THEINCREASING RATE OF HIV infection among US women and their biologicvulnerability to sexually transmitted diseases (STDs) are serious publichealth problems. 1,2 Besidesmutual monogamy with an uninfected partner, the male latex condom remains thebest method of protection from HIV and other STDs for sexually activeindividuals. 3–5 Consequently, HIV prevention messages for women focus on promoting consistentand correct use of condoms during sex with their malepartners.
Despite evidence that condoms are effective inreducing transmission of HIV/STD, women at risk for contracting HIV/STD oftenreport that they use condomsinfrequently. 6–10 Understanding the factors associated with condom use can help in the design ofeffective prevention messages. Findings show that women’s experiences innegotiating and using ofcondoms, 9,11 theirperceived social norms and socialsupport, 13,14–20 their perceived risk ofHIV/STD, 6,14 and theirself-efficacy in condomuse 13,21–22 are associated with their condom use behavior. Such psychosocial factors canbe influenced by public healthinterventions. 6,9,10,13,14,23–25
Thetheory of planned behavior, developed to explain behaviors, includinghealth-relatedbehaviors, 26,27 encompasses many categories of psychosocial variables known to be associatedwith condom use, particularly attitudes, subjective norms, and perceivedbehavior control. 28 Some researcherssuggests that adolescents underreport their STD history, whereas others havefound reporting to be reliable in this agegroup. 29–31 Finally, this study used the phrasing “has a doctor told you,”which may improve the accuracy of reporting. Similar disparate findings havebeen documented for self-reported condomuse. 32,33
Characteristicsof the sexual partnership also predict consistency of condomuse. 6,9,10,14,23–25,34 Furthermore, whereas STD history is associated with a number of risk behaviorsincluding alcohol use, current high-risk behavior, and other factorsassociated with condom use, differences in patterns of condom use by STDhistory have not been fullyinvestigated. 19,35
Theobjectives of this analysis were to use the theory of planned behavior as thetheoretical framework for evaluating psychosocial factors associated with malecondom use previously identified in the literature, and to assess theinfluence that partner type and STD history have on the association betweenpsychosocial factors and condom use. Specifically, it was hypothesized thatincreased condom use among women with a history of STDs is associated withhigher perceived risk and more assertiveness in usingcondoms.
Datafor this analysis were collected during the initial visit of participants in aprospective study of the effectiveness of male and femalecondoms. 36,37 Womenattending the STD clinics of the health departments of Jefferson and MadisonCounties (Alabama) comprised the population from which the sample was drawn.Those eligible for this study were between 18 and 34 years of age, were notpregnant or planning to become pregnant within the next 6 months, had notundergone a hysterectomy, and were not receiving long-term treatment withantibiotics. Recruitment began in July 1995 and was completed in August1997.
Women willing and eligible to participate in thisstudy were scheduled for an interview approximately 10 days after they wererecruited, thus allowing the clearing up of any STD diagnosed at therecruitment visit. Before the interview began, the women were informed aboutthe purpose of the investigation and the procedures in place to protect theirconfidentiality. After they had provided written informed consent, theparticipants met privately with a trained female interviewer, who administereda face-to-face 45-minute structured interview covering social-demographicvariables; sexual, reproductive, and medical history; and condom use in thepreceding 30 days. The psychosocial factors theoretically related to condomuse also were measured. Data from the initial interviews of the 1159 womenenrolled in the female condom acceptability study were used in thisanalysis.
The study protocol was reviewed and approvedinitially by the institutional review boards at the University of Alabama atBirmingham (UAB), the Alabama State Department of Public Health, and theCenters for Disease Control (CDC). The protocol was reviewed annually by theUAB and CDC institutional review boards thereafter.
