal program for anaemia in rural Tanzania. Int J Qual Health Care 2002;14:4418. 32 Munjanja SP, Masona D, Maxwell M, Mahomed K. A symphysial-fundal height nomogram for central Africa. Cent Afr J Med 1987;33:2932. 33 Neilson JP, Verkuyl DA, Bannerman C. Tape measurement of symphysis-fundal height in twin pregnancies. Br J Obstet Gynaecol 1988; 95:10549. 34 Challis K, Osman NB, Nystrom L, Nordahl G, Bergstrom S. Symphysisfundal height growth chart of an obstetric cohort of 817 Mozambican women with ultrasound-dated singleton pregnancies. Trop Med Int Health 2002;7:67884. 35 Theron GB, Theron AM, Odendaal HJ. Symphysis-fundus growth measurement followed by umbilical artery Doppler velocimetry to screen for placental insufciency. Int J Gynecol Obstet 2002;79:2634. 36 Berglund AC, Lindmark G. Health services effects of a reduced routine programme of antenatal care. An area based study. Eur J Obstet Gynecol Reprod Biol 1998;77:1939.

Commentary on Antenatal care programmes in low resource settings

Dr Majoko et al. report a cluster randomised controlled trial of a goal-oriented antenatal care model with the aim of better targeted but fewer visits in a rural setting. In Zimbabwe, as in many other parts of Africa, late booking and missed antenatal visits are common, and for this reason, it is not surprising that there were similarly small numbers of visits in the standard and the new care models. Having complete data on 78% of women is reasonable considering the setting with mobility of the population. Having crude follow-up data on 98% of women is a remarkable achievement. The new model aimed to deliver specic interventions and procedures at each visit. Haemoglobin and syphilis testing coverage were improved in the new model, although the change in the latter was not statistically signicant. The authors report that women were anxious about the reduction in number of visits, although data on the womens views were not systematically reported. Clinicians working in settings where hypertensive disorders of pregnancy are a common cause of maternal mortality and morbidity have expressed concern that 4-week spacing of visits after 32 weeks may fail to detect the onset of pre-eclampsia in time to prevent serious complications. A possible alternative, which has not been tested to our knowledge, is to teach pregnant women to do weekly urine testing at home. It might also be argued that clinic testing for proteinuria only when

810

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

Antenatal care in rural Zimbabwe

hypertension is detected, as in the current trial protocol, will fail to identify the less common but serious occurrence of gestational proteinuria preceding hypertension. However, the data on the occurrence of eclampsia in the current trial are reassuring. The Cochrane review (Villar et al. Patterns of routine antenatal care for low-risk pregnancy. Cochrane Database Syst Rev 2001) on antenatal care models includes two trials conducted in developing country settings. The rst trial was conducted in urban Zimbabwe by the same group of researchers who conducted the rural Zimbabwe trial reported here (Munjanja et al., Lancet 1996;348:3649). The second trial is the World Health Organization Antenatal Care trial conducted in Argentina, Cuba, Thailand and Saudi Arabia (Villar et al., Lancet 2001;357:155164). The main conclusion of the Cochrane review is that fewer but goal-oriented visits have similar clinical outcomes to traditional frequencies and cost less, although they may result in some women feeling that they do not receive adequate care. Antenatal care programmes should have specific goals, and the activities needed to reach those goals should be communicated clearly to caregivers. The minimum antenatal care package content is relatively clear. In different settings, there may be a need to include other activities based on epidemiological factors and local priorities. A particular problem in developing country settings, highlighted again in this report, is the lack of early antenatal booking to establish gestational age, treat syphilis, institute dietary supplementation and identify women requiring antiretroviral treatment. Research in South Africa has shown that many women access health services early in pregnancy for pregnancy confirmation, but these opportunities to institute formal antenatal care are not used, and when the women book much later in pregnancy, early pregnancy information is not available (Jeffery et al., S Afr Med J 2000;90:1536). A possible solution is to promote a policy whereby every pregnant woman attending a private or public health facility for the first time, even if not at a formal antenatal clinic, is offered the first visit package described in this trial and issued with a patient-held pregnancy record on which the early visit information is recorded. As with any practice, the implementation of a goal-oriented antenatal care programme should be audited. Such audit may be more important for a complex intervention such as an antenatal programme where several individual activities need to be tracked and to avoid ending up with a programme that contains the same old rituals in fewer visits. j lmezoglu* and G Justus Hofmeyr A Metin Gu

*The author is a staff member of the World Health Organization. The author alone is responsible for the views expressed in this publication and they do not necessarily represent the decisions or the stated policy of the World Health Organization.

2007 The Authors Journal compilation RCOG 2007 BJOG An International Journal of Obstetrics and Gynaecology

811