Since the detection of HIV transmission through breastfeeding, developed countries recommend formula feeding for HIV-infected women [1]. In resource-limited settings, where child mortality is mainly due to diarrhea, pneumonia, and malnutrition, the effectiveness of antiretroviral treatment (ART) in reducing the risk of HIV transmission through breastfeeding has largely resolved the dilemma on how HIV-infected women should feed their infants [2–4]. In such settings, the World Health Organization (WHO) recommends at least 12 months of breastfeeding with infant or maternal ART [5, 6]. Breastfeeding improves child health and development [7–17].

South Africa, a country with a high prevalence of HIV, recently phased out access to free formula milk in the public sector in support of breastfeeding for women living with HIV. Despite the evidence showing reduced risk of infants dying from infections when breastfed, especially in high HIV prevalence settings, breastfeeding rates remain low with challenges to promote breastfeeding by HIV-infected women [7–10]. We conducted a prospective cohort study in South Africa, when the Western Cape prevention of mother-to-child transmission program was phasing out access to free formula milk [18]. We found that few HIV-infected women chose to breastfeed; among those who did, many (50%) switched to formula feeding early (approximately four months following delivery) [18]. Developing simple interventions to promote and sustain continued breastfeeding by women is a public health priority.

Mobile phone text messaging (mHealth) is a simple, low-cost intervention that can promote health behavior change [19, 20]. Increasing mobile phone use in Africa stimulated research efforts on how to leverage mobile phones as a communication tool in healthcare. Text messaging improves adherence to medication among HIV-infected, diabetes, and tuberculosis patients [21–23]. Text messaging not only improves adherence to ART, but also reduces viral load and treatment interruptions [23, 24]. These trials suggest that specific characteristics of the text messages such as interactivity, timing, and content influence text messaging efficacy. Interactive weekly text messaging was superior to interactive daily text messaging [25]. One similarity between adherence to ART and infant feeding practice is that both have a strong behavioral modification aspect: HIV-infected people have to take medication to control the disease while breastfeeding women must modify their feeding practices to improve the health outcomes of their infants. These similarities justify applying information about text messaging from ART adherence studies to infant feeding practices.

Current research suggests that combining a number of approaches is more likely to influence behavior change than an individual approach [26, 27]. Home visits by community health workers and motivational interviewing are interventions known to influence behavior change [28, 29]. However, the former requires considerable human resources and may not be feasible in low-resource settings where the number of healthcare workers is constrained. Patient-centered approaches for negotiating behavior change outperform approaches that instruct patients to change behavior through providing advice [30, 31]. Motivational interviewing is a patient-centered approach that is less coercive and explores the patient’s readiness to change behavior and support the person’s commitment to do so in the preferred direction [32, 33]. Motivational interviewing was of benefit across many health problems, including HIV viral load suppression, body weight, and alcohol and tobacco use [34]. Motivational interviewing is more effective when combined with other interventions [35]. Little is known about the effect of mobile phone text messaging added to motivational interviewing on supporting adherence to breastfeeding among HIV-infected women.

This pilot trial protocol was written following the standard protocol items recommendations for interventional trials (SPIRIT) guidelines (see Additional file 1 SPIRIT checklist) [36].

Participants will be randomly assigned to study groups using a permuted block method. The randomization sequence will be generated using a random number generating program, with a 1:1 allocation ratio with blocks of different sizes to ensure a balanced allocation. Block sizes will be randomly permuted. Electronic sequence generation and random allocation will be done centrally. Women meeting inclusion criteria who consent to participate will be enrolled and immediately assigned a study arm in sequential order. The principal investigator will coordinate the procedures. Only data analysts will be blinded to participant allocation.

Women from peri-urban informal settlements will be invited to participate within 24 h of giving birth at a primary healthcare clinic in Cape Town, South Africa.

All women and their infants, irrespective of their study assignments, will receive health services and treatment according to the respective provincial guidelines applicable in the sector during the study. No deviation from existing guidelines will be caused by taking part in this study. Participants will be evaluated soon after giving birth and post delivery until cessation of breastfeeding or until end of the study. Children will be tested for HIV infection at time points already in routine practice (i.e. at birth and ten weeks). We will extract information on the children’s HIV status from the clinical and laboratory records.

