Sexually transmitted infections (STI), including infection with HIV, are a major cause of morbidity and mortality worldwide.1 2 STI other than HIV have been implicated in the transmission and acquisition of HIV infection and their treatment can play an important role in HIV prevention.3 4 Numerous large-scale, clinical trials are ongoing to test the efficacy of HIV-1 prevention interventions, including vaccines, microbicides, pre-exposure prophylaxis and the suppression of genital herpes simplex infection. Such trials typically aim to recruit persons at high risk of HIV-1 acquisition in order to acquire sufficient numbers of incident HIV-1 infections for adequate statistical power. Although many trials have included expanded STI screening for either research or patient care purposes,5 6 the incorporation of routine, basic STI services for study participants has not been uniformly adopted. When protocols call for long-term follow-up of persons at high risk, we present the argument that offering syndromic treatment alone is insufficient and should be supplemented with basic laboratory investigations and expanded risk assessment in the context of HIV-1 prevention trials.

We present our experience with a basic STI screening programme in the context of an ongoing cohort study in preparation for HIV-1 vaccine trials based near Mombasa, Kenya. Although the site has been participating in a multicentre, multicountry effort to identify volunteers at high risk of HIV-1 infection in Africa, the addition of a basic microscopy for STI screening programme was site-specific, as no such screening was available in the community. Our primary objective was to demonstrate the value of routine, basic STI screening at enrolment into an HIV-1 vaccine feasibility cohort study, with the simple hypothesis that STI screening at enrolment into the cohort would result in the increased diagnosis and treatment of STI among study participants. When several of our volunteers reported recent anal intercourse at enrolment, another site-specific element in our recruitment strategies became an expansion of the eligibility criteria to include a self-report of recent anal sex. Therefore, another objective became highlighting the importance of soliciting a history of receptive anal intercourse (RAI) in adults identified as “high risk”. We hypothesised that anal sex was not uncommon among high-risk adults and that participants with a history of recent RAI would be at increased risk of prevalent STI, including proctitis.

METHODS

Study population and data collection

In a clinic near Mombasa, Kenya, prospective volunteers at risk of HIV infection were identified and recruited into a cohort study in preparation for HIV-1 vaccine trials, starting in July 2005. Recruitment was aimed at high-risk groups such as female sex workers (FSW), persons with a recent STI or multiple partners, HIV-1-serodiscordant couples, or any other person who in the opinion of the investigator was at higher risk of HIV-1 infection. When a number of men and women recruited reported practising anal sex, the inclusion criteria were modified to add any person reporting anal sex in the previous 3 months, and recruitment of men who have sex with men (MSM) and women practising anal sex was intensified by peer educators. The National Ethical Review Committee of the Kenya Medical Research Institute approved this study and all participants provided written, informed consent.

Screening procedures at enrolment

A standardised, detailed sociodemographic and sexual behaviour history was completed through a face-to-face interview with a counsellor. When anal sex was practised, information on whether the participant took an insertive and receptive role (for men) was obtained. Subsequently, study physicians obtained a brief medical history and performed a physical examination, including a genital examination. A bimanual and speculum-aided pelvic examination was performed for women, during which vaginal and endocervical swabs were systematically collected for microscopic examination. Urethral swabs were collected for symptomatic men or if urethral discharge was present on examination. The presence of genital ulcers was recorded. Venous blood samples were collected for HIV-1 and syphilis serology.

From September 2006, male and female participants who reported RAI in the previous 3 months upon enrolment screening were asked about symptoms of anorectal pain, rectal discharge or bleeding, pain during sex, pain with defecation, or perianal ulcers or warts during the same period. Proctoscopy was offered to all participants reporting recent RAI to evaluate the STI burden of anal intercourse, and included inspection for perianal ulcers or condylomata, mucosa erythema or ulcerations and mucopurulent rectal discharge, as well as the collection of swabs of rectal secretions for Gram staining.

All participants received comprehensive risk-reduction counselling, free male or female condoms and water-based lubricants, and treatment of STI in accordance with WHO treatment guidelines.7 Symptomatic vaginitis and urethritis were treated at the enrolment visit. Women were provided with treatment for cervicitis if mucopurulent discharge was seen. The results of laboratory testing were available 10 days after each visit. Participants were asked to return for results and laboratory-confirmed STI were treated at this time. Tracing was initiated if a participant with an STI diagnosis failed to return for results. Participants who were HIV-1 positive at screening were offered either referral to a collaborating point of care or enrollment into a parallel HIV-1-positive cohort in which they received comprehensive HIV care and periodic STI screening. Follow-up of HIV-1-positive participants in a cohort with a similar appointment scheme has helped in retaining populations particularly vulnerable to HIV-1 infection while maintaining confidentiality about HIV-1 serostatus.8 Volunteers received a small reimbursement for travel expenses.

