HIV-Associated Histories, Perceptions, and Practices Among Low-Income African American Women: Does Rural Residence Matter?

Abstract

Objectives. This study compared HIV-associated sexual health history, risk perceptions, and sexual risk behaviors of low-income rural and nonrural African American women.

Methods. A cross-sectional statewide survey of African American women (n = 571) attending federally funded Special Supplemental Nutrition Program for Women, Infants, and Children clinics was conducted.

Results. Adjusted analyses indicated that rural women were more likely to report not being counseled about HIV during pregnancy (P = .001), that a sex partner had not been tested for HIV (P = .005), no preferred method of prevention because they did not worry about sexually transmitted diseases (P = .02), not using condoms (P = .009), and a belief that their partner was HIV negative, despite lack of testing (P = .04).

Conclusions. This study provided initial evidence that low-income rural African American women are an important population for HIV prevention programs.

In the United States, the incidence of HIV infection and AIDS diagnosis is increasing most rapidly among African American women.1 Compared with White and Hispanic women, African American women are disproportionately diagnosed with AIDS.1–4 This disparity has been observed throughout the course of the US AIDS epidemic.1

The geographic distribution of AIDS among African American women in the United States indicates that a vast majority of cases occur in urban epicenters and their surrounding communities. However, the diffusion of HIV to rural areas is an increasingly important issue.1,5–9 For example, a study of women residing in rural Alabama indicated a 170-fold increase in AIDS cases among African American women over a 10-year period.10 Evidence suggests that rural HIV epidemics most often affect women, particularly young adult African American women.5,11 Rural HIV epidemics also may be distinct from nonrural epidemics because rural communities may be less prepared to meet the prevention and treatment challenges imposed by the virus.12–14 Thus, an increased understanding of rural HIV epidemics is warranted. Although numerous studies have investigated correlates of HIV risk behavior among nonrural African American women15 and efficacious intervention programs have been developed,1,16,17 studies have not addressed how the AIDS epidemic has uniquely affected rural African American women.

Because of their geographic isolation from urban epicenters, rural African American women possibly could be less engaged by the potential threat of HIV infection than are their nonrural counterparts. For example, data from the National Health and Social Life Survey indicated that rural Americans were less likely than their nonrural counterparts to report any change in sexual behavior in response to the AIDS epidemic, including condom use.18 Also, a recent analysis of data collected from a national probability sample found that individuals living in rural areas were less likely to use condoms than were those living in large metropolitan areas.19 Yet, published studies have not reported specific comparisons between rural and nonrural African American women.

If rural African American women are relatively unengaged by the potential threat of HIV infection, they may be less receptive to the adoption of HIV-protective behaviors. An emerging behavioral theory, the Precaution Adoption Process Model, directly addresses this issue.20–23 This theory, previously applied to women's health behaviors,23 posits that people pass through 2 stages before they contemplate the overall benefit of protective action. The first stage is global awareness of the threat, and the second is personal engagement (i.e., perceiving the threat at a personal level). Although global awareness of HIV threat is probably widespread in this third decade of AIDS, many women may not perceive HIV threat at the personal level, particularly those who feel geographically insulated from the AIDS epidemic because they do not reside in or near AIDS epicenters.

The purpose of this study was to compare HIV-associated sexual health history, risk perceptions, and sexual risk behaviors of low-income rural and nonrural African American women. We chose to conduct this study exclusively with low-income women because this population is likely to experience disproportionately high rates of HIV infection.24 Because AIDS has predominately affected urban and suburban women, we hypothesized that rural women would report sexual health histories, risk perceptions, and sexual risk behaviors that suggest comparatively less personal engagement in the threat of HIV infection.

METHODS

Study Sample

Data from a statewide survey of women attending federally funded Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) clinics in Missouri were used for this study. More than 90% of the Missouri counties are rural. We used a stratified random sampling scheme to control the selection of rural and nonrural counties. The strata were rural, suburban, and urban counties. Based on guidelines from the US Census Bureau,25 rural counties were defined as those with a population of less than 50 000. Alternatively, urban counties were defined as those with a population of more than 500 000, and suburban counties were defined as those with populations ranging from 50 000 to 500 000. Within each strata, simple random sampling was used to select clusters (i.e., WIC clinics). We purposely oversampled rural counties.

Twenty-one counties were selected: 17 were rural and 4 were suburban. Because Missouri had only 2 urban counties, these were automatically included in the sample. To confirm that the rural counties selected were not “fringe” counties of metropolitan areas, we identified each county's rural–urban continuum code as most recently assigned by the Office of Management and Budget.26 Office of Management and Budget rankings are based on the proximity of counties to metropolitan areas. The Office of Management and Budget has ranked each US county on a continuum of 0 (greatest degree of urbanicity) to 9 (greatest degree of rurality). The mean ranking of the 17 rural counties was 7.

