Improving digital partner notification for sexually transmitted infections and HIV through a systematic review and application of the Behaviour Change Wheel approach

Objectives

To identify barriers and facilitators to using digital PN interventions in the published literature; to use a behavioural science approach to develop evidence based and theoretically informed recommendations.

Protocol registration

A study protocol with detailed methodology was pre-registered on PROSPERO.¹⁵ For justisfication of one deviation from this protocol, see Supplementary file S1 in Supplementary material.

Search strategy

A systematic search of articles published between January 2010 and July 2023 across eight research databases (CINAHL Plus, Cochrane Library, Embase, Medline, NHS Evidence, PsycINFO, Scopus, and Web of Science) was conducted (by SB), targeting the combination of three topics: (1) digital health; (2) STIs/HIV; and (3) PN/contact tracing. For exact terms, see Table S1 in Supplementary file.

Selection process

The results of the search were exported to EndNote ver. X9, and duplicates were removed. First, the titles, keywords and abstracts were screened and checked against the eligibility criteria in Table 1 (by SB and JS). Second, full texts were assessed against the criteria in Table 1 (see Fig. 1 for additional details). Uncertainty about studies meeting the inclusion criteria was discussed with JM and PF until a consensus was reached. The final included studies were categorised into: (1) barriers and facilitators for online notification of partners; and (2) barriers and facilitators for online management of partners (such as their testing and treatment following index patient testing positive).

Fig. 1.

PRISMA flow diagram of article search and selection.

Quality of contributing studies

The Mixed Methods Appraisal Tool ver. 2018 (MMAT)¹⁶ was used (by CW, JG, SB, PF) to assess the quality and appropriateness of the study methods.

Data extraction

Data analysis

Research Question 1: to generate evidence concerning the main barriers and facilitators to using digital PN interventions, verbatim barriers and facilitators were synthesised based on their similarity (by JM and SB). For example, ‘Did not have contact details’, ‘Tried to contact partner with no success’, ‘Not having email addresses for partners’ were synthesised into ‘Problems contacting partners’. Synthesised barriers and facilitators were then de-duplicated and recorded in a separate spreadsheet. This was reviewed by PF and disagreements were resolved by discussion until consensus was reached.

Extracting only data relating to digital PN, we identified 98 barriers and 54 facilitators. Many more individual barriers to PN were extracted than facilitators, as one paper reported only barriers.¹⁷ The individual barriers and facilitators were then synthesised to generate 30 high-level barriers and 35 high-level facilitators, of which 18 and 17, respectively, were selected (using selection criteria, see Supplementary file S2 in Supplementary file) to be taken forward to Behaviour Change Wheel (BCW) analysis to develop recommendations for future digital PN interventions. See Fig. 2 for an overview.

Fig. 2.

Selection of barriers and facilitators.

Research Question 2: the BCW approach,¹⁸ which incorporates the theoretical domains framework (TDF),¹⁹ was then used to generate theoretically informed recommendations on how to improve the use of digital PN interventions. The BCW approach is a way of systematically developing theoretically informed intervention content from a detailed understanding of the behavioural drivers of a particular behaviour¹⁸); in this case, using a digital PN intervention. Here, the de-duplicated, synthesised barriers and facilitators were critically appraised (by WK, JM and PF) for selection for BCW analysis using selection criteria (see Supplementary file S2 in Supplementary file).

Selected synthesised barriers and facilitators were coded (by JM and SB) with appropriate TDF¹⁹ domains. This coding was reviewed by WK. Disagreements were resolved by discussion until consensus was reached. The TDF brings together a plethora of behaviour change theories to classify drivers of behaviour into 14 domains (e.g. ‘Beliefs about capability’, ‘Knowledge’, or ‘Social influences’). It enables the theorisation of the drivers of behaviour in a pragmatic and useful way that allows subsequent work to systematically generate recommendations for future intervention content.

