WORLDWIDE, AN ESTIMATED 340 million curable sexually transmitted infections (STIs) occur annually, and vaginal infections (such as bacterial vaginosis [BV] and yeast infections) are even more common.1,2 Studies have shown that reproductive tract infections (RTIs) facilitate human immunodeficiency virus (HIV) transmission by increasing both infectiousness and susceptibility.3 Curable RTIs, when not treated properly, can also lead to other serious sequelae such as pelvic inflammatory disease, infertility, and negative health outcomes during childbirth.4,5 In many resource-poor settings, including public clinics in South Africa, RTIs are diagnosed using syndromic management guidelines.6 Syndromic management typically results in overtreatment of symptomatic women (except for genital ulcers) and no treatment of asymptomatic women.7 Improving the diagnosis and treatment of curable RTIs is therefore an important public health endeavor.
The gold standard for obtaining specimens to diagnose RTIs in women is for a clinician to take endocervical and/or vaginal samples during a speculum examination. However, this procedure is invasive, time-consuming, and requires a private clinic space with a gynecological examination table and sterile speculums. In some instances, it may not be culturally acceptable. Alternative sampling methods, such as self-obtained tampons, vaginal swabs, and urine samples, are therefore needed.8 Alternative sampling methods could also be used in large HIV prevention intervention studies or other studies requiring frequent testing for RTIs to improve study participation and cohort retention, simplify trial logistics, and reduce costs. Although self-sampling has been successfully used in a variety of settings with a variety of diagnostic tests,8–21 it has not been adequately validated as a screening strategy for multiple RTI pathogens in resource-poor settings.
The goal of this study (and a subsequent study) in a community health center in Gugulethu, Cape Town, South Africa, was to evaluate RTI sampling techniques that do not require a pelvic examination. In this first study, the validity, feasibility, and acceptability of self-sampling using vaginal swabs or tampons compared to clinician-obtained swabs during a speculum examination were assessed.
Materials and Methods
Study participants were recruited from the STI, family planning, maternal and child health, and general clinics within the NY1 Community Health Center in Gugulethu, which is a resource-poor, periurban area adjacent to Cape Town. Women were eligible to participate if they were 18 years or older, sexually active, not pregnant, and willing and able to comply with the study protocol and give written informed consent. They were excluded if they had taken antibiotics in the 4 weeks before study enrollment, were participating in a study evaluating a vaginal product, or had a history of sensitivity to latex.
Figure 1 shows the study design and flow of specimens. The aim was to enroll a total of 450 women: 300 women attending the NY1 Center for any reason other than to seek care for an STI or vaginal infection (non-STI group) and 150 women attending the NY1 clinic with STI symptoms (STI group). In each group, half the women (those with odd-numbered study identification numbers) were asked to self-sample using 1 tampon and the other half (with even-numbered study identification numbers), using 2 vaginal swabs. Additionally, a research nurse obtained multiple vaginal and endocervical swabs from all women in both groups during a speculum examination. Within each group, the order of specimen collection was further divided (with half using the self-administered method first and half the speculum examination first) in case the order affected the quality of the specimens. All specimens were tested for a variety of RTIs using the same laboratory tests for each type of specimen. The Research Ethics Committee of the University of Cape Town (UCT) and the Population Council’s institutional review board reviewed and approved the study.
General Study Procedures
After written informed consent was obtained, a counselor administered a structured questionnaire on demographics and sexual behavior, followed by counseling on RTIs (including HIV) and safer sex. Women were offered HIV testing and condoms free of charge if so desired. Next, a research nurse guided the woman through the self-sampling procedure and performed the pelvic examination in the preassigned order; all women were asked to wait 15 to 20 minutes between the 2 procedures, during which participants were interviewed about contraceptive and medical history. Papanicolaou smears and blood samples for syphilis testing were collected as a service to the participants. Menstruating women were asked to return after completing menses. Participants were interviewed about acceptability of the self-sampling and speculum examination procedures after they completed all clinical procedures at the sampling visit and again 2 weeks later, when they were asked to return to the study clinic to collect their test results. Some participants were also asked to participate in focus group discussions soon after they completed all study procedures. Although RTI treatment was mostly based on laboratory-confirmed diagnoses (any positive test on any type of specimen), genital ulcer disease, pelvic inflammatory disease, and severe cervicitis were treated immediately (before laboratory results were available), according to the South African syndromic management guidelines.6 Sexual partners of women with genital ulcer disease or laboratory-confirmed syphilis, gonorrhea, chlamydia, or trichomoniasis were also offered treatment.
