Legardy, Jennifer K. MPH*; Macaluso, Maurizio MD, DrPH*; Artz, Lynn MD, MPH†; Brill, Ilene MPH†
SEXUALLY TRANSMITTED DISEASES (STD) continue to be a major public health problem in the United States, underscored by alarming disparities by gender and race/ethnicity. Compared with men, women have higher morbidity and mortality rates from STD infections.1,2 STD rates are disproportionately higher among black women than in any other population group. Black women accounted for 64% of reported human immunodeficiency virus cases through 2001, and the majority of these infections were attributed to heterosexual transmission.3
Although both abstinence and a mutually monogamous relationship with a noninfected partner can afford complete protection against most STDs, consistent and correct condom use also can reduce the risk of infection among individuals whose behaviors place them at risk.4–7 Despite previous efforts to promote safer sex practices and consistent condom use to protect against infection, STD rates continue to be high and condom use lower than desirable, especially in high-risk populations. High-risk subjects cite reduced sexual sensation, being perceived as promiscuous by a partner, personal and partner dislike of condoms, and lack of condom availability as reasons for not using condoms.8–10
Mechanical failure (breakage, slippage) and incorrect use11 may reduce the effectiveness of condoms even when consistent use is achieved. Both mechanical failure and incorrect use are influenced by the skills and experience of the user; for example, condom breakage and slippage rates are very low among sex workers.6,12–14 Psychosocial barriers to condom use and condom use skills are typically targeted by behavioral interventions designed to promote safer sex.15–26 On the other hand, even groups at high risk for STD are not homogeneous about beliefs, knowledge, skills, and other key determinants of condom use. Thus, the impact of interventions may vary according to the characteristics of the individual. To address this research question, we analyzed data from a randomized trial of a skill-focused intervention that promoted correct and consistent use of the male condom.27
Methods and Materials
A detailed description of the study protocol has been published elsewhere.27,28 In short, this randomized, controlled trial was a subset of a prospective follow-up study of women who attended an STD clinic in Birmingham, Alabama, during 1992–1995. The purpose of the prospective study was to examine the effectiveness of using condoms and vaginal microbicides to prevent the transmission of STDs. The first 450 women in the prospective follow-up study were enrolled in the randomized trial; 23 were found to be ineligible after randomization. The randomized trial compared 2 behavioral interventions to promote consistent and correct barrier contraceptive use. Women who were between the ages of 18 and 35, were not on long-term antibiotic treatment, were not pregnant or planning to conceive in the next 6 months, and had not had a hysterectomy were eligible for inclusion. After signing the informed consent to participate in this portion of the study, they were asked to return for the initial visit.
At the initial visit, participants were randomly assigned to 1 of 2 intervention groups: enhanced or basic intervention (control). Next, they were interviewed, participated in the assigned intervention, had a clinical pelvic examination, were provided barrier contraceptives, and were trained to use a sexual diary. Monthly follow-up visits were scheduled for a 6-month period.
The components of the basic intervention counseling session included (a) brief advice about how to prevent STDs, (b) information about the effectiveness of barrier contraceptives, (c) free samples of barrier contraceptives (e.g., condoms, film, jelly, foam), and (d) instructional brochures on the proper use of spermicides and condoms.
In the enhanced intervention, the participant viewed a videotape that demonstrated successful strategies for negotiating condom use with partners and endorsed safer sex as the new social norm. During an individual counseling session, a nurse reviewed facts about STDs, corrected misconceptions about STD risk, recommended sexual abstinence or consistent use of barrier contraceptives, and discussed the efficacy of male condoms and vaginal microbicides. The nurse demonstrated correct use of condoms and vaginal microbicides on anatomically correct models and then allowed the participant to practice. Another key component of the enhanced intervention was to strengthen women’s communication and negotiating skills. Women role-played different ways to communicate with their partner about the use of barrier contraceptives and discussed strategies for convincing their partner to use a barrier method and overcoming partner resistance. In addition, women received a packet of materials, including educational brochures, product samples, a condom wallet, instructional booklet in comic book format, an assortment of condoms, greeting cards to give to their partners, and an additional packet of similar materials for a friend.
Data Collection Methods
This analysis used information from the recruitment interview, the initial visit interview, and the sexual diary. Sociodemographic information and reproductive history (i.e., prior birth control methods and STD prevention methods) were gathered during the recruitment interview. The initial visit interview collected more in-depth information on reproductive history, sexual behavior, as well as beliefs, attitudes, and experiences related to condom use.
