The Ballseye programme: a mixed-methods programme of research in traditional sexual health and alternative community settings to improve the sexual health of men in the UK

Aims

• To evaluate the public health and economic impact of different clinical approaches to STI testing in men.

• To develop an evidence-based algorithm for STI testing of asymptomatic men.

Specific objectives

• To compare factors associated with asymptomatic and symptomatic NCNGU.

• To determine whether or not symptom-based triage correctly identifies men at risk of STIs.

• To develop an evidence-based algorithm for the STI testing of asymptomatic men.

• To mathematically model the epidemiological and economic impact of removing microscopy from routine STI testing in asymptomatic men across a range of assumptions about pathogenicity and cause.

• To determine whether or not the removal of microscopy from routine STI testing in asymptomatic men is an appropriate strategy on cost-effectiveness grounds and to ascertain whether or not the resources associated with microscopy could be more cost-effectively used in alternative strategies that ensure better uptake of chlamydia and gonorrhoea screening.

Phase 1a: systematic review of the clinical consequences of asymptomatic non-chlamydial, non-gonococcal urethritis for men and their partners

This chapter contains information reproduced from Saunders JM, Hart G, Estcourt CS. International Journal of STD & AIDS (volume 22, issue 6), pp. 338–41, copyright © 2011 by SAGE Publications, Ltd. Reprinted by permission of SAGE Publications, Ltd. 40

Phase 1b: factors associated with asymptomatic non-chlamydial, non-gonococcal urethritis in men – findings from a case–control study

This chapter contains information reproduced from Saunders JM, Mercer CH, Sutcliffe LJ, Cassell JA, Estcourt CS. International Journal of STD & AIDS (volume 24, issue 8), pp. 627–31, copyright © 2013 by SAGE Publications, Ltd. Reprinted by permission of SAGE Publications, Ltd. 54

Mathematical modelling

We constructed a transmission dynamic mathematical model to synthesise evidence on the natural history and epidemiology of M. genitalium, quantify the parameters contributing most uncertainty and examine the potential public health impact (primarily increases in serious sequelae in women) of changes in the use of microscopy in asymptomatic men in genitourinary medicine (GUM) clinics, both ending its use with the 5% of patients who currently receive it and increasing its use by making it routine practice. Additionally, we explored the potential impact of a licensed specific (e.g. nucleic acid amplification) test for M. genitalium.

Methods

Model description

To represent the transmission dynamics of M. genitalium in a heterosexual population, we developed a deterministic compartmental model. The model population was stratified into male and female, with each having groups with high and low rates of sexual partner change to represent heterogeneity in sexual behaviour. Susceptible individuals (S in Figure 1) who become infected enter a latent state (L) before becoming infectious. Infectious individuals may be symptomatic and seek treatment prompted by their symptoms (I_SC), symptomatic without seeking treatment (I_SN) or asymptomatic (I_A). Others, both with and without symptoms, may seek care following partner notification. Successful treatment returns individuals to the uninfected susceptible state. Those unsuccessfully treated enter the treatment failure compartment (F) and will be retreated successfully or eventually recover naturally and return to the susceptible state or, in the case of women, may develop PID (P). In the absence of treatment, men eventually recover through natural immune processes and return to the susceptible state. Untreated women may also return to the susceptible state through natural recovery, but a proportion instead progress to PID, which may be symptomatic or asymptomatic. A proportion of women with symptomatic PID seek care because of symptoms and move into the treated PID state (T). All of those with PID may recover through natural processes and return to the susceptible state.

FIGURE 1.

Model flow diagram. Members of the population move through states as described in the text. F, failed treatment; I_A, infectious and asymptomatic; I_SC, infectious and symptomatic and care seeking; I_SN, infectious and symptomatic but not care seeking; L, latent infection; P, PID; S, susceptible; T, treated for PID.

Model parameterisation

Prior expectations of ranges for parameter values were defined, based on the literature, data from our previous studies and Barts Health NHS Trust sexual health services and expert opinion (Tables 3 and 4).

TABLE 3 - Model parameters 1a

F, female; m, male; max., maximum; min., minimum.

The table summarises prior expectations and posterior distributions of parameters, and sources of estimates.

Description	Gender	Symbol	Prior ranges	Mean	Posterior		Posterior mean	Source
					Min.	Max.
Organism-related parameters
Per-sex act transmission probability	F	β f	0.0–0.3	0.155	0.07	0.3	0.18	Anagrius et al.,7 Andersen et al.,58 Quinn et al.,59 Thurman et al.,60 Keane et al.61
	M	β m	0.01–0.3	0.155	0.08	0.3	0.23	Anagrius et al.,7 Andersen et al.,58 Quinn et al.,59 Thurman et al.,60 Keane et al.61
Natural recovery rate (per year)	F	γ F	0.3–2.4	1.2	0.46	2.37	1.23	Jernberg et al.,26 Cohen et al.,62 Oakeshott et al.,63 Smieszek and White64
	M	γ M	0.3–1.0	0.65	0.3	0.955	0.6	Jernberg et al.,26 Cohen et al.,62 Oakeshott et al.63
Proportion of infected who are symptomatic	F	z f	0.01–0.7	0.355	0.09	0.7	0.38	Falk et al.,5 Anagrius et al.,7 Lewis et al.65
	M	z m	0.1–0.7	0.4	0.1	0.57	0.2	Falk et al.,5 Anagrius et al.,7 Lewis et al.65
Treatment failure proportion		ξ	0.05–0.6	0.351	0.05	0.596	0.28	Bradshaw et al.,13 Jernberg et al.,26 Björnelius et al.66
Behavioural parameters
Proportion of contacts who are traced from GUM clinic		f gum	0.01–0.7	0.55	0.03	0.6	0.28	Turner et al.67
Proportion of contacts who are traced from general practice		f gp	0.01–0.5	0.255	0.01	0.5	0.19	Cassell et al.68
Proportion of patients who go straight to GUM clinic		prop gum	0.3–0.7	0.5	0.303	0.7	0.47	Cassell et al.69
Proportion of patients who go to GUM clinic from general practice		prop gpgum	0.1–0.4	0.25	0.103	0.4	0.24	Cassell et al.31
Proportion of infected seeking care who abstain from sex	F	α f	0.2–0.6	0.4	0.2	0.6	0.40	Mercer et al.,17 Irwin et al.,70 Fortenberry71
	M	α m	0.4–0.8	0.6	0.4	0.8	0.59	Mercer et al.,17 Irwin et al.,70 Fortenberry71
Proportion of symptomatic infected who go for treatment spontaneously	F	ρ f	0.01–0.455	0.6	0.01	0.87	0.26	Expert opinion
	M	ρ m	0.6–0.99	0.795	0.6	0.99	0.77	Expert opinion
Assortativity coefficient		ε	0.1–0.9	0.5	0.1	0.9	0.42	Expert opinion
Rate at which asymptomatic men go for screening spontaneously (per year)		σ M	0.001–0.138	0.089	0.005	0.132	0.076	Saunders et al.72
Amount of time before seeking care (days)		sk	0–20	10	0.05	19.76	9.46	Mercer et al.,17 Kretzschmar et al.73

TABLE 4 - Model parameters 2a

The table summarises prior expectations and posterior distributions of parameters, and sources of estimates.

Description	Symbol	Value	Source
Rate of ageing in and out	µ	1/22 per year	The model considers a 22-year age range
Natural recovery rate – treatment failure	γ fail	2 per year	Expert opinion
Natural recovery rate – PID	ϒγ Prec	0.25 per year	Expert opinion
Latent period	d lat	14 days	Jensen4
PID treatment duration	d tp	21 days	Expert opinion
Proportion of asymptomatic men screened with microscopy when at GUM clinic	η	5%	Carne et al.39
Specificity of microscopy	ν	97%	Smith et al.52
Sensitivity of microscopy	sens	80%	Smith et al.52
Delay before notifying partners	τ	10 days	Expert opinion
Proportion of population in low-risk activity class	nch	14%	White et al.,2 Johnson et al.74
Proportion of population in medium-risk activity class	ncm	30%	White et al.,2 Johnson et al.74
Proportion of population in high-risk activity class	ncl	56%	White et al.,2 Johnson et al.74
Partner change rate in high-risk activity class	r high	8.49	White et al.,2 Johnson et al.74
Partner change rate in medium-risk activity class	r medium	0.76	White et al.,2 Johnson et al.74
Partner change rate in low-risk activity class	r low	0.15	White et al.,2 Johnson et al.74
Sex-act frequency per 4 weeks	q	7	Expert opinion

The ranges were well defined for some parameters, as previous studies had acquired reliable data on their distributions. These parameters included lags associated with seeking care, receiving care and notifying partners; the proportion of patients abstaining from sexual activity when seeking care for symptoms; the proportion of partners traced; sexual partner change rates (from 2000); and treatment efficacy and failure rates. Other parameters such as the proportion of infections that are symptomatic, latent period and duration of infection, transmission probability and rate of progression to PID were less well defined either because of a lack of information about the parameter or a wide range of estimated values for it from different studies.

The model population consisted of 10,675,000 males and females aged 16–30 years (equal numbers of each sex, i.e. approximately 5 million each).

The model was fitted to KC60 surveillance data on annual diagnoses of NCNGU in men, collated by the Health Protection Agency (now Public Health England) from GUM clinics throughout the UK,1 of which 10–46% are estimated to be due to M. genitalium,75 and the prevalence of infection in women estimated from the study by Oakeshott et al. 63

Parameters that were varied in fitting were transmission probabilities (female-to-male and male-to-female), the proportion of incident infections that are symptomatic in females and males, the proportion of those with symptomatic infection who seek care, the rates of natural recovery from infection, the mixing coefficient (which varies how assortative sexual partner choice is) and the sensitivity of microscopy. These parameters were varied in combination across their prior ranges and the resulting model run compared with the target values for annual diagnoses in GUM clinics in men (10–46% of NCNGU diagnoses), and the prevalence in women derived from the study by Oakeshott et al. 63 Candidate parameter sets were accepted if both the criteria were met.

We examined the effect of stopping microscopy in the 5% of asymptomatic men in GUM clinics who still receive it, while continuing to use microscopy in symptomatic men in all GUM clinics. This means that asymptomatic NCNGU in men is not diagnosed and so not treated.

The incidence of TFI and ectopic pregnancy was estimated based on the incidence of PID in the model. Haggerty et al. 76 estimated that 18% of cases of symptomatic PID were followed by TFI; Westrom20,21 estimated the frequency to be 12%; Hillis et al. 15 estimated that the incidence of ectopic pregnancy following TFI was 8%. It is unclear how frequently TFI and ectopic pregnancy occur following asymptomatic PID; it may be that the risk of these sequelae is lower because of infection being less severe, but asymptomatic PID is unlikely to be treated and therefore likely to be longer lasting, which may increase the risk of serious sequelae.

Results

The posterior ranges for the fitted parameters are shown in Table 3. There is much uncertainty in the proportion of NCNGU caused by M. genitalium and in the natural history of infection, which leads to uncertainty in the effects of reducing the use of microscopy in GUM clinics.

Stopping microscopy in asymptomatic men has the immediate effect of reducing the quantity of microscopy procedures performed. The consequent reduction in diagnosis and treatment of asymptomatic infection results in an increase in the average duration of asymptomatic infection in men (as all now recover as a result of natural immune processes rather than some through treatment). This causes an increase in transmission to women and consequently an increase in transmission to men. Thus, there is more disease in women and men because of an increased incidence of symptomatic infection (as well as asymptomatic infection), leading to an increase in general practitioner (GP) and GUM consultations and an increase in the incidence of PID and consequent serious sequelae (TFI and ectopic pregnancy) (Figure 2 and Table 5). The most marked effect is in women, because symptomatic infection in men is still mostly diagnosed and treated effectively. Another reason for the increase in GP and GUM consultations – and reduction in partners managed as notified partners – is that asymptomatic infection in men is no longer being diagnosed and so their partners are no longer notified and women with symptoms who would previously have sought care as a notified partner will now seek care because of their symptoms and be treated as an index case instead. The reduction in treatment because of asymptomatic infections no longer being treated is partially offset over time by the increase in symptomatic infection that does get treated. Over both 10- and 20-year time horizons there are net reductions in the amount of microscopy performed, treatment given and notified partners managed and net increases in infection (both symptomatic and asymptomatic) in both sexes, the incidence of PID and serious sequelae and GP and GUM consultations of index cases.

FIGURE 2.

Impact of changing the use of microscopy in men in GUM clinics. (a) Total female infections; (b) total male infections; (c) PID cases.

TABLE 5 - Impact of changing the use of microscopy in men in GUM clinics

Outcome	Total cases over 20 years
	Likelihood-weighted cumulative incidence	Elimination of microscopy (excess cases occurring)
PID	2,374,364	2,379,943 (5578)
Ectopic pregnancy	189,949	190,395 (446)
TFI	284,924	285,593 (669)

The graphs in Figure 2 show the per cent change in the incidence of PID, TFI and ectopic pregnancy in women resulting from cessation of urethral smear microscopy in asymptomatic men in GUM clinics. In total, 25% of the simulations lie in the dark orange area; 50% lie in the mid-orange and dark orange areas; 75% lie in the light orange and mid-orange and dark orange areas; and 95% lie in the four shaded areas (orange areas and yellow area). The median of the simulations is represented by the line between the light-orange and mid-orange areas.

Economic analysis

Methods

To appropriately estimate the impact of testing and treatment on the future transmission of NCNGU, it is necessary to use an appropriate modelling approach for infectious diseases that can describe the transmission of infection between individuals, a transmission dynamic model (TDM). 77,78 In this study, a TDM describing the transmission of NCNGU in a population of 16- to 30-year-olds in England was constructed to examine changes in the use of microscopy in asymptomatic men in GUM clinics. This economic evaluation relies exclusively on the outputs from this model, along with secondary data describing resource use, and takes the form of a cost-effectiveness analysis carried out from a health-care provider perspective.

Model structure

The output used in this economic analysis is taken from a TDM, which has been described in full (pp. 13–17). In brief, this is a compartmental transmission model which describes heterogeneous sexual behaviour that was parameterised by behaviour data from a number of key UK surveys and national surveillance data and with the natural history of NCNGU being informed from data in the literature. The model describes the incidence and prevalence of symptomatic and asymptomatic infection, PID, care-seeking behaviour as a result of symptoms, partner notification and the possibility of treatment failure. The uncertainty of the parameters in the model was also factored into the model parameterisation.

