Article Text


Reducing the risk of sexually transmitted infections in genitourinary medicine clinic patients: a systematic review and meta-analysis of behavioural interventions
  1. D J Ward1,
  2. B Rowe1,
  3. H Pattison2,
  4. R S Taylor3,
  5. K W Radcliffe4
  1. 1Public Health, South Worcestershire Primary Care Trust, Worcester, UK
  2. 2Department of Psychology, Aston University, Birmingham, UK
  3. 3Public Health and Epidemiology, The University of Birmingham, Birmingham, UK
  4. 4Genitourinary Medicine, Heart of Birmingham Teaching Primary Care Trust, Birmingham, UK
  1. Correspondence to:
 Derek J Ward
 Birmingham and Solihull Health Protection Unit, Bartholomew House, 142 Hagley Road, Edgbaston, Birmingham B16 9PA, UK;


Objectives: Are behavioural interventions effective in reducing the rate of sexually transmitted infections (STIs) among genitourinary medicine (GUM) clinic patients?

Design: Systematic review and meta-analysis of published articles.

Data sources: Medline, CINAHL, Embase, PsychINFO, Applied Social Sciences Index and Abstracts, Cochrane Library Controlled Clinical Trials Register, National Research Register (1966 to January 2004).

Review methods: Randomised controlled trials of behavioural interventions in sexual health clinic patients were included if they reported change to STI rates or self reported sexual behaviour. Trial quality was assessed using the Jadad score and results pooled using random effects meta-analyses where outcomes were consistent across studies.

Results: 14 trials were included; 12 based in the United States. Experimental interventions were heterogeneous and most control interventions were more structured than typical UK care. Eight trials reported data on laboratory confirmed infections, of which four observed a greater reduction in their intervention groups (in two cases this result was statistically significant, p<0.05). Seven trials reported consistent condom use, of which six observed a greater increase among their intervention subjects. Results for other measures of sexual behaviour were inconsistent. Success in reducing STIs was related to trial quality, use of social cognition models, and formative research in the target population. However, effectiveness was not related to intervention format or length.

Conclusions: While results were heterogeneous, several trials observed reductions in STI rates. The most effective interventions were developed through extensive formative research. These findings should encourage further research in the United Kingdom where new approaches to preventing STIs are urgently required.

  • GUM, genitourinary medicine
  • ITT, intention to treat
  • MD, mean difference
  • RCTs, randomised controlled trials
  • RR, relative risk
  • sexually transmitted infections
  • behavioural interventions
  • health promotion
  • genitourinary medicine clinics

Statistics from

Recent reports highlight dramatic increases in sexually transmitted infections (STIs) in the United Kingdom1,2 alongside substantial increases in sexual risk behaviour.3 The government has responded with a national strategy for sexual health and HIV, which aims to improve the evidence base for prevention and sets targets for reducing new infections.4,5 Increased infection rates have placed considerable pressure on genitourinary medicine (GUM) clinics,3,6 which recorded over 1.5 million clinical episodes in England and Wales during 2002.1 Re-infection and re-attendance are thought to be common, but are not identified in routine data. A survey in England estimated that 44% of those diagnosed with gonorrhoea during 1996 had previously attended a GUM clinic.7 Reducing rates of re-infection among clinic patients by promoting behaviour change could contribute to current policy goals and improve public health.

We aimed to systematically review the evidence that behavioural interventions could reduce STIs in GUM clinic patients. A recent publication by the Health Development Agency8 identified two high quality systematic reviews that considered randomised controlled trials (RCTs) of behavioural interventions in clinical settings that reported changes to STI rates.9,10 Both observed considerable heterogeneity between trials and concluded that the effectiveness of interventions was related to their development through formative research. We sought to build upon this previous work using a systematic approach to identify trials, an explicit approach to appraising trial validity, and quantitative methods to explore possible sources of heterogeneity.11,12 We also sought to broaden the scope of existing reviews by considering measures of sexual behaviour change, relating these to infection risk.


