Abstract
Research on the determinants of condom use and condom non-use generally has relied on self-reported data with questionable validity. We identified predictors of recent, unprotected sex among 331 female sex workers in Madagascar using two outcome measures: self-reports of unprotected sex within the past 48 h and detection of prostate-specific antigen (PSA), a biological marker of recent semen exposure. Multivariable logistic regression revealed that self-reported unprotected sex was associated with three factors: younger age, having a sipa (emotional partner) in the prior seven days, and no current use of hormonal contraception. The sole factor related to having PSA detected was prevalent chlamydial infection (adjusted odds ratio, 4.5; 95% confidence interval, 2.0–10.1). Differences in predictors identified suggest that determinants of unprotected sex, based on self-reported behaviors, might not correlate well with risk of semen exposure. Caution must be taken when interpreting self-reported sexual behavior measures or when adjusting for them in analyses evaluating interventions for the prevention of HIV/STIs.
Keywords: Female sex workers, Condoms, Biological markers, Prostate-specific antigen, Africa
Introduction
Promoting the use of condoms among high-risk groups, including female sex workers (FSWs), is considered a cornerstone of controlling the HIV pandemic [1–3]. Accordingly, a vast body of research has been conducted on the determinants of condom use and non-use among FSWs. Condom use, among FSWs in Africa, has been associated with age, partner type, duration of sex work, number of clients, partner violence, prior HIV testing, and access to condoms [4–9]. Engaging in unprotected sex has been linked to binge drinking, charging more for the act, pregnancy intentions, knowing other women who have unprotected sex, older age of the partner, and sex at home [10–14].
Studies generally have identified determinants of condom use and condom non-use with self-reported data, which have questionable validity [15–19]. The detection of prostate-specific antigen (PSA) in vaginal fluid is a marker of exposure to semen within the previous 48 h [20–23]. Research suggests that almost all vaginal-swab specimens (97%) will be negative for PSA (defined as ≤1 ng of PSA/ml eluate) by 48 h after exposure to semen [23]. Thus, the detection of PSA can serve as a measure of recent sex unprotected by a condom or sex in which a condom either failed or was used incorrectly. We identified predictors of recent, unprotected sex (measured by both self-reports and PSA detection) and evaluated whether the associations between factors and the two measures of unprotected sex differed.
Methods
Study Procedures
The study was conducted among FSWs in two sites in Madagascar participating in an 18-month, randomized controlled trial of the effect of supplementing community-based promotion of male and female condoms with clinic-based counseling [24, 25]. Participants attending their last (18-month) study visit between December 2002 and May 2003 were asked to participate in this additional research, which consisted of answering eight questions about sex and condom use and having two vaginal swabs collected to test for PSA. Specifically, we asked participants about any sex acts (yes or no) with any clients, partners, or boyfriends during four timeframes: the previous 24 h, 48 h, 7 days and 14 days. For each timeframe, women also were asked to report whether any of the acts were unprotected (yes or no). As part of the main study visit on that same day, the participants also were administered a questionnaire, which included questions on potential correlates of unprotected sex (see below). Both questionnaires were administered by female study clinicians in a private setting within the clinic.
The same female clinician inserted a cotton-tipped swab into the posterior fornix and rotated it four times to collect the specimens for testing for PSA. After several hours of air-drying, staff sealed the swabs in individual transport tubes and froze them until data collection ended. Specimens were shipped on dry ice to the research laboratory at the University of North Carolina for processing. Only participants who gave verbal consent for the additional research were included. The consent process included an explanation that the purpose of the additional research was to improve ways of measuring when women have unprotected sex. Ethical review boards at Family Health International and the Madagascar Ministry of Health approved the additional research. All names and personal identifiers were removed before researchers at the Centers for Disease Control and Prevention (CDC) received the data, and the CDC institutional review board determined that the present analysis of anonymous data was exempt from review.
