Evaluation of a Text Messaging Intervention to Promote Preconception Micronutrient Supplement Use: Feasibility Study Nested in the Healthy Life Trajectories Initiative Study in South Africa

Introduction

Background

Women’s health and nutritional status before and during pregnancy are increasingly being identified as important determinants of their future health, pregnancy success, and the health of the next generation [1]. However, across high-, middle-, and low-income countries, both over- and undernutrition remain prevalent before and during pregnancy, and nonadherence to nutritional recommendations is a global issue [1-7]. Micronutrient deficiencies persist in South Africa, with anemia affecting 23% to 31% of women of reproductive age in urban settings [8,9]. In this context, preconception health and its consequences have also been identified as important knowledge gaps among young women [10].

Using digital spaces for health promotion and health behavior change is an increasing global phenomenon. In South Africa, radio is the most consumed form of media, with an estimated 37.8 million weekly listeners, of whom approximately 30% listen via mobile phones [11]. Although radio messaging has shown promise for improving health knowledge [12], health behavior changes may be more effectively achieved when messaging is targeted to specific population subgroups and tailored to the needs of the individual [13].

SMS text messaging interventions are a potential avenue for providing targeted and tailored behavior change support, reinforced by the high level of mobile phone penetration in many low- to middle-income settings. For example, approximately 96% of households in South Africa in 2018 had a mobile phone, and of the country’s population, 82% were estimated to have a smartphone subscription (smartphone penetration) [14]. Increasing data support the use of various SMS text messaging interventions to improve appointment attendance, medication adherence, and risk-related behavior change, mostly in the context of chronic diseases [15-19]. However, systematic reviews from low- to middle-income countries have found mixed or inconclusive evidence of the impact of SMS text messaging interventions, which are often not explicitly designed using behavior change theory [20-22]. In South Africa, SMS text messaging interventions have shown highly promising results for HIV care adherence, support, and education [23-25], as well as for pregnant and postpartum women, through the National MomConnect SMS text message health messaging program [26,27]. Data from high-income settings suggest that a 2-way SMS text messaging intervention can support nutritional behavior change, resulting in weight loss and maintenance in the postpartum period [28], and that a mobile app can effectively improve preconception nutrition [29]. However, to the best of our knowledge, the use of SMS text messaging to improve nutritional status during the preconception period has not been previously explored.

Objectives

The main aim of this study was to evaluate the feasibility of a tailored 6-month SMS text messaging intervention to support adherence to preconception micronutrient supplementation in women enrolled in the Healthy Life Trajectories Initiative (HeLTI) Bukhali trial, a complex preconception health trial in Soweto, South Africa. This included an evaluation of the acceptability, usability, perceived benefit, fidelity, cost, and indicative effects of the SMS text messaging intervention for promoting supplement adherence. The second objective was to evaluate the acceptability of a radio serial and accompanying Facebook page aimed at supporting health promotion in the trial setting, which was aired toward the end of the SMS text messaging intervention.

Methods

Setting and Population

This study was nested in the intervention group of the South African site of the HeLTI Bukhali study, a randomized controlled trial that aimed to evaluate a multifaceted intervention for promoting the health of women of reproductive age (18-28 years) [30,31]. Similar trials are ongoing in Canada, India, and China in collaboration with the World Health Organization. HeLTI Bukhali is based in Soweto, a historically disadvantaged township bordering Johannesburg, with approximately 1.3 million residents, and recruitment started in 2018. The intervention was delivered by research staff trained in healthy conversation skills [32], who distributed educational resource material and provide health feedback in terms of BMI, hemoglobin, blood pressure, hemoglobin A_1c to assess hyperglycemia, HIV testing, and mental health and facilitated sessions to support improved health behaviors around nutrition, physical activity, sleep, health monitoring (eg, HIV testing), and other goals identified by the participants. An aspect of the HeLTI Bukhali intervention is the provision of multimicronutrient supplements based on participants’ anemia status. Women receive a supplement containing, among other micronutrients, 27 mg iron twice per week if they are nonanemic (capillary hemoglobin ≥12 g/dL) and daily if they are mildly anemic (hemoglobin <12 g/dL), with women who are severely anemic (hemoglobin <7 g/dL) receiving referrals to receive the current standard of care, comprising further assessment, supplementation, and additional management as required. The trial’s control group was contacted once a month through telephone, SMS text message, or email to deliver information on life skills not directly related to health and had access to standard health services such as HIV and pregnancy testing. The intervention and control arms of the trial were included in the trial program for a total of 18 months and were followed throughout pregnancy and postpartum periods in case of pregnancy within the 18-month time frame [31].

