Menu
Menu
Menu
Menu
Background: Young people face special challenges to accessing needed sexual and reproductive health (SRH) information and support. With high penetration and access, mobile phones provide a new channel for reaching young people, but there is little evidence around the impact of digital interventions on improving health outcomes. The Adolescent/Youth Reproductive Mobile Access and Delivery Initiative for Love and Life Outcomes (ARMADILLO) study will assess the effect of an intervention providing SRH information to young people via text message on their mobile phones in Kenya and Peru. This protocol details an open, individually-randomized, three-arm trial lasting seven weeks to assess the effect of the ARMADILLO intervention on dispelling myths and misconceptions related to contraception. A secondary objective is to determine whether youth given access to SRH information via text message can accurately retain that information. Methods: Following a household enumeration, one eligible youth per household will be randomly selected for participation and randomized by computer into one of the three arms. Intervention arm participants will receive access to ARMADILLO content, control participants will receive no information, and ‘Arm 3’ participant interactions will vary by site: in Kenya, they will be alerted to various SRH domains and encouraged to learn on their own; in Peru, they will receive key content from each domain with the option to learn more if they wish. Participants will complete multiple-choice questionnaires administered by data collectors at baseline (prior to randomization), at intervention-period end (after week seven), and eight weeks after timing out of the intervention period. Discussion: Part of the Sustainable Development Goal commitment towards ensuring healthy lives and promoting well-being for all at all ages includes a commitment to ‘ensuring universal access to sexual health and reproductive health-care services, including for family planning, information and education’. If proven to be effective, interventions like ARMADILLO can bridge an important gap towards achieving universal access to SRH information and education for an otherwise difficult-to-reach group. Trial registration: This trial was retrospectively registered with the ISRCTN Registry and assigned registration number ISRCTN85156148 on 29 May, 2018.
The Kenyan site for this study will be within Kwale County, one of six counties in Kenya’s former Coast province. The total population of Kwale County was projected to be 713,487 persons in 2012, with young people aged 15–29 comprising 26% of the County’s total population [16]. In 2015, contraceptive prevalence in the county was estimated to be 38.3%–noticeably lower than the national level prevalence of 53.2% [17]. The adolescent birth rate in Kwale County is also higher than the national average, with 24.2% of women aged 15–19 having begun childbearing compared to the national level of 18% [18]. Data collection will take place in Matuga constituency and Ukunda,the most populated urban area in Kwale County. The Peruvian site for this study will be within the capital city of Lima, on the coast of Peru and home to 9.8 million people, or 31% of the population. About one-third of Peru’s adolescents live in Lima, with the highest proportions of youth living in peripheral areas similar to our study site [19]. In 2015, about one-third of Peruvian 15–19 year old females (32.1%) reported ever having sex and 13.6% of 15–19 year olds were pregnant or were already mothers [20]. The specific ARMADILLO study site will be Pampas de San Juan de Miraflores. Located in the southern cone of Lima, “Pampas” is one of the seven zones in the district of San Juan de Miraflores, which is one of Lima’s 43 districts. Pampas’ 46 settlements (pueblos jóvenes) are home to approximately 50,000 residents [21]. Eligibility criteria for the ARMADILLO study is as follows: In Kenya, participants will be between 18 and 24 years old. In Peru, participants will be between 13 and 17 years old. For privacy purposes, this stage requires participants to have access to their own mobile phones; the Kenya age range reflects formative phase findings that phone ownership dropped precipitously before age 18. In Peru, where younger youth have access to their own mobile devices, the lower age range of participants is a direct response to recommendations from health- and education- sector stakeholders in Peru, who requested that the ARMADILLO study be directed to a younger age group – specifically, 13–17 year olds – to be more in line with national government programs to reduce adolescent pregnancy. The ARMADILLO system has been describe elsewhere [13], but in brief, consists of a free, automated, menu-based and on-demand SMS platform that provides validated SRH information across a variety of youth-identified domains of interest, including puberty, relationships, sex, contraception, HIV/STIs, and rights. SMS or ‘text message’ was chosen as a mode of delivery due to the universal access of the channel, regardless of type of phone owned. The ARMADILLO architecture was developed in each study country using national and global guidelines around youth sexual and reproductive health, with message wording and content vetted by youth themselves during the study’s formative phase. This platform is meant to catalyse additional conversations and information- and service-seeking about SRH by providing brief, essential, guidelines- based information at the moment it is needed (with the ability to save the information for future reference). The structure of the ARMADILLO system, including how each domain is labelled, the sub-topics nested within each domain, and the language of the message itself differs by site. In Peru, ARMADILLO domains are organized as questions: ‘Who am I?’ ‘Who looks after me?’ and ‘How do I take care of myself?’ Meanwhile in Kenya, domains are thematically focused, for example: ‘Puberty and Anatomy’ and ‘Pregnancy Prevention’. Each domain has a menu of 5–10 numbered sub-topics, with a user indicating which sub-topic he/she wishes to learn more about by replying to the system with an SMS of that number. Finally, each queried sub-topic provides information through 2–3 SMS, totalling 320–480 characters. In Kenya, the system is in both English and Swahili (with users able to indicate their language preference). In Peru, the system is in Spanish. Those assigned to the intervention arm will be provided access to one new domain (SRH topic) every week (Day 1 of 7 of a given week) and with an SMS ‘quiz’ to maintain engagement at the end of that week (Day 7 of 7). These participants will have the ability to navigate freely across and between messages, accessing subdomains that interest them and ignoring those that don’t. At the start of the next week (Day 1 of 7), access to the previous week’s domain will close – meaning that the user can no longer access the information in that domain. However, any messages the user has already received will remain in his/her phone unless deleted by the user. At the same time, the next domain will be ‘unlocked’ and the user will receive a message alerting them to this new access. Progression through the intervention will be as follows: Some participants will be randomly allocated to the control arm. Control arm participants will emulate standard access to SRH information and so receive no messages on their phone during the intervention period. Instead, they will only participate in a baseline assessment (described above), and in endline and follow-up assessments (described below). The third arm has been developed in close collaboration with each study site, and aims to address specific questions from countries and/or the digital health literature. In Kenya, the ‘contact’ arm will allow for assessing whether changes in SRH outcomes are attributable to the content of the intervention or to the participant-contact nature of the intervention itself. As such, the ‘contact’ arm will match the number of system-initiated contacts without providing them access to ARMADILLO content. If the success of digital health interventions is merely in participant contact (i.e. encouraging users to consider a relevant SRH topic) rather than the informational content, we would expect no differences in outcomes between intervention and contact arms, and participants in both arms would likely have better outcomes than control participants. Therefore, participants in Kenya’s contact arm will not have access to ARMADILLO message content. They will, however, receive the same number of system-initiated contacts (pushed messages) that the intervention arm receives: specifically, they will receive messages alerting them to a new SRH domain at the beginning of each week, and the domain-specific assessment at the end of the week. Progression through the contact arm will be as follows: In Peru, push-message public health campaigns are the norm for a wide variety of sensitive (re: HPV vaccination) and non-sensitive (maternal health, nutrition, etc.) health campaigns. Bi-directional interventions do not exist; this will make the on-demand format of the ARMADILLO intervention arm novel. As such, Arm 3 for Peru will adopt the Peruvian ‘push system’ norm, answering a question important for adoption and scale up to national level in this context: how does free, active engagement (intervention arm) compare with passive receipt of messages (Arm 3) for engaging with users? Progression through the push arm will emulate existing public health campaigns in Peru and will be as follows: Messages pushed to phones incur no cost for the recipient. However, the study design requires intervention and Arm 3 participants to both send and receive messages. Sending SMS does incur charges for the study participants. Therefore, both intervention arm and Arm 3 will be zero-rated, with any SMS charges reversed billed to the study. Participants, during the enrolment process and periodically during the intervention, will be reminded that any interaction with the study is free. Participants will also have access to free airtime upon responding to the SMS quiz – this airtime will be credited to users by the study team. Participants will also be periodically reminded that they can opt out at any time by sending a ‘STOP’ command to a short code. This opt-out option helps