This study was approved by the Medical Research Council of Zimbabwe (MRCZ) and the University of Washington IRB. No deviations from the protocol after approval will be implemented without prior approved amendment by these IRBs except where it may be necessary to eliminate any immediate hazard to study participants. In such case, the deviation will be reported to the IRBs as soon as possible. Written informed consent will be obtained from all participants prior to enrollment in the study.
Data Safety and Monitoring Board:
We will form an independent Data Safety and Monitoring Board (DSMB). The DSMB will be available to review and advise on any concerning findings, such as social harms or other adverse events linked to the study, distress evoked among participants that might outweigh the potential benefits of the study, difficulties in recruitment, difficulties in study site participation, or any other issues that threaten the scientific or ethical foundation of the study. The DSMB will have the power to stop the study at any time. The investigators will also reach out to the UW IRB and members of the Medical Research Council of Zimbabwe in the event of concerns related to participant well-being or study validity, to consult on whether study procedures need to be modified or halted.
Using a prospective, randomized control trial (RCT), our intervention compares two groups of clients with cell phones: 1) routine care (control group) and 2) clients who receive and respond to a daily text with in-person follow-up only if desired or if an AE is suspected (intervention). Both arms complete a study-specific, Day 14, in-person, follow-up review for verification of self-reports (intervention) and comparison (control). The intended timeline is presented in Figure 1. The specific objectives are to:
Objective 1: Determine if 2-way texting can safely reduce VMMC follow-up visits through an un-blinded, prospective, non-inferiority, RCT in high-volume facilities providing VMMC. Two-way texting (2wT) will provide interactive, text-based follow-up. 2wT men healing without complication could decline in-person follow-up; those with suspected AEs will be referred to in-person care. We will compare the safety outcome of combined moderate or severe AE rate ≤ Day 14 post-VMMC and the workload outcome of average number of in-person follow-up visits between control and intervention arms.
Objective 2: Estimate the cost savings associated with 2wT over routine VMMC follow-up by determining the programmatic costs of 2wT from a systems perspective, including the technology (48), healthcare worker costs, and client perspective (travel, text costs, missed work). We will estimate the incremental intervention costs relative to routine practice to quantify gains in healthcare efficiency for scale-up and adoption.
Objective 3: Assess the acceptability and feasibility of 2wT for further scale-up via qualitative interviews with VMMC healthcare workers and brief quantitative interviews with 2wT clients inform intervention acceptance. Meetings with local researchers and collaborators will further assess feasibility, adaptation, open-source collaborative development, and system integration for replication and sustainability in Zimbabwe and the region.
Studies that demonstrate the impacts of mHealth interventions too often have evaluated technologies that are made ‘from scratch’ and as a result are not robust enough to merit widespread replication (40, 49). By partnering with a well-established non-profit mHealth organization, Medic Mobile (http://medicmobile.org/), and integrating with their existing software platform, open-source community and current efforts to integrate with existing health information systems (HIS) throughout sub-Saharan Africa, our team and proposed intervention are well positioned to scale up and sustain any promising results. Since 2008, Medic Mobile has been a leader in the global mHealth community (50-52), equipping more than 13,500 health workers serving over 8 million people across 23 countries. The Mobile Medic Toolkit is an Android-based application that supports texting in any language and works with or without internet connectivity on basic phones, smartphones, tablets, and computers (53). The existing, well-proven, app-based Toolkit that will be the basis for 2wT provides an automated and prioritized list of upcoming tasks, guiding health worker through actions (e.g., response waiting, referral)The Toolkit provides real-time progress indicators such as texting delivery rates and response rates. Data from mobile users are replicated to Medic Mobile web app and analytics tools for real-time response. The platform is highly configurable, currently supporting evidence-backed workflows and program implementation related to ensuring safe deliveries (54), tracking tuberculosis patients (55), boosting immunization rates (56) and monitoring stocks of essential medicines (57). These adaptable tools are free, open-source, and developed using human-centered design with input from people delivering care in the hardest-to-reach communities. While this technology has never been used in a VMMC program and requires adaptation of the clinical content, the existing 2wT software toolkit already contains the robust messaging features discussed in this proposal. The software adaptation process to alter and test alternative clinical content and the local specifications requires little additional software development. The adaptation process will be completed by collaborator, Holeman, with additional technical and training support provided by Nairobi-based technical experts. The automated versus manual texts are presented in Figure 2.
