Article Text
PDF
Abstract
Introduction HIV care engagement is lower among black sexual-minority men relative to other racial/ethnic groups of sexual-minority men. Being in a primary relationship is generally associated with more successful HIV care engagement across various populations. However, among black sexual-minority men, the association between primary relationship status and HIV-related outcomes is inconsistent across the HIV care continuum. Given the ubiquity of mobile technology access and use among racial/ethnic minority communities, leveraging mobile technology for HIV care engagement appears a promising intervention strategy. This paper outlines the protocol of the LetSync study, a pilot randomised controlled trial of a mobile health app intervention developed using the Framework of Dyadic HIV Care Engagement to improve care-engagement outcomes among black sexual-minority male couples living with HIV.
Methods and analysis Eighty black sexual-minority men in couples (n=160) will be enrolled to pilot test the LetSync app. At least one member of each dyad must be both HIV-positive and self-identify as black/African-American. Couples will be randomised to either a waitlist-control arm or an intervention that uses relationship-based approach to improve HIV care engagement. We will assess feasibility and acceptability of trial procedures and intervention protocols based on predefined metrics of feasibility and acceptability. Execution of the study will yield the opportunity to conduct analyses to test the measurement and analysis protocol on antiretroviral therapy adherence by comparing the intervention and waitlist-control arms on self-reported and biological (hair sample) measures of adherence.
Ethics and dissemination Study staff will obtain electronic consent from all participants. This study has been approved by the University of California (UCSF) Institutional Review Board. Study staff will work with the Community Advisory Board at the UCSF Center for AIDS Prevention Studies Board to disseminate results to participants and the community via open discussions, presentations, journal publications and/or social media.
Trial registration number NCT04951544.
- HIV & AIDS
- public health
- telemedicine
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
Mobile health interventions have traditionally focused on a single user’s experience and outcomes, which LetSync will challenge by harnessing couples’ resilience and ability to problem-solve together, both of which impact dyadic coordination and, in return, can improve HIV care engagement.
Involving participants in the app development process can allow for higher chance of acceptability of future iterations.
The remote nature of our study breaks down barriers to participation such as travel, time and expenses.
This intervention does not allow users to directly engage with the healthcare system.
Due to this being a couples’ study, it is possible that couples can break up during study participation which can impact feasibility results of the app.
Introduction
Black sexual-minority men (ie, gay, bisexual and other men who have sex with men (MSM)) account for 26% of 37 968 new HIV diagnoses in the USA in 2018 and 37% of new diagnoses among all MSM.1 2 Black MSM also show the least favourable HIV care engagement outcomes (ie, testing, linkage to and retention in HIV care, viral suppression) relative to other racial/ethnic groups of MSM.3 4 Suboptimal adherence to antiretroviral therapy (ART) can lead to transmission and detrimental clinical outcomes.5 6 Based on current data, it is estimated that one in two black MSM will be diagnosed with HIV during their lifetime.7 8
National estimates show that a third to a half of black MSM with HIV are in a primary relationship,9–11 which is associated with favourable outcomes in healthcare engagement via social support pathways.12–15 Dyadic approaches are part of a multilevel intervention approach; yet, they remain poorly understood among black MSM.16 Emergent evidence show that black MSM in couples help each other engage in HIV care and treatment but that many do so inconsistently.17 18 Additional characteristics of the dyad may moderate the effect of a primary relationship on HIV care engagement.14 15 19–21 For example, black couples with HIV may engage in joint problem-solving, a collaborative problem-focused approach to coping with stress, and dyadic coordination, or the synchronisation of activities and behaviours necessary in HIV care and treatment.18 22
With >75% of the US adult population owning smartphones,23 mobile health (mHealth) has emerged as a promising tool in healthcare including HIV prevention, care, and management efforts.24–26 Although mHealth has been shown to be feasible, acceptable and effective among black MSM,26–35 no dyadic mHealth interventions exist for this population even as black MSM face many unique barriers to care and treatment.36 Compared with white MSM, black MSM are 20% less likely to be linked to, engaged and retained in HIV care due to social and structural inequities such as racial discrimination,37 access to ART,38 39 food and housing insecurity40 41 and overcriminalisation and policing of black communities.39 Low retention rates can also be explained by inequities in the healthcare system, such as experiencing stigma and shame from healthcare providers.42 Black sexual-minority couples show great interest in using a couples-based app to facilitate joint problem-solving to coordinate care and treatment activities, and provided ideas for the app features they want.22 36 In contexts where same-sex relationships are highly stigmatised, black sexual-minority couples may appreciate an app that focuses on their primary romantic relationships.