The exclusion criteria for HeLTI Bukhali were a diagnosis of type 1 diabetes, cancer, or epilepsy; the presence of an intellectual disability that hinders informed consent; and being unwilling or unable to consent. For this feasibility study, only women recruited into the preconception intervention group of the trial were included. This study evaluated 2 distinct remote media approaches (SMS text messaging and radio combined with Facebook) in the same trial.

As indicated in the timeline in Figure 1, the first 120 women enrolled in HeLTI Bukhali between November 1, 2020, and February 2021, who consented to the SMS text messaging substudy, received the SMS text messaging intervention. This SMS text message–receiving group was asked to participate in this study’s focus group discussions (FGDs), in which the acceptability of the health promotion radio serial was also evaluated.

‎View this figure

In addition, for the preliminary evaluation of the indicative effect of the SMS text messaging intervention, a comparison group receiving the Bukhali intervention but not receiving the SMS text messages (SMS text messaging control group from here on) comprised the first 120 women who were recruited to the HeLTI Bukhali intervention arm from November 2019 onward until the target sample size was attained without any matching the SMS text message–receiving group. The time frame of the SMS text messaging control group recruitment was chosen to avoid seasonal variability in micronutrient status. This also avoided the major hiatus in data collection in response to the COVID-19 crisis that started in March 2020. Follow-ups for participants in the SMS text message–receiving and SMS text messaging control groups were conducted between May and October 2021 and May and December 2020, respectively. The delays in follow-up in both groups were largely attributable to the COVID-19 pandemic.

The HeLTI Bukhali Text Messaging Intervention

The HeLTI Bukhali SMS text messaging intervention and message bank were developed based on existing evidence and the behavior change theory. Textbox 1 provides an overview of the SMS intervention according to the TIDieR (Template for Intervention Description and Replication) checklist. The Health Actions Process Approach (HAPA) [33] was the main guiding theoretical approach for developing the SMS text message library. The HAPA is a sociocognitive model encompassing 2 phases: the motivation phase, including outcome and risk perception and self-efficacy, and the volitional phase, including action-oriented processes such as action planning and maintenance of self-efficacy. HAPA has previously been shown to successfully inform medication adherence and UK-based SMS text messaging interventions [28,34,35]. The message bank was developed with input from young women in Soweto, including lessons from formative work for HeLTI Bukhali around barriers to supplement use [31], and was tested for any content or technical issues before this feasibility study.

As described in Textbox 1, participants received 2 messages every week for 24 weeks, with messages comprising health literacy, adherence, and side effect reporting. From the time participants were recruited to HeLTI Bukhali, SMS text messages were sent and received using an automated SMS text message delivery platform—Twilio Inc. To avoid nonparticipation in the 2-way SMS text messages because of a lack of funds, participants were automatically supplied with 5 South African Rand (ZAR 5, around US $0.30) prepaid minutes of phone credit (“airtime”) weekly, which sufficiently covered the cost of responding to an SMS text message through any available provider, using FlickSwitch SIMControl South Africa (Cape Town). The successful delivery of the SMS text messages and airtime could be monitored through these respective platforms and was evaluated through qualitative data collection.

The Phila Impilo Yakho Kangcono Radio Serial

The Phila Impilo Yakho Kangcono (translating to “Live Your Best Life”) radio serial was designed as media support for the HeLTI trial, and an overview based on the TIDieR checklist is provided in Textbox 2. The weekly radio serial targeted young people and aimed to promote well-being and health. It was aired in isiZulu and English between March and June 2021 on a Sunday youth show on Ukhozi FM, South Africa’s largest radio station, and on Jozi FM, a local radio station popular in Soweto, from July to September 2021. An associated Facebook page was created and referred to in the radio episodes [36]. The content of each episode was designed by an independent production team working with and guided by young adults (both men and women) from Soweto and the research team.