A research nurse observed each self-sampling procedure but did not interfere. Women in the tampon group were asked to insert a nonapplicator minitampon for light menstrual flow (Lil-lets, Accantia, Solihull, U.K.). After 10 minutes, they were asked to remove it, place it (while holding it by its string) into a sterile 50-ml universal container (UC) prefilled with 12 ml of phosphate-buffered saline (PBS), and push it down with a clean swab to the bottom of the tube. A research nurse later pulled the tampon out of the tube using its string, placed it into a disposable funnel inserted into a clean tube, and squeezed it against the side of the funnel using her gloved right index finger to collect a minimum of 2.5 ml tampon-derived fluid. Women in the vaginal swabs group were asked to insert 2 cotton-tipped swabs (20 cm in length) high in the vagina and swirl them around the vaginal vault while in a squatting position. They were asked to mark with their finger up to where the swab had been inserted. The research nurse subsequently measured the insertion depth. She then placed one of the swabs into a tube containing 3 ml of PBS and squeezed it against the side of the tube. The second swab was placed into a Digene Hybrid Capture 2 (HC2) DNA Collection Device (Digene Corporation, Gaithersburg, MD) containing 1 ml transport medium and left inside until further processing in the HPV laboratory. Fluid obtained after squeezing the specimens in PBS was aliquotted as follows: 0.5 ml in 2 ml Amplicor CT/NG Transport Medium (Roche Molecular Diagnostics, Branchburg, NJ), 2 drops in 5 ml modified Diamond’s medium (MDM) for T vaginalis culture and 0.5 ml in 1 ml Digene HC2 DNA Collection Device fluid (tampon-derived fluid only), and 1 drop to prepare each slide (slides were air dried and fixed in 80% methanol before transport). A clean swab was inserted into the remaining fluid and placed into 5 mg Stuart’s transport medium for yeast culture.
During each speculum examination, a research nurse collected 3 vaginal swabs, an endocervical swab from the Amplicor CT/NG Specimen Preparation Kit, an endocervical brush from the Digene HC2 DNA Collection Device, and a cytobrush to prepare a Papanicolaou smear. The order in which the vaginal swabs were taken was rotated in case the order would affect the quality of the specimens. Immediately after collection, 1 swab was placed in Stuart’s transport medium, 1 was used to inoculate MDM medium, and 1 to prepare a slide.
HPV testing was performed at the Institute of Infectious Disease and Molecular Medicine, Faculty of Health Sciences, UCT. All other tests were conducted at the Department of Microbiologic Pathology, Faculty of Medicine, University of Limpopo (Medunsa Campus). Specimens were shipped to each laboratory at the end of each workday. The laboratories were masked to the sampling status of the specimens they received. The average time between specimen collection and final processing in the laboratory was 7 days for the UCT Virology laboratory and 2 to 3 days for the Medunsa laboratory. Specimens to be tested for HPV were hand delivered at room temperature to the UCT laboratory and immediately frozen at −20 °C until testing. Specimens in Amplicor CT/NG and Stuart’s transport media were kept cold, and MDM media and slides were kept at room temperature, from collection until processing in the Medunsa laboratory. Upon arrival, Saboraud’s Dextrose (Sabdex) and MDM plates were inoculated and incubated immediately. Sabdex plates were checked for growth after 24 hours, and MDM plates were checked daily for 7 days.
Clinician-obtained and self-obtained specimens were tested using the same laboratory tests as follows: N gonorrhoeae and C trachomatis by COBAS Amplicor CT/NG Test (Roche Molecular Diagnostics), T vaginalis by MDM culture, BV by Nugent scoring of a Gram-stain slide,22Candida species by Sabdex culture, and high-risk HPV types by the Digene HC2 High-Risk HPV DNA Test (Digene Corporation). All women were tested for syphilis using the Immutrep Rapid Plasma Reagin test and the Immutrep Treponema pallidum Haemagglutionation Assay (both by Omega Diagnostics, Alva, Scotland, UK). Blood of women who chose to be tested for HIV was tested with the Abbott HIV 1/2 ELISA (Abbott Diagnostics Division, Wiesbaden-Delkenheim, Germany), positive tests were confirmed with the Capillus HIV 1/2 test (Trinity Biotech, Bray, Ireland), and discrepant results were resolved using the Murex HIV Ag/Ab Combo Assay (Abbott Diagnostics Division).
