Abstract
Background:
Prostate-specific antigen (PSA), a biomarker of vaginal semen exposure, is less susceptible to bias than self-reported condom use behaviors. We examined the agreement of self-reported recent condomless sex (RCS) within couples and how these reports related to PSA detection.
Methods:
We analyzed data from a study conducted in Vietnam, 2017–2020, of 500 different-sex couples using condoms, and no other contraceptive method, to prevent pregnancy for six months. We assessed enrollment and 6-month data from vaginal swabs and questionnaires from both partners. We calculated Prevalence-Adjusted Bias-Adjusted Kappa (PABAK) to evaluate agreement of men’s and women’s reports. Among couples with detected PSA, we assessed partner-concordance of RCS reporting.
Results:
At enrollment (n=499), 79.8% of couples reported no RCS, 16.4% reported RCS, and 3.8% had partner-discordant reports (PABAK 0.93; 95% CI 0.91, 0.97). At 6 months (n=472), 91.7% reported no RCS, 5.7% reported RCS, and 2.5% had partner-discordant reports (PABAK 0.98; 95% CI 0.96, 1.0). Among couples with detected PSA at baseline (11%, n=55), 36% reported no RCS, 55% reported RCS, and 6% had discordant reports; at 6 months (6.6%, n=31), 58% reported no RCS, 35% reported RCS, and 3% had discordant reports.
Conclusions:
We observed high agreement regarding condomless sex within couples in a population using condoms as contraception in Vietnam; however, a high proportion of couples with detected PSA had both partners reporting no RCS, indicating that concordant reporting of no RCS is does not indicate lack of semen exposure.
Keywords: Condoms, Sexual Partners, Prostate-Specific Antigen, Contraception, Biomarkers
Short Summary
Couples using condoms in Vietnam had high between-partner reporting agreement of recent condomless sex. When compared to a measure of vaginal semen exposure, partner agreement alone was not a dependable indicator of the absence of semen exposure.
Introduction
External (“male”) condoms are an effective method of prevention for sexually transmitted infections and pregnancy; however, accurate measurement of condom use is difficult. The simplest and most cost-effective method of collecting data on condom use is via self-report, but these measures are susceptible to social desirability bias, potentially resulting in misreporting of condom use.1–4 In research examining reports from both members of a sexual dyad, self-report has demonstrated weak agreement when comparing men’s and women’s reports of condom use (Cohen’s kappa 0.30–0.43).5 In cases of partner-discordant condom use reports, it is unclear which partner’s report should take precedence.
Using biomarkers to assess recent condomless sex (RCS) overcomes common methodological challenges. Prostate-specific antigen (PSA) is a biomarker present in semen that is detectable in vaginal fluids for up to 48 hours following semen exposure. Vaginal PSA positivity is a validated method of detecting recent semen exposure from penile-vaginal sex and is less susceptible to bias compared to participant self-report of sex and condom use.6 Although PSA is only detectable for up to 48 hours following semen exposure, this short time interval is congruous with self-report measures, which often ask about only the last sex act. Although the use of PSA is a validated measure of semen exposure which is less prone to bias than to self-reports of condomless sex, it is often not practical to implement for a multitude of reasons, including cost, invasive nature of specimen collection, desires to capture exposure outside the immediately preceding 48 hours, and that PSA is only a validated measure of semen exposures for acts of penile-vaginal intercourse.
Previous research has assessed the relationship between individual self-report and PSA positivity,7–9 but there is a lack of literature addressing how this relationship is represented within the sexual dyad. Our objective was to examine the agreement of self-reported RCS between partnered men and women and to assess how these reports relate to PSA detection. Further, we assessed the characteristics of men and women who reported no RCS to clarify how individual and relationship characteristics are associated with semen exposure in the absence of reported RCS.
Materials and Methods
We analyzed data from a randomized controlled trial, collected between June 2017 and February 2020 at a provincial health center in Thanh Hoa, Vietnam.10,11 The trial enrolled 500 established different-sex couples who wanted to avoid pregnancy for the next six months and were not using another method of contraception besides condoms or intending to start another method in the next six months. The trial’s goal was to assess the effect of a condom containing an erectogenic drug on sexual pleasure and sustained condom use.