The time horizon for the economic analysis was 20 years, although this was subject to sensitivity analysis. It was felt that a time horizon that was longer than this would not be appropriate, as assumptions made in the transmission model regarding heterosexual behaviour amongst 16- to 30-year-olds are unlikely to be valid the further into the future the analysis is conducted. A discount rate of 3.5% was applied to the costs and outcomes in accordance with National Institute for Health and Care Excellence (NICE) guidelines. 79

All settings where sexual health services are provided were initially considered for inclusion in this economic analysis, from general practices and GUM clinics to pharmacies and educational settings. However, guidelines detailing the specific pathways and resources used at different sexual health service settings were sparse, with the most reliable clinical data and cost data found in the literature being related to general practice and GUM settings. Therefore, in this study the methodological focus was narrowed to NCNGU in primary care and GUM clinics.

Testing pathways for economic analysis

Two different pathways were compared in terms of their resource use and costs, each representing alternative approaches to the testing and treatment of patients with asymptomatic NCNGU. These pathways represent current practice with respect to the implementation of microscopy screening in men and the impact of withdrawing the microscopy test for asymptomatic men in the GUM setting. The two pathways are referred to in this study as ‘current practice’ and ‘no screening’.

Initially, a patient can be either infected or non-infected with M. genitalium and either symptomatic or asymptomatic. The patient may attend either a general practice or a GUM clinic for screening. Symptomatic patients are index patients with symptoms resulting from their underlying infection, such as dysuria and urethral discharge.

In the general practice setting, regardless of the strategy, all patients (asymptomatic and symptomatic) receive the NAAT for chlamydia and gonorrhoea. A proportion of the patients are then referred from the general practice setting to a GUM clinic for further diagnosis or treatment, for example those who are symptomatic or who have more complex sexual health needs.

In contrast, in the GUM setting the diagnostic approach varies depending on which strategy is being considered and whether a patient presents with symptoms or not. For the no screening strategy, microscopy screening is not offered to asymptomatic patients and these patients receive an NAAT only. All asymptomatic patients in all locations in the current practice scenario receive a NAAT, with one in 20 asymptomatic male patients receiving a microscopy test in a GUM setting (value derived from McClean et al. ’s16 national GUM clinic audit). During the course of the consultation all symptomatic patients in a GUM setting are also provided with partner notification leaflets and condoms, with the aim of attracting more contacts with infected patients to be tested and treated. The testing pathways considered in this study are summarised in Tables 6 and 7.

TABLE 6 - ‘Current practice’ testing scenario

Group	Setting	Testing strategy
		Men	Women
Asymptomatic patients	General practice	NAAT only	NAAT only
	GUM clinic	NAAT; 5% of all asymptomatic patients receive microscopy39	NAAT only
Symptomatic patients	General practice	NAAT only	NAAT only
	GUM clinic	Microscopy and NAAT	NAAT only

TABLE 7 - ‘No screening’ testing scenario

Group	Setting	Testing strategy
		Men	Women
Asymptomatic patients	General practice	NAAT only	NAAT only
	GUM clinic	NAAT only	NAAT only
Symptomatic patients	General practice	NAAT only	NAAT only
	GUM clinic	Microscopy and NAAT	NAAT only

In this analysis, treatment can be deemed either a success or failure. Successful treatment indicates that a patient is no longer infected with M. genitalium and cannot transmit infection to their sexual partners. Treatment failure indicates that there has been a failure of the drug treatment to clear the STI infection. Female patients who fail treatment or who are not treated can develop PID, a proportion of which is treated. Female patients untreated for PID may go on to experience symptomatic PID, infertility or an ectopic pregnancy.

Assumptions and parameterisation

This cost-effectiveness analysis is parameterised through secondary sources, which are described below. It was necessary to make some pragmatic clarifying assumptions to carry out the analysis. All assumptions were confirmed and agreed with clinical experts within the team before the analysis was carried out.

The assumptions made were:

• Symptomatic patients are index patients with symptoms resulting from their underlying infection.

• HIV and syphilis blood tests are administered 5% of the time in a GP setting (expert opinion).

• All patients in all settings receive a NAAT test.

• A GP consultation takes 11.7 minutes, which is the average for a surgery consultation. 80

• 50% of patients in a GUM setting are seen by a doctor and 50% by a band 7 nurse (expert opinion).

• Partner notification is conducted by a band 7 nurse with all index patients (symptomatic) in a GUM clinic setting who are positive and this takes 12 minutes. 49

• No formal partner notification is conducted in a general practice setting, with just brief ‘words of advice’ being given, which are not considered in this economic analysis.

• Taking a case history takes 5 minutes for asymptomatic patients in a GUM setting (expert opinion – study team).

• Taking a case history takes 10 minutes for symptomatic patients in a GUM setting (expert opinion – study team).

• Examination of a female patient in a GUM setting takes 10 minutes (expert opinion – study team).

• Examination of a male patient in a GUM setting takes 5 minutes (expert opinion – study team).

• A single dose (1 g) of azithromycin is given as treatment for NCNGU.

• For all microscopy tests implemented it takes 10 minutes for a laboratory technician to obtain and report the results (expert opinion – study team).

• The treatment for PID considered in this study is 500 mg of ceftriaxone in a single dose followed by 100 mg twice daily of oral doxycycline plus 400 mg twice daily of metronidazole for 14 days. 81

• All notified partners are assessed and presumptively treated in a GUM location and are asymptomatic (expert opinion – study team).

The non-cost parameters used in this cost-effectiveness analysis are shown in Table 8.

TABLE 8 - Resource use and costs

Parameter	Value (range)	Source
Proportion of times a HIV test is delivered alongside a NAAT test in a GUM setting	83% (71–97%)	Irwin et al.70
Proportion of times a syphilis test is delivered alongside a NAAT test in a GUM setting	84% (72–97%)	Irwin et al.70
Proportion of PID cases that give rise to ectopic pregnancy	(99/1309) 7.6% (6.4–8.8%)	Donovan,14 Iwuji et al.82 Based on number trying to conceive after laparoscopy-diagnosed PID. Range calculated from a beta distribution taking values at 5% and 95% parameterised using the method of moments83
Proportion of PID cases that give rise to infertility	18% (15–21%)	Donovan,14 Cohen et al.62 Range calculated from a beta distribution taking values at 5% and 95% assuming that the standard error = mean/1083
Proportion of PID cases that are symptomatic	56% (30–89%)	Value from posterior mean of the infectious disease model
Delay from PID to manifestation of infertility/ectopic pregnancy	5 years (1–15 years)	Expert opinion – study team

The costs used in this analysis were valued at 2011/12 UK prices. The cost of partner notification was adjusted to 2011/12 prices using the pay and price index for Hospital & Community Health Services. 80 Unit staff costs were obtained from Unit Costs of Health and Social Care 2013. 80

The unit costs of the resources used in this economic evaluation are described in Table 9 and the application of each of the unit costs to the output from the dynamic transmission model is shown in Table 10. The ranges for the costs were established by, first, describing the variation in the length of time of a procedure through the use of a gamma distribution, with the mean = the standard error and taking the values at the 5% and 95% points (with the minimum consultation time set to 2 minutes), and, second, when more than one member of staff is assumed to contribute to an examination, assuming that the cheapest and most expensive members of staff conduct the examination for the lower and upper values of the range respectively. Some of the resources in Table 9 do not have ranges as their unit costs are fixed and do not vary.

TABLE 9 - Resource use and costs

BNF, British National Formulary; HRG, Healthcare Resource Group.

Number	Resource	Unit cost (£)	Range (£)	Source
1	NAAT	9.15		£7.35 for a swab, 2005 prices = cost of hands-on time + equipment and consumables cost per test;49 2012/13 prices = £9.15
2	HIV test	8.36		Rapid test kit £5–1184 (mid-point 2012/13 prices)
3	Syphilis test	2.00		EIA, assume £2
4	Microscopy test (including staff costs)	7.00		HRG DAPS07 Microscopy85
5	Laboratory technician (10 minutes of staff time) to obtain and report results of microscopy test	3.50	0.35–10.50	Clinical support worker nursing (community) (£21/hour)80
6	Staff time to give results of microscopy in GUM setting (5 minutes of staff time)	6.67	1.33–20.00	5 minutes with nurse advanced (£80/hour)29
7	Azithromycin – drug cost for treatment	2.17		250-mg tablets, four-tablet pack (£2.17)86
8	General practice visit (excluding testing costs)	44.85	3.83–134.17	Includes direct care staff costs (with qualification costs, £230/hour); assume 11.7-minute surgery consultation29
9	Cost of partner notification in GUM setting – administered to all symptomatic patients	16.00	1.33–48.00	12 minutes with nurse advanced (£80/hour)29
10	Partner notification leaflets plus condoms given out during partner notification	1.00		Assumed cost
11	Asymptomatic female at GUM clinic – case history plus examination (13 minutes of staff time), 50% with band 7 nurse and 50% with GP	31.50	2.67–129.00	Nurse advanced cost per hour in surgery excluding qualification costs (£80/hour) + GP cost excluding direct care staff costs (without qualification costs) (£172/hour of patient contact)29
12	Symptomatic female at GUM clinic – case history plus examination (18 minutes of staff time), 50% with band 7 nurse and 50% with GP	42.00	2.67–172.00
13	Asymptomatic male at GUM clinic – case history plus examination (10 minutes of staff time), 50% with band 7 nurse and 50% with GP	21.00	2.67–86.00
14	Symptomatic male at GUM – case history plus examination (15 minutes of staff time), 50% with band 7 nurse and 50% with GP	31.50	2.67–129.00
15	Cost of treating PID	16.20		Ceftriaxone 500-mg single dose (1-g vial) (£9.58); doxycycline 100 mg twice daily for 14 days, 100-mg 8-capsule pack (£1.05 × 4); and metronidazole 400 mg twice daily for 14 days, 21-tablet pack (£1.21 × 2)86
16	Ectopic pregnancy	729.00		MA18C medical termination of pregnancy – < 14 weeks’ gestation, elective inpatient85
17	Infertility	579.01		£428 in 2003.87 Hospital & Community Health Services pay and price index80 used to inflate to 2012 prices

TABLE 10 - Application of resource usage to model states from the dynamic transmission model

Model state	Costs applied	Notes
GP consultation symptomatic females	1–3, 8
GP consultation symptomatic males	1–3, 8
GUM consultation asymptomatic females	1–3, 11
GUM consultation symptomatic females	1–3, 9, 10, 12
GUM consultation asymptomatic males	1–3, 9, 10, 13
GUM consultation symptomatic males	1–3, 9, 10, 14
PID treated	15	Only considers the actual treatment, with consultation/staff costs having been considered elsewhere
Microscopy	4–6	For asymptomatic and symptomatic men only
All treatment for NGNCU	7	Includes true positives and false positives treated in error
Partner-notified female partners of males	1–3, 11
Partner-notified male partners of females	1–3, 13
Ectopic pregnancy	16	Number based on total number with untreated PID and the delay from PID to manifestation
Infertility	17

Outcomes

The main outcome measure for this evaluation was the cost per case of PID averted. The second outcome measure was the cost per major outcome averted (MOA), with a major outcome defined as a case of symptomatic PID, a case of ectopic pregnancy or a case of infertility. All major outcomes are reported for completeness.

Analysis

The base-case scenario uses the mean results of 215 parameter sets from the dynamic transmission model and applies the resource costs to obtain the baseline deterministic results. Results are presented in terms of incremental cost-effectiveness ratios (ICERs), which presents the additional cost per additional unit of outcome for one strategy compared with another, in this case current practice compared with no screening.

Sensitivity analysis

To examine the impact of uncertainties in the model parameters and the assumptions made on the model results, sensitivity analysis was conducted. Two scenarios were examined to assess the impact of increasing and decreasing the costs applied in the study:

• Minimise costs. All costs were set to a minimum by taking the lowest realistic length of time for each consultation (minimum 2 minutes). When two staff members undertake a consultation at baseline, only the lowest paid was assumed to carry out the consultation.

• Maximise costs. All costs were set to a maximum by taking the highest realistic length of time for each consultation. When two staff members undertake a consultation at baseline, the highest paid staff member was assumed to carry out the entire consultation.

In addition, three further outputs from the transmission model were also analysed to assess their impact on the model results. These were the median results from the 215 parameter sets of the TDM and the upper and lower results from the 95% range of values.

Further one-way sensitivity analysis of key parameters was also investigated, with particular attention paid to parameters that were estimated through expert opinion. In addition, the time horizon was also varied to show its impact on conclusions drawn from the model.

Mean output from the transmission model

The graphs in Figure 3 provide a summary of the mean output from the transmission model that was utilised in this economic analysis at baseline for each of the two scenarios considered in this study.

FIGURE 3.

Output from the transmission model used in the baseline economic analysis with variation in the testing strategy. (a) GP consultations; (b) GUM consultations; (c) cases of PID; (d) true positive patients with NGNCU being treated; (e) cases of infertility; and (f) cases of ectopic pregnancy.

It can be seen from Figure 3 that increasing the coverage of microscopy leads to a reduction in the annual number of consultations in both general practice and GUM settings. Moreover, current practice also has a positive impact on reducing the number of cases of PID, the number of cases of infertility and the number of cases of ectopic pregnancy. It also lowers the number of true-positive patients with NGNCU being treated because of its impact on onward transmission.

Discussion

This economic evaluation utilised the output from a TDM to assess whether or not it is feasible to remove the use of microscopy to test asymptomatic men for NGNCU or whether or not the current practice of limited microscopy coverage for asymptomatic men is a relatively cost-effective option. The primary outcome measure used in this study was case of PID averted, with the secondary outcome being MOA (symptomatic PID, infertility or ectopic pregnancy).

The results of the base-case analysis indicate that the current practice of giving approximately 5% of asymptomatic men in the GUM setting a microscopy test has an incremental cost of £15,800 per case of PID averted, meaning that this strategy invests approximately £15,800 to generate one additional case of PID averted compared with a no screening strategy in which only symptomatic men are screened. In terms of MOA, current practice has an incremental cost of £50,200, meaning that this strategy invests approximately £50,200 to avert one major outcome.

