The relative clinical and cost-effectiveness of three contrasting approaches to partner notification for curable sexually transmitted infections (STIs): a cluster randomised trial in primary care

The null hypothesis

Provider referral and contract referral offer no advantage over patient referral alone in PN for curable STIs in the primary care setting.

In order to answer the research question for the main trial, the following objectives and outcomes were established.

Primary outcomes of the randomised controlled trial

• Number of partners per index patient treated for chlamydia and/or gonorrhoea/non-specific urethritis/pelvic inflammatory disease.

• Proportion of index patients testing negative for the relevant STI at 3 months.

Secondary outcomes of the randomised controlled trial

• Number of partners per index patient presenting for treatment.

• Proportion of index patients having at least one partner treated.

• Number of main, casual and ex-partners per index patient tested for the relevant STI.

• Number of main, casual and ex-partners testing positive for the relevant STI.

• Number of current partners tested for a human immunodeficiency virus (HIV) infection by 3 months.

• Number of index patients tested for a HIV infection by 3 months.

• Time to definitive treatment of index patient for the relevant STI.

• Time to definitive treatment of current partner for the relevant STI.

• Uptake by index patients of ‘contract’ and ‘provider’ referral for one or more partners, within the relevant randomised groups.

• Patient-related factors impacting on PN or STI disclosure to main, casual and ex-partners.

Target population

The target population was as follows:

• all 16- to 24-year-olds testing positive for chlamydia after chlamydia screening

• all patients (aged ≥ 16 years) diagnosed with a curable STI following clinical presentation

• all patients (aged ≥ 16 years) diagnosed elsewhere and attending the practice for antibiotic treatment.

Exclusion criteria

• Any patient aged < 16 years.

• Any patient unable to fully understand the trial information.

• Any patient who required translation of trial information.

• Any patient who required advocacy.

Setting

The setting of the trial was a diverse sample of primary care practices, both in specific localities [primary care research network (PCRN) practices in the South East region of England] and nationally [recruited via the Medical Research Council (MRC) General Practice Research Framework and the PCRN].

Patient and public involvement

A focus group of six sixth-form college students were consulted regarding their opinion on the participant information resources and recruitment methods. In addition, two University College London medical students who were members of the ‘Sexpression’ group (a network of student-led projects that teaches about relationships and sex education in local communities across the UK) agreed to sit on our Trial Steering Committee and advise on participant information resources and recruitment methods.

Approach to recruitment

Participants were to be recruited through letters of invitation, and personal invitation on attending the practice as follows:

1. Practices were asked to mail out letters to 300 randomly selected 16- to 24-year-olds at the beginning of the pilot (taken from the practice list), inviting them for a chlamydia test (as part of the NCSP). The invitation letter also mentioned the trial.

2. Clinical staff (usually practice nurses) approached potential participants, either (a) at the time of first attending the practice either with symptoms of a suspected or presumptive STI (symptomatic patients) or (b) at the time of chlamydia testing (asymptomatic patients, most of whom are tested as part of the NCSP) (Figure 2). The clinic staff were asked to explain that the practice was taking part in a study, as part of which additional assistance might be given to patients needing PN for a STI. It was explained that the practice had been randomly allocated to a group which would help either with providing care for STI patients up to the recommended national standard or with additional options for support. Patients were told the allocation of the practice. It was also explained that, should the patient be diagnosed with a STI, participation in the trial would mean, in all cases, that they would be contacted by an experienced HA. This HA would, depending on the trial allocation of the practice, assist them in their plans and actions to inform and obtain care for both current and ex-partners. Patients agreeing to be recruited at this stage provided personal contact details through which the study HA could communicate with them.

FIGURE 2.

Process of enrolling a participant.

Randomisation

Cluster randomisation at practice level was chosen for two reasons. First, there was a strong likelihood that clinical practice would be influenced by participation in the trial – with randomisation at patient level, practitioners who considered that ‘provider referral’ or ‘contract referral’ had advantages might more readily suggest that patients randomised to ‘patient referral’ attended a specialist clinic. Second, randomisation of patients who attend unpredictably in the middle of a busy surgery is more challenging than randomisation of patients with chronic disorders, while practice randomisation reduces this difficulty.

Consent at time of test versus consent at time of diagnosis

Two approaches to consent were piloted: consent at time of test (CAT) and consent at time of positive diagnosis (CAD). It was anticipated that CAD would reduce the workload for the health professional. However, a previous failed trial demonstrated the difficulty of achieving recruitment at the same time as a positive diagnosis is communicated to the patient. 32

We obtained written informed consent from all adults (aged ≥ 16 years). Where written informed consent was not possible (i.e. the patient had been tested for chlamydia, but no researcher was immediately available to take consent), consent was taken over the telephone soon after the patient had taken the test. In this case, potential participants were asked to agree to a researcher calling them to take consent. A ‘reason-for-test’ form included information on why patients were testing, whether or not they were interested in participating in the trial, contact information and signature (if they were happy for consent to be taken over the telephone).

Data collection

There were a number of data collection points and Table 1 summarises the data captured from each stage of the research process. The data specification was adapted from the accelerated partner therapy (APT) trial specification in order to allow for direct comparison between trials. 33

Substudies

Three substudies were included in the PN programme to enhance the trial. We proposed to explore patient-related determinants of PN with a view to advising on targeting different approaches of PN. The economic evaluation proposed to determine the cost-effectiveness of alternative methods of PN, and the mathematical modelling study proposed to address the problem of the right skew in the number (and nature) of sexual partnerships.

Setting

Six pilot practices, accounting for approximately 10% of the sites required for the main trial, were recruited to take part in the pilot. These practices were identified through the MRC General Practice Research Framework register and PCRN South East as having a sufficient population of young people in the target population. Three practices were from the South East, one was from Yorkshire and Humberside, one was from the East Midlands and one was from the South West. The three practices piloting each approach were randomised to patient, provider and contract referral arms. In addition, two approaches to consent were piloted: CAT and CAD.

Target recruitment

Based on the original sample size calculations for the main trial we aimed to recruit 25 patients per year per practice. It was, therefore, estimated that it should be feasible to recruit 5–10 patients per practice over 3 months. This would require writing invitations to 300 patients in each practice. Each practice sent invitations to approximately 300 16- to 24-year-olds, who were selected at random from the practice register.

Patients were invited by letter to take a chlamydia test at the practice and the trial was mentioned in the invitation letter. Sixteen- to 24-year olds who were attending the practices for other reasons were to be invited to test opportunistically, as were patients who were aged ≥ 16 years who presented with a symptomatic STI or had been diagnosed elsewhere and were attending the practice for treatment.

Process

Resources

Practices reported that the training session was well organised and training materials were comprehensive and easy to use. The web tool was felt to be simple and easy to use, and we did not experience any problems with incorrect data input from practices. However, staff did report that they often forgot to invite patients to screen and that there was insufficient time to approach patients during appointments, in addition to the reluctance described above.

Management

Five out of six practices reported very little interaction with the local NCSP staff and office. They were not aware of NCSP targets and there was some misapprehension that the trial team were part of the NCSP.

Practice nurses reported that it was difficult to engage their colleagues in the trial and to promote chlamydia screening throughout the practice and that often they were working alone. They reported that very few 16- to 24-year-old patients were seen in their practices, and that they did not see many chlamydia cases.

We were unable to assess the web tool as a PN management tool, as we did not have sufficient numbers of diagnosed patients to be managed through the pathway.

Scientific

In developing manuals in consultation with the HAs who would provide PN, it became apparent that it was hard to define a process of contract referral in a way that was clearly distinct from provider referral. HAs described their practice as a process of negotiation with the patient, leading towards the choice of a mode of PN.

Because insufficient numbers of patients were diagnosed with chlamydia we were unable to assess the PN pathway.

Methods

Since our application for funding, the NCSP had moved on in its policy and practice. Given the degree of local variation in its implementation, we were keen to identify novel and effective approaches to increasing testing rates. We therefore undertook a literature review, aimed at identifying alternative approaches to increasing STI testing in general practice and generally optimising our approach to recruitment.