Identification of trials

We searched seven databases of completed and ongoing research during January 2004 with no date or language restrictions (Medline, CINAHL, Embase, PsychINFO, Applied Social Sciences Index and Abstracts, Cochrane Library Controlled Clinical Trials Register, and the National Research Register). Searches used terms describing sexually transmitted infections or sexual behaviour combined (AND) with terms for prevention, health promotion, behavioural interventions, counselling, and psychological therapies. A full version of this report is available online.12

Inclusion and exclusion criteria

We included RCTs of behavioural interventions that aimed to reduce the risk of STIs in patients attending GUM or equivalent clinics.12 Infection rates were the primary outcome of interest, but we also included trials describing changes to self reported sexual behaviour. We excluded trials that provided educational materials alone, or reported only knowledge, attitudes, or behavioural intentions. Two reviewers independently assessed trials for inclusion using a standardised form.12

Trial quality

Two reviewers independently assessed trial quality using a standardised form based on the Jadad scoring system.12,13 This considers the adequacy of randomisation, concealment, masking, completeness of follow up, and the use of intention to treat (ITT) analyses.

Data abstraction and analysis

All data were abstracted using a standardised form.12 STI diagnoses were ascertained from laboratory or clinic records, or the results of screening visits. Sexual behaviour change was ascertained from self completed questionnaires or interviews. Dichotomous outcomes were expressed as relative risks (RR) while results for continuous outcomes were expressed as the differences between intervention and control groups’ mean within-subject change from baseline (mean difference, MD).12,14 For infection related outcomes and sexual risk behaviours, a RR less than 1 or a negative MD indicates a lower rate of infections or risk taking in the intervention group relative to control. In contrast, for protective behaviours a RR greater than 1 or a positive MD indicates greater adoption of the protective behaviour in the intervention group relative to control. Infection related outcomes were analysed as ITT while behavioural outcomes represent on-treatment analyses (in many cases no information was reported on those lost to follow up).

Where outcome data were consistently available across studies, results were pooled using random effects meta-analyses (Review Manager, version 4.2, Cochrane Collaboration, 2003). Heterogeneity was explored with stratified meta-analyses using covariates defined a priori (study population, intervention characteristics, and trial quality).12


Search for trials

We included 31 papers reporting 14 trials (identified by principal citation, fig 1).12 Most rejected RCTs did not recruit patients from a relevant setting. Initial agreement on inclusion was good (κ = 0.70) with disagreements resolved by discussion.

Figure 1

 Flow of trials included and excluded from this review. (✶No further details reported on those papers or trials not reporting evaluations or RCTs.)

Study characteristics

Population and setting

Most trials were conducted in the United States (table 1). Two trials recruited over 60% of all participants.15,16 Four trials recruited only males17–20 and three only females.21–23 Three trials focused on adolescents or young women.21,23,26 Most trials restricted entry to those with a recently diagnosed STI,17,18,20–25,27 and/or reporting high risk sexual behaviours.16,18,26 The largest trial did not restrict entry on these grounds and reported a 32% baseline STI rate.15

Table 1

 Characteristics of included trials

Trial quality

Jadad scores were low (⩽3 out of 5) reflecting an expected lack of masking and limited descriptions of trial methods.17,21,26–28 Recruitment, adherence, and follow up rates were low (weighted mean 45% n = 12, 66% n = 10, and 69% n = 12), especially for interventions lasting over two sessions. Recruitment was greatest for small group interventions. Five trials reported differential follow up rates,17,18,23,24,26 though the group with greatest attrition was inconsistent. We labelled five trials as higher quality using a post hoc definition that required a Jadad score of two or three, masking of outcome assessors (seven trials), and non-differential follow up.15,16,19,20,22

Experimental and control interventions

Intervention format and length varied considerably (see table A on STI website One trial also compared an enhanced intervention with a brief one based on US best practice guidelines for individualised risk reduction counselling.15,29,30 Most other US trials adapted these guidelines for their control interventions, which are more structured and detailed than usual care currently provided in UK GUM clinics.12

Experimental interventions explored many similar themes, including risk perception, barriers to safer sex, and triggers to unsafe sex. Individually based interventions concentrated on accepting and negotiating condom use21,23 and personal goal setting,15,20,24,26 using roleplay to develop skills.20,24,26 In addition, group interventions explored self esteem in the context of social expectations16,18,25 and increasing self efficacy.16,22,28 Modelling behaviours and skills was an important aspect of many interventions.16,17,19,22,25,28

