Impact evaluation of a digital health platform empowering Kenyan women across the pregnancy-postpartum care continuum: A cluster randomized controlled trial

Abstract

Background: Accelerating improvements in maternal and newborn health (MNH) care is a major public health priority in Kenya. While use of formal health care has increased, many pregnant and postpartum women do not receive the recommended number of maternal care visits. Even when they do, visits are often short with many providers not offering important elements of evaluation and counseling, leaving gaps in women's knowledge and preparedness. Digital health tools have been proposed as a complement to care that is provided by maternity care facilities, but there is limited evidence of the impact of digital health tools at scale on women's knowledge, preparedness, and the content of care they receive. We evaluated a digital health platform (PROMPTS (Promoting Mothers in Pregnancy and Postpartum Through SMS)) composed of informational messages, appointment reminders, and a two-way clinical helpdesk, which had enrolled over 750,000 women across Kenya at the time of our study, on 6 domains across the pregnancy-postpartum care continuum.

Methods and findings: We conducted an unmasked, 1:1 parallel arm cluster randomized controlled trial in 40 health facilities (clusters) across 8 counties in Kenya. A total of 6,139 pregnant individuals were consented at baseline and followed through pregnancy and postpartum. Individuals recruited from treatment facilities were invited to enroll in the PROMPTS platform, with roughly 85% (1,453/1,700) reporting take-up. Our outcomes were derived from phone surveys conducted with participants at 36 to 42 weeks of gestation and 7 to 8 weeks post-childbirth. Among eligible participants, 3,399/3,678 women completed antenatal follow-up and 5,509/6,128 women completed postpartum follow-up, with response rates of 92% and 90%, respectively. Outcomes were organized into 6 domains: knowledge, birth preparedness, routine care seeking, danger sign care seeking, newborn care, and postpartum care content. We generated standardized summary indices to account for multiple hypothesis testing but also analyzed individual index components. Intention-to-treat analyses were conducted for all outcomes at the individual level, with standard errors clustered by facility. Participants recruited from treatment facilities had a 0.08 standard deviation (SD) (95% CI [0.03, 0.12]; p = 0.002) higher knowledge index, a 0.08 SD (95% CI [0.02, 0.13]; p = 0.018) higher birth preparedness index, a 0.07 SD (95% CI [0.03, 0.11]; p = 0.003) higher routine care seeking index, a 0.09 SD (95% CI [0.07, 0.12]; p < 0.001) higher newborn care index, and a 0.06 SD (95% CI [0.01, 0.12]; p = 0.043) higher postpartum care content index than those recruited from control facilities. No significant effect on the danger sign care seeking index was found (95% CI [-0.01, 0.08]; p = 0.096). A limitation of our study was that outcomes were self-reported, and the study was not powered to detect effects on health outcomes.

Conclusions: Digital health tools indicate promise in addressing shortcomings in pregnant and postpartum women's health care, amidst systems that do not reliably deliver a minimally adequate standard of care. Through providing women with critical information and empowering them to seek recommended care, such tools can improve individuals' preparation for safe childbirth and receipt of more comprehensive postpartum care. Future work is needed to ascertain the impact of at-scale digital platforms like PROMPTS on health outcomes.

Trial registration: ClinicalTrials.gov ID: NCT05110521; AEA RCT Registry ID: R-0008449.

Fig 1. Retention at antenatal and postpartum follow-up of the cohort of pregnant women recruited at baseline.

This figure synthesizes the baseline recruitment of pregnant women who visited health facilities for antenatal care and tracks their retention at antenatal and postpartum follow-up, which occurred in the final weeks of pregnancy (i.e., around 36–42 weeks of gestation) and 7 to 8 weeks post-childbirth, respectively. Women recruited at or beyond 36 weeks of gestation at baseline were not approached for antenatal follow-up, while participants whose pregnancy had ended with birth of a living newborn were approached but ineligible. Meanwhile, for postpartum follow-up, all eligible and consented participants at baseline were approached and only a handful were ineligible due to being fewer than 7 weeks postpartum.

Fig 2. Map of study facilities.

This figure displays the location of the 40 health facilities in the study. The study facilities were distributed across 8 counties (Kajiado [4 facilities], Kilifi [6], Kirinyaga [2], Kisii [8], Meru [8], Narok [4], Nyeri [4], and Siaya [4]). Facilities are color-coded by treatment (blue) vs. control (gray) status. For 33 of the 40 study facilities, the closest facility was 10+ kilometers away. For the remaining 7 facilities, the nearest facility belonged to the same study arm. The distance between each facility and the nearest facility of the opposite arm was, on average, 32 kilometers. Map data courtesy of https://open.africa/dataset/kenya-counties-shapefile/resource/0b78f25e-494e-4258-96b8-a3ab2b35b121; available under a Creative Commons Attribution license.

Fig 3. Conceptual framework of PROMPTS’s potential impacts on the pregnancy-postpartum care continuum.

This figure offers a conceptual framework of the potential impacts PROMPTS could have across the pregnancy-postpartum care continuum. During both the antenatal and postpartum periods, improvements in knowledge (e.g., through informational messages) could lead to improvements in routine care seeking and danger sign recognition/care seeking. During the antenatal period, messages containing preparatory nudges may enhance preparation for childbirth, while during the postpartum period, messages containing care recommendations and content of care reminders may enhance newborn care and postpartum care content for infants and mothers. PROMPTS is also comprised of automated appointment reminders and a two-way clinical helpdesk, which may influence routine and danger sign care seeking.

Fig 4. Antenatal and postnatal care volume, by intervention arm.

This figure displays the increase in volume of ANC visits and PNC visits among participants recruited from treatment facilities just at or above the national-guideline-based thresholds of ≥4 ANC visits throughout pregnancy and ≥2 PNC visits during the first 6 weeks postpartum, following the first postpartum week. ANC, antenatal care; PNC, postnatal care.

Fig 5. Impacts on the content of postpartum care received by mothers and their newborns.

This figure depicts adjusted and unadjusted treatment effects for the 6 component measures (all denoting the share of participants for whom the respective outcome was present) incorporated in the postpartum care content index. The figure reveals that overall improvements in the domain were driven primarily by improvements in care that mothers received (e.g., more frequent discussions about their own health and discussions about family planning).