Corresponding Author Moke Magoma, Department of Obstetrics & Gynaecology, Bugando Medical Centre and Teaching Hospital, P.O. Box 1370, Mwanza, Tanzania. Tel.: +255 754 284 691; E-mail: firstname.lastname@example.org
To determine the effectiveness of birth plans in increasing use of skilled care at delivery and in the postnatal period among antenatal care (ANC) attendees in a rural district with low occupancy of health units for delivery but high antenatal care uptake in northern Tanzania.
Cluster randomised trial in Ngorongoro district, Arusha region, involving 16 health units (8 per arm). Nine hundred and five pregnant women at 24 weeks of gestation and above (404 in the intervention arm) were recruited and followed up to at least 1 month postpartum.
Skilled delivery care uptake was 16.8% higher in the intervention units than in the control [95% CI 2.6–31.0; P = 0.02]. Postnatal care utilisation in the first month of delivery was higher (difference in proportions: 30.0% [95% CI 1.3–47.7; P < 0.01]) and also initiated earlier (mean duration 6.6 ± 1.7 days vs. 20.9 ± 4.4 days, P < 0.01) in the intervention than in the control arm. Women's and providers' reports of care satisfaction (received or provided) did not differ greatly between the two arms of the study (difference in proportion: 12.1% [95% CI –6.3–30.5] P = 0.17 and 6.9% [95% CI –3.2–17.1] P = 0.15, respectively).
Implementation of birth plans during ANC can increase the uptake of skilled delivery and post delivery care in the study district without negatively affecting women's and providers' satisfaction with available ANC services. Birth plans should be considered along with the range of other recommended interventions as a strategy to improve the uptake of maternal health services.
Déterminer l'efficacité du planning familial à augmenter l'utilisation de soins qualifiés à l'accouchement et dans la période postnatale parmi les visiteuses des services de soins prénataux dans un district rural à faible taux d'occupation des unités de santé pour l'accouchement, mais à adoption élevée des soins prénataux dans le nord de la Tanzanie.
Essai randomisé par grappes dans le district de Ngorongoro, dans la région d'Arusha, incluant 16 unités de santé (eight par bras). Of 905 femmes enceintes fréquentant les consultations prénatales (CPN) à 24 semaines de gestation et au-delà (404 dans le groupe d'intervention) ont été recrutées et suivies jusqu’à un mois au moins après l'accouchement.
L'adoption de soins qualifiés à l'accouchement était de 16.8% plus élevée dans les unités d'intervention que dans les unités contrôles [IC 95%: 2.6–31.0; P = 0.02]. L'utilisation des soins postnataux durant le premier mois de l'accouchement a été plus élevée (différence entre les proportions: 30.0% [IC 95%: 1.3–47.7; P < 0.01]) et a également été initiée plus tôt (durée moyenne 6.6 ± 1.7 jours contre 20.9 ± 4.4 jours, P < 0.01) dans le bras d'intervention que dans le bras témoin. Le report de satisfaction pour les soins (reçus ou fournis) des femmes et des prestataires ne différait pas beaucoup entre les deux bras de l’étude (différence de proportion: 12.1% [IC 95%: −6.3 to 30.5] P = 0.17 et 6.9% [IC 95% −3.2 to 17.1] P = 0.15, respectivement).
La mise en œuvre de planning familial dans la clinique prénatale peut augmenter l'adoption de soins qualifiés à l'accouchement et après l'accouchement dans le district étudié sans affecter négativement la satisfaction des femmes et des prestataires, avec les services de soins prénataux disponibles. Le planning familial devrait être pris en compte avec la gamme des autres interventions recommandées comme une stratégie visant à améliorer l'utilisation des services de santé maternelle.
Determinar la efectividad de los planes de parto para aumentar el uso de cuidados especializados en el momento del parto y durante el periodo postnatal, entre pacientes recibiendo cuidados prenatales en un distrito rural del norte de Tanzania, con un bajo nivel de ocupación de las unidades sanitarias para partos, pero una alta aceptación de los cuidados prenatales.