Across all of the sensitivity analyses undertaken, current practice was never found to be cost saving but was always found to have a positive impact on reducing cases of PID and major adverse outcomes. Varying the outputs from the TDM provided a range of values for the outcomes in this study. For the case of PID averted, the ICER values ranged from £10,500 to £39,400, whereas for MOA the ICER values ranged from £33,800 to £125,000. By varying the time horizon of the analysis it was found that shorter time horizons made the intervention less cost-effective.

By implementing extreme scenarios in which the costs were minimised and maximised to the greatest extent, the ICER values for case of PID averted and MOA ranged from £9600 to £34,300 and from £30,500 to £109,400, respectively. However, these results must be approached with caution as it is very unlikely that these minimum and maximum costs across multiple parameters used in this analysis would be seen in practice.

This study has utilised the output from a well-parameterised dynamic model that describes the transmission of NGNCU in a population of males and females aged 16–30 years in England. Uncertainty in this model has been considered through the use of multiple parameter sets, whereas the results from this economic evaluation have been subject to extensive sensitivity analyses. Inevitably this has led to the range of plausible values obtained from the economic model being quite wide, although this does help to give confidence to the validity of the conclusions that might be drawn from this model.

A weakness of this study was the inability to conduct probabilistic sensitivity analysis for both the economic parameters and the parameters utilised in the TDM. Although it was possible to conduct probabilistic sensitivity analysis for just the economic parameters while maintaining the output from the TDM at constant values, the results describing the probability of a strategy being below a specific acceptable threshold would be meaningless.

To our knowledge this is the first economic analysis related to NGNCU in any setting and thus comparisons with the results from similar economic studies are impossible.

It is suggested that UK decision-makers are unlikely to fund an intervention if it has an ICER of > £30,000 per quality-adjusted life-year (QALY). 79 However, for the outcome measures used in this study, there are no accepted threshold values from which a conclusion can be drawn on whether an intervention should be accepted or not. It is therefore necessary to link the results here to the acceptance threshold values for the QALY to draw conclusions from this economic analysis.

Taking mean values from the transmission model, the ICER for the strategy of screening compared with no screening was £15,800 per case of PID averted and £50,200 per MOA. This means that it costs an additional £15,800 to avert each additional case of PID and an additional £50,200 to avert each major outcome as a result of screening compared with no screening. For case of PID averted, for this to be acceptable when using the maximum acceptance threshold of £30,000 per QALY from NICE, a case of PID averted would have to result in > 0.53 QALYs gained. This means that having PID would have to be equivalent to > 6 months in a state of death. Likewise, for MOA, this would have to result in > 1.67 QALYs, meaning that having a major outcome would have to be equivalent to > 18 months in a state of death for microscopy screening to be deemed cost-effective.

These observations can be extended by comparing them to previous estimates of the impact of PID sequelae on quality of life. Smith et al. 88 found that, using time trade-off amongst respondents with a previous history of PID, the mean valuations for health states associated with PID were 0.79 [standard deviation (SD) 0.34] for ectopic pregnancy, 0.69 (SD 0.37) for pelvic pain and 0.76 (SD 0.34) for infertility. These values suggest that the mean QALY gain to avert a case of pelvic pain (the state with the reported greatest negative impact on quality of life) that lasted 1 year would be 0.31 QALYs. However, as noted above for the results described here, for current practice to be cost-effective a MOA must result in a gain of > 1 QALY, suggesting that current practice is far from being cost-effective.

Given the comparisons described above, and acknowledging the uncertainty in the results as described above, we tentatively suggest that the current practice of ad hoc testing of asymptomatic men in GUM locations is unlikely to be cost-effective. Considering the results at baseline in this study, if all microscopy testing for asymptomatic men in GUM locations was to be withdrawn then this would have the potential to save > £5M (discounted) over a 20-year period. This could then be better spent expanding testing and treatment regimes for different diseases in areas that are more cost-effective.

One of the major issues related to any testing and diagnostic strategy is the impact of the testing pathway on patients. Patients may suffer from anxiety while waiting for the results of a test or may incur societal costs as a result of having to take time off work to attend for testing. There are also issues specific to the context of STIs, with patients worried about the stigma of attending for testing and receiving a positive diagnosis. In the testing and diagnosis context, future work should focus on these issues to better quantify their impact on patients. The impact of these issues could then be included in economic studies such as this, to better describe the true impact of the complete testing pathway on patients.

One limitation of the analysis of this screening strategy and many analyses like it is that no account of the knock-on effects of testing are considered. As a result of patients coming into contact with health-care professionals (HCPs), there is the opportunity to provide them with health-care advice and also intervene in other diseases, which is outside the scope of this analysis. For example, in this study there is the possibility that patients may be identified as having syphilis or HIV infection, which may lead to quicker access to treatment and a reduction in onward transmission for these conditions. It would obviously be extremely difficult to incorporate more than one infectious disease into the same transmission model, but it would certainly be more realistic were this to be possible.

Background

Across England, reported diagnoses of genital C. trachomatis infection, the commonest STI, continue to rise. 90 The proportion of infections diagnosed in primary care settings has increased90 and, in 2013, 58.2%91 of all cases of chlamydia in people aged < 25 years in England were diagnosed outside the specialist GUM setting, where partner notification services may not be routinely available.

Recent mixed-methods and mathematical-modelling studies suggest that improving the effectiveness of partner notification, the process of identifying exposed sex partners and offering them testing and treatment,92 could have a more marked impact on population control of chlamydia than widening access to testing. 93 The aims of partner notification are twofold: to reduce the burden of infection by reducing reinfection from untreated sex partners and to reduce onward transmission. However, the optimal partner notification strategy for people with chlamydial infection is unknown. A recent systematic review of RCTs of partner notification91 for STIs causing cervicitis and urethritis suggests that expedited partner therapy (EPT), in which the doctor provides the index patient with antibiotics or a prescription to give to their sex partner,94 is more effective than simple patient referral [verbal advice that the partner(s) should attend] in preventing index reinfection and achieves treatment of a higher proportion of sex partners. However, EPT does not appear to be more effective than enhanced patient referral (simple patient referral supplemented by written advice, longer verbal explanation and/or internet resources) in preventing index reinfection and is not associated with higher numbers of partners treated. 93

UK prescribing guidance does not support EPT in this form95 as it does not include a medical assessment of the sex partner. We have previously developed a new form of EPT known as APT, which uses health adviser telephone-led and pharmacist-led partner notification interventions and includes medical assessment of the sex partner. 95 These meet UK prescribing guidance and have shown promise in the specialist setting. 95,96 Furthermore, modelling studies suggest that APT could play an important role in reducing index reinfection and the prevention of costly reproductive health consequences, especially in women,93,94 by reducing the time to successful partner treatment. 95

Successful control of chlamydia will need effective methods of partner notification, irrespective of the setting in which the infection is diagnosed. Partner notification is known to be problematic in primary care68,97 but has been shown to be achievable in the context of a chlamydia screening trial. 98 Barriers to effective partner notification include lack of knowledge/skills, lack of time and organisational issues (e.g. partner is not patient of index practice). 99 Although GPs appear to be broadly supportive of APT methods of partner notification,97 this type of intervention may not directly translate to primary care settings in which staff may have fewer specific sexual health skills and service users may be different from those attending specialist services.

Objectives

The objective of this study was to determine the feasibility, acceptability and preliminary evidence of effectiveness of APT for women diagnosed with chlamydia in the non-specialist setting by conducting a pilot RCT of APT in contrasting primary care settings in England.

Trial design

We carried out a three-arm pilot RCT of two APT interventions – APTHotline and APTPharmacy – compared with routine care (patient referral) with allocation in a 1 : 1 : 1 ratio.

Participants

Inclusion criteria were women aged at least 16 years, diagnosed with genital C. trachomatis infection (index patients) and at least one contactable male sex partner in the last 6 months. Exclusion criteria for index patients were known HIV-positive status, co-infection with other STIs as these would require tailored partner notification and/or different epidemiological treatment of partners and an inability to understand English. Exclusion criteria for male sex partners were symptoms of complicated infection, allergy or contraindications to azithromycin and an inability to understand English as we did not consider that it was medically appropriate to conduct a telephone clinical assessment if someone was unable to fully comprehend the assessment.

Settings

The trial took place between 1 September 2011 and 31 July 2013 in two areas with diverse and contrasting patient populations: inner East London, which has a deprived young and ethnically mixed population [six general practices and two community (non-specialist) CASH services, supported by five pharmacies], and the south coast of England, which serves non-metropolitan populations, some of which are distant from GUM clinic provision and which therefore face contrasting challenges in access to services [six general practices and one community (non-specialist) CASH service, supported by nine pharmacies]. The pharmacies provided broad geographical coverage around each recruitment site.

Recruitment of health services

General practices and CASH services within the relevant geographical areas with the highest chlamydia screening and positivity rates were identified by the researchers and approached by e-mail from the chief investigator to the lead GP or lead clinician (CASH service) to introduce the study. A week later a researcher telephoned the lead GP/clinician to ascertain his/her level of interest in participating. A date was arranged to present the study to the whole general practice/CASH service and an agreement to take part in the study was obtained. Some general practices and one CASH service routinely devolved the provision of chlamydia test results and management for women aged < 25 years to the local National Chlamydia Screening Programme (NCSP) office. When this was the case, we worked with the relevant NCSP offices, as directed by the general practices/CASH service involved.

Independent or small-chain pharmacies (administratively easier than large-chain businesses) were initially chosen by the researchers for their proximity to good transport links (train, tube or bus) and their geographical coverage within the study area. A researcher then either telephoned or walked into each pharmacy to introduce the study to the lead pharmacist. Interested pharmacists were e-mailed further detailed information about the study and a researcher telephoned a week later to seek their agreement to take part. With some chain pharmacies, if the lead pharmacist was interested in participation the researchers also had to approach his or her business development manager to obtain permission for the pharmacy to participate.

Initially, six general practices, two CASH services and five pharmacies across Tower Hamlets Primary Care Trust (PCT) and City & Hackney PCT (London) and four general practices and five pharmacies across NHS East Sussex Downs & Weald PCT were recruited. However, during the trial there were extensive changes in NHS organisational structures100 and research had a low priority within the general practice settings. We decided to recruit a third CASH/NCSP service outside London (Portsmouth Sexual Health Services CASH Clinic and NCSP) with high positivity rates and a further four pharmacies in the Portsmouth area.

Participant recruitment

Development and features of the online patient and data management tool

We applied a user-centred approach to the design of a bespoke web-based tool to support the APT patient pathway in primary care, enabling rapid communication between the general practice, CASH service and the supporting hotline and pharmacy services. This was based on a similar tool used in a previous trial101 and a tool used in a related study of the primary care management of STIs. 102 It relies on a secure website, compliant with NHS encryption and data transfer protocols.

We included a pre-design phase involving discussions with primary care health professionals and sexual health clinic health advisers to determine their needs, current activity and work habits, to ensure that the web-based tool could be successfully integrated into their daily work practice. The web tool was then developed according to our specification by EpiGeneSys, a software company affiliated with Sheffield Clinical Trials Unit.

We chose a web-based format so that it did not require software installation or downloads. Data were collected in real time and all partner notification and other study outcomes were recorded. In addition, the web tool removed patient-identifiable data, which enabled the trial manager to see inputted data from all study sites, allowing real-time monitoring of recruitment and follow-up.

Other potential advantages of the web tool included a full audit trail, which allowed monitoring of the input and modification of records and users’ interactions with the system; access for health advisors to all index and sexual partner(s) reports to allow follow-up of partner notification; and the ability to download non-identifiable data by statisticians and health economists working on the study.

Interventions

All index patients were offered standard partner notification for all of their partners whereas those randomised to either of the APT options were offered APT in addition to standard partner notification for all of their partners. We define standard partner notification pragmatically as the routine method of partner notification offered by that service. In general practice this consisted of simple patient referral in which the GP or practice nurse advised the index patient of the need to inform her sex partners to undergo testing and obtain treatment. 81 Within the CASH services, enhanced patient referral was the standard partner notification offer. Enhanced patient referral consists of simple patient referral supplemented by written or online information about chlamydia and/or specific information on local health services at which the partner(s) could obtain appropriate care. 81 It is important to note that, as partner notification for these settings did not take place within the context of a GUM clinic but instead remotely, a direct and immediate offer of STI and HIV testing on first contact between the partner and the service did not take place. STI and HIV testing therefore depended on engagement either with self-sampling (urine sample kits for C. trachomatis and N. gonorrhoeae testing were included in the APT pack) or later attendance at a sexual health service for all three intervention arms.

Once the index patient gave consent the health professional began the partner notification consultation and filled in data using the web tool. Once the consultation was finished the health professional clicked a tab on the web tool that randomised the index patient’s sexual partners to receive routine partner notification, routine partner notification plus APTHotline or routine partner notification plus APTPharmacy. If either APTHotline or APTPharmacy was allocated, the web tool simultaneously and automatically sent a text message to the index patient with a unique PIN, which she was able to forward on to her sexual partners/contacts as her sexual contacts needed to quote this unique PIN to receive APTHotline or APTPharmacy treatment. If the index patient was allocated to routine care she did not receive a text message.

Details of the interventions are shown in Figure 4.

FIGURE 4.

The APTPC trial interventions. APT pack – contained prepackaged azithromycin (1 g), condoms, chlamydia information leaflet, a urine sample collection kit for C. trachomatis NAAT with instructions to provide the sample before taking the antibiotics, a prepaid postal envelope and sample packaging for returning the sample to the study clinic and a patient information leaflet about the study. APTHotline – sex partner undergoes telephone consultation with a sexual health clinic health adviser to determine eligibility for APT and receives an invitation for future clinic-based HIV and syphilis screening. Sex partner or his representative collects the APT pack from the clinic reception. Sex partner posts back his completed C. trachomatis NAAT urine sample kit (contained in the APT pack). Results and future clinical care are managed by a specialist clinic. APTPharmacy – sex partner undergoes consultation with sexual health-trained community pharmacist to determine eligibility for APT and receives an invitation for future clinic-based HIV and syphilis screening. Pharmacist gives sex partner the APT pack at the time of consultation. Sex partner posts back his completed C. trachomatis NAAT urine sample kit (contained in the APT pack). Results and future clinical care are managed by a specialist clinic. Standard partner notification (control) – for the purposes of this study we define standard partner notification as the HCP advising the index patient to notify his/her sex partner of the need for treatment. In the CASH services this was supplemented by written information about chlamydia and the provision of details of local sexual health services to the index patient for her to give to her sex partner. APTPC, APT delivered in primary care settings; PGD, Patient Group Direction; PN, partner notification; STS, syphilis screening.