The two largest trials cited the theory of reasoned action and, less specifically, social cognition theory as the basis for their interventions.15,16 Social cognition theory asserts that attitudes to a behaviour, perceived support for that behaviour, and self efficacy determine intentions and subsequent actions.31 Other investigators also cited cognitive models such as the health belief model21 and AIDS risk reduction model,22,24 which encourage participants to label themselves as vulnerable, commit to change, take action, and maintain change.32 Three trials cited the information motivation behavioural skills model,19,25,26 which asserts that personalised information, motivation, and behavioural skills are prerequisites of adopting protective behaviours.31 Finally, two trials cited the transtheoretical model of behaviour change.18,22 This represents change in terms of ordered stages allowing health promotion messages to be tailored to an individual’s readiness to change.

Trial results and quantitative data synthesis

Laboratory confirmed STIs

Four out of nine trials considering this outcome observed a greater reduction in their intervention groups relative to controls,15,16,22,25 a result that was statistically significant (p<0.05) for two.15,22 Pooled results do not indicate an overall effect (RR 1.00 (95% confidence interval 0.81 to 1.23)). However, visual examination of the forest plot and statistical testing suggests heterogeneity between trials that prevents over-reliance on summary measures of effect (fig 2, p = 0.001). In addition, a funnel plot of these results suggests evidence of publication bias (see fig A on STI website, p = 0.003 Egger test). One further trial also reported a statistically significant effect on reducing gonorrhoea and urethral discharge, but did not present the original data (reported RRs 0.39 and 0.36).17

Figure 2

 Results for laboratory confirmed STIs, clinically diagnosed STIs, self reported STIs, and self reported symptoms at trial end points. CI, confidence interval; n, number of events; N, number of participants; RR, relative risk. Laboratory confirmed STI includes gonorrhoea, chlamydia, syphilis, HIV, trichomonas (two trials14,22), non-specific urethritis (one trial14), chancroid and lymphogranuloma venereum (one trial18). Clinically diagnosed STIs also includes non-specific urethritis16,18,22,23 or mucopurulent cervicitis,22 pelvic inflammatory disease,22,23 trichomonas, first presentations of genital warts, herpes simplex, or hepatitis B infection,16,23 scabies, or pediculosis pubis,23 and presumptive treatment of either gonorrhoea or chlamydia on clinical grounds.18,22 Results for the trial by Kamb relate to the enhanced intervention.13 Equivalent results for the brief intervention are RR 0.82 (95% CI 0.68 to 0.99).

Trials reporting significant effects had among the greatest adherence and follow up rates.15,22 Generally, high quality trials were more likely to report a protective effect, though pooled results were not significantly different (high quality RR 0.89 (0.70 to 1.12), low quality RR 1.29 (0.89 to 1.87)).15,16,20,22 Pooled results for trials enrolling both sexes15,16,24,25 (RR 0.90 (0.78 to 1.04)) or young women21,22 (RR 0.97 (0.41 to 2.32)) were greater than results for the remaining trials.18,20

Intervention format or length was not associated with trial results (group based RR 0.94 (0.70 to 1.25) versus individually based RR 1.16 (0.76 to 1.75)). Similarly, Kamb found that both brief and enhanced interventions were equally effective when compared to a short information session.15 We did find evidence that the theoretical basis of interventions was related to effectiveness, though differences in pooled results were not statistically significant (social cognition models RR 0.91 (0.71 to 1.16) versus other trials RR 1.19 (0.84 to 1.68)). In addition, extensive formative research, including interviews, focus groups, input from community representatives, and pilot testing, was reported by the four most effective trials.15,16,22,25

Clinically diagnosed STIs

Four studies considered clinically diagnosed STIs (fig 2). One trial found no reduction in any individual diagnosis25 while others observed greater STI rates in their intervention groups compared to controls.18,20,24 Pooling of these results suggests an overall effect (RR 1.23 (1.01 to 1.50)) and there was no evidence of statistical heterogeneity (p = 0.58).

