Sexual risk reduction interventions for patients attending sexual health clinics: a mixed-methods feasibility study

Population

Young people aged 16–25 years or MSM groups.

Study design

Randomised controlled trials of brief interventions for reducing sexual risk that could be implemented in SH clinics in the UK were considered. Within this definition, interventions were divided into those that involved one session of < 30 minutes and those that involved between two and six 30-minute sessions. The analysis was restricted to RCTs to increase the likelihood that evidence of effects would be robust.

Intervention

Interventions in the following settings were included: social networking sites, primary care, emergency care settings, community treatment settings (e.g. GUM and SH clinics) and educational settings (including schools and colleges). Waiting-room, self-delivered, clinician-delivered and digital interventions were included.

Outcome

Biological (e.g. STI incidence), sexual behaviours (e.g. condom use or number of sexual partners), testing (e.g. home-based, appointment booking and clinic visits). Outcomes were measured at a minimum of 60 days’ follow-up. (Note that we included studies that did not show statistically significant effects on primary or secondary outcomes.)

Setting

The UK and other high-income settings.

Population

Studies focused exclusively on victims of sexual or domestic abuse or violence; those in prison, psychiatric facilities or nursing homes; or individuals/communities with no fixed address.

Study design

Studies without a randomised control group; animal models; narrative reviews, editorials and opinions; non-English-language papers; and reports published as meeting abstracts only, or when insufficient methodological details were reported to allow critical appraisal of study quality. Systematic literature reviews were not included in the review, but their reference lists were searched for relevant RCTs. Studies published prior to 2000 were excluded.

Intervention

All non-tailored interventions (e.g. social marketing campaigns providing free access to condoms), interventions for adherence to medical treatment (e.g. adherence to antiretroviral therapies), interventions for couples and family/parent-centred interventions.

Outcome

Only psychological changes evaluated, such as attitude change, or a follow-up of < 60 days.

Setting

Low- or middle-income settings.

Surveillance (Genitourinary Medicine Clinic Activity Data Set v2)

This is a national mandatory reporting data set for England: all level 2 and level 3 SH services are required to submit their SH patient episodes to PHE. It covers an estimated 600 services (including SH services) and reports STI diagnoses. The data set contains 12 variables (Table 4): demographics, attendance information and any episode activity and diagnoses. Data from the reporting periods quarter 1 (2013) to quarter 3 (2016) were used. This data set is referred to as ‘v2’ throughout.

Enhanced surveillance (Genitourinary Medicine Clinic Activity Data Set v3 pilot 2)

This data set was generated by PHE during a pilot conducted from July 2015 to June 2016 in five SH clinics: Bedford (Brook), Bristol (GUM), Croydon (GUM), Barnet (GUM) and Southend (GUM). This data set contains the same 12 variables from v2 and an additional 18 questions on recent sexual behaviours, drug and alcohol use, and previous diagnoses and attendance (see Table 4). This data set is referred to as ‘v3p2’ throughout.

Variable selection

Deprivation was included as quintiles, based on the UK indices of multiple deprivation111 derived from the patient’s postcode. Age was included in the model as a categorical variable; plotting the relationship between age and STI diagnosis showed that the association was not linear. We described the number of prior STI diagnoses reported, both longitudinally and from patient report. Within this cohort of young people, there were very few non-chlamydia prior diagnoses, and we therefore included prior chlamydia infection only in the model. Ethnicity and location of birth contained a large number of categories, 15 and 9, respectively, adding 23 degrees of freedom to the model. Many of the categories contained < 5% of the patient population; therefore, these variable categories were collapsed to ensure that there were more balanced categories for modelling. Drug use and problematic alcohol use were excluded because of high numbers of missing data, and sexual orientation was excluded for having too little heterogeneity.

Table 7 describes the variables and categories that were included in the primary analysis.

Sensitivity analyses

A model fitted using a forward stepwise approach, using a p-value threshold of 0.2, did not exclude any of the variables and therefore had the same model performance.

A model was fitted using multiple imputation. The following variables underwent 10 imputation rounds using chained equations: location of birth, ethnicity, deprivation, number of partners, new partners and condom use. The model had a pseudo-R² of 6.6% and a c-statistic of 0.688; the predicted risks ranged from 1% to 68%. Overall, this showed worse discrimination than the model that included categorised missing values.

A model including demographic data only, and fitted using the v2 data set (1,045,373 observations), showed considerably poorer model performance, with a pseudo-R² of 1.4% and a c-statistic of 0.590. The predicted risk of STI diagnosis was limited, ranging from 2% to 24%, reflecting poor discrimination. A typical high-risk individual based on demographics alone would be an 18- to 19-year-old black Caribbean male, born in Europe and living in an area of high deprivation (predicted risk 23%).

Variable selection

Similarly to the young person’s model, within the MSM model, age and deprivation were included as categorical variables, and ethnicity and country of birth were reduced to fewer categories because of the lack of heterogeneity within the sample. Within this cohort of MSM, a variety of prior STI diagnoses were reported, including HIV, syphilis, chlamydia and gonorrhoea. Many of these contained too few records to be included as individual predictors; therefore, a single binary variable indicating STI in the prior 12 months was used. Problematic alcohol use was excluded for having too many missing data.

Table 10 summarises the variables in all the models from this point forward.

Primary analysis

The model was fitted, using categorised missing values, with 36 degrees of freedom and may therefore be overfitted (Table 11). In the MSM model, the only significant demographic predictors of STI diagnosis were being of South Asian ethnicity (OR 2.53, 95% CI 1.05 to 6.10) and being born in Europe (OR 2.46, 95% CI 1.26 to 4.78). Significant behavioural risks included having had condomless anal sex in the previous 3 months (OR 1.95, 95% CI 1.39 to 2.73) and any drug use in the prior 3 months (OR 1.89, 95% CI 1.31 to 2.74).

The model had reasonable performance, with a pseudo-R² of 7.0% and a c-statistic of 0.676. The Hosmer–Lemeshow test showed good model fit (p-value = 0.224). The model predicted probabilities range from 3% to 71%, with a mean of 16%. Using a risk score threshold of 30% would result in one in five patients being classified as being at high risk of STI diagnosis, with a sensitivity of 38.7% and specificity of 84.8% (Table 12 and Figure 4).

FIGURE 4.

Model performance graphs for the primary MSM model. (a) Receiver operating characteristic curve of STI diagnosis, c-statistic = 0.676; and (b) linear prediction (log odds) of STI diagnosis. This article was published in EClinicalMedicine, vol. 4–5, King C, Hughes G, Furegato M, Mohammed H, Were J, Copas A, et al. , Predicting STI diagnoses amongst MSM and young people attending sexual health clinics in England: triage algorithm development and validation using routine clinical data, pp. 43–51, Copyright Elsevier 2018. 110

This article was published in EClinicalMedicine, vol. 4–5, King C, Hughes G, Furegato M, Mohammed H, Were J, Copas A, et al. , Predicting STI diagnoses amongst MSM and young people attending sexual health clinics in England: triage algorithm development and validation using routine clinical data, pp. 43–51, Copyright Elsevier 2018. 110

Sensitivity analyses

Using a forward stepwise approach to the model, with a p-value threshold of 0.2, excluded age, deprivation quintile, number of partners and ethnicity. This model was favoured according to the BIC statistic, but had a poorer discrimination (c-statistic = 0.658) and model fit (pseudo-R² = 5.8%); BIC tends to favour parsimonious models, that include fewer explanatory variables.

A model was fitted using multiple imputation, which underwent 10 imputation rounds using chained equations of location of birth, ethnicity, deprivation, number of partners, sex with a known HIV-positive partner, condomless anal sex and drug use in the prior 3 months. The model had a pseudo-R² of 6.8% and a c-statistic of 0.676; the predicted risks ranged from 4% to 71%. This model showed very similar performance and discrimination to the model that included categorised missing data, and similar direction and magnitude of relationships with the outcome.

A model including demographic data with only the v2 data set (245,863 observations) showed very poor model performance, with a pseudo-R² of 0.5% and a c-statistic of 0.553. The range of predicted risk of STI diagnosis was limited (7–23%), reflecting poor discrimination. A typical low-risk individual based on demographics alone would be a South Asian aged > 65 years living in an area of low deprivation who was born in Asia (predicted risk 7%). This is contradictory to the v3p2 model, in which being South Asian was one of the main risks for STI diagnosis.

Participant selection

We purposively recruited a range of HCPs to include service leads, HAs, doctors and nurses. Providers were targeted to reflect different types of clinics, sizes, geographic locations and client mixes. Selection of clinics was done through individual contacts and through random selection from the list of clinics provided by PHE, which was done in Stata. In total, we aimed to conduct 30 interviews.

Recruitment

Interviews were pre-booked following an invitation sent by e-mail. Participants were contacted up to three times by e-mail before they were considered as not interested in taking part.

Data collection

The interviews were conducted by telephone. Interviews lasted approximately 30 minutes in total, and consent was taken verbally at the start. The interviews were audio-recorded and transcribed by a professional service (Way With Words, London, UK).

Analysis

Analysis used the framework approach, a deductive approach that allows for a more structured approach to data analysis based on predetermined aims and objectives as well as accommodating emerging themes. Content analysis was conducted independently by two researchers (CK and AR) and themes were agreed through discussion until consensus was reached. Both predetermined concepts used for developing the topic guides and emergent themes arising from the data informed the process of identifying the key thematic categories to be used in data coding. 124,125

Study population

All SH services that report to the PHE GUMCADv2 reporting system were eligible to participate, that is level 1, 2 and 3 services within England. An estimated 570 services were reporting to PHE at the time of the survey, and a list of clinic contacts was provided by PHE. A supplementary list of clinic contacts was provided by Tom Nadarzynski [Brighton and Sussex Medical School (BSMS), 18 November 2015, personal communication] and used to update contact information.

Data collection

Providers were sent an e-mail containing study information and the link to the web survey. Up to five e-mail reminders were sent over a 6-month period (December 2015–June 2016): three generic and two personalised e-mails. The link to the survey was also distributed in the delegate packs at the BASHH 2016 annual conference. The survey was developed in Opinio (Version 7.3, Opinio, ObjectPlanet, Oslo, Norway), which is hosted on University College London (UCL) servers, and was designed to take 10 minutes. The survey was piloted by two independent clinicians who work in level 3 services to check for understanding and language. No personally identifiable information was collected.

Analysis

The survey was analysed using descriptive statistics, adjusted for clinic type and location. All analysis was done using Stata 13.