In Zimbabwe, 81% of people already had mobile phone subscriptions in 2014 (43), indicating that the technology infrastructure exists and uptake is high. 2wT will be implemented in existing VMMC sites in the Chitungwiza District, a district purposefully selected for its high VMMC volume clinic locations and Norton Hospital which is in Chegutu District. Chitungwiza District, a suburb of Harare, has an estimated 300,000 eligible, HIV-negative men ages 15-29; only 14% of eligible men have been circumcised to date. We will implement 2wT in up to 5 ZAZIC VMMC sites in Chitungwiza- Seke North clinic, Seke South clinic, Zengeza clinic, Chitungwiza central hospital and CitiMed hospital, sites averaging 50-200 VMMC/month. Norton Hospital performs on average 200-300 VMMC/month. As of March, 2017, all VMMC clients are surgical, rather than PrePex device-based, clients.
The City Health Director who manage the national VMMC program at the Provincial level in conjunction with our implementing partners will facilitate access to the clinics/hospitals where ZAZIC provides ongoing support for VMMC. We will also seek permission from the facility manager to conduct the study at that particular clinic. It would be expected that demand creation would happen in anticipation of the study implementation.
Each site will establish local recruitment and screening methods that operationalize protocol-specified requirements for eligibility determination in a manner that is tailored to and most efficient for the local study setting and target study population. In brief, information about the study will be disseminated through the Zimbabwe Ministry of Health and Child Care (MoHCC), ZAZIC, and HCW at the selected sites. VMMC demand creation as part of routine ZAZIC program practice will support study recruitment. HCW recruited for participation in the study will be reached at their workplaces (the health care facilities) following communication between the study team and the site leadership. VMMC clients will be recruited in the VMMC clinic area. Recruitment will be managed by a specifically-trained research coordinator who receive payment for their roles as part of routine or study work. This research coordinator will meet with VMMC patients to sensitize them about the opportunity to participate in a study of text-based follow-up; those meeting eligibility criteria will be individually informed by 2wT staff about the opportunity to participate in the study. Interested patients will be referred to the site study coordinator for study enrollment and informed consent. This person will meet with patients in a private setting, further explain the study, confirm study eligibility, and seek informed consent.
VMMC client recruitment, voluntary consent, and enrollment will be managed by the on-site 2wT study coordinator before VMMC surgery. Eligibility criteria for VMMC clients are: 1) 18 years or older; 2) possession of own phone at enrollment; 3) provides contact details (phone, address, next of kin); 4) requests surgical VMMC; 5) willing to follow MoHCC VMMC protocols; 6) willing to come in Day 14; 7) willing to respond to a questionnaire administered by phone 42 days after circumcision; and 8) no intraoperative adverse event during routine VMMC. Men without cell phones and those who chose PrePex will be excluded as PrePex requires a device-removal visit 7 days after placement and has distinctly different follow-up protocols. As phone sharing practices are common, informed consent will ascertain whether eligible men have consistent access to a phone to receive messages about AEs. Participants who refuse enrollment will be asked to provide informed consent for data collection about reasons for non-participation. We expect to enroll and randomize 722 men into a 1:1 ratio of intervention and control within 6 months, allowing observation of the primary outcome by end of year 1. On Day 14, all men in both arms who return for follow-up will receive a $5 airtime credit. Men who have an inter-operative AE during routine VMMC will be withdrawn from the study as these men will have known, additional follow-up risks and mandatory in-person visits.