To generate recommendations, WK matched the theorised barriers and facilitators to intervention functions, using the BCW,¹⁸ and Behaviour Change Techniques (BCTs), using the Behaviour Change Technique Taxonomy ver. 1 (BCTT ver. 1; a set of 93 unique active ingredients of behaviour change interventions²⁰). BCTs were then operationalised (by WK) considering Acceptability, Practicability, Effectiveness, Affordability, Side-effects and Equity (APEASE criteria).¹⁸ These operationalised recommendations were then reviewed by JM. Disagreements were resolved by discussion until consensus was reached.

Finally, considering the APEASE criteria once more¹⁸ recommendations were further assessed and reviewed by a wider multidisciplinary team (JS, CE, JG) including sexual health clinicians, a medical sociologist and a principal health advisor (sexual healthcare professional responsible for PN within a large geographical area).

All BCW coders completed the UCL Centre for Behaviour Change BCTT Online Training (http://www.bct-taxonomy.com/). All stages of analysis were additionally reviewed by PF, a senior behavioural scientist with considerable expertise in using these tools across a range of diverse topics.

Study selection

The search identified 1548 studies, 1124 remained after de-duplication. After screening titles and abstracts, 112 studies were assessed for full-text eligibility, following which eight studies remained. For an overview of the study selection process, see the PRISMA flow diagram in Fig. 1.

Study characteristics

A summary of the article characteristics is in Table 2. Populations addressed included HIV pre-exposure prophylaxis (PrEP) users,²¹ cisgender men and transwomen,²² men who have sex with men (MSM),^17,23,24 people recently diagnosed with chlamydia^9,25 and the general population.²⁶ The eight articles used a range of qualitative and quantitative approaches. Five studies related to hypothetical digital PN interventions;^{17,21–23,26} only one reported solely on an intervention that had been delivered²⁵ and two concerned both real and hypothetical interventions.^9,24

Quality appraisal

Most articles satisfied all, or the majority, of MMAT criteria, as detailed in Table S2 in Supplementary file. As such, none were excluded from further analysis. All eight studies focused on PN (chlamydia, n = 2; HIV, n = 1, syphilis, n = 1; HIV and syphilis, n = 1; and STIs in general, n = 3) and none focused on partner management. Understanding the barriers and facilitators to using digital PN interventions was complicated because PN was addressed in different ways in different studies. In three studies, it was not possible to distinguish whether barriers and facilitators to PN related specifically to digital or non-digital modalities.^9,17,26 In three other studies, it was possible to distinguish barriers and facilitators related specifically to digital or non-digital modalities.^23–25 Finally, two studies reported only barriers and facilitators to digital PN interventions^21,22 (Table 2).

Main findings

Few studies have focused on digital PN interventions and there are considerable gaps in the literature around partner management, HIV, and non-hypothetical studies within digital PN. Of the eight studies identified as relevant, all focused solely on notification and not on partner management (i.e. testing and treatment), and are consistent with the narrative review by Golden et. al.⁶ on partner services for HIV/STIs in high-income nations. Here, most studies investigated bacterial STIs, with only two examining HIV; and there was a large proportion of hypothetical preference studies. There was also a high degree of heterogeneity in relation to study participants and STIs addressed. Further, the reporting of specific barriers and facilitators to using digital PN interventions was poor, with many studies amalgamating barriers and facilitators across both digital and non-digital modalities (e.g. face-to-face, phone calls, SMS, email).

There were multiple, diverse, re-occurring barriers and facilitators to using digital PN interventions with considerably more barriers than facilitators reported in the literature. Key barriers and facilitators to using digital PN interventions included factors associated with index patients’ levels of knowledge of PN itself or of STIs and HIV, or beliefs about the health risks associated with STIs/HIV and the imagined consequences of PN; and factors associated with the technological aspects of digital PN interventions such as concerns about safeguards and imagined links into future face-to-face sexual health care. These barriers may help to explain why acceptability of, intention to use, and uptake of digital PN interventions is often low and highly varied.^4,6,14 Many of these important barriers and facilitators were amenable to change and were reflected in the wide-ranging, multi-levelled recommendations our behavioural analysis generated.

