A Short Message Service (SMS) increases postpartum care-seeking behavior and uptake of family planning of mothers in peri-urban public facilities in Kenya

Background It is estimated that one third of maternal deaths in Kenya in 2014 could have been prevented by more timely care-seeking. Mobile health interventions are increasingly being recognized as tools for the delivery of health education and promotion. Many maternal deaths occur in the first few weeks after delivery and mothers who are given adequate care in the postpartum period have better health outcomes. Kiambu County, Kenya has a high level of literacy and phone ownership amongst mothers delivering in public hospitals and was chosen as a site for a postpartum short message service intervention. Methods Women were recruited after delivery and randomized to receive a package of mobile messages or standard of care only. Messages covered danger signs, general postpartum topics, and family planning. Endline phone surveys were conducted at 8 weeks postpartum to assess knowledge, care seeking behavior and family planning uptake. Analysis was conducted using Stata and is presented in odds ratios. Results Women who received the danger sign messages were 1.6 times more likely to be able to list at least 1 danger sign and 3.51 times more likely to seek treatment if they experienced postpartum danger signs. There was no significant difference in routine postpartum care seeking or care seeking behaviors concerning newborns. Women who received family planning messages were 1.85 times more likely to uptake family planning services compared to controls and 2.1 times more likely to choose a long-acting method. Conclusions Simple, low-cost mobile interventions can support women in the early postpartum period when the information is targeted to particular points in the postpartum continuum. Additional research is needed to understand the interplay between healthcare providers and mobile health interventions. Health policy makers should consider direct mobile interventions for women as an option for supporting positive maternal health outcomes in certain populations.