Participant selection

We purposively sampled young men and women and MSM who were attending NHS SH services. The recruitment framework categorised MSM by age and young people by age and gender, with equal recruitment across two clinic sites. We targeted 15 heterosexual young people and 20 MSM (total = 35).

Recruitment

Participants were recruited from two SH clinics: Claude Nicol Centre, Brighton, and Mortimer Market Centre (MMC), London. Participants were approached in the clinic waiting room and given a study information sheet to read before deciding to take part. Participants were offered a £20 high street voucher as a thank you for taking part. Interviews were scheduled to take place either on the day of recruitment or at a future time.

Data collection

Interviews were conducted by researchers in person within the clinical setting. Interviews were designed to last 30 minutes, and were piloted with members of the PPI group to check for understanding and sensitivity. Written consent was taken prior to the interview starting, and the interviews were audio-recorded and then transcribed using a professional service.

Analysis

We used the same analysis methodology as described for the HCP interviews.

Study population and sample size

We recruited young people and MSM who were attending a NHS SH clinic, aiming for a representative sample of attenders within these groups. DCEs are not amenable to conventional power calculations in advance of developing the instrument. However, other studies using DCE methods to assess preferences for health care have typically included 200 participants. 126 As we planned subanalyses in young people and MSM, we aimed to recruit 350 participants.

Recruitment

Patients were recruited from three SH clinics: Claude Nicol Centre, Brighton; MMC, London; and Archway, London. Participants were approached in the clinic waiting room and given a study information sheet to read before deciding to take part.

Instrument design

The questionnaire used a ‘labelled’ rather than generic design. Four modes of brief behavioural intervention were included in the final design: ‘talking’ to someone [meaning talking therapies such as counselling and motivational interviewing (MI)], an ‘e-mail or text containing health advice’, an ‘online session by yourself’ or an ‘online group session’ (Table 13). A fifth option, ‘opt out’, was also offered. The attributes included type of contact, type of activity involved in each session, length and number of sessions, and the person who mediates the sessions. Note, however, that each attribute was not necessarily applicable to each intervention; for example, a person is not needed to mediate an e-mail/text-based intervention. The number of sessions (one to six) and their length (15 minutes to 1 hour) were deliberately small to reflect the brief nature of the interventions shortlisted.

The pilot questionnaire was generated using an orthogonal approach and set to 12 choice tasks given 12 degrees of freedom in the design using the NGene V1 software (ChoiceMetrics, Sydney, NSW, Australia). It was completed by 24 clinic attendees. The pilot design required participants to make two choices per DCE question. The first included an ‘opt-out’ option; this was omitted in the second (referred to as a ‘forced choice’ question). This two-stage approach was included to evaluate the concern that a large number of participants would ‘opt out’. However, the forced choice question was removed from the final design, as only a minority of responses indicated a preference not to participate. The final instrument was produced using a d-efficient approach using priors from the pilot. Participants were asked to complete all 12 DCE questions. Eight versions of the questionnaire were produced, in each of which the order of the DCE options and questions was different.

Data collection

Once patients had signed a consent form, they were given the paper questionnaire and asked to complete this while in the waiting room. The questionnaire was designed to take 10 minutes. Participants were asked to provide limited demographic and risk behaviour information, including age, gender, ethnicity and sexual orientation. The questionnaire was piloted with patients in one clinic to check for understanding. Data were entered into a Microsoft Access^® 2013 (Microsoft Corporation, Redmond, WA, USA) database.

Data analysis

Analysis used conditional logistic (CLOGIT) and latent class models (LCMs). CLOGIT models were the basic form of analysis, but because the results are presented for the ‘average’ respondent, they do not address issues of heterogeneity. LCMs address heterogeneity by assuming that the population of interest consists of a number of prespecified latent classes with a probability that each individual belongs to each class. Likely ‘membership’ of each class is estimated as a function of prespecified covariates: born in the UK (yes/no), having tested for a STI within the past year (yes/no), previously diagnosed STI (yes/no) and risk group (heterosexual 16–20 years of age, heterosexual 16–25 years of age, MSM 16–25 years of age, MSM 26–50 years of age and MSM ≥ 51 years of age). The number of classes was determined by selecting the number of classes in the model with the lowest BIC and examination of the standard errors on the coefficients.

All results are presented as ORs and 95% CIs based on robust standard errors given that each participant provided multiple responses. All attribute levels were dummy coded (1 for group membership, 0 otherwise) except when estimating the alternative specific constants (ASCs). The ASCs represent the extent to which people prefer one of the intervention options or opting out when all other factors are disregarded. That is, they indicate the strength of preference for each individual label. For the ASCs, effects coding was used (1 for group membership, –1 otherwise) to avoid confounding with the base levels on the main attributes. ‘E-mail or texts’ was used as the reference option in all analyses. Statistical analyses were performed using Stata 14 and NLOGIT 5 (NLOGIT 5, Econometric Software Inc., New York, NY, USA); the scenario evaluation was undertaken using Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA).

Current services

Most staff reported a mixture of appointment and walk-in services, with clinics varying in how patient pathways are set up. Level 2 services, many of which are nurse led, have set clinic times for specific procedures (e.g. coil fitting) that can require involvement of different specialties and are therefore appointment based. Self-check-in and booking was mentioned by staff from two GUM clinics:

Before they see the doctor or the nurse, there will be a, sort of, kiosk that will ask them pertinent questions in a way just to save time. So the majority of the history taking, if you like, will be done on the, sort of, electronically by patients.

Doctor, level 3

Several clinic staff reported having specific services and pathways for young people and MSM groups. The age cut-off for ‘young’ varied from 16 to 19 years, and the change in the pathways included additional questions, assessments for vulnerability and speaking to a HA:

An MSM who’s in his 20s or 30s, whatever, with symptomatic, so we have a policy of do-not-turn-away, we need to see that person and treat them.

Nurse, level 3

Other services were offered by participating clinics, with little standardisation in how the services were set up; these included contraceptive clinics, drug and alcohol services, psychological services and condition-specific (e.g. warts) services. Referrals to external services also varied between clinics, including GUM services (by level 2 clinics), sexual assault or domestic abuse, drug and alcohol services, charities [such as London Friend or the Terrence Higgins Trust (THT)] and other clinical specialties (e.g. psychosexual).

A variety of triage methods and their purposes were described by staff from different clinics. Examples of how triage rules varied included ‘MSM with greater than X number of partners’ (Doctor, level 3) or ‘Somebody who is displaying a sexual behaviour where there is multiple partner change’ (Doctor, level 3). Not all clinics had set rules, with triage lacking standardisation:

Well we have guidelines. We have sort of GUM guidelines, departmental guidelines, but it’s down to the individual doctor or nurse, seeing the patient, to decide whether someone should see the health advisor.

Doctor, level 3

One participant reported having an electronic triaging system, similar to the proposed approach in this project:

The system, the way it’s devised, does flag up to say this patient needs to see a health advisor because of risk A and B and it lists it down for you, what the clinician in the room has ticked.

HA, level 3

Clinic staff reported offering a range of SH promotion and risk reduction interventions, some more formally than others. Many of these activities were not specifically funded, but were done within existing resources:

So we just get paid as a level 2 sexual health screen regardless of whether we offer an intervention or not.

Nurse, level 2

Informal interventions included condom distribution and general health promotion messages, which were reported as being done by all clinic staff:

I mean, I think, really, sexual health promotion is just sort of integral to every kind of consultations so, in a way, some degree of sexual health promotion should be happening in every consultation.

Doctor, level 3

Various national campaigns were being delivered by clinics, such as the C-card initiative,127 and the Sex Positive campaign by Brook. 128 More formal risk reduction interventions focused on one-to-one sessions and outreach or educational services. One-to-one sessions in level 3 services were generally performed by HAs or counsellors, rather than all clinical staff:

So to a certain extent, the majority of staff have had some training in motivational interviewing [. . .] if someone starts to need more intensive motivational interviewing interventions, they’re referred to the health advisors.

Doctor, level 3

The one-to-one intervention method mentioned most often was MI. Brook specifically reported a longer 6-week educational programme about self-esteem and SH.

Proposed triage

The proposed Santé approach was generally seen as something that was done already, and this resulted in some respondents not being sure of the utility:

. . . we already have, well, it’s not a tool, but we have a means to ask people, so if anything was going to be developed that had a chance of being used it would have not to increase the length of time.

Doctor, level 3

However, this led others to state that it could be acceptable:

I think that would work because we do triage forms which give us a little bit of a clue.

HA, level 3

Potential barriers to the proposed triage included increasing the time needed with patients, how well the score would perform, training required in using it, how the patient referral would work, and issues in adapting EPR systems. On the other hand, several opportunities were highlighted, such as the perceived benefits of standardisation and accurate prediction, ease of having an EPR-based system, and potential patient acceptability.

Proposed interventions

We presented the following intervention types to the HCPs, and asked for both the opportunities and barriers to potentially implementing them in their setting: videos in the waiting room, group sessions, online resources, including mobile phone ‘apps’, and single and multiple one-to-one sessions.

Health-care professionals gave mixed opinions on videos, with the practical ease of implementing them and having a potentially receptive and captive audience given as opportunities:

It’s an easy way for people to kind of . . . people aren’t doing very much, so it’s quite a good time to kind of drill it in.

HA, level 3

However, there were concerns over the lack of targeting and appropriateness for diverse waiting rooms:

. . . we have a very heterogeneous waiting room for the walk-in clinic, you know. The challenge, I guess, would be how you target that, or do you have a number of different ones for different risk groups.

Doctor, level 3

Patient group sessions as an intervention format were, on the whole, not well received (e.g. ‘I think that’s a non-starter’ – HA, level 3). The barriers to using group sessions focused on resource issues, with a lack of appropriately trained staff, staff time and clinic space, and general disruption to the clinic running smoothly being potential issues. HCPs also anticipated low patient acceptance:

. . . personally, if I was a patient, I’d run out screaming if somebody tried to get me to do some group work when I’m sitting in a clinic that I might feel slightly uncomfortable about, anyway.

Nurse, level 3

Positive aspects to group sessions were highlighted, although were mostly assigned to specific risk groups and support group models. Another positive aspect was being opportunistic and engaging with patients while they are at the clinic:

Catch them while they’re waiting you know, they haven’t got anywhere to go.

Nurse, level 2

One HCP reported offering a group intervention and another reported that they had done so in the past.