Atotal of 3531 eligible women were identified. Of these women, 2702 agreed toparticipate in the study, and 1159 attended an initial study visit. Agreementto participate in this study was more likely among eligible women who wereblack, less educated, unmarried, receiving an income from welfare programs,characterized by a history of STDs, and involved with a higher number of sexpartners during their lifetime. There were essentially no differences betweeneligible women who did not participate (n = 1543) and those who did (n= 1159), with the exception that the participants who had a highernumber sex partners during their lifetime were more likely to have a historyofSTDs. 38
Interviewitems were developed to measure important components of the theory of plannedbehavior that could be associated with male condom use(Table 1). Theoutcome variable was the proportion of coital acts protected with a malecondom during the previous 30 days, by partner type. Condom use informationwas based on self-report, and data were obtained separately for main and otherpartners. The participant was asked whether she had a main partner(“like a husband or boyfriend”), how many coital acts she hadexperienced with him during the previous 30 days, and in how many of thoseacts her partner had used a condom. Next, she was asked whether she had anyother partners as well as how many coital and condom-protected acts she hadexperienced with such partners during the preceding 30 days.
Principalaxis factor analysis was used to construct the scales, using the hypothesizedconceptual framework as a guide for including items in specific scales. Inthis analysis, items within each domain were factor analyzed separately withprincipal axis analysis and an orthogonal rotation. Scales were constructedwith items that discriminated well between the factors that had an eigenvaluegreater than 1. Cronbach α was computed to estimate the reliability of allthe scales. Some summary indices were constructed by counting the number oflike responses across items.
Demographic and behaviorfactors included age, duration of the relationship with the main partner, druguse risk behavior, number of sex partners during the lifetime, and violence inthe relationship. Lifetime drug use risk behavior was measured by the numberof different high-risk drug use behaviors the woman had demonstrated such assmoking crack, shooting up, and sharing needles. A high score on the drug usescale indicates riskier behavior. The different types of violence in therelationship were numbered by counting the positive responses to sevendifferent types of violence on a checklist that the woman could haveexperienced in herrelationship. 39
Theattitudes associated with condom use included five scales: (1) perceived needfor male condom use in a faithful or trusting relationship with the mainpartner (α = 0.84), (2) perceived need for male condom use in afaithful or trusting relationship with other partner(s) (α = 0.82),(3) perceived need for condom use in a long-term relationship (α =0.73), (4) convenience of male condom use (α = 0.74), and (5)consequences the woman experienced requesting condom use, as measured countingpositive responses to a checklist of negative consequences a woman mightexperience in asking any sex partner to use acondom.
Perceived condom norms were measured with twoitems about peers’ condom use with main or other partners (α= 0.63). A high score indicated that condom use was more normativeamong peers.
Perceived behavior control was measuredwith seven scales: (1) The acceptability of the woman’s requests forcondom use was measured by the number of positive responses to three itemsassociated with partner response to such requests. (2) The number of times thewoman put the condom on the man in the past 30 days was measured byself-report. (3) The perceived STD/HIV risk from the main partner was measuredby two items. (4) The perceived HIV/STD risk from the other partner also wasmeasured by two items. Both scales had high reliability (α = 0.88main partner, α = 0.91 other partner). (5) Who usually applied thecondom was determined by a single item and scored on a 5-point Likert scale,with a higher score indicating that the man was more likely to apply thecondom. (6) Self-efficacy in condom use with the main partner was measuredusing five items related to condom use in different situations. The scale hada reliability coefficient of 0.79. (7) Self-efficacy in condom use with otherpartners was measured with two items with a reliability coefficient of 0.63.For both items, a low score indicated higher condom useself-efficacy.
The measure of STD history used responsesfrom data collected during the initial study visit at the clinic. Eight itemswere used to construct this variable. The women were asked whether a doctorhad ever told them that they had hepatitis, genital warts, herpes (oral orgenital), gonorrhea, syphilis, chlamydia, or trichomoniasis. If a respondenthad been told she any of these diseases, she was classified as having ahistory ofSTDs.