Laboratory Methods
We used Abbot IMx microparticle enzyme immunoassay (Abbott Laboratories, Abbott Park, IL) for detecting PSA in the vaginal swab eluates. After each vaginal swab was inserted into a 1.5 ml microcentrifuge tube, with 1 ml of phosphate-buffered saline, the contents were allowed to elute at room temperature for 5–15 min. We vigorously rotated individual specimens and pressed the swabs against the side of the tube to maximize the recovery of the sample. Capped tubes then were centrifuged at 13,000×g for 5 min before 250 µl of the resulting supernatant was dispensed into an IMx reaction cell. We did not perform the dilution and retesting necessary for quantifying concentrations above 100 ng PSA/ml. To have a threshold consistent with that of Macaluso and colleagues (who used three times the volume for eluting vaginal swab specimens), we defined a positive result as >3 ng PSA/ml vaginal swab eluate [23].
Methods for diagnosing sexually transmitted infections in this study have been reported elsewhere [24, 25]. Gonorrhea and chlamydial infections were diagnosed with ligase chain reaction testing (Abbott LCx Probe System, Abbott Laboratories, Abbott Park, IL, USA), and trichomoniasis was diagnosed by culture using InPouch (Bio-Med, White City, OR, USA).
Statistical Analysis
We assessed predictors of engaging in recent unprotected sex, using two outcome measures: self-reports of having had unprotected sex within the previous 48 h and the detection of PSA in vaginal fluid. Potential correlates included randomization group (peer-only counseling vs. peer and clinic counseling); study site (Tamatave vs. Antananarivo); age (highest quartile [≥35 years] vs. younger age); location from which clients were typically recruited (street only vs. bar, market, other); single without a steady partner (yes vs. no); current hormonal contraception use (yes vs. no); prevalent chlamydial infection (yes vs. no); prevalent gonorrhea infection (yes vs. no); prevalent trichomoniasis infection (yes vs. no); number of sexual partners in previous seven days (0–3 vs. ≥4 [median, 4]); sipas (emotional, non-client partner) in previous seven days (0 vs. ≥1); number of clients in previous seven days (0, 1–4 vs. ≥5 [median among those with any clients in previous seven days, 5]); when using condoms with clients, FSW usually suggests their use (yes vs. no); when using condoms with clients, FSW usually decides between male and female condoms (yes vs. no); symptoms of STI since last study visit (yes vs. no); told partner to get STI treatment since last study visit (yes vs. no); reported talking about male and female condoms with clinic provider (yes vs. no); and reported talking about male and female condoms with peer educator (yes vs. no).
We simultaneously fit models for the two outcome measures using bivariate logistic regression [26]. The dataset was augmented to have two records for each FSW (i.e., one for each outcome measure). Generalized estimating equations (using a logit link function and unstructured working correlation matrix) were applied to control for the correlation between outcome measures from the same woman. We chose this approach because assessing whether two models differ requires a direct comparison of the models; it is generally insufficient to declare that a difference exists simply because certain factors are significant in one model but not in the other. We first performed bivariable analyses by fitting individual models with the outcome type (self-report or PSA test results), the potential predictor, and the two-way interaction between the outcome type and the potential predictor. We then constructed a full initial model, which included parameters for outcome type, all potential predictors, and the two-way interactions between outcome type and each potential predictor. We reduced the full model using manual backwards elimination, in which variables that were not significant at the 0.05 level were removed; variables were only removed if they were non-significant for both outcomes. We report the P-values for contrasts, testing the differences in the associations between each potential predictor and the two outcomes. All analyses were performed by using SAS, version 9.1 (SAS Institute, Cary, NC).
Results
Of the 347 women who were recruited to participate in this additional research at their last study visit, all consented to participate. The analysis is restricted to the 331 women who provided both outcome measures (i.e., self-report and PSA data). The mean age of FSWs was 29 years (range, 17–52). Most FSWs (81%) solicited their clients on the streets, and few used oral contraception (6%) or injectable contraception (13%) for pregnancy prevention (Table 1). At least one breakage or slippage of the male or female condom during sex in the previous 30 days was reported by 8.5% of FSWs. Condom malfunctions generally were reported for the use of male condoms rather than female condoms (data not shown). Overall, 29.0% (n = 96) of FSWs reported unprotected sex in the previous 48 h and 38.1% (n = 126) had PSA detected in their specimens.
Table 1.