Outcome measures

The primary outcome, determined for each contactable partner, was whether they were considered to have been treated within 6 weeks of the index diagnosis. Accordingly, primary outcome partners for whom treatment information was unavailable were considered untreated and there was, by definition, no loss to follow-up.

Secondary outcome measures were determined for the index patient and for each partner and included whether the partner was notified that they were at risk of infection; partner uptake and acceptability of partner notification modalities, number of partners treated per index patient; number of contacts notified per index patient; time to partner treatment; and C. trachomatis reinfection/persistence rates in index patients 4–6 weeks after treatment.

We also collected resource data to inform our health economics evaluation, which are reported in full on pp. 45–48.

Ascertainment of outcomes

Health advisers staffing the APTHotline and pharmacists providing treatment entered clinical and partner notification data directly onto the web tool at the time of the patient interaction. Health advisers conducted follow-up telephone assessments with all index patients 4–6 weeks after treatment to elicit outcomes of partner notification for partners who did not access APT, to obtain routine clinical data and to explore the acceptability of the interventions. To determine reinfection/persistence, we sent index patients who could be reached for telephone follow-up and who agreed a urine collection kit for C. trachomatis NAAT testing, which they mailed to the study laboratory for testing.

Process evaluation

We carefully monitored the recruitment process by observing clinical staff at the different sites as they offered the trial to eligible patients, visiting pharmacies to observe practice and monitoring hotline consultations. We conducted informal interviews with a range of clinical staff involved with the study. We explored issues of acceptability and preference in more detail by undertaking limited qualitative telephone interviews with a sample of female index patients and with male sex partners. A preliminary thematic analysis was undertaken of the qualitative data.

Sample size

The study was designed as a pilot trial, in accordance with the Medical Research Council (MRC) framework for the development of complex interventions,103 with randomisation of index patients on an individual basis, recognising that there would be clustering of partners by index patient. The unit of analysis was the individual male partner. We assumed that 25% of male partners would be reported (by index or health-care service) as treated via routine partner notification in primary care in comparison to 40% of male partners who would be reported as treated via either of the APT options. We chose 80% power and a 2% significance level (to take account of the need to make comparisons across three arms: routine care, APTHotline, APTPharmacy) and applied a 10% design effect (to take account of clustering of an average of 1.5 male partners per female index patient). We therefore sought to recruit 400 partners to participate across the 10 practices in the two PCTs.

However, in light of recruitment rates being lower than anticipated, a pragmatic decision was made during the trial to seek to obtain outcome data, including index follow-up interviews, for at least 200 partners in total across study arms.

Randomisation

Index patients were randomised in a 1 : 1 : 1 ratio to one of the three study arms by simple computer-generated unrestricted randomisation within the web tool. The randomisation applied to all contactable partners identified by the index patient.

Blinding

It was not feasible for participants or researchers to be blind to the intervention type during implementation or evaluation. However, all statistical analyses were conducted blind.

Statistical analyses

Analysis of the primary outcome was by intention to treat with all partners not reported as treated assumed not to have been notified (therefore, by definition, there was no loss to follow-up). As a secondary outcome, we reported the proportion of contacts in each arm treated by each partner notification method (routine, APTHotline, APTPharmacy, other). Analysis was based on complete cases. However, as noted, the primary outcome relates to whether or not partners were considered treated and those for whom information was unavailable were analysed as if untreated. We also report the proportions of partners considered treated, considered untreated and with an unknown treatment status (by whether or not the index patient was followed up). Similarly, for the secondary outcome relating to partner notification, for ORs and statistical testing those partners for whom notification was unknown were considered not notified, but we also report the proportions notified, not notified and unknown (index followed up or not).

To calculate adjusted ORs logistic regression was used. Unadjusted and adjusted ORs are reported with 95% CIs and corresponding p-values. ORs were reported for comparisons between the hotline arm and routine partner notification and between the pharmacy arm and routine partner notification arm. Adjusted ORs were calculated adjusting for the age and ethnic group of the index patient. The analysis of outcomes determined for each partner, such as the primary outcome, was based on robust standard errors to acknowledge the clustering of partners by index patient. All analysis was conducted in Stata 13.

Registration

The trial was registered with a UK Clinical Research Network Study Portfolio ID number of 10123. Ethical approval for the trial was obtained from East London REC 1.

Recruitment and follow-up

Of 357 chlamydia-positive women screened for eligibility, 49 were found not to be eligible, 67 declined to participate and 42 were not included for other reasons. In total, 199 women were randomised, who in total identified 313 contactable partners (Figure 5). A total of 199 index patients were randomised.

FIGURE 5.

Consolidated Standards of Reporting Trials (CONSORT) diagram for the trial.

Baseline characteristics

By definition there was no loss to follow-up in the primary outcome. However, some secondary outcomes required further information from an index patient telephone follow-up, which did not occur for roughly one-quarter of index patients (53/199), affecting 28% of partners (87/313). This loss to follow-up was somewhat more common in the pharmacy partner notification arm.

Table 17 shows the baseline characteristics of the index patients randomised and their contactable partners. These were similar across arms. Index patients were relatively young, with a median age of 21 years overall, the mean number of sexual partners over the last 6 months was 1.7 and the mean number of contactable partners was 1.6, indicating that the vast majority of partners were contactable.

Findings

For the primary outcome, Table 18 shows that the proportion of partners known to have been treated varied little between study arms (from 39% to 46%) and the ORs in Table 19 show that, with or without adjustment, there was little evidence of a difference between either APT method and standard patient referral. Specifically, the unadjusted and adjusted ORs (95% CIs) for partner treatment in the hotline arm relative to standard care were 0.91 (0.48 to 1.73) and 0.64 (0.35 to 1.18), respectively, and for partner treatment in the pharmacy arm relative to standard care were 0.90 (0.65 to 1.27) and 1.06 (0.78 to 1.45), respectively.

Table 18 also shows that partners whom the index patient thought were untreated could be divided into two groups: those who were known to be untreated (7%) and those whom we assumed were untreated for the purposes of this analysis because the index had no information about them. When the number of partners known to have been treated was examined per index patient (Table 20), the mean was 0.7.

Table 18 shows that around 70% of partners in each study arm were known to have been notified in each study arm and the ORs in Table 19 confirm that there is no evidence of a difference between either APT arm and standard patient referral.

Table 18 also shows that among partners treated in the hotline arm only 21% were treated via the hotline and of those treated in the pharmacy arm only 30% were treated via a pharmacy.

Secondary outcomes

Process evaluation

Main findings

Similar proportions of partners were reported to have been treated across the three arms of the trial, in each case fewer than half. Although overall outcomes for partner notification were similar between the three arms, only a minority of those in the APTHotline and APTPharmacy intervention arm actually used that modality and their availability did not appear to improve outcomes.

The online patient and data management tool was acceptable to clinical staff and feasible for the referral of patients for partner notification support, including in the intervention arms, and for outcome measurement. However, recruitment to the research study within the clinic setting was challenging. The need to seek individual informed consent from index patients at an often emotional time and often by telephone acted as a barrier to participation for both patients and HCPs, whereas rapid changes in health services created challenges to clinical services in terms of prioritising research. The numbers of sex partners reported by women were comparable to the numbers reported by women attending specialist sexual health services104 and the majority of sex partners were contactable. However, fewer than half of the sexual partners were confirmed to have been treated across the trial and the addition of the APT interventions did not appear to improve outcomes. Uptake of the interventions was poor and most of the successful partner notification was achieved using standard methods rather than by APT.

No sex partners were known to have attended for HIV and syphilis testing although a small number of sex partners who took up the interventions returned a postal test kit for chlamydia and gonorrhoea. Despite good levels of index patient follow-up, including reminders about retesting, return of postal test kits for reinfection testing at 4–6 weeks post treatment was low.

Findings in the context of published literature

Effective partner notification in primary care is recognised as difficult to achieve97,99 but the importance of improving the effectiveness of partner notification as a strategy for improving population control of C. trachomatis is well recognised. It is increasingly recognised that partner notification outcomes should be measured per partner, rather than per index patient,105 as this reflects the true potential for the prevention of further transmission. However, current guidelines for partner notification, and newly available surveillance reports in England, continue to focus on outcomes per index patient. 106 Our choice to use partner-focused notification outcomes makes it hard to compare our data with audit data, which are typically reported as number of partners treated per index. Nevertheless, the suboptimal rates of partner notification described in this pilot trial are superior to those reported in similar settings. 68,104,105

Beyond the availability of partner-focused outcome data, there are further conceptual and empirical challenges with regard to the comparison of our data with those from other settings. Primary care and other non-specialist settings have not historically had a routinely available partner notification infrastructure and partner notification outcomes have typically not been measured or recorded at all. The provision of a partner notification service is of itself likely to produce a substantial ‘Hawthorne effect’ by normalising the discussion of partner notification. This may on its own be sufficient to motivate and empower an index patient to attempt discussion of the need for testing and treatment with his/her partner(s). Ideally, we would compare our data, at least for general practice, with data from a completely undeveloped service, but this is not in fact possible. We can nevertheless compare our data with limited data from community services, including the NCSP.

Our previous study of APT in specialist sexual health/GUM settings did use partner-focused outcome measures and suggested that APT could provide an improvement in partner notification outcomes compared with standard patient referral. 101 However, there were important differences in the design, inclusion criteria and exact nature of the interventions, which make direct comparisons inappropriate. Our initial APT study allowed index patients to select the method of partner notification that they thought would best suit each sex partner, whereas this study randomised patients. In routine clinical practice, index patients choose from a range of partner notification options and we believe that removing the element of choice by randomising patients to one intervention may not be the optimal design for sensitive interventions such as partner notification.

Limitations

Implementation of a pilot trial of this size, which required a large number of recruitment sites, at a time of unanticipated and wide-ranging health service change was extremely difficult. We had to scale down our original sample size to provide more realistic recruitment requirements for each service and support recruitment centrally. Our primary outcome was largely index reported rather than sex partner verified, which introduces uncertainty around the robustness of outcome ascertainment. Although we included a biological outcome (index patient positivity 4–6 weeks post treatment), in practice it did not prove useful as postal retesting had a low uptake rate despite telephone reminders.

We do not know whether the partners of women who declined to take part, or who did not take part for other reasons, were more or less likely to be treated. Our sample could therefore be biased in either direction with respect to the target population of index patients and partners.

Although APT did not appear to reduce the time to partner treatment, some centres recruited sex partners who attended with the index patient (contrary to protocol), which would bias the measures of routine partner notification to make it appear much faster.

Implications for public health and clinical practice

The care pathways that we developed, all of which used a novel online patient and data management tool, provide a feasible and acceptable infrastructure for the onward referral of patients diagnosed with STIs in general practice and other community settings to receive support with partner notification.

Outside the research setting, where there are major barriers to recruitment and randomisation at an emotionally taxing time, our model interventions appear to be capable of providing acceptable approaches to partner notification support. These could usefully form part of the portfolio of partner notification choices aimed at improving outcomes for all kinds of partner. It is increasingly recognised that novel interventions should not be seen as standalone strategies, but rather in the context of a portfolio of partner notification modalities that collectively improve overall partner notification outcomes.

The low uptake of follow-up STI or HIV testing is notable and suggests that these modes of partner notification which do not require direct engagement with a clinical service that can provide comprehensive testing may be unsuitable for higher-risk populations. Although in the future this may be addressed through postal self-sampling as a means of delivering more comprehensive testing, they are not suitable either for individuals or for populations who have been identified as being of higher risk, such as men who have sex with men (MSM) and MSM partners of heterosexual women. 107,108

Implications for research

Overall partner notification outcomes were superior to previously reported partner notification measures in similar settings and so further work is required to optimise the uptake of APT outside specialist services. An in-depth understanding of the reasons for poor uptake of the interventions, that is, whether it was the index patient failing to forward the text or partners not finding APT acceptable, will be important for future partner notification research, irrespective of setting. A cluster design in which whole services are randomised to the offer of an APT intervention is likely to facilitate uptake, improve staff familiarity, normalise the interventions and minimise the negative impact of health service changes on ongoing clinical staff-dependent research. However, this may be difficult to implement as large numbers of clinics may be required to accommodate the cluster design and this may prove excessively costly. In addition, given the difficulties in achieving individual consent, a non-consented trial could be considered, which, although controversial, has been undertaken in the field of chlamydia screening. 99,109

Research should also be undertaken to explore how to identify and reach higher-risk partners who may benefit from a more comprehensive range of sexual health services.

Introduction

The aim of the economic analysis was to compare the costs and outcomes for the three arms of the pilot RCT in terms of a preliminary cost–consequence analysis. As described in the first part of this chapter, the pilot trial compared two APT interventions (APTHotline and APTPharmacy) with routine partner notification (standard treatment) in a primary care setting. The success of any new trial or intervention must be balanced against the resources required to achieve the desired outcome; thus, any additional resources required must be evaluated in terms of any additional benefit that can be attributed to them.

Methods

The pilot RCT has been reported in the first part of this chapter. In brief, women diagnosed with chlamydia with at least one contactable male partner were randomised via a web tool to receive (1) APTHotline, which involved the telephone assessment of sex partners by a clinic-based qualified health advisor; (2) APTPharmacy, which involved partner assessment by a trained community pharmacist; or (3) standard partner notification, defined as the routine method of partner notification offered by the service. Those randomised to receive the APT treatment options were offered APT in addition to standard partner notification. The primary outcome for the pilot RCT was the proportion of contactable partners considered treated within 6 weeks of the diagnosis of the index patient.

For the economic analysis the costs and outcomes associated with the two APT treatment strategies were compared with the costs and outcomes associated with routine partner notification (standard treatment). This involved the collection of data on resource use and costs for each strategy, alongside their outcomes. The economic analysis was conducted alongside the pilot trial and the perspective adopted was that of the NHS and so only direct health service costs were included.