For health care workers (Objective 3) within the 6 study sites, inclusion criteria are: employees posted at the site; are at least 18 years of age or over; provide health care services to patients as part of the VMMC programs; and are able to provide written informed consent. Healthcare workers will be excluded if they are not willing to participate or not willing to be recorded
Study participants are described in Table 1. There will be 50 men enrolled in the pilot who will be enrolled in texting and not be randomized. These men will not be included in outcome analysis. There will be 722 men in the full study randomized in a 1:1 ratio of texting and control for 361 men in texting and 361 in control (routine care).
We will conduct a rapid situation analysis with healthcare workers, VMMC clients, and stakeholders to assess suitable responses to VMMC client texts, setting standards for text responses and in-person follow-up. We will modify existing usability surveys for this public health context. A small pilot with 50 VMMC clients who will be enrolled in the texting intervention will include usability testing with both 2wT clients and nurses implementing the 2wT system, illuminating system experiences from both perspectives. Usability results will inform in-box modification, message format preferences (SMS or WhatsApp) and optimal message delivery (timing, frequency, language preferences). Experience from study coordinator and other team members will add detail to inform 2wT adaptation and modification of Standard Operating Procedures before implementation. We will examine local infrastructure (e.g. electricity and cell network), and explore 2wT cost reduction options (e.g., text bundling, free-call back numbers) and adapt accordingly
Routine VMMC care (control arm):
For the 361 men randomized into the control arm, ZAZIC follows all MoHCC protocols based on WHO guidelines (8) including routine surgical VMMC follow-up on post-surgery days 2, 7 and 42 (Table 2). Patients may seek care outside scheduled visits for suspicion of AEs at any healthcare facility at any time but most often return to their VMMC site. Referral cards for VMMC clients provide local numbers for patients to text, call, or request a call back for emergencies. A standardized approach is used to assess, identify, and record the severity of AEs (46). All VMMC care, from assessment of all AEs through complete healing, is provided free to clients from MoHCC. Clients who do not return to the clinic for follow-up on Day 2 or Day 7 are traced: 3 attempts by phone and then up to 3 attempts at in-person tracking after which they are considered lost to follow-up (LTFU) (47). There is no tracing for Day 42 visits. For the purposes of this study, control arm VMMC clients will be asked to come in on Day 14 for an additional follow-up visit. No active follow-up is provided at Day 14.
VMMC care procedures (2wT arm):
We will conduct a prospective, un-blinded, randomized control trial (RCT) among VMMC clients in a 1:1 ratio of control to intervention. Study participants and clinic staff are not masked to treatment. For the 361 men randomized into the intervention arm, men in the 2wT will receive routine VMMC surgical care and counseling, including referral cards for emergencies. 2wT clients will receive automated daily texts from days 1-14 (Table 2). Texting language may be in English, Shona, or Ndebele at the choice of the participant. It is free to receive call and texts; it costs $0.05 to send a SMS in Zimbabwe (58). If they respond that they suspect no adverse event, no immediate follow-up action will be taken. If a 2wT VMMC client responds affirmatively to any daily text that he suspects an AE, a VMMC nurse will exchange modifiable, scripted texts with the client to determine the symptoms, frequency, and severity. Then, if deemed necessary, the client will be asked to return to clinic the following day or earlier if an emergency is suspected. AE management will adhere to MoHCC routine care. If 2wT patients do not respond to texts on Day 2 or Day 7, the same MoHCC tracing process will be activated, after which they will be considered LTFU. All study participants will be asked to come to the clinic for study-specific, Day 14 follow-up to review healing and verify adverse event reporting. Day 14 was chosen for verification because 95% of all AEs within the ZAZIC VMMC program are reported Day 14 or earlier (47), suggesting that most AEs have occurred by this time point. In a previous field study of AEs, the most common AEs of bleeding and infection were found a mean of 6.7 and 9.0 days, respectively, after VMMC (20), further supporting the 14 day period used in this and a previous study(13). The Day 14 review will be conducted by routine VMMC providers according to MoHCC review guidelines. At Day 42, we will implement a brief text-based survey with 2wT clients to ascertain complete healing, providing stronger inferences at study completion.