Characteristics of female sex workers in Madagascar (N = 331)
| Characteristics | N (%) |
|---|---|
| Randomization group | |
| Peer-only counseling | 165 (49.9) |
| Peer counseling and clinic counseling | 166 (50.2) |
| Study site | |
| Antananarivo | 166 (50.2) |
| Tamatave | 165 (49.9) |
| Age | |
| 17–34 years | 246 (74.3) |
| ≥35 years | 85 (25.7) |
| Location where recruit clients | |
| Street only | 268 (81.0) |
| Bar, market or other | 63 (19.0) |
| Single without steady partner | |
| Yes | 117 (35.4) |
| No | 214 (64.7) |
| Current hormonal contraception use | |
| Yes | 61 (18.4) |
| No | 270 (81.6) |
| Prevalent chlamydial infection | |
| Yes | 31 (9.4) |
| No | 300 (90.6) |
| Prevalent gonorrhea | |
| Yes | 40 (12.1) |
| No | 291 (87.9) |
| Prevalent trichomoniasis | |
| Yes | 35 (10.6) |
| No | 296 (89.4) |
| Sexual partners in previous 7 days | |
| 0–3 | 149 (45.0) |
| ≥4 | 182 (55.0) |
| Sipasa in previous 7 days | |
| 0 | 189 (57.6) |
| ≥1 | 139 (42.4) |
| Clients in previous 7 days | |
| 0 | 48 (16.3) |
| 1–4 | 111 (37.8) |
| ≥5 | 135 (45.9) |
| When using condoms with clients, FSW usually suggests their use | |
| Yes | 119 (36.2) |
| No | 210 (63.8) |
| When using condoms with clients, FSW usually decides between male and female condoms | |
| Yes | 83 (25.3) |
| No | 245 (74.7) |
| Symptoms of STI since last study visit | |
| Yes | 63 (19.0) |
| No | 268 (81.0) |
| Told partner to get STI treatment since last study visit | |
| Yes | 235 (71.0) |
| No | 96 (29.0) |
| Reported talking about male and female condoms with clinic provider | |
| Yes | 212 (64.1) |
| No | 119 (36.0) |
| Reported talking about male and female condoms with peer educator | |
| Yes | 310 (93.7) |
| No | 21 (6.3) |
STI sexually transmitted infections
Sipas are emotional, non-client partners
In the bivariable analysis, we found two predictors of both self-reported unprotected sex and having PSA detected in vaginal fluid: (1) having more clients in the previous seven days and (2) when using condoms with clients, FSW usually decides between male and female condoms (Table 2). Self-reported unprotected sex also was associated with seven factors that were not related to having PSA detected: study site, age, having more sexual partners in the previous seven days, having more than one sipa in the previous seven days, when using condoms with clients, FSW usually suggests their use, being other than single without a steady partner, and not having talked about male and female condoms with a provider at the clinic. The only factor that was associated with PSA, but not associated with self-reported data, was prevalent chlamydial infection.
Table 2.
Bivariable analysis of factors associated with recent, unprotected sex among female sex workers in Madagascar
| Predictors | Self-reports |
Detection of PSA |
P-valuea | ||
|---|---|---|---|---|---|
| OR | 95% CI | OR | 95% CI | ||
| Randomization group | |||||
| Peer-only counseling | 1.3 | 0.8–2.1 | 0.8 | 0.5– 1.3 | 0.12 |
| Peer counseling and clinic counseling | Referent | Referent | |||
| Study site | |||||
| Antananarivo | 1.9b | 1.2–3.2 | 1.4 | 0.9– 2.2 | 0.24 |
| Tamatave | Referent | Referent | |||
| Age | |||||
| 17–34 years | 1.9b | 1.0–3.5 | 1.2 | 0.7–2.0 | 0.15 |
| ≥35 years | Referent | Referent | |||