To be eligible for participation, women needed to be 18–45 years of age, speak Vietnamese, and be in a monogamous relationship for at least the past six months with her current male partner. Breastfeeding, known pregnancy and wanting a pregnancy in the next six months were exclusion criteria. Couples were ineligible for participation if either person was known to be HIV-positive or had a contraindication to use of the novel condom (i.e., history of low blood pressure or heart condition; current use of medication for anemia, blood pressure, erectile dysfunction (man only), migraines, headaches, or glaucoma; inflamed or broken skin that the condom could come into contact with (man only); or latex allergy or sensitivity).
After providing written consent, couples were randomized to use either standard condoms or intervention condoms for six months. This provides an ideal study population for the present analysis, which seeks to understand the behaviors and reports of couples using condoms to prevent pregnancy.
At the enrollment visit, men and women were independently administered enrollment questionnaires, and a study clinician collected a vaginal swab from women. Women were asked to return for follow-up at 2, 4, and 6 months post-enrollment to complete follow-up questionnaires and have another vaginal swab collected. Men were asked to return once, 6 months after enrollment. Given the potential for participants’ low literacy, all questionnaires were administered via face-to-face interviews. Participants were interviewed without their partner present by trained female interviewers. Women were counseled about condoms for pregnancy and disease prevention at enrollment and at 2 and 4-month visits. Institutional review boards at The Ohio State University and the Hanoi School of Public Health provided ethical oversight.
Self-reported recent condomless sex
Self-reported condomless sex in the last two days (day of interview to 2 days prior) was identified separately for men and women. We classified participants as reporting recent condomless sex (RCS) if they responded “today,” “yesterday,” or “day before yesterday” to the survey question asking about the last time that they had vaginal sex with their partner without using a condom for the entire act. The survey defined vaginal sex as “sex in which the penis goes into the vagina” (see Table, Supplemental Digital Content 1, for survey questions and response options related to recent condomless sex).
Measurement of Semen Exposure
The presence of semen was determined through the detection of PSA in women’s vaginal fluid samples, with a minimum level of 1ng/mL required for detection. The specimens were collected using a double-headed 1-mL rayon swab, stored at −70°C on-site and later shipped to the University of North Carolina laboratory on dry ice. The laboratory processed one of the swabs per batch and tested it for total PSA using the Architect Total PSA test (Abbott Diagnostics, Abbott Park, IL) following established procedures.12
Statistical Analyses
These analyses use data from the enrollment and 6-month visits because those are the time points at which men also completed survey instruments, including reporting of RCS. We categorized couples into three categories based on the concordance of men’s and women’s reports of RCS: partner-concordant reporting of RCS, partner-concordant denying of RCS, and partner-discordant reporting. We describe the study population stratified by concordance category at enrollment, using counts and percentages. Characteristics assessed include women’s age, men’s and women’s educational attainment, location of residence, parity, and wealth index.
Participants reported their demographic characteristics at the enrollment visit. We categorized men’s and women’s education as lower secondary or less (completed 9th grade or less), upper secondary (grades 10–12), or higher (post-secondary education). Women reported on the number of times they have given birth to a live infant. Women identified their area of residence as city (population ≥50,000), town, or countryside/rural. We constructed a wealth index using principal component analysis, implementing methods similar to those used by the Demographics and Health Survey.13,14 The index comprises household assets reported on women’s enrollment questionnaires, including water sources, toilet facilities, household possessions, cooking fuel, and the number of sleeping rooms. Wealth index is included as a percentile (1–100) when used as a covariate and reported in quartiles.
We compared self-reported RCS for men and women at enrollment and 6 months and calculated Cohen’s Kappa (κ) and Prevalence-Adjusted Bias-Adjusted Kappa (PABAK) to assess the level of agreement between partner reports. PABAK is the preferred statistic when the marginal prevalence of the outcome (self-reported RCS) is far from 50%.15,16 We then assessed partner concordance stratified by PSA positivity at enrollment and 6-month visits and present counts and percentages.
In order to assess individual and couple characteristics associated with semen exposure when recent condomless sex is not reported, we then calculated risk ratios for women’s semen exposure among two nonexclusive subpopulations: couples in which the man did not report RCS, and couples in which the woman did not report RCS. We used log-binomial generalized estimating equations and presented adjusted and unadjusted estimates. Adjustment variables included in these regressions are woman’s age, men’s and women’s education, parity, wealth, and area of residence. Although no difference was found in semen exposure between study arms assigned the standard or investigational condoms, study arm assignment was included as a covariate in adjusted models.10 We conducted statistical analyses in Stata (BE 17.0, StataCorp LLC, College Station, TX).