The 23 selected counties contained 29 WIC clinics. Each clinic director was contacted by the principal investigator and solicited for his or her cooperation in the study. This procedure yielded a high participation rate: 27 of the 29 (93%) clinics agreed to participate. WIC clinics in 21 counties served as data collection sites. Women receiving WIC benefits from the 27 clinics were eligible to participate in the study if they were aged 18 years or older and consented to study participation. The Committee for the Protection of Human Subjects at Indiana University and the Internal Review Board for the Missouri Department of Health approved the study protocol.

Data Collection

Every 2 months, women who received WIC benefits came to the clinic to pick up their vouchers. Thus, to ensure that all women had a chance to participate in the survey, we chose to conduct the data collection phase during a 2-month period. From February through April 1998, all eligible women entering WIC clinics were asked if they would be interested in participating in a brief survey about women's health practices. Women who agreed to participate were given a self-administered survey and a preaddressed, postage-paid envelope for the return of the survey. Surveys were anonymous; however, they were coded to indicate rural, suburban, or urban location. Incentives for survey participation were not provided.

About 90% (4117) of the women solicited agreed to participate in the study, and 58% of these women returned a survey in the mail (n = 2391). Despite the lack of incentives, this return rate approximated previously reported rates obtained from studies of HIV- or AIDS-associated sexual behavior that used large probability samples.27–30 Although the statewide survey was designed to measure reasons that low-income women do not always use condoms for the prevention of HIV infection, the purpose of the current analysis was to compare HIV-associated sexual health histories, risk perceptions, and sexual risk behaviors of low-income rural and nonrural African American women. Thus, the data analyzed for this study represent only those collected from African American women responding to the statewide survey (n = 571).

Measures

The survey instrument was developed on the basis of the underlying theoretical model, a review of the relevant literature, and a series of ad hoc key-informant interviews followed by pilot testing. The interviews were held among approximately 12 low-income women from the target population. Findings from these women provided guidance in the refinement of language used in the questionnaire as well the overall structure of each question. The findings also were useful in establishing the relevance and reading level of the questions. Subsequent pilot testing of the instrument was conducted among 30 lowincome women from the target population. After completing the questionnaire, women were interviewed to determine which questions were uncomfortable to answer or were difficult to understand. Findings from this phase of the study provided the research team with feedback that confirmed acceptability and adequate comprehension of the questions.

Twelve measures hypothesized to vary by rural and nonrural residence were assessed (Table 1 ▶). Six measures were relevant to women's sexual health history, 2 were relevant to women's perceptions of risk, and 4 were relevant to women's sexual risk behaviors. With one exception, all variables were assessed with nominal measurement. The exception—frequency of condom use—was assessed with a 5-point Likert scale that ranged from “never” to “always.” For measures that involved a time-limited recall period, a period of 2 months was selected to enhance accurate reporting.31,32 The survey also included an item that assessed whether women were currently infected with HIV. Women who indicated HIV infection (n = 5) were not included in the analyses.

TABLE 1.

—Bivariate Associations Between Geographic Location and Selected Measures Among Rural and Nonrural African American Women (n = 571)

	% Rural	% Nonrural	PR	95% CI	P
Sexual health history
Ever diagnosed with an STDa	29.9	38.6	0.77	0.58, 1.03	.07
Ever diagnosed with syphilis or gonorrhea	9.0	17.4	0.51	0.29, 0.92	.02
Not counseled about HIV during last pregnancyb	25.0	13.3	1.89	1.27, 2.80	.002
No HIV or STD counseling from care providerc	97.5	93.3	1.04	0.98, 1.10	.31
Ever been HIV tested	52.6	57.4	0.92	0.75, 1.12	.38
Past or current partner has not been HIV tested	79.8	65.6	1.22	1.08, 1.37	.004
Risk perceptions
No chance of becoming infected with HIV	55.9	47.4	1.18	0.99, 1.41	.09
No preferred way to prevent HIV or STD infection because “I don't worry about HIV or STD”	18.8	11.6	1.63	1.04, 2.54	.03
Sexual risk behaviors
Never used HIV prevention methods	13.2	11.2	1.17	0.67, 2.06	.57
Never used condoms, past 2 mod	57.8	47.2	1.22	1.01, 1.48	.048
No condom use: believee partner is HIV negativef	27.2	17.4	1.56	1.02, 2.40	.04
No condom use: partner has tested HIV negativef	35.9	49.0	0.73	0.54, 0.99	.03

Note. PR = prevalence ratio; CI = confidence interval; STD = sexually transmitted disease.