| Location where recruit clients | |||||
| Street only | 1.5 | 0.8–2.9 | 1.3 | 0.7–2.3 | 0.26 |
| Bar–market or other | Referent | Referent | |||
| Single without steady partner | |||||
| Yes | 0.5b | 0.3–0.8 | 1.0 | 0.6–1.5 | 0.02 |
| No | Referent | Referent | |||
| Current hormonal contraception use | |||||
| Yes | 0.7 | 0.4–1.3 | 0.8 | 0.4–1.4 | 0.78 |
| No | Referent | Referent | |||
| Prevalent chlamydial infection | |||||
| Yes | 1.4 | 0.6–3.0 | 3.9b | 1.8–8.6 | 0.04 |
| No | Referent | Referent | |||
| Prevalent gonorrhea | |||||
| Yes | 0.9 | 0.4–1.9 | 1.6 | 0.8–3.0 | 0.27 |
| No | Referent | Referent | |||
| Prevalent trichomoniasis | |||||
| Yes | 0.7 | 0.3–1.6 | 0.6 | 0.3–1.3 | 0.79 |
| No | Referent | Referent | |||
| Sexual partners in previous 7 days | |||||
| 0–3 | Referent | Referent | 0.26 | ||
| ≥4 | 2.1b | 1.3–3.5 | 1.5 | 1.0–2.4 | |
| Sipasc in previous 7 days | |||||
| 0 | Referent | Referent | <0.01 | ||
| ≥1 | 4.5b | 2.7–7.5 | 1.5 | 1.0–2.4 | |
| Clients in previous 7 days | |||||
| 0 | Referent | Referent | |||
| 1–4 | 2.3 | 1.0–5.4 | 2.3b | 1.1–4.8 | 0.97 |
| ≥5 | 2.5b | 1.1–5.8 | 2.1 | 1.0–4.3 | 0.70 |
| When use condoms with clients, FSW usually suggests their use | |||||
| Yes | 1.8b | 1.1–2.9 | 1.4 | 0.9–2.3 | 0.54 |
| No | Referent | Referent | |||
| When use condoms with clients, FSW usually decides between male and female condoms | |||||
| Yes | 2.1b | 1.2–3.5 | 1.6b | 1.0–2.7 | 0.52 |
| No | Referent | Referent | |||
| Symptoms of STI since last study visit | |||||
| Yes | 1.7 | 0.9–3.0 | 1.1 | 0.6–1.9 | 0.26 |
| No | Referent | Referent | |||
| Told partner to get STI treatment since last study visit | |||||
| Yes | 1.7 | 1.0–2.9 | 1.0 | 0.6–1.7 | 0.11 |
| No | Referent | Referent | |||
| Reported talking about male and female condoms with clinic provider | |||||
| Yes | 0.6b | 0.4–1.0 | 1.0 | 0.6–1.6 | 0.06 |
| No | Referent | Referent | |||
| Reported talking about male and female condoms with peer educator | |||||
| Yes | 0.5 | 0.2–1.3 | 1.0 | 0.4–2.5 | 0.08 |
| No | Referent | Referent | |||
CI confidence interval, FSW female sex worker, OR odds ratio, PSA prostate-specific antigen, STI sexually transmitted infection
Recent, unprotected sex measured by self-reports of unprotected sex within the previous 48 h or by detection of PSA in vaginal fluid
P-value for difference in the associations between the factor and the two outcomes for recent unprotected sex
P-value < 0.05
Sipas are emotional, non-client partners
In the multivariable analysis, three factors were associated with self-reported unprotected sex (Table 3). Younger women had a greater odds of self-reported unprotected sex compared to women 35 years of age or older (adjusted OR [aOR], 2.1; 95% CI 1.1–4.0). Current hormonal contraception use was protective against unprotected sex (aOR, 0.4; 95% CI 0.2–0.9). Finally, women who reported having one or more sipas in the previous seven days, had an odds of reporting unprotected sex 4.8 times (95% CI 2.9–8.1) that of women without a sipa. The sole predictor of having PSA detected was having a prevalent chlamydial infection (aOR, 4.5; 95% CI 2.0–10.1). Only the association with having one or more sipas in the previous seven days was significantly different between the models for the two outcomes (P-value <0.01).
Table 3.
