Article Text
PDF
Abstract
Objectives To document adaptations that were made to the implementation of the two-way texting (2wT) randomised controlled trial (RCT) for voluntary medical male circumcision (VMMC) in South Africa and to provide a nuanced discussion on the differences between adaptations and fidelity in this context.
Design We conducted a qualitative study using the Framework for Reporting Adaptations and Modifications in Evidence-based Implementation Strategies (FRAME-IS) to examine 2wT adaptations. We reported adaptations to the 2wT intervention using two steps. First, we categorised adaptations in a shared study-specific Google Docs that documented participant engagement with the 2wT system, tracked daily RCT implementation notes, reported software bugs and noted reminder emails about adaptations for the research team. Second, we conducted a qualitative assessment of the influence of adaptations on project outcomes via 10 periodic reflection meetings with VMMC implementers. Reflection documentation included notes from field observations, meeting minutes and informal partner check-ins to complete adaptation documentation. Using the FRAME-IS as a codebook, adaptations were categorised.
Setting The RCT was conducted in rural and urban VMMC clinics in the North West and Gauteng districts of South Africa.
Participants Implementation scientists and VMMC implementers who implemented the 2wT pragmatic randomised controlled trial (pRCT) were participants for the adaptation study.
Primary and secondary outcome measures The primary outcome measure was the adaptations that were made during the implementation of the 2wT pRCT. The secondary outcome measures were fidelity and rigour of implementing adaptations to the 2wT pRCT.
Results Between June 2021 and February 2022, 13 adaptations were identified in three phases during the implementation of the 2wT pRCT. The first phase of adaptations aimed to augment study recruitment, including conducting weekend VMMC recruitment camps, using mobile outreach services in the rural site, adding two urban sites to increase recruitment, using weekly WhatsApp calls for updates with all implementing teams, using virtual meetings to implement the 2wT strategy remotely during COVID-19 restrictions and allocating one clinician to work outside of normal working hours. The second phase of adaptations further enhanced enrolments, including adding two local language translations in the usability survey for 2wT men and contributing a portion towards the salary of the implementing staff by the research partner. The third phase included the exclusion of two rural clinics as recruitment sites due to inconsistent mobile phone networks, adding another layer of data quality checks to ensure data quality, training non-clinical counsellors to help with enrolling clients, retraining of staff in the rural site with high staff turnover and using both primary and alternative phone numbers for enrolment to reduce loss to follow-up.
Conclusions This study made adaptations to the 2wT pRCT without compromising the fidelity of the study. The 2wT pRCT balances rigour (fidelity) and relevance (adaptation). Adaptations should not be confined by rigour but should also not go unchallenged or unverified. We conclude that fidelity can be maintained with adaptations that are implemented to close the gap between research in the laboratory and practice.
Trial registration number The trial from which this study was conducted, ‘Expanding and Scaling Two-way Texting to Reduce Unnecessary Follow-Up and Improve Adverse Event Identification Among Voluntary Medical Male Circumcision (VMMC) Participants in the Republic of South Africa’, was registered at ClinicalTrials.gov (ID: NCT04327271) on 31 March 2020.
- HIV & AIDS
- Digital Technology
- Implementation Science
- Randomized Controlled Trial
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
In capturing adaptations in real time, rather than through interviews as is often the case with studies on adaptations, we have identified more and different adaptations than if we had waited until the end of the trial period.
Although the adaptations were implemented to accommodate the COVID-19 pandemic, the pandemic in turn spurred these adaptations, and two-way texting (2wT) became the right mobile health tool at the right time.
Implementation science (IS) is designed to improve routine healthcare, but the capacity to carry out an IS adaptation study of a 2wT pragmatic randomised controlled trial (pRCT) in resource-constrained settings was limited by inequality, poverty, human resource capacity, overburdened healthcare clinics and so forth.22
In conducting an IS pRCT within both the emergency context of COVID-19 and longstanding contextual constraints embedded in the South African healthcare system, it became difficult to command and control the routine setting to the minute detail to achieve the rigour demanded by a randomised controlled trial.