address a lingering ethics concern that a person may choose to exclude themselves from a study because their access to messages has compromised their comfort or security. As such, should a member of either SMS-receiving arms indicate via shortcode that they wish to be unenrolled, they will be immediately and automatically unenrolled from the study. Arm 3 and intervention participants in both sites will progress through domains sequentially, based on numbers assigned to each domain. However, participants will be randomized as to which number domain they start with, in order to account for participant recall bias at the endline assessment as a result of having seen domains at different points across the seven weeks. For example, a participant may be randomly assigned to start at Domain 3, meaning they would cycle through Domains 4–7 in the following weeks before finishing with Domains 1 and 2. All outcomes are linked to domains and content from the ARMADILLO system. The primary outcome, dispelling myths and misconceptions about contraception, will be assessed using an index of 8–10 of the most salient myths and misconceptions about contraception [22] (for example: family planning causes cancer, family planning can make a woman barren, etc.) in a given site. The index will be country-specific (though certain myths are likely to be relevant in both settings) and developed based on a review of relevant literature and a series of focus group discussions conducted in each study area prior to the start of this research. Those in the intervention arm are hypothesized to believe fewer contraception myths than those in control arm and Arm 3 and should therefore have a significantly lower index score at endline than they did at baseline, compared to the other two groups. Secondary outcomes of interest concern change (and retention of change) in knowledge of contraception; knowledge of puberty/anatomy; knowledge of HIV/AIDS and its transmission; attitudes around engaging in sexual activity (with self and others); attitudes around intimate partner violence; attitudes around family and peer support; and previous behaviour around sex and contraception use. These concepts will be measured using scales and indices that each consist of multiple questions to comprehensively measure the concepts. These have been pulled from a number of validated survey instruments [22–28] including (but not limited to) the Guttmacher Survey of Young Adults [29] and the Demographic and Health Survey (DHS) Program [30]. A sample size of 705 participants per site will provide 80% power to detect a 10% change in mean number of myths believed from baseline to endline, assuming that baseline level of belief is 55% (or .55, with an assumed standard deviation of .30) [22] and accounting for a dropout rate of up to 20% in each site. The sample size is calculated to allow for pairwise comparisons between all three groups. The slight increases in sample sizes in each site are to allow for balance in sample size across site-specific age groups as well as for males and females. We will use household-based surveys and multi-stage random sampling. First, recent satellite images or a recent census will be used to identify and enumerate all of the households in the study zone. A random sample of blocks of households will then be selected. Next, the research team will carry out a census of households to enumerate all eligible (those meeting the eligibility criteria for this study) youth in each selected block. Following the enumeration, a member of the study team in each site will randomly generate (using a computer-based random number generator) a list of potential participants to be sampled. To minimize contamination, only one youth from each household will be able to participate in this study. The list will be a random generation of 1) the households to be sampled and 2) one eligible youth to be recruited from within that household. Data collectors (none of whom will be involved in the randomization process described above) will be sent to a specific geographic area with the list of households. Recruitment will take place over a one month period. If the youth selected is not at home (and/or, in the case of Peru, the parent/guardian is not at home), the data collector will make an appointment for a second visit. Upon visiting the household, the individual will be consented on the spot. In the case of Peru, where participants are under age 18, parent/guardian consent will also be obtained at this point. If that individual does not wish to participate, no other eligible member of the household may be substituted. Following recruitment, consenting, and completion of the baseline survey, the data collector will leave the household and each participant will be randomized off-site in a 1:1:1 allocation ratio using a computer-based randomization tool to intervention developed using Node.js and docker. The randomization will be overseen by a research team member does not interact with participants. Once randomized, the participant’s seven-week interaction with the relevant arm will begin the following day. Across all arms, outcomes will be assessed via an in-person questionnaire, administered