Statistical Analysis of Quantitative Data
Questionnaire data were summarized as proportions or means and stratified by self-sampling method (tampon versus swabs) and recruitment group (STI services versus non-STI services). Differences between self-sampling methods and recruitment groups were compared using Fisher’s exact or z-tests for unordered categorical variables, Kruskal-Wallis test for ordered categorical variables, and Mann Whitney rank sum test for continuous variables. Sampling performance was determined by calculating sensitivity, specificity, positive predictive values (PPV) and negative predictive values (NPV) for each sampling method and laboratory test. The following gold standards were used in these calculations: for N gonorrhoeae and C trachomatis, COBAS Amplicor test positive on clinician-obtained endocervical swab from the Amplicor CT/NG Specimen Preparation Kit; for T vaginalis, positive MDM culture using clinician-obtained vaginal swab; for BV, a Nugent score of 7 to 10 on the Gram-stain slide that was prepared using a clinician-obtained vaginal swab; for Candida species, a positive Sabdex culture using a clinician-obtained vaginal swab; and for HPV, a positive Digene HC2 test using a clinician-obtained endocervical brush from the Digene HC2 DNA Collection Device. Procedures were considered valid if there was excellent agreement (defined as a κ value of at least 0.7) in test results obtained with clinician-collected and self-collected specimens.
Focus Group Discussions
Nine focus group discussions were held after completion of the clinical procedures to solicit women’s opinions about and experiences with self-sampling and speculum examinations: 3 with women who self-sampled with swabs; 2 with women who self-sampled with tampons; 2 with women who had participated in another study at NY1 center, during which they underwent speculum examinations every month for a year; and 2 with young women who had not participated in any studies. The discussions were held in Xhosa using a discussion guide, were taped, transcribed verbatim, translated into English, and coded manually.
Between January and August 2002, 476 women were assessed for eligibility and 450 were enrolled. All 450 women underwent a clinician-directed speculum examination; 222 self-sampled using vaginal swabs and 228 using a tampon. About half (227) completed the self-sampling procedure first, and the other half (225) started with the speculum examination. One hundred sixty-one women (36%) elected to undergo an HIV test.
Table 1 shows demographic, sexual, and contraceptive behavior and genital hygiene characteristics by self-sampling method and recruitment group. There were no statistically significant differences between the 2 self-sampling groups, but women recruited in the STI group were significantly younger, more likely to have multiple sex partners, their partners were more likely to have multiple partners, and more likely to use tampons and sanitary pads, than women in the non-STI group (Table 1). Participants were on average 29.9 years old and completed 9.5 years of schooling. The majority were married or living with a steady partner, with a mean of 1.4 children. Condom use was low, with 36.6% reporting any condom use in the last year. The most commonly used family planning method was hormonal injectables. The majority of women (73.3%) reported to use sanitary pads for menstrual hygiene and 14.2% tampons.
Prevalence of RTIs
Table 2 shows the prevalence of RTIs (based on clinician-obtained specimens), presumptive diagnoses, and self-reported symptoms by self-sampling method and recruitment group. The HIV prevalence among the 161 women who chose to be tested was 25.5%. There were no statistically significant differences in laboratory-confirmed RTIs, presumptive diagnoses, and self-reported symptoms between the self-sampling groups (Table 2). The order in which vaginal swabs were taken appeared not to have affected the results either (data not shown). Although women seeking STI services were much more likely to self-report symptoms and receive presumptive diagnoses, the only RTIs that they were significantly more likely to have were Candida species and N gonorrhoeae (Table 2). They were significantly less likely to have syphilis.
Self-Sampling Performance by Pathogen
Performance characteristics for each self-sampling method—compared to clinician sampling—by RTI pathogen are shown in Table 3. Sensitivity was 80% or better for BV, C trachomatis, and N gonorrhoeae and close to 80% for Candida species, with no significant differences between the 2 self-sampling methods. Sensitivity for high-risk HPV was good for vaginal swabs (79.8%) and moderate for tampons (59.5%). The worst (and dissatisfactory) performance was found for T vaginalis, with a sensitivity of 54.5% for vaginal swabs and 28.0% for tampons. All tests performed on self-collected specimens demonstrated excellent specificity when compared to clinician-obtained samples (82.6%–100%). Similarly, excellent agreement was found between self-collected samples and clinician-collected samples for all infections listed in Table 4, with the exception of T vaginalis and high-risk HPV. For these 2, tampons demonstrated less agreement than vaginal swabs (Table 4).