Laboratory procedures

Trichomonas vaginalis was diagnosed by the identification of motile trichomonads on a saline wet preparation of vaginal secretions. Vaginal candidiasis was defined as the identification of organisms consistent with Candida species on a 10% potassium hydroxide slide. Bacterial vaginosis was diagnosed by microscopic analysis of a Gram-stained vaginal smear; a Nugent’s score of 7 or greater was considered positive for bacterial vaginosis.9 Cervicitis was defined as the presence of 30 or more polymorphonuclear leukocytes per high power field on a cervical Gram stain. Urethritis was defined as the presence of five or more polymorphonuclear cells per high power field on a Gram-stained urethral smear. As a result of limited resources, T vaginalis and other pathogens were not tested specifically in men. The detection of Gram-negative, intracellular diplococci consistent with Neisseria gonorrhoeae was used to define gonococcal infection. Proctitis was defined as the presence of five or more polymorphonuclear cells per high power field on a rectal Gram stain.10 Laboratory technicians underwent training in the University of Washington’s Mombasa-based research laboratory, with periodic quality control of readings. Quality assurance of the STI screening programme was locally supported and was not part of the multicentre study procedures.

HIV-1 was tested on site by two rapid test kits (Determine, Abbot Laboratories, USA; Unigold, Trinity Biotech plc, Ireland). Discrepant rapid HIV-1 test results were resolved using a fourth-generation HIV antigen/antibody assay performed at the Kenya AIDS Vaccine Initiative laboratories in Nairobi.11 Screening for syphilis was performed at the KEMRI laboratories in Kilifi using rapid plasma reagin, with positive samples confirmed by a Treponema pallidum haemagglutination assay.

Data analysis

Data were analyzed using Stata 9.2 and were summarised using frequencies for categorical variables and medians with interquartile ranges for continuous variables. Sensitivity, specificity and predictive values of symptoms for the diagnosis of various genital tract infections were calculated according to standard formulae, and 95% CI were calculated using exact binomial estimation. Missing data (eg, from participants who refused a genital examination, or from visits before the addition of questions on anorectal symptoms in September 2006) were excluded from the analysis. Associations between reported recent sexual behaviour, prevalent genital tract infection and HIV-1 or syphilis were evaluated using multivariate logistic regression for records with complete clinical and laboratory data. A forward stepwise model was created, retaining variables in association with the outcome (p<0.2) or resulting in a greater than 10% change in the unadjusted odds ratio. p Values were two-tailed unless otherwise noted, with values less than 0.05 considered statistically significant.

RESULTS

Characteristics and STI prevalence of study population

A total of 897 participants at risk of HIV-1 infection were enrolled in the study cohort between July 2005 and June 2007. A total of 304 (34%) study volunteers self-identified at enrolment as commercial sex workers, 316 (35%) as MSM, 169 (19%) as having multiple recent sexual partners, 58 (7%) as having STI symptoms, 28 (3%) as having an HIV-1-seropositive partner and 22 (2%) were peer leaders who helped with recruitment, but were at slightly lower risk. The sociodemographic and sexual behaviour characteristics of this cohort, including the use of condoms and family planning methods, are presented in table 1; in this table, “regular” partners are those identified as persons with whom there is an ongoing relationship, whereas “casual” partners are persons with whom there is no expectation of a continuing relationship. The overall prevalence of STI symptoms and signs, as well as laboratory-diagnosed STI in men and women at cohort enrollment, is presented in table 2. For the analysis of symptoms and signs, 22 women and 36 men had incomplete data and were excluded. For the analysis of laboratory-confirmed STI, 37 women and 50 men were excluded as a result of incomplete data.

Value of microscopy

Of the 32 women diagnosed with trichomoniasis, 20 (62%) had not complained of vaginal discharge. Of the 34 women diagnosed with cervicitis, 20 (59%) reported no vaginal discharge and 29 (81%) had no cervical mucopus on speculum examination. Pelvic inflammatory disease was diagnosed clinically and treated in 13 women, of whom three reported no vaginal discharge. Of the 27 cases of laboratory-diagnosed urethritis, 18 (67%) had no compatible symptoms. The sensitivity, specificity and predictive values of genital tract symptoms for the laboratory diagnosis of an STI at enrolment are presented in table 3.