Based on our experiences in phase 1, the review focused particularly on interventions that could improve chlamydia testing/screening rates for young people up to 24 years old in general practice or similar settings; the use of incentives (for patients or health-care staff) to improve testing rates and/or recruitment in the field of sexual health; clinical pathways used to organise the testing of young people for STIs in general practice or similar settings; general practice staff experiences, behaviour and attitudes to STI testing; and the use of general practice by young people.

Three broad sources of evidence were reviewed, of which the findings are presented in Table 6. The review was targeted, but did not take the form of a formal systematic review.

First, we studied a review published in 2006, which summarised evidence for various elements of chlamydia screening and testing programmes relevant to the UK health system. 5 Second, we searched PubMed for peer-reviewed publications since 2005 on chlamydia testing and screening, using relevant search terms derived from the review in order to identify recently published settings. We reviewed full-text versions of original research relating to the UK and settings where general practice has a broadly similar role (e.g. Australia). Third, we reviewed policy documents, newsletters and conference abstracts focused on England, which described the implementation of chlamydia screening and testing in the general practice setting and included evaluation of effectiveness. The studies we identified are summarised in Table 6.

Findings

Expert advice from a marketing specialist

As a result of the literature review and poor uptake of testing in phase 1 of the trial, we felt we should explore the possibility of using incentives for the trial. In addition to the literature review summarised above, we spoke to a marketing specialist with substantial experience in running national campaigns, marketing and public relation projects. He suggested that we consider incentives for both the patients and practice staff as described in the sections following (Harvey Atkinson, University of Brighton Students’ Union, 2011, personal communication).

Survey of young people

We also conducted a short survey over a 2-week period by snowball recruitment using a social networking site, members of a youth group, and members of a sexual health-oriented group (Sexpression) to explore incentives for young people. A short questionnaire, consisting of nine questions, was posted online for young people aged 16–24 years to access and respond to anonymously (a hard copy was also made available).

One hundred and seventy-three young people took part, the majority of whom were in full-time education (74%, n = 128). Twenty per cent were not in any form of education or training (n = 34) and 6% were in part-time education (n = 11).

A large majority of respondents (83%, n = 144) said that they would take a chlamydia test if a doctor or nurse asked them to during a routine appointment. This was more than those who said they would if the receptionist asked them to test (61%, n = 105) or if self-test kits were available in the reception (66%, n = 115). Of the 17% (n = 29) who would not take a test, 38% (n = 11) said they would change their mind if offered an incentive. One participant who would not take the test or take part in the study was concerned that sexual health issues would appear on their medical record and they also expressed concern that ‘you don’t get paid to see a doctor so why get paid for this?’ This response demonstrated potential barriers to using incentives.

We asked all participants which incentives would make them more likely to take a test and participate in the study. One hundred and seventy-one participants replied to this question, of whom 42% (n = 72) said that incentives would not make a difference. Of the remaining 99 participants, 88 (89%) said that a cash incentive would make them more likely to take a chlamydia test and be part of a study, 64 (65%) said that shopping vouchers would, and only 13 (13%) said that mobile phone credits would make a difference to their participation (respondents were able to select more than one option).

When asked the lowest value of incentive that would make them more likely to participate, 39% (n = 58) of 172 participants still responded that no amount would make a difference, although this was less than for the previous question. A £5 incentive was the most popular option among those who said that an incentive would make them more likely to participate, with 34% of participants choosing this response (n = 58). Seventeen per cent of participants replied that £10 would be the minimum they would accept (n = 29), 6% (n = 11) chose £2, and 4% chose £1 (n = 7). The majority of respondents (76% of 167, n = 127) stated that they would prefer all participants to receive the same incentive, rather than a small incentive to take part and the chance to win a much larger prize (lottery incentives).

In conclusion, over 40% of the sample stated that incentives would not make a difference to their decision whether or not to participate in the study, and more than 80% of respondents would participate if asked by their doctor or nurse. This suggests that incentives for patients may not be an effective way to boost testing and recruitment. It should be noted, however, that this group included a high percentage of students still in full-time education, which may have biased the results, as they may not be representative of all 16- to 24-year-olds.

Practice interviews

During phase 2 of the pilot, members of the trial team visited every practice in the pilot to establish the practice staff’s experiences and views on barriers and attitudes to recruitment. The practice staff found the trial materials and web tool easy to use. However, there appeared to be a number of myths held by the practice staff which affected their ability to test for chlamydia. Most stated that they saw very few young people in their practice, and it was generally believed that chlamydia was not prevalent in their area. There also appeared to be no systematic way of identifying potential participants within the practice and some staff were unclear about the best way to approach young people to test for chlamydia and enter the trial.

Staff also reported having no time to recruit patients, and communication difficulties in approaching patients to test for chlamydia when they were attending for a completely different health matter.

Regional training nurse focus group

The research co-ordinator also ran a focus group with the General Practice Research Framework’s regional training nurses (RTNs) to discuss their views on the trial and recruitment. The key findings were that the RTNs also believed the myths (that people would not want to test and young people do not attend the practice). They were confused over the NCSP and how this could run in parallel with the trial. They also suggested asking young people what would encourage them to test.

Implications

Results of this focus group and the practice interviews suggested that the myths in practice needed to be addressed, that the NCSP involvement be clarified, that a systematic way to identify patients be established in each practice in the most effective and time-efficient way and that staff were clear on the best way to approach young people to screen and invite them into the trial.

Recruitment report

This study had been funded on the assumption that a large number of chlamydia tests would be taken within general practice, in accordance with national data showing a large and growing proportion of all chlamydia tests taking place in this setting. However, the low number of tests we were experiencing was surprising. We therefore collaborated with the Health Protection Agency (HPA) to explore the distribution of chlamydia testing within practices in more detail, since these data were not publicly available.

The Chlamydia Vital Signs Indicator 2010/11 measures the proportion of the 15- to 24-year-old total population tested for chlamydia outside GUM clinics. The target set by the Department of Health was 35% for testing undertaken during the period 1 April 2010 to 31 March 2011. Data for monitoring the vital signs indicator are based on NCSP data returns and testing outside GUM not reported to the NCSP. In 2010–11, 25.2% of the population aged 15–24 years were tested for chlamydia (1,733,220 tests) based on data reported to the NCSP. 46

Data received from the NCSP reported 6319 participating NCSP practices across England, of which around one-third (n = 2895) registered no positive chlamydia test results between April and December 2010. The range of tests performed by these practices over this period was 1–176, with a mean of nine tests per practice. Only eight practices performed between 332 and 2336 tests, averaging 797, and obtained over 30 positive results.

This indicates that there were very few NCSP practices testing for chlamydia at the rate the trial required to conduct the planned study (Table 7). Positivity varied little by practice, and stood at 6% at the time of the study.

What are the general practice consultation rates of young people?

Table 8 reports data from a QRESEARCH® (University of Nottingham, Nottingham, UK) report showing the patient consultation rates with GPs and nurses of 15- to 19-year-olds and 20- to 24-year-olds over the period 1998–2009 in England. 47

Although measures of dispersion are not given in this report, the data indicate that a substantial majority of people within the 15–24 years age group are visiting their registered practice for a GP or nurse appointment at least once per year.

Phase 2 recruitment rates

During the second phase of the pilot (March 2011 to June 2011) testing increased by 100% (from 31 to 62 tests). However, this accounted for only 11% of the 16- to 24-year-olds attending the practices over that period (Table 9).

Practice champion

A member of staff in the practice was identified as the practice champion to motivate and be a point of contact for other practice staff. Practices were encouraged to have champions for each staff group (e.g. nurse champion, GP champion).

Additional trial materials

We provided additional trial materials for the practice team. Information leaflets were designed specifically for the GPs and primary care team. They included information about the trial and benefits to the practice and challenged some of the myths we found within general practice (i.e. that they do not see many people of the target age group, that young people do not want to be tested for chlamydia in their general practice surgery and that their practice population do not get chlamydia). We previously used the NCSP posters in practices to encourage young people to test; however, we introduced trial posters designed to be eye-catching for the practice waiting and consulting rooms. The poster invited 16- to 24-year-olds to take part in the study and to ask a GP or nurse about the Spread the Word health study.