Self reported STIs

Two trials considered self reported diagnoses and found no significant effects.23,26 However, one trial observed an effect on self reported symptoms that was greater than that for laboratory confirmed infections.16

Self reported condom use

Six out of seven trials reporting consistent condom use (condoms always used or used at every sexual encounter) observed a greater increase in their intervention groups than controls.15,16,18,19,23,24 Examination of the forest plot and statistical testing do not indicate heterogeneity between trials (fig 3, p = 0.61) and the pooled result suggests an overall effect (RR 1.17 (1.10 to 1.25)). In contrast with laboratory confirmed infections, there is no evidence of publication bias (see fig B on STI website, p = 0.98 Egger test).

Figure 3

 Results for consistent condom use (condoms used (nearly) always or no unprotected sex) at trial end points. CI, confidence interval; n, number of events; N, number of participants; RR, relative risk. Outcomes refer to self reported sexual behaviour for the previous 1–12 months. Results for Imrie relate to the previous 1 month.16 Equivalent results for 12 months were RR 1.23 (95% CI 0.93 to 1.62).

Pooled results do not suggest that effects were related to trial quality (high quality RR 1.19 (1.11 to 1.28) v low quality RR 1.12 (0.97 to 1.29)), intervention format (group based RR 1.18 (1.05 to 1.33) v individually based RR 1.16 (1.06 to 1.28)), intervention length, the use of theory, or formative research. Results for trials recruiting females (RR 1.40 (0.88 to 2.20)) were greater than for trials recruiting mixed populations (RR 1.19 (1.10 to 1.27)) or males alone (RR 1.14 (0.92 to 1.42)).

Five trials considered the proportion of sexual encounters protected by condoms.16,19,22,24,28 Two found increased rates in intervention subjects relative to controls (table 2). Other trials reported inconsistent effects on refusing unsafe sex19,28 and using condoms with recent partners.23,24

Table 2

 Results for behavioural outcomes at trial end points (excludes consistent condom use, fig 3)

Sexual partners

Five out of seven trials reported fewer sexual partners among their intervention groups relative to controls (table 2),17,22,24,26,28 while two reported fewer partners among controls.19,25 Two of these trials also reported a reduction in sexual encounters with strangers or non-monogamous partners,26 and more avoidance of sex with partners suffering from known STIs.22 However, another intervention had no effect on homosexual men reporting unprotected sex with a partner of different or unknown HIV status.18


Several trials of behavioural interventions in sexual health clinic patients observed a greater reduction in laboratory confirmed STIs among their intervention groups relative to controls despite using different intervention formats in different populations. Successful trials were larger and of higher quality than those reporting no reduction in infection rates, and the most effective interventions were based on social cognition or related theoretical models and developed through extensive periods of formative research. We also found evidence that behavioural interventions increase consistent condom use, though effects on other aspects of sexual behaviour were inconsistent. There was no evidence that an intervention’s success in reducing infections or increasing consistent condom use was related to its format or length.

Key messages

  • Interventions that could modify behaviour in GUM clinic patients (a high risk group) could play an important part in reducing recurrent STIs and thereby improve public health

  • Two large high quality trials of behavioural interventions have demonstrated statistically significant reductions in STI rates despite using different intervention formats and targeting different populations

  • Most trials also demonstrate an increase in self reported condom use, though effects on other aspects of sexual behaviour are inconsistent

  • The most successful interventions were based on social cognition models and underwent careful development through extensive formative research. These results should be used to direct future research in a UK setting

One trial observed no difference in effect between a multiple session enhanced intervention and a two session brief one.15 This brief intervention was similar to the control intervention used by many trials,20–22,24,25 including one that successfully reduced infection rates.22 In effect, the control groups in these trials received an intervention that goes beyond current UK practice, making it difficult to generalise results or transfer effect estimates to a UK settings.

The quality of identified studies was frequently poor. Masking is difficult to achieve in trials of behavioural interventions, but we did consider masking of outcome assessors in our quality assessment as this may reduce bias.33 We included only RCTs in this review, which could have resulted in the omission of potentially relevant information. However, a suitable unbiased control group is essential to avoid ascribing sexual behaviour change to the effects of the intervention rather than the diagnosis of an STI itself.32,33 In addition, we found evidence of publication bias for the primary outcome (laboratory confirmed infections), though we attempted to identify all publicly funded research in the United Kingdom.

