WILSON, TRACEY E. PhD*; KOENIG, LINDA J. PhD†; WALTER, EMMANUAL MD‡; FERNANDEZ, ISABEL PhD§; ETHIER, KATHLEEN PhD†; FOR THE PERINATAL GUIDELINES EVALUATION PROJECT
State University of New York Downstate Medical Center, Brooklyn, New York; the †Centers for Disease Control and Prevention, Atlanta, Georgia; ‡Duke University Medical Center, Durham, North Carolina; and the §University of Miami School of Medicine, Miami, Florida
The Perinatal Guidelines Evaluation Project was funded by the Centers for Disease Control and Prevention (CDC) through cooperative agreements U64CCU412273 (Duke), U64CCU112274 (Yale), U64CCU412294 (Miami), and U64CCU212267 (SUNY). All procedures and consent forms were approved by the institutional review boards at the Centers for Disease Control and Prevention and each participating university and hospital. The authors thank the members of the study staff who assisted with data collection and project management and the women who participated in the study. Members of the Perinatal Guidelines Evaluation Project Group include Howard Minkoff, MD, Jack DeHovitz, MD, and Tracey Wilson, PhD (New York); Rachel Royce, PhD, and Emmanuel Walter, MD (North Carolina); Mary Jo O'Sullivan, MD, and M. Isabel Fernandez, PhD (Florida); Jeannette Ickovics, PhD, and Kathleen Ethier, PhD (Connecticut); and Linda Koenig, PhD, R. J. Simonds, MD, Jan Moore, PhD, Margaret Lampe, RN, Daniel Whitaker, PhD, Daphne Cobb, MS, and Ken Dominguez, MD (CDC).
Correspondence: Tracey E. Wilson, PhD, Department of Preventive Medicine and Community Health, State University of New York, Downstate Medical Center, Box 1240, 450 Clarkson Avenue, Brooklyn, NY 11203. E-mail: Tracey.Wilson@downstate.edu
Received for publication January 3, 2003,
revised May 21, 2003, and accepted May 27, 2003.
THE DEGREE OF PROTECTION afforded from unintended pregnancy and sexually transmitted disease (STD) varies with the form of contraception used. Methods that are most efficacious for preventing pregnancy such as oral contraceptive pills are not effective for preventing STD, whereas those that are most effective for preventing disease transmission such as condoms can be less useful for pregnancy prevention, depending on how consistently and correctly they are used. For example, although condoms are useful for preventing transmission of HIV and many other STDs, 1 their failure rate in relation to pregnancy prevention is as high as 14% over the first year of use. 2 Many women find this failure rate less than acceptable, and consequently, might opt to use multiple methods of protection to avoid both unwanted pregnancy and disease. The use of 2 methods, when one of these is a male condom, is currently the best available option for those who desire simultaneous protection from both pregnancy and STDs.
Many studies, however, have noted that one consequence of using multiple methods, or “dual use,” is decreased overall protection from STDs as a result of inconsistency of condom use. 3 For instance, results from many studies have repeatedly revealed a lower level of condom use consistency among women who also report either being infertile/sterilized or who use a hormonal method of contraception. 4-9 These studies indicate that disease prevention is inextricably linked with reproductive concerns and suggest an important paradox. Specifically, by using a strategy that confers maximum protection from pregnancy, individuals might be placing themselves at greater risk for disease through less effective condom use. The implications of these findings are of considerable importance in light of behavioral data indicating increased reliance on dual methods among women in the United States. 10
In general, however, most of these studies share a very basic limitation in that they fail to measure whether dual use over a defined time period occurs episodically (ie, at each instance of sexual intercourse) or whether different methods are used in different situations (eg, a condom can be used as a “backup” method in situations in which oral contraceptive doses are missed or when other methods are not available, or when intercourse occurs with a secondary rather than a primary partner). In other words, these analyses fail to assess behavioral patterns on an event-specific level. It would be expected that these 2 groups vary considerably in the consistency of condom use and in factors that are associated with these behaviors. These differences would, in turn, be important in designing behavioral interventions targeting maximal condom use consistency in dual-use circumstances.