Development and implementation of practice ‘action plans’

Practice action plans were developed and tailored to each practice. Specific objectives were identified in consultation with each practice to help improve recruitment. The action plans covered objectives, actions required, implementation dates and means of verification. Practice action plans were finalised after additional training (if required) and practice visit. Objectives within each action plan included:

• establish a visible practice champion(s)

• encourage team building: ensure the trial is and remains a high priority in the practice; ensure feedback to the whole practice on progress; ensure that enough trial materials are kept in consulting rooms; answer the queries of other members of staff about the trial; motivate and encourage staff (which could include using practice targets)

• display trial materials prominently and make them easy for young people to see and pick up

• highlight all patients aged 16–24 years coming in for appointments and alert the relevant staff to offer a test during the appointment (this could include computer alerts on the practice system or a note at time of booking)

• reinforce that all chlamydia testing was to be carried out by the practice and not through the NCSP process, to avoid confusion (confirming the laboratory process to be used)

• have practice champion remind all clinic staff weekly to give out trial information and ask all 16- to 24-year-olds to take a test and refer to research nurse for consenting (and considering the most effective way to do this, e.g. e-mails, face to face, practice meetings)

• use a banner at self check-in screen to invite all 16- to 24-year-olds to take part in the trial

• upload information about the trial on the practice website

• communicate earnings per patient to the practice staff (practice champion to cascade to colleagues)

• give monthly feedback on performance indicators and discuss any barriers the practice staff may have (research nurse to ensure that information is fed back)

• educate the wider team and engage all practice staff about the trial (practices were offered a visit by a member of the trial team to present to the practice staff, slides were also provided if the practice champion wanted to inform staff at a practice meeting, the general practice surgery/primary care information leaflet was distributed to practice staff)

• have reception give out test kits (or kits including trial material placed on counter to return to reception or nurse/GP)

• have receptionists hand out patient information leaflets and trial material

• have practice staff use scripts to help explain the trial (to be used at their own discretion)

• include a chlamydia test on registration to new patients aged 16–24 years.

Development of educational materials for practice staff

Scripts were prepared for each staff type to help explain the trial. Additional information, sourced from NCSP, was also provided on how to talk to young people, young people’s attitudes towards sex and relationships, and normalising testing.

Feedback of testing rates and footfall

Feedback was provided to practices on testing rates and footfall (i.e. the number of people in the target age group attending that practice). We collated individual pilot practice footfall and testing data, which demonstrated to each practice that a large number of young people were attending the practices taking part in the pilot, but only a small number were being tested (data on footfall and positivity rates were reported in the primary care team information sheet).

Practice meetings

The trial team offered to talk to the practice as a whole about the trial to dispel the myths around chlamydia testing and discuss specific practice barriers. Two of the practices requested a meeting.

Feedback on payment structure

Payment details of all screening and trial recruitment activity to date were sent to the practice manager, nurses and lead GPs in the five currently active pilot practices, excluding one other practice that dropped out of the study as a result of staffing issues. Practices’ letters included what they could have earned if they had reached their target. This was based on their estimated attendance (i.e. if screening 20% of the attendees had been achieved, this is what you could have earned). They were given information on how many participants the practice had recruited in the previous month and how many could have been recruited, based on their attendance figures. The difference in actual earnings and potential earnings through service support costs was considerable in some cases (e.g. actual £445.95 compared with potential £16,243).

Additional training of staff

Additional nurses were trained in one practice to help boost recruitment by increasing the number of nurses who could obtain consent.

Newsletter

A newsletter was sent to all participating practices in August 2011 showing that testing had increased but the proportion testing positive was low. The newsletter stated this was to be expected in general practice, as we know practices would (on average) have to test 20 people to get one positive test (there had been concern by some practices that they were not seeing many patients diagnosed with a STI). The newsletter encouraged practices to continue to screen all 16- to 24-year-olds opportunistically. It also highlighted that data show that the positivity rate is highest in slightly older men, 20- to 24-year-old males (8.6%),27 and that national figures show an average of two to three general practice visits per year from this age group. 49 We also presented the 20% footfall targets.

Engagement of practices with chlamydia screening office

The engagement of practices taking part in our trial with their local CSO varied greatly. One practice did not know that it was registered as participating in the chlamydia screening programme, while others did not know who their local co-ordinator was, and sometimes there had been no engagement prior to or during the trial. By contrast, in another case the local CSO was already closely engaged with the practice and attended the practice meeting when the proposed trial was presented to the wider primary care team. In this practice the programme manager and the screening promotions manager were very supportive, offering to manage testing and reporting processes differently to ensure that the results were directed back to the practice where the patient would be managed. In addition, they shared the results of a marketing strategy they had adopted locally to encourage testing.

Variation in process

Laboratory contracts

Contract arrangements between laboratories and local CSOs were found to be unclear and varied. In some cases, while the trial team was setting up processes, contracts were under review and being renegotiated. In general, however, laboratory managers were more than willing to facilitate trial requirements.

Where practices opted not to redirect results from NCSP laboratory forms, the cost of the chlamydia test was not always covered by the block contract with NCSP and was charged to the practice. One laboratory insisted on a confirmation letter from the local CSO, as contractually it was obliged to provide results directly to the local CSO; another required the trial team to provide it with the actual number of tests processed during the trial so that it could charge the local CSO directly. In another practice, the trial team was requested to meet the testing costs, as the laboratories would not allocate costs to the local CSO or the individual practice.

Prescribing and dispensing

Whilst the NCSP directive is to provide free treatment to all 15- to 24-year-olds, in practice this was not always the case. In three out of 17 practices medication was administered free of charge directly to the patient. These were practices with a locally enhanced service contract where a nurse prescriber administered medication during appointments where patients returned for treatment.

In other practices this was not the case. The trial team advised that medication should be prescribed under a patient-specific directive and administered free of charge. We provided information on how to manage a patient-specific directive, although this proved challenging, as practices were not familiar with using them. Practices were reluctant to store medication on site and did not have suitably trained staff available to dispense medication to individuals.

Clinical governance at the interface between practices and local chlamydia screening offices

The stand-alone character of CSOs on occasion made it challenging to organise an effective way of working with practices. Earlier in this chapter, we described the concerns expressed by local CSO staff about the study. In one locality, these proved unresolvable, and a general practice surgery choosing to take part in the study was not permitted to use NCSP forms.

In relation to this locality, we had a high-level meeting with the director of the NCSP. It was confirmed on all sides that it was not in the power of the NCSP team to direct any local CSO to approve the study. Nevertheless, with appropriate NHS Research Ethics and Research and Development approvals in place there was nothing to stop the practice from taking part in the study, but the tests and positives would not count within the NCSP targets. This put us in the uncomfortable position of appearing to undermine the local NCSP targets by working in parallel. In a pilot stage this was not a major problem, but had we been in a position to scale up this could have presented substantial conflicts between NIHR-sponsored research and the NHS delivery settings it aims to support.

Patients under 16 years

There was general concern across the local chlamydia screening areas regarding the management of patients under 16 years if we were managing the testing and care on their behalf. This was in fact easily remedied, as patients below this age group were not eligible for the trial and would be directed through the local CSO in the normal way, as they would not have been consented.

User testing

The web tool system was tested by the trial team and reviewed by the Interact Lab at the University of Sussex. Feedback was also given by health-care professionals (including research nurses, GPs and HAs). Users reported the system as easy to use; some suggestions were made to improve the database.

System access

Only the system administrator (trial manager) was able to add new users to the system and had the responsibility for assigning logins to authorise users. The system administrator did not have access to any patient-identifying information. The trial statistician was also able to download anonymised data. Each user was assigned a level of access appropriate for their role. General practice surgery staff could access only their own practice information. Practice staff could view only the data that were entered by the general practice surgery. They were unable to view the HA consultations of their patients.

