Abstract
Background:
Alcohol and drug use by women is related to high-risk sexual practices and protective behaviors.
Objectives:
To determine sexual risk and protective behaviors using information about women’s drug use immediately before or during sex.
Methods:
Latent class analysis using PROC LCA in SAS software was used to determine classes of women using both past 30-day drug use and before or during sex. Participants were recruited from a community-based research site located in a low socio-economic area of Los Angeles County and completed the Risk Behavior Assessment which elicits information on drug and sex risk behaviors.
Results:
The Risk Behavior Assessment, and HIV and sexually transmitted infections testing was obtained on 812 women. Five distinct groups were identified by PROC LCA: An Abstinent group comprised of 26% of participants; an Alcohol and Marijuana group (16%); an Amphetamine group (11%); a No Sex-with-Alcohol group (37%); and a Poly Drug group (11%). Multinomial logistic regression revealed that sexual behaviors and condom use were different across the five groups: The Alcohol and Marijuana group had a higher odds of vaginal intercourse, while the No Sex-with-Alcohol group was most likely to use condoms for vaginal intercourse. The Poly Drug group had the highest risk for anal intercourse while the Amphetamine and Poly Drug groups had high proportions of women with injection-drug using and men-who-have-sex-with-men sexual partners.
Conclusion:
Identifying women based on drug use immediately before or during sex can help providers understand prevention and risk-reduction practices and interventions for drug-using women.
Keywords: women, drug use, STI and HIV risk in women, condom use, alcohol use
Background
Women’s risk for human immunodeficiency virus (HIV) and sexually transmitted infections (STIs) has been associated with sexual behaviors, such as multiple sexual partners and unprotected vaginal and anal intercourse (VI, AI), partners who are known to be HIV-positive, as well as having an injection-drug-using (IDU) partner. In a recent study of STIs comparing men and women across the U.S., women accounted for more cases of Chlamydia, gonorrhea, herpes simplex-2 and Trichomonas, while men accounted for higher numbers of cases of HIV, hepatitis B (HBV) and syphilis (1); however women still accounted for 19% of HIV cases (2). Heterosexual anal intercourse has been associated with increased risk of STIs and was more prevalent among women who used drugs, especially stimulants such as amphetamine and crack cocaine (3–7). Multiple sexual partners conferred additional risk of HIV and STI acquisition, especially among ethnic minority women, as did concurrent sexual partners and partners recently released from incarceration (8–10).
While much is known about sexual and drug use risks in women, there is still much to be understood about drug use in the context of women’s sexual partnerships. Previous research has delineated high-risk subgroups of women who use drugs in the context of sex trading or sexual partnerships. Specifically, an early study of sex-trading women and HIV found that women who only traded for money were more likely to use condoms (11) and research has focused on reasons for sex-trading (12, 13) or whether sex trading occurred before or after IDU initiation (14).
One study of women and IDU found that women often did not have control over obtaining, preparing or injecting drugs and were reliant on sexual partners to assist them (15). Having a sexual partner who was an IDU has often been why some women transitioned from non-injection to IDU (16–18). Other studies have shown that women make themselves vulnerable to sharing of injection equipment and disease when they are unable to inject themselves (19).
Several studies have used latent class analysis (LCA) to explore drug use patterns, especially in men-who-have-sex-with-men (MSM) (20, 21) as well as non-IDUs in South Africa (22). A recent study from California used both type of drug use and route of administration to construct latent classes (23). The strength of LCA as an analytical approach is that it is person-centered and can help to identify subgroups of women using multiple measures, for example, drug use behaviors especially before or during sex. Our study is similar to others that have used alcohol and drug use variables to partition individuals into distinct subgroups, and then used those groups to investigate risky and other sexual behavior outcomes (24–26)
The current study sought to investigate the following research questions: (1) can partitioning women’s drug use profiles into multiple distinct drug use groups help us understand patterns of drug use immediately before or during sex among these groups? (2) how does this help us understand risk reduction practices among the women? To address these questions, we utilized latent class analysis (LCA) and multinomial logistic regression to explore data from women that were collected as part of a 4-year National Institute on Drug Abuse (NIDA)—funded project of point-of-care (POC) rapid testing for HIV and STIs.
Methods
Sample
The sample consisted of women who participated in a structured interview session and received rapid and confirmatory laboratory testing for HIV and STIs. All participants were recruited through the testing programs at the Center for Behavioral Research and Services (CBRS) in Long Beach, California, following a protocol approved by the California State University, Long Beach (CSULB) Institutional Review Board (IRB). CBRS is located in a low socioeconomic area of the South Bay of Los Angeles County with the majority of individuals recruited being at high risk for both non-injection and IDU, sexually transmitted infections and HIV. The informed-consent process included an overview of the study, including what was expected of participants in terms of instruments, blood and oral specimen collection. After the study was explained, participants could ask questions about any aspect of the study. Participants signed the written informed consent form, and were provided with a copy for their records. No data or specimen collection was initiated until the participant had all her questions answered and signed the informed consent form.
All participants were assessed for eligibility using a screening form developed for the study and based on behavioral risk groups as stipulated by the Office of AIDS Programs & Policy (OAPP) of the Los Angeles County Department of Public Health (LACDPH). For women, behavioral-risk groups (BRGs) were defined as: (1) IDU with verified track marks (e.g., visible signs of injection) (27), and (2) women who reported at least two male partners in the last two years or engaging in AI, sex trading, or sex with a man who has sex with men (MSM), an IDU, or an HIV positive man. The prevalence and correlates of heterosexual anal intercourse in women have been previously reported (5, 6, 28) as has women’s use of lubricants (29). All participants were at least 15 years of age at the time of informed consent.
STI Testing
HIV and STI testing were performed by nationally certified phlebotomists with training in HIV pre- and post-test counseling and HIV and STI rapid testing. STI testing included blood and urine specimens for syphilis (both rapid plasma reagin [RPR] and Treponema pallidum particle agglutination assay [TPPA]), gonorrhea, Chlamydia trachomatis, and hepatitis B and C, and HIV. These laboratory test results were used as distal outcomes in our analysis. Methods of recruitment and outcomes for HCV and syphilis have been previous reported (30, 31).