Results
One couple was excluded due to missing data (missing man’s education), leading to an analytic sample of 499 couples. The complete sample was first described in Nguyen et al. (2022).10 At enrollment, 398 couples were partner-concordant denying of RCS (79.8%), 82 were partner-concordant reporting of RCS (16.4%), and 19 couples were partner-discordant (3.8%). At 6 months, 433 couples were partner-concordant denying of RCS (91.7%), 27 were partner-concordant reporting of RCS (5.7%), and 12 couples were partner-discordant (2.5%).
At enrollment, the mean age of participating women was 34.2 years, with a range of 21 to 46 years, and a majority of women between the ages of 30 and 39 years (Table 1). The study population was primarily composed of couples in which the woman had a parity of two or more, including 78.1% of partner-concordant reporting of RCS, 85.4% of partner-concordant denying of RCS, and 100% of partner-discordant couples. About half of the study population resided in urban environments, with 57.3% of partner-concordant reporting of RCS, 57.0% of partner-concordant denying of RCS, and 42.1% of partner-discordant couples living in cities with a population of at least 50,000. The majority of women had received education beyond the high school level, with 75.6% of partner-concordant reporting of RCS, 74.1% of partner-concordant denying of RCS, and 63.2% of partner-discordant couples having a female partner with higher education. Similarly, 68.3% of partner-concordant reporting of RCS, 67.8% of partner-concordant for denying of RCS, and 57.9% of partner-discordant couples had a male partner with higher education.
Table 1:
Descriptive characteristics of participating couples by self-reported recent condomless sex at enrollment
| RCS+a Concordant | RCS− Concordant | Partner Discordant | All Couplesb | |||||
|---|---|---|---|---|---|---|---|---|
| n | (%) | n | (%) | n | (%) | n | (%) | |
| 82 | (16.4) | 398 | (79.8) | 15 | (3.0) | 499 | (100.0) | |
|
| ||||||||
| Woman’s Age | ||||||||
| 21–29 years | 31 | (37.8) | 91 | (23.5) | 7 | (36.8) | 127 | (26.2) |
| 30–39 years | 39 | (47.6) | 222 | (57.2) | 9 | (47.4) | 269 | (55.5) |
| 40–50 years | 12 | (14.6) | 75 | (19.3) | 3 | (15.8) | 89 | (18.4) |
|
| ||||||||
| Woman’s Parity | ||||||||
| Fewer than 2 births | 18 | (22.0) | 58 | (14.6) | 0 | (0.0) | 76 | (15.4) |
| 2 or more births | 64 | (78.1) | 340 | (85.4) | 19 | (100.0) | 419 | (86.7) |
|
| ||||||||
| Area of Residence | ||||||||
| City (population ≥50,000) | 47 | (57.3) | 227 | (57.0) | 8 | (42.1) | 280 | (56.6) |
| Town | 13 | (15.9) | 68 | (17.1) | 6 | (31.6) | 87 | (17.6) |
| Countryside/Rural | 22 | (26.8) | 103 | (25.9) | 5 | (26.3) | 128 | (25.9) |
|
| ||||||||
| Woman’s Education | ||||||||
| Lower Secondary or Less | 5 | (6.1) | 36 | (9.0) | 2 | (10.5) | 43 | (8.7) |
| Upper Secondary | 15 | (18.3) | 67 | (16.8) | 5 | (26.3) | 85 | (17.2) |
| Higher Education | 62 | (75.6) | 295 | (74.1) | 12 | (63.2) | 367 | (74.1) |
|
| ||||||||
| Man’s Education | ||||||||
| Lower Secondary or Less | 10 | (12.2) | 41 | (10.3) | 4 | (21.1) | 54 | (10.9) |
| Upper Secondary | 16 | (19.5) | 87 | (21.9) | 4 | (21.1) | 105 | (21.2) |
| Higher Education | 56 | (68.3) | 270 | (67.8) | 11 | (57.9) | 366 | (67.9) |
|
| ||||||||
| Wealth Quartile | ||||||||
| 1st Quartile (Lowest) | 20 | (24.4) | 98 | (24.8) | 8 | (42.1) | 125 | (25.4) |
| 2nd Quartile | 22 | (26.8) | 103 | (26.0) | 2 | (15.8) | 127 | (25.8) |
| 3rd Quartile | 23 | (28.1) | 126 | (31.8) | 4 | (21.1) | 152 | (30.8) |
| 4th Quartile (Highest) | 17 | (20.7) | 69 | (17.4) | 4 | (21.1) | 89 | (18.1) |
Recent condomless sex (RCS) is defined as self-reported penile-vaginal sex without the use of a condom in the past 2 days.