Multivariable analysis of factors associated with recent unprotected sex among female sex workers in Madagascar
| Predictors | Self-reports |
Detection of PSA |
P-valuea | ||
|---|---|---|---|---|---|
| aOR | 95% CI | aOR | 95% CI | ||
| Age in years | |||||
| 17–34 | 2.1b | 1.1–4.0 | 1.2 | 0.7–2.0 | 0.11 |
| ≥35 | Referent | Referent | |||
| Current hormonal contraception use | |||||
| Yes | 0.4b | 0.2–0.9 | 0.6 | 0.3–1.1 | 0.54 |
| No | Referent | Referent | |||
| Prevalent chlamydial infection | |||||
| Yes | 1.7 | 0.8–3.9 | 4.5b | 2.0–10.1 | 0.07 |
| No | Referent | Referent | |||
| Sipasc in previous 7 days | |||||
| ≥1 | 4.8b | 2.9–8.1 | 1.6 | 1.0–2.5 | <0.01 |
| None | Referent | Referent | |||
OR adjusted odds ratio, CI confidence interval, PSA prostate-specific antigen
Recent, unprotected sex measured by self-reports of unprotected sex within the previous 48 h or by detection of PSA in vaginal fluid
P-value for difference in the associations between the factor and the two outcomes for recent unprotected sex
P-value < 0.05
Sipas are emotional, non-client partners
As a sensitivity analysis, we repeated the analysis after defining a positive PSA result as 15 ng/ml vaginal swab eluate. The use of this higher threshold had a negligible effect on the effect estimates and did not change any of the conclusions from the multivariable analysis.
Discussion
We found different predictors of unprotected sex in the multivariable analysis, depending on the outcome measure used. Self-reports of recent, unprotected sex were associated with younger age, no hormonal contraception use, and having a sipa in the previous seven days. In contrast, recent, unprotected sex, measured with PSA detection, was associated only with having a prevalent chlamydial infection. Prior research conducted among FSWs and non-FSWs in four West African countries demonstrated an association between PSA detection and prevalent gonorrhea and chlamydial infections and Mycoplasma genitalium [27].
With a few exceptions, we measured predictors using self-reported data, for which the validity could not be assessed. Interestingly, the sole predictor of having PSA detected, chlamydial infection, was based on an objective, laboratory test. The association between self-reported condom use and a reduction in STI risk has not been consistently demonstrated [28]. However, the present findings suggest that the fallibility of self-reported data could have been responsible for the lack of correlation in prior studies.
The main strength of this research involves the use of PSA to measure recent unprotected sex. To our knowledge, this is the first study to evaluate factors associated with unprotected sex among FSWs by using both self-reports and a biological measure of semen exposure. Previous research has highlighted potential problems of relying on self-reported data when measuring coitus and condom use [29–31]. For example, in a substudy conducted among 135 women participating in a trial evaluating a candidate microbicide, 78% admitted in follow-up audio computer-assisted self-interviews (ACASI) that they had previously misreported behaviors at least once during face-to-face interviews for the larger study [19]. Notably, more women described past over-reporting than under-reporting of unprotected sex. If the data from the ACASI are valid, these findings suggest that we cannot even assume that bias will always be in the direction of under-reporting of risky behaviors.
Studies have detected apparent under-reporting of unprotected sex by comparing self-reports to the detection of PSA. A previous report, based on the present study, demonstrated that 29% of the FSWs who reported not engaging in unprotected sex within the previous 48 h, had PSA detected [16]. Similar studies have found PSA in specimens of 36% of FSWs in Guinea, 11% of FSWs in Kenya, and 15% of women in Zimbabwe, who denied recent unprotected sex [17, 18, 32]. Because PSA immediately begins to clear from the vagina following exposure (with 71% of women testing negative for PSA at 24 h after semen exposure) [23], these estimates should be interpreted as measuring the lower bound of under-reporting of unprotected sex. On the other hand, because of the decay period for PSA, the use of the biomarker likely underestimates the occurrence of unprotected sex in the previous 48 h. Thus, differences in the predictors of the two methods of measuring unprotected sex might be explained, at least in part, by this clearance time.
Although the variables that were significantly associated with the two outcomes differed, the only variable with a significant difference in predictive value was having a sipa in the previous seven days. With a larger sample size, additional variables might have emerged in the multivariable analysis as significant predictors for each of the individual outcomes. Thus, we cannot rule out the possibility that the predictors of the two outcomes might have shown more agreement—had the study been larger. Another study limitation is that if the level of condom use was very low, and if most FSWs were engaging in unprotected sex, our findings might demonstrate predictors of the frequency of sex rather than predictors of unprotected sex per se. Finally, because the questionnaires did not collect data on condom breakage or slippage in the previous 48 h, we cannot assess whether PSA detection was the result of a condom malfunction. However, given the relatively small proportion of women who reported malfunctions in the prior 30 days, the occurrence of these events within the previous 48 h might have been rare. Furthermore, the results of the sensitivity analysis suggest that the differences in the two outcomes are not explained by low levels of semen exposure.