Participants used the diary to record information prospectively on each coital act. They were asked to code the diary every day, even if they did not engage in a sex act. For each coital act, the women recorded a few key variables: type of intercourse (vaginal, oral on the men, oral by the men, anal), type of protection (no barrier, male condoms, foam, jelly, suppository, vaginal film, diaphragm), type of partner (regular, casual, or new), and any problems (e.g., breakage, slippage, partial use). Women were asked to bring the sexual diary with them to each follow-up visit, along with unused barrier contraceptives and the wrappers of condoms used during the month. This allowed study administrators to verify the number of sexual acts that were protected.
The University of Alabama at Birmingham institutional review board (IRB) and the Alabama Department of Public Health IRB approved all procedures and forms before recruitment began. All names and personal identifiers were removed before researchers at the Centers for Disease Control and Prevention (CDC) received the data. The CDC IRB determined that the present analysis of anonymous data was exempt from review.
The primary objective of these analyses was to assess the influence of participants’ background characteristics on 2 potential effects of a skill-based intervention: (a) consistent condom use (condom use was reported for every act of vaginal intercourse during the interval) and (b) problem-free, consistent use of the male condom (no user problem such as incorrect use, breakage, or slippage were reported during the interval).
We used data triangulation29 to assess whether the information that was reported in the sexual diary was reliable. We cross-referenced the number of condom wrappers that were returned at each follow-up visit with the number of sexual acts in which the women reported using a condom in the sexual diary and with their responses to questions asked retrospectively at each follow-up interview. If there was no difference between the product count form, the sexual diary, and the follow-up questionnaire, the condom-use information was considered to be reliable. There was high (88%) concordance between alternative measures of consistent condom use (κ = 0.76, 95% confidence interval (CI), 0.73–0.79), and there was no indication that receiving the enhanced intervention caused women to over report consistent condom use (data not shown). Thus, the analysis presented in this report uses the diary-based assessment of condom use.
This analysis was restricted to women who reported vaginal intercourse during the follow-up period. Initial frequency tabulations were made to compare the distribution of participants in the basic intervention and the enhanced intervention groups by potential determinants of condom use (age, race, education, household income, marital status, parity, number of lifetime partners, alcohol and drug use, STD history, and prior use of birth control). The association between the intervention and potential determinants of condom use was measured using risk ratios (RR). Stratum-specific RRs were calculated to assess the potential for a participant’s baseline characteristics (i.e., demographic, reproductive, and behavioral) to modify the effects of the intervention on consistent condom use and problem-free, consistent use. Because attrition was a potential threat to the validity of the findings, we also evaluated the distribution of the number of visits attended by intervention group and the distribution of consistency of condom use and months of follow-up, stratified according to whether a woman completed the study protocol or withdrew during follow-up. We assessed these associations both overall and within strata of selected baseline characteristics that modified the estimates of the interventions effectiveness. We used stratified contingency tables and χ2 statistics to formally assess whether attrition was differential by intervention group or by consistency of condom use.
Binomial regression was used to evaluate the intervention’s effectiveness on consistent use and on problem-free, consistent use as a function of multiple baseline characteristics. To assess the heterogeneity of intervention effects (RRs) across strata of baseline characteristics, we used Breslow-Day χ2 tests for stratified tables and included appropriate interaction terms in regression models. Generalized estimating equations were used in these models to explicitly model the correlation between multiple follow-up interval visits pertaining to the same person.30
Four hundred twenty-seven women were randomly assigned to either the enhanced (N = 213) or basic intervention (control) group (N = 214). The 2 groups had similar baseline characteristics (Tables 1–3). Approximately 71% of the women were 21 to 30 years old, and more than 87% were black. Sixty-eight percent of the enhanced intervention group and 74% of the control group were single. About two-thirds of participants (65% enhanced, 61% basic) had some high school education. Only 45% of the women enrolled in the study were employed, and, of those, 75% earned ≤$600/month. The majority of participants used some form of modern birth control before entry in the study; however, only 77 participants had used condoms as their only form of birth control.