^aIncludes chlamydia, gonorrhea, syphilis, human papillomavirus, and herpes.

^bAmong 505 women who had been pregnant and could recall whether or not they were counseled.

^cAmong 264 women who had a care provider under Missouri's managed care plan.

^dAmong 441 women who reported having sex with a man (or men) in the past 2 months.

^eDespite lack of partner testing for HIV.

^fAmong 351 women who had sex in the past 2 months and did not always use condoms.

Data Analysis

Rural vs nonrural residence served as the primary correlate of interest. The relation of this correlate to the measures shown in Table 1 ▶ was established through several sequential steps. First, because of skewed responses, the ordinal measure of condom use was collapsed to form a dichotomous measure (never vs some use). Strength of bivariate associations between geographic residence (i.e., rural vs nonrural county) and measures of sexual health history, risk perception, and sexual risk behaviors was assessed by prevalence ratios. To control for confounding effects, demographic variables were tested before conducting the primary analyses. Also, 2 potential confounders relevant to the analysis of frequency of condom use were assessed: pregnancy status and frequency of sexual intercourse.

Measures of sexual health history, risk perception, and sexual risk behaviors achieving a screening level of bivariate significance (P < .10) were sequentially tested for significance in the presence of the observed covariates. Thus, a separate logistic regression analysis was conducted for each outcome measure achieving bivariate significance. This process yielded adjusted odds ratios, 95% confidence intervals, and corresponding P values.

RESULTS

Characteristics of the Sample

Of the 2391 women responding to the statewide survey, 24% self-identified as African American. Among the African American women, the majority (72%) were from urban counties, with 4% and 24% coming from suburban and rural counties, respectively. Table 2 ▶ provides a comparison of demographic and other potentially confounding variables (i.e., current pregnancy status and frequency of sex) between African American women from rural and nonrural counties. As shown, only 1 variable (education level) was identified as a potential confounder.

TABLE 2.

—Analyses of Differences Between Rural and Nonrural African American Women (n = 571)

	Rural	Nonrural	P
Mean age (SD)	26.2 (7.6)	26.3 (6.7)	.85
Monthly household income < $600	48.2%	45.3%	.51
Monthly household income < $1200	85.1%	79.4%	.11
Education level = high school graduation	75.2%	59.3%	.0001
Had sex only with husband, past 2 mo	17.5%	15.7%	.63
Currently pregnant	22.6%	21.4%	.75
Mean frequency of sex, past 2 mo (SD)	10.8 (18.8)	8.3 (10.1)	.15

Bivariate Associations

Table 1 ▶ shows the percentages of rural and nonrural women reporting various measures of sexual health history, risk perception, and sexual risk behaviors. Table 1 ▶ also shows the bivariate associations between geographic residence and the assessed measures. As shown, 7 of the 12 measures achieved significance at P < .05.

Logistic Regression Analyses

Table 3 ▶ includes the adjusted odds ratios for measures that achieved the screening level of bivariate significance. After adjustment for the observed covariate (level of education), the 2 measures that achieved a significant screening level of significance (P < .10) remained nonsignificant. Adjusted odds ratios indicated that rural women were about 2 times more likely than urban women to report (1) a lack of HIV counseling during their last pregnancy, (2) that a sex partner had been tested for HIV infection, (3) no preferred HIV prevention methods because they did not worry about becoming infected by HIV or sexually transmitted diseases, (4) never using condoms, and (5) not using condoms because they believed that their current partner was HIV negative, despite lack of HIV testing. Rural women were about 50% less likely to report (1) ever having gonorrhea or syphilis and (2) not using condoms because the current partner had tested negative for HIV infection.

TABLE 3.

—Multivariate Associations Between Geographic Location and Selected Measures Among Rural and Nonrural African American Women (n = 571)

	AORa	95% CI	P
Sexual health history
Ever diagnosed with an STDb	0.72	0.48, 1.10	.13
Ever diagnosed with syphilis or gonorrhea	0.49	0.25, 0.94	.03
Not counseled about HIV during last pregnancyc	2.28	1.27, 3.78	.001
Past or current partner has not been HIV tested	2.07	1.24, 3.45	.005
Risk perceptions
No chance of becoming infected with HIV	1.31	0.88, 1.94	.17
No preferred way to prevent HIV or STD infection because “I don't worry about HIV or STD”	1.87	1.09, 3.23	.02
Sexual risk behaviors
Never used condoms, past 2 mod	1.77	1.15, 2.75	.009
No condom use: believee partner is HIV negativef	1.83	1.03, 3.24	.04
No condom use: partner has tested HIV negativef	0.58	0.35, 0.95	.03

Note. AOR = adjusted odds ratio; CI = confidence interval; STD = sexually transmitted disease.