Ensayo aleatorizado, en conglomerados, en el distrito de Ngorongoro - región de Arusha - que incluyó 16 unidades sanitarias (eight por brazo). Se reclutaron 905 mujeres embarazadas que recibían cuidados prenatales y que estaban de al menos 24 semanas de gestación (404 en el brazo de la intervención), y se les realizó un seguimiento hasta al menos un mes después del parto.
La aceptación de unos cuidados de parto especializados era un 16.8% mayor en las unidades de intervención que en las de control [IC 95% 2.6–31.0; P = 0.02]. El uso de los cuidados de postparto durante el mes siguiente al nacimiento era mayor (diferencia en las proporciones: 30.0% [IC 95% 1.3–47.7; P < 0.01]) y comenzaron antes (duración media 6.6 ± 1.7 días versus 20.9 ± 4.4 días, P < 0.01) en el brazo de la intervención que en el brazo control. Los informes sobre la satisfacción por el servicio, tanto de las mujeres como de los proveedores (recibidos y ofrecidos), no diferían mucho entre los dos brazos del estudio (diferencia en proporción: 12.1% [IC 95% −6.3 to 30.5] P = 0.17 y 6.9% [IC 95% −3.2 to 17.1] P = 0.15 respectivamente).
Implementar el plan de parto durante los cuidados prenatales puede aumentar la aceptación de un parto especializado y de cuidados de postparto en el distrito del estudio, sin afectar de forma negativa la satisfacción de las mujeres y de los proveedores con los servicios prenatales disponibles. Los planes de parto deberían tenerse en cuenta junto con todas las demás intervenciones recomendadas como una estrategia para mejorar la aceptación y uso de los servicios de salud materna.
Skilled care during and immediately after delivery and emergency obstetric care (EmOC) are key strategies for reducing maternal and neonatal mortality (Starrs 1997; WHO et al. 2009). Yet in many countries in sub-Saharan Africa, substantially more women attend antenatal care (ANC) than deliver with skilled providers in health facilities or receive postnatal care (WHO/UNICEF 2003; Stanton et al. 2007). In Tanzania, for example, more than 90% of women attend ANC at least once, but only 50% of women deliver under the care of skilled providers; and just 31% of delivered women seek postnatal care (PNC) within 48 h of delivering (National Bureau of Statistics & ICF Macro 2010). Measures to address this discrepant pattern have not yielded results fast enough to meet national and global efforts to achieve Millennium Development Goal 5 (MDG 5) to improve maternal health (Tanzania National Strategy for Growth & Reduction of Poverty 2005; United Nations 2004), and many complex factors still prevent women from seeking care (Jahn et al. 1998; Kowalewski et al. 2000; Allen 2002; Mpembeni et al. 2007; Mrisho et al. 2007; Gabrysch & Campbell 2009; Kruk et al. 2009; Magoma et al. 2010).
One suggested approach for increasing coverage of skilled delivery care is the inclusion of birth plans (often referred to as birth preparedness and complication readiness plans) in routine ANC (WHO 2006). A number of countries in sub-Saharan Africa have adopted WHO's focused antenatal care model (WHO 2008) which promotes birth plans as a strategy for improving women's health seeking behaviours for timely and appropriate care during pregnancy, labour, delivery and the postnatal period (WHO 2006).
We searched for published articles in MEDLINE, Cochrane, EMBASE and POPLINE databases to identify quasi experimental, randomised controlled trials, systematic reviews and meta-analyses published in English in which birth plans had been implemented to increase skilled care at and after delivery in developing countries. Search terms were as follows: randomised controlled trials; intervention; birth preparedness; birth plans, complication readiness; skilled delivery care; skilled delivery attendance; health facility utilisation for delivery; postpartum; postnatal and after delivery care. Only articles published from January 2000 to September 2011 were included. Cross-sectional studies and reports on the same subject as well as those whose methods were not well explained were excluded.