The current study addresses this issue through an examination of condom use patterns in a cohort of HIV-infected and at-risk women. This population is ideally suited for the study of this issue, given the varying levels of disease transmission and prevention concerns likely to exist between the 2 cohorts. Specifically, we delineate 1) patterns of condom-only versus dual method use behaviors, 2) the proportion of dual-method users who use both methods at every occurrence of vaginal sex versus those who do not use them together (“simultaneous” vs. “alternating” dual users), and 3) factors that differentiate these distinct groups. Finally, we highlight the importance of understanding and targeting dual-method use on an event-level basis as a tool for the promotion of safer sexual behaviors.
Materials and Methods
The Perinatal Guidelines Evaluation Project HIV and Pregnancy Study (PGEP) was a prospective cohort study designed to examine behavioral, clinical, and psychosocial factors related to pregnancy in the context of HIV-1 infection in the United States. Enrollees in the full study included 634 pregnant women recruited from 31 prenatal care clinics across several cities in 4 states, including Miami, FL, Brooklyn, NY, and various locations throughout Connecticut and North Carolina. Women were identified either through provider referral or through a screening form administered in prenatal care clinics, and began study participation between October 1996 and October 1998. Within each geographic area, clinics were chosen based on their ability to recruit the largest number of HIV-positive pregnant women. Women in the HIV-seronegative cohort were also recruited from prenatal clinics, and were matched within each state to within 5% of the proportion of HIV-seropositive women on 1) HIV risk factors (ever injected drugs, had sex with a man who injects drugs, used crack cocaine, and/or exchanged sex for drugs or money); 2) self-reported primary racial/ethnic identity (black, Latino, white, or other); and 3) registration for prenatal care either before or at or after 20 weeks gestation.
All women in the PGEP completed standardized interviews that were implemented by trained study personnel at 3 time periods: in the prenatal period (at or after 24 wks gestation but before delivery), 6 weeks after delivery, and at 6 months postpartum. The current analyses are based on attitudes and behaviors reported at the 6-month interview; 486 (77%) women completed this interview (258 HIV-seropositive, 228 HIV-seronegative). Women who completed the 6-month interview were older (mean, 28 vs. 26, P <0.05), more likely to be black (83% vs. 74%, P <0.01), and had more pregnancies before recruitment (mean, 3 vs. 2, P <0.05) compared with women who did not complete the follow-up interview. No statistically significant differences were detected between those lost to follow up versus those completing the study protocol as a function of HIV serostatus or education level (high school level or higher). Interviews were administered by trained study staff and conducted in Spanish, Creole, and English, reflecting the languages most frequently spoken in our cohort. Study procedures were approved by institutional review boards at both the Centers for Disease Control and Prevention and at participating academic and medical institutions. Detailed study design and methods are described elsewhere. 11
Sociodemographic information was collected at baseline. At the 6-month interview, women were asked to report on several other variables. These included the prevalence of sexual activity since the birth of their child (we focused on heterosexual partnerships to study patterns of male condom use), the prevalence of male condom use, and the consistency of condom use (always versus not always). We asked women whether they had used any other contraceptive methods during that time, including tubal ligation, and to identify what other methods were used. When women reported using condoms plus at least one other method, we asked whether condoms were used together with other contraceptive methods during each act of vaginal intercourse or whether there were instances in which these methods were used separately. Women who reported using these methods together during episodes of intercourse were classified as “simultaneous” users; other dual-method users were categorized as “alternating” dual users.
We also asked participants to describe drug use behaviors (prevalence of alcohol, cocaine, heroin, or other injection drug use since the baby's birth) and how “upset” the respondent would feel emotionally if she were to become pregnant in the next 6 months (not at all or a little upset vs. very upset). Finally, we created a broad index reflecting the risk of STD/HIV disease transmission to and from male partners based on whether 1) the woman reported 2 or more male sexual partners since the baby's birth, and 2) whether any of these partners could be classified as having increased risk for STD/HIV (ie, a man the woman knows or suspects injects drugs, has sex with other men, has been in jail or prison, has HIV or AIDS, or with whom she has exchanged sex for money or drugs). Women reporting 2 or more partners or at least one risk factor were classified into the higher risk category, whereas those with only one partner and who reported no known risk factors were classified into the lower risk category. This variable generally reflects risks of either STD/HIV transmission to or from an infected partner, and could also represent risk of HIV superinfection among those who already have HIV.