Some clinic staff reported that they referred patients to online resources and apps; specific examples included dedicated online education tools for psychosocial issues. There was generally a positive attitude to using digital interventions, across staff and clinic types. The main barriers concerned patient motivation and uptake and a current lack of tools to which to refer patients:

There’s so much else to distract them on the internet, but unless it’s something they enjoy doing, the learning is not going to happen unless it’s couched in a very user-friendly, quick, vehicle.

Doctor, level 3

Opportunities for digital interventions included their accessibility, perceived patient preference and minimal staff delivery time required:

We have quite an IT-savvy patient group, I would say, so something like that might appeal.

HA, level 3

Yeah, well, they love apps. I mean we suggest apps. I’m quite an elderly nurse now but even I know to suggest apps.

Nurse, level 2

Brief one-to-one sessions of MI were mentioned as something that was offered by all the GUM and Brook clinics that we engaged with. However, it was also an intervention that providers highlighted had a lot of challenges. One participant identified a lack of evidence as a challenge; this was raised more frequently for MI than for other intervention types. The current needs associated with one-to-one sessions focused on costs and staff resourcing:

So I think the clinical time, availability of time in the clinic is probably the biggest challenge.

Doctor, level 3

Patient motivation was also viewed as a barrier, that is, whether or not the patient would be open to the intervention:

With behavioural interventions, if people are referring into that service, if you’ve got to work out whether the patient is really ready for this intervention, because if the patient is not ready for it, it’s just no point doing it.

HA, level 3

The main opportunity that was raised for one-to-one sessions was the flexibility that these offer, and the ability to tailor sessions to individual risks and needs. Many of the HCPs expressed that they felt that the brief sessions were effective, even if this is hard to demonstrate:

Yes I know that’s probably not the most cost-efficient. But I think that’s probably the most effective method of risk reduction, because it is tailored to the actual patient’s needs and you have time to explore what their risk is.

HA, level 3

Similar opportunities and barriers were raised around a series of one-to-one sessions, with HCPs highlighting the constraints of time and resources available in clinic to deliver these:

Yes, that’s a great idea, but we’ve never had capacity to do that.

HA, level 3

There was a perception that this was a good intervention format and that it could be effective, provided that patients were motivated:

But if it’s something that perhaps is reserved for people who are seen as particularly high risk, and particularly amenable to this sort of intervention, then it would have a place.

Doctor, level 3

Implementation challenges

Financial and staffing constraints were raised frequently as barriers to the delivery of current services, as well as being an anticipated barrier for delivering novel triage pathways or interventions. One approach currently taken to limited budgets was self-sampling:

All this quick checking and self-assessment has started as a result of changes in funding and competition in sexual health services [. . .] that’s where that’s all heading.

HA, level 3

There was a perception that commissioners focused on treatment rather than prevention for STIs and that evidence was needed for a service to be commissioned:

Commissioners, I think, will not fund anything that hasn’t been shown to be effective. And so I think you’ll have to demonstrate in some way that it is effective and not just that it’s acceptable.

Doctor, level 3

Continuity of care was deemed both important and lacking by HCPs, with issues associated with how services are commissioned:

. . . one of the problems that we face, generally, is that drug and alcohol services generally tend to be borough-based, and all the patients we see come from everywhere.

Doctor, level 3

Additional services or improvements that were desired by HCPs included outreach for homeless people and sex workers, improved drug services, community education and post-exposure prophylaxis (PEP) follow-up pathways.

Current services

Two respondents reported not offering any health promotion or risk reduction intervention services, both of whom worked in level 3 services. Table 15 describes the services currently being offered by SH services in England.

The least common health promotion activities offered were videos and apps, whereas leaflets and brief one-to-one sessions were relatively common in both level 2 and 3 services at the time of the survey in 2015/16. Five respondents reported previously showing educational videos; reasons for stopping included lack of funding (n = 3), no observed impact (n = 1) and the materials no longer being available (n = 1). Lack of funding was cited as the reason for one clinic ceasing to offer an app, and another for ceasing an online intervention. HCPs cited a lack of trained staff time as the main reasons for stopping in clinics that previously offered ‘talking interventions’ (i.e. one-to-one, multiple sessions of MI and group sessions).

Triage

The majority of clinic staff (77%) reported triaging on the basis of SH risk; this was less common in level 2 services (68% vs. 84%). No respondents reported using an automated (or algorithm-based) triage decision; rather, a nurse, doctor or HA always made the decision, along with patient input. Comments in the free-text section about an automated system were mixed, with many reporting that it is ‘not necessary’, but others stating that it would be useful, for example: ‘a good thing as long as not too long and time consuming’. HCPs were asked what three factors they considered most important for assessing patient risk of STIs. Sexual orientation, number of recent partners and the types of sexual activity reported (e.g. condomless sex) were the most commonly selected factors, and this was consistent between level 2 and 3 services. These are all variables included in the GUMCADv3 tool.

Overall, 14 responses stated that the SH service did not have any EPR system [level 2 = 2 (13%), level 3 = 12 (18%)]. Of those with EPR systems, 45 (55%) reported having it amended, of whom one-quarter stated that it was very difficult to do; 9% said that they were not able to amend their system despite trying to do so.

Intervention barriers and opportunities

Of the interventions not currently offered by clinics, the most desired by level 3 services were online learning materials (67%) and mobile ‘apps’ (64%). Group sessions were the least popular, with only 18% of level 1 and 2 services and 23% of level 3 clinics expressing any desire to offer them (Table 16).

For those interventions that clinic staff expressed an interest in delivering, the main barriers to and motivations for delivery are presented in Table 17. Most of the barriers were related to funding and staff time for delivery, whereas the motivations were around potential effectiveness and uptake (rather than practical reasons). This suggests that if digital or video-based interventions were developed, clinics would be able (and want) to deliver these.

Trial feasibility

Respondents were asked if they would be interested in taking part in a cluster RCT of brief behavioural interventions, and whether or not being randomised at the clinic level would be acceptable. Nearly half of the clinics (48%) expressed an interest in trial participation, but only 40% reported being comfortable with randomisation. There was no statistical difference between level 2 and level 3 clinics (38% vs. 41%, respectively).

Risk perception

We asked participants about their self-perception of sexual risk, as a way to introduce the concept of our proposed triage approach and to gain an insight into potential barriers to triage. Trust, or lack of trust, of a sexual partner was frequently associated with service users’ perceptions of their sexual risk, driving sexual risk behaviour decisions such as using condoms; this was common among both young people and MSM. The concept of causal and regular partners was much more common among MSM, and the types of sexual behaviours practised within these partnerships were different; for example, not using a condom was something done only in a relationship. Young people discussed different levels of trust between people they knew and, for example, someone they met via the dating app Tinder (Match Group, LLC, Dallas, TX, USA).

Use of condoms, a major aspect of risk perception in young people, was reported as being influenced by peers. Young women specifically discussed an inability to negotiate the use of condoms, or were described by young men as not negotiating condoms:

. . . it would be a big rarity if a girl basically asked me to put on a condom.

Male, 21–25 years

There was a perception that peers were not using condoms and that normalised the behaviour; concern about pregnancy was more of a motivation for their use than STI risk. For MSM, on the other hand, condom use was seen as a matter of preference. Condom use was also circumstantial, and non-use was often seen as a one-off event:

I like to class myself as someone who is safe with sex but I am not perfect.

MSM, 16–25 years

Other factors that shaped how people felt about their personal risk included drug and alcohol use, and perceptions about the general prevalence of STIs in the population and in overseas partners.

Risk was seen as dynamic by both groups and was related to relationship status, age or maturity, and a personal scare experienced by either the individual or someone they know. Common views were that being in a relationship is associated with lower risk and that risk decreases with age: ‘I’d like to think that the older people do get the more kind of cautious and aware of STIs they are’ (female, 16–20 years). In the case of young people more specifically, university was seen as a distinct period of taking more risks, but that this risk was self-limited during this period. In MSM, self-esteem was specifically mentioned as affecting risk behaviours:

. . . if your self-esteem is quite low, it’s quite possible to engage in un-safer practice than if your self-esteem is quite high.

MSM, 26–50 years

Proposed triage

We asked participants about being offered services based on standardised risk assessment, and they had mixed views. Positive aspects of triaging included the process of having a score acting as an intervention in itself, as it may raise awareness about risks that service users had not considered previously:

. . . it’s something that people might not like, but you, kind of have to know, it’s better to know.

Female, 21–25 years

Participants’ trust of HCPs also meant that they would listen to them, as they were seen as knowing best, although this would rely on the triage being well explained (no ‘technical jargon’) and trusted:

. . . you’re a registered health-care professional, so I trust your reasoning.

Female, 16–20 years

It was acknowledged that triage might act as a ‘shock factor’, and that, despite not being supportive of it at the time, service users might ‘reflect’ afterwards, although this would not necessarily lead to behaviour change:

I wouldn’t consider changing my behaviour actually, I would just see it as, yes, a warning.

MSM, 16–25 years

Concerns raised with triage included anxiety associated with a classification of ‘high risk’. Alternatively, telling users something they already knew was considered redundant:

. . . either way I’m going to get tested, so I don’t know why they tell people really.

Female, 21–25 years

Some MSM expressed concern that they were pigeonholed simply based on being MSM and belonging to a particular demographic group:

. . . with gay culture being so sleazy you just sort of expect to be high risk all the time.

MSM, 16–25 years

Onward referral based on triage raised concerns that services should be available to everyone, with the denial of services or rigidity of referrals criteria not liked. However, many noted that the offer of supportive services was positive and that service streamlining made sense.

Proposed interventions

Similar to the HCP interviews, we asked participants for their thoughts on different intervention formats and any preferences for waiting room videos, group sessions, online materials and apps or one-to-one sessions. Service users raised multiple concerns about videos in the waiting room, suggesting that they would make people feel awkward in mixed waiting rooms or increase their anxiety:

. . . having had sex with 20 different people last night, they don’t need a video saying ‘don’t be promiscuous’.

MSM, 16–25 years

Opportunities, however, focused on the notion that education is good, and a SH setting is the correct setting for SH education:

Why not?’ Information is a good thing. It’s a sexual clinic, so that’s why people are there, to talk about sex.

MSM, 26–50 years

Recommendations about content included that text information or statistics would be desirable and short advert or campaign-type clips would be acceptable, but that content should not be graphic.

Group sessions were not viewed favourably, with many participants stating that they would not take part despite seeing the role that they could have. Privacy and confidentiality were the primary concern with this intervention format, as many did not want to talk about SH in a group despite considering themselves open:

. . . you share funny stories with your friends, and I do talk about sex quite a lot with my friends, but not about this part.