Analysis

We compared costs and outcomes for all three trial arms separately in a cost–consequence analysis. 110 This form of analysis was deemed appropriate as the analysis was based on a pilot RCT. A cost–consequence analysis involves the assessment of costs and outcomes in a disaggregated manner to see if there is any strategy that shows clear dominance. 112 Dominance is said to occur when a strategy costs less but is more effective than another strategy in terms of the outcome achieved. A strategy is seen to be dominated when it costs more and is less effective than an alternative strategy. The methods of analysis followed those adopted in a previous study of APT in a specialist care (GUM clinic) setting,96 with some adjustments to reflect differences in trial design. The primary cost–consequence analysis was based on the outcome of whether the index patient’s partner(s) was considered to have been treated within 6 weeks of index diagnosis (intention to treat). In addition, the proportion of partners treated by each APT treatment option (APTHotline, APTPharmacy) within each study arm is presented.

A series of one-way deterministic sensitivity analyses were conducted. A probabilistic sensitivity analysis was not appropriate as this was a pilot trial and the results are illustrative and preliminary. We analysed uncertainties around all key cost and outcome parameters, specifying plausible ranges using information from the trial and from the literature. These analyses included (a) varying the costs associated with the web tool from £1 per player randomised to £20, to explore the potential impact of different take-up rates and web tool costs during roll-out; (b) including costs associated with recording information for index patients on the web tool, using information recorded in the trial; (c) increasing costs associated with the APTHotline consultations for a scenario in which hotline staff were less able to carry out other tasks between calls to the APTHotline service; (d) adjusting the cost of standard treatment; (e) including training costs for staff involved in inputting information on the web tool, using data recorded in the trial; and (f) varying the take-up of the APT strategies and standard partner notification. This included the application of estimates of take-up from a previous study in a specialist setting. 96 Further sensitivity analyses were carried out but are not reported.

In the reporting that follows, all cost data are presented in UK pounds in 2012/13 prices.

Results

The results of the trial have been reported in the first part of this chapter and are summarised in Table 18. Table 22 shows that there was little difference in the proportion of partners considered treated within 6 weeks of index diagnosis between the standard treatment arm (46/102, 45%) and the APTHotline (39/111, 35%) and APTPharmacy (46/100, 46%) arms. The table also shows that, among partners considered treated in the APTHotline arm, only 7% (8/111) received treatment via the hotline service and in the APTPharmacy arm only 14% (14/100) received treatment via a pharmacy. Other partners received treatment through standard partner notification routes such as GUM clinics or GP services.

The web tool and APT packs were common to both the APTPharmacy and APTHotline arms. The total cost of the APT pack was estimated to be £8.41 (Table 23). Full breakdowns of the health service costs associated with the APT treatment strategies are presented in Tables 24 and 25. Taking into account the likely uptake, the ongoing cost associated with the web tool was estimated to be £200–222 for the APT treatment arms. The cost associated with the text messaging service was £5.44 for the APTHotline arm and £5.20 for APTPharmacy (see Tables 24 and 25). The average cost per partner treated for the APTHotline arm was estimated to be £35.19 and that for the APTPharmacy arm was £32.47.

The breakdown of costs for standard partner treatment is shown in Table 26. The average cost per partner treated for the standard treatment arm was £30.10. The standard treatment arm had the lowest average cost per partner treated of the three trial arms.

Table 27 presents a summary of the total costs and outcomes for the APT treatment arms compared with the standard treatment arm. The number of partners considered to be treated is shown as well as information about the number of partners treated by the two APT treatment options.

The cost–consequence analysis examined the costs of all three arms and compared their outcomes in terms of the number of partners considered treated. The results in Table 27 show that there was little difference in the proportion of partners considered treated between the standard treatment arm and the APT treatment arms. The APTHotline and APTPharmacy arms had a higher average cost per partner treated than the standard treatment arm, primarily because of the additional costs associated with the web tool and text messaging, which were not required for partners receiving standard treatment. The total cost associated with the APTPharmacy arm was higher than the total cost associated with the APTHotline arm because of the higher costs associated with the pharmacist carrying out partner consultations. However, the cost per partner treated in the APTPharmacy arm was lower than that for the APTHotline arm because of the higher number of patients receiving treatment (the denominator). Both the APTHotline arm and the APTPharmacy arm had a low proportion of partners treated by APT treatment services (7% and 14%, respectively), with the majority of partners receiving treatment from other services.

These preliminary results do not suggest that the additional costs associated with the APT strategies can be balanced against increased effectiveness in achieving partner notification, as the proportion of partners considered treated in the standard treatment arm was similar to the proportions considered treated in the APT treatment arms and the cost was lower. It is also evident that a high proportion of partners considered treated in the APTHotline and APTPharmacy arms received treatment using standard partner notification methods (e.g. general practice or specialist sexual health services).

Discussion

The results of this preliminary cost–consequence analysis suggest that, although APTHotline and APTPharmacy have been demonstrated to be cost-effective alternative strategies to routine partner notification in specialist settings,96 further evidence is needed about their potential within a primary care setting. The proportion of partners treated was lower than was demonstrated by a previous trial of APT in specialist services, which meant that the higher fixed costs associated with the APT treatment strategies were not offset by a higher proportion of partners treated (as was found in the trial in the specialist setting). The reasons why partners did not elect to access APT treatment strategies in the APT treatment arms requires further investigation. However, overall partner notification outcomes were superior to those reported in previous studies in similar settings, suggesting that further work is needed to explore strategies to optimise partner notification outside specialist settings.

This is the first study to collect and analyse costs and outcomes for APT in a primary care setting in the UK, as previous studies have been concerned with APT within a specialist care setting. Its main strength is that costs and resource use were collected prospectively within a pilot RCT. There are several limitations associated with the study. The ability to follow up index patients varied by study arm, which had an impact on the denominators used to calculate the proportion of partners treated. In addition, the primary outcome relied on index-reported results, rather than being sex partner verified, which potentially reduces the robustness of outcome measurement. To ensure consistency between study arms a number of assumptions were made within the economic analysis. Although these assumptions were tested within the sensitivity analysis, further exploration of such factors would be required before the interventions were rolled out.

The poorer uptake of APT treatment strategies in the primary care setting compared with previous findings in a GUM setting may reflect the fact that for the current study index patients were randomised to a treatment arm and were not able to choose a partner notification approach. Further analysis is planned that will examine the likely impact of allowing index patients to choose a treatment strategy for their partners in a primary care setting. This will use data previously collected on partner type and treatment strategy selection to assess the likely impact of such factors on cost-effectiveness.

Aims

• To explore the medical, sporting and social venues that young men find acceptable and feasible for accessing STI screening and to determine the models of screening that young men consider acceptable and feasible in those contexts (phase 1).

• To undertake a pilot RCT of two screening interventions in sports settings (phase 2).

Specific objectives

• To determine young men’s usage of medical, sporting and leisure venues.

• To determine which venues young men would find acceptable and feasible for accessing STI screening.

• To develop through qualitative research and consumer and stakeholder consultation two feasible and replicable interventions for delivering STI screening in football club venues.

• To determine the acceptability to young men and feasibility of football trainer-led STI and HIV screening.

• To undertake a pilot RCT of football captain-led STI screening in two contrasting football clubs in different geographical areas.

• To determine the uptake of STI screening by young men in football club settings.

• To obtain cost data for the football captain-led STI screening strategies to use in a preliminary economic evaluation.

Objectives

• To determine young men’s usage of medical, sporting and leisure venues.

• To determine which venues young men would find acceptable and feasible for accessing STI screening.

Here, we report findings from a stratified random probability survey that explored the medical, social and sporting venues in which men aged 18–35 years and resident in the UK find it acceptable to access self-collected testing kits for STIs and HIV infection. We also aimed to determine whether or not those men who play football would find their football venues acceptable as pick-up points for self-testing kits in an attempt to further understand the acceptability and feasibility of using football settings to engage men in STI testing.

Methods

Our study formed part of the National Centre for Social Research (NatCen) quarterly social research survey (Omnibus). 144 This is a stratified random probability survey of adults aged > 16 years in the UK. A multistage sampling design technique is used for the Omnibus survey. First, postcode sectors are ordered according to government office regions and the national statistics socioeconomic classification before selecting 153 sectors. Twenty addresses are then selected from the postcode address file for each of the 153 postcode sectors. This gives a total sample size of 3060 addresses. Finally, a single participant aged > 16 years is selected at random from these addresses (although questions relating to our study were delivered only to men aged between 18 and 35 years). Appropriate selection and calibration weights are applied to correct for the unequal probability of selection in households of different numbers of occupants and to ensure that the weighted distributions match population totals. The Omnibus survey is conducted in accordance with the Social Research Association ethical guidelines. 145

We developed 10 questions exploring the use of general and sexual health care; key sexual risk behaviours; participation in sporting activities; and the acceptability of self-collected STI and HIV testing in a variety of medical, social and recreational settings. When applicable we used questions validated for the National Surveys of Sexual Attitudes and Lifestyles (Natsal). 146 Surveys were delivered using a combination of computer-assisted personal interviewing and self-interviewing techniques. Questions were piloted with researchers at NatCen prior to inclusion in the survey and the survey programme was tested for correct routing; internal range and consistency error warnings were also created.

Data were collected in three waves between January and October 2010. Selected participants were sent an invitation letter prior to the interview together with an unconditional £5 voucher. Interviewers called at each address on at least six, and a maximum of nine, separate occasions at different times of the day and week, including evenings and weekends, before an address was recorded as a non-response. The first three calls were conducted after 6 p.m. on Monday to Thursday or at a weekend. In an attempt to increase participation and the accuracy of the data recorded, more sensitive questions were completed by participants without the interviewers seeing their responses and data were immediately and confidentially stored on the laptop computers. Interviews lasted between 25 and 30 minutes.

Coding of the data was performed by researchers at NatCen. Free-text responses to questions were back-coded when possible into existing codes for that question. New code frames were created for open questions from responses given in initial interviews.

Results

The three waves of the survey had an overall response rate of 53%. The median age of men was 28 years, with 130 men (38.9%) aged 18–24 years, 124 men (28.2%) aged 25–29 years and 157 men (32.9%) aged 30–35 years.

Discussion

Overall, men appeared well engaged with health care, with almost all being registered with a GP and three-quarters having seen their GP in the last year. Awareness of sexual health appeared high, as almost one-third of men had been screened for STIs and one-fifth had been tested for HIV. The most acceptable venues for young men to pick up self-collected STI and HIV test kits were health-care settings (general practice, sexual health clinics, pharmacies), whereas sports, social and recreational venues were acceptable to a smaller proportion of men. Football (soccer) was the most popular sport played and around half of men who played football would find their football venue an acceptable place to access STI and HIV testing kits.

To our knowledge this is the first stratified random probability survey of young men in the UK to determine the acceptability of various settings for accessing self-collected STI and HIV test kits. The findings directly informed the settings and intervention for the subsequent pilot cluster RCT. The study provides generalisable data that should be beneficial to those involved in researching, developing and delivering STI services for men in health-care and non-health-care settings within the UK, particularly in the context of low uptake of testing in men reported by the NCSP. However, we do not know the reasons for refusal to take part in the survey or how those who declined differed from participants. Although the questions that we developed did not undergo formal psychometric testing, when possible we used questions that had been validated for use in the highly regarded Natsal surveys. 146

Although many studies have shown that using non-traditional and sports settings to screen for STIs is feasible,134,147 few have focused on the acceptability of different settings for men. Lorimer et al. ’s148 study of willingness to participate in a non-medical approach to chlamydia screening found that men in particular valued the possibility of screening in these settings. Furthermore, the uptake of screening varied by setting,136 supporting our finding that the acceptability of sports settings was greater in those who had actually engaged in sporting activities over the last month. Anonymity appeared to be a key factor in determining the acceptability of screening in a qualitative study of young men’s experiences and perceptions of chlamydia screening commissioned by the NCSP. 149 Men in that study rejected many of the proposed sports and social venues for fear of the stigma of being seen to take a test. Men also appeared to perceive a degree of incongruity between attending these locations for recreation and the health message of screening. 149

Our finding that most men had seen their GP in the last year is in keeping with the findings from other studies. 150 Taken together with the high acceptability of general practices for accessing STI and HIV testing kits, this highlights the importance of, and potential for, chlamydia and HIV screening in general practice, especially among younger men. 119 However, current rates of STI and HIV screening in general practice are low, which could suggest a reluctance on the part of the health-care provider to offer testing. 140

A substantial proportion of men in our study had previously tested for STIs (28.7%) and HIV (19.8%), mostly within specialist sexual health settings. Among those aged 18–24 years, over one-quarter reported that they had previously been tested for STIs; however, NCSP data suggest a coverage of around 12% for those aged 16–24 years in England and Wales. 151 This probably reflects testing in GUM services among participants in our survey. Our data reveal a testing rate for HIV that is higher than equivalent data from the most recent Natsal, which found that 6.63% (95% CI 5.14% to 8.52%) of men aged 16–24 years and 14.8% (95% CI 13.0% to 16.7%) of men aged 25–34 years reported having had a blood test for HIV. 139 Both our data and the second National Survey of Sexual Attitudes and Lifestyles (Natsal-2) data139 excluded blood donation as a reason for a HIV test, although the Natsal data are now more than a decade old.

It is important that venues in which STI screening and testing kits are offered are acceptable to target populations. Although it is now possible to deliver testing in non-clinical settings, this research highlights that, among men aged 18–35 years, in Great Britain at least, it is the traditional health-care settings that are most acceptable as pick-up points for self-testing kits. Young men frequently access primary care and we feel that there is considerable potential to engage more men in STI and HIV testing through general practice. 138 NCSP data show that only 25% of tests in men are carried out in ‘core’ health settings. 151 In contrast, our research shows that these are the preferred access points for as many as 80% of men. Based on our findings, there is clearly a mismatch between where services are currently provided and the settings in which men prefer to access STI and HIV screening. However, non-traditional settings are acceptable to a minority of men and may be important in reaching men who would otherwise not seek STI screening and for men who perceive their local sexual health services to be inaccessible, The importance of appropriate analysis of the cost-effectiveness and public health impact of the later pilot cluster RCT cannot be underestimated, to enable resources to be used most appropriately. This is especially important when considering the potentially low rates of chlamydia detected among the men screened in non-health-care settings. 119

Phase 2a: Qualitative survey of men’s preferences for sexually transmitted infection screening

Phase 2b: the SPORTSMART study – a pilot randomised controlled trial of sexually transmitted infection screening interventions targeting men in football club settings with a preliminary cost–consequence analysis and behavioural survey

Methods

Trial design

We used a cluster RCT design. We allocated two clubs to each of our three trial arms: team captain led and poster STI screening promotion (arm one), sexual health advisor led and poster STI screening promotion (arm two) or poster-only STI screening promotion (control/comparator arm three).