Guided by the Framework of Dyadic HIV Care Engagement (figure 1),18 22 initial designs were created for a dyadic mHealth application (app) intervention called LetSync, for let’s synchronise’, to target dyadic coordination and joint problem-solving skills to improve retention in care and ART adherence. LetSync aims to facilitate among couples the dyadic coordination and joint problem-solving necessary for optimal engagement in HIV care among black MSM.
Framework of Dyadic HIV Care Engagement.
This protocol paper describes the pilot randomised waitlist-controlled trial to assess the feasibility and acceptability of the study protocols and procedures, assess the feasibility and acceptability of using LetSync and test measurement and analysis protocols on preliminary data of app use on ART adherence. Fully developing a couples-based mHealth intervention will require that we translate findings to inform LetSync designs and iteratively develop, refine and pilot-test prototypes for a large-scale, future efficacy trial.
Methods and analysis
Setting and participants
LetSync is a single-site, pilot randomised waitlist-controlled trial with the primary goal of assessing feasibility and acceptability of the mobile app, LetSync, among 80 black sexual-minority couples (n=160) living in the USA. The sample size was chosen to be adequate to gauge feasibility and acceptability while remaining feasible for a pilot. Participants will be randomised to immediately begin the intervention or wait 6 months. A waitlist-control design (figure 2) will allow us to evaluate two versions of LetSync, a later version iteratively refined based on feedback about the previous version.43 LetSync will be developed by a third-party app developer to be compatible with both iOS and Android.
Timeline of LetSync intervention.
Participation in the study will last 14 months, with assessments conducted at baseline, 6, 8 and 14 months. We will collect feasibility and acceptability data, as well as preliminary data on ART adherence as measured by antiretroviral (ARV) concentrations in hair. Participants will consent to the study and complete an initial baseline survey online. Study staff will communicate with participants through text, email, phone and Zoom. The University of California, San Francisco (UCSF) Institutional Review Board (IRB) has reviewed and approved this study.
Eligibility
Black MSM who are at least 18 years of age, living with HIV in the USA and in a primary relationship with another man for at least 2 months will be eligible to participate. A primary relationship will be defined as a commitment to someone over and above anyone else that has lasted at least 3 months and includes a sexual relationship.44
At least one member of the couple must be African-American/black and living with HIV (Index) who is either not on ART or is <100% ART adherent as assessed via a three-item adherence measure.45 Their primary partner can be of any race or ethnicity, and any HIV status. Among couples where both partners meet eligibility as an Index, one will be chosen at random to be the Index. Both members of the couple must own or have access to a smartphone.
We will exclude individuals who (1) report fear of intimate partner violence resulting from participation as assessed at screening,46 47 (2) are unwilling or unable to disclose HIV status to primary partner or (3) are presenting evidence of severe cognitive impairment that would prevent comprehension of study procedures assessed during informed consent.
Patient and public involvement
Prior to the design of LetSync, investigators conducted formative research with black sexual-minority men in the San Francisco Bay Area. Black sexual-minority couples showed strong mHealth preferences and interest in using a mobile app to facilitate joint problem-solving to achieve optimal HIV care engagement.36 We will also assemble a Community Advisory Board of black sexual-minority couples to obtain feedback on LetSync prototypes and develop LetSync V.1.0.
Study procedures
Recruitment
We will use a multipronged recruitment approach that includes in-person and virtual engagement. In addition to the San Francisco Bay Area, we will prioritise recruiting from US cities with the highest prevalence of HIV among black MSM (eg, Atlanta, Georgia; Los Angeles, California; Washington, District of Columbia; Houston, Texas). We will attend virtual events hosted by community-based organisations serving black/African-American and/or sexual-minority communities impacted by HIV/AIDS, placing targeted online advertisements on social media (eg, Facebook) and asking clinics that serve black MSM with HIV to distribute flyers. We will also recruit from within UCSF clinics via the UCSF Recruitment Letter Services. We will also contact participants of other UCSF studies who gave consent to be contacted.