Binomialregression was used to evaluate the association between potential predictorsof condom use and the proportion of protected coital acts, by partner type.Each partner-specific response of women who reported any sexual activityduring the previous 30 days was considered a binomial experiment, the numberof trials being the number of partner-specific acts and the number ofsuccesses being the corresponding number of condom-protectedacts.
A two-stage approach was used to develop themultiple binomial regression models. In the first stage, preliminary analyseswere conducted to identify psychosocial and selected demographic variablesassociated with the two outcomes (condom use with main and other partners).The association between condom use with both main and other partners andpsychosocial factors was assessed using a Pearson correlation coefficient.This correlation coefficient was weighted by coital frequency. In stepwiseregression analyses, a P value of 0.20commonly is used as the cutoff in selecting variables for inclusion in furthermodels. Given the large number of factors evaluated, a critical value of 0.01was selected to minimize dependence on chance. This value was selectedaccording to a selection criterion of 0.2 for each of the 19 variablesevaluated.
In the second stage of the analysis, all thevariables selected in the previous step were included in the binomialregression models. Regression models were fit separately according toself-reported STD history and partnertype.
Mostof the women who participated in this study were young and black (84%). On theaverage, they were 24 years of age and had been in a relationship for 27months (Table 1). Amedian income of $300 to $600 per month was reported by the 45% indicatingthat they were employed. Approximately 27% of the women had less than a highschool education. One third (37%) had a high school education, and theremaining 36% had more than a high school education. Approximately 49% of thewomen reported that they currently used condoms for birth control. Accordingto their self-reports, 104 women had engaged in sex with a partner other thantheir main partner during the preceding 30days.
Table 2 presents the bivariate associations between the twooutcomes and the psychosocial factors significantly associated with at leastone of the partner types. In these analyses, the proportion of protectedcoital acts was inversely associated with age, increased duration of therelationship, increased lifetime drug risk behavior, and increased number oftypes of violence. The proportion of protected acts was positively associatedwith increased education, perceived need for male condom use intrusting/faithful relationships, perceived need for condom use in long-termrelationships, convenience of male condom use, perceived condom use norms,acceptability of condom use requests, perceived partner risk, and condom useself-efficacy.
Factorsinversely associated with the proportion of protected coital acts involvingother partners were application of the condom by the man and lifetime drug usebehavior. Other variables positively associated with condom use were theperceived need for condom use in long-term relationships, the convenience ofusing male condoms, the perceived condom use norms, the number of times thewoman put the condom on the man in the past 30 days, the application of condommost often by the man, and condom use self-efficacy. One factor identified inthe theoretical model (i.e., the consequences of requesting condom use) wasnot associated with either outcome.
The binomialregression analysis of protected coital acts with the main partner included 13predictors (Table 3).This analysis included 793 women who reported having sex at least once withtheir main partner. This accounts for 5978 acts of intercourse and 2148 condomusages. Items were included in the multivariate models if they weresignificant at a P value less than0.01 in the bivariate analysis. When a scale parameter of 1 was used, themodel was overdispersed and did not provide adequate fit to the data. Toaccount for the overdispersion, the scale parameter was estimated using thePearson χ 2 statistic divided by the degrees offreedom. The model that accounted for the overdispersion provided adequate fitto the data. Of the 13 factors identified in the bivariate analysis, 12remained significant in the multiple regression analysis. The fit of the modelwas not improved with the inclusion of STD history(χ 2 diff = 2;P > 0.05). Additionally, nointeraction terms improved the fit of the model.