Ethics Approval

The human research ethics committee (Medical) at the University of Witwatersrand approved this study (M171137 and M1811111). Participants gave written informed consent before enrollment into the study and provided additional written informed consent before participating in the recorded FGDs.

Outcomes

This mixed methods study adopted a parallel-convergent approach, using a combination of quantitative and qualitative data sources to address feasibility outcomes [37]. The framework for evaluating young women’s perceptions of the HeLTI Bukhali SMS text messaging intervention and the radio serial and associated Facebook pages are outlined in Table 1 and are based on the existing literature on process evaluation [38,39]. The main outcomes of interest for evaluating women’s perceptions of the SMS text messaging intervention were acceptability, usability and level of interaction, perceived benefit, intervention cost and fidelity, and indicative effects. For the Phila Impilo Yakho Kangcono radio serial and associated Facebook page, the main outcome was the acceptability of the radio serial and the associated Facebook page.

Data Collection

HeLTI Bukhali Text Messaging Intervention

Qualitative data on the acceptability, usability, interaction with, and perceived benefit of the SMS text messaging intervention were collected during 5 FGDs, including 2 to 9 participants. These were conducted by trained research staff from Soweto using a topic guide developed by the research team and were organized and attended by LMS. Participants were invited at random from those receiving SMS text message notifications 4 to 8 months after the initiation of the study. Refusals to participate in the FGDs were because of having moved away from Soweto, having other commitments, or being ill on the day of the FGD. Refreshments were provided at the FGD, which took 1 to 2 hours. The FGDs were conducted in English and in the participants’ preferred language and were audio recorded (Philips Digital Voice Recorder DVT4110 and DVT1150). The recordings were transcribed verbatim but were anonymized and translated into English where necessary.

To evaluate the indicative effect of SMS text messaging intervention, data were captured and managed using REDCap (Research Electronic Data Capture; Vanderbilt University) [40]. Data were collected by trained research assistants at the research unit in Soweto. At baseline, a validated questionnaire was used to gather information on participant age, attitudes toward and confidence in supplement use, parity, country in which the participant was born, their home language, education level, employment status, and food security [41]. Participants were categorized as at risk of food insecurity if they answered yes to one of the 3 questions in the food security questionnaire (“Does your household ever run out of money to buy food?”; “Do you ever cut the size of meals or skip meals because there is not enough money to buy food?”; “Do you go to bed hungry because there is not enough money to buy food?”) and as food insecure if they answered yes to ≥2 of the questions, as previously described in our setting [42].

At a follow-up visit conducted approximately 6 months after initial recruitment for both the SMS text message–receiving and SMS text messaging control groups, a questionnaire was used to gather data on self-reported adherence to the micronutrient supplements (1=I never took them to 5=I took them every time), participants’ perceived understanding of the condition of anemia, perceived health improvement because of the supplement, fears and side effects related to the supplement, and main reasons for not taking the supplement. At both baseline and follow-up, hemoglobin levels were measured using a Hemocue 201 device. As Soweto is situated >1700 m above sea level, hemoglobin was altitude adjusted by –0.5 g/dL, according to the World Health Organization recommendation [43], as has previously been evaluated in our setting [44]. The difference in hemoglobin levels from baseline to follow-up (Δhemoglobin) was calculated (hemoglobin follow-up – hemoglobin baseline).

Data obtained from the log of the automated SMS text messaging system also provided information on participant interaction with the intervention (rate of response to 2-way SMS text messaging features) and intervention fidelity (number of messages successfully received, technical failures, and airtime received by participants). The study records provided information on the total intervention costs per participant.

Phila Impilo Yakho Kangcono Radio Serial

During the 5 FGDs described in detail previously, all participants played 1 to 2 episodes of the Phila Impilo Yakho Kangcono radio serial and were shown the associated Facebook page to ensure exposure among all participants. Subsequently, data were collected on the acceptability of the radio serial. These data were recorded and transcribed as described previously.