Female, 21–25 years

Among MSM this was seen as an issue particularly if someone is shy or isolated:

. . . they cannot talk at home, they cannot talk at school, and they cannot talk in church and they cannot talk to their own best friends, so they are not going to start talking here.

MSM, 26–50 years

These concerns were associated with the group or other situations not being appropriate to their circumstances, and a belief that there would be judgement about lifestyle. MSM were also concerned about the possibility of bumping into someone they knew, and about HIV status affecting how someone might participate. On the other hand, the ability to learn from others’ experiences was seen as valuable.

Generally, having some form of digital-based intervention (e.g. website, social media or app) was positively viewed, although respondents felt differently about the various formats. Although all participants expressed concerns about apps, this was more an area of concern for young people, who typically reported sharing mobile phones, with friends and family often looking at them, or leaving mobile phones out ‘on the table’:

. . . people use my phone, so they would know my business.

Female, 16–20 years

The apps were considered somewhat redundant if there was a website available:

. . . most information I can find it online, I don’t need an app just for that [. . .] it’s not like you need to check it every day.

MSM, 16–25 years

In favour of apps, however, were their convenience and immediacy, especially if they could do more than just provide information (e.g. book appointments or provide remote clinical consultation), or if they could give wider health information. Social media were less popular, with concerns over Facebook (Facebook, Inc., Menlo Park, CA, USA) or Twitter (Twitter, Inc., San Francisco, CA, USA) not being anonymous and therefore lacking privacy:

I wouldn’t share my, as much personal information as if I come here and I talk to someone.

Female, 21–25 years

However, some MSM said that they had followed online pages about SH and had seen adverts about SH promotion advertised through Facebook. More specifically, MSM-targeted apps or websites such as Grindr (Grinder LLC, West Hollywood, CA, USA) were suggested as good places to have SH information. Online formats were viewed as convenient as they could be accessed as and when needed; however, there was a concern that they needed to be a reliable source to ensure anonymity.

One-to-one ‘chatting’ interventions were well received, and most participants indicated that these were their favoured option. The face-to-face aspect and the element of human interaction were considered important (‘the thing about, you know, chatting to a human is, they’re receptive’, male, 16–20 years), and the ability to ask questions in a tailored session. The need for the session and what people expected to get out of it was related to their trust in the HCP, for both the referral and what they would share in the session. Negative aspects of this intervention were the possibility of being embarrassed, not seeing it as being needed and being inconvenient. Most participants said that between 15 and 30 minutes would be a good duration. Carrying out sessions over the telephone received mixed reactions, with many saying that they preferred face-to-face contact and the privacy afforded by a clinic, whereas others saw value in the convenience:

You get to speak to a real person, you can do it from the comfort of your own home.

Female, 21–25 years

Conditional (fixed-effects) logistic regression analysis

‘Talking’ was chosen as the most preferred intervention option on 40% of occasions. The next most frequently chosen option (27% of responses) was the ‘e-mail or text’-based design. The opt-out option was preferred on < 10% of occasions (Table 20).

The CLOGIT model explained more of the variation in the data than a model with no independent variables (likelihood ratio chi-squared test, p < 0.0001). McFadden’s pseudo-R² was 0.13, indicating that the model fitted the data moderately well. 129 It also predicted 42% of choices correctly, and the signs on the model coefficients were logical, offering a degree of plausibility to the underlying model (e.g. people preferred shorter to longer sessions).

The analysis showed that respondents generally preferred interventions over ‘opting out’ (Table 21), with ‘talking’ interventions being the most clearly favoured option (OR 1.45, 95% CI 1.35 to 1.57, vs. ‘e-mail or text’). Face-to-face group sessions were generally preferred less than individual face-to-face sessions (OR 0.66, 95% CI 0.57 0.78) or ‘one-to-one telephone calls’, although the latter comparison did not achieve statistical significance (OR 0.87, 95% CI 0.73 to 1.02). Respondents generally preferred fewer sessions to more sessions, and shorter sessions were more highly valued than longer sessions. Respondents indicated a strong preference for sessions to be facilitated by HCPs rather than by peers (OR 0.53, 95% CI 0.46 to 0.60), but they did not express a clear preference on the type of HCP.

Latent class model

Three classes were identified for the LCM, which predicted a higher proportion of correct choices than the CLOGIT analysis [73% (2419/3312)] and was more flexible in terms of the options it predicted. None of the sociodemographic variables was predictive of class membership, and these were therefore omitted from the final model.

Participants in the classes were similar in terms of how they valued the number and length of sessions, the choice of facilitator and whether or not meetings were one to one or group based. However, they differed in terms of their preferred intervention method (Figure 5). Participants who were more likely to be in a LCM class 1 (60%) favoured ‘talking’ interventions, although all other options were preferred to ‘nothing’. Those who were more likely to be in class 2 (14%) had a general preference for ‘opting out’, although their next strongest preference was for ‘talking’ interventions. Respondents who were more likely to be in class 3 (26%) demonstrated a preference for ‘e-mail or text’-based interventions over any other option.

FIGURE 5.

Latent class model results. a, ORs are effects coded, otherwise they are dummy coded; label percentages indicate the proportion of respondents likely to be in each class; log-likelihood –3,547; McFadden’s pseudo-R² = 0.33. Reproduced from Miners A, Llewellyn C, King C, Pollard A, Roy A, Gilson R, et al. Designing a brief behaviour change intervention to reduce sexually transmitted infections: a discrete choice experiment. Int J STD AIDS (vol. 29, iss. 9), pp. 851–60,130 copyright © 2018 by SAGE Publications. Reprinted by permission of SAGE Publications, Ltd.

Reproduced from Miners A, Llewellyn C, King C, Pollard A, Roy A, Gilson R, et al. Designing a brief behaviour change intervention to reduce sexually transmitted infections: a discrete choice experiment. Int J STD AIDS (vol. 29, iss. 9), pp. 851–60,130 copyright © 2018 by SAGE Publications. Reprinted by permission of SAGE Publications, Ltd.

Participant selection

We used a sampling framework, with quota sampling based on sexual orientation, age and gender (including MSM and young people) to gain a broad range of opinions. Four groups were defined: MSM aged < 25 years, MSM aged > 25 years, young heterosexual women and young heterosexual men. Approximately six to eight participants (n = 24–32 in total) were invited to attend each group.

Recruitment

Participants were recruited from the SH service and community settings through posters and leaflets for group discussions. Young heterosexual men were purposefully approached and recruited from SH clinics for individual interviews because of a lack of uptake for group discussions. All participants were recruited in Brighton for pragmatic reasons.

Data collection

Written consent was taken prior to the focus group discussions or interviews starting, and all discussions were anonymous. Each focus group discussion was planned to last 45–60 minutes. Discussions were facilitated by two researchers trained in focus group methodology (AR and AP) and were transcribed verbatim. Interviews were planned to be 30 minutes and were conducted by a single researcher (AR, CK or Sarika Desai). Patient participants received £20 as recompense in line with current practice.

Analysis

We used the same analysis approach as described in Chapter 4 (i.e. content analysis, using a framework approach).

Brief one-to-one consultation

Reasons for choosing this intervention included:

• it being one of the effective approaches identified in the WP1 systematic review

• it being a clear front-runner from the process of WP3

• it being already in place, with most of the clinics interviewed currently offering these

• most participants having favourable opinions about it

• it needing to be conducted in a personal and private space.

In the case of MSM, three effective trials used a one-to-one approach with counselling, MI and personalised cognitive counselling. 69,79,84 Two effective interventions among young people used this approach;56,73 one trial included both young people and MSM. 56 The intervention manuals were available for three73,79,84 out of these five effective trials. Through round table discussion, the conclusions from WP3 were used to refine and select the intervention components from these trials to make decisions about the format and content of sessions. For example, personalisation was considered important to patients, and, therefore, this influenced the decision about how the session would be structured. We decided that the one-to-one approach would use the most appropriate components of all the available manuals for both MSM and young people, and apply these to the needs of both groups. This was based on the finding that, generally, the needs, risk perception motivations and preferences for delivery were similar between young people and MSM. It also supports the flexible nature of the one-to-one approach.

In the interests of equitable delivery, the intervention was designed to be offered to all service users at high risk, regardless of their motivational assessment, a key barrier to delivery raised by providers and service users. The focus of the sessions would then initially be on identifying ‘what aims need to be achieved in this session and how to achieve them’, including a checklist of conversation topics and their level of motivation. Consultation tasks might include risk assessment (e.g. patients’ individual risk level, the kind of problems they encounter) and provide normative or attitudinal information based on their risks. In the case of those who are already motivated, that is those who understand their risks and want to change their behaviour, this process could be skipped and the focus changed to ‘what you can do’, for example use condoms. The sessions would draw on MI principles but not attempt to deliver MI per se. The lack of trained staff time was a key barrier to the delivery of interventions. It was therefore concluded that the intervention should be consistent with MI principles, but without requiring intensive MI skills training. This would make it more likely to be feasible.

We sought service user input on the session duration and proposed that the intervention would be a single session designed to last 20–45 minutes, dependent on need and motivation. The decision to have a single session was based on the need for the intervention to be pragmatic and able to be delivered within existing resources. Service users also suggested that multiple sessions would be a barrier to patient attendance and engagement. We acknowledge that service users may be referred for health promotion or partner notification on future clinic visits; even though we decided on a single session, we planned for these to be adaptable and that additional booster sessions could be offered through the use of action plans to be included in patients’ records on an EPR system. These decisions and the overall design attempt to standardise what many HCPs reported they were already doing, and adds evidence-based structured elements to enhance this.

Motivational interviewing is a collaborative, goal-oriented style of communication with particular attention given to the language of change. It is designed to strengthen personal motivation for, and commitment to, a specific goal by eliciting and exploring the person’s own reasons for change within an atmosphere of acceptance and compassion. We prioritised a focus on listening for Change Talk within Motivational Interviewing (DARN-C). 137 Identifying and working with patients’ change talk is essential for moving from exploration and motivation to directional change. The consultation was intended not to add additional consultations to the clinic service, but rather to provide a way of structuring usual discussions with high-risk clinic attendees as identified in a standardised way by the triage algorithm.

Brief one-to-one manualisation

In order to provide a pragmatic, effective and time-efficient intervention, the one-to-one consultation was structured to suit HAs’ existing experience with MI by providing a sequence of five linked key steps (Figure 7) that target specific needs and provide a menu of effective intervention strategies (Box 4).

FIGURE 7.

Santé task list: leading the patient through change. The five-step pathway for young people and MSM (based on strategies described in Box 2).