Includes four couples in which one partner did not report on RCS at enrollment and are therefore not included in stratified enrollment statistics.
Men’s and women’s self-reported RCS showed excellent agreement, with Cohen’s kappa of 0.90 (95% CI 0.85, 0.95) at enrollment and 0.91 (95% CI 0.83, 0.99) at 6 months, and PABAK of 0.94 (95% CI 0.91, 0.97) at enrollment and 0.98 (95% CI 0.96, 1.0) at 6 months (Table 2). Of the 55 PSA-positive couples at enrollment, 20 (36%) were partner-concordant denying of RCS and 3 (5.5%) were partner-discordant. At 6 months, 31 couples were PSA-positive, and among those 18 (58%) were partner-concordant denying of RCS and 1 (3%) was partner-discordant. (Table 3)
Table 2:
Self-reported recent condomless sex for men and women at enrollment and six months
| Women’s Reports | ||||||||
|---|---|---|---|---|---|---|---|---|
| RCSa Reported | No RCS Reported | Unreportedb | ||||||
| Men’s Reports | n | (%) | n | (%) | n | (%) | PABAKc | (95% CI) |
|
| ||||||||
| Enrollment Visit (n=499) | 0.94 | (0.91, 0.97) | ||||||
| RCS Reported | 82 | (16.4) | 6 | (1.2) | 0 | (0.0) | ||
| No RCS Reported | 9 | (1.8) | 398 | (79.8) | 3 | (0.6) | ||
| Unreported | 0 | (0.0) | 1 | (0.2) | 0 | (0.0) | ||
|
| ||||||||
| Six-Month Visit (n=472) | 0.98 | (0.96, 1.00) | ||||||
| RCS Reported | 27 | (5.7) | 2 | (0.4) | 0 | (0.0) | ||
| No RCS Reported | 3 | (0.6) | 433 | (91.7) | 0 | (0.0) | ||
| Unreported | 0 | (0.0) | 3 | (0.6) | 4 | (0.9) | ||
Recent condomless sex (RCS) is defined as self-reported penile-vaginal sex without the use of a condom in the past 2 days.
Participants who did not answer all components needed to assess recent RCS.
Prevalence-Adjusted Bias-Adjusted Kappa (PABAK) measures agreement between men’s and women’s reports of RCS. This calculation excludes couples in which either partner is “unreported.”
Table 3:
Concordance of men’s and women’s self-reported recent condomless sex by PSA positivity at enrollment and 6 months
| Enrollment Visit (n=499) | 6-Month Visit (n=472) | |||
|---|---|---|---|---|
| n | (%) | n | (%) | |
|
| ||||
| PSA+ | ||||
| RCS+a Concordant | 30 | (54.6) | 11 | (35.5) |
| RCS− Concordant | 20 | (36.4) | 18 | (58.1) |
| Partner Discordant | 3 | (5.5) | 1 | (3.2) |
| Unreportedb | 2 | (3.6) | 1 | (3.2) |
| Total PSA+ | 55 | (100.0) | 31 | (100.0) |
|
| ||||
| PSA− | ||||
| RCS+ Concordant | 52 | (11.7) | 16 | (3.6) |
| RCS− Concordant | 378 | (85.1) | 415 | (94.1) |
| Partner Discordant | 12 | (2.7) | 4 | (0.9) |
| Unreported | 2 | (0.5) | 6 | (1.4) |
| Total PSA− | 444 | (100.0) | 441 | (100.0) |
Recent condomless sex (RCS) is defined as self-reported penile-vaginal sex without the use of a condom in the past 2 days.
At least one partner did not report on RCS.
Among participants not reporting RCS, men’s and women’s education, women’s parity, and area of residence were not significantly associated with semen exposure (Table 4). Among both men and women not reporting RCS, the risk of semen exposure was significantly higher if the couple had partner-discordant reports (men’s aRR: 3.72, 95% CI: 1.14, 12.11; women’s aRR: 3.51, 95% CI: 1.22, 10.08). Couples in which women were 30–39 years old had a significantly lower risk of semen exposure than those in which the woman was 21–29 years of age in both the population where men reported no RCS (aRR 0.44, 95% CI 0.21, 0.88) and where women reported no RCS (aRR 0.41, 95% CI 0.19, 0.87). A similar reduction in risk was seen when comparing couples with women ages 40–50 to those 21–29 years, but this relationship was not statistically significant. An increase in wealth index was associated with decreased risk of semen exposure among couples in which men did not report RCS, with the highest quartile having a risk of semen exposure 39% that of the lowest quartile (aRR 0.40, 95% CI 0.16, 1.00).