Data

Anonymous data were collected on all chlamydia test results carried out in the practice during the time of the trial. Once a patient tested positive, additional information was collected by the GP/nurse in the practice. Data were entered directly onto the web tool. The web tool allowed for real-time monitoring of testing, recruitment, results and PN. For each potential participant added onto the web tool, a unique identifier was generated automatically. Any edits made to the data required the user to provide a reason for the change, allowing for a full audit trail of the data. The system logged all user interactions with the system and input of and modification of records (date- and time-stamped). The web tool complied with the requirements of the E6 Guideline for Good Clinical Practice (1996)62 developed by the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use and the Data Protection Act 1998. 63

Consistency checks were employed at data entry stage to avoid inconsistencies and improve the integrity of the data. Regular checks of the database by the trial team ensured that any missing data were entered in a timely manner.

Overview

We used a qualitative and participatory approach, as we considered the analysis of simulated clinical practice with reflective discussion the best available approach. 66 During a 1-day workshop, experienced PN practitioners were observed while contributing to focus groups, actor-assisted role plays and a plenary discussion. All discussions and field notes were recorded and used to undertake a thematic analysis.

Development of workshop

We developed the workshop as a multidisciplinary research team. Participants completed a short online survey to establish their views of current practice. This helped us to refine focus group topic guides and establish areas requiring clarification. We designed three role plays of commonly encountered, uncomplicated PN scenarios for Chlamydia trachomatis based on existing guidance, discussion with practitioners and informal review of recent literature.

Selection and recruitment of participants

Ten participants, a major part of whose job was PN, were purposively selected via the network of the study’s lead HA.

Data collection: focus groups, role play and plenary discussion

The first focus group session established the participants’ perceptions of PN, which were further explored in actor-assisted role plays. These were followed by researcher-led discussion using topic guides and subject matter created dynamically in-session. A second focus group session specifically explored the practitioners’ and actors’ (as patients) perceptions of PN exemplified in the role play and implications for practice. In the final plenary session, an emerging view on how to deal with any ambiguity was established.

The focus groups, role plays and plenary session were digitally recorded and transcribed. Contributing materials included the online questionnaire, observer field notes and flipchart sheet notes.

Background of partner notification practitioners

Ten out of twelve invited practitioners attended the workshop: HAs (n = 8) and senior nurses (n = 2), working in GUM clinics (n = 6), young people’s services (n = 2) or community outreach clinics (n = 2). The locations ranged from large urban areas to smaller towns across England, in London, the South East, the South West, and Yorkshire and the Humber. Practitioners’ experience ranged from 2 to > 20 years, with equal male–female representation.

Theme 1: partner notification in practice

Theme 2: scenario building

Practitioners invested considerable time and effort in helping patients deal with difficult questions that disclosure might raise within the partnership, such as infidelity. Practitioners used motivational interviewing and challenged patients’ blocking strategies, helping patients to build scenarios and enabling them to project into the future and identify any anticipated regret. Practitioners also suggested creative solutions to introduce the subject to partners or elicit telephone numbers for previous partners.

When you do those patient referrals, sometimes [the patients] want to do them but they do not necessarily know the best ways. So I always say, ‘Have you thought about how you might bring the subject up?’ or ‘You might want to say . . .’

HA7, male, south coast city, GUM and community clinics

It’s our job to help you so it’s not about forcing you to do anything; it’s just working with you really. Obviously you can choose to tell them yourself but obviously you have been a bit worried about that. What we can do is we can let them know on your behalf if it helps?

HA4, female, London, GUM clinic

Is it possible to say, ‘Look, if we are going to be starting to think about trying for a baby maybe I’d better go for a sexual health check up, and check that I have not got anything before we start because I’ve had sex with people before you?’

HA3, female, London, GUM clinic

The first thing people often say is ‘Hang on a moment, why did you not tell me?’ And that creates a whole other dynamic for you.

HA5, male, London, GUM clinic

Approach it very calmly, say that in your history, ’cos it’s a new relationship, as someone that you like to consider your sexual well-being so actually while you were away you decided to have a check up. And then you can decide what you want to tell him the outcome of that was . . . would you rather he found out from a third party or from you?

HA5, male, London, GUM clinic

And it’s a fairly new relationship, so I do not think that it’s without . . . I do not think it’s outside the expected realm that this is a possibility in a new relationship that an infection could be there. You would not actually have to disclose that . . . he knew you had a previous partner just before him, you would not really have a reason to have to tell him there was any overlap at all. He would not gain anything from knowing that.

HA2 female, London, GUM clinic

You can just maybe talk to some work guys, just to get his phone number. Just say he owes you £5, track him down like that.

HA4, female, London, GUM clinic

Theme 3: movement between strategies

Although practitioners were able to define the three distinct PN methods, it emerged through role play that there was often a fluid movement between methods as a response to individual patient need. This movement between strategies was enacted through either a follow-up call to check on progress (usually 2 weeks) or a shorter check-back call (2 days) for patients who needed time to consider their options. A time frame for the call was agreed with the patient at the initial consultation.

I can give you a little while to think about it, a couple of days, and then give you a ring back and see how you are feeling about it.

HA1, female, northern city, community clinic

Theme 4: contract or delayed provider referral?

Participants did not regard contract referral as normal practice for common bacterial STIs. Discussions across the day demonstrated some ambiguity and inconsistency around what constituted a contract referral. Some practitioners described contract referral as an agreement made with the patient that they would follow them up within a set time frame to check their progress, and that this agreement constituted a contract.

So they are expecting a call, so in effect that’s a kind of contract that we will [follow up]. I’m not saying you have got to do it by this time, but I’m definitely going to follow you up.

HA8, female, large southern town, GUM clinic

We tell all our patients that we will call them in a couple of weeks’ time just to see how they are and we wrap it round saying, ‘Did you have any problems after the tablets? Were you okay?’ But we always tell them [in advance] and say, ‘And then we can see how you’re getting on with telling your partners.’ So, in a way, the contract referrals are implicit in the normal way that we work because of the checking we do at 2 weeks. If they have not been able to do it then, then we’ll [offer provider referral].

HA3, female, London, GUM clinic

However, this is not contract referral as previously defined (see Table 15). The practitioner has not agreed with the patient that their partner will be notified directly by the practitioner if they have not attended by a certain time. Instead they have agreed a follow-up call to check on progress. The key difference is that the patient still must agree to provider referral, rather than this having already been agreed or contracted. An indication of whether contract referral has taken place is whether the practitioner collected partner details during the initial conversation for subsequent use if needed. Three of ten practitioners said they would initially request partners’ names so that they could refer to them clearly in conversation with the patient, but none would ask for further partner contact details until an offer of provider referral had been accepted.

Findings in relation to published literature

We found conflicts between practice and guidance. Contract referral as presented in the SSHA manual states:

A hybrid approach may be employed where an initial patient referral is followed up by a provider referral after an agreed period of time, if the contact has not attended.

(p. 20)64

This implies that a contract is made with the patient during the initial consultation that the practitioner will contact a partner directly if they have not attended by an agreed period of time – no questions asked.

However, guidance for patient referral states:

It is important to negotiate a back-up plan during the first interview, if possible (for example, ‘If he’s not been within x days/weeks should I contact him directly, or speak to you again? . . . Is it ok to ring you? . . .’)

(p. 32)64

If the patient agrees to the practitioner contacting the partner directly as a back-up plan, this would constitute a contract referral as defined earlier in the manual. If the patient agrees to be called back and subsequently takes up the offer of provider referral, this would fall under ‘provider referral’ in the guidance, despite differing from a provider referral agreed at the initial consultation. Confusingly, some participants described this incorrectly as ‘contract referral’, even though there was no initial agreement for the practitioner to contact the partner directly.

If a separate category of delayed provider referral was introduced into the guidance this could clarify between a provider referral offered initially and one offered as part of a back-up plan, as is common practice. This would help practitioners to accurately categorise this type of PN offer as a ‘delayed provider referral’ and not as a ‘contract referral’.

Guidance also recommends that for patient referral:

A follow-up interview may be necessary if there is no record of the contact having attended. The purpose of this is to check progress, gather any additional data and repeat the offer of provider referral if the index patient is having difficulty. There is evidence that many patients who initially opt to inform their own partners subsequently agree to provider referral at follow-up interviews.

