It has been widely documented that routine antenatal care (ANC) visits is one of the most effective methods in reducing maternal and child mortality and morbidity. Despite calls by the World Health Organization (WHO) for a minimum of 4 ANC visits, two-thirds of women living in Africa are able to attend ANC clinic only once during their gestational period [1,2]. It is estimated that 287,000 maternal deaths occur globally each year due to preventable complications during pregnancy and childbirth, 95% of which occur in sub-Saharan Africa and Southern Asia [3,4]. Furthermore, approximately 240 women per 100,000 live births die in low-income countries compared with only 16 women per 100,000 live births in high-income countries. African women experience 24 times the risk of intrapartum stillbirth compared with women in high-income settings [5-7]. Globally, there are around 3 million neonatal deaths and 2.65 million stillbirths each year [7]. In Kenya, the annual maternal mortality rate is 488 per 100,000 live births, of which only 47% of women are able to attend all 4 recommended ANC visits and the majority of them are from urban areas [8-10]. A further decline is seen in the remote rural areas of Kenya such as the northern and arid lands (NAL) whereby only 40% of pregnant women attend 4 or more ANC visits [8].

Routine immunization among children, on the contrary, is one of the most effective public health interventions of the 20th century. It has dramatically reduced global child morbidity and mortality [11]. Despite this, an estimated 19.3 million children did not receive vaccination for diphtheria, pertussis, and tetanus (DPT3) in 2010, one-third of which live in Africa, resulting in approximately 1.5 million preventable deaths every year [12]. In Kenya, approximately 52 infants per 100, 000 die every year, with only 77% receiving routine immunizations [8-10]. New and innovative strategies are required to improve ANC attendance among pregnant women and to increase routine immunization coverage among children.

Mobile health (mHealth) is defined as mobile communication technologies to support health care [13,14]. Currently, there are approximately 7 billion mobile phone subscribers globally, 89% of which are in low- and middle-income countries (LMICs) [15,16]. In Kenya, for example, mobile phone penetration was 77% in 2012, with SMS text messaging (short messaging service, SMS)-based communication on the rise [17]. The recent surge in mobile phone penetration alongside reduced cost has improved the feasibility of mHealth programs in LMICs [18]. Available data outlines mHealth’s potential and ability to connect health care services to pregnant women and mothers by passing different hurdles that would otherwise keep various health services inaccessible due to cost, distance, and accessibility [18,19]. SMS text messaging–based mHealth interventions can play a significant role in ensuring that the expectant mothers are followed throughout the continuum of care till completion of routine immunization coverage of their child [19].

Existing literature on mobile phone use in Kenya focuses on mobile phone penetration, trends in uptake of mobile communication services, corporate market share, mobile consumer-based services available, ANC visits improvement, SMS text message reminders and conditional cash transfer to improve routine immunization coverage, potential for mHealth interventions based on phone ownership and penetration, and engagement of care [20-26]. Few studies, however, examine the demographic characteristics of mobile phone users, their usage patterns, ownership, and mobile phones’ potential for improving engagement with health care providers and access to health services, specifically in a remote and rural population.

There is a vast difference in health indicators for NAL compared with rest of the Kenya, particularly for health-seeking behavior, maternal and child mortality, and immunization coverage. In this study, we aimed to assess the feasibility of implementing an SMS text messaging–based mHealth intervention intended to connect expectant mothers and child caregivers with their health care providers in rural and remote regions of Kenya. Our objectives were to examine region-based differences in mobile phone access and mHealth perceptions among visitors to antenatal and routine immunizations clinics in 8 regions of northern Kenya.

Between October 13, 2014, and February 20, 2015, participants were recruited from antenatal and routine immunization clinics at 8 health facilities in northern Kenya as part of a Grand Challenges Canada phase 2 “transition to scale” and Amref Health Africa as part of government of Kenya–led AphiaPlus/Marisha program for the health strengthening in the region [26]. Six participating government health facilities, 2 each in Isiolo, Marsabit, and Samburu counties were located in Kenya’s remote NAL, and 2 clinics each were located in Laikipia and Meru counties, which are less remote and more densely surrounding northern region (central highlands). The study received clearance from Amref Health Africa’s Ethics and Scientific Review Committee on September 1, 2014. Figure 1 is a graphic depicting distribution of all 8 sites.

The study questionnaires included sections on patient demographics, mobile phone usage patterns, the feasibility of an SMS text messaging–based mHealth intervention, as well as accessibility to primary health centers. Because the proposed mHealth intervention (WelTel) used weekly SMS text messages and allowed for shared phone access, questions related to this were included [27]. Inclusion criteria comprised pregnant women visiting the antenatal clinic, caregivers attending the routine immunizations clinics for children, and the ability to provide informed consent. The formal sample size calculation was not done; however, a target of 280 completed surveys was set. Consecutive convenience sampling was used to recruit patients whereby health care workers at the point of triage referred patients to study staff as patients arrived for their scheduled ANC and routine immunization appointments.