Measures
The behavioral assessments included the following:
Risk Behavior Assessment.
The Risk Behavior Assessment (RBA; 1993) was designed by grantees of the National Institute on Drug Abuse and is a structured interviewer-administered instrument which measures HIV risk factors such as sexual behavior (oral, vaginal, anal, and whether a condom or other barrier was used), gender of sexual partners, and sexual preference and gender self-identification of the participant; drug use in the 30 days prior to interview, and time spent in jail; as well as other HIV risks. The reliability and validity of the RBA has been extensively studied (32, 33).
Lubricant Trailer (LT).
The LT elicits information on a wide variety of sexual practices, including use and type of lubricants and condoms (29).
Behavioral Science Aspects of Rapid Test Acceptance (BSARTA) Screener.
The BSARTA Screener was specifically developed for this study and elicited information on risk factors identified by OAPP to distinguish women at sexual risk and included items on whether a sexual partner was known to have previously had sex with another man, and other items not captured by the RBA.
Drug Use Variables
A total of 15 variables from the RBA were used as input into the latent class analysis; seven of the variables elicited information on 30 day (past month) use of alcohol, marijuana, crack, powder cocaine, heroin, opiates other than heroin, and amphetamine. An additional seven variables elicited information on the use of these same seven drugs immediately before or during sex in the past month. An additional variable was created to designate abstinence from use of all drugs, including before or during sex. Inclusion of an abstinent group has been found to be of value in other research of drug-using women (34). All variables were coded dichotomously for use in PROC LCA (ever/never past month) such that any non-zero number of days in the past month was coded to ever (35).
Sexual behaviors, both those that used condoms and those that were condomless, were captured based on information elicited for the past 30-day period for VI, AI, and giving oral sex. Sexually transmitted infections and HIV positivity were assessed by laboratory test results based on blood specimens, and as noted previously were used as distal outcomes in analysis.
For purposes of constructing the latent class model, the drug use variables, monthly use and drugs used immediately before or during sex, were the inputs to the model. The outcomes appraised using the latent classes were the sexual behaviors and the blood test results for STIs.
Analytical Approach
Analyses were conducted using SAS 9.4 to determine latent classes using PROC LCA (35). We used a three-step approach that includes 1) fitting the LCA model; 2) assigning each participant to the latent class with the modal posterior probability for that individual; and 3) estimating the relationship between the assigned class membership and outcomes of interest. Variables, coded dichotomously, entered into PROC LCA and several possible numbers of latent classes were explored, including two, three, four and five class models. Models were evaluated using the techniques proposed in Collins and Lanza (36) using the G-squared, Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC) fit statistics, with the final model chosen to reflect both the objective fit criteria as well as interpretability of the latent classes. We modeled distal outcomes of laboratory results and self-report of STIs with latent class membership (37–41) and used multinomial logistic regression was used to model sexual behaviors and condom use across the latent classes.
To test the assumption of local independence of the latent class groups, we used the methods of Dziak and Lanza (42); in PROC LCA the local independence assumption is treated as fundamental to the definition of the classes.
Results
Table 1 provides the characteristics of the women (N = 812) on demographic, sex and drug use risks, as well as laboratory and self-report on a variety of STI tests. Five groups were identified by the latent class analysis of our sample (Table 2). Any variable with an item response probability greater than or equal to .50 is bolded in the table in accordance with recommendations (36). Fit statistics for the 2, 3, 4 and 5 group models can be found in Table 3.
Table 1.
Characteristics of Women in Sample (N = 812)
| Variable | n (%) | Mean (SD) |
|---|---|---|
| Ethnicity | ||
| Black | 406 (51) | |
| White | 187 (23) | |
| Hispanic | 135 (17) | |
| Other/mixed | 73 (9) | |
| Education | ||
| Less than high school | 304 (37) | |
| High school | 291 (36) | |
| Greater than H.S. | 217 (27) | |
| Self-reported homelessness, Past 30 days | 379 (47) | |
| Lives in own house, apt. | 231 (29) | |
| Lives in other house, apt. | 282 (35) | |
| Other living place, (hotel, shelter, streets) | 283 (35) | |
| Income, past month | ||
| Less than $500 | 430 (54) | |
| $501-$999 | 266 (34) | |
| $1000-$1999 | 75 (9) | |
| >$2000 | 23 (3) | |
| Marital status | ||
| Single | 464 (59) | |
| Married/partnered | 209 (27) | |
| Other | 114 (14) | |
| Sexual orientation | ||
| Heterosexual | 498 (63) | |
| Lesbian | 32 (4) | |
| Bisexual | 255 (33) | |
| Ever been told by health care provider you have (self-report): | ||
| Hepatitis B | 49 (6) | |
| Gonorrhea | 208 (26) | |
| Syphilis | 87 (11) | |
| Genital Warts | 69 (9) | |
| Chlamydia | 241 (30) | |
| Traded sex for money, ever lifetime | 75 (63) | |
| Traded sex for drugs, ever lifetime | 396 (49) | |
| Traded sex for drugs, past month | 97 (12) | |
| Traded sex for money, past month | 183 (23) | |
| Sex with MSM* | 193 (25) | |
| Sex with known HIV+ partner | 46 (7) | |
| HIV Risk Perception | ||
| 0% (no chance) | 235 (30) | |
| 25% (some chance) | 261 (35) | |
| 50% (half chance) | 123 (16) | |
| 75% (high chance) | 123 (16) | |
| 100% (sure chance) | 36 (5) | |
| Currently Pregnant? | ||
| Yes | 28 (4) | |
| Positive Laboratory test results | ||
| HIV 1 serum | 18 (2) | |
| HCV | 121 (15) | |
| HBV (Core AB) | 166 (20) | |
| Syphilis (non treponemal) | 17 (2) | |
| Syphilis (treponemal) | 100 (12) | |
| Sources of income** | ||
| Paid job, salary | 115 (14) | |
| Welfare, food stamps | 367 (46) | |
| Social security/disability | 183 (23) | |
| Spouse, friends, family | 196 (25) | |
| Sell, trading goods, barter | 65 (8) | |
| Ever injected any drug | 143 (18) | |
| Ever in drug treatment, any modality | 450 (56) | |
| Age, in years | 38.4 (11.92) | |
| Times injected any drug, last 30 days | 4.13 (18.38) | |
| # sex partners, last 30 days | 3.38 (7.09) | |
| # drug injector sex partners, last 30 days | 1.06 (2.49) | |
| Past 30-day drug use, # days used | ||
| Alcohol | 6.58 (9.39) | |
| Marijuana | 5.06 (9.87) | |
| Crack | 2.92 (7.47) | |
| Cocaine | 1.00 (4.32) | |
| Heroin | 0.92 (4.05) | |
| Speedball (heroin + cocaine) | 0.23 (1.55) | |
| Opiates other than heroin | 1.31 (4.89) | |
| Amphetamines | 2.74 (7.06) | |
| Drug use before/during sex, times used past 30 days | ||
| Alcohol | 4.05 (9.28) | |
| Marijuana | 2.80 (10.04) | |
| Crack | 1.28 (6.21) | |
| Cocaine | 0.56 (3.04) | |
| Heroin | 0.76 (4.94) | |
| Speedball (heroin + cocaine) | 0.15 (1.45) | |
| Opiates other than heroin | 0.37 (3.05) | |
| Amphetamines | 1.93 (7.73) | |
Note:
MSM man-who-has-sex-with-men.
