Impact of Birth Preparedness and Complication Readiness Interventions on Birth with a Skilled Attendant: A Systematic Review

Background Increased preparedness for birth and complications is an essential part of antenatal care and has the potential to increase birth with a skilled attendant. We conducted a systematic review of studies to assess the effect of birth preparedness and complication readiness interventions on increasing birth with a skilled attendant. Methods PubMed, Embase, CINAHL and grey literature were searched for studies from 2000 to 2012 using a broad range of search terms. Studies were included with diverse designs and intervention strategies that contained an element of birth preparedness and complication readiness. Data extracted included population, setting, study design, outcomes, intervention description, type of intervention strategy and funding sources. Quality of the studies was assessed. The studies varied in BP/CR interventions, design, use of control groups, data collection methods, and outcome measures. We therefore deemed meta-analysis was not appropriate and conducted a narrative synthesis of the findings. Results Thirty-three references encompassing 20 different intervention programmes were included, of which one programmatic element was birth preparedness and complication readiness. Implementation strategies were diverse and included facility-, community-, or home-based services. Thirteen studies resulted in an increase in birth with a skilled attendant or facility birth. The majority of authors reported an increase in knowledge on birth preparedness and complication readiness. Conclusions Birth Preparedness and Complication Readiness interventions can increase knowledge of preparations for birth and complications; however this does not always correspond to an increase in the use of a skilled attendant at birth.

. Therefore we set out to systematically review the literature, including qualitative studies, for the effect of BP/CR on increasing SBA [19].
The aim of this systematic review is to review the literature of BP/CR interventions and assess its effect on increasing SBA [19].
As there are several ways to implement and evaluate BP/CR interventions, we formulated the following key research questions to guide our review: 1. To what extent does BP/CR result in increasing skilled birth attendance?
2. What strategies are used to implement BP/CR? 3. What methodologies are used to measure the effectiveness of BP/CR? Findings in this paper are also included in the WHO recommendations on heath promotion interventions for maternal and newborn health 2015 [20].

Methods
In order to systematically synthesize the body of evidence, we followed the guidelines for systematic reviews of the Cochrane Handbook for Systematic Reviews of Interventions [21], the PRISMA statement [22] and the guidelines published by the National Health Service (NHS) Center for Reviews and Dissemination [23]. Details on the specific review methodology can be found in a prior publication (S1 File) [19]. The study protocol was registered at PROSPERO (no: CRD42012003124). Additional methodological considerations not mentioned in the study protocol or which were adjusted during the review process are described below.

Literature search and selection process
We developed a search strategy (S2 File) for three electronic databases: PubMed, Embase and CINAHL. A wide range of search terms was used for high sensitivity as we anticipated that BP/ CR terminology had only recently been used in publications. Originally we searched articles published between January 1987 and October 2012. However, this resulted in many irrelevant articles, in which concepts and interventions related to BP/CR were difficult to identify. We therefore excluded studies published before January 2000 and limited our search to the English language. We also manually searched grey literature and reviewed a database that included results of a systematic mapping of research on maternal health in low-and middle-income countries published from 2000 to 2012 [24,25]. The latter was limited to Arabic, English, French, Spanish, Japanese and Portuguese.

Inclusion criteria
Studies were included if they were RCTs, quasi-experimental studies or comparative cohort studies which met the following criteria: • Study population: pregnant women, women who recently gave birth, husbands of pregnant women or of women who recently gave birth, health care providers, traditional birth attendants (TBAs), all adults in the community (in low-and middle-income countries) • Interventions: including BP/CR components, which could be facility-based, communitybased or home-based implemented both as single intervention or as a package of interventions.
• Comparison: outcome reported must be compared with the outcome in any comparison group, • Outcomes: birth with SBAs or facility births, maternal and neonatal mortality and morbidity, ANC with a skilled provider and knowledge of danger signs, implementation of BP/CR plan elements such as saving necessary funds, transport arrangements, etc We excluded interventions that focused on increasing the quality of ANC provision and studies on facility training without the objective of increasing BP/CR. We also excluded descriptive studies on BP/CR, which did not evaluate any BP/CR intervention, but merely described barriers to BP/CR or use of SBAs.