Table 4:
Relative risk of semen exposure among participants not reporting recent condomless sex
| Man Reports No RCSa,b | Woman Reports No RCSa,b | |||||||
|---|---|---|---|---|---|---|---|---|
| Unadjusted | Adjustedc | Unadjusted | Adjustedc | |||||
| RR | (95% CI) | RR | (95% CI) | RR | (95% CI) | RR | (95% CI) | |
|
| ||||||||
| Partner’s Report | ||||||||
| No RCS Reported | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
| RCS Reported | 5.82 | (2.39, 14.17) | 3.72 | (1.14, 12.11) | 2.69 | (0.41, 17.91) | 3.51 | (1.22, 10.08) |
|
| ||||||||
| Woman’s Age | ||||||||
| 21–29 years | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
| 30–39 years | 0.53 | (0.28, 1.02) | 0.44 | (0.21, 0.88) | 0.50 | (0.26, 0.97) | 0.41 | (0.19, 0.87) |
| 40–50 years | 0.56 | (0.24, 1.33) | 0.47 | (0.19, 1.18) | 0.52 | (0.21, 1.29) | 0.44 | (0.17, 1.16) |
|
| ||||||||
| Woman’s Education | ||||||||
| Lower Secondary or Less | 1.31 | (0.49, 3.54) | 1.16 | (0.32, 4.16) | 1.47 | (0.54, 4.00) | 1.50 | (0.40, 5.63) |
| Upper Secondary | 1.56 | (0.75, 3.23) | 1.42 | (0.59, 3.38) | 1.61 | (0.78, 3.29) | 1.62 | (0.66, 4.00) |
| Higher Education | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
|
| ||||||||
| Man’s Education | ||||||||
| Lower Secondary or Less | 1.39 | (0.61, 3.18) | 1.02 | (0.31, 3.34) | 1.24 | (0.50, 3.10) | 0.88 | (0.24, 3.29) |
| Upper Secondary | 1.02 | (0.47, 2.24) | 0.82 | (0.37, 1.81) | 0.99 | (0.46, 2.11) | 0.74 | (0.31, 1.73) |
| Higher Education | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
|
| ||||||||
| Woman’s Parity | ||||||||
| Fewer than 2 births | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
| 2 or more births | 2.28 | (0.72, 7.25) | 3.13 | (0.94, 10.41) | 2.10 | (0.66, 6.65) | 3.41 | (1.00, 11.60) |
|
| ||||||||
| Area of Residence | ||||||||
| City (population ≥50,000) | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
| Town | 0.81 | (0.30, 2.13) | 0.59 | (0.22, 1.56) | 0.52 | (0.18, 1.45) | 0.38 | (0.14, 1.06) |
| Rural | 1.18 | (0.61, 2.23) | 0.99 | (0.49, 1.99) | 0.89 | (0.44, 1.80) | 0.71 | (0.33, 1.53) |
|
| ||||||||
| Wealth Category | ||||||||
| 1st Quartile (lowest) | ref. | -- | ref. | -- | ref. | -- | ref. | -- |
| 2nd Quartile | 0.69 | (0.31, 1.53) | 0.58 | (0.25, 1.34) | 0.78 | (0.33, 1.80) | 0.59 | (0.23, 1.47) |
| 3rd Quartile | 0.82 | (0.39, 1.69) | 0.81 | (0.32, 2.00) | 1.09 | (0.52, 2.29) | 0.89 | (0.35, 2.26) |
| 4th Quartile (highest) | 0.51 | (0.19, 1.38) | 0.40 | (0.16, 1.00) | 0.51 | (0.17, 1.54) | 0.40 | (0.14, 1.12) |
Relative risk of semen exposure among couples in which the [man/woman] did not report recent condomless sex (RCS), defined as self-reported penile-vaginal sex without the use of a condom in the past 2 days, calculated using generalized estimating equations for enrollment and 6-month visits.
Men’s analysis includes 410 couples at enrollment and 436 at 6 months; Women’s analysis includes 405 couples at enrollment and 438 at 6 months.