We assume that a sample size of 60 participants is large enough to determine feasibility [38]. The information from this pilot trial will inform the design and provide initial estimates of effect for sample size calculation for the main trial.

Data collection tools will include a baseline questionnaire and a follow-up questionnaire (at weeks 2, 6, 10, and 24 post randomization). Our questionnaires were developed using a validated WHO infant feeding questionnaire [39]. The questionnaires will be used in English and will be administered by a research nurse or counsellor fluent in the participant’s language. In instances where study participants speak only local languages, the researcher nurse will translate questions directly. We will inquire about feeding practices during the previous 24 h and during the prior week. Study follow-up, SMS messages, and responses will be recorded on a study log weekly. Telephone communications with participants in the usual care group not receiving text messages will be recorded in a similar study log. Clinic records will be reviewed for clinically relevant data. Study follow-up will continue until complete cessation of breastfeeding or until end of study.

To maintain participant confidentiality, only a coded number will identify all questionnaires, reports, and other records. A unique patient identification number will be used to link participant-specific data. All paper records will be kept in a locked file cabinet at the research site and at the Centre for Evidence Based Health Care, Stellenbosch University. Access will be limited to research staff. Electronic files will be password-protected. Clinical information will not be released without written permission of the participant, except if required by the ethics review committee. Participant names will be stored separately. No reports will link individual names with person level data. The research nurse or counsellor will receive training prior to beginning of the trial. A standard operations manual will be available to staff for reference on operational details.

Data will be transcribed from the paper format into an electronic database. We will check data to ensure that entered values are acceptable, required fields are completed, and items are consistent with other related items in the database. We will verify with source documents for any discrepant entries. A record of the database update will be kept, identifying information about the person who made the changes, date, changed values, and comments.

The study will be conducted according to the ethical guidelines and principles of the International Declaration of Helsinki, South African Guidelines for Good Clinical Practice, and the Medical Research Council Ethical Guidelines for Research [41, 42]. Participants will provide written informed consent. Participants will receive R150 (~US $10) for their participation time, transport, lunch on study visit days, and any other study-related costs. There are no anticipated physical risks involved in participating in the study; however, participants will be insured by Stellenbosch University’s research policy in the event of some form of physical injury occurring as a direct result of taking part in this study. The Stellenbosch University Human Research Ethics Committee approved the study protocol (reference no. N16/09/11). This pilot trial will not have a Data and Safety Monitoring Board. The study will not have a codebreaking rule for randomization as the study is unblinded, except for data analysts. All investigators declare no competing interests. If there are any protocol modifications, the ethics committee, trial registry, and trial participants will be informed. Study results will be presented in aggregate format in technical reports and journal publications.

The pilot trial evaluates the feasibility of conducting a larger trial on communication and support approaches that may improve adherence to exclusive and continued breastfeeding by HIV-infected women. Several clinical trials on mobile phone text messaging showed improved treatment adherence for chronic medication, including diabetes, HIV, and tuberculosis. Text messaging and motivational interviewing will allow participants to access personalized adherence advice and support. Text messaging is simple and popular among participants and healthcare providers and this may maximize its use in real practice settings. Additionally, our study is unique in that it tests the ability to combine a number of approaches to influence behavior change.

The study will be conducted in a setting where adherence to breastfeeding is poor. This risks finding larger differences between the intervention and control groups than what could be found in settings with better baseline adherence to breastfeeding. However, we assume the intervention would show a similar trend in direction of effect, in settings with better adherence to breastfeeding. Another disadvantage of our study design is that both healthcare providers and the population whose behavior we are attempting to change (including those in the usual standard of care group) are sensitized to the potential benefits of the intervention during enrolment. We will record the frequency with which participants attend the primary healthcare clinics during follow in order to identify any dose-response relationships between interventions and behavior change.

The design and analysis plan will allow us to assess the effect of the intervention on secondary outcomes. The results of this pilot trial will inform further development of a larger trial on use of mobile phone text messaging plus motivational interviewing to improve breastfeeding practices of HIV-infected women in resource-limited settings. While our measure of behavior change is based on mothers’ self-report rather than more objective measures that accurately predict infant feeding behavior, it is a widely used approach for measuring change in infant feeding practices.

The pilot trial is not yet recruiting participants. Protocol version 1 25/06/2017.