(p. 32)64

The guidance to ‘repeat the offer of provider referral’ at the follow-up call suggests that an offer of provider referral should have already been made at the initial planned patient referral consultation.

These ambiguities seem to occur as a result of attempting to use distinct definitions for PN methods in the guidance to describe the fluid PN strategy adopted by many practitioners in practice. Some practitioners said they found the term ‘contract’ unhelpful, as its meaning overlaps with negotiating a provider referral, which was closer to how they saw their work.

Other findings

Implications for the health adviser call to a patient

• A short list of four questions that can be worked through in 2 minutes should be considered.

• The first scenario presented by the research team was felt to be too wordy and complex for a patient to process. A short introduction is necessary in order to reduce the risk of losing the patient early on. The telephone conversation should also adopt the ‘chunking and checking’ methods used in communication skills so as to not encourage the patient to switch off.

• Evening telephone calls to patients must be planned for in cases where the patient is unable to take a call during the day.

• A decision should be made whether the patient is able to call the HA back at a convenient time on a set number. HA were concerned that a Withheld Number call to a patient does not set the right tone and does not eliminate the possibility of it being interpreted as a hoax call. There is a preference for revealing the caller number.

• Faced with reluctance to allocate time to talk to the HA, it was unclear how many times the HA should attempt to contact index patient?

• All patients should receive a follow-up call from a HA after the PN discussion (Figure 9).

FIGURE 9.

Telephone call timeline for patient referral (calls 1 and 3) and provider referral (calls 1, 2 and 3).

Strengths and weaknesses

The 1-day workshop allowed us to collect novel observational data on the working practices of PN professionals in a UK setting. These findings describe the current practice of experienced PN practitioners for the first time, and have implications for both clinical and research communities.

A limitation of the study is that we were unable to observe real-life clinical practice. The logistical and ethical barriers to obtaining consent from patients for the taping of initial conversations about sexual partnerships following a STI diagnosis are very challenging. We considered an analysis of simulated clinical practice with reflective discussion the best available approach. A further limitation is that the group was small, and may not have sampled the full range of organisational cultures in clinics offering PN.

Implications for the main study

The questions the workshop aimed to answer were:

1. Are the three PN methods as described in the SSHA manual clear to PN practitioners?64

   1. In theory, practitioners understood the definitions of the three distinct PN methods. However, there was considerable ambiguity about the precise meaning of contract referral in the context of common bacterial STIs. Practitioners used only two methods: patient and provider referral. Practitioners did not use either method exclusively but moved between strategies responding to patient need, often delaying an immediate offer of provider referral and negotiating a delayed provider referral with patients to optimise PN.

2. In practice, are the three methods distinct and feasible to deliver?

   1. The two methods used of patient and provider referral were found to be feasible and operational. In the latter case there is an issue with the time delay until the check-back call, which is agreed with the patient on an individual basis. This delay ranged between 2 days and 14 days. A final timeline for HA calls was agreed, as shown in Figure 9.

3. Under what circumstances do practitioners offer specific PN strategies?

   1. While approaches to individual clients varied, the overall approach to negotiation of PN in practice was remarkably consistent across the group. Adopted PN methods varied depending on the STI diagnosed and patient-related factors. All practitioners tailored their offer of PN to individual patient need and were often guided by their instincts. Patient referral was the preferred method, with practitioners using motivational interviewing techniques to help patients make their own decisions to preserve their current relationships and protect their own health.

   2. Provider referral was more often used for clients such as gay men and young people, for whom it was seen as more effective. Some practitioners, however, found provider referrals difficult to manage because they believed that it took away patient choice and autonomy.

   3. Contract referral (as defined in guidance) was regularly used for chronic and serious STIs including HIV, hepatitis B and C, and sometimes syphilis, but not often for common STIs.

4. If there are conflicts between guidance and practice, how might these be resolved?

   1. Existing guidance may need to be modified to reflect our findings and we propose consultation on the following advice and recategorisation of PN methods:

      1. Patient referral: no change.

      2. Immediate provider referral: patient agrees from the outset that the HA may contact a partner immediately.

      3. Delayed provider referral: patient needs some time to consider the offer of provider referral and agrees to a call back (2 days) to discuss this or, if during a routine follow-up call (2 weeks) the patient has not managed PN themselves, provider referral is offered.

      4. Contract referral is reserved for blood-borne viruses and syphilis and removed from the guidance for bacterial STIs.

This will have implications for training and assessment of competencies for all the professional groups with an interest in PN.

Unanswered questions and further research

There remains a marked lack of qualitative or operational research on PN with HAs or other specialists in PN, and we are therefore unable to provide a detailed comparison with similar studies. Interestingly, there remains no published three-way comparison between provider, contract and patient referral, although provider and contract referral have separately been compared with patient referral. 9–11 This absence suggests that there may be operational overlap between contract and provider referral not reported previously, which has implications for both clinical practice and research evaluation of different approaches to PN.

Economic objective

What are the costs of the patient and provider referral models of PN as used in the pilot and how do these costs compare with other proposed/existing pathways that might be used for the purpose of PN?

Methods

We compared the costs associated with the pilot PN pathways with two separate published studies that investigated the costs (and outcomes) associated with PN strategies compared with current practice in the UK.

The two studies with which we compared the pilot pathways are: (1) the NIHR-funded ClaSS project, which included a nested PN trial;23 and (2) the MRC-funded APT studies. 33,68 In both studies, resource use data were collected alongside a primary clinical study and unit costs were applied. The average cost per case of PN as reported in the published ClaSS and APT studies is naturally affected by the success in terms of relative effectiveness of any particular strategy. For example, if a particular PN strategy successfully reaches and treats more sexual partners than an alternative strategy at the same cost, the cost per partner treated will be lower than the alternative, as more partner(s) have been treated. In the current study, it is not appropriate to compare across these different strategies in terms of effectiveness because the approaches apply to different individuals in different geographical settings and such a direct comparison will cause bias.

Furthermore, the ClaSS project represented a randomised controlled trial while the APT study was an exploratory trial and was not randomised. Instead we attempted to consider the typical pathway for one index case and their respective partners. We therefore assumed the average resource use (e.g. average time of consultation with the health-care professional relevant to that pathway) and applied unit costs to the average resource use likely to be incurred by one individual and their partners on that pathway. In order to provide a consistent approach to the following comparative analysis of costs it was necessary to make some assumptions prior to conducting the analysis.

Assumptions

1. All pathways begin from the diagnosis of and discussion with the index case and include the costs and resources associated with the index case to ensure consistency across pathways. The inclusion of the index patient is required because some pathways include PN advice at this initial stage and fair disaggregation of associated resource use would not be possible.

2. Based on results from the ClaSS study, we have assumed that each index case generates 1.5 partners.

3. All index patients and their partners comply with all aspects of the pathway.

4. All index cases and partners receive the same treatment and tests where required, and so the costs of tests and treatment are not included in any of the pathways. We included only resource use associated with trying to deliver PN (that is, the associated staff costs).

5. The ClaSS project and APT study pathways assessed the success of PN in their respective studies by recording how many partners were consulted or treated as part of the outcome. The cost of follow-up telephone calls to the index to assess the outcome for the partner, if required, was assumed to be a cost of the research and not part of the intervention and, therefore, was not directly recorded. To achieve consistency with the current study, resource use and associated costs have been included for these studies to represent the required follow-up to index cases or partners to assess whether or not PN has been achieved. These costs are assumed to be the same as those incurred by the current study.

We present a brief summary of the ClaSS study, the APT study and the patient and provider pathways used in the pilot of the current study. For both pathways in each of the three studies we explain how we have determined resource use and applied unit costs.

Partner notification in the Chlamydia Screening Studies

The primary objective of the PN trial, which was an integral component of the ClaSS study, was to explore effectiveness of PN advice provided in a primary care setting compared with referral to genitourinary clinics. The latter was and still is the current practice.