Only five intervention studies and one systematic review were identified: Shefner-Rogers and Sood (2004), Hossain and Ross (2006), McPherson et al.(2006), Moran et al. (2006), Brazier et al. (2009). Of the intervention studies, all were non randomised field interventions. Two were conducted in West Africa Moran et al. (2006), Brazier et al. (2009) and the remaining three in Asia Shefner-Rogers and Sood (2004), Hossain and Ross (2006), McPherson et al. (2006). In summary, four studies had institutional or skilled delivery care as their outcome (two in Asia and two West Africa). The increase in institutional or skilled delivery care ranged from 0% to 28% with studies in West Africa reporting the highest improvement (19.0% and 28.0% respectively). Two studies, both conducted in Asia, reported improvements in EmOC access as an outcome (2.0% and 23.9%). Only one study reported postpartum care uptake as an outcome, with reported increased utilisation of care of 17%.
The reviewed studies share some common characteristics: most employed multiple interventions which were often not similar and differed in the reported outcomes and measures of intervention effect. The interventions were at different levels: multiple interventions at the community level (Hossain & Ross 2006; Moran et al. 2006; Brazier et al. 2009) or birth plans at the community level with pre- and/or post intervention interview for impact assessment (Shefner-Rogers & Sood 2004; McPherson et al. 2006), thereby precluding the attribution of any effectiveness to birth plans alone. In a review of methodological issues in the measurement of birth preparedness in support of safe motherhood, Stanton (2004) found that few studies had examined the effectiveness of birth preparedness interventions, and the studies which existed were flawed in their study designs and sample size calculations Stanton (2004).
We thus concluded that robust empirical evidence lacks for the effectiveness of birth plans in increasing skilled care at delivery and after delivery as well as satisfaction with care among both care providers and women particularly in countries characterised by divergent patterns of coverage of ANC and SBA. We undertook a cluster randomised trial in a rural district in northern Tanzania with the objective of determining the effectiveness of birth plans in increasing skilled care during and after delivery. Our study sought to provide such evidence in a typical situation of high use of antenatal care and low utilisation of both skilled delivery and after delivery care, and where birth plans were rarely implemented despite being part of the routine ANC package.
Study sites and participants
The study was conducted in Ngorongoro district, Arusha region. The population consists of Maasai (80%), who are semi-nomadic pastoralists and Watemi (18%), who are mostly peasant farmers. On the basis of 2002 district census(National Bureau of Statistics 2002) and the crude birth rate of 39.0 per 1000 for rural Tanzania (National Bureau of Statistics & ICF Macro 2010), about 5000 deliveries are estimated to occur in the district each year.
Ngorongoro has a total of 20 health units. Eighteen of these, two hospitals and 16 dispensaries, offer MCH services. We focused our study on the 16 dispensaries and excluded the two hospitals. The 16 dispensaries included in the study had a total of 23 workers of various cadres involved in the delivery of MCH services: nursing officers (2), midwives (5), MCH nurses (8), nurse auxiliaries (4) and clinical officers (4). The distribution of the cadres in the two arms of the study was the same with the exception of midwives. There were two midwives in the control arm of the study and three in the intervention arm.
All women at 24 weeks gestation or later were recruited over a period of 4 months, and followed up to at least 1 month after delivery. Women were excluded if they planned to deliver outside the district or move out of the district after delivery and before the follow-up interview.
The intervention involved the introduction and promotion of birth plans by care providers during ANC to prepare women and their families for birth and complication readiness. This included discussions on planned place of delivery, the importance of skilled delivery care for all women, transport arrangements to the delivery site or during an emergency, funding arrangements for delivery or emergency care services if needed, identification of possible blood donors, identification of a birth companion if desired and appropriate, and support in looking after the household while the woman was at the health facility. Strategies for overcoming barriers to accessing skilled delivery care and recognising danger signs during pregnancy, labour and the postpartum period were also discussed. Providers in the intervention units were given a birth plan implementation guide with instructions on how to assist women to formulate and achieve their birth plans.