Digital online intervention

Reasons for choosing this intervention type included:

• It being one of the effective intervention types identified in the systematic review.

• It being favourable to service providers because of the limited demand on available resources.

• It being favourable to service users because of the convenience. However, participants were worried about the confidentiality of an app on their phone and therefore generally preferred websites.

• Some clinics were already referring participants to websites and/or are in the process of developing digital services.

Intervention mapping identified effective trials that used some form of digital intervention: three trials58,63,68 that were effective in young people and five8,80,83,85,86 that were effective in MSM. However, of these eight trials, intervention materials were available for only two (Figure 8). None of the effective interventions identified in WP1 is currently available online. The Downs et al. 63 website was developed in the USA and uses videos to present relationships and SH education, but at the time of our study this web page was being updated and charged individual users for access. The need to adapt the page to a NHS setting, the individual user fees and cultural differences in the content were barriers to its use. The second intervention with available content was from Mevissen et al. ,68 which consisted of a virtual clinic consultation. However, the virtual clinic was developed in the Netherlands; adaptation would have required considerable resources to translate the content into English and adapt it to the NHS setting, and it was not available free of charge for research use.

FIGURE 8.

Summary of RCTs available for adaptation in the IM process. YP, young people.

Therefore, there were no evidence-based digital interventions that could be used ‘off the shelf’. We consequently decided to develop a web page that could function as a portal to selected web resources, and be piloted to measure acceptability and engagement, although this would not be an evidence-based risk reduction intervention. We decided that this would include customised links to trusted web resources, which would tailor content to users via brief demographic screening questions, as the personal aspect was important to service users. Placeholders from legitimate bodies (e.g. NHS Choices) were discussed for inclusion, and we sought recommendations from HCPs and our PPI group.

Triage process

Most participants felt that the offer of a one-to-one intervention could be appropriate and acceptable (especially to ‘others’ and ‘the younger ones’), but was contextualised with significant considerations when participants spoke of their own perspectives. At least two participants, both young males, were anxious that the offer of an intervention by a HCP would be ‘scary’ and make them feel ‘stressed’. However, such an offer was felt to be more acceptable when a diagnosis underlined the relevance and value of the offer. When offering an intervention was seen to be normal practice, and was conducted confidentially, this was seen to reduce the anxiety of being specified:

It depends if you’re being singled out or if it’s the same for everyone . . . If they say, because of X, Y and Z, we think this and it’s done in a normal, generic way – that will be more relevant.

MSM, > 25 years group

The most frequent factor cited as a positive influence on the acceptability of an offer was patients’ perception of the value of or need for an intervention. Motivating factors were identified or articulated as perception of risk (and potential impact of infection), a STI diagnosis and recognition of the clinic’s supportive and person-centred approach (‘not just tick a few boxes’). Across MSM and young people there were repeated expressions of the desire for services that addressed them as individuals with specific and distinct needs. Participants also speculated that the refusal by some patients to engage with personal risks would be a key factor in their response to the offer of an intervention, and this was considered to be a greater concern for younger patients, particularly according to older MSM.

One-to-one consultation

The offer of a HA appointment was considered acceptable by participants, but was seen to be viable only when the patient recognised that there was sufficient need or value and that the content would be relevant. Participants expressed little appreciation of the value or purpose of speaking to a HA in the absence of an immediate and explicit need, such as a STI diagnosis.

Younger participants in particular discussed the value of and preferences for the timing of a HA intervention. There were diverse and divided opinions across and within groups, but most said that they would prefer that an intervention would start straight away, primarily because of other demands on their time. Younger participants also acknowledged the risks of forgetting and not coming back, and not wanting to provide telephone numbers to clinics.

The ideal length of a HA consultation, according to participants, varied between 20 minutes and 1 hour, with most participants suggesting around 30 minutes to be appropriate and acceptable. Web-based interaction with HAs, such as private webchat or telephone interactions as an alternative to face-to-face appointments, was popular with all groups. However, the older MSM group specifically valued speaking one to one on the telephone, whereas the younger groups tended to prefer webchat interactions.

There were mixed responses to the provision of written ‘action plan’ cards as part of an intervention. Although the value of clarification and a reminder of key points was identified, some participants found the cards ‘nannyish’, and several participants suggested that paper versions (even at credit card size) would be easily lost or discarded because of confidentiality concerns. The provision of the same information by text message was considered a more effective and preferred method.

Digital online intervention

Participants across all demographics valued the internet as a source of immediate, accessible information, but it was not typically identified as an arena for exploring sexual behaviour. However, participants clearly expressed a desire for interactive content over static information, which would enable them to ask questions specific to their individual needs. The screening page offered as an example (which asked for age, sex and sexual orientation) was acceptable and valued by all groups and interviewees. However, any use of login details was seen as frustrating and off-putting.

There was a high level of respect for NHS web content across all groups, and it was considered reliable and factual. This, however, went hand in hand with a prevalent perception that NHS web pages would also be wordy, static and unengaging: ‘. . . the NHS is gloriously boring and matter of fact’ (MSM, > 25 years). There was widespread recognition that a risky degree of unreliable information exists on the internet; however, the availability of interactive content (The Mix and Brook’s ‘text/web chat’) appeared to over-ride concerns about the reliability of information on these (mostly) unfamiliar sites. Younger participants expressed enthusiasm for the youth-focused presentation of these sites, and the opportunity to ask questions via text/webchat appeared to over-ride or displace previously expressed caution about reliability.

The MSM focus groups were shown the following web pages: THT, GMFA, NHS Choices (‘Sexual health for gay and bisexual men’, www.nhs.uk/live-well/sexual-health/sexual-health-for-gay-and-bisexual-men/) and Men’s Safer Sex. 138 Familiarity with some of these sites may have influenced responses, but a key issue that was raised repeatedly by participants was the need for them to personally identify with sites. The site for men of any sexual orientation was largely dismissed because of the prominent image of a woman’s lips, but was thought potentially useful to younger bisexual men or men not fully embracing a gay identity: ‘This definitely looks predominantly targeted to straight people’ (MSM, > 25 years).

Participants appeared to be less engaged with sites that primarily offered static information and advocated more dynamic content. The description of risks as ‘slip-ups’ in the site for men of any sexual orientation was appreciated as an alternative to ‘risk’ when discussing future experience, and the use of narrative to engage readers was briefly raised in the older MSM group. But even in this group there was little understanding of or interest in behaviour change interventions and most discussion centred on websites as sources of information:

. . . that is, for most people, what happens. They slip up. And they may get a reminder to plan, so I go, ‘all right, yes.’ Because, again, it’s for that person to kind of go, ‘All right, yes, I’ve got to plan this.’

MSM, > 25 years

The THT page was largely dismissed and disregarded by MSM participants as being visually too busy and oriented to the charity’s fundraising needs:

THT is clearly about health fundraising. That first page is just saying to me, ‘We’re a charity’, and I would not have even thought about going to there for social advice.

MSM, > 25 years

The GMFA page was acceptable to some, but others (especially younger participants) considered the presentation of a man in underwear too sexual:

. . . going back to the idea again, of wanting to be represented; I am put off by the whole beautiful people thing and to me, it just seems too sexy to be educational.

MSM, > 25 years

The young people were shown the following web pages: FPA, Brook, NHS Choices (‘Live Well: Sexual health’ www.nhs.uk/common-health-questions/sexual-health/), The Mix and Seventeen Days (women only). Preferences did not differ greatly between male and female participants. The FPA site was dismissed in the female group with little comment and the male one-to-one interviews took little notice of this page, referring only to its potential to provide factual information, and its off-putting status as a charity:

Like maybe, people will get the wrong idea of, ‘Oh I have to donate something to the cause’, or something. For me personally if I see something, ‘oh charity, oh no, I need to avoid it’.

Young person, male

Interactive opportunities were more popular than static information pages, and the option of a web chat facility on the Brook home page was enthusiastically focused on by both men and women for two key reasons: the flexibility to contact someone at the patient’s convenience and the arm’s-length method of engagement, which was frequently cited as preferable to meeting a HA:

I personally think the web chat would be like the best. Being able to just go straight on it and just have a professional writing back to you, like with an answer because, obviously, you can search all day and still not have an answer for something. So I think that is the best.

Young person, female

‘The Mix’ page was popular with the young female group and most male interviewees. It was seen as being addressed specifically to younger people, and this targeted relationship was a significant motivator of interest. One young male participant was familiar with this site as it was promoted in his school. Positive discussion focused on key aspects, including its varied content addressing diverse aspects of younger people’s lives (mental health, STIs, contraception, pornography). The home page tabs ‘Your Voices’ and ‘Mental Health’ were each mentioned as a valuable feature, despite their content not being visible in the single page presented to the focus group:

It’s got everything on it. It’s not just banging one thing. It’s got drugs and alcohol and everything.

Young person, male

The home page of the Seventeen Days website, which was shown only to the women’s groups, was swiftly dismissed for the lack of apparent interaction and immediacy of the narrative format and content, and the discussion moved on to the value of getting quick answers to STI-related issues:

Yes, and they [Seventeen Days] could be talking about all stuff that you didn’t want to even discuss.

Yes, you just wanted to know one thing and that could be at the end. Maybe too much to go through.

Young people, female

Triage

The first step of the intervention package is the application of the triage algorithm presented in WP2 (see Chapter 3). To balance clinic resources with sensitivity and specificity, the risk threshold was set to refer approximately 15% of MSM and 5% of young people. The triage was designed to be conducted through the clinic EPR system by any member of clinical staff seeing a patient.

Online material

Feedback from service users indicated that a screening page used to direct users to more tailored online material was acceptable and addressed the desire for services to feel personalised. The proposed links received mixed opinions, so we reviewed the links with further input from the PPI group. The final lists of links included were, for young people, The Mix, MenSS138 and BISH (www.bishuk.com) and, for MSM, THT, GMFA and NHS Choices. Multiple approaches were proposed for advertising and referring service users to the web pages that we recommended, for example sending a link to patients as part of appointment reminder text messages, displaying posters in clinic waiting rooms with the web link and encouraging HCPs to direct patients to the web page during appointments. All of these methods were included in the intervention manual.

One-to-one consultation

The approach proposed to service users during this phase was deemed acceptable in discussions across the wider project team and was not amended based on feedback.

Participants

All MSM and young people attending the clinics during the pilot period were eligible to be triaged and offered the appropriate intervention. Service users who lacked capacity to self-complete the triage (e.g. could not read English) were excluded; however, a member of the research team was present to help with any technical barriers to using the triage assessment tool.