Adjusted for woman’s age (continuous, except when modeling risk associated with women’s age), men’s and women’s education, area of residence, wealth index percentile, number of live births (continuous, except when modeling risk associated with women’s parity), and parent study assignment.
Conclusions
In the present study, we aimed to examine the agreement of self-reported recent condomless sex (RCS) between partnered men and women and to assess how these reports relate to PSA detection. We found that there was a high level of agreement between men and women’s self-reported RCS, as measured by the Prevalence-Adjusted Bias-Adjusted Kappa (PABAK) statistic. However, we also found that a substantial proportion of those who had positive PSA results also had partner-concordant reports of no RCS.
The high agreement between men’s and women’s reports of recent condomless sex is a significant finding, as previous research has shown poor agreement between men’s and women’s reports of condom use.5 Previous research assessing men’s and women’s reports of sexual behaviors has used partner concordance as an indicator that reports were more likely to be accurate.17 However, these findings suggest that partner-concordant reports may not necessarily indicate that reporting is accurate.
A dyadic analysis of men’s and women’s reports of contraception use in twelve countries by Beker et al. (2001) concluded that women’s self-reports are likely more accurate than men’s.18 This is consistent with our finding that among couples in which men report no RCS, the relative risk of semen exposure associated with a partner reporting RCS is of greater magnitude than among couples where women report no RCS but their partner does.
One potential explanation for this discrepancy is that partner-concordant reports are more indicative of greater sexual communication within the dyad than of accurate reports. Couples who discuss their beliefs and expectations regarding sexual behaviors may be more inclined to provide similar answers, regardless of actual behavior. Additionally, the discrepancy between RCS reports and PSA suggests that couples may not be aware of, or may not consider certain sexual behaviors as RCS, such as withdrawal or other forms of contraceptive failure. It is also possible that couples are aware of their RCS behaviors but may underreport them due to social desirability bias, where couples may be reluctant to report RCS due to cultural or societal norms or because they were participating in a study in which they were counseled to use condoms. We took measures to reduce the impact of social desirability bias, including training interviewers on proper techniques and interviewing participants privately without their partner present; however, due to the sensitive nature of sexual behaviors, the impact of social desirability bias cannot be excluded.
Qualitative research has provided several explanations for disagreement between PSA results and self-reported condom use. Interviews with 11 women attending an STI clinic in Kingston, Jamaica who reported no recent condomless sex but had positive PSA results provide possible explanations for this disagreement.19 Six participants reported plausible explanations for PSA positivity, including two participants who reported instances of external ejaculation which may have come in contact with the vagina, three who suspected or confirmed condom slippage or breakage, and one who reported that no condom was used for subsequent intercourse following condom use. The remaining five interviewees provided no explanation, or the explanation provided was deemed implausible. Of the six who provided a plausible explanation, only one did so before learning their PSA results, with further disclosures occurring after rapport was built with the interviewer. These results indicate that, in addition to social desirability bias, misreport could be driven by a lack of specificity in questions. The authors suggest that researchers should ask not only about confirmed condom failure but also about suspected slippage or breakage.
The detection of semen exposure through PSA testing offers improved accuracy compared to self-reported measures, as it can also capture instances of incorrect condom use or condom failure. A study assessing self-reports of condom use among commercial sex workers found detectable PSA (≥ng/dL concentrations) among 21% of those who reported no sex and 39% of those who reported only protected sex in the preceding 48 hours.7 Another study of sexually active monogamous couples found an estimated breakage rate of 0.4% for latex condoms and a slippage rate of 0.9% per 6 menstrual cycles of use.20 They found detectable levels of PSA following 65% of condom breaks, but at levels approximately 50% of that found from condomless sex. They also detected PSA following 41% of slip-offs, at levels approximately 80% lower than that found from condomless sex.
This study’s use of PSA as a biomarker of semen exposure provides a valuable point of comparison for self-reported RCS behaviors. The overall rate of PSA-positivity was low, which is unsurprising due to the speed of clearance of PSA from vaginal fluids; however, a high proportion of PSA-positive couples reported no RCS. Interestingly, although the overall rate of PSA-positivity declined from enrollment to 6 months, a similar number of PSA-positive couples had concordant reports of no RCS at enrollment (n=20) and 6-months (n=18); the reduction in PSA-positivity was driven by a reduced frequency of PSA-positive couples with concordant reports of RCS, a change that we would expect because the enrollment data were collected based on behaviors before the couples entered a study about condom use. The discrepancy between reported RCS and semen exposure highlights the need for more accurate and comprehensive measures of RCS behaviors among clinicians and researchers for whom the assessment of biomarkers is unavailable or impractical. This is an important consideration for future research, as it highlights the importance of developing methods for assessing RCS behaviors that are both accurate and practical for use in different settings.