Providers in the intervention health units received didactic and practical training over the course of two half days on the birth plan intervention. Providers in the control health units were trained for half a day on how to collect the required information in their arm. Providers in both the intervention and control health units participated in a separate half day review of the elements of the focused ANC model. All providers were followed up in their respective health units fortnightly, and any providers who experienced difficulties with implementing the intervention were provided with support as needed.
When developing the birth plans, providers asked each woman about her choice of place of delivery and explored the reason(s) for the choice. With the woman's consent, her male partner or any other person she identified as involved in her care was invited to attend subsequent discussions on birth plans to enable the woman to deliver at the health unit (dispensary or hospital) of her choice. The plans were completed in writing; one copy was kept by the woman and a second was retained at the dispensary where she received ANC.
Women were interviewed twice: on the day of recruitment and around 1 month after delivery upon exit from postnatal clinics or at home, depending on the woman's choice. The interviews were conducted by trained data collectors fluent in the two local languages (Ma or Kitemi) using pretested questionnaires in Kiswahili. Data were collected on socio-demographic characteristics, past and present obstetrical history including antenatal, delivery and postnatal care and on the women's birth plans. Information was also collected on the components of ANC defined in the Tanzanian national ANC guidelines and those that were assessed in the 2004–2005 Demographic and Health Survey (DHS) (National Bureau of Statistics & Macro International 2006).
The study was implemented in three phases. The initial phase involved qualitative data collection to gain an understanding of the factors facilitating or detracting from women's ability to receive skilled delivery and postnatal care. Results from this phase have been reported elsewhere (Magoma et al. 2010). The results of the formative research phase were used to inform the development of the data collection tools, birth plan instrument and training of care providers and data collectors. The third phase involved trial implementation. A process evaluation of the implementation of birth plans and routine ANC was conducted for quality control purposes. Results from the process evaluation have also been reported Magoma et al. (2011).
The primary outcome measure was delivery by the individual woman in a health unit and was ascertained through women's reports, and cross-checked against health unit delivery records. Secondary outcomes included PNC attendance and satisfaction of women and providers with care received and provided, respectively. PNC attendance within 1 month of delivery was determined from women's reports and cross-checked against women's antenatal care cards and health unit records.
Sample size estimation
We calculated the sample size using Hayes and Moulton's (2009) method, which takes into account the coefficient of variation (k), the number of events, the expected effect and the power of the study. We assumed a coefficient of variation of 0.36, calculated by dividing the standard deviation of utilisation of health units for delivery (estimated at 2.5% from the range of delivery rates in the available health units in the district) by the average delivery care utilisation at health facilities of 7% in the district. Both figures were based on the 2006 district annual report Ngorongoro District Council (2006). The sample size formula indicated that given 16 clusters, an average sample size of 45 women per cluster would be required to detect an absolute increase in skilled delivery care of 10% (from 7% to 17%, significance level 5%, two sided) with a power of 80%. To account for a potential loss to follow-up of 20%, the sample size required was set to 900 women.
Sequence generation and allocation concealment
Health units were stratified into two groups based on the level of skilled delivery utilisation of ANC attendees in the year preceding the study (two facilities with at least 3% utilisation; and 14 facilities with <3%). The facilities were randomly assigned to the intervention or control arms in a 1:1 ratio using computer-generated random numbers by a statistician in London who had never visited the study district or study health units. Health units consented to take part prior to randomization. All eligible patients were invited to participate in the study, and all but two consented. The study did not allow blinding of either providers or women who participated in the study to the treatment allocation.