Feasibility of Self-Sampling
For each pathogen, only 0% to 0.9% of the clinician-obtained specimens and 0% to 2.4% of the self-obtained specimens were considered invalid. A few of these specimens contained insufficient material to prepare a slide for Nugent scoring (1 clinician-obtained vaginal swab, 3 self-collected vaginal swabs, and 7 tampons); others leaked in transit and could not be used for subsequent testing (4 samples from clinician-obtained swabs, 5 from self-collected swabs, and 6 from tampons). The research nurses observed that 98.9% of the woman did not experience any problems with self-sampling. Only 2 women asked questions during the procedure, and 1 woman had to repeat it because one of her swabs broke. Swabs were inserted an average of 7.4 cm inside the vagina.
Acceptability of Self-Sampling
Table 5 summarizes the quantitative acceptability results. The overall experience with both self-sampling and speculum examinations was rated as good or okay by the majority of women. Women in the swab group and women recruited from STI services rated the self-sampling experience slightly better than the other women (P = 0.048 and 0.032 respectively). Only 3.3% of the women experienced pain during self-sampling, with no differences between the 2 self-sampling groups, but a significantly higher proportion of women in the STI group reporting pain (6.7%) than in the non-STI group (1.7%; P = 0.010). The proportion of women who experienced pain during the speculum examination was much higher, with 24.7% of women in the STI group and 10.4% of women in the non-STI group reporting pain (P <0.001). When asked which sampling method the study participants would choose in the future, about half of them chose self-sampling (50.1%) and the other half a speculum examination (46.1%). Most study participants (63.5%) thought that women should be able to self-sample at home and take the specimen to a clinic. When asked to choose a sampling method again, this time with the home sampling option included, 31.1% chose self-sampling at the clinic, 24.0% self-sampling at home, and 42.4% a speculum examination at the clinic. The most frequently mentioned reason for choosing self-sampling in general was that it is easy and fast, and for self-sampling at home, that it can be done in full privacy in one’s own time and circumvents long waiting times at the clinic. Women who chose clinic sampling said that they would worry about making mistakes without supervision, would want to be able to receive treatment right away, might forget to self-sample or take the specimen to the clinic, or considered the clinic safer and more hygienic than the home environment. The most important reason for preferring a speculum examination was that a clinician can evaluate the woman visually for internal abnormalities while collecting samples. Older women were less likely to choose self-sampling than younger women (OR = 0.97 for each additional year; P = 0.019), but no differences were found by levels of education or parity (data not shown). Women were asked for their sampling preference again when they came back to collect their test results 2 weeks later. Compared to preferences reported at the sampling visit, more women in all groups reported that they would choose self-sampling (Table 5).
The quantitative acceptability findings were confirmed by the focus group discussion results. Most women in all groups reported not to like speculum examinations (because they find it embarrassing to spread their legs in front of a clinician and they find the speculum “cold and painful”), but many admitted that speculum examinations are the only way for a clinician “to see the inside of a woman,” for example, to check for cervical cancer. Women who preferred speculum examinations wanted to be checked “inside” by a qualified clinician, did not trust their own ability to take adequate specimens, and wanted to have an opportunity to ask questions. Women who preferred self-sampling wanted privacy, disliked speculum examinations, disliked the long waiting times at public clinics, and believed that self-sampling is empowering. Women tended to prefer the self-sampling method that they had tried in the study, with swabs being slightly more popular. Some women feared that a tampon may “lodge inside,” but others feared that swabs may break while in the vagina or may be inserted too deep.