No compensation will be provided for healthcare workers. A $5 cell phone credit will be given for both control and intervention arm participants on Day 14 visit in appreciation of their time.
Training of Study Personnel:
VMMC staff at study sites will be informed of the study and briefed on study protocols. A 2wT study coordinator will be trained in confidentiality, protection of human subjects, enrollment, and data collection methods, working to not interfere with routine VMMC flow. For technology training, an experienced Medic Mobile trainer from Nairobi will work with the local study coordinator and team members in-person to implement and learn to maintain the Medic Mobile toolkit independently. We believe that achieving local maintenance and ownership of the system is feasible because of Medic Mobile’s considerable investments in open source collaboration and in designing software that is user friendly and intuitive enough to be adapted and maintained by local staff. As part of 2wT, one experienced VMMC nurse will serve as Study Coordinator in Harare and will manage the texting database, including text and voice follow-up communication with intervention clients. Zimbabwe-based co-investigators and interested partner organization researchers will also be invited for Medic Mobile training.
We will create and allocate 361 “texting” and 361 “routine” (ensuring 10 extra per group for withdraws) envelopes. Envelopes were block randomized in groups of 20 using a manual shuffling to ensure a near random order. The full set of 722 will be numbered before distribution to sites. A set of 20 extra envelopes will be created -- 10 per arm – to allow for withdraws with replacement.. Each group assignment envelope will be drawn by the site coordinator and then opened with the participant. Security envelopes will be used to help prevent selection of assigned group. Subjects within each block are randomly assigned to treatment conditions.
For Objective 1, the RCT will apply hypothesis testing for non-inferiority to examine the outcome of interest: AE rate (moderate or severe) occurring ≤ Day 14. A non-inferiority trial is appropriate as we wish to test our assumption that 2wT with reduced in-person follow-up is as safe (not clinically nor statistically inferior) as routine VMMC (59, 60), but with additional advantages including efficiency and lower direct and indirect costs to providers and patients. The non-inferiority margin, based on statistical and clinical considerations, is the maximum difference between the rate of AEs ≤14 Days in control and 2wT arms where we would conclude that 2wT is not inferior to routine care. We set that margin to 1.6% which would create a non-inferiority cutoff of 2% AEs or less in 2wT. The AE proportion cutoff for 2wT non-inferiority was set at a conservative 2% AEs to match the lower bound of AE rates reported by previous rigorous studies (1-3) and because an AE rate of 2% is regarded as a commonly-used standard of VMMC safety (61, 62), including in Zimbabwe (42). Including a possible 10% LTFU, 361 participants per arm provides 90% power at an alpha of 2.5% to detect AEs greater than the non-inferiority margin of 1.6% (Table 3). We would conclude that 2wT is not inferior if we can rule out an AE rate of greater than 2%.
Measures and Analysis:
During informed consent, participants will be asked to give permission to use data from their routine VMMC medical records, including the VMMC register and client intake form (CIF). Data from paper CIFs includes VMMC number, age, circumcision type, eligibility criteria, pre-procedure assessment, and adverse events during the procedure. Variables of interest from CIFs and register will also be entered using Medic Mobile software and verified through data checks, easing merging with the SMS database data management. De-identified, coded data will be shared between researchers via secure networks.
For Objective 1, the 2wT safety outcome of interest is cumulative AE rate (moderate or severe) ≤ Day 14. Incidence of AEs before Day 14 will be extracted from routine VMMC data for both 2wT and control. Incident AEs on Day 14 will be identified, classified, and graded for severity using routine MoHCC protocols (9) and recorded on routine VMMC AE forms. We will compare cumulative rates of any moderate or severe AE ≤ Day 14 between groups using Fisher’s exact test as the expected number of AEs is low. The rates will be calculated per arm as: (# moderate + severe AEs)/ (total # VMMC clients who attend/responded to 2, 7 or 14 Day follow-up visits or texts). Multivariate logistic regression models (any AE v none) will quantify the magnitude of difference, adjusting for any potential confounders. To determine follow-up visit reduction, we will compare the mean number of in-person visits for intervention and control using a t-test. A multivariate linear regression model will further quantify the effect of intervention on visit reduction, adjusting for potential confounders. Secondary outcomes include: AE rates on Day 14, texting response rate, time between 2wT AE text reporting and follow-up, severity of AEs. We will perform appropriate summaries of missing data and LTFU, testing for systematic differences in response by study arm. Data will be exported from the study database into STATA 12.0 for analysis. We hypothesize that 1) 2wT is non-inferior to routine follow-up for patient safety and that 2) 2wT will reduce unnecessary follow-up over routine care.
