Adoption of Mobile Phone Messages for Delivery and Newborn Care in Bangladesh

In Bangladesh, mobile phones have been adopted as a health communication tool to improve maternal and child healthcare. To understand what works and what doesn’t work for mobile phone based messages around delivery, postpartum and newborn care in resource limited settings we interviewed 33 women who enrolled in an educational service that provides twice weekly voice or text messages to pregnant women and mothers of 0-11 month old babies and had participated in a survey. This follow up qualitative exploratory study showed that women appreciated receiving messages around newborn care and nutrition information over pregnancy messages. Women in lowincome households faced challenges accessing messages on shared phones while those with low literacy and limited technological knowledge preferred to receive voice messages over text messages. Husband’s endorsement of the service improved women’s adoption of the messages. Knowledge on additional consultation service and information on how to enroll in the service for later pregnancies was found to be low. Some participants were reluctant to pay for educational messages and avoided the calls. Women’s healthcare practices suggested growing awareness on biomedical practices although women from lowincome households were more likely to follow traditional unskilled birthing and newborn care practices at cultural influences unless experienced complications. Besides providing contextual messages, a holistic response is required that includes; training local birth attendants, sensitizing female family members who organize the home based deliveries, and establishing a subsidized referral system to improve birth related health outcomes in lowincome households.

Mobile phone subscriptions have sky rocketed in Bangladesh; in 2012 total subscribers increased from 86.5 to 97.1 million in a year amongst a population of almost 150 million people (BTRC, 2012). There is a reliable platform for mHealth interventions and increasing mobile phone ownership among women (Khatun, Heywood, Ray, Bhuiya, & Liaw, 2016;Tran et al., 2015). However, there is little evidence on the effectiveness of mobile phone based educational services on delivery, and pre-and postnatal care in Bangladesh. Over the last two decades numerous mobile phone based interventions in Bangladesh did not extend beyond the pilot phase (Ahmed et al., 2014).
The "Aponjon" service, deployed in 2012, is an example of a global public-private initiative endorsed by the Government of Bangladesh that provides national and internationally endorsed guidelines on best practices around pregnancy, delivery and postpartum care. Culturally appropriate messages designed by national and international experts, are disseminated to pregnant women and mothers of 0-11 months old babies (Alam, D'Este, Banwell, & Lokuge, 2017;Rajan et al., 2013). The effectiveness of Aponjon messages in influencing behavioral outcomes has been measured (Alam et al., 2017;Chowdhury, Shiblee, & Jones, 2019). A survey conducted in 2014 on 476 subscriber women showed no or limited effect on delivery and postpartum healthcare practices such as delivery by a skilled birth attendant, initiation of breastfeeding right after birth, delay of bathing of newborn baby and completion of recommended postnatal check-ups (Alam et al., 2017). Factors such as women's higher education and wealth, older age and lower parity were associated with an increase in birthing at health facilities (Alam et al., 2017).
However, little is known about what prevented some women from following the recommended practices delivered through Aponjon's mobile phone messages, especially in low income households.
In order to better understand the adoption of mobile phone messages around maternity, newborn and infant care, a sub-sample of respondents from the aforementioned survey were recruited for in-depth interviews (Alam et al., 2017). This exploratory study aimed at understanding women's perception of message content, barriers to adopt messages and socio-cultural context of women's birthing and childcare practices. Our study provides further indepth evidence on the adoption of mHealth services, their relationship to MNCH behavioral change in Bangladesh and, more broadly, on the usefulness of mHealth in similar resource-poor settings where the rapid uptake of mobile phone based education interventions is yet to be fully assessed.