Outcomes

The primary outcome was the proportion of eligible men accepting the offer of screening (intervention uptake). Secondary outcomes were the proportion of screened men who tested positive for chlamydia and/or gonorrhoea and health service costs (reported later and in Jackson et al. 114).

Statistical analysis

We reported the primary outcome with a 95% CI based on a robust standard error that acknowledges the clustering of participants by club.

Club recruitment

We identified all potentially eligible amateur clubs in appropriate geographical areas from the AFC listings available on the internet. 153

Randomisation

Prior to randomisation we divided clubs into three pairs. Pairing was based on a description of the club memberships’ ethnicity, age, education status and size, as described by early qualitative work132 and club representatives’ reports, and was performed to achieve approximate balance across pairs in these characteristics. The pairs of clubs were then allocated to one of the three study arms by the lead study statistician by random permutation. Clubs (and thus participants) were unblinded directly following study arm allocation. It was not feasible that clubs or investigators be blind to the intervention type during implementation or evaluation.

Team captain and health adviser recruitment

During the recruitment phase the trial co-ordinator explained to the club contacts that two of the participating clubs would be randomly allocated to a captain-promoted screening intervention and so all participating clubs needed to have at least one captain willing to promote the screening intervention among two teams in each club.

Health advisor selection

Based on our preclinical qualitative work,132 we recruited a male health advisor to deliver our STI screening promotion. The health advisor was also in the same age group as the football players involved in the intervention and so the distinguishing difference between the self-selected team captain and the health advisor was that the health advisor was a medical professional from outside the club.

Delivery of the interventions

The trial co-ordinator e-mailed the club contacts prior to the football matches with brief details of the screening event. On the day, the trial co-ordinator put up posters in all participating clubs and set up the test kit collection boxes in the club changing rooms just prior to the players’ arrival. The interventions were delivered according to randomisation during the usual pre- or post-match team briefing. Interventions were as follows:

1. Captain and poster screening promotion. The team captain delivered a standardised brief screening promotion talk of < 5 minutes’ duration (Table 36) and then handed each player a test kit (Figure 6) and answered any questions from participants.

2. Health adviser and poster screening promotion. A sexual health adviser from the study clinic delivered the standardised brief screening promotion talk of < 5 minutes’ duration and then handed each player a test kit and answered any questions from participants.

3. Poster-only screening promotion (comparator arm). Posters were displayed that the men were free to read, with kits readily available, but no verbal information was given.

TABLE 36 - Screening promotion content

Topic	Message
Health promotion message	Chlamydia and gonorrhoea are common and often present without any sign that anything is wrong (asymptomatic)
Screening information	Confidential; right to refuse; how contacted with results
Test kit instructions	Step-by-step instructions
Further research participation	Invitation to follow-up qualitative interview

FIGURE 6.

SPORTSMART self-sampling kit.

Men who wished to participate completed a sample kit according to the instructions provided and placed the completed kit back into the secure collection box. Alternatively, men could take the kit away with them for later completion and post it back to the clinic in a discreet postage-paid package. All clinical follow-up, including provision of test results via text message, was carried out by clinic staff according to routine standards of care.

Process evaluation and additional data collection

Captains and the sexual health advisor completed a ‘report-back’ form directly after each intervention. Information gathered included the number of men in the changing room and exposed to the intervention and their views of implementing the intervention. In addition, the trial co-ordinator took field notes to describe the circumstances of each intervention (including weather, match outcome, timing of intervention) to assess the fidelity of the interventions in practice.

Resource use data were collected prospectively for use in the health economics analyses, which we report on pp. 90–92.

Participants (players) were invited to take part in a telephone semistructured interview to explore their views of the interventions within a month of participating in the initial screening event.

At the time of the SPORTSMART survey distribution (see below), captains of the participating football teams were prompted by the trial co-ordinator to ask players who took part in the intervention if they would like to participate in a one-to-one telephone interview with a researcher to explore their experiences of receiving STI screening promotion as part of the trial. A list of players and captains who agreed to participate was provided to the researcher, who then telephoned, and explained the purpose of the qualitative interviews, gained informed consent and organised a date and time for the telephone interview.

Data collection

Individual semistructured telephone interviews were conducted with the players who had consented. The interviews varied slightly according to whether the participant was a player or a captain of a football team; however, all interviews explored the players’ basic demographic characteristics, their views and experiences of the process of the intervention they had received (or gave if the participant was a captain) at their club, their thoughts of having STI testing sample collection kits available at the football club, their views of the contents and use of the kits, whether or not they had previously tested for a STI, their experiences of testing for an STI elsewhere and their preferences for future STI testing. All interviews were audio recorded, lasted approximately 30 minutes and were conducted 4–6 weeks after the SPORTSMART intervention by a female researcher. Participants were given a £30 online store voucher for their participation.

Data analysis

Interviews were transcribed verbatim and analysed using the framework approach. 154 Transcripts were read and reread by four researchers and coded into broad themes based on the research objective and interview content to create an initial coding framework. This framework was further discussed and modified within the research group. Two members of the research group then systematically applied these codes to the transcripts. Reliability was enhanced by double coding and comparing a subset of transcripts. Few discrepancies emerged and, when they did, consensus was negotiated.

Two weeks after the intervention was completed, all (playing and non-playing) club members aged at least 18 years were invited to take part in a brief, self-administered, anonymous pen-and-paper survey questionnaire (the SPORTSMART survey) to assess club members’ sexual risk behaviour and previous STI testing history to inform estimates of the public health impact of offering screening in these settings. We report these findings on pp. 98–101.

Sample size

We aimed to recruit 200 men to estimate the overall acceptance of screening rate with acceptable precision, considering a wide range of possible rates because of the lack of directly relevant evidence from previous studies. Specifically, this sample size would allow us to estimate the rate within 7% if the rate was 50% (i.e. a 95% CI of 43% to 57%) or within 5% if it was either higher or lower (85% or 15%), assuming minimal variability between clubs.

Consent

Club managers gave consent for clubs’ involvement in the study. Signed informed consent was obtained from captains before the intervention. Football team members opted in to screening by completing the kit offered, but could opt out of the intervention at any time.

Ethical approval for this study, (including the SPORTSMART survey) was given by the National Research Ethics Service (reference number 13/SC/0029).

Results

Recruitment

Recruitment of football clubs was conducted between October and December 2012. Clubs were contacted by the trial co-ordinator by e-mail and/or telephone to assess interest and eligibility. Five of the 18 clubs initially identified had invalid contact details. Of the remaining 13 clubs, five did not respond and eight (62%) indicated that they were willing to participate. Six were chosen based on the willingness of a club representative to meet with the study co-ordinator and fully discuss the study objectives; the remaining two clubs were placed on a reserve list (Figure 7).

FIGURE 7.

Participant flow in the SPORTSMART trial. POL, popular opinion leader.

Acceptability of screening and sexually transmitted infection positivity

The interventions were implemented between February and April 2013. Across the three arms 153 men in six clubs participated in the trial and 90 (59%, 95% CI 35% to 79%) accepted the offer of screening (Table 37). Acceptance rates varied considerably by club (see Table 37), but the aggregate rates were broadly comparable across the arms: captain led: 28 out of 56 (50%); health professional led: 31 out of 46 (67%); control: 31 out of 51 (61%). The variability within arms was greater than the variability between arms. The majority of test kits were completed within the clubs and only one was returned by mail. There were no positive tests for chlamydia or gonorrhoea from any of the study arms.

TABLE 37 - Screening kit uptake among participating clubs

Study arm	Club	Players in changing room	Completed kits returned	% return
Health professional led	1a	24	10	42
Health professional led	1b	22	21	95
Captain led	2a	26	10	38
Captain led	2b	30	18	60
Control	3a	24	20	83
Control	3b	27	11	41
Total	All	153	90	59

Process evaluation

The AFC club listings were a useful initial resource to identify clubs, but contact details for individual clubs were often incorrect and some club websites contained out-of-date contact information. We do not know whether the reason why five clubs did not respond was because of a lack of interest in the study or our failure to establish an appropriate means of communication.

A number of cancelled and rescheduled matches meant that we were unable to deliver the interventions to as many players as planned. Evaluation of field notes and report-back forms from captains and the sexual health adviser confirmed that the interventions were delivered in a standardised way across all study arms and captains felt comfortable delivering the short intervention. However, the poster comparator arm was unintentionally ‘enhanced’ by some captains, who actively publicised the availability of STI screening at the club prior to the day by including details of the research in their weekly team information e-mail and encouraging players to participate. There were no adverse effects from this research.

Men’s views on the acceptability of the SPORTSMART intervention

Thirteen men (10 players and three captains) agreed to participate in the follow-up interviews. Two captains and six players from two different football clubs received the poster intervention, four players from two different teams received the poster and HCP intervention and one captain from one team gave the poster and the captain-led intervention.

Men were aged between 21 and 31 years and all had previously tested for a STI. Seven had tested at a specialist GUM clinic, three at university, four at their GP surgery and one at school. Eight men (one married) described themselves as in a monogamous relationship of ≥ 12 months; three men described themselves as single with no new sexual partners in the past 3 months; one man reported ‘four to five’ new sexual partners; and one man reported one new sexual partner during the last 3 months.

The following themes emerged from the analysis.

Delivery and content of the intervention

Some men who had experienced the poster and HCP-led intervention felt that the HCP brought some legitimacy to STI testing at football clubs, as they perceived the HCP to be more knowledgeable and better able to give advice and reassurance before and after testing for a STI:

if someone just said, ‘oh there’s a few things there or there’s a few checks there’, you feel a bit, mmm, you know, you’re a bit wary. But when someone’s there to explain to you the ins and outs, know what I mean, it makes it a bit comfortable.

When the health professional delivered the screening intervention message the men seemed to better appreciate the importance of testing for a STI:

I thought it was a good in terms of importance, I remember he told us like statistically someone in here will have it and some people thought that was a bit strong. It’s true I guess.

He obviously sold it as the idea that these things can be very serious and definitely worth getting checked for. That’s pretty much it. It was sort of quite a strong message. It’s the sort of thing that you should be doing and it’s definitely worthwhile getting checked.

They also thought the length of the screening intervention message was about right and liked the tone in which it was delivered:

What I did like was the fact that – I can’t remember the gentleman’s name – he was very matter of fact, just came in, ‘right this is what you need to do’. It wasn’t a lecture or anything like that. It was very concise.

Poster-promoted screening

All of the men from the clubs that had been assigned the poster-only intervention had been made aware that there would be STI testing kits available at the football club, despite guidance from the trial co-ordinator that testing kits should just be made accessible to the men should they want to test after reading the poster, rather than being ‘promoted’ in any way. Some men said that they had been told the week before by e-mail as part of the regular team newsletter whereas others said that they had been encouraged to test by the team manager or captain; thus, without consciously doing so these managers and captains were acting like our captain-led intervention based on the popular opinion leader theory. However, the players seemed happy with this approach:

I quickly, briefly read over the poster and then just explained it to the team, and just sort of said, ‘OK guys, it’s obvious there’s a couple of posters in the changing room’. There was a drop-in bin, and a box full of the kits. Everyone was kind of wondering what it was all for, although I had warned them before. I just explained, you know, ‘All you need to do, piss in a pot, fill out the form, and it will be sent off and you’ll get your results back’.

Captain

He told us on the day [the manager] that it was on, then he came up with a couple of forms with more detail more information about it and I then had to pass it on to the team. He just told me exactly what was going on, ‘the bins are in there, get everyone who wants to to take part’. From my point of view, I thought it was kind of . . . well the important thing is I think the club wanted to do it so it was important thing for us. So I kind of took it on to introduce it to everyone. I think that probably made sure that everyone was involved because it was almost like no one wanted to be left out.

Captain

Yeah we walked in and then the manager gave a small speech on how we need to do some testing and read some leaflets. And we did, we all went in the toilet separately and we did a test.

the e-mail said boys, this weekend we gonna have STD [sexually transmitted disease] tests in the changing room, part of this survey and it’s a pilot scheme for testing lots of young men in one place at a time. And that the football team will be compensated for taking part so please boys, if you can, if you don’t mind, please do the test.

Poster and captain-promoted screening

We were unable to obtain the opinions of players who received this intervention, but we did interview one captain who gave the message to the players. He was very enthusiastic about STI testing being made available at the football clubs as he had previously tested for a STI at a GUM clinic and found it embarrassing and time-consuming:

Well I thought it was a good idea to be honest. Personally for me, and a couple of other boys were sort of saying as well, going down to the clinic is, well once you’re in the clinic it could take up to 2, 3 hours depending how busy the clinic is. And also sort of, it’s quite embarrassing to be seen there. It’s not a good thing to be seen in a clinic, so sometimes, you know, I’ve needed to go but I’d probably avoid going in case I bumped into somebody I knew. But having it at the football if a couple of boys that did the test at football then the other boys seem to and it sort of, it’s not really embarrassing really at football because you’re with the boys.

After going through the message to convey with the researcher, the captain said that he felt comfortable giving the message and subsequently felt that it was well received. However, he thought that a few of the less mature men who usually hang round in a gang felt it embarrassing, whereas the older men found it easy. This seemed to be because the captain himself felt more comfortable giving the message to men whom he had perceived as having the same values as himself:

well I got there a bit early, and [the researcher] made me aware of what was happening and went through it with me so when the boys came in I just sort of spoke to them about what was happening, and then when they were all sitting down and ready I just sort of repeated myself again and just said to complete one of the bottles and I think the majority did, we only got a few where some of them were a bit sort of were against it. I don’t know why. I think a few of them were saying it was intrusive, but I don’t think it was the way it was presented. I think it was just what the test was for and stuff.