Receptive anal intercourse

RAI in the previous 3 months was reported by a total of 191 men (36%) and 64 women (18%); 57 (89%) of the women reporting recent RAI identified themselves as FSW. From September 2006, clinicians documented anorectal symptoms and offered proctoscopy to all participants engaging in RAI to screen for rectal STI. Anorectal symptoms including pain and bleeding were common among persons reporting RAI (reported by 33%); rectal discharge was less common (reported by 10% of men and 2% of women). Participants reporting anorectal symptoms were more likely to agree to undergo proctoscopy (61%) compared with those who were asymptomatic (30%, p<0.001). Men were more likely to accept proctoscopy than women, irrespective of symptoms (45% versus 26%, respectively, p = 0.024). Among the 69 participants who had a complete anorectal examination by proctoscopy, mucoid or mucopurulent discharge was seen in 14 (20%) participants, mucosal erythema in 21 (30%) and ulcerations in five (7%). No anal ulcers were seen in persons diagnosed with syphilis, although only two such persons underwent proctoscopy. Proctitis was diagnosed by Gram stain in four (7%) participants, all of whom were men; three were symptomatic.

Table 4 presents associations between prevalent HIV infection and either recent RAI or prevalent STI at enrolment among men and women in the cohort. RAI was strongly associated with HIV-1 in men (adjusted odds ratio (aOR) 3.8, 95% CI 2.0 to 6.9), but this association did not reach significance in women (aOR 1.2, 95% CI 0.5 to 2.5). Prevalent HIV-1 infection was significantly associated with anogenital ulcers, anogenital condylomata and confirmed urethritis in men. In women, prevalent HIV-1 infection was significantly associated with anogenital condylomata and pelvic inflammatory disease; associations with vaginal trichomoniasis and syphilis were of borderline significance. Of note, two-thirds of the women with active syphilis on enrolment reported recent RAI. In a separate multivariate analysis of factors associated with syphilis in women, recent RAI remained significantly associated with prevalent syphilis on enrolment (aOR 12.9, 95% CI 3.4 to 48.7, full analysis not included).

DISCUSSION

This study was conducted in a group of high-risk young adults recruited into a cohort study in preparation for HIV-1 vaccine trials and was therefore selected on the basis of their high-risk behaviours, including commercial sex work. In this population, the burden of untreated STI identified at enrolment was relatively high, even with the use of limited screening methods. Not surprisingly, basic microscopic examination of vaginal, cervical, urethral and rectal secretions identified a large number of STI in persons not complaining of discharge, who would not normally have been treated using a syndromic approach. The limitations of the syndromic approach are widely recognised,7 12^–14 especially for the detection of cervicitis, which is often asymptomatic in women.15 We found that both vaginal trichomoniasis and urethritis would also have frequently been missed if symptoms alone were used to identify STI in our cohort. Microscopy also allowed us to avoid treating all women with vaginal discharge and a positive STI risk assessment for cervicitis, as recommended by WHO syndromic treatment guidelines in the absence of microscopy.7 This practice leads to increased cost and the potential for adverse effects and drug resistance.12 Therefore, we conclude that the addition of basic microscopy can be a valuable intervention in adults who are identified for cohort recruitment based on their presumed higher risk of HIV-1 infection.

Clearly, additional cases of STI would have been identified by expanded screening, including culture for T vaginalis and N gonorrhoeae or nucleic acid amplification testing for N gonorrhoeae and Chlamydia trachomatis. For example, the prevalence of gonococcal cervicitis was 5% at enrolment into a Mombasa-based FSW cohort in which culture has been routinely available.16 Using only microscopy in our cohort, we found no cases of gonococcal cervicitis by Gram stain alone. In a recent circumcision trial of mostly heterosexual Kenyan men, nucleic acid amplification testing identified gonococcal urethritis in 2% of participants and C trachomatis in 5%; T vaginalis was detected by culture in 2%.6 We found a 1% prevalence of gonococcal urethritis by Gram stain alone and were unable to identify C trachomatis or T vaginalis infections. Because herpes simplex infection is an important cause of genitourinary disease and increases the risk of HIV-1 transmission,17 specific testing for herpes simplex 2 infections would also be valuable for counselling and patient management purposes. Although our data demonstrate that even low-cost microscopy can have considerable value in high-risk populations, expanded screening would be needed to estimate the prevalence of specific infectious aetiologies and further reduce the STI burden in our study population.