Intervention
Any woman with access to a personal or shared phone is eligible to enroll in the service during any stage of pregnancy or after childbirth. They are enrolled by community health workers, Non-Government Organization (NGO) clinics or service call center (Rajan et al., 2013). Subscriber women receive two messages (voice or text) per week with an optional additional message for their husband or other family members, appropriate to their gestational stage (Rajan et al., 2013). After childbirth pregnant women are upgraded to a "newborn service" by call center agents on confirmation of delivery. The service terminates when the infant reaches his/her first birthday, however, any subscriber can opt out earlier (Alam et al., 2017). There is an additional hotline service where subscribers can consult with medically trained doctors on maternal and newborn health issues at a regular call rate (Alam, Banwell, Olsen, & Lokuge, 2019). The voice messages are in Bangla and are one minute long while each text message typically contains 161 characters in transliterated Bangla format to support a full range of mobile phones (Alam et al., 2017). A subscriber can retrieve missed messages (voice) by dialing the service and following an automated voice-prompted menu. The messaging service is free for low income households and costs USD 0.02 (BDT 2.3) for others (Alam et al., 2017). Are you experiencing water break? I will talk more on delivery symptoms next week.
If you are a working mother, probably you are getting ready to get back to work. You can store breast-milk in a cup before you leave for work. The milk will be okay for the next 8 hours. In your absence, someone can feed the baby breast-milk with a spoon. Remember to inform the caregiver about the feeding procedure and timetable.

Study Area
Bangladesh is located at the top of the Bay of Bengal and shares borders with India in its' north, west and east and with Mymanmar in south-west (Ministry of Health and Family Welfare, 2015). The majority of the population is Muslims (89.3%) with the remainder (10.7%) comprised of Hindus, Buddhists, Christians and other religious groups (DGHS, 2012). Administratively the country is divided into 8 large divisions and 64 districts (BBS, 2019). For this study five districts were selected from the northern (Bogra), south-east (Chittagong, Laxmipur) and southern (Bagerhat and Patuakhali) part of Bangladesh in order to maximise diversity in culture, language and geographical location of participants. All participants in this study were initially recruited to the service by their designated community health workers.

Interviews
All study participants were subscribers to the "Aponjon" service and had participated in a previous survey in 2014 when their last born child's age was between 3-18 months (Alam et al., 2017). One hundred and ten respondents from the original survey, living in selected study areas were approached. Of these 35 women were able to be re-contacted by phone almost two years after the survey and 33 agreed to be interviewed and two declined.
Approximately 25% of dialed phone numbers were disconnected, and 43% of participants did not answer after two contact attempts. If the phone belonged to a family member who was away from home, an alternative time was arranged to discuss the research with the subscriber woman before taking her consent for the interview. The first author interviewed the 33 women over the phone using either Bangla or a local dialect (for respondents in Chittagong) between December 2015 and January 2016. Interviews, lasting between 15-35 minutes, were recorded with the consent of the participant after an information sheet was read to them. Participants were given their rights to not answer any question and withdraw from the research without any consequences. The study received approval from an International Review Board (IRB) and an IRB exemption from related organizations for conducting the study in Bangladesh.
Information on the participants' socio-economic status such as age, level of education, family expenditure and pregnancy history was collected. Open-ended interview questions were used to explore women's experiences of Aponjon, mobile phones in general, and healthcare seeking behavior for pregnancy, birthing and pre-and postnatal period. Women were asked a range of questions such as: "Can you tell me your experience regarding Aponjon service and messages (who accessed the messages, usefulness of information, barriers to receiving messages)?", "Would you recommend the service to others or re-enroll for next pregnancy, why or why not?", "Can you tell me about your life during pregnancy (check-ups, type of provider, preparedness for delivery, nutrition, vaccination, family custom)?" "Can you tell me about the last birth and previous births (selection of delivery place and birth attendants, complications, people involved in decision making), "Can you tell me what (feed) the baby was given right after birth and the reason behind it?", "Can you tell me when did the baby receive the first bath and who initiated it?", " Can you tell me how you took care of the umbilicus of new born baby and what was the source of your information?", "Did you see anyone for check-up or symptoms within the first 42 days after childbirth? Why or why not?", "Who do you usually contact for your or infant's health?" and "Have your children been immunized?".
Interviews were transcribed verbatim and translated into English by first author. The transcripts were uploaded into Atlas.ti for storage, organization and analysis (Friese, 2012). Thematic analysis was used to identify pertinent and common experiences across participants. This involved repeated readings of transcripts by first author, and coding to develop themes. The analysis was guided in part by the interview questions related to women's experiences of Aponjon, maternal and newborn care practices and socio-cultural contexts in which the service was received. A number of sub-themes were identified inductively and then organized under three key themes (Braun & Clarke, 2006).