Screening
Study staff will provide a brief overview of the study to prospective participants, answer any questions and complete an eligibility screening over the telephone. Targeted online advertisements will link to an online prescreener that interested individuals can take to see if they qualify. Only those who are potentially eligible (based on screener responses) will be contacted by study staff. Ineligible responses will be recorded along with the reasons why (eg, not living with HIV, not in relationship with a man).
Consent/Enrolment
If found to be eligible on screening, individuals will be sent an informed consent form online. Eligible individuals will be instructed to read the consent form in full and ask any questions they may have prior to giving consent. Study staff will be available to respond to any questions or concerns and to ensure comprehension.
Intervention
Randomisation
After obtaining informed consent from both members of the dyad, we will randomise couples to the intervention or waitlist-control groups using a randomisation-plan generator.
Intervention content: LetSync
To enhance the couples’ capacity for HIV care engagement, LetSync was designed based on problem-solving therapy. Problem-solving therapy consists of distinct steps to help identify problems one may have, possible solutions to follow and the advantages and disadvantages to each.48 Problem-solving therapy has shown to be effective in other mHealth interventions (eg, iProblemSolve, a goal-setting app targeting individuals).49
The defining feature of LetSync is ‘My Action Plan’, which will guide the Index to arrive at a tailored action plan that addresses a component of HIV care engagement. The Index will identify current HIV care engagement and general health-related issues, choose strategies for addressing the issues (strategies already extant in the app plus new strategies the user can add) and evaluate those strategies in terms of likelihood of implementation. The Action Plan, which is composed of the strategies the user identified as most likely to be implemented, can then be shared with their partner through the app. The Action Plan will contain features to encourage the Index and their partner to engage in joint problem-solving and dyadic coordination. For example, partners will be prompted to make suggestions to Action Plans, download the Action Plans into their own mobile calendars, view goals and progress, coordinate activities around goals and appointments and share encouragements.
Timeline
The study timeline will be split into four time points (T): T1 (baseline), T2 (6 months), T3 (8 months) and T4 (14 months) (figure 2).
At T1, participants in the intervention and waitlist-control arms will receive hair-sample collection kits in the mail with necessary supplies, an electronic link to an instructional video and a pre-paid envelope for returning samples.50 51 Participants in the intervention arm will receive an electronic link to the baseline survey and will be scheduled their first study visit, which will occur via videoconference (eg, Zoom). At the first study visit, study staff will give an overview of the study, answer any questions and assist the participant in installing the app on their phone and provide necessary instructions for app use. The intervention group will use LetSync V.1.0 for 6 months.
At all three subsequent time points (T2–T4), participants in both arms will receive a text or email informing them that the next study assessment is due, along with the link to complete the assessment. Simultaneously, we will mail all participants a hair-sample kit.
Between T1 and T2, we will collect data on acceptability and feasibility and use this to revise LetSync V.1.0 and update it to LetSync V.2.0.
At T3, participants in the waitlist-control arm will attend a videoconference during which study staff will offer an overview of the study, answer any questions and assist the participant in installing and using LetSync V.2.0. Meanwhile, the participants in the intervention arm will continue to use LetSync V.1.0.
At T4, we will conduct virtual exit interviews with participants from both arms over the phone or via videoconference. During exit interviews, we will ask for feedback about the randomisation procedures to inform future randomised controlled trial (RCT) procedures. Interviews will be audio-recorded for transcription and data analyses.
Incentives
Participants will receive a US$50 cash card, payment through a cash app or reloadable debit card on completing each survey, an additional US$50 on receipt of hair samples at T1, T2, T3 and T4, and US$30 for completing the exit interview at T4. Altogether, each member of the couple can receive up to $430.
Outcomes
Primary outcome
The primary outcome is ART adherence. We will measure ARV levels in hair samples across all four time points. Additionally, assessments at each time point will measure engagement in HIV care using a comprehensive behavioural composite of engagement in HIV care.52
Feasibility of app/intervention
At T2 and T4, we will assess feasibility based on metrics in table 1 and metadata (eg, number of times the Action Plan was shared between partners, frequency of encouraging messages exchanged). We will code and tabulate these interactions to analyse dyadic HIV care engagement by, for example, the volume and sequence of activities planned. Participants can report glitches and other issues at any time through a reporting feature in the app or study website, or by contacting the study staff. All reports of issues will be tabulated.