Study selection
Our search yielded 5552 records, of which 3665 remained after removal of duplicates (Fig 1). All abstracts and titles were searched and screened in duplicate and independently by ASM, YR and MvE. Of the 3665 records, 2991 were found not relevant or published before 2000. Remaining records (n = 674) and additional records identified (n = 12) were compared against JHPIEGO's BP/CR matrix [10] to determine whether the study's conceptualization aligned with the definition of BP/CR used in this review. Two reviewers reviewed the remaining 171 full text records independently for inclusion (ASM, YR, MvE or LS). Reference lists of the included records (n = 21) were hand searched for potentially relevant sources, yielding 9 additional records. Three additional studies were identified after presentation of our preliminary results at the WHO Technical Consultation on health promotion interventions for maternal and newborn health [26]. Disagreements on inclusion of studies were resolved by discussions with third parties (JS, JvR and AP). Included were 33 records covering 20 separate intervention programmes (e.g. some interventions or studies produced multiple articles).

Data extraction
We tailored the NHS Center for Reviews and Dissemination data extraction table to fit our research questions [23]. Data extracted included setting, study design, outcomes, funding sources and intervention description. Data extraction by ASM was checked for accuracy and completeness by YR and LS. If additional data was needed, reports or data files were acquired by contacting authors and / or searching for study reports online.

Quality assessment
ASM, YR, MvE and LS independently assessed the quality of the studies. Risk of bias for quantitative studies was assessed using the Effective Public Health Practice Project (EPHPP) quality assessment tool [27]. The quality of qualitative studies was assessed using the eight criteria developed by Walsh and Downe (2006) [28].

Data analysis
Quantitative results were summarised in summary of evidence tables. The studies varied in BP/ CR interventions, design, use of control groups, data collection methods, and outcome measures. We therefore deemed meta-analysis was not appropriate and conducted a narrative synthesis of the findings. [29].

Results
All 20 programmes consisted of interventions wherein BP/CR comprised one element, either as a component (e.g. of ANC education), a sub-intervention (e.g. of a behavioural change strategy) or a primary intervention (e.g. administering a BP/CR package). See tables 1 and 2 for characteristics of included studies. The study designs include one RCT, three cluster RCTs,  seven pre and post comparative studies with a control group, one pre and post study without control, seven one group before and after evaluations and one qualitative study. Three quantitative studies also had a qualitative component. Five of the 19 studies with a quantitative component received a moderate and 14 a weak rating. Assessment of four qualitative studies resulted in one moderate and three weak ratings (S1 Table).
The studies were conducted in sub-Saharan Africa (n = 7), South East Asia (n = 12) and Central America (n = 1). The Maternal Neonatal Health (MNH) programme supported by the Johns Hopkins University Centre for Communication Programs (JHU/CCP) in Guatemala, Nepal, Indonesia and Burkina Faso and the Skilled Care Initiative of Family Care International in Burkina Faso, Kenya and Tanzania were multi-country programmes.
Results of studies are presented in Tables 3 and 4. We distinguished between BP/CR programs that aim to increase SBA for all births and those promoting SBA in case of complications. The latter took place in contexts with extremely low SBA and where the majority of births take place at home; consequently BP/CR messages are different and focused on care seeking for complications and the intervention also contributed to ensure safe birth practices at home.