When prompted by the interviewer to explain why he thought some men found it intrusive, the captain gave this explanation:

Well to me it was the boys that were, the younger boys that were, it was like not involved in gangs but sort of . . . the sort of I don’t know how to explain it. I wouldn’t say they were involved in gangs but the young, street boys. But I think the older players found it easier. I think the younger lads found it a bit more embarrassing.

Captain

It is important to note that, even though the captain refers to some participants as ‘boys’, all men were aged ≥ 18 years.

Reasons for testing

Several of the men mentioned the inconvenience of testing at other services, including making appointments and the time required to attend those appointments. They explained that they had done the test in the trial because it was simple and easy to do, with easy to understand instructions. One of the men who had identified himself at risk said that he would not go to his GP or GUM clinic as he found it embarrassing and had done it once before and found it ‘horrible’, whereas testing at the club he found easy and felt comfortable ‘around his mates’:

This is quite easy ‘cos you’re at football, everyone’s doing it and you just do it. It’s a lot less worrying. Whereas if you were to book an appointment to go and do it, you’d probably, I don’t know, I’d personally get a bit nervous.

Interestingly, few men tested because they thought that they were at risk of having a STI, saying that they just did it because everyone else was doing it. The following dialogue is with a player who reported four to five sexual partners in the last 3 months:

Well I suppose I had no reason for doing it, but everyone else was doing it, there was that, and certainly it was that because I had a test not so long ago, so it wasn’t that I was looking to do it relatively soon, but I think it was a sensible thing to do. It was a case of it was there, the opportunity was there, everybody else was doing it, so that was it really the thinking behind it.

So do you think you would have been at risk of chlamydia since your last test in January until now?

No. Well I can’t tell you 100% but I don’t think so no.

OK, so have you had any new sexual partners since January?

Mm yes.

And that was protected or unprotected sex?

That was unprotected.

OK, I was just wondering why you thought you weren’t at risk?

Actually when you put it like that . . . actually, I suppose you always assume it’s not going to happen to you don’t you, when you put it like that I suppose I am at risk.

Feelings about testing

When men were asked how they felt about STI testing at the football club, the majority of them said that they were very comfortable testing with their football colleagues in an all-male sports environment. They did not feel embarrassed to be testing and despite doing a test with all of the other team members they perceived the whole process to be more private and confidential than going to a GUM clinic or their GP:

No, I think maybe if it had been just 11 guys at work and stuff, it would be a bit weird but we’re all quite good friends and football teams are quite confident with each other anyway, like a shower, we all to do on a Saturday, so like putting a pee in a pot isn’t too much of an issue really.

I would say no one was embarrassed. I guess probably the nature of the culture, so you know, it’s a football club and we’re all close mates. We chat about this sort of stuff all the time. Well, not about this sort of stuff, but we chat about sex and women all the time. We’re fairly open. We shower together. So actually it’s probably as good an environment as you’re gonna get, if that makes sense?

I think it’s a good thing in that it’s, well it seems like it’s a, you know, it’s a targeted, it’s a group that, sort of men between the ages of 18 and 30 so this, I found it accessible as well. And at first I had my sort of, I wasn’t sure how confidential it would be but then when I saw that there was just a drop box, sort of no one, yeah, I thought it was good because it was confidential.

Preferences for future testing

Overall, the men interviewed would prefer to be tested at a football club in future, because of the ‘all lads together’ relaxed environment. They also preferred the ease and simplicity of the tests and the convenience – not having to travel to a doctor or a clinic and the fact that the tests were available in the changing rooms next to the toilets so they could use them when they were changing:

I’d probably, being a boy, I’d prefer it how I have just done it with the football team. I don’t know, I suppose it’s more confidential if you know what I am saying.

In what way?

Don’t know, I suppose if you go to your doctor you’re doing it face-face sort of thing. I suppose if you’re doing it like how your team done it, then I suppose it makes people feel a bit more relaxed and stuff. It certainly made me a bit more relaxed than going to the doctor’s or something.

I thought it was very good actually, because I think a lot of people, probably myself included, should get tested a lot more often than we do. But actually we all lead very busy lives and we’re probably a little bit lazy. But having it there, it’s very easy to do, and it was a good thing. It was well received.

Methods

Results

The outcomes by trial arm are reported in Table 37 and are summarised below. It is evident that, although there was considerable variation in the uptake rates between individual clubs, the results were broadly comparable across study arms.

The cost of the SPORTSMART testing kit was common to all study arms. The kit included equipment for testing and registering as a patient. The overall cost of the kit was £5.66, which included the cost of several specially designed elements to make the pack as attractive as possible to the players (Table 38).

Specimen collection boxes were used in all locations and had a football-themed design to encourage players to deposit their samples. The costs associated with the boxes are shown in the Table 39.

Full costs for each of the intervention arms are shown in Tables 40–42. The results of the pilot trial suggested that total costs were similar across all of the intervention arms, with the total costs of the captain-promoted screening intervention estimated to be £2489.37 compared with £2735.61 for the health advisor-led arm and £2535.61 for the poster-only arm. Overall costs were similar because the highest proportion of costs related to fixed costs, such as staff time for recruiting and briefing the clubs and the equipment required for delivering the promotion.

The aim of the cost–consequences analysis was to examine the costs for all three intervention arms and compare these with the main outcome of screening uptake (Table 43). The average cost per player tested was comparable across the trial arms using the base-case results, with the average cost per player tested for the captain-led promotion estimated to be £88.91, compared with £88.25 for the sexual health advisor-led screening promotion and £81.79 for the poster-only (control) arm. Because overall costs are similar, the average cost per player screened will naturally be affected by the relative effectiveness of any particular intervention model. For example, if a particular intervention arm results in more players accepting screening than an alternative strategy at the same cost, the cost per player screened will be lower relative to the alternative as more players have been screened. However, as uptake rates varied considerably at the club level, our ability to estimate uptake definitively for any single intervention arm is limited.

Discussion

This was an exploratory economic evaluation comparing the costs and outcomes of alternative models for promoting screening amongst young men. The results as a whole suggest that all of these methods of screening promotion are acceptable to players within amateur football clubs, with 153 men receiving the intervention and 90 accepting the offer of screening (59%, 95% CI 35% to 79%). The overall costs associated with the intervention arms were similar. The outcome of average cost per player screened was comparable across all arms, with the average cost per player tested for the captain-led promotion estimated to be £88.91, compared with £88.25 for the sexual health advisor-led screening promotion and £81.79 for the poster-only (control) arm. This outcome is obviously affected by the estimate of the number of players accepting screening and, as our ability to estimate uptake for any single intervention arm is limited, drawing conclusions about the relative costs and consequences of the interventions is not justified. No intervention model can be judged to be dominant and the average cost per player tested can be seen as comparable across all of the study arms.

It might have been expected that the poster-only control and captain-led arms would have significantly lower costs than the health advisor arm as no member of staff was required to deliver verbal information to the players. However, costs were actually found to be similar because of the need for a member of clinic staff to deliver the materials to the study location and ensure the safe return of testing samples to the clinic, to meet clinical guidelines. In the event of a roll-out of the trial, a satisfactory alternative to the presence of a clinic member of staff might be found, which would make the costs associated with these arms slightly lower.

This study has highlighted several limitations, as is typical of exploratory research. It was difficult to draw firm conclusions about the relative cost-effectiveness of the interventions in the trial as screening uptake could not be estimated with precision for any single intervention arm because acceptance rates varied considerably at the club level and the overall target sample size was not achieved. This was because of difficulties in recruitment linked to poor weather and the need to reschedule a number of matches. Within the economic analysis some assumptions were made about how the interventions would operate if they were rolled out and, although these were examined within a sensitivity analysis, they would need to be tested on a larger scale. A one-way deterministic sensitivity analysis was carried out as this is a preliminary economic analysis alongside a pilot trial and a full probabilistic sensitivity analysis would not be appropriate because of the small sample size and the heterogeneity of uptake rates at club level. Further uncertainties around cost and outcome parameters would need to be analysed if a full RCT were conducted. Finally, it was intended that players in the control arm would be uninfluenced by team captains. However, the process evaluation revealed that team captains in this arm had informally encouraged players to participate in screening, because they were so enthusiastic about the study. This is likely to have affected the results for the poster-only arm.

The strength of this study is that detailed data on costs and resource use were collected, which can inform similar interventions in this area and enable comparisons with other research findings. In addition, data were collected on the numbers of players who attended the screening promotion events and thus it is possible to estimate the level of uptake of screening. Often with such health promotion interventions, the number of people exposed to a particular intervention is unknown, which limits the type of economic analysis possible.

Very little information exists about the cost-effectiveness of screening programmes in non-clinical settings. A recent systematic review of studies concerned with chlamydia and gonorrhoea screening outreach programmes identified only three studies that included information on costs. 155 Buhrer-Skinner et al. 164 calculated a cost per test carried out for outreach clinics in Australia but did not include staff time, transport and set up costs. Morris et al. 165 calculated the costs associated with two Californian youth programmes but did not include additional time associated with volunteer input. Detailed costings were also provided by a study evaluating the cost-effectiveness of a multifaceted community intervention to increase screening in Stockholm. 166 However, because of the nature of the intervention, which involved a large-scale publicity campaign and expanded access to testing facilities, it is difficult to compare these results with those of the SPORTSMART study. The findings of a costing study of chlamydia screening within primary care suggest that the costs per case screened for the SPORTSMART study are higher than would usually be expected in a UK primary care setting. 102

However, the additional public health benefits associated with outreach activities would also need to be taken into account. In addition, achieving adequate coverage of screening in primary care, especially in men, is challenging.

Conclusions

This preliminary economic evaluation has shown that similar costs and outcomes were demonstrated across all three study arms. This suggests that further evidence is needed about the most cost-effective way to encourage screening in a football setting. The results may suggest that simply displaying posters and providing testing equipment offers the best value for money compared with providing formal verbal information to promote screening. However, the fact that the control arm was unintentionally ‘enhanced’ by enthusiastic team captains needs further investigation. It may suggest that team captains can have an important influence on screening uptake, irrespective of whether or not they take on a formal role in promoting the intervention. This has implications for economic analyses, as the potential costs associated with their informal and formal input need to be carefully considered. Additional evidence is needed about the public health benefits associated with screening interventions in non-clinical settings so that their cost-effectiveness can be fully evaluated. The potentially higher costs associated with ‘outreach’ activities need to be weighed against their wider health benefits so that policy makers can be fully informed about which approaches offer the best value for money. This may involve consideration of a broader range of outcomes that can capture both the health and non-health benefits potentially associated with public health interventions.

Objective

• To examine the potential public health benefit of the SPORTSMART intervention by comparing the sociodemographic and behavioural profile of men who completed the SPORTSMART survey with that of similarly aged men in the general population.

Methods

Two weeks after the STI intervention was completed, all club members aged at least 18 years were invited to take part in a brief, self-administered, anonymous pen-and-paper survey questionnaire to assess their sexual risk behaviour and STI testing history, as well as collect standard sociodemographic data. Club managers, secretaries and team captains offered the survey to all male members aged at least 18 years and completed surveys were returned to a locked box on display in the club premises. The 20-item questionnaire is provided in Appendix 2. As far as possible, questions used the same wording as in the third National Survey of Sexual Attitudes and Lifestyles (Natsal-3),146,167 permitting comparisons with this general population survey.

Results

In total, 212 men completed the SPORTSMART survey, of whom 192 were aged 18–44 years. This corresponds to an estimated response rate of 61.9% of the total number of questionnaires left with the football clubs. In the Natsal-3 sample, 409 men were in this age group and resident in London. The estimated overall response rate for Natsal-3 was 57.7% and the co-operation rate was 65.8% (of all eligible addresses contacted). 146,167

Discussion

Men attending amateur football clubs who completed the SPORTSMART survey were more likely to report some sexual risk behaviours, and they seemed aware of their increased STI risk, relative to men in the general population. These differences may explain the larger proportion of SPORTSMART men reporting STI testing in the past year. This suggests that the SPORTSMART initiative is unlikely to reach men who would not otherwise use sexual health services. However, the data also suggest that men who participate in amateur football may benefit from the opportunity to test for STIs and to receive health promotion messaging in non-traditional venues such as football clubs. Guidelines recommend STI testing on change of sexual partner,29,31 yet, although almost half of the men in the SPORTSMART sample reported two or more partners in the past year (and thus had experienced partner change), only around one in five reported STI testing during this time. As football is the highest participation team sport in England,137 there are potential health benefits for both the population and the individual from providing novel opportunities for STI testing. Furthermore, the football clubs that participated in the SPORTSMART study were all situated in London, which as a large city in the UK has a number of opportunities to access free sexual health care. It is highly likely that initiatives such as SPORTSMART may therefore yield an even greater public health benefit in more geographically remote settings with limited access to sexual health care. 156

It is important to consider whether or not methodological differences between SPORTSMART and Natsal-3 may account for some of the findings observed. For example, the SPORTSMART sample data may be subject to greater reporting bias as, although they were collected using an anonymous self-completion survey, the men completed the surveys alongside their peers. The direction of this bias is unclear as over- and under-reporting of behaviours are both plausible. It is also necessary to question the impact that participation in the SPORTSMART intervention may have had on participants’ responses in the subsequent survey. STI screening promotion featured in all three arms of the SPORTSMART intervention and so social desirability bias is probable. Because of the anonymous nature of the survey it is not possible to know which men who completed the survey also participated in the SPORTSMART intervention and so it is not possible to determine whether the uptake of testing through the SPORTSMART study varied according to risk behaviour and/or STI testing history.