Data Analysis

Qualitative data analysis was informed by the thematic analysis methods described by Braun and Clarke [45]. The analytic process and presentation of results followed the phases set out in the framework method, which is a codebook approach to thematic analysis [46]. The coding of transcripts combined a deductive approach based on the study objectives and the adapted process evaluation framework [38,39] outlined in Table 1, with inductive analysis to allow for flexibility in incorporating unforeseen findings. LMS read the transcripts to familiarize herself with the data. The initial coding of the transcripts was completed by LMS using MAXQDA software (version 20.4.1; VERBI GmbH), and preliminary themes and framework matrices were reviewed by and discussed with LJW, SK, and MM to refine the analysis.

Quantitative data analysis was performed using STATA (version 13.0, StataCorp). Baseline characteristics and SMS text message response data were described as numbers and percentages for categorical variables and mean and SD or median and IQR for continuous variables. Outcomes at the 6-month visit, including change in hemoglobin from baseline (Δhemoglobin) and change in anemia status, were compared between the participants receiving SMS text messages and the SMS text messaging control group using the Student t test (2-tailed) or Wilcoxon rank-sum test, depending on the normality of the data. The baseline characteristics of participants lost to follow-up in the SMS text message–receiving and SMS text messaging control groups were described and compared with followed-up participants using statistical significance testing. Mixed linear modeling was used to explore the time point–adjusted association among altitude-adjusted hemoglobin, intervention exposure, time from baseline to follow-up, and the interaction between intervention exposure and time. A second, adjusted model was run adding variables that were statistically significantly different between the SMS text message–receiving and SMS text messaging control groups at baseline.

Results

The HeLTI Bukhali Text Messaging Intervention

Participant Baseline Characteristics

Figure 2 shows the participants of this study. For the SMS text messaging intervention, 72.5% (87/120) of participants in the SMS text message–receiving group and 81.7% (98/120) of participants in the SMS text messaging control group had their data collected at the follow-up visit. Of these, 90.8% (79/87) in the SMS text message–receiving group and 90.8% (89/98) in the SMS text messaging control group had hemoglobin data available from their follow-up visits. Reasons for loss to follow-up included withdrawal from HeLTI Bukhali (not specifically from the SMS text messaging substudy; 17/240, 7.1%; 13/120, 10.8%, in the SMS text message–receiving, and 4/120, 3.3%, in the SMS text messaging control group) and relocation or an inability to trace (38/240, 15.8%). The main baseline characteristics in the followed-up versus lost to follow-up group are provided in Table S1 in Multimedia Appendix 1, and the only statistically significant difference between these 2 groups was the number of participants born in South Africa (185/185, 100%, in the followed group vs 53/55, 96%, in the lost to follow-up group).

Table 2 shows the main baseline characteristics of the study participants. The median age was 21.5 (IQR 19-24) years, and 30.8% (74/240) of the participants had anemia (hemoglobin <11.9) at baseline. More SMS text message–receiving participants were unemployed and fewer had graduated from high school than the SMS text messaging control group participants. Moreover, the number of participants with mild or severe anemia at baseline was higher in the SMS text message–receiving group than in the SMS text messaging control group (50/120, 41.7% vs 24/120, 20%).

‎View this figure

Objective 1: Feasibility of the Bukhali Text Messaging Intervention

Acceptability

In terms of acceptability, participants liked receiving SMS text messages on their phones, which they expressed using frequently, and reported that the content was acceptable (Table 3).

The participants suggested changes that would make the SMS text messages more acceptable and useful. For example, participants suggested sending SMS text messages earlier in the morning and on days when they had to take micronutrient supplements (such as Mondays). Participants taking a daily dose of micronutrient supplements expressed the potential added value of receiving reminder messages every day. However, some participants described adherence messages as repetitive, resulting in them not always being read. Although some participants expressed that the health literacy messages were too long, others liked them and described reading them in full. Participants gave suggestions unique to their personal preferences and needs, such as receiving messages at night or preferring phone calls.

Interaction and Usability

On the basis of quantitative analysis of the data log of participants’ use of the 2-way SMS text messaging system, the participants’ interactions with the 2-way SMS text messaging intervention in terms of response rate were low, with only 59.2% (71/120) of participants replying to any 1 message (Figure 3). Use declined over the 6 months of the intervention, and 10.8% (13/120) of participants still receiving messages replied in week 24.