Study participants were recruited from postnatal wards in three public health facilities in Kiambu County, Kenya, which is located 13–31 kilometers north of Nairobi City County. Facilities were selected on the basis of providing maternal and infant health services to the surrounding neighborhoods; were considered ‘mid-volume’ as they averaged at least 300 deliveries per month based on the Kenya Health Information System; and served populations of both semi-rural and peri-urban centres. The final facilities selected for the study were Kihara Sub-District hospital, Ruiru Sub-District hospital and Tigoni District hospital. Women between 18 and 40 years old were eligible for inclusion if they had delivered vaginally at one of the three study sites and had access to a mobile phone with which they could receive SMS messages. Women were not eligible for inclusion if they had complications in their delivery (including undergoing a cesarean section) or if their infants showed signs of illness/complications prior to discharge, such as jaundice, abnormal breathing, or heart conditions. The list of complications that excluded women from participation was compiled by an obstetrician and defined as complications that could require specialized care. Research assistants approached all eligible women being discharged (~24 hours after delivery) from the facility and gave them information describing the study. Potential participants were taken through a voluntary, written informed consent process and signed an informed consent form if they chose to enroll in the study. Women could withdraw from the study at any time and stop the messages by sending the word “STOP” to the shortcode from which they received messages. A randomized controlled trial was conducted with four study arms. Following enrollment, participants were randomized at the individual level into one of four arms and uploaded into the SMS system by a research assistant not involved in recruitment or enrollment. Randomization was done by utilizing a random number generator in Stata (2009 StataCorp, College Station, TX) to create a list of numbers 1 to 1000 in a random order. The numbers 1–250 were assigned to the 1st study arm, 251–500 assigned to the 2nd study arm, 501–750 assigned to the 3rd arm, and 751–1000 assigned to the 4th Arm. The random number list (and associated study arm number) was copied into the enrollment sheet. As baseline surveys were completed, the phone numbers were copied in the exact order of completion into the enrollment sheet and assigned the corresponding study arm based on the random number in the column next to it. Research assistants involved in recruitment and enrollment were not aware of the number allocation. The phone numbers were then uploaded into the SMS system by group. Arm 1 was the control group (Control), where participants were enrolled in the study but only received the current standard of care (no SMS). Each of the three intervention arms had a different combination of SMS content. Arm 2 received the “postpartum checklist” (PPC) the week following discharge, which consisted of Yes/No questions regarding their postpartum state. Arm 3 received the “postpartum checklist” plus general postnatal care messages and reminders in the 4 weeks after discharge (PPC+PNC). Arm 4 received the “postpartum checklist” plus family planning messages/reminders between 4 and 6 weeks after discharge (PPC+FP). At the time of enrolment, all women were informed about the randomization process, and that there was a possibility of placement in the control group. A visual of the message timing for the different intervention arms can be viewed in Fig 1. All SMS were sent in Swahili and were completely free for the participants to receive and to respond to, with the intention of lowering accessibility barriers for low-income women. All messages were developed using international guidelines and academic publications with input from clinicians at Jacaranda Health. The components of the checklist and their sources are described in McConnell et al, 2016 [22]. New mothers received four messages per day, on days 2, 3 and 4 post-discharge. Each message was in the form of a yes/no question and was designed to prompt mothers to self-assess for maternal infection, secondary postnatal hemorrhage, postnatal preeclampsia, insufficient breastfeeding, jaundice in newborns, and local infection. At the end of each set of messages the mother is encouraged to indicate if she has observed the described symptoms in herself or in her baby. If the mother indicated that she observed the symptoms described in the messages, she received a referral message encouraging her to visit the hospital. The PNC message stream included 11 health messages and were sent every 3 days, starting at day 6 up to day 36 after discharge. The messages included information related to breastfeeding, baby care, and family planning. The messages also encouraged women to adhere to routine postnatal follow-up care and the baby’s immunization schedule. The FP message stream included 6 family planning messages, began 30 days after discharge, and continued every three days through day 45 after discharge. The messages included information related to healthy birth spacing of 2 years, modern methods of contraception, and a reminder that one can get pregnant again soon after birth before having periods. All message packages were tested through a series of focus groups in Kiambu County, each with 8–12 postpartum women who delivered at public hospitals. The focus groups first probed for the types of information women felt they needed after delivery and combined this with information that was prioritized by clinicians. The subsequent focus groups then tested the messages for language clarity and understanding, preferences around timing/frequency of messages, and preferences around language choice. We aimed to recruit 900 women in order to have 80% power in detecting a 10% change in postnatal care attendance among mothers, while also accounting for an 18% loss to follow up. Enrollment took place over 5 months, beginning in November 2017 and concluded in March 2018. Research assistants approached all eligible women on the day they would be discharged from the facility and gave them information describing the study. If a woman consented to participate, she completed the in-person baseline survey administered by a research assistant at the hospital before discharge and was enrolled in the study. Endline survey data was collected 8 weeks after discharge through a phone survey. Endline data collection concluded in May 2018. All hospitals were enrolling simultaneously during the enrollment period. Survey data were recorded on a tablet using SurveyCTO software (2016 Dobility, Inc, Cambridge, MA). Calls were attempted 4 times before the participant was deemed unreachable. Data was analyzed in STATA 11 (2009 StataCorp, College Station, TX). Outcomes related to danger sign knowledge and care seeking related to danger signs compared all participants who received the postpartum checklist (PPC group, PPC+PNC group, PPC+FP group) together versus the control group who did not receive any messages. Outcomes related to general postnatal care (immunization, mother’s postpartum checkup, and child wellness visits) compared women who received the PPC+PNC messages versus all other participants grouped together (PPC, PPC+FP, Control). Outcomes related to family planning (uptake of family planning, planning to uptake family planning, LARC method use) compared women who received the PPC+FP messages versus all other participants grouped together (PPC, PPC+PNC, and Control). All demographic data was described using Pearson’s chi2 tests. Univariate logistic regression models were run, followed by mulitvariate logistic regression models controlling for potential confounders. Outcomes are reported as Odds Ratios (OR) with their 95% confidence intervals (95% CI). Ethical clearance for the study was obtained by the Amref Ethics and Scientific Review Committee in Kenya under protocol number P355-2017. Interested participants provided written informed consent and were alerted to both the potential benefits and risks, as well as the fact that they may be randomized to not receive messages. At all points during the study, participants could voluntarily decide to withdraw from the study or opt out from receiving the SMS messages by sending the word ‘STOP’.

N/A

• Dima Health