Sums to greater than 100% as there can be more than one source of income.
Numbers may not sum to N=812 due to missing data.
Table 2.
Latent Class Model of Monthly Drug Use Immediately Before or During Sex for Women
| Latent Class |
|||||
|---|---|---|---|---|---|
| Abstainers | Alcohol and Marijuana | Amphetamine | No Sex-with-Alcohol | Poly Drug Use | |
| Latent class prevalence | .26 | .16 | .10 | .37 | .11 |
| N | 210 | 125 | 75 | 315 | 87 |
| Item response probabilities | Probability of a “Yes” Response | ||||
| Monthly alcohol use | .00 | .83 | .57 | .84 | .93 |
| Monthly marijuana use | .00 | .98 | .53 | .29 | .81 |
| Monthly crack use | .00 | .16 | .03 | .31 | .69 |
| Monthly cocaine use | .00 | .04 | .00 | .01 | .77 |
| Monthly heroin use | .00 | .00 | .05 | .05 | .58 |
| Monthly opiate use | .00 | .06 | .15 | .07 | .55 |
| Monthly amphetamine use | .00 | .02 | .99 | .08 | .63 |
| Alcohol with sex | .00 | .70 | .39 | .46 | .77 |
| Marijuana with sex | .00 | .98 | .22 | .00 | .53 |
| Crack with sex | .00 | .09 | .00 | .12 | .52 |
| Cocaine with sex | .00 | .00 | .01 | .00 | .52 |
| Heroin with sex | .00 | .00 | .02 | .00 | .44 |
| Other opiates with sex | .00 | .00 | .07 | .00 | .28 |
| Amphetamine with sex | .00 | .00 | .88 | .00 | .59 |
| Abstinent | .98 | .00 | .00 | .00 | .00 |
Notes: Item response probabilities > .50 in bold to facilitate interpretation. The probability of a “No” response can be calculated by subtracting the item-response probabilities shown above from 1.00.
Table 3.
Comparison of fit statistics for baseline latent class analysis model conducted on community-recruited women (N = 812)
| Number of latent classes | 2 | 3 | 4 | 5 |
|---|---|---|---|---|
| Log-likelihood | −4401.46 | −3978.30 | −3795.85 | −3669.87 |
| G-squared | 3049.64 | 2203.33 | 1838.43 | 1586.47 |
| AIC | 3111.64 | 2297.33 | 1969.43 | 1744.47 |
| BIC | 3257.32 | 2518.21 | 2260.49 | 2115.73 |
| CAIC | 3288.32 | 2565.21 | 2323.49 | 2194.73 |
| aBIC | 3158.88 | 2368.96 | 2060.43 | 1864.86 |
Note: AIC Akaike information criterion, BIC Bayesian information criterion, CAIC consistent
AIC, aBIC adjusted BIC; for all fit statistics, smaller is better.
Group 1:
The first group, identified as the Abstainers group accounted for 26% of participants and was characterized by no use of any drugs in the past 30 days and no use of any drug or alcohol before or during sex. Abstinence refers to drug abstinence not sexual abstinence.
Group 2:
Identified as the Amphetamine group, these women were characterized by a very high probability of use of amphetamine in the past 30 days as well as amphetamine use immediately before or during sex. There was also a >.50 probability of use of alcohol and marijuana in this group, which was 10% of respondents.
Group 3:
This group is identified as the Alcohol and Marijuana group, and was characterized by a very high probability of both alcohol and marijuana use in the past 30 days and a very high probability of use of both alcohol and marijuana before or during sex. No drugs other than alcohol and marijuana, either on a past 30-day basis or before or during sex had a >.50 probability of use in this group. Approximately 16% of the respondents were classified into this group.
Group 4:
This group is identified as the No Sex-with-Alcohol group and was characterized by a high probability of 30-day use of alcohol only and it was not used immediately before or during sex. This group was comprised of 37% of the respondents. The No Sex-with-Alcohol group engaged in sexual activity without drugs or alcohol, so was not sexually abstinent.
Group 5:
The final group is identified as the Poly Drug group and was characterized by > .50 probabilities of use all drugs (marijuana, crack, cocaine, heroin, other opiates, amphetamine) and alcohol in the past 30 days, as well as >.50 probabilities of also using alcohol and almost all drugs before or after sex. This group represented 11% of the sample.