A high burden of RTIs was found in both symptomatic and asymptomatic women in Gugulethu, Cape Town, which is consistent with findings from other studies that were conducted in the same area.23 Syndromic management would have resulted in overtreatment of symptomatic women and undertreatment of asymptomatic women, even in this research setting with well-trained and adequately supervised research nurses.24
The sensitivity, specificity, PPV, and NPV values in Table 3 should be interpreted with caution. The gold-standard procedures in this study are not perfect, and infections may have been detected by one sampling method and not the other due to the location of the infection (endocervical, vaginal, and/or urethral). However, the concordance results in Table 4 are consistent with the results in Table 3. We therefore conclude that self-sampling with tampons and swabs resulted in satisfactory validity for N gonorrhoeae, C trachomatis, BV, and Candida species when tested with molecular tests or microscopy. Studies in STI patients in Vienna, women in rural Australia, and military women and adolescents in the United States also showed that self-sampling in combination with molecular tests provided valid results for N gonorrhoeae and C trachomatis.8,10–14 Studies in clinics in the United Kingdom and the United States, a rural community in Uganda, and a hospital in Durban, South Africa, showed that self-sampling in combination with microscopy provided valid results for BV.15–18
For high-risk HPV, self-sampled tampons performed worse than self-sampled swabs (sensitivity was 59.5% and 79.8%, respectively). This is in contrast with results from several other studies using molecular techniques, which showed that tampon-derived samples generally contained more amplifiable DNA than vaginal swabs.19–21 The hypothesis was that this could be due to better absorption capacity of tampons and the fact that tampons touch a larger area within the vagina than a swab. However, in our study tampons were placed in 12 ml PBS, expressed, and 0.5 ml of the expressed fluid was added to 1 ml Digene HC2 DNA Collection Device fluid, whereas swabs were inserted directly into 1 ml Digene fluid. The higher dilution factor for tampons than swabs, the potential direct inhibition by PBS, and/or the longer delay between sampling and insertion in Digene fluid might have negatively influenced our results. We achieved better sensitivity results—and comparable results for tampons and swabs—when a subsample of self-sampled specimens was tested for various HPV types by the Roche Reverse Line-Blot assay (data not shown).
We did not achieve good results for T vaginalis with self-sampling combined with testing by culture: the sensitivity of self-sampling compared to clinician sampling was 54.5% for swabs and 28.0% for tampons. T vaginalis culture could be negatively affected if too much time passes between specimen collection and culture medium inoculation. However, in our study, not more than 20 minutes passed between specimen collection and transport medium inoculation. More likely explanations are that self-sampling and dilution in PBS before medium inoculation may have produced a limited number of organisms in the specimen, or PBS may have had a direct inhibitory effect on T vaginalis. Both hypotheses would also explain why tampons performed worse than swabs: swabs were inserted in only 3 ml and tampons in 12 ml PBS. We believe that the use of self-sampling to test for T vaginalis should not be abandoned based on our results. Other studies have found satisfactory results for T vaginalis when self-sampled specimens were tested by molecular tests10–12,14 or by InPouch culture system (Biomed Diagnostics, San Jose, CA).16
In this study, the self-sampling procedures carried out by the participant herself were limited in scope, but they nonetheless enabled clinic staff to evaluate women for RTIs in the absence of a pelvic examination. Both self-sampling methods were considered as acceptable, if not more acceptable, than speculum examinations. Different women found different aspects of self-sampling and speculum examinations desirable, and no one refused to undergo a particular sampling method. Many women mentioned that speculum examinations are unpleasant but would always be needed to some extent, because, unlike self-sampling, they allow the clinician to view the vagina and cervix.
We conclude that self-sampling should not replace speculum examinations in all circumstances but should be explored further (possibly in combination with self-testing using simple dipstick tests) as a strategy to improve screening for RTIs in resource-poor settings. Self-sampling and self-testing (in a clinic setting but also at home) may be particularly useful for young women who do not yet need cervical cancer screening, who are at high risk for RTIs, but may not seek RTI screening to avoid having to go to a clinic or undergo a speculum examination. Self-sampling could also replace some speculum examinations in large HIV-prevention intervention studies or other studies requiring frequent testing for RTIs to improve study participation and cohort retention, simplify trial logistics, and reduce costs.
1. World Health Organization. Global Prevalence and Incidence of Selected Curable Sexually Transmitted Infections: Overview and Estimates. Geneva, Switzerland: WHO, 2001.
2. Joint United Nations Programme on HIV/AIDS. Report on the Global HIV/AIDS Epidemic 2004. Geneva, Switzerland: UNAIDS, 2004.
3. Way PO, Schwartländer B, Piot P. The global epidemiology of HIV and AIDS. In: Holmes KK, Sparling PF, Mardh PA, et al., eds. Sexually Transmitted Diseases. New York, NY: McGraw Hill, 1999:77–94.
4. Westrom L, Eschenbach D. Pelvic inflammatory disease. In: Holmes KK, Sparling PF, Mardh PA, et al., eds. Sexually Transmitted Diseases. New York, NY: McGraw Hill, 1999:783–809.
5. Cates W Jr, Brunham RC. Sexually transmitted diseases and infertility. In: Holmes KK, Sparling PF, Mardh PA, et al., eds. Sexually Transmitted Diseases. New York, NY: McGraw Hill, 1999:1079–1087.