Intervention

The intervention consisted of three components, as described in more detail in Chapter 5 (triage, online intervention, one-to-one intervention). Participants in the intervention clinics were triaged using a risk prediction tool, which was self-completed on a stand-alone tablet computer. The triage result code was printed on a ticket and participants were asked by a member of study staff to give the ticket to the HCP they saw. Based on their risk score, participants were eligible to be offered one or both interventions:

1. High intensity – patients who scored highly in the triage algorithm for sexual risk were eligible to be offered the high-intensity intervention. This was a brief one-to-one session with a trained member of the health-care team (expected to be a HA in GUM clinics). This intervention was designed to be delivered in a single session, lasting up to 45 minutes, and on the same day as the participant’s clinic visit.

2. Low intensity – this was offered to all patients, whether they scored above or below the threshold for referral for the high-intensity intervention. This was a web page designed specifically for the pilot trial, containing targeted SH information that could be accessed either during the clinic visit or at home later (www.santeproject.com – note that this web page is no longer active).

Service users in control clinics received standard care, which could include the offer of a behavioural intervention, for example a consultation with a HA.

In the intervention clinics, the HCP whom patients saw for their appointment could decide whether or not to refer a ‘high-risk’ user to the high-intensity intervention. The triage tool was set to refer approximately 15% of MSM service users and 5% of young people. This meant that MSM with a predicted risk of STI diagnosis of > 24% and young people with a predicted risk of > 22% were classified as high risk, based on the development data set used in WP2. These thresholds were deemed to be feasible in terms of the numbers of service users being referred, and giving a reasonable balance between sensitivity and specificity.

The project protocol and intervention package were presented to all clinical staff in intervention clinics during a routine staff meeting. Participating HAs had a training session (designed to be 2 hours) on the intervention manual, including the use of role play.

Data collection

Triage data were self-completed by service users and were collected using Android tablets through a custom-built ODK Collect (ODK Development Team, version 1.4) form. The variables collected were the same as those presented in Table 6. Anonymised data were uploaded to a secure server. Data about the one-to-one intervention process were recorded in the EPR system in Brighton and on paper forms in Archway. Engagement with the digital component of the intervention package was monitored using Google Analytics (Google Inc., Mountain View, CA, USA) during the pilot period. Process data for the number of service users attending the clinics during the pilot period, when available, were collected within the clinic existing EPR system.

Sample size

Based on historical clinic patient data on attendances provided as part of the GUMCAD data return, we estimated that a total of 5500 eligible patients would attend four clinics during the 1-month intervention period (Figure 11). Based on 70% of patients in the intervention period being triaged, an average of 10% being eligible for the high-intensity intervention, and 60% then being offered interventions, we expected 231 patients to be offered the high-intensity intervention. Assuming that 50% (expected 95% CI 44% to 57%) accepted the intervention, and then 50% completed it, we estimated that 58 patients would complete the intervention. The assumptions for the proportions triaged, offered and accepting the intervention were purposefully conservative to reflect the potential for poor engagement with the trial from HCPs and service users.

FIGURE 11.

Flow chart for patient flow through the clinic.

Analysis

We planned to describe the proportion of patients who went through each phase of the intervention process, comparing clinic types and patient demographics, adjusted for clustering at the clinic level. All analyses were conducted in Stata 14.

Participant selection

We sampled young men and women and MSM who were attending SH services and completed the triage process. We sequentially recruited service users who scored as ‘high risk’, aiming to recruit 24 service users across two clinics. HCPs were purposefully recruited from participating clinics to represent both HAs and other clinical staff, with two focus group discussions planned for each participating site.

Recruitment

Service user participants were recruited from participating clinics. They were approached in the clinic waiting room and given a study information sheet to read before deciding to take part. Participants were offered a £20 high street voucher for taking part. Interviews were either scheduled to take place on the day of recruitment or scheduled for a future time by telephone.

Health-care providers were e-mailed to invite them to take part in a group discussion. When there was not enough interest to form a group, individual interviews were conducted.

Data collection

Interviews were by telephone and group discussions were conducted within the clinical setting. Interviews were designed to last 20 minutes, and the group discussions 45 minutes. Written consent was taken at the point of recruitment for service users and prior to the discussion starting for providers. Interviews and discussions were audio-recorded and then transcribed using a professional service.

Analysis

We used the same analysis methodology as described in Chapter 4.

Participant selection

All young people and MSM attending the recruitment clinics during set time periods were eligible for recruitment. There were no sample targets in terms of demographics.

Recruitment

Participants were approached in clinic waiting rooms and given information about the study by a member of the study team or clinic staff. Recruitment was conducted in specified time blocks at clinics until enough patients consented. The patient’s preference for returning to the clinic or being sent a postal self-sample kit was recorded and patients self-completed information about their age, ethnicity, sexual orientation and contact details. The clinic patient ID was used as an anonymised study ID that could be linked to clinic records for processing results. The process was standardised across intervention and control clinics, and information posters were displayed in all clinics. A total of 75 and 100 patients were targeted for recruitment from control and intervention clinics, respectively (Figure 12 – n2).

FIGURE 12.

Sample size for patient follow-up at 6 weeks. a, n2 = total for all participating clinics; therefore, 400/4 = 100 in intervention clinics and 300/4 = 75 in control clinics. (a) Intervention clinics and (b) control clinics.

Data collection

Recruitment data were collected using Android tablets, through a custom-built ODK Collect survey form. Encrypted data were uploaded to a secure server. The follow-up web survey was created using SnapSurvey (Snap Surveys, London, UK), with no identifiable information requested. Patients were e-mailed the survey link up to three times. For the follow-up STI screen, patients either posted a self-collected sample, which was then processed by The Doctors Laboratory (London, UK), or returned to the clinic. The results were e-mailed to a NHSMail e-mail account, and patients were sent negative results via text message from a study telephone number. Any positive results were sent to the study co-ordinator at the recruiting clinic for follow-up and treatment in accordance with local protocols. All follow-up data were entered into a Microsoft Access 2013 database and processed using Stata 14.

Analysis

We described the frequencies and proportions, by clinic, age group, sex and ethnicity for each stage of the follow-up. Proportions were compared using chi-squared tests and multivariable logistic regression.

Triage

The original proposal aimed to embed the triage tools developed in WP2 within the EPR systems of clinics for the pilot trial. We discussed this proposal with two different EPR software providers [MillCare (Belper, UK) at Brighton and RioMed (Eastleigh, UK) at MMC and Archway], but these companies were unable to implement the triage within either of these systems.

In Brighton, the feedback from the EPR provider was that the implementation of a triage tool as we developed would be feasible, but would take considerable coding and configuration and a minimum of 6 months’ development and testing. Therefore, we did not pursue adaptation of the EPR at Brighton because of the time constraints. It was also noted that the EPR provider raised concerns over our approach, as, despite the triage tool parameters having already been collected within the EPR system, the real-time processing would be computationally intensive (e.g. deprivation quintile is derived from a patient’s postcode).

At MMC and Archway, the EPR provider indicated that the implementation of the triage within the system was achievable and anticipated 3 months to develop and pilot the system. Concerns regarding the coding or configuration were not raised. Over an 8-month period the EPR provider developed three iterations of the triage; however, none of these was deemed practical, with issues of double data entry, burdensome navigation through the patient record and the removal of compulsory fields. Therefore, we were unable to pilot test the triage within the EPR system at these sites. This indicates that, were this to be pursued, a programme of software development would need to be supported and funded by the NHS provider.

For the purposes of the pilot trial, we developed a stand-alone Android tablet-based system using open-source software (ODK Collect). This altered the patient pathway from that originally proposed: the triage tool was self-completed by patients prior to their appointment, rather than being conducted as part of the consultation with a HCP. This system was in place for 1 month before the pilot began so that it could be integrated into the clinic and to allow providers to become familiar with the study protocol.

The triage tool was completed 1064 times, representing 16% of patient attendances during the pilot trial period. Sufficient information to complete the triage process was provided by 1030 (97%) patients, of whom 612 (59%) were either young or MSM. As study staff asked service users to complete the triage prior to their clinical appointment, we were unable to distinguish our target groups and, therefore, aimed to have all patients take part. In both settings, a higher proportion of young people than of MSM was triaged (Table 23).

The triage tool was set to a risk referral threshold that we had planned would identify 5% of young people and 15% of MSM to be at high risk, based on the model performance from WP2. However, at both sites, the triage process identified a considerably higher proportion than this (Table 23). Notably, there was a higher proportion of high-risk young people at the Archway clinic and a higher proportion (and attendance in general) of high-risk MSM in the Brighton clinic. These risk profiles of the two clinics is not necessarily surprising and probably reflects that the demographics and sexual history of participants in the pilot trial differ from those of participants included in the development data set. There was also a lower number of missing sexual history and behaviour data in the pilot than in the WP2 data set, and this may have changed the triage tool’s performance. The threshold for defining a high-risk patient could be adjusted for clinics according to the capacity to provide a high-risk intervention, but this poses specific challenges for a standardised approach to triaging.

Santé web page

During the pilot period, the project web page was advertised in four ways in both pilot clinics: posters displayed information about it in the clinic waiting rooms, the web address was printed on the triage tickets, HCPs in the clinics were informed about the web page and it was printed on the action plan cards given during the one-to-one sessions.

A total of 24 unique users visited the web page during the pilot, out of a potential 6805 patients (0.4%), with one visitor returning. On average, visitors stayed on the web page for 38 seconds, and stayed on the home page. Two users accessed the web page using a mobile phone, and all other site visits were from computers. Figure 15 demonstrates the age and gender of the site visitors.

FIGURE 15.

Age and gender of the Santé web page visitors (derived from Google Analytics). (a) Age, 100% of total sessions; and (b) gender, 100% of total sessions.

One to one

At Archway, one HA agreed to work additional bank shifts during the piloting period, meaning that we could deliver the intervention package 2 days per week during May 2017. In Brighton, the HAs implemented the intervention as part of their routine practice from 2 May to 2 June 2017. Training was delivered in two brief (1-hour) sessions at Brighton with the whole HA team present, and in a single session at Archway (1 hour). The training included an overview of the study protocol, an explanation of the intervention manual, and brief role-play exercises using the five steps of the one-to-one consultation and action plan cards.