Results
Background information of the participants is provided in Table-3. Participants were mostly young mothers aged less than 25 years. Participants exemplified a wide range of educational background ranging from having never attended school to completing secondary education or more. Women were mostly housewives, meaning they were reliant on their husbands and other family members for living and decision making on major expenditures. Almost half of the participants came from poor households (monthly income USD 118.15). Although half of the participants were enrolled in pregnancy, in general participants were more likely to remember information about their newborn baby's nutrition, development, growth and details on caring than their personal health. Automated mobile phone messages according to women's gestational stages helped new mothers to cope with their journey in motherhood while experienced mothers were reminded of the recommended practices. Often participant women shared the messages with elderly female family members who were also involved in taking care of the baby.

Involvement of husbands
In low-income households where women were less educated and not involved in income generating works, husband's encouragement was essential for women's adoption of the service. Participant women were able to utilize the service better and remembered to follow the recommendations when their husbands were regularly involved in receiving and relaying messages:  Barron et al., 2018;mWomen, 2012;Obasola, Mabawonku, & Lagunju, 2015;Watson, Sabumei, Mola, & Iedema, 2015). Additionally, messages were often not accessed because participant women had forgotten to keep adequate balance in the phone or did not find the information interesting. However, lack of personal phone and coordination with family members in arranging shared phone sets stood out to be the foremost barrier that prevented participant women from getting regular weekly information (Huq, Azmi, Quaiyum, & Hossain, 2014;Khatun et al., 2017;Poorman, Gazmararian, Elon, & Parker, 2014;Tran et al., 2015). Participant women, who did not own their phones or shared phones with family members, relied heavily on the phone owners to access the messages. Because voice messages were programmed to be sent out on a specific time of specific days of the week according to subscriber women's preference during the time of enrollment, it was essential that the phone owning family member passed on messages to the subscriber women:

Literacy
Literacy for mHealth can be of two types: the ability to operate mobile phones efficiently and the ability to comprehend voice or text messages. Although Aponjon provides voice and text messages at the same price, most respondents opted to receive voice messages which only required basic literacy to receive the phone call. However, interviews suggested low literate participant women could not follow voice messages and preferred consultation of health workers. On the other hand the transliterated text messages, which had the advantages of being stored and read at leisure period over voice messages, were welcomed by women who received higher education and were able to send or receive text messages on their own (Datta, Ranganathan, & Sivakumar, 2014;Khatun et al., 2016):

"I think for [low-literate] people in the village "uthan boithok" [group counseling organized at someone"s courtyard] is better where Shebika [health worker] discusses different topics and issues with a group of people. How many people [can] use the mobile phone tell me? So, it is better if they can gather and talk
to the Shebika about their problems." Participant woman 28.

Service Knowledge
Although some participants accessed the Aponjon medical consultation service for urgent or non-urgent symptoms and clarified issues with these doctors, many participants were not aware of such service. Yet some participants were unaware of the schedules of messages every week while some others did not know how to contact the service providers by dialing the short code for service related issues: "I wanted to enroll for my second baby. But Health worker Apa who enrolled me for my first child did not visit me for the second one. I didn't know how to enroll again." Participant woman 11.

Willingness to Pay
Although subscribers are generally informed about the pricing policy prior to enrollment, we found mixed opinion from the participants. Participants who frequently accessed the messages found the service affordable.
On contrary pricing policy of the service may have deterred some participants from accessing the messages:

Traditional Practices
Although Aponjon advises subscribers to organize safe delivery with skilled birth attendants and to follow essential newborn care guidelines, cultural influences may prevent women from adopting recommended practices. It was common for our participants from low income households to report using a mix of biomedical and traditional non-western practices during birth and the postnatal period. Similar to past studies that explored delivery practices among low-income households in Bangladesh, there were a few participants who had deliveries at home with untrained birth attendants; a tradition that had been passed down in families for generations (Choudhury & Ahmed, 2011). Other reasons for a home based delivery were cost of delivery at hospitals, fear of C-sections, convenience and lack of family members to accompany participants to hospital. Before organizing their deliveries at home most participant women confirmed with designated health workers that their pregnancy was normal. Participant women had little autonomy during home based deliveries; mostly deliveries were organized by their mothers or mothers-in-law. Husbands sometimes became involved when there were complications during delivery and a pregnant woman had to be removed to a health facility. Mothers or mothers-in-law usually handpicked the Dai (traditional birth attendant) depending on her years of experience and success at delivering live births in the community rather than her professional training. Although participants could not always confirm whether the Dai who assisted her during delivery had received any formal training, felt confident in their capabilities.
Essential neonatal practices such as wiping, drying and wrapping the newborn baby, proper cord care, delaying of first bath of the baby for at least 24 hours (delay is recommended up to 72 hours in Bangladesh) and initiation of breastfeeding right after birth (usually within an hour) requires assistance from people involved in the delivery process (National Institute of Population Research and Training, 2016). These practices were overruled because traditional remedies and adherence to cultural rituals were trusted due to the influence of elderly female family members; a finding consistent with previous studies on delivery and newborn care practices in Bangladesh (Darmstadt, Syed, Patel, & Kabir, 2006): Although antenatal care visits helped participants understand their pregnancy progression, participants who had deliveries at home with untrained Dais, were less likely to receive postnatal care visits, particularly if they had relocated to their parents' house before delivery and had lost contact with their health worker. Postnatal visits were deemed necessary only when there was a perceived need for maternal or neonatal illnesses. It was common among participants from low income households to adopt traditional sources of care, homeopathy or remedy from the grammo daktar (village doctor) for newborn babies, considering no or minimal side effects.

Biomedical Services
In general, there is an increasing normalization of westernized medical care practices in pregnancy, childbirth and postnatal period, irrespective of mobile phone based educational messages. Our participants had access to formal healthcare providers such as community health workers, paramedics or medically trained doctors who provided advice and treatment. Often, previous negative birthing experiences such as a stillbirth, delivery related complications or delay in reaching the hospital encouraged participants to access health facilities in the hope of a better outcome next time. After a facility based delivery, women, who otherwise would follow traditional practices at home guided by an elderly family member, adopted recommended postpartum health practices at doctor's advice:

"Nowadays the doctors advise [after delivery at hospital] not to put anything [mustard oil] on the umbilical stump and gives a medicine [instead] if it does not dry up." Participant woman 13 "Even if it is a normal delivery, the baby won't be given a bath right after bath….. Whatever you want to do now, you have to first consult a doctor." Participant woman 25
Whether the births took place at home or at a health facility, participants confirmed that they were aware of free vaccines at the monthly local health camps due to the house visits from vaccine providers and phone-call reminders from the national immunization program, and had never missed an immunization schedule for their children.
However, cost determined the type of facility women visited for pre-and postnatal cares. While subsidized or free services at public or NGO-run maternal facilities remained a feasible option for participant women from low income households, wealthier or more highly educated women were able to visit gynecologists often and organize delivery at private hospitals.

