Maternal and child health handbook to improve continuum of maternal and child care in rural Bangladesh: Findings of a cluster randomized controlled trial

This study aimed to evaluate the effectiveness of maternal and child health handbook (MCH) enhanced by mobile tools and to generate evidence informing the adoption of the program in Bangladesh. A cluster randomized controlled trial (RCT) has been implemented in Lohagora of Narail District and Dhamrai of Dhaka District. Unions of the study settings were randomly allocated in either one of three groups: (1) Intervention 1 using both mobile platform and MCH, (2) Intervention 2 using MCH alone, or (3) the Control. A total of 3,002 participants were recruited. The interventions were designed to promote two-way communications between pregnant women/their families and community health workers by an empowering approach. A total of 3,002 pregnant women were recruited. As the results, the interventions both significantly improved the utilization of CoC, although the overall proportion of CoC was relevantly low: 2.79% in the Control (95% CI: 1.37–3.54%), 6.16% in Intervention 2 (95% CI: 4.67–7.86%), and 7.89% in Intervention 1 (95% CI: 6.29–9.90%). Neonatal mortality rate with and without CoC was 5.43 per 1,000 (95% CI: 3.63–9.57 per 1,000) and 34.8 per 1,000 (95% CI: 24.3–45.4 per 1,000), respectively. Our study indicated the effectiveness of the interventions by leveraging MCH and a mobile platform to promote uptake of CoC throughout prepartum, intrapartum and postpartum/neonatal periods, potentially bringing long-lasting benefits to mothers and their offspring. The explicit approach is expected to guide policy makers to adopt MCH interventions in primary healthcare strengthening at the community level. Trial registration: UMIN000025628 Registered June 13, 2016.

Introduction order to inform updates of the MCH in the context of Bangladesh and policy making for the targets of SDGs related to maternal and neonatal health.

Study settings and participants
The community-based cluster RCT (trial registration: UMIN000025628) was conducted in two upazilas (administrative regions in Bangladesh), Dhamrai in Dhaka District, Dhaka Division and Lohagora in Narail District, Khulna Division from February 2017 to August 2018. The demographic characteristics of the study sites was summarized in Supplementary file (S1 File). The study period covered the duration from the start point at which the pregnant women were identified and recruited to the end point, when the participants came through the fourth week after giving birth, or terminated the pregnancy due to miscarriage, stillbirth and neonatal / maternal mortality. The cluster, namely the unions in each Upazila, rather than the individual, was subjected to the cluster randomized sampling. These unions in the targeted study sites have homogeneous socioeconomic characteristics such as income level of household, annual birth rate, accessibility to primary healthcare, literate rate, school attendance, and hygiene conditions in household. The sampling and randomization process was briefly described in the protocol and Supplementary file (S1 File). The selected unions were randomly allocated to either 1) the intervention that combined mobile phone communication with MCH, 2) the intervention using MCH alone or 3) the control, where the proposed intervention was voluntarily implemented after the study.
The target population was pregnant women aged 15 to 49 years living in the selected settings and expected to give birth between 1 August 2017 and 31 July 2018. The eligible criteria include: i. currently having a good health status, without any maternal complication; ii. living and planning to give a birth in the study settings during the period from February 01, 2017 to August 30, 2018 and iii. willing to participate to the proposed study with agreement to the informed consent. The pregnant women were basically identified from two routes: the registration list in the upazila health complex; and field activities of the community health workers (local NGO staffs), as they have close interactions with the rural residents.
The study also included healthcare providers at the community. We also included health professionals in each upazila as required. A total of 3,002 participants were finally recruited, including 998 for the intervention 1, 1,001 for the intervention 2 and 1,003 for the control. Table 1 summarizes the participants in the study settings. Details of study design and sampling issues were described in our published protocol [26].