While the research team conducted a pRCT and the clinical teams did their everyday jobs to ensure relevance, adapting the approach to account for local constraints also required balancing between rigour and relevance.
Background
Implementation science (IS) involves finding ways to maximise the adoption and uptake of known and tested evidence-based interventions (EBIs) in practice.1 To aid in the adoption of EBIs, researchers and implementers may make use of implementation strategies, which cover the ‘how to’ part of delivering the EBIs.2 These strategies may include methods of provider training, use of tool kits, checklists, guidelines and user manuals.3
In acknowledging the difference in settings and environments in which EBIs are implemented, it is imperative that these implementation strategies be tailored for each specific setting for the efficient delivery of the EBI.3 In this study, an adaptation was defined as altering the delivery of the pragmatic randomised controlled trial (pRCT) considerately and deliberately to improve its fit or effectiveness in the rural and urban contexts.4 The adaptations we made reflected diverse processes of change that were responsive to unanticipated challenges that arose during the implementation of the pRCT.5 These adaptations were required to fit the pRCT, its budget, timelines and staffing needs in the rural and urban environments.6 When adaptations were made to the pRCT, the original design of the intervention was left unaltered, thereby maintaining fidelity. We defined fidelity as staying true to the original design of the pRCT by implementing the two-way texting (2wT) intervention as we intended as developers.7
Adaptations and fidelity are concepts that are carefully negotiated in randomised controlled trials (RCTs). RCTs require implementers to remain consistent with the core elements of EBI through the rationale, process and outcome of the intervention—maintain fidelity. Our pRCT, implemented in real-world, routine, rural and urban community settings, required implementation scientists to make adaptations due to evolving changes in the population characteristics, research and implementing agencies and/or community contexts.8 Our pRCT aimed to evaluate the effectiveness of a 2wT intervention in real-world settings rigorously, not strictly in research settings, to understand better and bridge the divide between research and routine contexts.9 We used adaptations as tools to achieve this.
We conducted a pRCT in South Africa between 2021 and 2022, applying the process of planned or purposeful adaptations to the content and delivery of the intervention10 in response to emerging and evolving COVID-19 constraints and opportunities.10–12 The pRCT aimed to determine if a mobile health (mHealth) innovation, 2wT can safely reduce postoperative follow-up after voluntary medical male circumcision (VMMC) while reducing provider workload.
In this paper, we describe adaptations to the delivery of a mHealth-focused pRCT in response to the volatile environment during the COVID-19 lockdown period in South Africa.
Methods
The 2wT intervention
The 2wT messaging platform was built using Medic’s open-source Community Health Toolkit. The intervention was described in detail previously.10–13 In brief, the 2wT system comprises four core components at the provider and patient levels: (1) hybrid automated and interactive patient-to-provider messaging over the first 13 days of post-VMMC follow-up, (2) SMS text messaging–based triaging of clients by nurses (eg, for reassurance, referrals to care and follow-up in case of no contact), (3) daily client monitoring via SMS text messaging, and (4) longitudinal patient records (potential adverse events (AEs), AE follow-up and referral confirmation) and reporting (eg, client response rates). These features of 2wT were designed to support a streamlined workflow, reinforce high-quality VMMC services and generate data to monitor the programme delivery.10–13
For the pRCT, 1093 men were randomised 1:1 across two arms and divided equally across urban and rural districts, with 553 men assigned to the 2wT intervention arm.10 Recruitment for the study commenced on 7 June 2021, and follow-up was completed on 21 February 2022. Males aged 18+ were followed up for 14 days by either routine, postoperative, in-person visits on days 2 and 7 (control) or 2wT daily messaging with a VMMC nurse in lieu of clinical reviews (intervention). All participants returned for a study-specific review on postoperative day 14 to determine healing status and record AEs. 2wT aimed to support patients to monitor wound healing on their own and to empower them to opt in for physical follow-up visits only if necessary. 2wT visits and AE outcomes were compared between groups.