Data were entered using the SPSS statistical package (version 16) and later transferred into STATA (version 9) for cleaning and analysis. Analysis was by intention to treat. We assessed intervention effectiveness by calculating the absolute difference in the proportion of care utilisation and satisfaction between the intervention and control arms of the study using unpaired t-test statistics taking into account the clustering effect. Information on covariates likely to affect utilisation of health units for delivery care was collected, and adjustments were made in cases of imbalances in the two study arms. Wealth quartile and ethnicity were not balanced and were adjusted for in calculating the effectiveness of the intervention on the primary outcome. These potential confounding variables were entered into a logistic regression model to examine their association with the primary outcome measure (skilled delivery). The difference between the predicted and observed in each cluster (the residual) was then compared using a unpaired t-test to provide the intervention effect adjusted for these covariates as described by Hayes and Moulton (2009).
Ethical review and approval were sought from WHO, the London School of Hygiene and Tropical Medicine, and the National Institute for Medical Research in Tanzania. Ngorongoro district health, administrative and traditional authorities were informed and their approval for the study sought. Written informed consent was requested and obtained from all participating women. The trial was registered with the Australian New Zealand Clinical Trial Registry (number ACTRN 12609000268246).
Recruitment of women into the trial (Figure 1) commenced on 1st December 2008 and ended on 31st March 2009, with follow-up continuing up to August 28th 2009. All recruited women were successfully followed up. Approximately, 60% of women in both the intervention and control arms lacked formal education (Table 1). Utilisation of health units for delivery in the antecedent pregnancy was also low (12.8% in the intervention vs. 14.3% in the control). Most women initiated ANC late in the second trimester.
Table 1. Baseline characteristics of women in the intervention and control arms of the trial on the effectiveness of antenatal birth plans in Ngorongoro, rural Tanzania
Intervention N = 404
Control N = 501
Unless otherwise stated, number indicates number of women' in the specific study arm and those in brackets are the respective percentages.
Place of initial interview
Place of follow-up interview
Respondents' mean age in years ±SD
25.3 ± 6.0
24.8 ± 6.5
Woman' ethnic group
Women's level of education
At most primary completed
Households' wealth quartile
Antecedent delivery in a health unit (%)
39 [12.8] (n = 304)
53 [14.3] (n = 372)
Gestation age in months at recruitment (mean ± SD)
6.9 ± 1.0 (n = 404)
6.8 ± 0.8 (n = 501)
Gestation age in months at initial antenatal care attendance (mean ± SD)
5.4 ± 1.2 (n = 404)
5.8 ± 1.2 (n = 501)
Distance in km from residence to nearest health unit offering delivery services (mean ± SD)
5.5 ± 7.8 (n = 404)
5.6 ± 7.5 (n = 499)
In general, the intervention was implemented as per protocol. Changing of clinics was infrequent. Only two women in the intervention arm migrated and sought care from other units in the same study arm. The number of women attending ANC on any given day was low (three in the intervention vs. two in the control), allowing individualised birth plan discussions. Most birth preparedness topics did not feature prominently in the provider-patient dialogue in the control health units Magoma et al. (2011).
Women in the intervention arm of the study reported that they discussed most elements of the birth plan with their providers (Table 2). Of note, more women in the intervention arm than in the control arm discussed identifying their preferred delivery site and making transport arrangements. The length of the ANC consultation in the intervention arms was more than double that in the control arm (average time ±SD 40.1 ± 5.0 min vs. 19.9 ± 6.5 min P < 0.0001; information not shown in tables).
Table 2. Birth plans components provided to women in the intervention and control arms reported at initial interview
Care component reported
Intervention (n = 8)
Control (n = 8)
Difference in proportion
Number in each row denotes % of women in each arm that reported to have had the discussion with the provider on a particular care component and the respective 95% confidence interval.