Mean (standard deviation) age at first intercourse was 15.5 (2.4) years in the enhanced intervention group and 15.5 (2.6) years in the control group, and the mean age of first pregnancy was 17.8 (3.0) years in the enhanced group and 17.8 (3.2) years in the control group (data not shown). For both groups, more than half of the women had had ≤1 child. About 85% of women in both groups had a history of STDs (Table 2). The majority of women reported that they rarely or never used alcohol and drugs before having sex (91% enhanced, 93% basic) in the previous month (Table 3). At enrollment, most women reported having 1 sex partner (72% enhanced, 62% basic) during the previous month.
Attrition was substantial; 39% of the enhanced intervention group and 44% of the basic intervention group withdrew before completing the study (P = 0.37). Although there was some variability in the length of the interval between visits, most intervals were 4 to 6 weeks in duration. For simplicity, we refer to all intervals as “months of follow-up.” At the first month’s follow-up visit, the proportion of women achieving 100% consistent condom use was 45% in the enhanced intervention group compared to 30% in the basic intervention group (P = 0.008). Both rates gradually increased with each follow-up visit. By the 6-month follow-up visit, the proportion of consistent users increased to 57% in the enhanced intervention group and to 40% in the basic intervention group (P = 0.01). This increase in consistent use during follow-up did not appear to be a result of selective attrition of inconsistent users (P = 0.16). During the first month of follow-up, 77% of the consistent users in the enhanced intervention group experienced problem-free, consistent use compared to 80% in the basic intervention group (P = 0.7). By the sixth month of follow-up, however, the rate of problem-free, consistent condom use had increased to 84% in the enhanced intervention group and 80% in the basic intervention group (P = 0.6) (data not shown).
Effect of the Intervention on Consistent Condom Use
On average, the consistent-use rate was 60% higher in the enhanced intervention group than in the basic intervention group (RR, 1.6; 95% CI, 1.4–1.8). Among women in the basic intervention group who were ≤20 years old at enrollment, 29% of 100 months of follow-up were consistent-use months (Table 1). By contrast, the proportion of consistent-use months among women of the same age category in the enhanced intervention group was 69% (RR, 2.4; 95% CI, 1.7–3.4). The proportions of consistent-use months were 20% and 90% higher for the enhanced intervention group among women aged 21 to 25 and ≥26 years, respectively. Also, the effect of the intervention was stronger among women who received food stamps (RR, 2.0; 95% CI, 1.6–2.4) than among women who did not (RR, 1.4; 95% CI, 1.2–1.6) (RRs significantly different, P = 0.009).
Age at sexual debut appeared to modify intervention effectiveness. Among women who first engaged in sexual intercourse on a regular basis at 15 to 16 years of age, those in the enhanced intervention group were about 2 times more likely to use condoms consistently compared with those in the basic intervention group. However, the intervention effect was the opposite (about 40% more consistent use in the basic intervention group) among women whose sexual debut was at ≥20 years of age (Table 2).
The effect of the enhanced intervention on consistent condom use was higher among women who had 3 or more children (RR, 2.3; 95% CI, 1.6–3.2) than among women who had fewer children, although this difference did not achieve statistical significance (P = 0.07). Also, among women who used a user-independent birth control method (i.e., tubal ligation, intrauterine device, Norplant, Depo-Provera) (RR, 2.1; 95% CI, 1.7–2.7) or oral contraceptives (RR, 1.8; 95% CI, 1.4–2.2) the intervention was more effective than among women who used no contraceptive method (RR, 1.3; 95% CI, 0.8–1.9) (P = 0.01).
The intervention appeared to have a strong effect among married women who reported having had sexual intercourse only at their homes during the 30 days before enrollment (RR, 4.6; 95% CI, 2.2–9.6). It was less effective among women who were married and did not always have sexual intercourse at their homes (RR, 0.8; 95% CI, 0.4–1.7) (Table 3). Among women in the enhanced intervention group who responded “definitely yes” to the question about whether they would ask their partner to use condoms next time they had sex, the intervention had a small but statistically significant effect (RR, 1.2; 95% CI, 1.1–1.4), while the effect appeared to be the largest among women who answered “definitely no” (RR, 2.8; 95% CI, 1.5–5.3) (P = 0.002). This apparently counterintuitive pattern of intervention effectiveness was inversely correlated with rates of condom use in the control group, which were highest (47%) for women who answered “definitely yes” and lowest (11.5%) for women who answered “definitely no.” Thus, the intervention appeared to be least effective in a subgroup that was very likely to use condoms consistently even in the absence of the intervention. Also, the intervention effect appeared to be higher among women who always or sometimes used alcohol and drugs before sex (RR, 5.3; 95% CI, 2.0–13.9) compared with women who rarely or never did (RR, 1.6; 95% CI, 1.4–1.8) (P = 0.01).
