DiFranceisco, Wayne MA*; Pinkerton, Steven D PhD*; Dyatlov, Roman V PhD†; Swain, Geoffrey R MD, MPH‡§
Since its inception in 1985, publicly funded HIV counseling and testing (C&T) has remained the single largest and most expensive program for HIV prevention in the United States.1,2 As of 1992, the Centers for Disease Control and Prevention (CDC) estimated that approximately one third of the adult population in the United States (60 million men and women) had been tested for HIV and that at least one half of these tests had been performed at federally funded C&T sites.3 Currently, more than 2 million tests are performed annually at approximately 11,000 publicly funded locations in the United States, including dedicated HIV C&T sites, sexually transmitted disease (STD) clinics, drug treatment centers, hospitals, and prisons.4-7 Although CDC funding priorities have shifted in favor of a variety of other types of health education and risk reduction interventions, the CDC's support for local HIV C&T programs continues to be substantial. In the 1998 fiscal year, the CDC awarded more than $91 million to health departments throughout the United States and its territories in support of their HIV C&T programs. This constituted approximately 35% of the CDC's total annual expenditures for local prevention activities.6
The primary objectives of HIV C&T programs are (1) to provide individuals with the opportunity to learn their HIV serostatus and offer referrals for medical and psychosocial services to those infected and (2) to provide behavior change counseling to help clients avoid future infection or, for those already infected, to prevent them from transmitting HIV to others.8 With regard to the core activities of diagnosis and referral, HIV C&T has had a widely beneficial impact. Evidence regarding its effectiveness as a prevention intervention has been inconclusive at best, however. Some studies have detected significant decreases in unprotected sex among HIV-seropositive persons as long as 1 year after diagnosis. Similar results have been obtained from studies of serodiscordant couples. With few exceptions, however, these evaluations have observed little or no behavior change among individual C&T clients who tested HIV-seronegative as little as 3 months before follow-up.9-14
Several conclusions have been drawn from these research findings. These include the observation that HIV testing may have the unintended consequence of actually reinforcing preexisting behavior patterns among some clients, who interpret negative test results as validation for their high-risk activities.15,16 A related concern is that some individuals' continuing uncertainty or denial regarding their risky practices may induce them to adopt repeated testing as a kind of post hoc “prevention” strategy.16,17 Still another explanation for the low incidence of behavior change among seronegative C&T clients maintains that large numbers of these individuals may already have been practicing lower levels of risk behavior before HIV testing. Accordingly, it is argued that C&T resources need to be prioritized such that higher risk clients receive more aggressive intervention, whereas individuals with lower risk exposure attend standard education-only sessions or, where feasible, receive referrals to private service providers.15,17,18
A recent meta-analytic review of the literature evaluating HIV C&T programs offers a somewhat different perspective.2 Although Weinhardt et al's analysis2 replicates the association between serostatus and risk reduction after HIV testing that was found in previous studies, their multivariate models reveal several other significant predictors of behavior change. Such independent factors include sample seroprevalence, client age, individual volition for testing, and the length of time to follow-up. These findings suggest that the effectiveness of HIV prevention interventions undertaken in conjunction with testing is not predicated solely on test outcome. Evidence from research on the CDC's Project RESPECT tends to support this reasoning.19 This latter study found that even moderately “enhanced” C&T consisting of a brief 2-session program of interactive counseling based on behavioral science theory was more efficacious in reducing risk behaviors among clients than the didactic information-only counseling approach that the authors assert remains the standard at most C&T sites.
The present investigation sought to explore these issues further, focusing specifically on the association between safer sex practices and the length of time since HIV testing, among a sample of sexually active STD clients. We hypothesized that individuals who had recently been tested for HIV, regardless of the result, would report higher rates of condom use, even after controlling for a variety of other predictors of safer sex. These control variables were selected from 5 broad domains, including sociodemographics, sexual orientation, and relationship status; HIV status and testing history; STD history and recent infection status; motivations for HIV testing and/or behavior change; and previous exposure to HIV education and prevention activities.
MATERIALS AND METHODS
Participants and Setting
The investigation was based on a cross-sectional survey of clients of the Milwaukee Health Department (MHD) STD Clinic. The MHD STD Clinic, which is centrally located in the downtown area of the city, sees approximately 8000 clients annually. Although the clinic provides services to all populations at risk for HIV and other STDs, its clientele is predominantly composed of inner-city African-American men and women. Study participants were recruited from among clients with appointments for HIV and/or STD diagnosis or treatment. Those who agreed to participate completed an anonymous self-administered questionnaire in a designated area of the clinic's waiting room. Each respondent was paid a $5 stipend for completing the survey. Between September 1999 and February 2000, a total of 401 clients (201 men and 200 women) were recruited.