Information on danger signs in pregnancy
17.2 [2.7 to 31.8]
Information on the danger signs in labour, delivery and after
40.8 [24.3 to 57.2]
Discussion on identifying a delivery place
52.1 [29.4 to 74.8]
Discussion on transport arrangements
59.9 [35.1 to 84.6]
Discussion on financial arrangement for transport and services
66.4 [48.7 to 84.0]
Discussion on company to the delivery site
70.9 [54.0 to 87.8]
Discussion on blood donor
75.2 [64.5 to 86.0]
Discussion on arrangement for someone to help with household chores
66.8 [47.8 to 85.8]
Discussion on birth preferences
67.0 [47.7 to 86.3]
Dialogue on emergency care seeking
22.6 [0.6 to 44.5]
Discussion that all pregnancies carry risk
34.2 [5.2 to 63.3]
Discussion on identifying a place for postpartum care
19.8 [−6.3 to 45.8]
35% of women in the intervention arm and 20% in the control arm delivered in a health unit. Prior to adjustment for baseline imbalance between the two arms, there was no strong evidence of a difference (P = 0.21). However, after adjustment for wealth quartile and ethnicity, there was some evidence that women in the intervention arm were more likely to deliver in a health unit than women in the control arm (adjusted difference in proportions 16.8%, 95% CI 2.6–31.0, P = 0.02) (Table 3).
Table 3. Assessment of the effectiveness of an antenatal birth plans intervention in Ngorongoro, rural Tanzania
Results from both the unadjusted and adjusted analysis indicate that women in the intervention arm were more likely to seek postnatal care within 1 month of delivery than women in the control arm (unadjusted absolute difference 30%, 95% CI 11.3–46.7, P = 0.004 vs. 31.3%, 95% CI 15.4–47.2 in the adjusted analysis). Women in the intervention arm of the study who sought postnatal care tended to do so within 1 week, whereas in the control arm, the delay was approximately 3 weeks (Table 3). More women in the intervention group attended postnatal care within the first 48 h than their counterparts in the control (22.7% vs. 2.2%, difference in proportions 20.5% 95% CI 14.3–26.8, P < 0.0001; results not shown in tables). Providers' and women participants' reports on satisfaction with care provided or received did not differ significantly between the two study arms, and satisfaction with care received was high in both arms (Table 3).
The birth plan intervention in this study was implemented at health units by ANC providers with no additional inputs in health infrastructure or personnel. Although the average time for initial and subsequent ANC consultations was relatively higher in the intervention clinics, it fell within the recommended WHO time, indicating that birth plans are feasible in this setting. The intervention appears to have improved both the utilisation of health units for delivery and postnatal care within 1 month of delivery without any significant negative effect on providers' and women's satisfaction with the ANC they provided or received. Women's similar levels of satisfaction with the ANC provided in both study arms would need further exploration in future studies, in particular using qualitative data. High regard for the ANC services available in the study setting, low expectations of care, problems with the instrument/tool used or chance are possible explanations of this finding.
We believe our study is unique in its method – the use of a cluster randomised trial and the introduction of the single birth plan intervention. Previous studies have been inconclusive on the effectiveness of birth plans because they were introduced as part of a package of interventions (Hossain & Ross 2006; Moran et al. 2006; Brazier et al. 2009) or were evaluated using a pre/post interview approach with no control arm (Shefner-Rogers & Sood 2004; McPherson et al. 2006). Unlike most previous studies on the topic, our study also assessed the effectiveness of the intervention on uptake of postnatal care.
Initiation of ANC early before 14th week gestation is emphasised in the focused ANC guidelines in Tanzania, but women in the study initiated care when they were around 5 months pregnant. Although similar to the country average, the timing may not allow implementation of all components of care, including birth plans. Furthermore, results from this study suggest that the intervention may increase uptake of initial PNC within 1 week of delivering. The current Tanzanian postnatal guideline advise women to attend four visits with an emphasis on the first visit within 48 h of birth and subsequent visits on the 7th day and 4th and 6th weeks Tanzania Ministry of Health (2002) but just 31% of women receive the initial PNC within 48 h National Bureau of Statistics, & ICF Macro (2010). Fewer women involved in this study sought initial PNC within this period.