In the FGDs, participants described their main reasons for not using the reply function as being too busy, lack of interest, lack of airtime, not understanding whether they were required to reply, and a technical error (Table 4). Participants expressed that a fuller explanation of the reply function and the received weekly prepaid airtime was needed. Similarly, participants expressed that being too busy, a lack of interest, and technical issues were the main barriers to receiving and reading SMS text messages. In terms of technical issues, some participants received the message in 2 parts or only received half of the message, possibly because of the length of some of the health literacy messages.

‎View this figure

Perceived Benefits Versus Unintended Consequences

Participants described experiencing benefits from the intervention, including feeling supported by the study team, being reminded to take the micronutrient supplements, and learning new concepts from the health literacy messages (Table 5). Unintended consequences associated with the health literacy messages that were described included feelings of fear about the information in the side effect messages and feeling patronized (made to feel “stupid”) by the messages. This may be because the SMS text messages reinforce content from the HeLTI Bukhali resources, and for participants with limited health literacy, this may feel that it highlights their lack of knowledge and understanding.

In general, participants did not feel the need to respond to the messages, although positive feelings about communicating through the reply function were expressed by one participant. In addition, some participants expressed feeling guilty for not responding to the messages or feeling like they might have been judged if they replied that they had not taken their supplements.

Fidelity and Cost of Intervention Delivery

On the basis of the quantitative data log of received messages, airtime, and technical errors, there was only one participant to whom the SMS text messages were not delivered as intended because of a technical error, which was resolved at week 16 of the 24 weeks after updating the participant’s number and provider. As indicated in Table 4, in the FGDs, participants sometimes reported receiving messages in 2 parts or only receiving half a message, particularly when using nonsmartphone models. Some also reported having technical problems when trying to reply to the message. However, the exact cause or number of cases in which this occurred could not be determined.

In terms of receiving weekly prepaid airtime, a few participants experienced technical problems, with 4.2% (5/120) of participants experiencing repeated technical errors. Of these, 3 were because of the initial recording and use of the wrong cell network by the research team. One was because of a change in the phone number. For these 4 participants, errors were resolved within 2 months of onset. For the fifth participant with a repeated technical error, technical failure could not be resolved, likely because they had a nonchargeable SIM card.

On the basis of the cost log for the intervention, delivering the SMS text messaging intervention cost ZAR 71.65 (US $4.65) per participant or, on average, ZAR 245.64 (US $15.93) per week for the 35 weeks that SMS text messages were sent. When including the weekly prepaid airtime, the intervention cost per participant was ZAR 266.26 (US $17.27) or ZAR 912.91 (US $59.21) per week.

Indicative Effect of the Bukhali Text Messaging Intervention

This section reports the results of preliminary, nonrandomized comparisons between the SMS text message–receiving and SMS text messaging control groups. As shown in Figure 4, at follow-up, self-reported adherence was higher in the SMS text message–receiving group (42/63, 67% reported taking supplements every time) than in the SMS text messaging control group (33/85, 39%; P=.02). In addition, a larger percentage of the SMS text message–receiving group (32/63, 51%) strongly agreed that the supplements had a positive impact on their health than the SMS text messaging control group (36/85, 42%; P=.09; Figure 5). The main reasons for sometimes or often missing supplements in both groups were traveling (and therefore not having access to the supplement) and forgetting; however, fewer participants in the SMS text message–receiving group reported missing their supplements for these reasons. Moreover, a lower number of participants in the SMS text message–receiving group reported having fears about taking the supplements than in the SMS text messaging control group (4/63, 6% vs 16/77, 21%; P=.02), although the number of participants reporting a side effect in the SMS text message–receiving group was nonstatistically significantly higher than in the SMS text messaging control group (13/64, 20% vs 9/84, 11%,). In addition, statistically significantly more participants in the SMS text message–receiving group agreed or strongly agreed that the micronutrient supplements were explained to them well (59/64, 92% vs 66/87, 76% in the SMS text messaging control group; P=.03).

‎View this figure

‎View this figure

The mean difference in hemoglobin from baseline to follow-up (Δhemoglobin) was negative in the SMS text messaging control group (mean −0.46, SD 1.59), whereas in the SMS text message–receiving group, Δhemoglobin was positive (mean 0.52, SD 2.18; P=.002). This difference in hemoglobin levels between the 2 groups remained statistically significant when the 2 participants with severe anemia at baseline were excluded from the SMS text message–receiving group (P=.003).