Metrics and thresholds to assess feasibility of the LetSync app
We will monitor rates of recruitment and effort (eg, number of staff hours), number of screenings, proportion eligible and agreed to enrol, number of participants who withdraw after being randomised to condition and reason(s) for withdrawal and the number of participants who complete each time point. We will record the number of rescheduled, cancelled and missed visits to inform estimation of future staffing needs. Using call/time logs, we will record the frequency and mode of contact with participants, when, and for how long. During remote visits, staff will complete a checklist and take notes on study proceedings such as the procedures implemented, amount of time spent and participants’ reactions. These data will inform modifications to the intervention and protocols of a subsequent, full-scale efficacy trial.
We will compare HIV clinical outcomes and dyadic capacity measures between the two arms in exploratory analyses. We will evaluate feasibility and acceptability of LetSync V.2.0 in the waitlist-control arm and evaluate persistent use of LetSync V.1.0 over 14 months in the intervention arm.
Feasibility of hair sample collection
Feasibility of hair collection will be evaluated by: (1) the number of samples per participant received by the study, (2) the time difference between when remote hair samples were due versus when samples were received by the study and (3) rates of verifiable ARV results. Staff will document when hair collection kits were sent and received.
Acceptability
Acceptability will be evaluated via a measure of app usability,53 and self-reported satisfaction with security and privacy of app use, study procedures and design and remote hair collection. During the exit interview, we will ask participants about what was convenient/easy versus inconvenient/difficult regarding remote study participation. The threshold for acceptability will be 80% of participants reporting being satisfied with the app content and delivery format. Table 2 contains examples of items used to capture each measure.
Items and measures to assess acceptability of the LetSync app
Throughout the intervention, we will contact participants in both arms monthly via text, call and/or email. We will check in about their experiences of using the app, along with troubleshooting app-related issues and sending in hair samples.
Data collection and analysis
Quantitative data collection and analysis
Assessments at baseline, 6 months, 8 months and 14 months will be administered online and will measure HIV care engagement using a comprehensive behavioural composite of engagement in HIV care52; engagement and retention in care using the Index of Engagement in HIV Care (eg, “How well do you follow through on your HIV care when things in your life get tough?”)54; and self-reported ART adherence (eg, “In the last 30 days, on how many days did you miss at least one dose of any of your medication?”)55 and viral suppression (eg, “Was your last viral load detectable or undetectable?”). Guided by our conceptual framework (figure 1),22 we will measure dyadic capacity using the Dyadic Coping Inventory,56 Couple Health Support, Partner Support for HIV Treatment57 and relationship factors using the Power Imbalance in Couples Scale,58 and the Couple Sexual Satisfaction Scale (Conroy AA, Development and Validation of the Couple Sexual Satisfaction Scale for HIV and Sexual Health Research, Under Review). We will also assess individual-level factors as indicated by our conceptual framework, including the HIV Stigma Scale.59
Frequency tables will be generated for all clinical outcomes. One-way frequency tables will be generated for the number of rescheduled, cancelled and missed visits. Relative frequencies will be calculated for the number of participants enrolled in the study, those who were eligible in general and lost to follow-up. We will also tabulate and summarise acceptability outcomes in one-way frequency tables.
We will fit linear mixed models to continuous outcomes (eg, ARV levels in hair) and fit generalised linear mixed models to discrete (eg, viral suppression) and non-normally distributed continuous outcomes (eg, self-reported ART adherence) to model outcome data. These analyses will include couple sero-status (sero-concordant HIV-positive vs sero-discordant) as a covariate as required by the stratified randomised design.60 61 Following guidelines in the literature62 63 and from NIH,64 hypothesis testing will de-emphasised. Instead, we will perform these analyses to ensure that all measures and procedures are well established to perform a subsequent efficacy trial.
Qualitative data collection and analysis
At T4, staff will conduct remote exit interviews with all participants. Exit interviews will explore participants’ experiences with the study protocol and procedures. Interviews will be audio-recorded and professionally transcribed.