The costs incurred by each PN strategy are presented from the perspective of the NHS. Costs were originally obtained in pounds sterling at 2003 prices and subsequently updated to 2005 prices for the ClaSS report. We have used reported resource use from the ClaSS study and applied wages and costs that apply for 201169 for the purpose of the current report. Practice nurses recorded the total duration of the consultation, which included the time taken to give results and treatment, explain the study, obtain consent, and conduct randomisation followed by either PN or referral. Published data on the duration of GUM clinic consultations for PN were used.

In the nested PN trial for the ClaSS project, nurse-led PN at the practice for the index patient while the patient is receiving their own result was estimated to add just a few minutes extra on to the consultation time. For individuals in the comparator arm, where the index was advised to seek PN advice at the GUM clinic, the appointment with the nurse at the general practice surgery was for the purpose of receiving treatment and the result only. The index patient then required an additional appointment at the GUM clinic for PN advice. Partner(s) were also required to attend the GUM clinic for treatment and advice.

A slight adjustment was made to the ClaSS project results which are presented here. The initial appointment in the ClaSS project was estimated to take almost 42 minutes for those individuals randomised to the nurse-led PN strategy and 38.8 minutes for those randomised to the GUM PN strategy, but both these timings included time for randomisation and consent to the study. In the current study, the explanation of the study and taking of consent were estimated to take between 10 and 15 minutes. We have assumed the mid-point of this range (12.5 minutes) and deducted this from all the timings.

Partner notification in the Accelerated Partner Therapy study

The objective of the APT study was to compare costs and outcomes associated with three alternative methods of expedited PN in the primary care setting. The cost analysis carried out alongside the exploratory trial was conducted from the perspective of the NHS and considered only direct health service costs.

Following the pre-determined criteria, eligible index patients who had given consent to participate in the study were offered one of three methods of PN: (1) APTHotline, telephone assessment of their sex partner by a clinic-based nurse-qualified HA; (2) APTPharmacy, assessment of their sex partner by a trained community pharmacist; or (3) routine clinic PN, patient referral which included infection-specific information, advice that the sex partner should attend the clinic for testing and treatment and, in one clinic, a standard letter detailing antibiotic treatment options for the sex partner to give to his/her GP if appropriate. Each index patient was asked to choose which method they preferred for each contactable sexual partner. The index was instructed to provide the partner with all relevant information. Once the partner engaged with the allocated PN method, the appropriate health-care professional explained the study and sought consent from the partner to participate. Any partner who did not like the method of PN chosen for them by the index patient could default to routine PN (PN at GUM clinic as per ClaSS study).

In the APTHotline group, either partners collected the treatment pack from the clinic reception or the index patient could take the pack to them, which occurred if the sex partner completed his/her telephone assessment before the index patient had left the clinic. Partners in the APTPharmacy group received their treatment packs from the trained community pharmacist at the time of their consultation.

Engagement in the APT process was apparent when the partner adhered to the method of PN chosen for them by the index case. In some cases it was necessary for the index case to be followed up for confirmation about where, if at all, the partner received APT or routine care. All resource use incurred as a result of the APT strategies was collected prospectively.

Results

Pathways 1–6 present the pathways for each of the six strategies under comparison. Tables 18–20 present the resource use in terms of items such as length of appointment, number of telephone calls, etc. The unit costs applied to these resource items such as relevant staff salary or cost of call and equipment is appropriately applied to each item.

The cheapest strategy is the ClaSS project strategy of providing PN advice to the index at the practice when the index receives the result of their own test (strategy 1), with the partner(s) being assumed to attend the local GUM clinic to get their own PN advice and treatment. This strategy has an estimated cost of approximately £42.00 per index, assuming the index case has 1.5 partners.

The ClaSS project strategy of PN at the GUM clinic (strategy 2) is the next cheapest strategy. This strategy assumes that the index case is identified at the practice but that the index then attends the GUM clinic to receive PN advice and the partner also subsequently presents at the GUM clinic for assessment and treatment. Thus there are additional health service interactions required, adding to the costs for this strategy which are estimated as £48.31 per index (assuming 1.5 partners).

Strategy 5 (patient referral) and strategy 3 (the APT study strategy of PN via a hotline) are roughly equal in cost. In the patient referral strategy (strategy 5), the patient assumes the responsibility of contacting the partner and the average cost per index of this strategy is £52.53 (assuming 1.5 partners). In the APT strategy (strategy 3), the index is assumed to be identified as positive at the GUM clinic and the partner is requested (by the index) to telephone the study hotline for assessment and advice about how to get treatment (each index is assumed to have 1.5 partners). The average cost per index case for strategy 3 is approximately £52.80. In both these strategies the effort of contacting the partner is incurred by the index.

Strategy 4 (PN via pharmacy) assumes the index receives PN advice at the GUM clinic during diagnosis but requires the partner to attend the pharmacy for assessment and treatment. The additional cost of the pharmacist consultation contributes to this strategy being slightly more expensive than the preceding ones. This strategy costs approximately £56.38 per index case (assuming 1.5 partners).

Strategy 6 (provider referral) is the most costly strategy. Like strategy 2 (PN at GUM clinic), this strategy requires additional interactions with the health service in that after initial diagnosis of the index (irrespective of the setting) the index does not receive PN advice at that point, but a subsequent and entirely new interaction is required as the HA receives an alert to contact the index, provide them with an assessment and discuss PN advice. In the provider referral strategy all the costs of contacting the partner(s) are borne by the health service and not the patient.

Thus strategy 6 (provider referral) is the most costly strategy, with an estimated cost of approximately £85.98 per index (assuming 1.5 partners). For this reported strategy it is assumed that all partners are contacted by the provider at the wish of the index, although in the strategy index cases could potentially choose a mixture of patient and provider referral approaches for their partners. If a mixture was the preferred approach then it can be assumed the average cost for the strategy would be somewhere between the results of strategy 6, where it is assumed in this case that provider referral is applied to all partners, and the results of strategy 5, where patient referral is applied to all partners (Figures 10–12).

FIGURE 10.

Chlamydia Screening Studies pathways. (a) PN at general practice surgery; and (b) PN at GUM clinic.

FIGURE 11.

Accelerated Partner Therapy pathways. (a) PN via hotline; and (b) PN via pharmacy.

FIGURE 12.

Pilot pathways. (a) Patient referral; and (b) provider referral.

Discussion

The estimated costs of the alternative strategies of PN range from approximately £40 to £90 per index case, based on the assumptions that all index cases have 1.5 partners requiring PN and there is complete compliance of all index patients and their partners in adherence to whatever PN approach is indicated by the particular strategy, and excluding costs associated with the actual test and treatment for everyone involved.

The least costly strategy appears to be that of PN at the practice (strategy 1) as offered during the ClaSS project. This costs approximately £42.00 per index case. This strategy costs less than any other because the index receives PN advice at the same time as being diagnosed (i.e. receiving their own result) and the costs of contacting the potentially infected sexual partner(s) are borne entirely by the index, which is asked to inform their sexual partner that they should attend the GUM clinic for their own assessment and advice. There is no cost incurred by the health-care provider in this strategy related to contacting the partner, although the partner is ultimately assumed to attend the GUM clinic, and this cost is included.

Strategy 6 (provider referral) is the most costly strategy because the health service bears the costs of contacting the partners (or potentially some of them). If the index did not choose provider referral, even though offered, this strategy would cost no more than the patient referral strategy (strategy 5).

Strengths and weaknesses

The strength of this analysis is that it is the first to directly compare a full range of potentially feasible alternative strategies for PN advice to index patients and their partners, in terms of costs. The costs associated with some pathways were estimated as part of previous studies, and for the purpose of comparison some assumptions were required. However, in all cases, the assumptions were made in advance of any calculations and have not been adjusted in the light of the results.

A weakness in the analysis is that the comparison does not represent the costs associated with a ‘head to head’ comparison in a primary study. Furthermore, the costs presented are estimates and the detail of their differences should be interpreted with caution. The limited data available and the limitations imposed by the halting of the wider trial mean that the costs are estimates based on relatively small numbers and present an indication only of potential differences. Appropriate information with which to estimate confidence intervals around resource use, such as length of time for appointments, was available for some resources used but not others. Therefore, we have not presented confidence intervals for the estimated costs. Detailed scrutiny through in-depth sensitivity analysis would be inappropriate given the available data.