Exploring this finding further with a mixed linear model, we found that the altitude-adjusted hemoglobin increased more between baseline and follow-up in the SMS text message–receiving group than in the SMS text messaging control group, as indicated by the interaction term in the model (1.03, 95% CI 0.49-1.57; P<.001; Table 6), although absolute the hemoglobin level was lower in the SMS text message–receiving group when adjusted for time (−0.70, 95% CI −1.13 to −0.28; P=.001). These results remained statistically significant and were not attenuated after correcting for differences present at baseline (employment status and level of education; Table S2, Multimedia Appendix 1).

The Phila Impilo Yakho Kangcono Radio Serial Intervention

Objective 2: Acceptability of the Radio Serial and Facebook Page

Participants expressed that they found the content of the messages educational and relatable and that they enjoyed the narrative aspect of the radio serial as a way of communicating health messages (Table 7). The participants reported not having previously heard the radio serial on the national radio station (Ukhozi FM) or on the local station (Jozi FM).

Advantages and Disadvantages of SMS Text Messaging, Radio, and Facebook Page

An overview of the advantages and disadvantages of the different health messaging media based on the FGD data is provided in Table 8. Although participants expressed concerns about the accessibility of data-requiring messaging platforms such as Facebook and had not visited the Facebook page associated with the radio serial before the FGD, participants liked that it provided a community with whom they could interact.

Discussion

Principal Findings and Implications

In this nested feasibility study, the HeLTI Bukhali SMS text messaging intervention, aimed at improving preconception adherence to micronutrient supplements, was found to be acceptable, beneficial, and usable by participants. Indicative effect results suggest a promising potential effect of the SMS text messaging intervention on self-reported supplement adherence and changes in hemoglobin from baseline to follow-up, although these are not based on a randomized comparison. These findings are in line with evidence from high-income countries on the acceptability and potential of SMS text messaging interventions to support behavior change in various life stages and contribute to the existing evidence, although less conclusive, for SMS text messaging interventions from low- and middle-income countries [20,21,27,28]. Finally, the health-promoting radio serial was found to be acceptable for young women.

Although participants reported finding the SMS text messages easy to understand and helpful, we received several suggestions for adapting the SMS text messaging intervention. This included a preference for receiving messages in the early morning on days they were meant to take their supplement. Although personalizing the timing and frequency of SMS text message delivery could negatively affect the scalability of the intervention [47], it may be feasible to send more frequent messages to those participants taking supplements daily, as suggested in the FGDs. As the SMS text messaging intervention was automated, it required few resources and was not demanding on implementation staff, increasing its practicality in limited-resource settings such as ours. However, introducing a nonautomated aspect to the intervention, for example, through a help desk, which participants could reach by replying to a number, could help manage the technical challenges experienced by participants and facilitate communication with the study team. Existing evidence also suggests that a nonautomated, personalized element may increase interaction with and the effect size of the intervention, although this remains to be tested from a cost-effectiveness perspective [48]. It is important to establish the feasibility and affordability of such a service in low- and middle-income settings, given that automated systems could be more scalable and sustainable in resource-constrained environments. Furthermore, in the context of the HeLTI Bukhali trial, there are likely other intervention priorities that would take precedence over such a service, particularly in terms of the allocation of human resources. However, there could be less resource-intensive ways of integrating these learnings into the intervention implementation, such as reminding those delivering the intervention to follow up more intentionally about the delivery of SMS text messages with participants.

In our study, the 2-way reply function of the SMS text messaging intervention was underused, with only 10.8% (13/120) of participants using the reply function in the final week of the intervention. A study evaluating a 2-way SMS text messaging intervention for weight loss in postpartum women in a high-income setting found good engagement with a comparable yes or no SMS text message reply option, with most participants replying to over half of the messages and high engagement contributing to the outcome of interest [28]. In a low- to middle-income setting, some evidence suggests that a 2-way SMS text messaging intervention could be effective if engagement is high enough [49]; however, a study of a 2-way SMS text messaging intervention for improved HIV care in pregnant and postpartum women in Kenya found a similarly low response frequency [50]. Qualitative research from Uganda indicated similar barriers to responding to a 2-way SMS text messaging intervention, as expressed by our participants, including a lack of airtime, being too busy, and phone-specific issues [51]. Whether these barriers can be minimized through more extensive instructions, explanations, participant encouragement at intervention initiation, or the use of alternative media such as unstructured supplementary service data would require further assessment before this aspect of the intervention can be recommended in our setting. Despite the low response rate, the SMS text messaging intervention was found to be acceptable and showed a potentially indicative effect. Therefore, a 2-way function may not be an essential component of an intervention aimed at improving supplement intake, as opposed to more complex behavior change targets.