Our evidence and theory-based recommendations include that digital PN interventions should enable and empower index patients to make choices about the kinds of PN that work for them and their partner(s). This means ensuring that digital PN interventions educate and model how to notify partners and also enable access to psychosocial support when doing so if it is needed. Digital PN interventions need to be as simple, convenient and usable as possible, complying with explicit and implicit ‘industry’ standards (e.g. the quality of visual content) and where possible, integrating into other existing digital systems (e.g. dating apps). Our recommendations also emphasise that unless the social and cultural context of PN is addressed, future digital PN interventions are unlikely to gain traction or be effective in the long term. Reinforcing messages about the positive aspects of PN (for individuals, partners and communities) and our collective responsibility to enact it will be crucial going forwards. If these upstream psychosocial and sociocultural issues are not addressed, downstream technological innovation has little chance of increasing the use of digital PN interventions and improving PN outcomes. These recommendations are designed to directly overcome the barriers and enhance facilitators to engaging with digital PN interventions identified in the literature, to improve the uptake of digital PN interventions.

Implications for practice

Using these recommendations could considerably enhance digital PN interventions. However, the multi-levelled nature of the recommendations means that delivering them will necessarily involve working in partnership with a variety of stakeholders with diverse skill sets and expertise. The recommendations are relatively systemic, speaking simultaneously to both the upstream (e.g. social context) and downstream (e.g. digitally mediated prompts) determinants of using digital PN interventions. A piecemeal approach where any one recommendation is enacted is unlikely to gain traction or deliver substantial change in PN outcomes. Given the heterogeneity of contexts in which the contributing studies were conducted, we suggest that our recommendations should be explored locally and with relevant stakeholders (e.g. industry partners, diverse patient groups, healthcare professionals and CBOs). This should ensure they are operationalised in coherent, culturally appropriate, locally tailored and co-produced ways.

Strengths and limitations of the current study

The paper has many important strengths. Throughout the study, we used multiple experts to code, review and audit data and analysis (e.g. quality appraisal, data extraction, and coding using the TDF, Intervention Functions, and BCTT ver. 1). We also used a wide range of disciplinary, and professional perspectives, to ensure the translation of our behavioural science findings into practical recommendations. We drew upon well established, theoretically-based ways of developing recommendations¹⁸ and have provided an auditable account of our analytic steps.

However, we restricted inclusion to articles in the English language, which may have excluded some relevant studies. In addition, the specifications of barrier and facilitators statements from a particularly heterogeneous set of original studies required some interpretative work and we did not weight barriers and facilitators according to the original study design or quality. Equally, although the studies identified were conducted across five different countries, we combined their findings into one data set without any sub-population analyses. It is feasible that important cultural or societal differences that may be important are overlooked in our synthesis of barriers and facilitators.²⁸ Additionally, the tools used to develop our recommendations for future intervention content are undergoing revision, including the content of the BCTTv1 and the way in which theoretical mechanisms can be theorised (https://theoryandtechniquetool.humanbehaviourchange.org/).

Future research

As new digital PN interventions are developed, it is crucial to compile a detailed evidence-base to evaluate their efficacy. This would include information about their content, the barriers and facilitators to their use, their population reach, the types of sexual partner for which they are used, and their impact on health inequalities. Furthermore, we should aim to understand the effects of digital PN interventions on non-digital sexual health services (e.g. changes in the pattern of their use and provision). Several of the behavioural science tools used here may be very helpful in achieving these evaluative goals.¹⁸ Research using high quality frameworks²⁹ is also needed to deliver cumulative insights into the implementation of digital PN interventions across diverse settings and nations as well as a deep understanding of the effects of such interventions, both positive and negative. Studies of new digital PN interventions should also develop and evaluate comprehensive partner management by the inclusion of testing and treatment, and opportunities for prevention, as appropriate. Future research should also address any similarities and differences in approaches needed for the different infections for which we use PN, notably curable bacterial STIs versus long-term treatable infections such as HIV. PN for HIV, for example, while poorly addressed within the literature reviewed here, presents particular benefits (e.g. the elimination of HIV transmission) yet simultaneously particular challenges (e.g. the criminalisation of HIV transmission, psychosocial impacts of potential HIV exposure).