The impact of variables that were significant modifiers of the intervention’s effectiveness in simple analyses was further assessed in a binomial regression model for repeated observations. Age at sexual debut (began having sex ≥once a month), substance use before sexual intercourse (past month), marital status and place of intercourse (at baseline), and intention to use a condom at next intercourse were included in the model, along with the intervention effect and the interaction terms to model effect modification.
Using this model, a number of factors were shown to modify the intervention effect. The effect was higher among women whose sexual debut occurred at an early age (RR, 1.9; 95% CI, 1.5–2.4), women who frequently engaged in alcohol and substance use before intercourse (RR, 5.4; 95% CI, 1.5–19.9), married women who had sex only at their homes (RR, 3.6; 95% CI, 1.5–8.4), and women who answered “maybe yes/maybe no/definitely no” to the question of whether they would ask their partner to use condoms the next time they had sex (RR, 2.5; 95% CI, 1.7–3.6). There was no significant intervention effect for women whose sexual debut occurred at ≥20 years of age (Table 4).
Effect of the Intervention on Problem-Free, Consistent Condom Use
No mechanical failure or incorrect use was reported during 82% of 409 consistent-use months in the enhanced intervention group and during 79% of 253 consistent-use months in the control group (not significantly different). Women in the enhanced intervention group did not have higher problem-free, consistent condom use rates compared with women in the basic intervention group (RR, 1.0; 95% CI, 0.9–1.1). Demographic, reproductive, and behavioral/psychosocial characteristics did not appear to modify the (null) effect of the intervention on problem-free, consistent condom use (data not shown).
Our findings show that the enhanced intervention significantly increased consistent condom use among participants, as did a previous intent-to-treat analysis of this intervention.27 Results from other randomized interventions targeting women at high risk for STDs have reported increases in condom use among participants.15,20,24,26 Our skill-based intervention helped some at-risk women achieve consistent (100%) condom use. This suggests that women who responded to the intervention became more capable of negotiating condom use or refusing sexual intercourse with men who were not willing to use a condom.
Although increasing consistent condom use is the typical objective of many STD prevention interventions, correct and problem-free condom use is important for ensuring their effectiveness. During months in which participants achieved consistent use, we found no significant difference between intervention groups with regard to the proportion of problem-free, consistent use months. Because the enhanced intervention was not directly administered to males, who are usually responsible for putting on condoms, our intervention had a limited potential to improve correct use or reduce mechanical failure rates. However, by promoting consistent condom use among nonusers and encouraging women to be active and put condoms on their partners, the enhanced intervention could in fact have resulted in more frequent usage problems. Our results indicate that the intervention did not detrimentally affect problem-free, consistent condom use.
This analysis focused mainly on whether the women’s baseline characteristics modified the effectiveness of the intervention. Indeed, we found signs of effect modification associated with several demographic and behavioral characteristics. Most notably, intervention effectiveness was greatest among women who were not at all confident they would use condoms the next time they had sexual intercourse. The intervention’s emphasis on skills training was apparently successful in improving these women’s self-efficacy, which translated into greater condom use. In contrast, the intervention was less effective among women who were very confident they would use condoms the next time they had sex; these women were already at a high level of consistent condom use (56%) at baseline but still far from an optimal level. Perhaps some were overconfident and failed to attend as closely to the intervention, or perhaps the intervention did not adequately address important predictors of condom use for this subgroup.
The intervention effect also appeared to be stronger among women who had initiated sex at an early age and those who reported recent and frequent substance use before sex. These women are commonly considered to be at high risk for STDs and may have been motivated by the intervention to reassess the magnitude of their own risk and to achieve consistent condom use.
Last, we determined that married women who had sex only at home were the least consistent condom users at baseline compared with all other groups. They may not use condoms as often because they do not perceive their risk or because condom use is not the norm in this population. The increase in consistency of condom use after receiving the intervention was larger in this group than in other groups defined by marital status and where they have sex. This may be a result of the intervention’s success in addressing misconceptions about STD risk and increasing participants’ perception of their own level of risk. By promoting this reassessment of personal risk, fostering the acceptance of new social norms, and improving the women’s negotiation skills, the intervention may have both motivated and enabled them to effectively negotiate condom use in their sexual partnership.