The major weakness is that very few positive index cases were identified to enable PN to be tested. We have compared the proposed pilot strategies with strategies from other studies, and it would not be appropriate to attempt to assess cost-effectiveness relative to other strategies, since the comparison was not head to head or randomised. Therefore, in order to facilitate the reporting of comparable costs for the pathways in this study, all strategies are assumed to be equally successful in achieving their objective. However, this is not likely to be the case and, as a result, costing studies alone cannot be helpful to decision-makers. Clearly the least costly strategies to administer will not necessarily save money if they are ineffective in achieving their desired objective. In contrast, the more costly strategies might be sufficiently more effective to justify the additional cost through a demonstrable increase in the success or effectiveness of the strategy. This has not been proved in this study. Indeed, the difficulty in trial recruitment and the failure to recruit enough infected individuals into the study was the reason that it is possible to present only an estimated comparison of costs. Thus, the major weakness of this study is that it has not been possible to assess the strategies in terms of their relative cost-effectiveness.

Conclusions

The results from this costing study suggest that alternative PN strategies are unlikely to differ much in terms of costs. However, the more responsibility is assumed by the health service to contact partners to facilitate PN strategies (and further follow index cases), the more costly the strategies will be.

Unanswered questions for future research

Whether or not the additional costs to the health service implied by strategies such as provider referral, which assume the most responsibility for contacting partners, are justified in terms of effectiveness and lead to a greater societal benefit by reducing the spread of the disease would need to be fully assessed in a cost-effectiveness study with appropriate population modelling.

Partner notification is an essential component of addressing and treating STIs. The conclusion from this study is that increasing the degree of responsibility assumed by the health service to reach partners will increase the cost. A justification for spending more of the health service’s limited resources to reach partners to achieve the outcome sought is required, which this study was unable to show.

Background

The planned economic data collection and analysis for this study was to be carried out during the main trial itself, and was not scheduled to have any role in the pilot trial. The objective of the pilot study was to explore the feasibility of recruiting individuals for chlamydia screening and PN in the general practice setting. When it became clear during the pilot that recruitment to the full trial may be problematic, the research team decided to explore a dedicated intensive recruitment approach.

Acknowledging that the full trial was unlikely to proceed without a better response to recruitment, and thus in an effort to salvage all possible information from the project’s pilot trial, the economic team considered it necessary to attempt to include a comparison to the intensive recruitment approach used in order to undertake an economic analysis. An economic evaluation is defined as a comparison of two or more interventions in terms of costs and outcomes. 67 Thus, the only comparison that could be considered which was pragmatic and would not add any additional burden to staff in practices (and to the research team who would recruit in the practices, and for whom recruitment was the main priority) was a simple and pragmatic ‘before and after’ (intensive recruitment) study.

In the pilot trial, a specific aim was to intensively recruit patients for chlamydia screening in general practices by introducing a suitably trained researcher into a practice for a 3-week period. The costs and outcomes associated with this approach could be measured and the success of intensive recruitment could potentially be evaluated for the purpose of a preliminary economic analysis, if it could be compared with the costs and resources used by, and the outcomes achieved, prior to the intensive recruitment phase.

Objective

What is the additional cost and success of intensively recruiting patients in general practices compared with the existing approach within any practice?

The data from ‘before-and-after’ comparisons are more appropriately presented as cost–consequence analyses, where costs and outcomes are presented in a disaggregated manner but in which no further analysis to achieve a cost-effectiveness result is undertaken. This is appropriate in this case because the data for the comparison are not derived from a controlled experiment.

Estimating costs and outcomes used in general practices in the pilot with regard to chlamydia screening and identifying positive cases

General practices in the UK manage a diverse range of primary care issues which might include antenatal care and advice, treating patients for minor illness, identifying infection and chronic illness and where necessary referring patients on to secondary care, to mention but a few. Screening and/or testing individuals for infection with chlamydia is one of many tests and services practices can provide. Although the practice readily provides this wide variety of services it must be realised that it is a finite resource: when a nurse or doctor provides an appointment for the purpose of screening and treating an individual who is positive for an infection such as chlamydia, it means that that particular appointment is not available to another patient who has a different need. Thus, providing a chlamydia test to someone who is not likely to be positive for chlamydia (a low-risk individual) means that not only is there a waste of the time and resources used in that particular appointment, there is also a forgone benefit for the individual who would have liked an appointment but did not get one and was given a place in the queue instead. In health economics, this is referred to as ‘opportunity cost’. Opportunity cost is the value of the consequences/outcome/benefits forgone by choosing to deploy a resource in one way rather than in its next best alternative use. 67 It is, therefore, appropriate in an economic analysis to estimate costs and outcomes associated with the intervention, even if the intervention is perceived to be a service that already exists and is already being paid for.

Methods

We provided practices with a short baseline questionnaire to explore their existing approach to chlamydia testing and PN. The questionnaire was developed in a relatively short time frame (when it was realised recruitment was poor) and was intended to be a pragmatic assessment of what the approach of a particular practice to chlamydia testing and PN was prior to the anticipated period of intensive recruitment. This was a one-off opportunity to assess the testing regime prior to the proposed intensive recruitment intervention, as the questionnaire was sent out just prior to planned intensive recruitment at the 11 practices. It was not possible given limited time and resources to monitor the exact throughput of testing and the result (positive or negative case of infection) and the questionnaire relied on the recall of the previous month of the member of staff completing the questionnaire. The data collection sheet used to assess the existing approach to testing is presented in Appendix 1.

The cost data applied to the resource use estimated by each practice are explained and presented in Appendix 2.

Results

Discussion of existing approach

The results suggest that general practices believed there was reasonable activity in testing individuals for chlamydia in the month prior to intensive recruitment. It had been planned to estimate costs associated with the reported testing activity. These costs are reported in Appendix 2 as a result of our concern with regard to their reliability. We believe to present results in terms of the average cost per test carried out or average cost per positive case identified would be misleading. The limitation here is likely to result from recall bias of the staff completing the questionnaire. It is also acknowledged that the pragmatic design of the questionnaire, in asking respondents to report a general band of activity, may have led to inaccurate reporting.

Concern over some of the assumptions necessary to estimate the costs associated with testing and the reliability of the data provided on testing by each practice suggests they should be viewed with some caution. This is discussed further in the Discussion section in reference to some other data recently made available.

Assumptions and framework for intensive recruitment

There are three different types of costs:

1. General costs including training, recruitment and set-up. These are distributed across all participating practices as a lump sum cost. In addition, each practice would receive service support costs for participating in the research.

2. Labour costs are already provided directly from the intensive recruitment team and not routine sources. The cost estimates for salary of the staff involved are specific to the grade of the researcher, the duration spent at the practice, and any equipment or disposables used.

3. There are no PN costs included because no data are provided on PN or its success.

All costs are presented in pounds sterling in 2011/12 prices.

Data for intensive recruitment approach based on costs of research staff

Data from 10 clinics were available to assess the impact of intensive recruitment. The staff time involved and the costs associated with the process are presented in Table 22. The details of staff time and costs are presented in Appendix 3. The total costs are apportioned across the clinics. The average length of the intensive recruitment at each clinic is between 15 and 18 days (weekdays).

The average numbers of days and hours spent at each clinic to carry out the process is presented in Table 23.

In Table 24, we present the costs for each clinic for the intensive recruitment period. In Appendix 3 we present the breakdown of these costs. The fixed costs of the process for set-up and training are apportioned equally and a fixed element applied to all practices. The time, resources and associated costs for each practice were recorded directly by the intensive recruitment team.

The variance seen between practices (from £1578 to £6408) is a result of the travel and accommodation costs for practices at a longer distance from the researchers’ base. The costs for practice 7 are considerably lower, as recruitment was managed by the local PCRN.