In Brighton, the EPR system was adapted to include information about whether or not the patient had been triaged, referred for the one to one and accepted. Of the 552 eligible patients who completed the triage process, records for 168 (30%) were entered into the EPR system, with this proportion being slightly higher for young people than for MSM (31% vs. 24%, respectively). A total of 21 (0.6%) potentially eligible patients (see Table 23) were recorded as having completed a one-to-one session; this represents an average of one patient per day. However, it is possible that more patients completed the one-to-one but were not recorded in the EPR. In the HA comments recorded within the EPR system, sessions took 5–15 minutes, but this was reported in only a few records. More general comments on the content of the sessions, of which examples are given below, showed that they covered a range of topics and that patients had a range of motivation around behaviour change:

Has made patient realise that it is important to know potential partners well and use condoms, especially in early stages of a relationship.

Young person, Brighton

Triaged wrongly as high risk; one episode of UPAI [unprotected anal intercourse] for which he accessed PEP. No other UPAI. No [chemsex]. Consistent condom use and informed.

MSM, Brighton

Discussed risk taking as part of PEP discussion. Patient acknowledges risk behaviour and it is usual for him to make decision of knowing partners. Due to alcohol this did not happen this time. Patient not choosing to make changes to this behaviour at this time.

MSM, Brighton

At Archway, of the 60 eligible patients who completed the triage, 19 (32%) were high risk and four (7%) completed the intervention. The four sessions were recorded as taking approximately 30 minutes each, but the intervention steps and action planning were conducted in only 3 minutes. The content of the sessions was reported as including condom use, peer pressure around risk behaviours such as drugs and contraception with young people, and PrEP, PEP and general STI knowledge in MSM. Among those who were triaged as high risk but did not attend the one-to-one session, reasons for non-attendance included the patients not waiting for the appointment, issues with the referral process in the clinic and HCPs not feeling that a referral was warranted.

Triage and referral to the one-to-one session

All the participants were recruited in participating clinics following their completion of the self-triage; participants indicated that the triage process was acceptable, although many appear not to have understood its purpose. Overall, they felt comfortable with the questions being asked (i.e. demographic and recent sexual behaviours) and found that filling in the questions on the tablet computer was straightforward. One of the main advantages was that the tablet was felt to be quick to complete and could save time. The discretion afforded by self-completion was also mentioned, but not as a key issue.

Although one participant raised a concern about whether or not the questions took all the relevant information about sexual risks into account (MSM, > 50 years), no one objected to being referred to a HA based on the triage tool. However, although conceptually acceptable, there was little evidence that the ticket-based system we piloted in lieu of an integrated EPR-based system was feasible. Only half the participants we interviewed gave the ticket to the HCP they saw – reasons for this failure were confusion about an unfamiliar process and forgetting, for example ‘We’re a bit distracted when we’re there’ (MSM, > 50 years).

There were mixed experiences among those participants who successfully passed their triage ticket to the HCP. For example, a young female participant (25 years old) came in for contraception and did not expect to talk about sexual risk. Some patients felt that the triage ticket prompted a helpful discussion about new sexual partners and could result in them agreeing to have a STI test. Others, however, reported that their HCP did not seem to want the ticket or discuss sexual risk, whereas others spoke briefly with the HCP about sexual risk: ‘I gave it to, like, my doctor, and she, did a little talk on sexual health and stuff’ (female, aged 22 years). Patients who did not give the tickets to their HCP generally reported discussing sexual risk with their HCP regardless.

Many of the patients who had come for contraception or routine check-ups neither expected nor were offered any health promotion interventions, such as leaflets or referral to a HA. Among patients who were referred to a HA based on the triage tool, reasons for accepting a referral varied, including the opportunity to talk about PEP or to have a general discussion about SH and risk behaviours:

I found it reassuring that there are services in place that would be, looking out for potential relapses with people that come into that clinic.

MSM, 25–50 years

I think that was good because I feel, if I hadn’t given her the slip, I might not have been told to go to the health person. I think that was good for me, just to speak to someone else as well.

Female, 18–22 years

Santé website

Although there was no indication that patients found the idea of the Santé website unacceptable, only one of the interview participants (who had also attended the HA one to one) had visited the web page. The young woman who visited the web page (18 years old) said she went to the website for additional information and was happy to do both a one-to-one session and visit a web page. Other participants reported wanting to find specific information from the internet; a young woman (21 years old), for example, had searched to find out more about her treatment but had not visited the intervention website.

A few participants indicated that they might check a website on their mobile phones while waiting for their appointments but would not go back to it afterwards. However, none of the participants had visited the website using the link advertised on posters in the clinics or the tickets, and most had not noticed the information printed on the ticket. Barriers to this were poor understanding that the ticket contained relevant information and not having the skills or facility to scan the QR code on their mobile phones. One young woman stated that she might have been interested in checking the website but the HCP took the ticket from her. Participants reported not seeing the posters among all the displayed information and being too distracted or busy reading or talking while they waited. Texting the web link (e.g. with the appointment reminder) was suggested as a more effective way to promote a website.

One-to-one session

Very few of the participants we interviewed were referred to see the HA as intended. Participants who had previous reported experience of one-to-one sessions typically saw them as worthwhile, giving them motivation to change and even resulting in behaviour change:

It was something that I was already thinking but it just pushed a bit more.

Female, 20 years

Participants were generally open to the idea of a one-to-one session with a HA. Only one young woman (aged 21 years) said that she was always asked to see a HA about her drinking and consistently declined as she does not consider it to be a problem. Some participants wanted a clear reason for seeing the HA and would be motivated to ask to talk to someone if they had specific questions or concerns. Others thought that they would accept the opportunity to talk to a HA if it was recommended to them, although this sometimes raised anxious concerns:

I think I’d be a bit nervous as to why they recommended me to one, but, like, I would go. If they’re recommending me to go see one, then I would.

Female, 17 years

Other barriers to attending a HA referral included time and the gender of the HA; for example, a young woman (aged 25 years) said that she would be more inclined to talk to a HA if it was a quick discussion, and wanted the HA to be the same gender as her.

Triage

The value and effectiveness of the triage algorithm was questioned by most participants, who generally considered this to be a ‘blunt’ and diluted alternative to the more sensitive interpersonal, face-to-face risk assessment. Among HAs there was suspicion and lack of confidence in the capacity of the triage tool to accurately predict risk or need:

As a system, I don’t think it works and it’s not the same as having someone in front of you with it and then, kind of, ascertaining the best way for using instincts.

Brighton, Doctor

I think it’s a completely different kind of experience and I think it depends what your measures or outcomes are, because I think sometime we’ll see a person in clinic, they’ll have a really good experience with us, it will be a really human experience and actually they might then come back a month later and tell us about something completely different that they hadn’t mentioned at the first visit because they feel safe, they’ve had a good experience. Whereas that is quite a, sort of, impersonal . . .

London, HA

The HAs’ experiences of the pilot led them to feel that the tool had sometimes referred patients who were subsequently assessed to have not needed referral and failed to refer patients whom they subsequently identified as high risk. Conversely, the Brighton HA group also stated that many of the patients referred via the triage tool were patients from the waiting room who had already been booked in to see the HAs. This was seen to both assert the effectiveness of the triage tool in identifying appropriate patients and undermine the purpose and value of the tool:

And I think we’re picking up the ones anyway that would engage, so they’re the ones that, kind of, tend to be known to us and they engage and they’re getting through to us in other ways. And the people that won’t engage are the people that won’t engage, no matter what triage system you’re using.

Brighton, HA

There was a further challenge to this method in the routine practice in Brighton, where all individuals < 18 years old are routinely screened and receive a thorough assessment. This raised questions regarding the context of the pilot, as the role of an additional triage assessment was not clear:

And with the young people they have an assessment anyway, really thorough from a safeguarding perspective, so all under 18s and that bring up anything and everything to do with risk. So that’s going to happen anyway, when they’re in the room. So, I don’t think you need something on top of that, with regards to young people, do you?

Brighton, HA

Several advantages or opportunities offered by the triage tool were also discussed, including the potential for the triage tool to enable patients to highlight a risk profile or other issues that they might not feel able to raise face to face. This private disclosure was seen as avoiding a potentially embarrassing discussion with a member of staff:

So, what I’m saying is they may say something different to the consultant, nurse, whoever they see, and they may be saying what they really feel [overtalking]. If you see what I mean? Because sometimes it’s easier to do this than it is to talk to someone face to face. So, it’s tricky, yes, you’re going to get that I think.

London, HA

The Brighton HA group also revealed that the piloted triage tool was potentially more effective than the current process of asking patients to tick key risk assessment questions on a form at reception.

Another potential advantage of a systematised triaging that the London group proposed was the process and structure that it could provide to new members of staff who might lack confidence and skills. The systematised triage was thought to prompt staff to both identify and refer patients who come in for non-STI issues but who have risk profiles that would otherwise go unassessed:

I suppose a good thing is that it might make a clinician think more about a patient, particularly if they’re repeat attenders for something, and they think that they’re really settled and stable and don’t need any intervention, this might actually pick it up. So, it might make people explore things more as opposed to thinking, ‘Oh, this person’s a repeat attender for something,’ it might make them look into it more.

London, HA

I rarely see the contraception service patients even though we’re supposed to be integrated service. So, I would say it really helped clinicians to identify that. But the whole assessment was still down to them: do they feel this person would need to be referred?

London, HA

Santé website

We had limited feedback on the Santé web page as consenting participants either were not aware of it or had not used it. At least two of the Brighton HAs were not aware that a Santé website existed, and none of the others made reference to having used the Santé website. There was some brief discussion about the potential value of referral to websites:

. . . there are some aspects or some topics maybe that people feel OK to do online, but I think in the area that we’re in, in sexual health it’s such a sensitive issue . . . it’s quite a vulnerable position to be in to start talking about . . . So, when it comes to online, I’m not sure how that will translate, but for some people that might be just what they want, because for them, maybe talking doesn’t help them or they don’t feel it benefits them, and it might be a starting point for some people online.

London, HA

One-to-one sessions

This feedback focused on the HA experience, as the non-HA staff had little experience of the intervention. HAs from Brighton and the London HA interviewee made conflicting comments about their use of the one-to-one sessions, stating that ‘It’s what we do anyway,’ but also commenting that the manual provided useful structure and format for these interventions. HAs in Brighton also discussed their existing use of MI as part of their routine work, although they acknowledged the limited amount of MI training that the HA staff had received. Much of the training had taken place many years ago, and HAs subsequently asked the study team for additional training in MI.

Two of the Brighton HAs stated that they were concerned about the limitations of the intervention manual and felt that it repeated what was already part of their routine practice. They had therefore ignored it (‘I must admit I didn’t really use it’), whereas another Brighton HA participant referred to the value of the manualised MI approach for the structure it provided:

I like the part when it gets to the work, I like the MI, I like just being able to, kind of, code or prove what you’re doing or, I’m not saying that very well, but measuring what we do somehow.