The findings of this analysis should be interpreted in light of certain limitations. First, the study population was predominantly black women in their twenties. Thus, our results may not be generalized to other age or racial/ethnic categories. A second potential limitation is attrition. The basic intervention group experienced a higher attrition rate than the enhanced intervention group. However, the group differences did not achieve statistical significance and are unlikely to explain any of the findings. In addition, attrition was not associated with consistent condom use during follow-up, thus making it unlikely that the assessment of intervention effectiveness was substantially distorted.
The study limitations were offset by considerable strengths. Data triangulation allowed us to validate self-reporting of consistent condom use in the sexual diary. By cross-referencing the sexual diary with the product count forms, we were able to identify overreporting and reduce misclassification of condom use. Other studies have shown that validity of interview data is strengthened by data triangulation.31,32
The analyses presented here do not fully benefit from the effect of randomization as the study participants were not randomly assigned to the intervention groups within blocks defined by baseline characteristics. Thus, our findings should be interpreted as observations on the variability of the intervention effect, not as the results of a trial designed to study interactions. Nevertheless, the randomized design of this trial strengthens the internal validity of the comparisons, as attested to by the close similarity of the intervention groups about baseline characteristics.
In combination with the findings of the intent-to-treat analysis of this intervention trial,27 this analysis highlights 2 key issues when designing and evaluating public health programs intended to promote condom use for the prevention of STDs. First, increasing condom use remains a primary focus for STD prevention, but interventions may not be adequately addressing this issue if use effectiveness (i.e., ensuring correctness of use) is not improved. Second, an intervention may be most beneficial when it is designed to address the specific needs of the target population. We found that our intervention was most beneficial among women who may have had a low perception of their risk and who had low self-efficacy about condom use. Whereas it is desirable to develop programs based only on interventions of proven effectiveness, such interventions may have been designed for populations with a particular profile of behavioral deficits. Thus, it may be also advisable to carry out preimplementation needs assessment to verify that the intervention design matches the needs of the population.
In conclusion, our intervention effectively increased both the rate of condom use, and the proportion of months in which participants maintained consistent condom use. The study also demonstrated that intervention effects were modified by certain baseline demographic, reproductive, and behavioral characteristics of the participants. The nature of, and reasons for, effect modification in intervention research are rarely studied. Our findings indicate that this type of research may help clarify the differential impact of interventions on specific population subgroups and may assist in the design of more effectively tailored interventions.
1. Ebrahim SH, Peterman TA, Zaidi AA, Kamb ML. Mortality related to sexually transmitted diseases in US women, 1973 through 1992. Am J Public Health 1997; 87:938–944.
2. Sexually transmitted diseases in women: a threat to women’s health. J Midwifery Womens Health 2002; 47:288–289.
3. Centers for Disease Control and Prevention. HIV/AIDS surveillance report 2001. 2001; 13:1–44.
4. Centers for Disease Control and Prevention. Update: barrier protection against HIV infection and other sexually transmitted diseases. Morb Mortal Wkly Rep 1993; 42:589–591.
5. Cates W, Stone KM. Family planning, sexually transmitted disease and contraceptive choice: a literature update: part I. Fam Plann Perspect 1992; 24:75–84.
6. Albert AE, Warner DL, Hatcher RA, Trussell J, Bennett C. Condom use among female commercial sex workers in Nevada’s legal brothels. Am J Public Health 1995; 85:1514–1520.
7. Roper WL, Peterson HB, Curran JW. Commentary: condoms and HIV/STD prevention-clarifying the message. Am J Public Health 1993; 83:501–503.
8. Rosenthal SL, Biro FM, Succop PA, Baker JG, Stanberry LR. Reasons for condom utilization among high-risk adolescent girls. Clin Pediatr (Phila) 1994; 33:706–711.
9. Weinstock HS, Lindan C, Bolan G, Kegeles SM, Hearst N. Factors associated with condom use in a high-risk heterosexual population. Sex Transm Dis 1993; 20:14–20.
10. Jadack RA, Fresia A, Rompalo AM, Zenilman J. Reasons for not using condoms of clients at urban sexually transmitted disease clinics. Sex Transm Dis 2002; 24:402–408.