The interventions
The two interventions were designed to promote two-way communications between pregnant women/their families and CHWs by an empowering approach. The intervention 1 applied mobile platform, including text and / or audio messages and phone calls if necessary, which were combined with MCH, while the intervention 2 utilized MCH alone. Contents of MCH encompassed the general profile of pregnant mother, menstrual history and history of previous pregnancy (if any), records of health education and consulting, records of conditions/health status, healthcare utilization and clinical results during pregnancy, delivery and postnatal/neonatal period, as well as information on common complications and signs of danger, on health seeking for mothers and babies, and on daily care and nutrition. In the two interventions, MCH was distributed to each participant at the point of recruitment. Every two months the enrolled pregnant women and their families and CHWs were organized for a community meeting, where health education, consulting/advice and anthropometric measurements were provided to accompany the discussions on seeking health services for mothers and babies and the application of MCH. Additionally in Intervention 1, besides MCH and community meetings, user-friendly mobile messages were developed and sent according to the gestational age (GA), including reminders of antenatal and postnatal care visits and facility-based delivery, list of locations of skilled birth attendants and hospitals, GA-specific health issues, daily care and nutrition during pregnancy, intake of iron tablet and folic acid, support from husband and families during pregnancy and lactating period, signs of danger, signs of labor, and postnatal/neonatal care. Audio messages and phone call were also used for follow-up and consulting/advice, as necessary. For those participants in households without mobile phones, the community staffs made regular visits to their home according to their GA to provide equivalent information.
The primary study target was the pregnant women, and their families (husbands, motherin-law and / or mothers) were also invited to participate to the interventions, as they were substantially involved in daily home-based care during and after pregnancy, as well as seeking and utilization of healthcare services during pregnancy, childbirth and neonatal period in the study settings. By leveraging the community network, pregnant women and their families were organized for health education / health promotion activities, and a linkage with the community health workers were established to provide advice and referral when necessary. Through the network built, the interventions also closely involved communications among health care providers at different levels: for example, those healthcare staffs at the community were able to provide supports and advice and refer those with maternal complications to health professionals at the upper-level facilities (Upazila and District), and health professionals at the referral level were able to provide necessary guidance and training as well. The study procedure was briefly described in the protocol and Supplementary file (S1 File).

The expected outcomes
The expected outcomes were neonatal death, fetal death (stillbirth/miscarriage), preterm birth, low birthweight, maternal pregnancy complications and referral, antenatal care visits for at least one time (ANC1), antenatal care visits for at least four times (ANC4), antenatal care visits for at least six times (ANC6), facility-based delivery (FBD), mode of delivery, utilization of postnatal/neonatal care (PNC), and the experience of health education. The definition of neonatal deaths followed the standard employed by WHO, that is, death within the first 28 days of life. By referring Bangladesh Demographic and Health Survey (BDHS) 2014, neonatal deaths and fetal deaths were determined from the complete birth history from mothers and recorded by our trained staffs [28]. The continuum of care (COC) for mothers and babies in the study referred to healthcare services during pregnancy, at birth and after birth, and the variable was then created by combining ANC4, FBD and PNC.

Data analysis
For data analysis, univariate analysis was first performed to explore the characteristics of variables. In the comparison of each variable, the equality of covariates of the three groups at baseline was examined by a stratification. To examine the effect of the interventions on the targeted outcomes, we carried out multivariate logistic regressions. We considered potential correlations in the participants from the same unions. Then, generalized estimation equations (GEE) with an independent correlation structure were performed. The covariates were adjusted by using ratio residuals for each cluster obtained from the logistic regression models.
For the fixed effect and the mixed effect of the cluster, namely the unions, we also performed generalized linear mixed models (GLMM) for comparison, in which the union ID was subject to the randomization. Risk ratios (RR) and their 95% confidence interval (CI) were calculated for the targeted outcomes in the intervention groups compared to the control group, after controlling demographic and socioeconomic factors of the pregnant women, such as age, education, household income, and those related to healthcare accessibility. Data analysis was performed using Stata 15.0.