The current study was conducted among FSWs in Madagascar who had participated in an 18-month study of condom promotion. Furthermore, the validity of participant responses could be dependent on a range of factors (e.g., rapport with the interviewers, comprehension of the questionnaire, order of questions or order of visit procedures). Thus, these results might not be generalizable to other populations or even to the same population when studied in other contexts.
The differences in predictors of the two outcomes identified in the present study, suggest that determinants of unprotected sex or condom use on the basis of self-reported data might not correlate well with the risk of semen exposure. The variables significantly associated with self-reported unprotected sex might be more appropriately interpreted as predictors of the reporting behavior, rather than the behavior of interest (i.e., unprotected sex). Thus, we cannot rule out that prior studies purporting to identify determinants of condom use or non-use, instead, might have identified correlates of the reporting of these behaviors. Given the substantial apparent misreporting in measures of unprotected sex, the use of more objective outcomes is crucial for understanding behaviors and other factors related to HIV/STI risk. In the meantime, we must take caution when interpreting self-reported sexual behavior measures or when adjusting for them in multivariable analyses evaluating interventions for the prevention of HIV/STIs.
Acknowledgments
The Microbiology Core Laboratory of the North Carolina Sexually Transmitted Infections/Topical Microbicide Cooperative Research Center, funded by U.S. National Institutes of Health (NIH) contract U19 AI031496, and the Sexually Transmitted Diseases Clinical Trial Unit, funded by U.S. NIH contract N01 AI075329, supported this research. The U.S. Agency for International Development (cooperative agreement #AID/CCP-3079-A-00-5022-00) also supported this work.
Footnotes
Disclaimer The findings and conclusions in this report are those of the authors and do not necessarily represent official views of the Centers for Disease Control and Prevention.
Contributor Information
Maria F. Gallo, Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA, mgallo@cdc.gov
Markus J. Steiner, Family Health International, Research Triangle Park, NC, USA
Marcia M. Hobbs, Departments of Medicine and Microbiology & Immunology at the University of North Carolina
Mark A. Weaver, Family Health International, Research Triangle Park, NC, USA
Theresa Hatzell Hoke, Family Health International, Research Triangle Park, NC, USA.
Kathleen Van Damme, Department of Medicine, University of North Carolina, Chapel Hill, Chapel Hill, NC, USA.
Denise J. Jamieson, Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA
Maurizio Macaluso, Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA.
References
- 1.Lowndes CM, Alary M, Meda H, et al. Role of core and bridging groups in the transmission dynamics of HIV and STIs in Cotonou, Benin, West Africa. Sex Transm Infect. 2002;78(Suppl 1):169–177. doi: 10.1136/sti.78.suppl_1.i69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mayaud P, Mabey D. Approaches to the control of sexually transmitted infections in developing countries: old problems and modern challenges. Sex Transm Infect. 2004;80(3):174–182. doi: 10.1136/sti.2002.004101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Plummer FA, Nagelkerke NJ, Moses S, Ndinya-Achola JO, Bwayo J, Ngugi E. The importance of core groups in the epidemiology and control of HIV-1 infection. AIDS. 1991;5(Suppl1):S169–S176. [PubMed] [Google Scholar]