Brighton, HA

The London HA interviewee in particular valued the manual for its structure and the focus on goal setting, which they felt guided their approach and focused attention on action planning. They also suggested that the focus on action plans provided structure for patients too:

. . . just having to do it 100%, thinking about it 100% and doing it 100% was more how it, kind of, focused the consultations, so that it helped to elicit three goals that the person wants, so, it really made the consultation more structured actually . . . I think in the normal run of the clinic that wouldn’t be the same way which we would deal with it. So, it, made it more structured for me personally.

London, HA

Although the London HA welcomed the use of the credit-card-sized action plans, none of the Brighton HAs had used the cards or appeared to be aware of them. As with attitudes towards the manual, the Brighton HAs described a tension between routine practice already being similar to the intervention process and the action plans not being acted on:

It’s what we would do anyway. The couple of times that I have done it I’ve forgotten to do an actual action plan, because I wouldn’t necessarily do that.

Brighton, HA

Again, the London HA interview identified advantages in the structure that the intervention and action plans provided, and related these to both the HAs and the patients:

I thought it was a very comprehensive tool, I really did, and a very comprehensive tool, it just, highlighted the bits that you needed to do; and, as I say, the action plan I think was really for me the best part of it.

London, HA

Implementation barriers

A number of implementation barriers were raised during discussions, including the ticketed triage process and limited staff training. The use of tickets was seen to be a key ‘leakage’ point at which both patients and staff lost potential referrals. The tickets, which were a workaround in place of the initially planned integrated EPR system planned, were seen as confusing to patients, and they were lost, abandoned or ignored by patients and were lost or forgotten by staff. The use of these tickets was introduced as a workaround because of the delays and technical barriers to adapting EPR systems, and this was seen to have introduced several significant barriers to the pilot trial implementation. Integration into the EPR system was identified as a valuable solution that would have avoided many of these ‘leakage’ points:

I found them screwed up on the lab floor, stuffed in between people’s notes, sometimes in the wrong place. Or people just randomly putting notes on my desk, you know, ‘Do I give this to you, what do I do with it?’

Brighton, HA

I think, yes, people who would’ve been referred to me . . . basically, because there were few issues with just getting everybody understanding what to do with that bit of paper.

London, HA

The engagement of doctors, nursing staff and HAs in briefings and training about the conduct of the trial appears to have been inconsistent, fractured and of limited impact. Although all participants said that they found ways to make things work, patients were lost to the study because of confusion about the triage tickets, and the intervention manual was not always read or properly applied by HAs. Among the HAs there was considerable variation in the engagement with the trial, and questionable skills/capacity of HAs to administer a MI intervention. The capacity of many HAs to deliver MI was questioned by some HAs themselves and by participants in the London group:

I think MI is often talked about as something that everyone does, and I think everyone can do it but equally it can be a very filtered, watered down version of what could be most useful. So I think it’s good to make sure if you are offering it, that it’s being done by somebody that’s really experienced in doing it.

London, Psychologist

Although most staff in the two pilot clinics had been informed about the pilot’s aims and objectives and all HAs had been through some training on the intervention manual, concerns were raised about the effectiveness of the briefings and training. In the Brighton discussions, a senior nurse had joined the clinic after the pilot had begun and highlighted the necessity of effective, ongoing introductions for new staff. This was also recognised to be an issue for junior doctors, who may arrive after initial briefings:

I think if the staff that are seeing them [patients] have a grip of the basic principles of the study, that’s more persuasive than them just going, ‘Oh, I’m not really sure it’s something to do with offering you an intervention if you’re a high risk.’ There’s consistency, isn’t there? Everyone saying the same thing, in the same sort of way.

Brighton, Doctor

The triage and intervention were generally accepted as viable in these busy clinic environments, and the trial appeared to have had only a very limited impact on the day-to-day clinic work. Even HAs appear to have found the trial and intervention to have been acceptable within their workload, although this may be related to their opinion that most referrals were of patients they would already have expected to see and the limited numbers of triaged patients who actually made it through to HA appointments. The Brighton HA group suggested that the ‘Brighton Express’ clinic for those aged < 18 years, which provides routine and thorough face-to-face assessment of the needs of younger patients, would have been negatively affected by the trial.

Web survey

A total of 990 e-mails were sent out to 389 service users. Thirteen (3%) e-mails bounced as the e-mail address provided by the service user was incorrect (see Figure 16). Of the 376 service users who received the link, 133 (35%) completed the survey, and an additional 17 (5%) participants started but did not complete the survey. On average, participants responded to the third e-mail reminder and took a median of 2 minutes to complete the survey.

Young people were significantly less likely to complete the web survey than MSM (OR 0.39, 95% CI 0.25 to 0.61), and the overall completion rate ranged between clinics from 36% at Archway to 19% at Croydon. Table 26 presents factors associated with survey completion in MSM and young people.

Among young people, women were more likely to respond to the web survey than men [adjusted odds ratio (aOR) 4.35, 95% CI 1.59 to 11.88]. Participants recruited from Croydon were significantly less likely to respond, even when the age, gender and ethnicity of the participant were taken into account (aOR 0.22, 95% CI 0.07 to 0.66). In MSM, no demographic factors were identified that predicted whether participants were more or less likely to respond to the web survey, although there were trends towards older and white MSM responding. Interestingly, MSM from Croydon had the highest response rate.

Sexually transmitted infection screen

We offered two options for the 6-week follow-up: returning to clinic for a ‘quick check’ or being posted a self-sample kit. Typically, ‘quick check’ STI screening appointments involved the completion of a very short questionnaire, self-collection of samples and minimal interaction with clinic staff. In some of the clinics we were unable to offer a ‘quick check’ appointment because of limitations with the booking system. Of the 406 participants recruited, 385 opted for the postal self-sample kits and provided address details. The return rate was higher among those who were sent kits (27%) than among those who opted to return to the clinic (3/21, 14%). Among the 108 participants who were successfully screened, there were two positive tests for chlamydia and no gonorrhoea diagnosed. We aimed to send the self-sample kits at 6 weeks following recruitment into the study; on average samples were returned at a median of 9 weeks (IQR 8–11 weeks).

Young people were significantly less likely to return the self-sample kit than MSM (OR 0.45, 95% CI 0.29 to 0.72), and the overall completion rate ranged between clinics from 34% in MMC and Chelsea to 15% in Darlington. Table 27 presents factors associated with survey completion in MSM and young people.

Among young people, the only factor significantly associated with completing a STI screen was being female (aOR 3.32, 95% CI 1.32 to 8.38); the association with being older was borderline. Among MSM, although no associations were statistically significant, there was again a trend towards older MSM being more likely than younger MSM to complete the screen (63% in those > 45 years vs. 0% in those aged 16–20 years).

Individual randomisation

Elements of the intervention package could be well suited to individual randomisation, such as the triage being randomly applied to different patients. However, there are still concerns about contamination between the intervention and control patients because of the service-wide nature of the intervention. Employing study staff to deliver the intervention could mitigate this risk but would have cost implications, and this implementation method would need piloting.

Step-wedge roll-out

As the intervention package was in principle acceptable and used evidence-based elements, the intervention package could be routinely implemented within clinics, if commissioners agreed to support it. A stepped-wedge trial, which did not involve randomisation or the need for the level of standardisation that a RCT would require, could allow for some adaptation of the intervention within each clinic. If this was combined with changes to GUMCAD, as currently being implemented for the Impact trial of HIV PrEP,146 then the outcome measures could be collected as part of routine data collection. This would allow for an evaluation of real-world implementation. However, this approach would be unable to establish the effect size, and requires funding agreement from commissioners.

We did not conclude that either of these alternative designs would be feasible for a trial within existing resources. Either would require additional developmental work and piloting.

Digital intervention

A digital risk reduction intervention would need to be developed or adapted from one of the trialled interventions for which materials are available. Following the development process, additional piloting work would be needed to improve and incentivise engagement, for example in terms of the processes used in the pilot (e.g. specific text message promotion).

One-to-one session

One concern with the one-to-one session, which was supported by our pilot work, is that because it is similar to HA’s current practice there may be issues with intervention fidelity. Furthermore, different SH providers have different levels of training. Some MI training is required, but it is not universally provided, even to HAs. Extensive training was considered unfeasible within current clinics’ resources for the pilot. We designed the training to be pragmatic within existing clinic resources, assuming a baseline level of MI experience among staff, which was not always found. Additional work would be needed to evaluate how well this training module could be implemented, the gaps it would leave and how well those HCPs implemented the intervention session, rather than defaulting to their usual practice.

Electronic patient record-based triage tool

Although implementing the triage tool within clinic EPRs seems feasible, we were unable to actively demonstrate this within the pilot. A key challenge in this was communicating how the triage should be presented in the front end of the system (i.e. what the HCP interacts with). Therefore, further work that includes the participation of clinical staff and engages with multiple different EPR providers would be needed. An important aspect of this would be how to standardise, to a sufficient level, both the ways in which the data are captured and processed and the end-user experience. In addition, further work on the external validation or refinement of the triage tool is needed, using more complete data to demonstrate whether it could be usefully rolled out either in clinics or in online pathways.

Follow-up

Because of the relatively poor follow-up rates, further investigation and piloting of different methods (e.g. telephone call reminders) and potential incentives (e.g. vouchers for samples returned) to improve follow-up rates would be needed. Specifically, these would need to assess whether heterogeneity in follow-up increased or decreased by location and type of clinic, as well as the service user demographics.

Economic evaluation

We had intended to estimate the cost of delivering the intervention package as part of the pilot trial but were unable to collect the data we needed to do this. At the outset, it was envisaged that the intervention would be delivered within existing resources by reallocation of staff time, in particular from existing work with patients. As such, an economic analysis might be less useful. However, it was clear that, to deliver the intervention, existing resources were not sufficient and an economic evaluation is needed. One of the main barriers we faced in conducting the pilot trial was delivering the one to one, and having a clear understanding of what resources a clinic would require to deliver the intervention package could have improved our ability to make conclusions on feasibility.

Dissemination

Further communication between service providers, commissioners and service users is needed if the proposed intervention approach is to be trialled or undergo further piloting. A barrier we faced was the disconnect between service providers’ and users’ preference for risk reduction and what commissioners were prioritising for funding. We plan to disseminate the findings from this project to both service providers and commissioners, through this report, academic publications and conference presentations, to encourage this communication.