Women were excluded from analysis if they reported that they were pregnant at the 6-month interview (n = 11) or if they were not sure if they were currently pregnant (n = 2); thus, 2.7% of the sample was removed for either reason. An additional 17 women were excluded (3.5%) as a result of missing data across behavioral outcomes of interest (sexual and contraceptive behaviors). The total sample included 456 women, or 94% of those completing the 6-month interview. Correlates of dual-method use as a function of HIV serostatus were analyzed among those women reporting at least one episode of vaginal sex since the birth of their child (n = 361) using logistic regression. We also used multiple logistic regression to assess correlates of dual-use type (simultaneous vs. alternating) among the 134 dual users who also provided data on consistency of condom use (n = 8 missing). To assess whether different types of dual-method users were more likely to also use methods of contraception that varied in the extent to which they were susceptible to adherence-related failures, we included in this model whether women's other methods included oral contraceptives, which require daily adherence, and whether they reported methods depending less on adherence (intrauterine device, Depo-Provera, Norplant, or whether they had a tubal ligation). Other variables in this model included HIV serostatus, alcohol use, pregnancy attitude, and perceived male partner risk.
Women in this study (n = 456) identified primarily as either black (67.4%) or Hispanic (22.5%), and over one fourth (28.9%) were born outside of the United States. The median monthly household income was $800.00, and the mean age of respondents was 28 years (standard deviation [SD], 6.6). There were no statistically significant differences in age, race/ethnicity, country of origin, or household income as a function of HIV serostatus. In the 6 months after birth, 361 (79.2%) of women reported engaging in at least one act of vaginal sex; those with HIV infection were less likely to report any sexual activity in the postpartum period than were those without infection (HIV-negative, 85%; HIV-positive, 74%; OR, 2.0; 95% CI, 1.25-3.21).
The majority of sexually active women reported using condoms with their male partners (n = 250; 69.2%), with 66% of these reporting always versus not always using condoms (Table 1). Those who reported consistent condom use were more likely to be HIV-seropositive (OR, 2.25; 95% CI, 1.30-3.87). Among sexually active women, 9% reported having 2 or more partners (mean partners, 1.5; SD, 8.6); the likelihood of having multiple sexual partners did not vary as a function of HIV serostatus. One third of women (33%) reported that their sexual partner had at least one risk factor for STD/HIV. As might be expected, HIV-seropositive women were more likely to report a partner with at least one risk factor (OR, 2.57; 95% CI, 1.63-4.05). Postpartum alcohol use was reported by 28% of women and postpartum drug use by 7%.
Dual use of condoms and another contraceptive method was reported by 39.3% of sexually active women, 29.9% reported condoms only, 20.5% reported no condom use but use of other methods, and 10.2% reported using no methods for disease or pregnancy prevention. As compared with condom-only users (HIV-negative, 20.9%; HIV-positive, 39.1%), dual users were not significantly more likely to be HIV-seropositive (HIV-negative, 31.3%; HIV-positive, 47.5%; OR, 0.81; 95% CI, 0.48-1.4). In contrast, those who reported using either no methods (HIV-negative, 15.9%; HIV-positive, 4.5%; OR, 0.15; 95% CI, 0.08-0.30) or using other methods excluding condoms (HIV-negative, 31.9%; HIV-positive, 8.9%; OR, 0.15; 95% CI, 0.06-0.36) were less likely to be HIV-seropositive as compared with condom-only users.