Study design and framework
We conducted a qualitative study using the Framework for Reporting Adaptations and Modifications in Evidence-based Implementation Strategies (FRAME-IS) summarised in figure 1. We documented adaptations to the strategies that were employed in implementing the 2wT EBI using four modules of FRAME-IS to describe (1) the 2wT intervention, the implementation strategies employed and the adaptations done; (2) what was adapted; (3) the nature of the content adapted and (4) the goal and the level of the rationale for the adaptation.14 We also assessed rigour and relevance in documenting adaptations.
The Framework for Reporting Adaptations and Modifications in Evidence-based Implementation Strategies (FRAME-IS) for documenting adaptations to the implementation of interventions.
Using the FRAME-IS as a codebook, adaptations were described and categorised. We documented adaptations to the implementation of the 2wT study using two steps: (1) we used qualitative data from Google Sheets to identify, categorise and describe adaptations and (2) qualitatively assessed the impact of adaptations on project outcomes via a review of programme reflection documentation (field observations, meeting minutes, informal check-ins, etc).15
Patient and public involvement
A Community Advisory Board was consulted during the conduct and design of the study and acted as community gatekeeper. The results of the study were disseminated to study participants and community stakeholders at two district health research conferences in 2023 and 2024. Dissemination to other stakeholders is planned through the local and international conferences and publications.
Ethical approvals
The study obtained ethics approval from the University of Witwatersrand Health Research Ethics Committee and the provincial research ethics committees in the two provinces where the study was conducted. The Data Safety Monitoring Board for the 2wT study reviewed interim safety at periodic intervals and on completion of the recruitment of 1093 men in the study.
Data collection methods and documentation
We used various data collection methods to document the implementation for the delivery of pRCT, as summarised in table 1.
Google Sheets for the 2wT arm
Step 1: Google Sheets and other primary data sources
Daily progress notes and other operational data were recorded on Google Sheets as part of routine study monitoring.
Google Sheets is a web-based spreadsheet that allowed the research team to create, update, modify and share the data online which we accessed through our Google accounts. The study nurse and the data capturer completed the Google Docs daily. This process documentation for the study showed enrolments per site, per arm and daily interactions with the participants. It was also used to triangulate data and was shared in real time.
For the 2wT arm, columns F through I were completed between day 1 and day 3 for the day 2 visit, and columns J through M were entered between day 6 and day 8 for the day 7 visit. This section recorded whether we made contact with the participant between day 1 and day 3 and is equivalent to the day 2 visit for routine care and between day 6 and day 8 for the day 7 visit. Only one of these columns in each section (columns F–I and columns J–M) could be answered with yes to indicate the type of contact that was first documented. If the participant did not respond to automated texts, text reminders and phone calls, then all fields could be answered with a no.
Other primary data sources included participant observations, training workshop notes, site visit reports, emails and communication through the study team’s WhatsApp. We used a data extraction sheet to document data on adaptations that were made to the implementation strategies, reviewing and coding content in Google Sheets related to adaptations made to the strategy, fit and effectiveness. Adaptations to the implementation of the 2wT study were extracted and coded from Google Sheets.
Step 2: periodic reflections
As part of informal or routine partner and stakeholder review meetings, we conducted 10 periodic reflections with implementers of the 2wT pRCT in the rural and urban sites to answer the following IS optimisation questions:
What component or part of the implementation to deliver the 2wT intervention was changed in this adaptation; in other words, what was the nature of the change?
Who was responsible for first suggesting or initiating this change? Was this the person or persons who implemented the change? If not, who implemented the adaptation?
When during the 2wT intervention was this adaptation first made?
Why was this adaptation made? Was this done to get more people to participate, to make the programme attractive to more settings, to increase its effectiveness, to make it easier to deliver, to make it easier to maintain or reduce costs, etc?9
The research manager conducted periodic reflections and observations in the form of 15–60 min meetings, at least once a month. These were lightly guided discussions by telephone, Zoom or site visit conducted with individuals (such as nurses, clinical associates, team leaders, data managers and recruiters), small teams (such as the surgical, data management or recruitment teams) or site teams (rural or urban teams) to observe, discuss and document real-time adaptations. The periodic reflections were audio-recorded and transcribed.