Results from this study indicate that introducing birth plans during ANC improved women's utilisation of health units for PNC within 1 month of delivery. Several factors may explain this result. Previous findings in the study population suggest that women prefer to seek skilled delivery care at hospitals rather than at lower-level health units Magoma et al. (2010) even though hospitals are less accessible and associated with higher opportunity costs (e.g. transportation costs, and the costs of leaving the household for the time required to reach and return from the hospital). In contrast, women tend to seek PNC at local dispensaries which are usually near their households. PNC also offers other incentives: Women can register their babies and pick up the card required for receiving free child health services during consultations (Magoma et al. 2010). Women's widely held trust in the efficacy of childhood immunizations may have been another motivation to seek PNC because child health clinics and PNC consultations are jointly organised. Attending antenatal and postnatal clinics are viewed by women in the study setting as empowering because these are rare opportunities to leave the household on their own accord and take control of their own health Magoma et al. (2010).
Strengths of this study are the 100% follow-up rate, implementation with no additional inputs in health care infrastructure and personnel, and triangulation of the sources of data to assess the primary and secondary outcomes. Although still lower than the national average of 50% (National Bureau of Statistics, & ICF Macro 2010), the improvement in utilisation of the available health units for delivery to approximately 35% and the 30% improvement in the utilisation of postnatal care are commendable in view of the study population and setting. Ngorongoro is a protected wildlife zone, with both game parks and wildlife areas. The associated laws and its remoteness limit the mobility of the district's residents and their access to health care services. Effective implementation of the intervention and follow-up of all participants were related to several factors. Developing the birth plans during routine ANC consultations allowed providers to address the individual needs of their patients. The data collectors were people from the same villages as the study participants, which facilitated the identification of women's location for the follow-up interview. TBAs and a strong network of village health workers in the district-assisted data collectors in locating women's homes for the postnatal interview.
The study had a number of limitations. There were few clusters per study arm. Larger studies in similar or different settings are required to confirm our findings. Additionally, the study population is unique to the Ngorongoro district, and the two ethnic groups have distinct practices regarding care seeking for delivery and postnatal care. Results from this study may therefore not be applicable to other settings or population groups in Tanzania or in other countries in sub-Saharan Africa. The choice of using 1 month duration for the secondary outcome of postnatal care utilisation is based on the prevailing practice in the district at the time of the intervention implementation (Magoma et al. 2010). Although guidelines for postnatal care were revised in 2004 to emphasise the initial visit in the first 48 h after delivery, they had not been implemented in the study district. Nevertheless, the finding that more women in the intervention units sought initial postnatal care earlier than those in the control health units is encouraging; and future studies should specifically examine the impact of birth plans on increasing utilisation of postnatal care within 48 h postpartum as an outcome. Clinics in other locations may experience a higher volume of patients, limiting provider ability to offer individualised counselling on birth plans. Lastly, contamination of the control arm cannot be ruled out as providers in the intervention units might have shared information about the intervention with their counterparts in the control health units.
We conclude that birth plans may be an effective strategy for improving women's utilisation of health units for delivery and postnatal care without negatively affecting care satisfaction of women or providers. Follow-up studies involving larger samples are needed to further develop the evidence base on the effectiveness of antenatal birth plans to increase utilisation of health units for delivery care in this and similar settings. A cost effectiveness study comparing different ways of doing birth plans such as community vs. antenatal care would be useful too.
We thank the World Health Organization/HRP (UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction) for funding this study through a grant to London School of Hygiene and Tropical Medicine. We also thank Ford Foundation for the additional funding to the principal investigator's scholarship for a PhD degree at London school of Hygiene and Tropical Medicine. We offer special thanks to all participants in this study: antenatal and postnatal clinics attendees, TBAs, elder men in the two major ethnic groups in Ngorongoro, maternal health care providers, heads and administrators in all health units involved and Ngorongoro District Health Administration. Our appreciation and gratitude to Symon Wandiembe and Erin Anastasi, London School of Hygiene and Tropical Medicine, for their assistance at various stages of this manuscript.