by a trained researcher, with data collected on a mobile phone or tablet. The questionnaires will be administered: All in-person study activities will take place in a private room at a location and time that is convenient for participants. Locations will allow for both visual and auditory privacy. A project staff member will explain the study activity and answer any questions prior to and/or following completion of the given study activity. Following recruitment and consent, all participants will complete a baseline, close-ended survey with questions linked to ARMADILLO content. Participants will be told that, should they be randomized to intervention or Arm 3, they can expect their first message from the system the following day. Following completion of the baseline survey, the participant is randomized into one of the three arms. During the last week of an individual’s intervention period (Week 7) the participant will be contacted to schedule an in-person assessment of all outcomes at the end of the intervention. Research team members will update participant contact information if necessary and alert participants that they will be invited back for a follow-up assessment in two months. There will be no contact with participants from any study arm in the eight weeks between the end of the intervention and follow-up. The ARMADILLO system will be offline in this period as well. Two months following a participant’s end-line assessment he/she will be contacted for an in-person, follow-up assessment of all SRH outcomes. Raffles and airtime may also be used to encourage participants to return for the follow-up assessment. Following the completion of the in-person assessment, returning participants in all arms will be provided with a short code to access the full ARMADILLO system. In the event that the system must undergo substantial revision prior to finalization, study participants will be told that they will be alerted via SMS when the final system is ready for access. All study results will be kept confidential by the team in either password-protected files for electronic data or locked cabinets for paper data. Only approved team members will have access to study results. A master list will be maintained that includes ID numbers that are uniquely assigned to each participant. Interview notes, consent forms, and digital files will be labelled only with these ID numbers. These master ID lists and informed consent forms will be stored together in a locked cabinet. Master ID lists and informed consent forms will be kept separately from any printed data related to the study. All data collected will be marked with the ID number of the relevant participant – this will be the only unique identifier for any data. All hard copy documents that contain study results will be stored in a locked file cabinet (separate from the Master ID lists) that is accessible only to key study personnel. Digital data files will be stored securely on a password-protected computer and on password-protected cloud storage such as Dropbox. Study materials will be destroyed after three years. The study coordinator for each site, under guidance of the PI, will be tasked with ensuring that all files (hard and digital) have been deleted at the appropriate time. Only the study coordinator and principal investigator for each site will initially have access to the locked cabinets and password protected storage devices used in this study. If both principal investigator and study coordinator are in agreement, access to data files on cloud storage can be granted to select research staff who will be participating in the data analysis. Any data related to the ARMADILLO system (intervention, control, and arm 3) will be stored on an instance of RapidPro, hosted by the technology partner, Ona. RapidPro, the open source communication platform of choice for this intervention, has the ability to passively track participants’ progression through various domains and message content. Data is available on what content an individual phone number accesses and how often. Tracking phone numbers also offers an unobtrusive mechanism for reducing contamination, allowing the research team to monitor the intervention to ensure that the only phone numbers accessing the content of the intervention and contact arms are those associated participants randomized to those respective arms. It should be noted, however, that within the ARMADILLO system, these phone numbers will not be linked with any identifying information on the participant. Additionally, study teams retain complete ownership of the data and of the account(s) on RapidPro. As agreed with Ona, only the study team will be able to grant access to the account to others, including Ona. Having complete ownership of the account, the study teams will be able to completely delete their own data once the project is finished. It should be noted that while the intervention makes use of mobile network operator (MNO) infrastructure (with text messages being relayed from RapidPro, then through an aggregator, to participants’ phones), there are no special privacy or confidentiality concerns to users arising from their respective carrier. MNOs