Data analysis
To explore adaptations in the pRCT, we reviewed participants’ engagement with the 2wT system, tracked the database for daily operational notes and reviewed reminder emails about adaptations to the research team and partners. We documented implementation choices, constraints and challenges of the 2wT intervention via a shared Google Docs that was accessible to all pRCT study staff and updated each weekday as per study protocol. At the completion of the pRCT, we qualitatively and descriptively reviewed and considered the adaptations to the various implementation strategies. We documented changes made to improve men’s engagement with the 2wT system, reviewed choices that improved implementation feasibility and updated to the implementation strategy that aimed to increase equity (eg, language and operation hours) of 2wT delivery in both rural and urban environments. The Google Sheets was consistently reviewed by principal investigators (PIs) to reduce missing adaptations and ensure accuracy. We documented adaptations that impacted on costs as well.
We employed two distinct sources of fidelity data: first, through direct observations during periodic reflections with implementers, leveraging their expertise and objectivity; second, by analysing clients’ engagement with the 2wT system, benefiting from real-time reporting. This dual-source approach allowed for comprehensive comparisons between fidelity information derived from implementers and clients. The weekly collection of detailed client data in the system facilitated a nuanced analysis of fidelity patterns over time, supplying valuable information to the PIs and adaptation team.16
Using the FRAME-IS coding manual, two researchers categorised the adaptations to the 2wT project using the categories within each FRAME-IS construct.17 We used the Google Docs and periodic reflections to identify, categorise and describe adaptations. Then, we qualitatively assessed the influence adaptations had on project outcomes. To assess influence, we examined the categories for each adaptation within the project in relation to the construct of perceived short-term impact, including impacts to reach, adoption and implementation, and whether the adaptation preserved or altered the intervention’s core elements or functions. The researchers used their depth of knowledge about the intervention to determine whether the adaptation had an influence or impact on outcomes.18
Thematic content analysis was used to analyse the data from periodic reflections and source documents (18). We created a spreadsheet of key themes derived from the FRAME-IS model, described the adaptation and answered the four categories of IS questions (what, who, when, why). Codes were compared between analysts and discussed to reach a consensus for documenting adaptations.
Ethics
This qualitative study was embedded in an RCT, ‘Expanding and Scaling Two-way Texting to Reduce Unnecessary Follow-Up and Improve Adverse Event Identification Among Voluntary Medical Male Circumcision (VMMC) Participants in the Republic of South Africa’, that is registered at ClinicalTrials.gov (ID: NCT04327271). The overall study, including this substudy, was approved by the Internal Review Board of the University of Washington (Study 00009703, PI: Feldacker) and the University of the Witwatersrand, Human Research Ethics Committee (Ethics Reference No: 200204, PI: Setswe). Participants in periodic reflections received comprehensive information regarding their voluntary participation in the study and signed a written informed consent form prior to study enrolment.
Results
The EBI being implemented was 2wT, a text-based follow-up method for VMMC tested through a pRCT in rural and urban settings in South Africa. The consort flow diagram (attached as an additional file) provides the flow of enrolment, allocation, follow-up and data analysis for the 2wT RCT. Table 2 dissects adaptations made to various implementation strategies employed to deliver the EBI. For each adaptation, we described the adaptation, indicated when the adaptation was made, who made it, at what level and the goal of the adaptation as shown in table 2.
Adaptations made to implementation strategies to deliver two-way texting (2wT) for voluntary male medical circumcision (VMMC), adopted from Framework for Reporting Adaptations and Modifications in Evidence-based Implementation Strategies (FRAME-IS)
Between June 2021 and February 2022, 13 adaptations were made across seven rural and urban sites where the 2wT pRCT was implemented. These adaptations (table 2) included tailoring, tweaking or refining some elements, changes in the packaging of materials and removing elements. Two sets of adaptations were made to ensure rigour/fidelity or relevance. In table 2 under goal of the adaptation, we label each of the adaptations either as an adaptation to (1) ensure rigour or fidelity in implementing the 2wT as it was designed or (2) ensure relevance of the study to the context and environment that prevailed. Of the 13 adaptations made to the 2wT study, six were adaptations to ensure rigour or fidelity and seven were adaptations to ensure relevance of the study.