^aAdjusted odds ratio; the reference category is nonrural.

^bIncludes chlamydia, gonorrhea, syphilis, human papillomavirus, and herpes.

^cAmong 505 women who had been pregnant and could recall whether or not they were counseled.

^dAmong 441 women who reported having sex with a man (or men) in the past 2 months.

^eDespite lack of partner testing for HIV.

^fAmong 351 women who had sex in the past 2 months and did not always use condoms.

DISCUSSION

Controlled analyses generally supported the study hypothesis. Fewer rural than nonrural women reported sexual health histories, risk perceptions, and sexual risk behaviors that suggested personal engagement in the threat of HIV infection. For example, fewer rural than nonrural women reported ever being diagnosed with gonorrhea or syphilis. Nonrural women may have been influenced by this adverse experience to the extent that diagnosis enhanced their perceptions of personal vulnerability to HIV. The lack of counseling about HIV during pregnancy also may have contributed to the lack of engagement among rural women. The finding that rural women were more likely to report that a sex partner had not been tested for HIV suggests that their partners also may have been unengaged by this threat.

In addition, lower condom use among rural women may be, at least in part, a result of lack of concerns about HIV. The finding that rural women were more likely to indicate that they did not use condoms because they believed that their partner was HIV negative is also important. Because this belief was based on something other than the partner's HIV test, the finding suggests that rural women may be more likely than nonrural women to “take their partners' word” that they are HIV negative. Similarly, rural women were less likely than nonrural women to report that they did not use condoms because their partner had been tested for HIV. Although nonuse of condoms on the basis of partner HIV testing could be considered risky (i.e., the partner may have lied about being tested, or the partner may have acquired HIV after the last test), this practice is the best public health alternative to consistent and correct condom use throughout the course of a relationship—an especially unrealistic goal in the context of long-term relationships, particularly those that involve intent to conceive a child.

A few related studies provide support for our study findings suggesting low personal engagement in the threat of HIV infection among low-income rural African American women. For example, a recent study reported that rural minority and low-income women living with HIV or AIDS typically had believed, before their diagnosis, that they could not get infected or that their sex partners were not infected.33 In another study of persons living with HIV or AIDS, participants were more likely to report that they had acquired HIV in a rural as opposed to a nonrural area.34 A study of predominantly African American adults attending a rural sexually transmitted disease clinic also provided evidence that rural women may commonly engage in behaviors that place them at high risk for HIV infection.35 However, none of these studies made analytic comparisons between rural women and their nonrural counterparts; thus, our findings represent a starting point for subsequent empirical investigations designed to identify behavioral differences between low-income rural and nonrural African American women.

Limitations

The findings of this study were limited by the use of self-reported measures. Also, the study did not assess a comprehensive range of HIV-associated risk and protective behaviors (e.g., injection drug use, recent HIV testing). A further limitation was that we did not assess the relative extent of the 2 groups' relocation between rural and nonrural areas. Differential relocation rates between rural and nonrural women could influence the study findings. In addition, the study sample was limited to low-income African American women who attended WIC clinics; thus, findings cannot be generalized to lowincome African American women who do not receive WIC benefits. Furthermore, the response rate of 58% could indicate a response bias, thus limiting the generalizability of the findings. Finally, the findings cannot be generalized to low-income African American women from states other than Missouri; however, the findings can be used to guide similar research efforts in other rural states. More research is needed with larger and more diverse samples of African American women, as well as with rural women from other racial/ethnic groups and diverse socioeconomic strata.

Conclusions

Prevention efforts tailored to African American women are an important public health priority because of this population's disproportionately high risk of HIV infection. This study provides initial evidence suggesting that low-income rural African American women are less engaged than their nonrural counterparts by the threat of HIV infection. In general, rural African American women believed themselves to be less susceptible to HIV because of personal or partner-related perceptions of protection from the epidemic. Given the potential diffusion of HIV from the high-concentration epicenters to rural areas, these perceptions may be highly problematic in regard to the adoption of protective practices such as condom use. Thus, rural African American women may be an especially important population for HIV prevention initiatives. Further research should identify factors that affect personal engagement in HIV prevention among rural African American women. Such research is critical to guiding the development and implementation of effective HIV prevention programs for this population.