in each country will have no way of knowing the content of the messages being sent to participants. MNOs will have no more ability to monitor users who happen to be ARMADILLO participants than they would regular users. After assessment of the randomization of participants, if there are no differences between arms in terms of sociodemographic characteristics, differences between arms in the contraception myths index score at endline will be assessed using standard methods—comparisons of proportions (chi-square tests) and means (t-tests) between arms and difference-in-difference techniques. If the randomization was not successful, statistical adjustment with multivariable regression will be used to assess intervention impact. It is envisioned that analyses will be conducted as intent-to-treat. Separately, similar secondary analyses will be conducted on knowledge gained and attitudes shifted in other SRH domains assessed, as well as the retention of both primary and secondary outcomes over time (through the follow-up period). The WHO HRP Review Panel on Research Projects (RP2), comprised of a committee of external reviewers, reviewed and approved the scientific and technical content of the study (protocol ID, A65892b). We then obtained ethics review and approval from the WHO Research Ethics Review Committee (ERC). Local ethics approval was also obtained from Institutional Review Boards of the Universidad Peruana Cayetano Heredia as well as the University of Nairobi/Kenyatta National Hospital. Those who voluntarily consent to participate in the study will be fully informed, as part of the consent process, that the study concerns SRH topics that they may consider sensitive and that it is possible that questions in the survey or message content could make the respondents feel uncomfortable. Participants will be informed, as part of the consent process that they can choose to ignore questions or leave the study altogether at any point if they feel uncomfortable, without repercussion. Data collectors will be the same sex of participants, in order to minimize any discomfort participants may have commenting on the subject matter. Participants will be encouraged to identify an environment in which they are comfortable speaking with the research team, and a time that suits them. Also, there are certain features of the ARMADILLO system itself which have ethical considerations worth mentioning. First, while ARMADILLO only purports to provide cursory information on the included topics, with the idea that the user accesses a different source for in-depth information and/or to take action, there is one notable exception: gender-based violence (GBV), for which a participant may require immediate support. The messages on gender-based violence will therefore contain direct linkages to site-specific resources from which to seek immediate support. Both Peru and Kenya have existing 24-h hotlines available to victims of violence. Additionally, both study sites have GBV services in the catchment area. As such, messages in the violence module will provide the relevant hotline phone number and information on where participants can seek services if needed/desired. Prior to the start of the study, each research team will liaise with GBV support and counselling centres in their respective catchment areas to be sensitized to the services available. At the point of consenting, all participants will be provided with a discrete (e.g. credit-card-sized) card that also contains the list of relevant hotlines and GBV services in the catchment area. Second, ARMADILLO was developed as an on-demand system to preserve the privacy of users. However, participants in this study must also receive occasional messages pushed to their phone (quizzes and topic/unlocked domain alerts), thereby sacrificing some of the privacy that a purely on-demand system would offer. In order to compensate, eligibility criteria requires that participants in this phase have their own mobile devices, addressing a concern of an SMS appearing on a shared phone when the study participant is not using it. A common concern in digital health interventions is whether there is a privacy-related risk to participants, in that an outside person (parent/caregiver, partner, friend) may view the messages on the participant’s phone against their wishes. This was a concern first flagged during the formative stage of research and so was thoroughly explored over the course of data collection. In both sites, formative data indicates that even if partners, peers and/or parents were to come across information from the study, there would be minimal – if any – repercussions. In Kenya parents, caregivers, and stakeholders engaged with the formative phase confirmed that their communities would view mobile phones as a valuable means of providing SRH information to young people. Meanwhile, in Peru, respondents indicated that health-oriented message campaigns were common place and would be acceptable so long as they appeared as ‘public service messages’. As such, privacy concerns are anticipated to be a non-issue; however, should we find otherwise over the course of data collection, the study will not continue.
N/A