- 4.Abdool Karim Q, Abdool Karim SS, Soldan K, Zondi M. Reducing the risk of HIV infection among South African sexworkers: socioeconomic and gender barriers. Am J Public Health. 1995;85(11):1521–1525. doi: 10.2105/ajph.85.11.1521. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kayembe PK, Mapatano MA, Busangu AF, et al. Determinants of consistent condom use among female commercial sex workers in the Democratic Republic of Congo: implications for interventions. Sex Transm Infect. 2008;84(3):202–206. doi: 10.1136/sti.2007.028324. [DOI] [PubMed] [Google Scholar]
- 6.Morris CN, Morris SR, Ferguson AG. Sexual behavior of female sex workers and access to condoms in Kenya and Uganda on the trans-Africa highway. AIDS and Behav. 2009;13(5):860–865. doi: 10.1007/s10461-008-9431-z. [DOI] [PubMed] [Google Scholar]
- 7.Stoebenau K, Hindin MJ, Nathanson CA, Rakotoarison PG, Razafintsalama V. “…But then he became my sipa”: the implications of relationship fluidity for condom use among women sex workers in Antananarivo, Madagascar. Am J Public Health. 2009;99(5):811–819. doi: 10.2105/AJPH.2007.118422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Voeten HA, Egesah OB, Varkevisser CM, Habbema JD. Female sex workers and unsafe sex in urban and rural Nyanza, Kenya: regular partners may contribute more to HIV transmission than clients. Trop Med Int Health. 2007;12(2):174–182. doi: 10.1111/j.1365-3156.2006.01776.x. [DOI] [PubMed] [Google Scholar]
- 9.Wang C, Hawes SE, Gaye A, et al. HIV prevalence, previous HIV testing, and condom use with clients and regular partners among Senegalese commercial sex workers. Sex Transm Infect. 2007;83(7):534–540. doi: 10.1136/sti.2007.027151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chersich MF, Luchters SM, Malonza IM, Mwarogo P, King’ola N, Temmerman M. Heavy episodic drinking among Kenyan female sex workers is associated with unsafe sex, sexual violence and sexually transmitted infections. Int J STD AIDS. 2007;18(11):764–769. doi: 10.1258/095646207782212342. [DOI] [PubMed] [Google Scholar]
- 11.Ntumbanzondo M, Dubrow R, Niccolai LM, Mwandagalirwa K, Merson MH. Unprotected intercourse for extra money among commercial sex workers in Kinshasa, Democratic Republic of Congo. AIDS Care. 2006;18(7):777–785. doi: 10.1080/09540120500412824. [DOI] [PubMed] [Google Scholar]
- 12.Moore AR, Oppong J. Sexual risk behavior among people living with HIV/AIDS in Togo. Soc Sci Med. 2007;64(5):1057–1066. doi: 10.1016/j.socscimed.2006.10.004. [DOI] [PubMed] [Google Scholar]
- 13.Tassiopoulos K, Kapiga S, Sam N, Ao TT, Hughes M, Seage GR., 3rd A case-crossover analysis of predictors of condom use by female bar and hotel workers in Moshi, Tanzania. Int J Epidemiol. 2009;38(2):552–560. doi: 10.1093/ije/dyn358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Yadav G, Saskin R, Ngugi E, et al. Associations of sexual risk taking among Kenyan female sex workers after enrollment in an HIV-1 prevention trial. J Acquir Immun Defic Syndr. 2005;38(3):329–334. [PubMed] [Google Scholar]
- 15.Chen MP, Macaluso M, Blackwell R, et al. Self-reported mechanical problems during condom use and semen exposure. Comparison of two randomized trials in the United States of America and Brazil. Sex Transm Dis. 2007;34(8):557–562. doi: 10.1097/01.olq.0000258487.38309.b9. [DOI] [PubMed] [Google Scholar]
- 16.Gallo MF, Behets FM, Steiner MJ, et al. Prostate-specific antigen to ascertain reliability of self-reported coital exposure to semen. Sex Transm Dis. 2006;33(8):476–479. doi: 10.1097/01.olq.0000231960.92850.75. [DOI] [PubMed] [Google Scholar]
- 17.Gallo MF, Behets FM, Steiner MJ, et al. Validity of self-reported ‘safe sex’ among female sex workers in Mombasa, Kenya—PSA analysis. Int J STD AIDS. 2007;18(1):33–38. doi: 10.1258/095646207779949899. [DOI] [PubMed] [Google Scholar]