Discussion
Our research has implications for women's adoption of mHealth services and the position of mHealth within the existing health system and broader socio-cultural context. Participants in our study recognized the relative advantage of signing up to the service and were part of a social network that inspired other potential pregnant women to sign up as well (Rogers, 1983). We found that women responded positively to, and were likely to adopt, some of the health information provided by Aponjon mobile phone messages, particularly those related to newborn nutrition and care of newborn babies that could be performed at a household level. We infer that mHealth messages coupled with nutrition supplements have the potential to improve nutritional deficiency among infants in Bangladesh (GAIN, 2014; National Institute of Population Research and Training, 2016) and we recommend a proper observational study design to understand the effects.
Our findings suggest that traditional birthing practices and post-natal care, such as home-based deliveries by untrained Dais, are increasingly being used in conjunction with, or are being replaced by biomedical services and recommendations (National Institute of Population Research and Training, 2017). While Aponjon's messages have supported this transition to a western health model, many of these practices are taking place independently. There has been continued investment in health infrastructure, medical equipment, capacity building in formal healthcare, improved health systems governance and monitoring, demand-side-financing schemes and an extensive door-to-door health worker campaign strategy by the Government and NGOs (El Arifeen et al., 2013;Ministry of Health and Family Welfare, 2015;MOHFW, 2018). Research elsewhere suggests that mHealth interventions are less likely to show significant improvements in outcome indicators in high-performing health areas (Hategeka, Ruton, & Law, 2019) and that targeted messages may increase awareness and readiness around delivery and special care for premature babies among pregnant women who have an increased risk of preterm births (Olivia Kim et al., 2019).
Therefore, we suggest mHealth interventions will work best with targeted groups who need help most, particularly adolescent women in context of Bangladesh. Teenage pregnancy remains a problem in Bangladesh as one-third of 15-19 year old girls are likely to experience pregnancy before reaching adulthood and are at risk of contributing to a pregnancy related mortality ratio that has doubled between 2010 and 2016 (Islam, Islam, Hasan, & Hossain, 2017;National Institute of Population Research and Training, 2017). Culturally acceptable mHealth messages on contraception and reproductive health that worked in other resource limited settings could reduce teenage pregnancies in Bangladesh (Vahdat, L'Engle, Plourde, Magaria, & Olawo, 2013).
Like other studies (Brinkel, Dako-Gyeke, Kramer, May, & Fobil, 2017;Khatun et al., 2017;Khatun et al., 2016), we found that mobile phone ownership, literacy, technological functionality and involvement of significant others has determined service adoption practices of participants in our study. The operational challenges of providing Bangla text messages encountered by Aponjon is likely to be overcome in coming years by the increasing availability of smart phones that support Bangla fonts and new legislation from Bangladesh Telecommunication Regulatory Commission (BTRC) asking telecom operators to send messages to their clients in Bangla fonts (Tribune, 2016(Tribune, , 2018. Nevertheless, subscribers need training to access SMS. Along with technological support, husband's endorsement of the service is crucial for availing the phone and support for entire period of child birth. A study in India found that there were improved pre-and postnatal visits and adherence to newborn babies' delayed bathing among women whose husbands passed on information from the voice messages (Hazra, Khan, & Mondal, 2018). We suggest that husbands should be encouraged to access the service during enrollment, to share information with their wives and to consult the hotline service for delivery emergencies and neonatal complications, especially in low-income households where women are not confident with using technology (Alam et al., 2019). Also, women should be encouraged to listen to the messages along with elderly female family members who coordinate delivery and neonatal care practices in low income households. Making the service accessible to the household acknowledges that birthing and raising children is a family rather than an individual practice.
Our findings suggest that subscription charges may discourage subscribers who are reluctant to pay for educational messages, although our research on Aponjon counseling line suggests families were willing to pay to consult doctors at the call centers (Alam et al., 2019) meaning that subscribers preferred personal interactions and real-time personal advice which is missing in one-way general messages (Rajan et al., 2013). While cash incentives (Muller et al., 2019) and free-talk times (Lund et al., 2012) for accessing health facilities and healthcare providers may improve delivery and new born indicators in low-income households, more should be done. A referral service linked with local health facilities and organized house-visits by local healthcare providers would improve health outcomes and retention of subscribers (Huq et al., 2014;Jo et al., 2014;Watson et al., 2015).
We identified a gap in the community health workers' recruitment information as most participants were unaware of the consultation service with doctors or how to contact the call center for service upgrades or re-enrollment. Some participants had provided mobile phone numbers of family members to recruiters that they did not access. Since community health workers remain a major medium for reaching and recruiting women in the remotest households to mHealth services, it is important that health workers understand how mobile phones are used by households, and that they inform subscribers clearly about the purpose of the messages, billing method of the service and how to access the messages, the consultation service and the service call center. This requires on-going training and monitoring (Rajan et al., 2013).
Our study is limited by a number of factors. First, we cannot exclude the possibility of recall bias among participants, as they were re-contacted almost two years after they were surveyed in 2014. Second, it was difficult to ascertain how frequently women accessed the messages especially on shared phones; accessing service data on participants' usage could add value to their self-reported experiences. Third, our study lacked experience of SMS subscribers in low-income households as they preferred voice messages over transliterated Bangla text messages.
Offering women the option of Bangla SMS may encourage low-income households to enroll in the text message service thus improving their experience with "missed messages".

Conclusion
This exploratory study which documents the experiences of early adopters of a mobile phone based education service around delivery and newborn care practices in Bangladesh can provide valuable lessons for mHealth implementers and policy makers in similar cultural settings. The service could be expanded to include messages on women's health and health tips for well-being of under-5 and older children. For example, there is a growing unaddressed problem of obesity among urban women and the double burden of malnutrition among children in Bangladesh (National Institute of Population Research and Training, 2016). mHealth services, if they are well delivered, monitored and thoroughly evaluated, may contribute to improving a range of health issues in resource poor settings.