Adaptations to ensure rigour or fidelity
Six of the 13 adaptations aimed to ensure rigour and fidelity of the study. These adaptations were to (1) increase the reach of 2wT or the number of patients receiving 2wT through weekend VMMC recruitment camps; (2) increase recruitment using weekly WhatsApp calls for updates with all implementing teams; (3) improve acceptability, appropriateness and feasibility, engagement and implementation of the 2wT by using check-in calls three times a week, weekly project update calls with all implementing teams; (4) increase engagement and feasibility of 2wT and improve the fit between the implementation effort and the needs of those receiving 2wT by adding two local language translations in the usability survey; (5) improve the sustainability of 2wT and increase the chances that 2wT remains in practice by implementing additional data quality checks and (6) increase adoption or number of clinicians recommending and patients using 2wT by retraining of staff members on recruitment, recording and follow-ups.
During the week, participants were working or attending school. To address this and other contextual factors, such as high unemployment rates and precarious employment opportunities, we adapted the 2wT strategy. We conducted camps on selected weekends, reaching men without disrupting their work or school schedules and significantly boosting study recruitment.
During COVID-19 lockdowns, the team used virtual meetings and digital technology to implement 2wT remotely to ensure rigour and fidelity.
Adaptations to enhance enrolments and ensure compliance with study protocols included adding two local language translations in the usability survey for 2wT men and contributing a portion towards the salary of the implementing staff by the research partner. During the pilot, the team ascertained that the usability study needed to be translated into local languages (ie, Setswana and isiZulu) to reach participants who were not fluent in English.
The team used virtual meetings and digital technology to implement and refine 2wT remotely to compensate for international and local COVID-19 travel bans, which restricted the possibilities of physical meetings. To address communication challenges arising from COVID-19 restrictions, the challenges of working across countries and provinces, and working in network-poor locations, various platforms—such as WhatsApp and virtual calls—allowed the teams to provide updates on progress, deal with challenges as they arose and adapt implementation accordingly.
The VMMC team had data quality monitors, but continuous monitoring revealed quality assurance gaps. To address this, the PIs added another layer to review the quality of the data collected. The research team also conducted in-person data reviews to deal with data quality issues.
Staff turnover became very high in the rural site during the early part of the intervention. New staff had to be trained and onboarded, and existing staff were retrained as needed. To accommodate high staff turnover rates, a robust training approach that is effective for quickly onboarding new staff is required.
Adaptations to ensure relevance
Seven of the 13 adaptations were to ensure that the study is relevant to the context and environment in which it was implemented.
South African COVID-19 level four restrictions required the decongestion of health facilities. At the rural site, we used mobile outreach facilities to reach clients in their community. This adaptation addressed geographic distances in rural areas, allowing for more men to be reached.
Recognising the slow enrolment in the urban area, the research team identified the need for additional sites. During the implementation stage, we proactively added two more sites to address the time constraints associated with the pRCT target recruitment numbers. These sites were selected based on the implementing partner’s already existing setup within public clinics, ensuring a smooth and efficient process (2).
Implementing staff initially viewed 2wT as a separate intervention from routine follow-up for VMMC, resulting in a reluctance to attend to clients after hours and additional time burdens for documentation of pRCT processes. To address this, a portion of the implementing staff’s salary was allocated from the 2wT budget.
Clients would contact clinicians after hours, on public holidays and weekends. One adaptation made was a roster for clinicians to attend to clients after hours to distribute the additional workload. Before the roster, all the extra hours would fall on one or two clinicians, creating unequal workloads.
The exclusion of two rural clinics as recruitment sites due to inconsistent mobile phone networks are as follows: adding another layer of data quality checks to ensure data quality, training non-clinical counsellors to help with enrolling clients, retraining of staff in the rural site with high staff turnover and using both primary and alternative phone numbers for enrolment to reduce loss to follow-up.
Two facilities in the rural area were excluded because these sites had no mobile phone network coverage during periodic scheduled electricity cuts called load-shedding, and enrolments could not be done. While offline enrolments in the Medic system were not possible at the time of the pRCT, developing future offline functionality could further adapt the intervention to network-poor contexts. These sites were also very far from the rural team hub, and routine client follow-ups were impractical.