Two previous reviews also reported inconsistent results, though they suggest that adherence rates and the development of an intervention through formative research were predictors of success.9,10 We identified five trials not included in these reviews,17,20,23,26,28 four of which report infection related outcomes, yet our findings suggest similar factors are important to an intervention’s effectiveness. Formative research seeks to identify the behaviours, motivations, and beliefs within the target population that lead to increased risk, and link these to the key elements of an intervention.10,32 The format, setting, delivery, and acceptability of the intervention can then be explored through pilot testing.32 Theory may provide a framework for this process, though models do not specify the inclusion of particular elements within an intervention, and we also found evidence that effective interventions were more likely to be based on theoretical models relating behaviour to individual cognitions.

Trials appeared more effective at increasing (consistent) condom use than reducing infections. This could represent a chance event, or bias resulting from the different methods of ascertainment. Alternatively, small changes in condom use may be an insensitive marker of overall STI risk, which is the sum of many behavioural factors.10,32 At the population level, increased condom use has been accompanied by increases in other risky sexual activities and a marked rise in infections.3 Further analysis of the NIMH trial data also suggests that the number of sexual partners remains an important risk factor for gonorrhoea infection regardless of changes to condom use.34 However, too few trials reported both behavioural outcomes alongside infection rates to explore this issue further.

This review highlights a need for research on behavioural interventions in a UK setting, where new approaches to reducing infection rates are urgently required. Heterogeneity between trials for the primary outcome of interest means that we cannot rely on simple summary measures of effect, but there is evidence that interventions can successfully change sexual behaviour and reduce infection rates if appropriately developed and targeted to a population. The success of interventions appears unrelated to their intensity. This is important, as the number of sessions is likely to be a major factor in determining cost. However, even the briefest intervention shown to be effective could not be introduced into routine GUM practice without further investment in premises, staff, and training. Information on the likely effect size, acceptability, and cost effectiveness of introducing behavioural interventions into the United Kingdom will be critical to those developing new services. However, this information cannot be simply extrapolated from existing evidence, which mainly derives from the United States. We propose that future studies should develop their experimental interventions by tailoring the approaches shown to be most effective in US settings, and crucially inform this process by a period of formative research that incorporates qualitative approaches. While this is likely to be time consuming and costly, without such an approach any proposed intervention is unlikely to be successful.


The idea for the review was proposed by KWR, and the review was planned by DJW and RST with assistance from all authors; DJW developed the search strategy and together with BR undertook the searches, appraised the articles, and extracted the data; DJW analysed the data with the help of RST and HP; writing up the report was principally done by DJW with input from all members of the review team; in particular, RST advised on systematic review methods; HP advised on behavioural interventions, their theoretical basis and application, and KWR advised on sexual health and GUM clinic practice.


View Abstract
  • The appendices are available as a downloadable PDF (printer friendly file).

    If you do not have Adobe Reader installed on your computer,
    you can download this free-of-charge, please Click here


    Files in this Data Supplement:

    • [view PDF] - Supplemental table. Experimental and control interventions employed by trials.
    • [view PDF] - Figure 4. Funnel plot for trials reporting change in laboratory confirmed STI rates (RR <_1 indicates="indicates" that="that" infection="infection" rates="rates" were="were" lower="lower" in="in" the="the" intervention="intervention" group="group" relative="relative" to="to" controls.--="controls.--" end="end" desc="desc" dc1="dc1" _2="_2" _--="_--">
    • [view PDF] - Figure 5. Funnel plot for trials reporting the proportion of participants reporting consistent condom use (RR >1 indicates that a greater proportion of the intervention group reported consistent use relative to controls).


  • DJW and BR are employed on the West Midlands Public Health Higher Specialist Training Scheme. At the time of undertaking this study, both were working at South Worcestershire Primary Care Trust. HP is employed by the School of Life and Health Sciences at Aston University, while RST is employed by the University of Birmingham and funded by the West Midlands Regional Public Health levy. KWR is employed by the Heart of Birmingham Teaching Primary Care Trust and works as a consultant in genitourinary medicine at the Whittall Street Clinic.

  • Funding: All contributors undertook this review as part of their normal employment and did not receive specific funding for this study, which was carried out independently of their employers.

  • Competing interest: The authors have no competing interests.

  • Ethical approval: Ethical approval was not required for this study.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Brief Encounters
    Rob Miller