Table 4 shows bivariate differences in demographic variables across the five groups. For example, women in the Alcohol and Marijuana and Poly Drug groups were more likely to be Black or African American and more likely to report their relationship status as partnered or married compared to the other groups. The Alcohol and Marijuana group was more likely to report living with children under the age of 18. All groups reported high proportions of homelessness, but the highest proportion at 81% was in the Poly Drug group. Differences among the groups were found on self-reported sexual preferences, with the Poly Drug group having the highest proportion of women identifying as bisexual; the Abstinent group was predominantly heterosexual.
Table 4.
Latent Class Membership and Demographic Characteristics of Women (N = 812)
| Abstainers | Amphetamine | Alcohol-Marijuana | No Sex-with-Alcohol | Poly Drug | ||
|---|---|---|---|---|---|---|
| n(%) | n(%) | n(%) | n(%) | n(%) | p | |
| Race/ethnicity | ||||||
| African American | 93 (44) | 15 (20) | 68 (54) | 179 (58) | 51 (59) | |
| White | 63 (30) | 35 (47) | 31 (25) | 45 (15) | 13 (15) | |
| Hispanic | 38 (18) | 17 (23) | 20 (16) | 48 (16) | 12 (14) | |
| Other | 16 (7) | 8 (11) | 6 (5) | 32 (11) | 11 (13) | .0001 |
| Marital Status | ||||||
| Single | 141 (67) | 35 (49) | 69 (56) | 177 (59) | 42 (49) | |
| Partnered | 49 (23) | 11 (15) | 43 (35) | 77 (26) | 29 (34) | |
| Other (divorced, widowed) | 19 (9) | 25 (35) | 11 (9) | 44 (15) | 15 (17) | .0001 |
| Currently living with children <18 | ||||||
| No | 145 (69) | 66 (88) | 80 (64) | 239 (76) | 74 (85) | |
| Yes | 65 (31) | 9 (12) | 45 (36) | 76 (24) | 13 (15) | .0001 |
| Currently homeless | ||||||
| No | 135 (65) | 42 (56) | 69 (56) | 157 (52) | 16 (19) | |
| Yes | 74 (35) | 33 (44) | 53 (44) | 147 (48) | 70 (81) | .0001 |
| Education | ||||||
| < High School | 68 (32) | 22 (29) | 40 (32) | 136 (43) | 38 (44) | |
| High School Graduate | 75 (36) | 27 (36) | 50 (40) | 116 (37) | 23 (37) | |
| >High School | 67 (32) | 26 (35) | 35 (28) | 63 (20) | 26 (20) | .01 |
| Sexual Preference | ||||||
| Heterosexual | 143 (70) | 42 (56) | 84 (68) | 208 (70) | 21 (24) | |
| Lesbian | 9 (5) | 2 (3) | 5 (4) | 9 (3) | 7 (8) | |
| Bisexual | 51 (25) | 31 (41) | 34 (28) | 80 (27) | 59 (68) | .0001 |
| Self-reported Hepatitis B | ||||||
| No | 205 (98) | 74 (99) | 118 (94) | 270 (89) | 85 (98) | |
| Yes | 5 (2) | 1 (1) | 7 (6) | 34 (11) | 2 (2) | .0001 |
| Self-reported Gonorrhea | ||||||
| No | 165 (79) | 63 (84) | 76 (61) | 238 (78) | 51 (59) | |
| Yes | 44 (21) | 12 (16) | 49 (39) | 67 (23) | 36 (41) | .0001 |
| Self-reported Syphilis | ||||||
| No | 185 (88) | 73 (97) | 102 (82) | 279 (92) | 72 (88) | |
| Yes | 25 (12) | 2 (3) | 22 (18) | 26 (8) | 12 (12) | .0067 |
| Used condoms for Vaginal Intercourse | ||||||
| No | 187 (89) | 67 (89) | 83 (78) | 229 (73) | 80 (92) | |
| Yes | 23 (11) | 8 (11) | 23 (22) | 86 (27) | 7 (8) | .0001 |
| Ever Anal Intercourse | ||||||
| No | 184 (88) | 58 (77) | 103 (82) | 279 (89) | 73 (84) | |
| Yes | 26 (12) | 17 (23) | 22 (18) | 36 (11) | 14 (16) | .07 |
| Used condoms for Anal Intercourse | ||||||
| No | 207 (99) | 75 (100) | 119 (95) | 303 (96) | 85 (96) | |
| Yes | 3 (1) | 0 (0) | 6 (5) | 12 (4) | 2 (4) | .15 |
| Known drug injecting sexual partner | ||||||
| No | 166 (79) | 34 (45) | 102 (82) | 270 (86) | 17 (20) | |
| Yes | 44 (21) | 41 (55) | 23 (18) | 45 (14) | 70 (80) | .0001 |
| Injected in past month | ||||||
| No | 210 (100) | 36 (48) | 124 (99) | 286 (91) | 26 (30) | |
| Yes | 0 (0) | 39 (52) | 11 (1) | 29 (9) | 61 (70) | .0001 |
| Known HIV+ Sexual Partner | ||||||
| No | 200 (95) | 70 (93) | 121 (96) | 298 (95) | 77 (88) | |
| Yes | 10 (5) | 5 (7) | 4 (4) | 17 (5) | 10 (12) | .11 |
| Ever traded sex for drugs | ||||||
| No | 139 (66) | 47 (63) | 88 (70) | 212 (69) | 24 (28) | |
| Yes | 71 (34) | 28 (37) | 37 (30) | 93 (31) | 63 (72) | .0001 |
| Ever traded sex for money | ||||||
| No | 126 (60) | 29 (39) | 59 (47) | 168 (55) | 24 (28) | |
| Yes | 84 (40) | 46 (61) | 66 (53) | 137 (45) | 63 (72) | .0001 |
| Sex for drugs past month | ||||||
| No | 208 (100) | 63 (84) | 112 (90) | 277 (91) | 36 (45) | |
| Yes | 0 (0) | 12 (16) | 13 (10) | 28 (9) | 44 (55) | .0001 |
| Sex for money past month | ||||||
| No | 196 (95) | 52 (70) | 89 (71) | 236 (78) | 38 (45) | |
| Yes | 11 (5) | 23 (30) | 36 (29) | 66 (22) | 47 (55) | .0001 |
| Sex with known MSM partner | ||||||
| No | 142 (74) | 33 (63) | 80 (82) | 210 (75) | 34 (48) | |
| Yes | 49 (26) | 19 (37) | 18 (18) | 71 (25) | 37 (52) | .0001 |
| Laboratory Test Results | ||||||
| HIV | ||||||
| Negative | 210 (100) | 75 (100) | 124 (99) | 302 (96) | 83 (95) | |
| Positive | 0 (0) | 0 (0) | 1 (1) | 13 (4) | 4 (5) | .004 |
| HBV (Core AB) | ||||||
| Negative | 191 (92) | 69 (92) | 92 (74) | 219 (70) | 73 (84) | |
| Positive | 17 (8) | 6 (8) | 33 (26) | 96 (30) | 14 (16) | .0001 |
| HCV | ||||||
| Negative | 195 (93) | 60 (80) | 109 (89) | 255 (81) | 68 (78) | |
| Positive | 15 (7) | 15 (20) | 14 (11) | 58 (19) | 19 (21) | .0007 |
| Syphilis (treponemal) | ||||||
| Negative | 191 (91) | 73 (97) | 117 (94) | 258 (82) | 73 (84) | |
| Positive | 19 (9) | 2 (3) | 8 (6) | 57 (18) | 14 (16) | .0001 |
| Syphilis (nontreponemal) | ||||||
| Negative | 208 (99) | 69 (97) | 120 (96) | 307 (97) | 87 (100) | |
| Positive | 2 (1) | 2 (3) | 5 (4) | 8 (3) | 0 | .20 |
Note. All totals may not sum to 812 due to missing data.