Several reports have shown that anal intercourse in both men and women may be more common than was previously thought in sub-Saharan Africa.18^–20 Despite strong local convictions that MSM behaviour, in particular, is incompatible with traditional African culture, recent studies in Senegal and Kenya have proved otherwise.8 21 22 Ferguson and Morris23 stressed the importance of specific, carefully worded questions to assess the occurrence of anal intercourse in women. Both recall and social desirability bias may influence responses; in a South African coital diary study, FSW reported a total of five anal sex acts per week compared with only one anal act per week when responding to a recall questionnaire.24 Despite increasing evidence that anal intercourse is an important and not uncommon risk factor for HIV/AIDS, questions on RAI initially included in the recently conducted national AIDS indicator survey in Kenya were rejected as being too offensive to ask (L Marum, personal communication). Unfortunately data on the general population practice of anal sex in Kenya remain elusive. In our high-risk cohort, we found that men and women frequently reported RAI in the previous 3 months. The frequency of condom use was lower for anal intercourse than for vaginal intercourse, as reported in previous African studies.22 25 The majority of women reporting RAI were sex workers, but other women did report this practice. Sex workers report receiving higher payment for RAI than for vaginal intercourse (unpublished observation); in addition, some women may use RAI 1as a form of birth control or to avoid vaginal sex during menstruation. Given that anal sex is not uncommon among high-risk adults, we recommend that STI screening should include questions on RAI and the diagnosis of proctitis when symptoms are present; this is particularly important in research settings and programmes aimed at reducing the risk of HIV-1 acquisition.

Unprotected RAI is reported to be the most efficient mode of sexual transmission of HIV among both MSM and heterosexual couples26 27 and increasing attention has recently been drawn to the role anal intercourse plays in HIV-1 transmission in sub-Saharan Africa.8 22 25 28 29 We have, like others, presented evidence that African MSM are at higher risk of HIV-1 infection.8 30 31 In our cohort men who practised RAI had a higher HIV-1 prevalence risk than those practising only insertive anal intercourse.8 Among women in our cohort, RAI was not associated with prevalent HIV-1 infection, but women practising RAI were more likely to have syphilis. It is not clear why this difference was found, although possibilities include differences in syphilis or HIV-1 prevalence among insertive partners, differences in condom use or sexual practices not captured by our questionnaires, or different biological susceptibilities between men and women. Questions about RAI have not been routinely included in studies of syphilis risk factors in African women32 33 and our research suggests that such questions are an important component of a sexual risk assessment in general.

Our STI screening programme was conducted in the context of a large, multicentre HIV-1 cohort study aimed at recruiting high-risk adults in preparation for HIV-1 vaccine trials. The study findings challenge general approaches to recruiting at-risk adults. First, to identify volunteers at the highest risk of HIV-1 infection, it may be preferable to enquire about specific sexual behaviour rather than rely on transactional sex, numbers of partners and condom use as the sole indicators. Second, volunteers reporting recent anal sex may have anal infections that should be managed appropriately for public health reasons. Recent guidelines on the management of proctitis recommend proctoscopy and the collection of Gram stain for all people reporting RAI.34 35 Third, HIV-1 prevention studies will target vulnerable and socially isolated people who lack access to appropriate healthcare. As STI treatment can reduce both the risk of HIV-1 acquisition and transmission,36 37 the incorporation of STI screening into prevention studies not only provides benefit to research participants but also to the communities at large. Finally, populations at higher risk of HIV-1 infection are increasingly more difficult to identify and retain, especially in areas in Africa where general population HIV-1 prevalence and incidence are declining.1 STI screening and treatment may aid in the retention of volunteers—the crux of any intervention study. We have not presented data on incident STI in this paper, nor have we been able to assess the optimal period for repeat STI screening in this population. Further research into optimal STI screening procedures, in particular for the screening of persons practising RAI, is clearly needed.

In conclusion, adults with high-risk sexual behaviour are at increased risk of both HIV-1 and other STI and may have a substantial burden of untreated STI upon entry into research studies. Basic microscopy is a valuable, low-cost component of STI screening that can identify asymptomatic infections and avoid overtreatment of STI based on non-specific symptoms. Soliciting a history of RAI in high-risk persons is important to identify persons in need of intensified risk-reduction counselling. Further research into appropriate methods for proctitis screening in developing countries would add to the value of such screening. We believe that improving STI screening should be integrated into HIV-1 prevention research targeting high-risk populations, as an important service to participants and their communities.