In terms of population-level health promotion messaging, our participants indicated a preference for the narrative and accessible nature of the Phila Impilo Yakho Kangcono radio serial. Radio health messaging has been found to increase health knowledge and awareness in the context of, among others, maternal health, HIV prevention research, and child health in low- and middle-income settings, including Southern Africa [52-56]. However, we found that not all participants listened to the radio frequently, indicating the need for research on the reach of such a radio-based health promotion campaign among young women in our setting. By combining media, directed health messaging (such as SMS text messaging) could be used to increase young women’s awareness of new radio and social media campaigns to improve their reach and effectiveness [57].

Strengths and Limitations

To the best of our knowledge, this is the first study in our setting to evaluate social messaging through individual-level SMS text messaging intervention and population-level radio serial for preconception health promotion. However, a limitation of this study is that more in-depth data on the acceptability, feasibility, and impact of the radio serial were not available as participants’ exposure was limited to 1 or 2 episodes, which they heard for the first time during the FGDs. Another limitation is the high rate of loss to follow-up and the presence of missing data, which threatens the validity of the findings regarding the indicative effect of the SMS text messaging intervention. Moreover, the loss to follow-up may indicate that collecting data on these end points might be challenging in the context of a full randomized controlled trial. Although a portion of the loss to follow-up can likely be explained by the unique circumstances of the COVID-19 pandemic, additional methods to limit withdrawal, increase participant motivation for participation, and prevent loss of contact (eg, through community engagement) could be useful to explore. The withdrawal and loss-to-follow-up rates were higher in the SMS text message–receiving group than in the SMS text messaging control group, which could be because of the greater impact of COVID-19 later in the pandemic (with data collected in 2021 vs 2020). Although there was no indication from the participants that they were withdrawing in response to the SMS text messaging intervention, further (randomized) analysis could help confirm this. The evaluation of the indicative effect of the SMS text messaging intervention was not conducted between the 2 randomized groups, and the baseline difference in hemoglobin levels between the SMS text message–receiving and SMS text messaging control groups, although adjusted for in the linear mixed model, may have affected our results. Therefore, a fully randomized controlled exploration of the effectiveness, as well as the cost-effectiveness, of the intervention is still needed.

The FGDs were moderated by trained research staff from Soweto who were not involved in the development of the intervention. The researcher performing the initial coding and qualitative analysis (LMS) was a cultural outsider as a White European woman, which may have affected nuanced aspects of data interpretation, although the outcomes were reviewed by other members of the research team [58]. LMS was also involved in the study implementation and management (although not the development of either intervention), which may have contributed to observer bias [59]. Nevertheless, this study contributes to the understanding of women’s perceptions of preconception health messaging strategies in our setting.

Conclusions

In this feasibility study, young women found that an SMS text messaging intervention aimed at increasing adherence to preconception micronutrient supplements was acceptable. In addition, mass media health promotion through a radio serial aimed at young people was found to be acceptable and relatable. Participants found the HeLTI Bukhali SMS text messaging intervention beneficial and useful. The intervention may also help improve self-reported adherence, attitudes toward supplements, and measured hemoglobin levels at follow-up, although these results are based on a limited, nonrandomized comparison. Although these findings suggest a previously unexplored potential for SMS text messaging interventions to support improvements in preconception micronutrient supplement adherence, further studies in the form of randomized trials are still needed. In addition, some refinements based on participant suggestions and reassessment of the 2-way SMS text messaging component are needed. Social messaging interventions for improving the health and nutrition of young women could consider how directed messaging (such as SMS text messaging) and mass messaging strategies (such as radio) may be used to complement each other to promote the health of young women in the preconception period.