Ethical consideration
The study was approved by the ethical committee of Bangladesh Medical Research Council (BMRC), Bangladesh. After stipulating the objectives, procedures, risks, benefits, confidentiality and voluntary both verbally and literally, the pregnant women who were willing to participate were asked for a signature in the informed consent form. For those unable to read and write, their substitutes (basically their families, relatives or neighbors) were asked to sign on the form on behalf. Additionally, for those aged below 18 years, their guardians were asked to accompany with, and then sign on behalf of them for the informed content. These consent procedures as described were approved by the ethical committee. The details of demographic and maternal characteristics of the participants were all listed in Supplementary file (S1 File). The average age of the participants was 23.53 years (SD: 4.67 years). The average gestation age at birth and birth weight was 37.4 weeks (SD: 1.8 weeks) and 2,841.24 g (SD: 439.64 g), respectively. The proportion of those live birth who undertook cesarean section was as high as 50.81%. Socio-demographical characteristics did not significantly differ between the intervention groups and the study settings at baseline.

Experience of health education
In the two intervention groups, all participants reported that health education was provided during pregnancy, while in the control group, there 36.5% participants reported no relevant experience. Regarding the (potential) usefulness of MCH for knowledge dissemination for mothers and babies, 99.9% and 81.2% of participants in the intervention and the control group, respectively, had a positive attitude.