During the busy winter season, clinicians were occupied attending to VMMCs and could not administer informed consent and capture client details on the Medic system. To address this, non-clinical counsellors were recruited to assist with client enrolment and capture them on the Medic system. Each intern was provided with a device and sufficient data to recruit, educate about the 2wT approach and enrol clients in the Medic system. This adaptation reduced clinicians’ workload while increasing the intervention’s reach.
On a few occasions, participants would change or lose their primary phones, which led to their being lost to follow-up. The team adapted the enrolment process to include collecting additional contact details to assist with tracing.
Discussion
Fidelity and adaptations
The researchers made 13 adaptations to the 2wT study, six were adaptations to ensure rigour or fidelity and seven were adaptations to ensure the relevance of the study. Conducting research under pragmatic settings enabled both researchers and implementers to better understand how the 2wT intervention is performed in diverse populations and settings. The FRAME-IS helped explore and explain the influence of adaptation to ensure, not reduce, fidelity. There are merits to emphasise both intervention fidelity for internal validity and encouraging acceptance of adaptations that boost external validity for diverse intervention contexts.18 We present three IS-related takeaways from our adaptation exploration.
First, fidelity and adaptation may not be opposing concepts; instead, there should be an exploration of how to achieve a balance between intervention fidelity and adaptation within interventions, allowing for adaptive interventions.16 Chambers and Norton (2016) challenge the assumption that the common trajectory of moving with fidelity from RCT to routine practice is best. Rather, they call for flexibility to consider the positive effect of adaptations made during intervention implementation, an openness for new data that could drive intervention efficiency gains, and recognition that intervention momentum may drive implementation fit even before the evidence base is solidified.19 They continue by proposing the adaptome with a more fluid concept of where, when and how evidence is gathered in support of an intervention’s evolving positive impact, allowing for considerations of adaptations to optimise the intervention and its implementation over time.19 In line with Chambers, adaptations to 2wT implementation were considered throughout the 2wT RCT, responding to needs to improve fit at clinician, clinic and organisational levels, showing awareness of, and openness to, opportunities for improving 2wT implementation over time and context. Indeed, responding to needs as they arose allowed the research team to stay attuned to contextual sensitivities, limitations and opportunities and (re)shape 2wT implementation to be an equitable, reliable and quality health solution.
Second, recognition that fidelity and adaptation can be concurrent, and not oppositional, forces along the pathway from research to practice is also promoted by the value equation developed by von Thiele Schwarz et al.20 In growing recognition that adaptation is necessary to optimise fit, a simplified summary of von Thiele Schwarz’s model suggests that the optimal intervention value (V) for multiple stakeholders is a combined product of an evolving intervention (IN) that reflects its broad context (C) and adaptive implementation strategies (IS) that drive intervention fit. Overall, the value equation suggests that implementers consider a broad view of intervention value, using adaptive implementation strategies to align EBIs with their contexts and be transparent in the processes to guide both internal and external considerations for intervention expansion.
Third, 2wT adaptations appear acceptable and in line with enhanced fidelity. The nature and content of the adaptations in this study, as supported by O’Connor et al, are acceptable adaptations.18 For the adaptation of the 2wT intervention, there was no deviation from the established implementation strategy, characterised by a lack of loosening in structure or departure (‘drift’) within the implementation. Likewise, there was no instance of drifting from the implementation strategy without subsequent return, such as ceasing to offer consultation or halting postoperative follow-ups.21 The adaptations also aimed to reduce the number of patients who were lost to follow-up, improving fidelity or the extent to which 2wT is delivered as intended. Pérez et al support the idea that fidelity and adaptation co-exist and that adaptations can impact the effectiveness of the intervention either positively or negatively.16 They further suggest that it is essential to look systematically at the aspects of an intervention that are being adapted and that implementation research should answer the question of how an adequate fidelity-adaptation balance can be reached.