We will read all individual transcripts and develop a codebook based on the interview guides, our theoretical framework and emergent themes. To establish intercoder agreement, a primary coder will apply codes based to a subset of transcripts to test and revise the codebook. A secondary analyst will apply the revised set of codes on a random subset of transcripts. Discrepancies in coding will be discussed by the team until an agreement is reached.
Power analyses
We estimated minimum detectable effect sizes (MDEs) for the assessments of feasibility and acceptability proposed to address the pilot RCT. We anticipate 80 couples (40 seroconcordant-positive and 40 serodiscordant per condition) at the beginning of the study and 64 couples at T4 following 20% estimated attrition. The effective sample size (ESS) will depend on the unit of analysis (couple vs individual), which participants are included in the analysis and when the outcome is measured. For instance, the enrolment proportion to assess feasibility is a couple-level variable measured at the outset of the study. Assuming α=0.05, power=0.80 and 70% enrolment for 114 couples contacted to yield 80 couples (70% of 114), the width of the CI for single enrolment proportions is 19% (standardised distance to the limit: 0.20). In contrast, acceptability scores will be measured at the individual level at the study end point among participants in each condition.
We also performed power analyses for proposed outcome analyses in order to supply additional information. For individual-level outcomes, the ESS will depend on the degree of within couple correlation of responses, ρ, within couples. We set ρ based on prior dyadic research in which the average within-couple correlation of virological control measurements was ρ=0.23. Accordingly, we lowered the ESS inputted for the power analyses to be ESS=N/DEFF, where N is the end point sample size and DEFF is the design effect or variance inflation attributable to using correlated data. DEFF is computed as 1+(M−1)×ρ, where M is the number of participants per dyad (ie, 2). Therefore, DEFF=1+(2−1)×0.23=1.23, so ESS=80×0.80=64/1.23=52. Under these assumptions, distance from the observed mean to the confidence limit is estimated to be 0.28. For longitudinal analyses to evaluate ART adherence, outcomes will be measured at the individual level at every time point among HIV+ participants in both arms. An 80% retention rate means 20×0.80=16 seroconcordant-positive couples yielding 32 HIV+ participants, where ESS=32/1.23=26 plus 20×0.80=16 serodiscordant couples yielding 16 HIV+ participants for a total end point study sample of 42 per arm. Assuming α=0.05, power=0.80 and four time points with r=0.30 correlation between repeated measures (in Dr Johnson’s study, the average within-subject r’s for ART adherence and viral suppression were 0.24 and 0.28, respectively), the minimum detectable standardised mean differences for continuous outcomes is 0.421. For binary outcomes, using the same inputs as above plus small, medium and large base rates of 10%, 30% and 50%, respectively, raw proportion differences range from 16.1% to 20.5% (standardised difference=0.422−0.429). H1–H3 will be directly tested by contrasts derived from the longitudinal analytic models. For H1 and H3, we estimated the MDEs of those contrasts by reassessing the power of the longitudinal analyses with only two time points. The resulting effect sizes ranged from 0.493 to 0.503, which are medium standardised effects. For H2, MDEs for a non-zero longitudinal change in a group mean or proportion range from 0.407 to 0.470, which are small to medium standardised effects. As noted previously, hypothesis testing will be de-emphasised in this pilot feasibility and acceptability study.
Discussion
This paper describes the protocol for a randomised waitlist-controlled pilot of a dyadic app intervention, LetSync, focused on black sexual-minority couples living with HIV. Barriers to HIV care for black MSM are multilevel, often at the social (eg, HIV stigma) and structural (eg, transportation) levels, while extant interventions target barriers at the individual level. LetSync addresses this gap by targeting, at the dyadic level, black MSM couple dynamics, emphasising the roles of dyadic coordination and joint problem-solving in improving HIV care engagement.
Although black MSM-centred mHealth interventions exist in general,32 65 there is a paucity of couples-based mHealth studies for this population despite the demonstrated power of dyadic coordination in care, and couples facing many unique barriers to care and treatment.