Our findings suggest that demographic factors such as age and socioeconomic status may also influence the reliability of self-reports of RCS. We found that couples in which the man or woman reported no RCS with women aged 30–39 had a 41–44% the risk of PSA positivity compared to couples in which the woman was 21–29 years of age. Similarly, in couples in which the man reported no RCS, couples in the highest wealth quartile had 40% the risk of PSA positivity, compared to those in the lowest wealth quartile. There are multiple potential explanations for this finding. It is possible that couples who are younger and of lower socioeconomic status have less knowledge about effective condom use, leading them to not recognize RCS behaviors accurately. Understanding how demographic factors may influence the accuracy of self-reported condom use is important for the development of interventions that are tailored to the specific needs of different populations.
One of the main study limitations is that while the presence of PSA in vaginal fluids is confirmatory of recent condomless sex, the absence of PSA does not indicate a lack of recent condomless sex. This means that while we were able to identify participants who reported not having recent condomless sex, we cannot confirm their self-reports in cases where PSA was not detected in vaginal fluids. Additionally, PSA is validated to detect semen in vaginal fluids up to 48 hours after condomless sex; however, our measure of self-reported recent condomless sex is not directly analogous to this. Self-reports are based on calendar days (today, yesterday and day before yesterday) rather than discrete times, so cannot say with certainty that the reported recent condomless sex occurred within the previous 48 hours, only that it took place within the previous two calendar days. Also, the use of PSA as a biomarker of semen exposure is limited to penile-vaginal intercourse and may not be able to accurately capture other forms of sexual activity. Reported RCS is impacted by coital frequency because couples engaging in more frequent coitus are more likely to have sex which occurs within the 48-hour reporting period, whereas couples with lower coital frequency are less likely to have sex acts captured in this window; therefore, couples with lower coital frequency who are having condomless sex may not be identified with this measure. Finally, reported RCS is itself subject to misreporting due to social desirability or recall bias, including gendered differences in reporting of coital frequency.21,22 It should also be noted that since it was conducted in Vietnam, the results may not be generalizable to other populations. Moreover, the study was conducted in a specific context in which there is a strong emphasis on family planning and reproductive health, which may not be the case in other settings.
A primary study strength was that the dyadic design allowed for the assessment of both men’s and women’s self-reports, which is rare in studies of this kind. The longitudinal design of the study also increased statistical power by providing up to two observations per couple in this analysis and allowed for observation of reporting behaviors from the enrollment to 6 months. Direct detection of semen exposure is also superior to self-reported measures of condom use in assessing risk because, in addition to capturing condom non-use, it also measures semen exposure due to incorrect use or condom failure.
This research highlights the need for more comprehensive methods of assessing semen exposure in future research. Future research should also focus on exploring the factors that influence self-reported condom use in diverse populations, as this would provide a more complete understanding of the accuracy of self-reported condom use in different contexts.
Supplementary Material
Acknowledgements:
The authors thank John Casterline for his comments on the manuscript.
Sources of Support: This study was supported by award number R01HD084637 from the National Institute of Child Health and Human Development, and by award number UL1TR002733 from the National Center for Advancing Translational Sciences, of the National Institutes of Health. Futura Medical donated the CSD500 condoms used in the parent study; they otherwise had no input into the study design or its report. The first author received a fellowship from the American Sexually Transmitted Diseases Association to complete these analyses. The content is solely the responsibility of the author and does not necessarily represent the official views of the funders.
Footnotes
Disclosure: The authors report no conflicts of interest.