Discussion
To our knowledge, this study is the first cluster randomized controlled trial to assess the effect of the MCH program enhanced by mobile platform This is also the first cluster randomized design for HBR to improve COC for mothers and babies in Bangladesh. Our findings indicated that the application of MCH improved uptake of multiple healthcare services, including antenatal care and postnatal/neonatal care, among rural pregnant women. The interventions increased ANC visits by using MCH and combining MCH and mobile platform, respectively. Although the overall proportion of at least four visits of ANC as recommended was relevantly low in the study settings, the figure in the two intervention groups, especially in the combined intervention, were better. A similar tendency was also observed in PNC. The combined  intervention further improved facility-based delivery and utilization of healthcare facility for complications during pregnancy and delivery. The multilevel GEE models identified statistical significance of these intervention effects, after adjusting potential confounders. Compared to the monitoring data of UNICEF which targeted the overall population during the study period in Bangladesh [29], our study which targeted pregnant women living in rural areas identified higher NMR of 29.7 per 1,000 (95% CI: 21.6-37.8 per 1,000). Although the estimated figure was lower in the two intervention groups, no statistical significance on the immediate efforts to reduce mortality and morbidities was identified. A possible reason for this could be the calculation of the study sample size was based on an NMR of 24.4 per 1,000 (derived from the final MDG report), while the indicator has been substantially reduced since then. We also acknowledged that unlike obstetric care practices, MCH does not have an immediate life-saving effect and that the universal access to good-quality obstetric and neonatal healthcare plays a key role in reducing NMR based on the success observed in Bangladesh and other developing settings [30,31]. On the other hand, consistent with the findings of a systematic review, [32] our analysis confirmed that a crucial determinant in reducing NMR was COC; both interventions showed a significant improvement. This suggests that MCH has a potential to improve neonatal survival through the promotion of utilization of COC for mothers and the newborn. In our study, MCH brought upon several benefits, such as health education, promotion of daily care awareness and practices, involvement of husband and family members and boosting communication between pregnant women and healthcare providers, especially CHWs, leading to better healthcare utilization during pregnancy, at birth and after birth. This was compatible to previous studies on MCH [16][17][18][19][20]. The interventions involved primary healthcare at the community as an inevitable aspect. In the intervention settings, and the local residents, including pregnant women and their families, were organized and networked, and community meetings aiming to strengthen participatory learning and action on preventive and care-seeking behaviors were also implemented regularly. Similar empowerment practices have proven to be effective in improving key behaviors and neonatal survival outcomes, although its mechanism may depend on local practices, capabilities and the responsiveness of health services [33]. In our study, during this empowerment process, MCH or MCH combined with the mobile platform were the key instruments. CHWs were mobilized to reinforce the linkage, deliver knowledge and primary care, organize the community meeting and bridge pregnant women and healthcare facilities, in order to accomplish the proposed interventions. To this end, the results suggested that MCH can be a useful tool to strengthen primary healthcare delivery in rural Bangladesh. The interventions largely filled the gap of health education during pregnancy and routine primary healthcare at the community level, and the (potential) usefulness of these interventions were definitely recognized among most participants.
Compared to MCH alone, the combined intervention achieved better utilization of COC, especially in terms of facility-based delivery and care seeking for complications during pregnancy and delivery. What works for this intervention were likely to be effective contacts and more frequent interactions between pregnant women and CHWs, such as sharing information and advising daily home-based care, together with seeking relevant healthcare based on individual needs and requirement. Text and voice messages complemented MCH in knowledge dissemination and deepening the understanding of the key contents of MCH. The high mobile coverage and the low costs in the study settings facilitated the intervention. The results added relevant evidence on the effectiveness of mHealth on improvement of maternal and neonatal outcomes and related care seeking by the high-quality study design, which were of lack in lowand middle-income countries [34], and suggested the value to apply these effective tools in primary healthcare at the community level.
Our study revealed the latest status of universal health coverage for mothers and neonates in rural Bangladesh. The uptake of ANC4 among rural pregnant women living in the study settings was considerably lower than that of the overall population identified by BDHS 2014 [35,36], but was comparable to that of community-based studies conducted in a rural area [37,38]. This can be explained by a substantial rural-urban gap in the uptake of maternal healthcare services [39]. Contrary to the stagnant progress in ANC uptake, our results suggested a fairly progressive uptake of PNC and FBD compared to previous surveys and estimates [40,41]. The overall low uptake of these maternal and neonatal services suggested a big room for improvement through strengthening primary healthcare as the frontline of health system [42], particularly in rural areas.
The incidence of cesarean section (CS) identified in our study was much higher compared to that in BDHS 2014 [43], and largely exceeded the optimal rate ranging from 5% to 20% [44], while we did not identify the significant impact of undertaking CS on neonatal mortality reduction. Although it is a life-saving measure in obstetric care, a high level of CS indicates a substantial proportion of the practice without medical indication, leading to wasting of scarce healthcare resources and a high health and economic burden, especially in low-and middleincome countries [45][46][47]. The mechanism of the high-level CS tended to be complicated, mixing motivations of both the supply and demand sides, and the decision of the mothers and their family may largely be affected by doctors due to poorly informed healthcare needs [43,48]. Our results suggested that this alarming phenomenon is emerging in not only urban areas, but also in rural areas recently, and an intervention by applying MCH and mobile platform had the potential to reduce the misuse. The emerging issues on CS in MCH for implementing health promotion/health education programs at community level are expected to be covered.
In interpreting these major findings, several issues should be carefully considered. The enrollment of the target pregnant women relied on self-report and local registration. Because of the variation in identifying pregnancy among the participants, gestational age at enrollment was diversified, causing differences in the participation duration. Moreover, our study was likely to be inevitably contaminated somehow, because the interventions and the outcomes cannot be masked, and there had been some previous NGO-driven health promotion campaigns and activities targeting the rural community in the study settings. However, there was no differences regarding these factors across the study settings and groups. Finally, because of the limited follow-up duration, our study did not observe the outcomes posterior to the neonatal period, potentially missing the overall effects of the target tools on maternal and child health.

Conclusions
In conclusion, our study indicated the effectiveness of the interventions by leveraging MCH and a mobile platform to promote uptake of COC throughout prepartum, intrapartum and postpartum/neonatal periods, potentially bringing long-lasting benefits to mothers and their offspring. These tools coordinated the interactions of pregnant women, their families and CHWs and their active engagement in primary healthcare at the community level, potentially contributing to better health outcomes. It is worth including these tools in primary healthcare to achieve universal health coverage for mothers and babies in rural Bangladesh.