Structured questionnaires were administered in either English or Kiswahili depending on a patient’s preference and proficiency with English by trained study staff having command over both languages. All participants included in the study were asked to sign the informed consent, and thumb impression was taken from those who were illiterate or could not sign. Questionnaires were administered in a private place to ensure patient confidentiality and reduce the likelihood of response bias. No incentive was provided for participating in this study.

Prior to the study, questionnaires were developed and underwent forward and backward translation to ensure semantic consistency, and were piloted at one of the field sites before being rolled out. The research staff received 2-day training on research ethics, providing informed consent, administering the study questionnaire, and the study protocols to be followed throughout the course of the study. Continuous monitoring by study coordinators occurred throughout the study to ensure that data were collected as per the study protocol. All data was double entered into an Access program and analyzed using IBM SPSS Statistics version 23, with a level of significance in reference to a 2-tailed, type 1 error set as <0.05. Univariate analysis was performed using the chi-square test or Fisher exact test for dichotomous variables and the Mann-Whitney test for continuous variables.

Distribution of the respondents among study sites and population demographics at the study sites are shown in Table 1. Overall, 284 participants completed the survey from the 8 health facilities. Specifically, 63.4% (180/284) from the NAL clinics (Samburu, Isiolo, and Marsabit counties), and 36.6% (104/284) patients participated from central highlands (Meru and Laikipia counties). Overall, 98.6% (280/284) of the participants were females, and the median age of participants was 26 years (interquartile range, IQR 23-30 years). Forty-four percent (125/284) of participants were pregnant women visiting the ANC clinic, whereas 56.0% (159/284) were caregivers of infants attending the immunization clinic for routine immunizations.

When compared with participants from central highlands, a higher proportion of participants from the NAL were more likely to be married (91.1% vs 80.8%; P=.01) and self-identify as nomadic (24.4% vs 6.7%; P<.001). There were large differences in the proportion of participants who had received formal education: 48.9 % (88/180) of respondents from NAL reported having no formal education compared with 1.9% (2/104) from central highlands (P<.01). Similarly, a much higher proportion of participants from the central highlands clinics could understand (98.0% vs 52.2%; P<.001) and read (91.3% vs 46.7%; P<.001) common Swahili words (“mambo,” “sawa,” and “shida”) when compared with participants from NAL. Fifty percent (142/284) of the participants were employed of which the most common occupation was cited as a business.

Overall, 74.3% of participants (211/284) owned a mobile phone across all sites. However, a significantly higher proportion of participants from central highlands owned their phones when compared with participants from NAL (94.2% vs 62.8%; P<.001). Out of those who did not own a mobile phone 53.4% (39/73) did have shared access to a mobile phone with (83.3% vs 51.5%; P<.001) citing shared access in central highlands as compared to NAL. This indicated that the majority of participants from both the central highlands and NAL had some access to a mobile phone (99.0% vs 82.1%, P<.001).

A higher proportion of those from NAL were more likely to experience network challenges (27.1% vs 14.6%; P<.001). However, 92.1% (35/38) in NAL and 93.3% (14/15) at adjacent site participants were able to overcome the network coverage issues and only 9.2% (26/284) reported of a problem with keeping their phone battery charged overall. Further participants from NAL were less likely to be able to send an SMS text message (55.0% vs 95.1%; P<.001) when compared with participants from central highlands. Of those who were not able to send an SMS text message, 93.7% (59/63) in NAL and 80.0% (4/5) in adjacent sites mentioned that a trusted individual could respond on their behalf. Tables 2 and 3 show phone ownership and usage characteristics and SMS text messaging and mobile phone related outcomes.

Despite regional differences, there was no significant difference in the proportion of participants from NAL and central highlands, who indicated that they would like to receive a weekly SMS text message from their health care provider (95.0% vs 97.0%; P=.52). Overall, more than 91.6% (230/251) of participants who had access to a mobile phone indicated that they would like to receive a weekly SMS text message from their health care provider. Of these, 51.2% (124/242) preferred SMS text message as their mode of communication, whereas 48.8% (118/242) favored a phone call. Regional variances existed. Participants (78.2% vs 31.9%; P<.001) in central highlands preferred SMS text message compared with NAL; however, participants (68.1% vs 21.8%; P<.001) in NAL preferred phone calls as opposed to central highlands. Further, 62.5% (157/251) of responders preferred receiving SMS text messages in Swahili. Moreover, 49.4% (124/251) cited no time-of-day preference for receiving SMS text messages, and 40.6% (102/251) responded in favor of receiving 1 SMS text message per week. The average amount of money spent on a mobile phone by patients in the survey every month was 664 KSH (US $6.57), equivalent to about 170 minutes of talk time or more than 626 SMS text messages. The mean distance of the participants’ home from health care center was 53 minutes (n=268/284).