6. South African Department of Health. Protocols for the Management of a Person with a Sexually Transmitted Disease, According to the Essential Drugs List. Pretoria, South Africa: South African Department of Health, 1998.
7. Pettifor A, Walsh J, Wilkins V, et al. How effective is syndromic management of STDs? a review of current studies. Sex Transm Dis 2000; 27:371–385.
8. Stary A, Najim B, Lee H. Vulval swabs as alternative specimens for ligase chain reaction detection of chlamydial infection in women. J Clin Microbiol 1997; 35:836–838.
9. Black CM, Marazzo J, Johnson RE, et al. Head-to-head multicenter comparison of DNA probe and nucleic acid amplification tests for Chlamydia trachomatis
in women with use of an improved reference standard. J Clin Microbiol 2002; 40:3757–3763.
10. Tabrizi SN, Paterson BA, Fairley CK, et al. Comparison of tampon and urine as self-administered methods of specimen collection in the detection of Chlamydia trachomatis, Neisseria gonorrhoeae
and Trichomonas vaginalis
in women. Int J STD AIDS 1998; 9:347–349.
11. Knox J, Tabrizi SN, Miller P, et al. Evaluation of self-collected samples in contrast to practitioner-collected samples for detection of Chlamydia trachomatis, Neisseria gonorrhoeae
, and Trichomonas vaginalis
by polymerase chain reaction among women living in remote areas. Sex Transm Dis 2002; 29:647–654.
12. Garrow SC, Smith DW, Harnett GB. The diagnosis of chlamydia, gonorrhoea, and trichomonas infections by self-obtained low vaginal swabs in a remote northern Australia clinical practice. Sex Transm Inf 2002; 78:278–281.
13. Gaydos CA, Crotchfelt KA, Shah N, et al. Evaluation of dry and wet transported intravaginal swabs in detection of Chlamydia trachomatis
and Neisseria gonorrhoeae
infections in female soldiers by PCR. J Clin Microbiol 2002; 40:758–761.
14. Wiesenfeld HC, Lowry DLB, Heine RP, et al. Self-collection of vaginal swabs for the detection of chlamydia, gonorrhea and trichomoniasis. Sex Transm Dis 2001; 28:321–325.
15. Morgan DJ, Aboud CJ, McCaffrey IMB, et al. Comparison of gram-stained smears prepared from blind vaginal swabs with those obtained at speculum examination for the assessment of vaginal flora. Br J Obstet Gynaecol 1996; 103:1105–1108.
16. Wawer MJ, Gray RH, Sewankambo NK, et al. A randomized, community trial of intensive sexually transmitted disease control for AIDS prevention, Rakai, Uganda. AIDS 1998; 12:1211–1225.
17. Schwebke JR, Morgan SC, Weiss HL. The use of sequential self-obtained vaginal smears for detecting changes in the vaginal flora. Sex Transm Dis 1997; 34:236–239.
18. Sturm PDJ, Moodley P, Nzimande G, et al. Diagnosis of bacterial vaginosis on self-collected vaginal tampon specimens. Int J STD AIDS 2002; 13:559–563.
19. Wright TC, Denny L, Kuhn L, et al. HPV DNA testing of self-collected vaginal samples to detect cervical cancer. JAMA 2000; 283:81–86.
20. Fairley CK, Chen S, Tabrizi SN, et al. Tampons: a novel patient-administered method for the assessment of genital human papillomavirus infection. J Infect Dis 1992; 165:1103–1106.
21. Harper DM, Noll WW, Bellonis DR, et al. Randomized clinical trial of PCR-determined human papillomavirus detection methods: self-sampling versus clinician-directed biologic concordance and women’s preferences. Am J Obstet Gynecol 2002; 186:365–373.
22. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosis of bacterial vaginosis is improved by a standardized method of Gram stain interpretation. J Clin Microbiol 1991; 29:297–301.
23. Coetzee N, Hoosen A, Blanchard K, et al. Baseline STI prevalence: Carraguard vaginal microbicide safety trial. Poster presentation at the 14th
Biennual Congress of the International Society for Sexually Transmitted Diseases Research, Berlin, Germany, June 2001.
24. Young T, Altini L, Jones H, et al. An evaluation of the quality of RTI case management within a research setting. Poster presentation at the 15th
Biennual Congress of the International Society for Sexually Transmitted Diseases Research, Ottawa, Canada, July 2003.