With respect to self-reported STIs (“Has a health care provider ever told you that you have the following?”) the Poly Drug and Alcohol and Marijuana groups were most likely to report they had ever been told they had syphilis (12% and 18%, respectively), and both groups were also more likely to report they had been told they had gonorrhea. The Alcohol and Marijuana and No Sex-with-Alcohol groups were more likely to have a positive laboratory test result for hepatitis B (core AB). Results of laboratory tests (not self-report) indicated that the Poly Drug and No Sex-with-Alcohol groups had the highest prevalence of positive test results on the treponemal test for syphilis. The Poly Drug group had the highest proportion testing positive for HIV and also had the highest proportion of HIV+ sexual partners compared to the other groups. Two of the five groups had high proportions of women with positive test results for anti-hepatitis C; the highest prevalence was in the Poly Drug group at 21%, followed by the Amphetamine group at 20%. The Poly Drug group had the highest proportion of injection drug-using partners at 72%, followed by the Amphetamine group at 60%.
Table 5 display the results of the multinomial logistic regression analyses. There were distinct sexual and risk and protective behaviors evidenced among the groups. Compared to the Abstinent reference group only the Alcohol and Marijuana group had a significantly greater odds of VI, while the Poly Drug group was significantly less likely to reported engaging in VI. The No Sex-with-Alcohol group was at significantly increased odds of using condoms for VI. The Poly Drug group had a greater odds of reporting AI compared to the reference group and none of the groups was more or less likely to report using condoms for AI. The Amphetamine group was the only one significantly more likely to report giving oral sex to a male partner compared to the reference group.
Table 5.
Adjusted multinomial logistic regression of sexual risk and protective factors associated with latent class membership (reference category = Abstinent)
| Abstinent | Amphetamine | Alcohol-Marijuana | No Sex Alcohol | Poly Drug | |||||
|---|---|---|---|---|---|---|---|---|---|
| OR | 95% | OR | 95% | OR | 95% | OR | 95% | ||
| Vaginal intercourse*** | Ref. | 1.34 | .57, 3.13 | 2.48 | 1.22, 5.05 | .89 | .55, 1.44 | .12 | .05, .32 |
| Condoms for VI*** | .59 | .23, 1.50 | 1.46 | .75, 2.87 | 2.54 | 1.42, 4.55 | 1.11 | .35, 3.47 | |
| Anal intercourse* | 1.04 | .49, 2.20 | .68 | .33, 1.40 | .72 | .37, 1.42 | 4.18 | 1.49, 11.75 | |
| Condoms for AI | .02 | .0002, 16.9 | 1.49 | .30, 7.40 | .41 | .08, 2.03 | .40 | .04, 3.82 | |
| Gave oral sex** | 3.27 | 1.48, 7.25 | 1.69 | .94, 3.02 | .76 | .45, 1.27 | 1.36 | .48, 3.82 | |
| Condoms for oral sex*** | 116.5 | .28, 47142 | 36.57 | .09, 14113 | 255.7 | .70, 93120 | 144.7 | .34, 61198 | |
Table 6 provides the probabilities of the distal outcomes of laboratory results and self-reported STIs across the latent classes for which sufficient power existed.
Table 6.
Adjusted multinomial logistic regression of HIV and STI laboratory test results associated with latent class membership (reference category = Abstinent)
| Abstinent | Amphetamine | Alcohol-Marijuana | No Sex Alcohol | Poly Drugs | |||||
|---|---|---|---|---|---|---|---|---|---|
| OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | ||
| HIV | Ref. | 1.19 | .35, 3.98 | 2.04 | .51, 8.08 | .06 | .0003, 16.27 | .04 | .001, 1.19 |
| HCV | 2.27 | 1.14, 4.51 | .61 | .28, 1.35 | 1.89 | .85, 4.17 | .60 | .30, 1.18 | |
| Syphilis, treponemal | .72 | .35, 1.45 | .05 | .01, .20 | .05 | .01, .30 | .54 | .28, 1.01 | |
| Syphilis, non treponemal | .12 | .0003, 50.49 | 58.16 | 9.81, 344.7 | 23.17 | 2.82, 190.48 | 2.35 | .42, 13.02 | |
| HBV*** | .25 | .13, .50 | .53 | .29, .94 | .08 | .02, .23 | .16 | .08, .29 | |
Significance tests based on 2*log-likelihood
<.05
< .01
<.001
Discussion
This study explored profiles of drug use behaviors in the context of sexual risk and suggests that it may be useful to assess the drug use profiles of women using the type of drug use before or during sex. Women in the Abstinent group were more likely to live with children under the age of 18, and a majority of them report their marital status as single. Women with substance abuse issues who also have minor children have fears that children will be removed from their custody if their drug use becomes known or they present for drug treatment (43). The women in the Abstinent group may be limiting drug and alcohol use to ensure they do not face criminal justice system involvement that could lead to loss of their children. A large proportion of women in the Alcohol and Marijuana group also reported living with children under the age of 18; like the abstinent group they may be avoiding illicit drug use in order to reduce potential loss of their children. These two groups also had the lowest proportions of known IDU sexual partners and HIV + sexual partners.