- 18.Minnis AM, Steiner MJ, Gallo MF, et al. Biomarker validation of reports of recent sexual activity: results of a randomized controlled study in Zimbabwe. Am J Epidemiol. 2009;170(7):918–924. doi: 10.1093/aje/kwp219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Turner AN, De Kock AE, Meehan-Ritter A, et al. Many vaginal microbicide trial participants acknowledged they had misreported sensitive sexual behavior in face-to-face interviews. J Clin Epidemiol. 2009;62(7):759–765. doi: 10.1016/j.jclinepi.2008.07.011. [DOI] [PubMed] [Google Scholar]
- 20.Graves HC, Sensabaugh GF, Blake ET. Postcoital detection of a male-specific semen protein. Application to the investigation of rape. N Engl J Med. 1985;312(6):338–343. doi: 10.1056/NEJM198502073120603. [DOI] [PubMed] [Google Scholar]
- 21.Kamenev L, Leclercq M, Francois-Gerard C. An enzyme immunoassay for prostate-specific p30 antigen detection in the postcoital vaginal tract. J Forensic Sci Soc. 1989;29(4):233–241. doi: 10.1016/s0015-7368(89)73257-6. [DOI] [PubMed] [Google Scholar]
- 22.Lawson ML, Maculuso M, Bloom A, Hortin G, Hammond KR, Blackwell R. Objective markers of condom failure. Sex Transm Dis. 1998;25(8):427–432. doi: 10.1097/00007435-199809000-00009. [DOI] [PubMed] [Google Scholar]
- 23.Macaluso M, Lawson L, Akers R, et al. Prostate-specific antigen in vaginal fluid as a biologic marker of condom failure. Contraception. 1999;59(3):195–201. doi: 10.1016/s0010-7824(99)00013-x. [DOI] [PubMed] [Google Scholar]
- 24.Feldblum PJ, Hatzell T, Van Damme K, Nasution M, Rasamindrakotroka A, Grey TW. Results of a randomized trial of male condom promotion among Madagascar sex workers. Sex Transm Infect. 2005;81(2):166–173. doi: 10.1136/sti.2004.010074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hoke TH, Feldblum PJ, Damme KV, et al. Randomised controlled trial of alternative male and female condom promotion strategies targeting sex workers in Madagascar. Sex Transm Infect. 2007;83(6):448–453. doi: 10.1136/sti.2006.024612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Fitzmaurice GM, Laird NM, Zahner GEP, Daskalakis C. Bivariate logistic regression analysis of childhood psychopathology rating using multiple informants. Am J Epidemiol. 1995;142(11):1194–1203. doi: 10.1093/oxfordjournals.aje.a117578. [DOI] [PubMed] [Google Scholar]
- 27.Pépin J, Fink GD, Khonde N, et al. Improving second-generation surveillance: the biological measure of unprotected intercourse using prostate-specific antigen in vaginal secretions of West African women. J Acquir Immune Defic Syndr. 2006;42(4):490–493. doi: 10.1097/01.qai.0000222286.52084.9c. [DOI] [PubMed] [Google Scholar]
- 28.Zenilman JM, Weisman CS, Rompalo AM, et al. Condom use to prevent incident STDs: the validity of self-reported condom use. Sex Transm Dis. 1995;22(1):15–21. doi: 10.1097/00007435-199501000-00003. [DOI] [PubMed] [Google Scholar]
- 29.Catania JA, Gibson DR, Chitwood DD, Coates TJ. Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull. 1990;108(3):339–362. doi: 10.1037/0033-2909.108.3.339. [DOI] [PubMed] [Google Scholar]
- 30.Noar SM, Cole C, Carlyle K. Condom use measurement in 56 studies of sexual risk behavior: review and recommendations. Arch Sex Behav. 2006;35(3):327–345. doi: 10.1007/s10508-006-9028-4. [DOI] [PubMed] [Google Scholar]
- 31.Institute of the Medicine of the National Academies (IOM) Board on Global Health Design considerations: adherence. In: Lagakos SW, Gable AR, editors. Methodological Challenges in Biomedical HIV Prevention Trials. Washington, DC: The National Academies Press; 2008. pp. 119–147. [Google Scholar]
- 32.Aho J, Koushik A, Diakité SL, Loua KM, Nguyen VK, Rashed S. Biological validation of self-reported condom use among sex workers in Guinea. AIDS Behav. 2009 doi: 10.1007/s10461-009-9602-6. [Epub ahead of print] [DOI] [PubMed] [Google Scholar]