11. Warner L, Clay-Warner J, Boles J, Williamson J. Assessing condom use practices: implications for evaluating method and user effectiveness. Sex Transm Dis 1998; 25:273–277.
12. de Graaf R, Vanwesenbeeck I, van Zessen G, Straver CJ, Visser JH. The effectiveness of condom use in heterosexual prostitution in the Netherlands. AIDS 1993; 7:265–269.
13. Wong ML, Chan RKW, Koh D, Wee S. A prospective study on condom slippage and breakage among female brothel-based sex workers in Singapore. Sex Transm Dis 2000; 27:208–214.
14. Rugpao S, Beyrer C, Tovanabutra S, et al. Multiple condom use and decreased condom breakage and slippage in Thailand. J Acquir Immun Defic Syndr Hum Retrovirol 1997; 14:169–173.
15. Belcher L, Kalichman SC, Topping M, et al. A randomized trial of a brief HIV risk reduction counseling intervention for women. J Consult Clin Psychol 1998; 66:856–861.
16. National Institute of Mental Health (NIMH) Multisite HIV Prevention Trial Group. The NIMH Multisite HIV Prevention Trial: reducing HIV sexual risk behavior. Science 1998; 280:1889–1894.
17. Ehrhardt AA, Exner TM, Hoffman S, et al. A gender-specific HIV/STD risk reduction intervention for women in a health care setting: short- and long-term results of a randomized clinical trial. AIDS Care 2002; 14:147–161.
18. Carey MP, Braaten LS, Maisto SA, et al. Using information, motivational enhancement, and skills training to reduce the risk of HIV infection for low-income urban women: a second randomized clinical trial. Health Psychol 2000; 19:3–11.
19. Dancy BL, Marcantonio R, Norr K. The long-term effectiveness of an HIV prevention intervention for low-income African-American women. AIDS Educ Prev 2000; 12:113–125.
20. DiClemente RJ, Wingood GM. A randomized controlled trial of an HIV sexual risk-reduction intervention for young African-American women. JAMA 1995; 274:1271–1276.
21. Gillmore MR, Morrison DM, Richey CA, Balassone ML, Gutierrez L, Farris M. Effects of a skill-based intervention to encourage condom use among high risk heterosexually active adolescents. AIDS Educ Prev 1997; 9(suppl A):22–43.
22. Gollub E, French P, Loundou A, Latka M, Rogers C, Stein Z. A randomized trial of hierarchical counseling in a short, clinic-based intervention to reduce the risk of sexually transmitted diseases in women. AIDS 2000; 14:1249–1255.
23. Kamb ML, Fishbein M, Douglas JM, et al. Efficacy of risk-reduction counseling to prevent human immunodeficiency virus and sexually transmitted disease: a randomized controlled trial. JAMA 1998; 280:1161–1167.
24. Kelly JA, Murphy DA, Washington CD, et al. The effects of HIV/AIDS intervention groups for high-risk women in urban clinics. Am J Public Health 1994; 84:1918–1922.
25. Scholes D, McBride CM, Grothaus L, et al. A tailored minimal self-help intervention to promote condom use in young women: results from a randomized trial. AIDS 2003; 17:1547–1556.
26. Shain RN, Piper JM, Newton ER, et al. A randomized, controlled trial of a behavioral intervention to prevent sexually transmitted disease among minority women. N Engl J Med 1999; 340:93–99.
27. Artz L, Macaluso M, Meinzen-Derr J, Kelaghan J, Austin H, Fleenor M, Hook III EW, Brill I. A randomized trial of clinician-delivered interventions promoting barrier contraception for sexually transmitted disease prevention. Sex Transm Dis 2005; 32:672–679.
28. Macaluso M, Artz L, Kelaghan J, Austin H, Fleenor M, Hook EW. Prospective study of barrier contraception for the prevention of sexually transmitted disease. Sex Transm Dis 1999; 26:127–136.
29. Flick U. Triangulation revisited: strategy of validation or alternative? J Theory Soc Behav 1992; 22:197.
30. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986; 4:121–130.
31. Upchurch DM, Weisman CS, Shepherd M, et al. Interpartner reliability of reporting of recent sexual behaviors. Am J Epidemiol 1991; 134:1159–1166.
32. James NJ, Bignell CJ, Gillies PA. The reliability of self-reported sexual behaviour. AIDS 1991; 5:333–336.