The Amphetamine and Poly Drug groups were more likely to report having known HIV+ sexual partners and IDU sexual partners; both groups had high proportions with hepatitis C, however the Poly Drug group had the highest proportion of HIV+ respondents. The high HIV prevalence in the Poly Drug group may also explain the significantly lower odds for this group in reporting VI. The No Sex-with-Alcohol group also had a relatively high proportion of HIV+ women, at 4%. This group also had the highest proportion of anti-HBV positive test results. For the women in this group, HIV and hepatitis B are more likely to have been sexually transmitted. The women in this group appear to have learned this, as reflected in both the lack of any drug or alcohol use immediately before or during sex, as well as the greater odds for use of condoms for VI. Previous studies have noted HBV and HIV are primarily sexually transmitted in non-IDUs (44, 45).
All groups had fairly high proportions of respondents who reported ever trading sex for drugs or money, with the No Sex-with-Alcohol and Abstinent groups having the lowest proportions. The Poly Drug group was the only group with a high proportion (55%) reporting trading sex for money in the past month; 55% of this group also reported trading sex for drugs in the past month.
The multinomial logistic regression model findings demonstrate that the Poly Drug group, the group with the highest proportion of HIV+ sexual partners, had a significantly lower odds of VI compared to the reference group, and a significantly higher odds of AI. For the women in this group, unprotected VI may still be a significant source of HIV risk, and they may believe they are mitigating that risk by engaging in AI.
Previous studies have shown that women underestimate the risk of receptive AI because it is not hetero-normative behavior. It is not clear whether AI occurs as a sex trading activity or as a “special” behavior reserved for long-term partners, and both are possible for the women in our study (4). The Amphetamine and Poly Drug groups had high proportions of women who have known IDU sex partners, and known MSM partners. Disclosure of HIV serostatus has been found to be correlated with perceptions of social support, and HIV+ individuals often disclose selectively to some partners but not to others (46). HIV+ men are more likely to disclose their status to primary partners with bisexual men being more willing to disclose to male partners than female partners (47). Information about their male partners may explain, in part, the low odds of VI and high odds of AI among the Poly Drug group and the increased odds of giving oral sex in the Amphetamine group.
Limitations
We did not have information on the type of sexual partner at the time drugs were consumed before or during sex. Having this information could have helped clarify the context of the sexual encounter given that there were high proportions of women in four of the five groups who reported ever trading sex for money or drugs. Information on marital status only partly accounts for sexual partners. Our study did not collect information on non-partnered sex, so the drug use before or during sex variables assume partnered sex, but this may not be the case. Our study did not address drug use and sexual behaviors over time, so we are not able to address potential changes in either type of behaviors. Because drug use variables were measured as of the past 30 days, it is possible that some of the women in the Abstinent or No Sex-with-Alcohol groups have previously been members of one of the other groups; this is possible because high proportions of women in all groups reported ever (lifetime) trading sex for money or drugs. The Abstinent, No Sex-with-Alcohol, and Alcohol and Marijuana groups should not be wrongly identified as women who have never been involved in the use of drugs other than alcohol and marijuana. Past 30-day use of marijuana, for purposes of this study, did not distinguish between illicit marijuana use and medical marijuana use. During the recruitment period, marijuana in California was available on a medical basis according to state law, though not yet legally approved for recreational use. Finally, condom use for VI and AI may not be the only protective sexual behaviors practiced by the sample. The study did not collect information on sexual practices of non-penetrative sex, such as mutual masturbation. We also do not know how the women knew that their partners were either HIV+ or an MSM; it is possible that partners self-disclosed to the women, or that the women made assumptions based on other indirect information.
Conclusion
One of the strengths of this study was the investigation of drug use before/during sex variables to construct the latent classes. Women’s combined use of alcohol and drugs before or during sex has an impact on their overall risks for HIV and STIs. Latent class analysis has been widely used in studies of substance use (48–50). One such study found that substance use with steady partners was associated with increased sexual risk behaviors (22), which is consistent with a theoretical framework in which alcohol or drug use leads to cognitive impairment, which in turn leads to disinhibition with respect to sexual behavior and/or inability to perform protective behaviors (51). While we cannot infer causation, the women in our study have high-risk steady partners (HIV+, IDU, sex with known MSM), but also show evidence that they may be attempting to moderate or reduce sexual risks. Interventions that incorporate information about drug use immediately before or during sex would be especially relevant to this group of high-risk women, especially within a skills-building context (52).
Acknowledgements
Funding was provided in part by grant R01 DA030234 from the National Institute on Drug Abuse.
Footnotes
Disclosures:
The authors report no relevant financial conflicts.