Effect on birth with a skilled attendant
Across multi-country programmes, i.e. Skilled Care Initiative and the MNH programme, results varied. The Skilled Care Initiative found increases in SBA in Burkina Faso and Tanzania, but not in Kenya [30,36,39]. Exposure to BP/CR interventions in Tanzania correlated with increased likelihood of seeking skilled care during childbirth. Of respondents exposed to ANC counselling on BP/CR 74% sought skilled care versus 64% of those unexposed (p<0.05) [39].
The MNH programme resulted in an increase in facility births or birth with SBAs in Burkina Faso and Guatemala [42,43]. No improvements were found in Nepal and Indonesia [40,41]. In Burkina Faso improvements were mainly due to an increase in births assisted by auxiliary midwives from baseline to endline (15.6% to 41.7%, p<0.05), which was higher for the exposed group. All authors of the MNH programmes reported an increase in knowledge of BP/CR and increase in BP/CR actions, however, this did not necessarily increase seeking skilled care [43].
In Tanzania, an intervention package, comprising training of Safe Motherhood Promotors and education on the importance of a birth preparedness plan through home visits, showed an increase of 51.4% in SBA post-intervention compared to 34.1% at baseline (P<0.05) [46]. 3), while the facility births in the control group increased from 4% to 15%. In India, a birth preparedness intervention geared towards families and communities resulted in an increase in births at primary care facilities (p<0.001) and government hospitals (p<0.001) [51].
Of the six studies of interventions aiming to increase access to care for complications, three resulted in increased facility births [54,59,55]. Hossain et al (2006) implemented a multi-stakeholder intervention consisting of facility-based interventions (facility upgrades and improvements in quality of care) and community interventions addressing birth planning and community mobilization to ensure timely recognition and referral of obstetric emergencies. The intervention site received all interventions, the comparison site only a facility upgrade and the control site received none. The intervention area showed an 8.1% increase in facility births (p<0.01 95% CI 7.2-9.0); however, both the control and comparison area had a higher preintervention facility birth rate. Darmstadt et al (2010) found a significant increase of facility births in the intervention area (from 12.1% at baseline to 20.2% at endline) compared to the control area (increase from 12.5% at baseline to 16.5% at endline-P<0,05) [54].
Most authors reported a statistically significant improvement in knowledge on BP/CR (Table 4). Mullany et al (2007) showed that couple counselling significantly improved knowledge compared to individual counselling and they suggested that immediate conversations between spouses might enhance knowledge retention [45]. Knowledge acquired was not always consistently related to the intervention as was shown by , who found that knowledge of danger signs was higher in the control group [40].
ANC attendance was not evaluated in all BP/CR studies. Study results varied from a significant increase in ANC attendance [36,39,42,47,48,54,43], earlier booking dates [46,48] to not any effect [40,45]. Different outcome measures and cut-off points for frequency or timing of ANC visits were used.
Of the five studies reporting on neonatal mortality, Kumar et al (2012) report significantly lower neonatal mortality in the BP intervention group [52] and Hodgins et al (2009) showed fewer neonatal deaths over time [50]. No significant difference was found in other studies. No. In the intervention area SBA increased from 27% at baseline to 28% at endline (p-value not provided, authors report non significant). In the control area there was higher increase from 30% at baseline to 37% at endline (P<0.05) Marginally: ANC visit at least 1 increased in the intervention group form 85% at baseline Intervention: baseline: 85% endline: 89% P<0.01. ANC visit >2 in the intervention group increased from 67% at baseline to 73% at endline<0.01. Similar changes were seen in control district. Birth preparedness counseling and information provided at ANC in the intervention group increased from 35% at baseline to 84% at endline (p < .001), However an increase was also seen in the control area from baseline:29% to endline:81% (p < .001) Family Care International Tanzania, 2007 [39] (Tanzania) Yes. In the intervention area SBA increased from 48% at baseline to 54% at endline (p = 0.01) compared to the control area with a decline from 38% at baseline to 31% at endline (p-value not provided, authors report non significant) Yes. Significant increase in exposed group (no significant change in unexposed area) for earlier ANC visit mean decreased from 7.0 to 6.1 months (p = 0.05) and ANC visit at least 1 increased from 88% at Baseline to 95% at endline (p < .001). Improvements were seen in endline counseling in BP/CR (Increase from 18% to 35% in the intervention district p<0.001) and from 24%-32% in the control district (p<0.01)) and advice on where to give birth increased in the intervention district (44% to 57%, p < .001) but declined in control district (48% to 42%, p < .01).

Sood et al, 2004 [40] (Nepal)
No. Birth assisted by a doctor increased from 11.6% at baseline to 34.4% at endline. However, this was higher for the unexposed group (42%) compared to the exposed group (29.3%). Births attended by a nurse decreased from 0.8% at baseline to 0.0% at endline.
Partially. For knowledge there was a significant increase in exposed compared to unexposed for: vaginal bleeding as danger sign during pregnancy. An increase, but no significant difference between exposed/unexposed mentioned for severe post partum vaginal bleeding, high post partum fever, awareness of community schemes for transport and funds. A reduction was seen in Knowledge of prolonged labour as danger sign both in all groups (due to inconsistent terminology used). No effect was seen for retained placenta as danger sign. For practice, >4 ANC clinics attended in all groups, effect of intervention: ns. Arrangements for safe childbirth increased in all groups, effect of intervention: ns