When theanalysis was conducted separately for women who did and those who did notreport a history of STD, some differences were observed(Table 3). Thisanalysis included 628 women who reported a history of STD and 165 who reportedno history of STD. Acceptability of condom use requests, condom useself-efficacy, and condom use convenience were significantly associated withcondom use in both groups of women. Importantly, women who reported a previousSTD were more likely to use a condom if they reported putting condoms on theirpartners, thought condoms were convenient to use, thought condom use wasimportant in long-term relationships, and perceived themselves at risk. Womenwho did not report a history of STD were more likely to have unprotected sexif they had been in a relationship longer and had experienced fewer types ofviolence in their relationship. They were more likely to use condoms if theyreported that condoms were important in trusting relationships(Table 3). Similarmodels were fit using the proportion of condom-protected coital acts withnon-main partners as the outcome. However none of the identified factors weresignificantly associated with condom use. Education could not be considered inthe no-STD history because of the small samplesize.
Thisstudy demonstrated three primary findings. First, this analysis generallysupported the expectations suggested by the theory of planned behavior. In anextensive meta-analysis, Sheeran etal 17 reported similar findings. Atleast one scale from each of three domains that characterize the theory ofplanned behavior was included in the finalmodel.
Second, self-reported STD history appears to bean important modifying factor associated with condom use. Most of the factorsidentified as relating to condom use with main partners differed by STDhistory. Only 3 of the 12 factors included in the model were significant inboth STD history groups.
Third, whereas a number offactors were identified as relating to condom use with regular partners,condom use with other partners was largely unexplained. This analysis,however, may have been limited by the small number of women reporting otherpartners.
Stratification by history of STD allowed amore careful analysis of factors associated with male condom use. Condom usewith the main partner decreased with increased duration of the relationshipand reported relationship violence. However, this was true only for women withno history of STD. For women who reported a history of STD, these factors werenot significant predictors of condom use with main partners. Conversely, someitems of significance in the regression model for women with a history of STDwere not significant for women with no history of STD. For example, the numberof times the women put the condom on the man, the perceived normative condomuse among peers, and the perceived partner risk were significant predictors ofcondom use for women who reported a history of STD, but not for women withouta history of STD.
The result of the analysis stratifiedby partner status also provides important information. The findings of thisstudy support those in the literature suggesting that the factors associatedwith condom use vary by partner type. Interestingly, this study could notidentify any factors associated for both the current main and otherpartner(s).
A potential weakness of this analysis isthat the measure of condom use was based on 30-day recall. These estimatescould reflect an overreporting bias resulting from the social desirability ofcondom use. In addition, the women in the study were recruited from STDclinics, so they may have reported recent increased condom use because ofsuspected STD. Comparison of women who did and those who did not participatein the study does suggest that the study participants were at higher risk. Theselection could explain why age of sexual debut and race or ethnicity were notsignificantly associated with condom use in thisstudy.
Conducting the analysis separately by reportedhistory of STD was a post hoc method for evaluating potential differences incondom use attributable to prior STD. Although some researchers have statedthat a self-reported history of STD is unreliable, the measure of STD historyobtained by naming specific STDs and asking the question “has a doctorever told you” is likely to improve the reliability of the measure. Inaddition, any misclassification of women into the no-history group would biasthe results toward reduction of any difference between groups. Analysis offactors associated with condom use among women with non-main partners byhistory of STD could not be conducted because of the small sample size.Further research needs to be conducted to identify the factors associated withcondom use in this group.
This analysis reports themagnitude of the association between psychosocial factors and condom use bySTD history. Previously published studies often have focused on differences incondom use by partner type but have not investigated the potential role of STDhistory. 6,9,10,23 This analysis of condom use with main partners found that condom useself-efficacy and the woman applying the condom were associated with increasedprotection regardless of STD history. The current findings support previousfindings on determinants of condom use and the need for theoretically basedinterventions addressing these importantpredictors. 40
Thefindings presented in this report demonstrate the need for including STDhistory as a modifying factor in condom use behavior. The differences infactors associated with condom use are likely to reflect the women’sassessment of their risk and the consequent modification of their behavior toreduce that risk based on previous STD experience. In analyses of condom use,the history of STD may be useful for clarifying what factors are important indetermining condom use among women, because STD appears to be an importantmodifying factor. In addition, STD history may be an important factor intargeting public health interventions to increase condom use. In a clinicalsetting, knowing the woman’s STD history can help to focus thecounseling session on identified risk factors. As with risk-reductioncounseling by partner type, this tailored approach may prove to be moreeffective in helping women reduce their STD riskbehavior.