In Table 24, outcome data for the intensive recruitment period showing number of cases screened and number of cases identified as positive are presented. Compared with the ‘existing’ activity data as reported by the prior to intensive recruitment survey, there is an overall increase in the number of screening tests in 6 of the 11 practices. In 4 out of 11 practices intensive recruitment testing was lower than what was reported prior to the intensive recruitment period. One practice did not, in the event, undergo a period of intensive recruitment.

In April 2013, new data became available from the chlamydia testing activity data set (CTAD), which is a relatively new (set up in 2011) universal disaggregate data set for the reporting of chlamydia testing data from all NHS and NHS-commissioned laboratories in England. We accessed the data relevant to the practices in our survey for the month prior to intensive recruitment. These data are presented alongside the results from the ‘before’ intensive recruitment survey and alongside the activity that resulted from intensive recruitment. These are presented in Table 25.

We compared CTAD figures for the month prior to intensive recruitment with the testing rates reported by practices. Three practices provided a reasonable recalled estimate of their activity (practices 9, 10, and 11). Although practices 9 and 11 might appear to have overestimated their screening rates, this is an artefact of the questionnaire in which the upper limit of the range offered was assumed.

Three results based on the questionnaire cannot be verified by the CTAD data. Practices 2, 4, 8 and 10 all underestimated their screening rates, and two practices overestimated their screening rates (practices 1 and 3) in the pre-intensive recruitment survey questionnaire.

All the results suggest that intensive recruitment is costly, but importantly it does not seem to produce a respectable number of positive cases of chlamydia identified, given the cost and effort.

The results of this study suggest that intensive recruitment helped to increase the number of tests carried out in some practices but not others, given the result of the survey carried out before intensive recruitment and the results of the CTAD figures. It is clear that the additional tests being carried out are not effective at identifying individuals who have the disease. The results suggest that in some practices there was an increase in the number of people who were tested but this did not lead to an increase in the number of positive cases identified.

Discussion

Although a comparison of activity before and after a period of intensive recruitment to screening was attempted, the results of our study suggest that a full economic evaluation was not required and a cost and consequence analysis was the most appropriate approach. The results show that the number of tests assumed carried out prior to intensive recruitment and reported through recall when completing the questionnaire was lower in the majority of practices (6 out of 11) before intensive recruitment took place. Thus, intensive recruitment for these six practices did improve the screening rate based on the questionnaire results. However, a comparison with the CTAD figures makes any inference from the effectiveness of intensive recruitment less clear. In 3 out of 11 practices it was implied that the screening rate was higher in the period before intensive recruitment than during intensive recruitment.

The results suggest that the intensive recruitment period may have been successful in a few practices in increasing the number of screening tests carried out in the general practice setting. The results suggest that the recalled number of positive cases detected in any practice in the month prior to intensive recruitment is likely to be an optimistic estimate.

However, it is clear that while intensive recruitment may have increased the number of tests carried out (although this is not convincing given the CTAD data), it did not lead to an improvement in the number of positive cases of infection being identified.

Costs associated with screening prior to the period of intensive recruitment were estimated and are reported in Appendix 2. The results of the costing exercise should be interpreted with caution, as a result of some of the assumptions relating to the number of tests carried out, who carried them out and the time devoted to them being carried out. However, these results suggest much higher costs associated with intensive recruitment.

The strength of this analysis is that it was the first primary study to collect cost and outcome data associated with active intensive recruitment of STI screening in a trial that was failing to recruit. All cost and resource use data, as well as all the clinical data on numbers of individuals who were screened and detected as positive and treated, have been collected in a primary research study. A limitation is that no robust comparator data exist with which to conduct an economic evaluation, and a further limitation is that these data were not collected as part of a controlled study. Furthermore, to estimate costs some assumptions were made about the time and resource use of the staff carrying out the tests.

The inferences of this study are that intensive recruitment was actually not effective at detecting positive cases. Intensive recruitment is shown to be costly and there is no clear improvement in terms of a resounding effect on numbers of positive chlamydia cases identified, as the positive cases identified remained zero in the majority of practices. The number of individuals undergoing a screening test may increase as a result of intensive recruitment, although the results of this study do not provide strong support for this. But it is clear that intensive recruitment is costly and does not achieve the desired objective of finding positive cases and treating them.

General practice

As previously described in Chapter 4, practice staff held a number of incorrect beliefs about young people’s attendance in practice, the acceptability to them of testing for chlamydia in this setting, and the prevalence of chlamydial infection. However, addressing these beliefs had little effect on recruitment.

A more fundamental barrier appeared to be the inability of nurses and doctors to prioritise, within a busy and highly reactive workplace culture, the promotion of asymptomatic testing for chlamydia. This was felt to be embarrassing for staff and potentially stigmatising to young people. Recruitment is easiest where patients are gathered in a designated clinic, which was not possible in this case.

We were able to demonstrate that use of an external researcher to recruit in the waiting room, as in a related Australian study, could enable opportunistic recruitment on a scale which did not appear to be possible within general practice itself, despite remuneration. Although this was not an inexpensive option, it should be considered by future researchers seeking to recruit young people opportunistically in primary care, especially on sensitive subjects.

The epidemiological basis for the study

We, our funders, and colleagues at the NCSP offices of the HPA had assumed that the considerable aggregate numbers of chlamydia tests that were being taken in primary care, and the growing proportion of all tests from that setting, would be sufficient to underpin this trial assuming careful selection of practices. However, none of us had realised the degree of dispersion of chlamydia tests within NCSP in a very large number of practices, and the very small number of practices that were currently doing sufficient tests to make a minimally useful contribution to recruitment. This finding presented a challenge that was to prove fatal to our study, given that we were not able to change testing behaviours substantially.

It also presents an interesting challenge for future studies in this area. The vast majority of practices continue to make tiny numbers of diagnoses (one to five per year), despite longstanding national policy commitments to developing STI care in general practice. This has implications for the appropriate orientation of future research in this field. Two service problems that need future research support are:

1. How can any care pathway be delivered on such an occasional basis, yet to ensure that people diagnosed with STI in primary care are offered an equivalent PN service to those diagnosed in sexual health-oriented services?

2. If general practice does not provide these tests, where, given current policy as discussed in Chapter 7, can they be accessed by the many young people who are not able to use a conveniently located sexual health service?

Partner notification and its evaluation in future trials

We were unable, in the event, to standardise or operationalise the proposed three arms of this trial. As previously noted, although several three-way trials have indeed been published,9–11 none to date has compared contract, provider and patient referral directly. We used an innovative participative methodology to explore HA practices. This work demonstrated that, in practice, provider and contract referral are subsumed together in a process of negotiation over time in the case of common bacterial STIs including chlamydia and gonorrhoea. This was close to a model of PN used in a trial by Cleveland in which provider referral was offered if a partner had not presented within 3 days. 10 In contrast, ‘contract referral’ in its standard meaning was used almost exclusively by these practitioners in managing infections with longstanding infectivity that were considered to have more serious health consequences, such as HIV, hepatitis B and C, and syphilis.

The structures of the National Chlamydia Screening Programme and their implications for future sexual health research in primary care

As previously noted, the majority of local CSOs managed patients testing positive from general practice without the practice taking part in this process. There was a culture of separation between these two services, which made it difficult for staff in CSOs to believe that a competent service could be provided from general practice for these patients, even though sexual health work takes place routinely in general practice. Structural features of the NCSP at the time of the study also meant that it was difficult for local chlamydia-screening staff not to see the study as a ‘competitor’ for their target numbers and – in the longer term – providing a model of service that undermined the future of the CSO in its form at the time of the study.

It will be important for commissioners of research to consider the need for a background of consistency and mutual support for future studies and to work with commissioners (now the local authorities) to facilitate research effectively. This will be needed to underpin best practice in all settings and also to enable audit, evaluation and research into complex interventions such as PN in community settings. Although research in this field is difficult, the absolute numbers and proportion of people diagnosed with STIs in primary care are substantial. 12,13,70 It is essential that continued research enable them to receive the high-quality care they deserve as the provision in primary care evolves. 55,57