Sood et al, 2004 [41] (Indonesia)
Marginal. Woman's reported use of a SBA at birth decreased from 64.4% and baseline to 58.9% at endline. This decline was mainly due to lower reported use of health facility midwives (18% to 7.6%). There appeared to be an increase in birth with a SBA among the exposed group with significant difference between exposed and unexposed groups. Hospital birth did increase from 7.1% at baseline to 9.0% at endline (p<0.05) This was higher for the exposed group (11.4%) than the unexposed group (5.7%) (p<0.000) Yes. Significantly higher awareness of vaginal bleeding as danger sign in pregnancy in all respondent categories, (e.g. women: 40.7% exposed group compared to 16.4% in unexposed group) and of vaginal bleeding during labour only in women (30,8% exposed to 12.3% unexposed), for postpartum bleeding significantly in all groups exposed compared to the unexposed. Significantly higher awareness of community assistance schemes in exposed group compared to unexposed. Emergency transport schemes were used more often by the unexposed. Knowledge of fever as danger sign decreased. For ANC visits there was no baseline data available for comparison (Continued) Yes. FB increased from baseline (30.5%) to endline in the unexposed group to 31.2% and in the exposed group to 54.7%. P<0.01 between baseline and follow up P<0.01 between exposed and unexposed Yes. Knowledge (of danger signs and community plans for transport and funds) increased significantly (between p<0.01 and p<0,05 for testing difference between exposed and notexposed), except for fever as danger sign. Seeking care for ANC visits in second trimester increased significantly among those exposed (
Skinner et al, 2009 [49] (Cambodia) Marginal. There was no baseline data collection in the intervention areas. Outcome data where extrapolated from existing data sources. Routine health facility data of the 10 facilities in the intervention area showed a 32% increase in the number of women giving birth with a midwife (2005 n = 271 and 2006 n = 357). The national average also increased in this period with 13% No. There was no baseline data collection in the intervention areas. Antenatal care visits increased to 22% according to existing data of the facilities.
Hodgins et al, 2009 [50] (Nepal) Marginal. Percentage of respondents who delivered in a health facility (among respondents with live birth) increased from 24.0% to 28.4% (OR 1.31 95% CI 1.10-1.57). In Banke the proportion rose markedly but in Jhapa, where the baseline rate was already high, there was little change.
Yes. Neonatal mortality decreased from 20/1000 (95% CI: 14 to 27) to 8/1000 (95% CI: 4 to 13) at endline. Adjusting for literacy and wealth differences between baseline and endline survey, as well as the cluster design, this yields an OR of 0.42 (95% CI: 0.24 to 0.72). Positive changes were seen in household practices for birth preparation. Setting aside money increased from 34.8% at baseline to 81.9% at endline (OR 9.78 6.93-13.80). Where 11.5% made arrangements for health facility delivery before birth at baseline, this increased to 19.9% at endline (OR2. 10  Yes. Household with a pregnant woman who had a birth plan in place increased from 0 at baseline to 48% at endline. Pregnant women who were able to identify 2 or more danger signs during pregnancy and delivery increased from 10% to 56% at endline. Obstetric complications attended at the district hospital increased from 4% to 15% at endline. A total of 44 of 52 communities had descriptions of action plans for transporting people with health emergencies to health facilities, and 12 (23%) had a specific system in place to implement the transport system (e.g. had collected funds) (Continued) between 2004 and 2012 suggesting increased interest in BP/CR and using the label of BP/CR after introduction by JHPIEGO.

Methodologies to measure effectiveness
Definitions of BP/CR varied from identifying a place of birth and preferred SBA, to preparing funds for complications, to arranging for (emergency) transport and knowledge of danger signs. Focus was either solely on the mother, or on both the mother and newborn. Across studies, household surveys were most frequently employed to evaluate programme effectiveness. Although interventions targeted different study populations (women, husbands and mothers in law), authors almost exclusively evaluated women's behaviour as primary outcome [62][63][64][65][66][67][68]. The study population was heterogeneous across studies, ranging from pregnant women who gave birth during the study period [45,52] to women who had 'recently' given birth. A number of multi-country programmes had outcome measures at family or health worker level. Some authors measured facility birth as an indicator for SBA. Methods to assess if women were birth prepared and complication ready differed greatly across studies, due to varying scales and index criteria used.