References
- 1.Satterwhite CL, Torrone E, Meites E, Dunne EF, Mahajan R, Ocfemia MCB, et al. Sexually transmitted infections among US women and men: prevalence and incidence estimates, 2008. Sexually Transmitted Diseases 2013;40(3):187–93. [DOI] [PubMed] [Google Scholar]
- 2.Aholou TM, McCree DH, Oraka E, Jeffries WLt, Rose CE, DiNenno E, et al. Sexual risk and protective behaviors among reproductive-aged women in the United States. Journal of Women’s Health 2017;26(11):1150–60. [DOI] [PubMed] [Google Scholar]
- 3.Reynolds GL, Latimore AD, Fisher DG. Heterosexual anal sex among female drug users: U.S. national compared to local Long Beach, California data. AIDS Behav 2008;12(5):796–805. [DOI] [PubMed] [Google Scholar]
- 4.Reynolds GL, Fisher DG, Rogala B. Why women engage in anal intercourse: results from a qualitative study. Archives of Sexual Behavior 2015;44(4):983–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Reynolds GL, Fisher DG, Napper LE, Fremming BW, Jansen MA. Heterosexual anal sex reported by women receiving HIV prevention services in Los Angeles County. Women’s health issues : official publication of the Jacobs Institute of Women’s Health 2010;20(6):414–9. [DOI] [PubMed] [Google Scholar]
- 6.Reynolds GL, Fisher DG, Laurenceau JP, Fortenberry JD. An electronic daily diary study of anal intercourse in drug-using women. AIDS Behav 2015;19(12):2325–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Javanbakht M, Gorbach P, Stirland A, Chien M, Kerndt P, Guerry S. Prevalence and correlates of rectal Chlamydia and gonorrhea among female clients at sexually transmitted disease clinics. Sexually Transmitted Diseases 2012;39(12):917–22. [DOI] [PubMed] [Google Scholar]
- 8.Harbison HS, Austin EL, Hook EW 3rd, Muzny CA. Is there a continuum of risk for sexually transmitted infections among African American women? Sex Transm Dis 2017. [DOI] [PubMed]
- 9.Fields JC, King KM, Alexander KA, Smith KC, Sherman SG, Knowlton A. Recently released Black men’s perceptions of the impact of incarceration on sexual partnering. Culture, Health & Sexuality 2018;20(1):55–68. [DOI] [PubMed] [Google Scholar]
- 10.Miller M, Korves CT, Fernandez T. The social epidemiology of HIV transmission among African American women who use drugs and their social network members. AIDS Care 2007;19(7):858–65. [DOI] [PubMed] [Google Scholar]
- 11.Kwiatkowski CF, Booth RE. Differences in HIV risk behaviors among women who exchange sex for drugs, money, or both drugs and money. AIDS Behav 2000;4(3):233–40. [Google Scholar]
- 12.Cobbina JE, Oselin SS. It’s not only for the money: an analysis of adolescent versus adult entry into street prostitution. Sociological Inquiry 2011;81(3):310–32. [DOI] [PubMed] [Google Scholar]
- 13.Morris MD, Lemus H, Wagner KD, Martinez G, Lozada R, Gómez RMG, et al. Factors associated with pathways toward concurrent sex work and injection drug use among female sex workers who inject drugs in Northern Mexico. Addiction 2013;108(1):161–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Strathdee SA, Philbin MM, Semple SJ, Pu M, Orozovich P, Martinez G, et al. Correlates of injection drug use among female sex workers in two Mexico–US border cities. Drug & Alcohol Dependence 2008;92(1):132–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wagner KD, Jackson Bloom J, Hathazi SD, Sanders B, Lankenau SE. Control over drug acquisition, preparation, and injection: implications for HIV and HCV risk among young female injection drug users. Addiction 2013;2013. [DOI] [PMC free article] [PubMed]
- 16.Tuchman E Women’s injection drug practices in their own words: a qualitative study. Harm Reduction Journal 2015;12:6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Young AM, Larian N, Havens JR. Gender differences in circumstances surrounding first injection experience of rural injection drug users in the United States. Drug and Alcohol Dependence 2014;134:401–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Argento E, Strathdee SA, Goldenberg S, Braschel M, Montaner J, Shannon K. Violence, trauma and living with HIV: Longitudinal predictors of initiating crystal methamphetamine injection among sex workers. Drug and Alcohol Dependence 2017;175:198–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tuchman E Women’s injection drug practices in their own words: a qualitative study. Harm Reduction Journal 2015;12(1):6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.McCarty-Caplan D, Jantz I, Swartz J. MSM and drug use: A latent class analysis of drug use and related sexual risk behaviors. AIDS Behav 2014;18(7):1339–51. [DOI] [PubMed] [Google Scholar]
- 21.Yu G, Goldsamt LA, Clatts MC, Giang LM. Sexual initiation and complex recent polydrug use patterns among male sex workers in Vietnam: A preliminary epidemiological trajectory. Arch Sex Behav 2016;45(4):975–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Trenz RC, Scherer M, Duncan A, Harrell PT, Moleko AG, Latimer WW. Latent class analysis of polysubstance use, sexual risk behaviors, and infectious disease among South African drug users. Drug and Alcohol Dependence 2013;132(3):441–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Roth AM, Armenta RA, Wagner KD, Roesch SC, Bluthenthal RN, Cuevas-Mota J, et al. Patterns of drug use, risky behavior, and health status among persons who inject drugs living in San Diego, California: a latent class analysis. Substance Use & Misuse 2015;50(2):205–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Fenaughty AM, Fisher DG, Cagle HH. Sex partners of Alaskan drug users: HIV transmission between white men and Alaska Native women. Women & Health 1998;27(1–2):87–103. [DOI] [PubMed] [Google Scholar]
- 25.Fenaughty A, Fisher DG. High-risk sexual behavior among drug users: The utility of a typology of alcohol variables. Sexually Transmitted Diseases 1998;25(1):38–43. [DOI] [PubMed] [Google Scholar]