Fourth, adaptation was necessary in the context of COVID-19 to enhance effectiveness in this real-world setting. The 2wT implementation adaptations largely responded to the changing landscape of COVID-19 restrictions and changes in policy for in-person reviews and reflected waves of client concerns in accessing healthcare services—especially voluntary ones like VMMC. Adapting the strategy to reach more patients with routine VMMC services, and therefore increase the likelihood of 2wT recruitment, was a near-contact challenge during this period. Moreover, 2wT adoption was likely enhanced during this period when healthcare workers and programmes sought methods or options to reduce clinician/patient contact. Furthermore, although clearly not in response to the global pandemic, the timeliness of the 2wT intervention cannot be understated. COVID-19 likely improved the fit between the implementation effort and the needs of those delivering 2wT. Rapid adoption of other virtual follow-up and tacit approvals for telehealth over potential for increased transmission also helped improve fidelity of the intervention. Growing understanding of the underlying cost advantages (less travel), improved equity (more language options for client communication, reduced client transport costs) and improved safety using 2wT likely created an enabling environment for adaptations and their acceptance.
Overall, adaptations to the 2wT intervention occurred at four levels. At the socio-political level, the adaptations were done to address existing national VMMC follow-up mandates such as reducing the number of physical visits to health facilities and replacing them with 2wT. At organisational level, adaptations were implemented to address available staffing or materials. For example, counsellors were trained to enrol and consent clients to relieve the load from clinicians so they can have more time to conduct circumcisions. The implementer level (programme and clinical teams) was for those charged with leading the implementation effort. The team leader of the implementation team could make adaptations or decisions on who does circumcisions or follow-ups or enrolments daily. The patient or recipient level was for the circumcised men who were benefitting as intended from the 2wT intervention. They could enrol in a language that was familiar to them and receive follow-up support via 2wT instead of in-person reviews (8).
Conclusions
The adaptations made to the 2wT pRCT in South Africa and the justifications provided support the idea that adaptations are common and inevitable to account for the needs of specific contexts. The results reflect that the reality of implementing mHealth interventions is a highly dynamic and adaptive process in which adaptations contribute to optimisation for maximum impact. There are merits to arguments for both fidelity and adaptation. We used the FRAME-IS model to reconcile the debate on fidelity and adaptation. Although the model is meant to be a flexible, practical tool for documenting adaptations to the implementation of EBIs, its use has helped to illuminate the pivotal processes and mechanisms by which implementation strategies exert their effects. We suggest that the FRAME-IS be used to help IS move toward a better understanding of the roles of fidelity and adaptation in the implementation process. Adaptations helped the study balance research rigour and relevance to the rural and urban environments where the intervention was implemented. Fidelity and rigour should not be the enemy of adaptation and relevance in closing the gap between research in the laboratory and practice.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants. This Multiple Principal Investigators (MPI) study was approved by the Internal Review Boards of the University of Washington (00009703, PI: Feldacker) and the University of the Witwatersrand, Human Research Ethics Committee (Ethics Reference No: 200204, PI: Setswe). Consent to participate in the study was obtained from all eligible clients. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors would like to thank the Departments of Health of the Gauteng and North West provinces, implementing partner, Right to Care, for enabling the participation of the clinical teams and the CHAPS study implementation team for their dedication and skills in study recruitment, data management and data capture. The authors would also like to thank the Medic research team for their support of the 2wT study implementation and adaptation process.
Footnotes
Contributors CF is the study guarantor. GS and CF conceptualised the study. CF, GS and JP acquired funding and provided the resources for the 2wT RCT and this study. GS and CF supervised this study as principal investigators and ensured adherence to the protocol. GS, CF and JP conducted the 2wT trial from which this study was created. GS, CF, FN and JP deliberated and agreed on the FRAME-IS as a methodology to guide the study. GS and FN collected data on adaptations into the Google Docs, project meeting records and from interviews with field teams and conducted formal analysis of the data. GS, CF, FN, JP, VN and SD contributed to the writing, review and editing of the manuscript.
Funding Research reported in this publication was supported by the National Institute of Nursing Research of the National Institutes of Health under award number 5R01NR019229, ‘Expanding and Scaling Two-Way Texting to Reduce Unnecessary Follow-up and Improve Adverse Event Identification Among Voluntary Medical Male Circumcision Clients in the Republic of South Africa' (principal investigators: CF and GS).
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.