1. Centersfor Disease Control and Prevention. HIV/AIDS Surveillance Report 8(1). Atlanta, GA: U.S. Department of Health and Human Services, Public HealthService, 1996.
2. Instituteof Medicine. The hidden epidemic: confronting sexually transmitteddiseases. Washington DC: National Academy Press, 1997.
3. Centersfor Disease Control and Prevention. Update: barrier protection against HIV andother sexually transmitted disease MMWR Morb Mortal Wkly Rep 1993;42:589–591,597.
4. Cates W, Stone KM. Family planning: sexually transmitted diseases and contraceptive choice: aliterature update–Part 1. Fam Plann Perspect 1992; 24: 75–87.
5. Pinkerton S, Abramson P. Effectiveness of condoms in preventing HIV Infection. Soc Sci Med 1997; 44: 1303–1312.
6. Soskolne V, Aral SO, Magder LS, Reed DS, Bowen GS. Condom use with regular and casual partners among women attendingfamily planning clinics. Fam Plann Perspect 1991; 23: 223–225.
7. Catania JA, Coates TJ, Kegeles S, et al. Condom use in multi-ethnic neighborhoods of SanFrancisco: the population-based AMEN (AIDS in multi-ethnic neighborhoods)study. Am J Public Health 1992; 82: 284–287.
8. Grinstead OA, Peterson JL, Faigeles B, Catania JS. Antibody testing and condom use among heterosexual AfricanAmericans at risk for HIV infection: the National AIDS Behavioral Surveys. Am J Public Health 1997; 87: 857–858.
9. Moore J, Harrison JS, Kay KL, Deren S, Doll LS. Factors associated with Hispanic women’s HIV-relatedcommunication and condom use with male partners. AIDS Care 1995; 7: 415–427.
10. Organista KC, Organista PB, Garcia de Alba JE, Moran MAC, Carrillo LEU. Survey of condom-related beliefs, behaviors, and perceived socialnorms in Mexican migrant laborers. J Community Health 1997; 22: 185–197.
11. Green J, Fulop N, Rocsis A. Determinants of unsafe sex in women. Int J STD AIDS 2000; 11: 777–783.
12. Santelli JS, Kouzis AC, Hoover DR, Polacsek M, Burwell LG, Celentano DD. Stage of behavior change for condom use: the influence of partner type, relationship, and pregnancyfactors. Fam Plann Perspect 1996; 28: 101–107.
13. Corby NH, Wolitski RJ. Condom use with main and other sex partners among high-riskwomen: intervention outcomes and correlates of reduced risk. Drugs Society 1996; 9: 95–96.
14. St.Lawrence JS, Eldridge GD, Reitman D, Little CE, Shelby MC, Brasfield TL. Minimizing participant attrition in panel studies through the useof effective retention and tracking strategies: review and recommendations. Am J Community Psychol 1998; 26: 7–28.
15. Buunk BP, Bakker AB, Siero FW, van den Eijnden RJ, Yzer MC. Predictors of AIDS-preventive behavioral intentions among adultheterosexuals at risk for HIV-infection: extending current models and measures. AIDS Educ Prev 1998; 10: 149–172.
16. Abraham C, Sheeran P. Modeling and modifying young heterosexuals’ HIV-preventivebehavior: a review of theories, findings, and education implications. Patient Educ Couns 1994; 23: 173–186.
17. Sheeran P, Abraham C, Orbell S. Psychosocial correlates of heterosexual condom use: a meta-analysis. Psychol Bull 1999; 125: 90–132.