- 26.Schonnesson LN, Atkinson J, Williams ML, Bowen A, Ross MW, Timpson SC. A cluster analysis of drug use and sexual HIV risks and their correlates in a sample of African-American crack coaine smokers with HIV infection. Drug & Alcohol Dependence 2008;97:44–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cagle HH, Fisher DG, Senter TP, Thurmond RD, Kaster AR. Classifying skin lesions of injection drug users: A method for corroborating disease risk DHHS Pub No (SMA) 02-XXX. Rockville, MD: Center for Substance Abuse Treatment, Substance Abuse and Mental Health Services Administration; 2002. [Google Scholar]
- 28.Reynolds GL, Latimore AD, Fisher DG. Heterosexual anal sex among female drug users: U.S. national compared to local Long Beach, California data. AIDS and Behavior 2008;12(5):796–805. [DOI] [PubMed] [Google Scholar]
- 29.Hess KL, Reynolds GL, Baig S, Fisher DG. Lubricant use by women in a community sample from Long Beach, California International Academy of Sex Research; 2012; Estorial, Portugal. [Google Scholar]
- 30.Hess KL, Fisher DG, Reynolds GL. Sensitivity and specificity of point-of-care rapid combination syphilis-HIV-HCV tests. PLOS One 2014;9(11):e112190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Fisher DG, Hess KL, Erlyana E, Reynolds GL, Cummins CA, Alonzo TA. Comparison of rapid point-of-care tests for detection of antibodies to hepatitis C virus Open Forum Infectious Diseases; 2015: Oxford University Press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Dowling-Guyer S, Johnson ME, Fisher DG, Needle R, Watters J, Andersen M, et al. Reliability of drug users’ self-reported HIV risk behaviors and validity of self-reported recent drug use. Assessment 1994;1(4):383–92. [Google Scholar]
- 33.Needle R, Fisher DG, Weatherby N, Chitwood D, Brown B, Cesari H, et al. The reliability of self-reported HIV risk behaviors of drug users. Psychology of Addictive Behaviors 1995;9(4):242–50. [Google Scholar]
- 34.Carter A, Roth EA, Ding E, Milloy M, Kestler M, Jabbari S, et al. Substance use, violence, and antiretroviral adherence: a latent class analysis of women living with HIV in Canada. AIDS and Behavior 2017:1–15. [DOI] [PubMed]
- 35.Lanza ST, Collins LM, Lemmon DR, Schafer JL. PROC LCA: A SAS procedure for latent class analysis. Structural Equation Modeling 2007;14(4):671–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Collins LM, Lanza ST. Latent class and latent transition analysis: With applications in the social, behavioral and health sciences Hoboken: NJ: John Wiley & Sons; 2010. [Google Scholar]
- 37.Dziak JJ, Bray BC, Wagner AT. LCA_Distal_BCH SAS macro users’ guide, version 1.1 University Park, PA: The Methodology Center, Penn State; 2017. Available from: http://methodology.psu.edu. [Google Scholar]
- 38.Bolck A, Croon M, Hagenaars J. Estimating latent structure models with categorical variables: One-step versus three-step estimators. Political Analysis 2004;12(1):3–27. [Google Scholar]
- 39.Vermunt JK. Latent class modeling with covariates: Two improved three-step approaches. Political Analysis 2010;18:450–69. [Google Scholar]
- 40.Bray BC, Lanza ST, Tan X. Eliminating bias in classify-analyze approaches for latent class analysis. Structural Equation Modeling: A Multidisciplinary Journal 2015;22(1):1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Dziak JJ, Bray BC, Zhang J, Zhang M, Lanza ST. Comparing the performance of improved classify-analyze approaches for distal outcomes in latent profile analysis. Methodology 2016;12(4):107–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Dziak JJ, Lanza ST. LCABootstrap SAS macro users’ guide, version 4.0 University Park: PA: The Methodology Center, Penn State; 2016. Available from: http://methodology.psu.edu. [Google Scholar]
- 43.Powis B, Gossop M, Bury C, Payne K, Griffiths P. Drug‐using mothers: social, psychological and substance use problems of women opiate users with children. Drug and Alcohol Review 2000;19(2):171–80. [Google Scholar]
- 44.Gyarmathy VA, Neaigus A, Miller M, Friedman SR, Des Jarlais DC. Risk correlates of prevalent HIV, hepatitis B virus, and hepatitis C virus infections among noninjecting heroin users. J Acquir Immune Defic Syndr 2002;30(4):448–56. [DOI] [PubMed] [Google Scholar]
- 45.Neaigus A, Gyarmathy VA, Zhao M, Miller M, Friedman SR, Des Jarlais DC. Sexual and other noninjection risks for HBV and HCV seroconversions among noninjecting heroin users. The Journal of Infectious Diseases 2007;195(7):1052–61. [DOI] [PubMed] [Google Scholar]
- 46.Kalichman SC, DiMarco M, Austin J, Luke W, DiFonzo K. Stress, social support, and HIV-status disclosure to family and friends among HIV-positive men and women. Journal of Behavioral Medicine 2003;26(4):315–32. [DOI] [PubMed] [Google Scholar]
- 47.Sullivan KM. Male self-disclosure of HIV-positive serostatus to sex partners: a review of the literature. Journal of the Association of Nurses in AIDS Care 2005;16(6):33–47. [DOI] [PubMed] [Google Scholar]
- 48.Grant JD, Scherrer JF, Neuman RJ, Todorov AA, Price RK, Bucholz KK. A comparison of the latent class structure of cannabis problems among adult men and women who have used cannabis repeatedly. Addiction 2006;101(8):1133–42. [DOI] [PubMed] [Google Scholar]
- 49.Carlson RG, Wang J, Falck RS, Siegal HA. Drug use practices among MDMA/ecstasy users in Ohio: a latent class analysis. Drug & Alcohol Dependence 2005;79(2):167–79. [DOI] [PubMed] [Google Scholar]
- 50.Green TC, Kershaw T, Lin H, Heimer R, Goulet JL, Kraemer KL, et al. Patterns of drug use and abuse among aging adults with and without HIV: a latent class analysis of a US Veteran cohort. Drug & Alcohol Dependence 2010;110(3):208–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Jones A, Christiansen P, Nederkoorn C, Houben K, Field M. Fluctuating disinhibition: Implications for the understanding and treatment of alcohol and other substance use disorders. Frontiers in Psychiatry 2013;4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Tross S, Campbell AN, Cohen LR, Calsyn D, Pavlicova M, Miele G, et al. Effectiveness of HIV/STD sexual risk reduction groups for women in substance abuse treatment programs: Results of a NIDA clinical trials network trial. Journal of Acquired Immune Deficiency Syndromes (1999) 2008;48(5):581. [DOI] [PMC free article] [PubMed] [Google Scholar]