The SPORTSMART survey used identically worded questions to those used in Natsal-3 for almost all of the characteristics and behaviours compared to allow valid comparisons. One notable exception is the question regarding STI testing. The SPORTSMART survey asked about testing for any STI whereas Natsal-3 asked about testing for particular STIs, including chlamydia. 115 Although this may therefore explain some of the difference observed, this is perhaps a reasonable comparison as a chlamydia test is the minimum test to be offered when testing for STIs, regardless of setting170,171

To ensure valid comparisons in terms of the men being compared it was necessary to limit the Natsal-3 denominator to men resident in London. Thus, although Natsal-3 has a large sample size overall, the number of participants eligible for this study may have limited the detection of statistically significant differences, especially in multivariable analysis. However, it was necessary to use this statistical technique to take account of the confounding effects of demographic and behavioural differences between the two samples. Furthermore, although the Natsal-3 survey is broadly representative of the British general population,28 with a response rate in line with that of other major social surveys completed in Britain around the same time,17,172 it is worth noting that the response rate was slightly lower among young men resident in London, the population group of interest for this paper. 30

In conclusion, men who completed the SPORTSMART survey were more likely to report some sexual risk behaviours and they seemed aware of their increased STI/HIV risk. This may explain the greater proportion reporting to have tested for STIs in the past year relative to men in the general population. Offering STI testing in amateur football clubs – regardless of mode of delivery124 – is therefore unlikely to reach men who would not otherwise test for STIs unless access to sexual health services is poor. 156 However, it does provide men who seem to be at a greater risk of STI transmission an alternative venue to test and receive health promotion advice. As well as increasing access, this approach may also help to normalise sexual health care for men, yielding a public health benefit. With the recent changes in sexual health service delivery and commissioning in England,13,75 including in terms of recognising STI prevention as just one component of sexual health and well-being173 and, in turn, sexual health as just one component of public health, this survey provides empirical evidence to inform the planning and delivery of services and health promotion interventions to maximise public health benefit across the piece.

Conclusions, recommendations for future research and implications for clinical practice The Invisible Man

Our research focused particularly on sexual health needs and provision among young heterosexual men. Throughout the programme we identified wider gender issues that create structural inequalities in sexual health provision and access. Gendered expectations of men’s health-seeking behaviour, along with the offer and provision of services in line with stereotypes of male behaviour and need that are out of line with evidence, often limit men’s ability to seek and use health care. Although there is growing understanding of gender in health,174 this largely focuses on the vulnerabilities of women. As the Chief Medical Officer report states, women need to be able to address their health needs in the workplace and across civil society. 174 Throughout our programme of research, which unusually focuses on the needs of younger heterosexual men, it grew increasingly clear that the health of men and women alike requires radical development in our understanding of the relationship between gender and health.

We identify and challenge an overarching myth that heterosexual men are both incompetent and unwilling to engage with health services. Our data comprehensively demonstrate that young heterosexual men are far more engaged with services than clinicians believe, or than media stereotypes suggest.

Myth and reality: young men’s relationship to conventional and novel sexual health services

Our findings from the national random probability sample survey (see Chapter 3) confirmed that men use, and prefer to use, conventional services, including general practice and GUM clinics, for their sexual health needs, with around three-quarters of young men surveyed having attended their GP in the past year. These findings are supported by the narratives of men who participated in the qualitative interviews (see Chapter 3). These men also placed value on accessing screening in a discrete, time-efficient way that fitted in with their normal routines and day-to-day life. The preference for conventional services may reflect cultural expectations with regard to access to and supply of health care along with a degree of perceived congruency between health-care settings and medical testing. Taken together with the high acceptability of general practice for accessing STI and HIV testing kits, this highlights the importance of, and potential for, chlamydia and HIV screening in general practice, especially among younger men,72 and aligns with the latest NCSP strategy. However, this approach has many inherent challenges, which we discuss below.

The finding that most men had seen their GP in the last year is in keeping with the findings of other studies. 107,108,150,159,175 However, young men’s high reported rates of attendance in general practice are not in line with the perceptions of general practice staff, who tend to substantially underestimate the frequency with which men attend general practice. 102 Current rates of STI and HIV screening in general practice are low, which could suggest reluctance on the part of the health-care provider to offer testing140 or a failure to recognise the attendance of young men as a common opportunity for health promotion.

The many barriers both to opportunistic chlamydia screening and HIV testing in general practice are well recognised. 102,160,176–178 As a result of strongly held beliefs that men are difficult to reach primarily because of non-attendance at traditional health-care venues, attempts to increase chlamydia screening uptake in men have to date focused on trying to create innovative and customised approaches rather than exploiting the potential of mainstream health services. 179 The NCSP data show that only 25% of tests in men are carried out in ‘core’ (general practice, GUM clinic, CASH service) health settings. 151 However, given our findings, and the considerable potential to engage more men in STI and HIV testing through general practice,138 it is likely that basic sexual health services could be more efficiently placed within existing frameworks, irrespective of how ‘non-innovative’ they may seem.

Although many studies have shown that using non-traditional and sports settings to screen for STIs is feasible,134,147 few have focused on the acceptability of different settings for men. Lorimer et al. ’s136 study of willingness to participate in a non-medical approach to chlamydia screening found that men in particular valued the possibility of screening in these settings. Furthermore, the uptake of screening varied by setting, supporting our finding that the acceptability of sports settings was greater in those who had actually engaged in sporting activities over the last month. 136 Our findings contrast with those of others which suggest that anonymity is a key factor determining the acceptability of screening. In a qualitative study of young men’s experiences and perceptions of chlamydia screening commissioned by the NCSP,125 men rejected many of the proposed sports and social venues for fear of the stigma of being seen to take a test. Men also appeared to perceive a degree of incongruity between attending these locations for recreation and the health message of screening. 125 The high uptake of screening within our trial and the narratives of the men who participated (see Chapter 3) suggest that, for some men, undertaking screening within the football team environment felt safe and private, facilitating uptake of the interventions.

The potential of new testing technologies

New technologies are available and widen the range of opportunities to test both within and beyond conventional sexual health services. These technologies are changing the face of established services, yet their evaluation has to date focused largely on issues affecting women. The roll-out of chlamydia testing through the NCSP and other initiatives has raised awareness of STI risk among men, as evidenced in our programme through both the qualitative and the survey work.

Men reported a willingness to use self-collected testing kits for HIV and STIs in the national survey and the high uptake of screening within the SPORTSMART pilot RCT adds support to this finding. The speed and convenience of self-collected samples, together with their ability to afford privacy (they can be collected in private and sent back to laboratories by post), aligns well with men’s views on the acceptability of screening models from the qualitative study and trial process evaluation (see Chapter 3).

Self-sampling kits can be used in a variety of settings, including general practice. However, the use of self-collected urine testing kits is appropriate only for asymptomatic men and would also miss infections at non-genital sites. This is particularly important for MSM, who are recommended to undergo multisite sampling, dependent on reported sexual risk. Opportunities for more comprehensive self-sampling do exist but are currently rather cumbersome: self-collected rectal and pharyngeal swabs appear to perform as well as provider-taken swabs and seem to be acceptable180–182 but to our knowledge have not been evaluated in general practice. Self-collecting specimens for syphilis testing is feasible using dried blood from a finger prick and appeared to be acceptable in one study among MSM recruited from non-clinical settings. 183

Faced by the challenges of implementing wider STI screening in general practice, other settings must continue to be considered. Offering STI screening using self-sampling kits in football clubs may have the potential to widen access to screening and offer a public health impact for men with limited local traditional services. However, men in our outer-London trial and survey (see Chapter 3) appeared to be aware of their STI risk to some degree and reported high levels of previous screening. Simply providing the opportunity to test in these settings appears to result in a similar uptake to interventions that rely on direct promotion by a HCP or club captain (popular opinion leader) and the costs of these approaches within our trial were similar.

Self-testing compared with point-of-care testing: is non-chlamydial, non-gonococcal urethritis a tipping point?

A range of novel testing technologies has recently emerged and is likely to continue to offer new opportunities both for self-testing (e.g. posting of samples) and point-of-care testing (results immediately available to patients). We should not, however, lose sight of the history of point-of-care testing and the implications of novel technologies. STI clinics have long offered a range of point-of-care testing, which has paradoxically declined as the availability of NAAT tests has grown. It is now less common for syphilis or gonorrhoea to be diagnosed through microscopy, although HIV rapid tests now provide quick initial results. We explored NCNGU both as a specific example of biomedical uncertainty as to its pathogenic consequences and as a potential example of outdated fetishism about the ‘point of care’, which may no longer be relevant to STI services. Depending who is asked, NCNGU may be seen as an example of overmedicalisation or as a potentially missed opportunity to prevent the onward transmission of infection.

Currently, men accessing health services are recommended to undergo urethral smear microscopy only if they have symptoms of urethritis63 (although some clinics continue the practice). This means that asymptomatic NCNGU ceases to be a diagnosable and treated condition. Our systematic review (see Chapter 1) shows that available evidence, albeit poor in quality, does not support an association between asymptomatic NCNGU and significant adverse clinical outcomes for men or their sexual partners. The mathematical modelling and cost-effectiveness analysis of removing all asymptomatic urethral microscopy screening suggests that, despite a small rise in adverse outcomes, it would be highly cost-effective to stop all urethral microscopy in asymptomatic men. (see Chapter 1).

These findings fill an evidence gap in the literature and support national clinical guidance in this respect but published studies on which this work is based are few and mostly of poor quality. However, emerging knowledge of the complexity of the urethral biome, together with likely routinely available tests for M. genitalium, should resolve some of the uncertainty around this multifactorial condition. Currently, the sexual health of men may be better served by diverting the resources funding remaining testing and treatment of men with asymptomatic NCNGU and their partners into increasing coverage of screening for STIs with established adverse health consequences.

Novel technologies that can be accessed online or otherwise beyond the settings that people attend because of symptoms or a perceived risk will themselves profoundly influence debates about what health services should be or should offer. Such technologies break down conventional distinctions between a motivated STI clinic visit and an opportunistic invitation to test in the context of some other health-care encounter. Register-based screening, so well suited to cancers with a well-characterised natural history, is not well suited to the disparate clustering of behavioural and lifestyle risks that make STI epidemiology so very uneven. The SPORTSMART study, in particular, raises complex issues about what counts as a health service in the age of online risk calculators and self-sampling.

Partner notification interventions in primary care and community settings

Our partner notification studies demonstrated that, in the age of novel communication technologies, there are trade-offs between partner notification and other prevention technologies. For example, the trade-off for novel partner notification models may be low rates of HIV/STI testing among partners, which may need to be addressed contextually with an eye to relative benefits in different populations.

We developed and evaluated through a pilot RCT two novel interventions for partner notification (APTPharmacy and APTHotline), offered in addition to a ‘routine’ control arm, to which patients were referred through a web tool (see Chapter 2). In all arms of the trial, reported partner notification outcomes were better than previously reported outcomes in similar settings but did not reach national standards for partner notification. Uptake of the specific interventions was low and addition of either type of APT did not appear to improve outcomes. The cost-effectiveness of APTPharmacy and APTHotline over routine partner notification was not demonstrated.

The care pathways that we developed, all of which used a novel online patient and data management tool, provide a feasible and acceptable infrastructure for the onward referral of patients diagnosed with STIs in general practice and other community settings to receive support with partner notification. Outside the research setting, where there are major barriers to recruitment and randomisation at an emotionally taxing time, our model interventions appear to be capable of providing acceptable approaches to partner notification support. These could usefully form part of the portfolio of partner notification choices aimed at improving outcomes for all kinds of partner. It is increasingly recognised that novel interventions should not be seen as stand alone but rather in the context of a portfolio of partner notification modalities that collectively improve overall partner notification outcomes.

The low uptake of follow-up STI testing or HIV testing is notable and suggests that these modes of partner notification, which do not require direct engagement with a clinical service that can provide comprehensive testing, may be unsuitable for higher-risk populations. Although in the future this may be addressed through postal self-sampling as a means of delivering more comprehensive testing, these modes of partner notification are not suitable either for individuals or for populations who have been identified as being at higher risk, such as MSM partners of heterosexual women. 184

If STI screening for men in primary care is prioritised, the need for effective partner notification becomes even more important. The challenges in achieving this should not be underestimated, as shown by the considerable difficulties experienced in this and previous large trials. 102

‘Queering’ sexual health services: a way forward?185

The myths that we encountered about heterosexual males’ health care-seeking behaviour, which have already been disproved in numerous studies, raise interesting issues about how health-care services and policy makers can avoid ‘othering’ men from the beginning of their independent engagement with health services in adolescence. Just as contraception services are a gateway to adolescent women using health services independently, sexual health needs are highly prevalent among young men, yet the health service response to their need if much more limited, and may indeed amount to denial.

By contrast, MSM are recommended to have regular HIV check-ups,117 which might be seen as a feminine and health-protecting behaviour, yet experience very poor outcomes in sexual and other domains of health, as documented in a recent Public Health England report. 186

Interestingly, the health service attitude to MSM and female use of services contrasts with how services tend to marginalise both the health, and the health-seeking behaviours, of heterosexual men. This paradox may provide some insights into ways forward that could address provider perceptions and structural inequalities that may make sexual health care less available and accessible to heterosexual men, even though study after study suggests that they are attending services unnoticed and without receiving health promotion.

It is important to note that this programme of work has focused on sexual health provision appropriate to the needs of heterosexual men. MSM experience a disproportionate burden of STIs and HIV and require more comprehensive suites of testing than those considered here, which are not generally available either in general practice or in enhanced sexual health services in the community. 187 Our work does not address their needs and further research is needed to optimise both provision of and signposting to specialist services for young MSM who may of course also be sexually active with women.

Future research priorities

1. Research to improve understanding of men’s collective behaviours with respect to health interventions and how these could be harnessed to increase uptake. The field could benefit from ethnography and from queer theory, which has been a major current in the exploration of gender, sexuality and society in the humanities. 185

2. Exploration of barriers to, and facilitators of, opportunistic STI screening for men attending general practice, including increasing understanding of why men are not opportunistically offered tests at times when they engage with health care for other reasons. Gendered expectations could be explored and addressed through action research.

3. Development of evidence-based interventions to increase offers of opportunistic STI screening for men attending general practice and development and evaluation of different pathways of access to testing kits in general practice.

4. Partner notification trials: further work is required to optimise the uptake of APT both within and outside of specialist services and to explore the linkage between specialist services and community services, including the trade-off with other priorities.

5. Randomised controlled trial of football club-based screening in geographical areas with limited access to sexual health services.

6. Development of interventions that identify and reach higher-risk partners who may benefit from a more comprehensive range of sexual health services.

7. Better understanding of the issues specific to screening MSM and, in particular, how, with the different epidemiology of STIs in MSM and the current narrow focus on chlamydia, this could negatively impact MSM’s sexual health.

MEDLINE search (18 August 2009)

Cumulative Index to Nursing and Allied Health Literature search (18 August 2009)