18. Baker SA, Morrison DM, Carter WB, Verdon MS. Using the theory of reasoned action (TRA) to understand thedecision to use condoms in as STD clinic population. Health Educ Q 1996; 23: 528–542.
19. Morrison-Beedy D. Correlates of HIV risk appraisal in women. Ann Behav Med 1997; 19: 36–41.
20. Zapka JG, Stoddard AM, McCusker J. Social network, support, and influence: relationships with drug use and protective AIDS behavior. AIDS Educ Prev 1993; 5: 352–366.
21. Forsyth AD, Carey MP. Measuring self-efficacy in the context of HIV riskreduction: research challenges and recommendations. Health Psychol 1998; 15: 559–568.
22. Luaby J, Semaan S, Cohen A, et al. Self-efficacy, decisional balance, and stages of change for condomuse among women at risk for HIV infection. Health Educ Res 1998; 13: 343–356.
23. Lansky A, Thomas JC, Earp JA. Partner-specific sexual behaviors among persons with both main andother partners. Fam Plann Perspect 1998; 30: 93–96.
24. Kline A, Kline E, Akin E. Minority women and sexual choice in the age of AIDS. Soc Sci Med 1992; 34: 447–457.
25. Macaluso M, Demand M, Artz L, Hook EW. Partner type and condom use. AIDS 2000; 14: 537–546
26. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Processes 1991; 50: 179–211.
27. Ajzen I, Driver BL. Prediction of leisure participation from behavioral, normative, andcontrol beliefs: an application of the theory of planned behavior. Leisure Sci 1991; 13: 185–204.
28. Reinecke J, Schmidt P, Ajzen I. Application of the theory of planned behavior to adolescents’condom use: a panel study. J Appl Soc Psych 1996; 26: 749–772.
29. Clark LR, Brasseux C, Richmond D, Getson P, D’Angelo LJ. Are adolescents accurate in self-report of frequencies of sexuallytransmitted diseases and pregnancies? J Adolesc Health 1997; 21: 91–96.
30. Orr DP, Fortenberry JD, Blythe MJ. Validity of self-reported sexual behaviors in adolescent womenusing biomarker outcomes. Sex Transm Dis 1997; 24: 261–266.
31. VanDuynhoven YT, Nagelkerke NJ, van de Laar MJ. Reliability of self-reported sexual histories: test–retest and interpartner comparison in a sexuallytransmitted disease clinic. Sex Transm Dis 1999; 26: 33–42.
32. Zellman JS, Weisman CS, Rompalo AM, et al. Condom use to prevent incident STDs: the validity of self-reported condom use. Sex Transm Dis 1995; 22: 15–21.
33. Fishbein M, Jarvis B. Failure to find a behavioral surrogate for STDincidence: what does it really mean? Sex Transm Dis 2000; 27: 452–455.
34. Sobo EJ. Finance, romance, and social support and condom use amongimpoverished inner-city women. Hum Org 1995; 54: 115–128.
35. Ericksen KP, Trocki KF. Behavioral risk factors for sexually transmitted disease inAmerican households. Soc Sci Med 1992; 34: 843–853.
36. Macaluso M, Demand M, Artz L, et al. Female condom use among women at risk of STD. Fam Plann Perspect 2000; 32: 138–144.
37. Artz L, Macaluso M, Brill I, et al. Effectiveness of an intervention promoting the female condom tosexually transmitted disease clinic patients. Am J Public Health 2000; 90: 237–244.
38. MacalusoM, Wang X, Brill I, et al. Participation and retention in a study of femalecondom use among women at high risk of STD. Sex Transm Dis. Inpress.
39. Straus MA. Measuring intrafamily conflict and violence: the Conflict Tactic (CT) Scales. J Marriage Fam 1979; 36: 13–29.
40. Kelly JA. Changing HIV Risk Behavior: Practical Strategies. New York: Guilford Press, 1995.