A search in the literature yielded only one couples-based mHealth study for black MSM. In 2010, an existing evidence-based intervention originally developed for heterosexual couples was adapted for black MSM to reduce sexually transmitted infections (STIs; including HIV and other STIs) and drug use outcomes. This adaptation was recently piloted with 34 MSM dyads with promising results.66 67 Of the seven couple-based HIV studies that have been conducted since the start of the HIV epidemic, only three have included MSM in general, and none included black MSM .66
Our study addresses the lack of couples-based interventions for black MSM in several innovative ways. It seeks to harness couples’ resilience and ability to synchronise problem-solving approaches, both of which are likely to impact dyadic coordination and joint problem-solving—thus improving HIV care engagement.14 It is also informed by our theoretical framework, the Framework of Dyadic HIV Care Engagement, which is formulated by preliminary and existing research. Rather than focus on single users’ experiences and outcomes, as is the case for most traditional mHealth designs (including HIV prevention),34 68 the design of LetSync targets the dyad where each user’s outcomes are dependent on the joint, collaborative, synchronised behaviours of both users. The dyadic level is often missing in multilevel HIV prevention efforts, but retention in care and ART adherence often occur in the dyadic context for black sexual-minority couples.14 Lastly, our study is the first of its kind to include the use of remote hair collection to measure ART adherence. Hair concentrations of ARVs are stronger predictors of virological suppression than self-reported adherence or plasma ARV levels in large cohort studies of patients with HIV.69 Self-collection of hair samples at home reduces travel time and expenses, and assessing our primary outcome via remote collection of hair is congruent with the mobile nature of the intervention.
There are several challenges to this study. Suboptimal app engagement poses a challenge in mHealth data collection. To optimise app engagement, we will programme pop-up reminders to appear on a weekly basis if the app has not been opened. We will assess the feasibility and acceptability of this feature during exit interviews. To minimise participant attrition, which is intrinsic to longitudinal designs, we will collect at least three methods of personal contact such as social media handles and additional phone numbers. We will also maintain regular contact with participants by sending reminders about virtual check-ins and sending in hair samples and asking about any app-related issues. Lastly, addressing break-ups is necessary as our study involves couples. If break-up occurs between screening and randomisation, the couple will become ineligible and referrals for support will be offered to both participants. If break-up occurs after randomisation, participants may still take part in the remaining data collection time points as scheduled, and the breakup will be noted in the retention and tracking study databases.
This paper documents the protocol for the LetSync study, which was designed to help couples work together to improve HIV-related outcomes. While the number of HIV-centred mHealth interventions have proliferated in recent years, very few exist that focus on black MSM in couples. mHealth for dyadic HIV care engagement holds promise in being cost-efficient and transcending common barriers to intervention and care, which our study aims to demonstrate. Findings from the proposed research are needed for a subsequent large-scale, RCT to test the efficacy of LetSync in improving HIV care and treatment outcomes among black MSM. These findings may inform future studies and protocols for other chronic conditions where the dyad is an important unit of intervention.
Ethics and dissemination
This study has been approved by the University of California (UCSF) Institutional Review Board (#15-18042). Informed consent will be obtained electronically (eg, via Qualtrics). Participants will be informed that their participation in the study is voluntary and that they may decline to participate for any reason without any negative consequences. Referrals for emotional support and mental health will be available.
Results of the pilot RCT will be disseminated through peer-reviewed publications, conferences and presentations and reports to participants and stakeholders. We will also hold Town Halls with the UCSF Center for AIDS Prevention Studies and symposia with community-based organisations that serve people living with HIV.
Ethics statements
Patient consent for publication
Acknowledgments
The authors would like to thank Sage Bionetworks for granting LetSync the Digital Health Catalyst Award.
References
Footnotes
Contributors JT designed the study, obtained funding, provided leadership in the execution of the study and contributed to revising the manuscript. TBN, LP, PS, EA and SK contributed to study conception. TBN and LP also contributed to trial design. The paper was drafted by HCK, and all authors, including DJB and RWW, read and approved the final manuscript.
Funding This research was supported by a grant from the National Institute of Mental Health R01MH118967 (JT).
Disclaimer The contents of this publication are solely the responsibility of the authors and do not represent the official views of the National Institutes of Health.
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 and analysis' section for further details.
Provenance and peer review Not commissioned; peer reviewed for ethical and funding approval prior to submission.