Discussion
Heterogeneity of study designs and BP/CR interventions, and lack of high quality evidence prevents making a pooled analysis. Although BP/CR interventions can increase knowledge of danger signs and preparations for birth and complications, this did not always correspond to an increased use in SBA at birth. Where an increase in SBA or facility birth was reported, BP/CR interventions were primarily part of a package of multiple interventions and involved multiple stakeholders, making it difficult to attribute the effect to the BP/CR component alone.
Interventions where BP/CR was a primary component can be better assessed in terms of causality, but what these results mean in a complex reality is unclear [45,46,69]. Many variables influence both programme interventions and outcomes such as female education and policy changes [48,70]. Active involvement of policy makers in BP/CR interventions facilitated implementation at the national level in some countries [30,[40][41][42][43]. We will analyse this further in a separate publication [70]. Increased use of SBA in BP/CR programmes within a package of interventions could be due to facility or infrastructure improvements, community-based behavioural change interventions, other factors, or due to interactions between all [30]. Although the JHPIEGO BP/CR matrix includes preparedness of facilities and health providers, BP/CR studies rarely focus on the supply side of skilled care [10]. Ensuring that services are equipped to meet the increased demand likely to be generated by BP/CR interventions is crucial. Advising community members to prepare for facility birth, while health services or health providers are not birth prepared or complication ready, or while local health systems are not ready for an increased caseload, can lead to an increase of in-facility complications or maltreatment. Consequently, negative in-facility experiences can increase delay in care seeking [5] and should be avoided as much as possible. Also, negatively contributing factors at ANC need to be addressed for proper BP/CR counselling such as insufficient human resources and time constraints [69,71,72].
Although most studies report increased knowledge of BP/CR, not all clarify whether this resulted in plans or actions. Knowledge alone does not equate to an increase in care-seeking behaviour, especially for maternity care services, often due to financial, structural, geographical or cultural factors [73,74]. Studies that focussed on 'knowledge on danger signs' and 'preparing transport and funds in the event of an emergency' predominantly aimed to increase access to Emergency Obstetric Care (EmOC) in case of complications. However, most births start uncomplicated and risk identification is unreliable [75]. Delays in reaching skilled care are partially caused by delayed recognition of signs and symptoms of labour onset [76,77]. We argue that BP/CR programmes should follow Safe Motherhood programmes in their shift towards the promotion of skilled care for all births and include education on the signs of uncomplicated labour to ensure timely preparations [78].
The strength of this review lies in its broad literature base, including published and unpublished studies (e.g. reports from NGOs). Although we limited our initial search to English language, the systematic mapping of maternal health research did not have this limitation. It is likely that we included all relevant studies by crosschecking our search results with this broad database and by being open for inclusion of additional articles at the WHO Technical Consultation [26]. Facility birth in many studies was used as an indicator for birth with SBAs, this must be interpreted with caution as many facilities may lack the availability or presence of SBAs [35]. Similarly only four studies presented their definition of an SBA and it is unclear if in the other studies a SBA was defined according to our definition. Despite wide spread promotion of BP/CR through the JHPIEGO and WHO publications, definitions and indicators of BP/ CR varied greatly across studies, therefore comparing studies was challenging which also prevented the possibility of conducting a meta-analysis.

Conclusion
Although BP/CR in theory is compelling as a strategy to increase birth with a SBA, robust evidence of the effect of BP/CR in itself on increasing birth with a skilled attendant remains limited. This review does suggest that BP/CR interventions in combination with other interventions have the potential to increase use of SBAs and to increase timely use of facility care for birth and obstetric and newborn complications. We argue that BP/CR interventions seem as strong as the weakest link in the continuum of maternal care pathway.

Recommendations
Clarification of definitions of BP/CR is needed to guide future programme implementation and evaluation. Expert meetings and internationally-agreed upon definitions and indicators for BP/CR could help. However as specific actions and messages required to prepare for birth and complications are highly context specific, it seems undesirable to aim for uniformity. Creating a flexible BP/CR definition that allows local adaptation is a step forward. Collaboration between target groups is a crucial step, and requires further study. An excellent way to locally define and implement BP/CR programmes would be to develop and study local BP/CR pathways collaboratively with target groups from community to policy level. The JHPIEGO matrix is a helpful tool to start this process. Study of this process and outcomes, should include mixed methods by transdisciplinary research teams [79].