For Objective 2, 2wT costs, we will calculate the relative costs and outcomes (effects) of intervention versus control, including costs for technology, healthcare worker time, and client considerations (travel, text costs, missed work). We will conduct both activity-based costing from the implementation perspective and from the technology perspective to extrapolate our results as costs that would be incurred by the MoHCC should they elect widespread scale up of 2wT. Approach for costing. First we quantify the total direct and indirect costs of 2wT deployment, a method previously used for technology cost assessment in Malawi (48), which includes comprehensive costs from installation and training, routine maintenance, healthcare worker time associated with 2wT, and staff efficiency gains/losses in addition to the direct cost of technology. Second, we will estimate incremental costs (incremental relative to routine practice) for the intervention. This component entails a micro-costing study using activity-based approaches for costs incurred (trainings, VMMC service provision, follow-up) and costs averted (health costs for providers and patients saved by reducing visits), adapting previous VMMC costing estimates (63) for 2017 dollars when appropriate. Cost data will also be collected from the study budget, public health clinic budgets, published government reports, and the health economics literature. Lastly, to estimate the cost savings associated with 2wT over routine VMMC follow-up, we will conduct a time-motion study (64, 65) to quantify time spent for VMMC follow-up, indicating potential time savings for providers and VMMC clients. One trained observer will record VMMC client/provider follow-up interactions for 5 days, recording times and activities using a pre-established checklist and a personal digital assistant (PDA) that assigns start and end times. During these same 5 days, client time from registration through visit completion will be recorded by a second observer. Data will be exported into STATA 12.0 for analysis. The combined, overall costs for the delivery of 2wT in public health clinics will be estimated and compared to routine follow-up. We hypothesize that 2wT will reduce the costs associated with VMMC patient follow-up over routine care.
For Objective 3, 2wT acceptability, using study recruitment and enrollment logs in addition to the texting database, we will describe levels of acceptance, participation, refusal and drop-out. We will carry out key informant interviews (KIIs) with up to 8 health care workers to gauge acceptability, satisfaction, identify facilitators and barriers to program success, and ascertain suggestions for intervention improvement. KIIs will be audio recorded and transcribed. Atlas.Ti software will be used to create a spreadsheet of key themes, perceived barriers, and suggested facilitators to the program from KIIs. We will also implement questionnaires at the Day 14 visit with a subset of 100 2wT VMMC clients in the main study who were randomized to texting to gauge satisfaction, estimate direct and indirect costs (time away from work, transportation costs), and ascertain suggestions for intervention improvement. Responses from these brief, self-administered, quantitative surveys with VMMC clients will be entered in Excel and frequencies explored in STATA 12.0. For feasibility, costing data will be combined with usability and acceptability information. Qualitative data will be entered, coded and analyzed as text documents in Atlas.ti 6. The first-line data quality assurance will be the responsibility of the I-TECH investigators. These files will also be sent to the team in Seattle, USA for secondary data quality assurance and analysis. The recordings of the interviews will be destroyed one year after the activity ends. However, the transcriptions will be kept for five years in compliance with University of Washington policy. These comprehensive data will be discussed at a final stakeholder meeting to disseminate study results, validate interpretations and refine recommendations for future scale up and modification of 2wT in Zimbabwe and beyond. Additional meetings between local researchers and the MoHCC will further determine feasibility for replication and scale-up. We hypothesize that acceptability, usability, and feasibility will be high, aiding program scalability.