Among 142 dual users in this sample, data on patterns of use (simultaneous vs. alternating) were available from 131 (92.2%) participants; 84 of these dual users (64.1%) were classified as being simultaneous users. As a validation of our measure of dual use, we regressed condom use consistency (always vs. not always) on whether women reported using condoms alone, simultaneous dual use, or alternating dual use. Compared with women who reported using only condoms during vaginal sex, simultaneous users were more likely to be consistent condom users (OR, 4.2; 95% CI, 2.5-7.1). On the other hand, alternating users were significantly less likely to report consistent condom use than were those who reported relying solely on condoms (OR, 0.2; 95% CI, 0.1-0.3).
Women who reported simultaneous dual use differed from alternating dual users on a number of characteristics (Table 2). After controlling for other variables in the model, simultaneous dual use versus alternative use was positively associated with being HIV-seropositive (adjusted OR, 2.7), with reporting that a pregnancy would be upsetting if it were to occur in the next 6 months (adjusted OR, 2.4), and with not using alcohol in the postpartum period (OR, 3.7). Patterns of dual use were not associated with partner risk factors, oral contraceptive use as the other reported method of contraception used, or with the use of Norplant, Depo-Provera, intrauterine device use, or tubal ligation as the other reported method.
Past research has indicated that among heterosexual women, the use of dual contraceptive methods, which theoretically confers the most effective means for the prevention of both unintended pregnancy and STD, results in significantly less condom use consistency when compared with single-method male condom use. 4-9 A more detailed analysis of this issue clarifies that this association only exists when dual-method use is characterized by alternating, nonsimultaneous use of both methods. Consequently, it is important to differentiate dual-method users who alternate methods across acts of sexual intercourse versus those who use dual methods simultaneously.
Nearly 40% of our population reported dual-method use, and the majority of these (64%) reported that these methods were used simultaneously during sexual encounters. Women reporting alternating dual-method use differed from simultaneous users in a number of ways. One way in which they differed was that simultaneous users were more likely to be HIV-seropositive. Thus, there was a lower use effectiveness rate among the cohort of uninfected women as compared with HIV-infected women, meaning that the uninfected women are less likely to report using the methods correctly and consistently.
A second difference between these groups is that simultaneous dual users were less likely to report alcohol use in the postpartum period. Intervention efforts geared toward consistent dual use, then, might require skill-building for condom use in cases in which alcohol is likely to be consumed before intercourse. Beliefs about emotional reactions to a pregnancy should one occur were also related to the likelihood of using 2 methods simultaneously; those who perceived that a pregnancy would have a negative emotional impact were more likely to use methods together during each act of sex. It is clear that dual use is motivated in part by pregnancy prevention; therefore, an exploration of the impact of an unplanned pregnancy could help motivate simultaneous use in this cohort.
Of interest, the potential risk of sexual partners in terms of STD/HIV transmission was not associated with simultaneous versus alternate use of dual methods. In addition, no differences were detected in the use of more effective contraceptive methods (eg, oral contraceptives, tubal ligation, Norplant, intrauterine device, Depo-Provera) as a function of dual-use pattern. It might be then that the motivation for consistent condom use among dual users is the desire for additional pregnancy protection in these groups, and not disease prevention. Research is needed to more clearly delineate psychosocial factors influencing the decision to engage in different patterns of dual-method use.
Women in this sample were either infected with HIV or were defined as being at risk for disease acquisition through self-reported risk behaviors. For this reason, our data do not reflect patterns of dual use among a broader population of women. In addition, concerns about disease and/or pregnancy prevention could play a far different role in decisions about dual use among those at lower risk for HIV disease acquisition or transmission. Research is needed to study patterns, rates, and determinants of dual-method use across diverse populations to target interventions appropriately to different risk groups.
Previous studies report that dual use in general puts women at risk for disease acquisition or for unintended pregnancy; our research can be used to illustrate the point that these relationships are restricted to a subgroup of dual users. Among some dual users, the degree of condom consistency is similar to or higher than rates seen among those who use condoms alone. A key point is that condom-use consistency is tied to whether dual use occurs during each act of sexual intercourse or whether there are times that one method might be used in place of another. These 2 groups differ in terms of pregnancy beliefs, alcohol use, and HIV serostatus. Interventions promoting protection from pregnancy and STD/HIV should explore the role of these issues in a woman's life.