References
- 1.Noar SM, Cole C, Carlyle K. Condom use measurement in 56 studies of sexual risk behavior: Review and recommendations. Arch Sex Behav. 2006. Jun 24;35(3):327–45. [DOI] [PubMed] [Google Scholar]
- 2.Fonner VA, Kennedy CE, O’Reilly KR, Sweat MD. Systematic Assessment of Condom Use Measurement in Evaluation of HIV Prevention Interventions: Need for Standardization of Measures. AIDS Behav. 2014. Dec 1;18(12):2374–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Agnew CR, Loving TJ. The role of social desirability in self-reported condom use attitudes and intentions. AIDS Behav. 1998;2(3):229–39. [Google Scholar]
- 4.Rao A, Tobin K, Davey-Rothwell M, Latkin CA. Social Desirability Bias and Prevalence of Sexual HIV Risk Behaviors Among People Who Use Drugs in Baltimore, Maryland: Implications for Identifying Individuals Prone to Underreporting Sexual Risk Behaviors. AIDS Behav. 2017. Jul 1;21(7):2207–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Koffi AK, Adjiwanou VD, Becker S, Olaolorun F, Tsui AO. Correlates of and Couples’ Concordance in Reports of Recent Sexual Behavior and Contraceptive Use. Stud Fam Plann. 2012. Mar;43(1):33–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Minnis AM, Steiner MJ, Gallo MF, et al. , Van Der Straten A, et al. Biomarker validation of reports of recent sexual activity: results of a randomized controlled study in zimbabwe. Am J Epidemiol. 2009;170(7):918–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.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. Aug;33(8):476–9. [DOI] [PubMed] [Google Scholar]
- 8.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. Jan 25;18(1):33–8. [DOI] [PubMed] [Google Scholar]
- 9.Woolf-King SE, Muyindike W, Hobbs MM, et al. Vaginal Prostate Specific Antigen (PSA) Is a Useful Biomarker of Semen Exposure Among HIV-Infected Ugandan Women. AIDS Behav 2016 217. 2016. May 25;21(7):2141–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Nguyen NC, Luong TN, Le VT, et al. Effectiveness of erectogenic condom against semen exposure among women in Vietnam: Randomized controlled trial. PloS One [Internet]. 2022. Feb 1 [cited 2022 Mar 23];17(2). Available from: https://pubmed.ncbi.nlm.nih.gov/35176037/ [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gallo MF, Nguyen NC, Luff A, et al. Effects of a Novel Erectogenic Condom on Men and Women’s Sexual Pleasure: Randomized Controlled Trial. J Sex Res. 2022. Nov 22;59(9):1133–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gallo MF, Snead MC, Black CM, et al. Optimal methods for collecting and storing vaginal specimens for prostate-specific antigen testing in research studies. Contraception. 2013. Jun;87(6):830–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rutstein SO, Johnson K. The DHS Wealth Index. Calverton, MD: ORC Marco; 2004. Aug. [Google Scholar]
- 14.Vyas S, Kumaranayake L. Constructing socio-economic status indices: how to use principal components analysis. Health Policy Plan. 2006. Nov 1;21(6):459–68. [DOI] [PubMed] [Google Scholar]
- 15.Byrt T, Bishop J, Carlin JB. Bias, prevalence and kappa. J Clin Epidemiol. 1993. May;46(5):423–9. [DOI] [PubMed] [Google Scholar]
- 16.Flight L, Julious SA. The disagreeable behaviour of the kappa statistic. Pharm Stat. 2015;14(1):74–8. [DOI] [PubMed] [Google Scholar]
- 17.Harvey SM, Bird ST, Henderson JT, et al. He Said, She Said: Concordance Between Sexual Partners. Sex Transm Dis. 2004;31(3):185–91. [DOI] [PubMed] [Google Scholar]
- 18.Becker S, Costenbader E. Husbands’ and wives’ reports of contraceptive use. Stud Fam Plann. 2001;32(2):111–29. [DOI] [PubMed] [Google Scholar]
- 19.Carter MW, Bailey A, Snead MC, et al. Exploring discordance between biologic and self-reported measures of semen exposure: a qualitative study among female patients attending an STI clinic in Jamaica. AIDS Behav. 2013. Feb 15;17(2):728–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Walsh TL, Frezieres RG, Peacock K, et al. Use of prostate-specific antigen (PSA) to measure semen exposure resulting from male condom failures: Implications for contraceptive efficacy and the prevention of sexually transmitted disease. Contraception. 2003. Feb 1;67(2):139–50. [DOI] [PubMed] [Google Scholar]
- 21.Lagarde E, Enel C, Pison G. Reliability of reports of sexual behavior: a study of married couples in rural west Africa. Am J Epidemiol. 1995. Jun 15;141(12):1194–200. [DOI] [PubMed] [Google Scholar]
- 22.Seal DW. Interpartner concordance of self-reported sexual behavior among college dating couples. J Sex Res. 1996;34(1):39–55. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.