Survey Assessment and Study Measures
The study assessment measure required approximately 30 minutes to complete and elicited information in several areas.
Sociodemographic and Other Background
In addition to items on their gender, race/ethnicity, age, education, and income, participants were asked to respond to questions about their sexual orientation (dichotomized as “gay or bisexual” vs. “heterosexual”) and current sexual relationship status (dichotomized according to whether a person reported that he or she was or was not “in an exclusive relationship with 1 other person”).
HIV and Sexually Transmitted Disease Status and Testing History
Respondents were asked about the total number of times they had been tested for HIV as well as the date (month and year) and result of their most recent serologic test. The number of months since one's most recent HIV test was computed by subtracting this test date (assuming the first day of the indicated month) from the date of the survey and then dividing the resulting total days by 30.42. Individuals who had been tested for HIV more than 3 times (the sample median) in the past were designated as repeated testers.
Respondents also were asked to report whether they had tested positive for 1 or more specific STDs at any time in the past and within the prior 3 months. Dichotomous measures were coded to indicate those respondents who had had a recent STD (“within the past 3 months”) and those who had “ever” had an STD (but excluding those respondents who also had tested positive for an STD within the past 3 months).
Motivations for HIV Testing and Behavior Change
Respondents were asked to indicate 1 or more factors that had motivated them to obtain their most recent HIV test. One of these noted concerns that their sexual behavior or that of their partner(s) may have placed them at risk for infection. Another set of questions assessing motivations for behavior change was derived from the Health Belief Model.20 These items elicited respondents' evaluations of self-perceived risk of HIV infection (“somewhat” or “very likely” vs. “not at all” or “only a little likely”), worries about becoming infected (“somewhat” or “very worried” vs. “not at all” or “only a little worried”), and perceived severity of AIDS illness (“worst thing that could happen” vs. “not the worst thing” or “not that terrible”). Persons who were already HIV-seropositive (n = 11) were instructed not to answer the “perceived risk” and “worries” items. To retain these previously positive cases for multivariate analysis, we assigned them to the modal response categories (low perceived risk and low worries) for these variables.
Previous Exposure to HIV Prevention Information and Activities
A series of items inquired about the respondents' exposure to HIV prevention education messages from various sources. These included receiving information about AIDS from news articles, television programs, radio programs, and teachers at school. Other items elicited whether respondents had ever engaged in discussions about AIDS with friends and family or whether they had ever participated in a formal HIV prevention program.
Respondents also were administered a battery of 11 items testing their knowledge about HIV/AIDS disease and risks for transmission (sample items: “You can get HIV by kissing an infected person” and “Having a sexually transmitted disease makes it easier to get HIV”). We computed an index of HIV/AIDS risk knowledge by summing the number of correct responses from each respondent. The mean number of correct responses on this index was 8.3 (SD = 1.7).
Sexual Activities and Condom Use
A series of items elicited information about the clients' sexual practices during the 3 months before their clinic visit, including number of vaginal and anal intercourse partners and occasions. Condom use was assessed using 2 items. On the first, respondents reported their frequency of condom use for all occasions of vaginal and anal intercourse during the past 3 months on a 5-point Likert scale (“never,” “sometimes,” “50% of the time,” “most of the time,” and “every time”). Because only a small percentage of the sexually active respondents (11.2%) reported consistent (“every time”) condom use over the prior 3 months, we operationalized the outcome measure as “frequent condom use,” which dichotomized respondents into those who indicated that they used condoms for intercourse at least “most of the time” during the prior 3 months (32.7%) versus those who reported using condoms less frequently or not at all. This definition of safer sex falls within the parameters of condom use frequency (ie, greater than 50% of the time) that Pinkerton and Abramson21 argue results in a considerably reduced risk for HIV infection. The second condom use outcome measure consisted of an item that asked the respondents whether or not they used a condom on the occasion of their most recent anal or vaginal intercourse. Although these 2 condom use outcome measures were not independent of each other, they were only moderately correlated (r = 0.44).
We performed univariate tests (contingency tables and logistic regressions) to evaluate the associations between the predictor variables and both condom use outcome measures. Odds ratios (ORs) and 95% confidence intervals (CIs) were computed for each relationship. Next, we examined the relative contributions among various predictors of condom use by fitting logistic regressions to the data for each condom use outcome variable. Each model analyzed the association between condom use and time since HIV testing, adjusting for individual variations in demographic background, HIV and STD testing histories, motivations for testing and/or behavior change, and previous exposure to HIV prevention education. Predictors that had obtained a significance level of at least P < 0.10 in the univariate analysis for either outcome were initially included in both stepwise regressions.
Characteristics of Selected Cases
Of the total 401 respondents, 70 had never been tested for HIV. Thirteen individuals reported that they had been tested but did not provide a test date and thus could not be included in the analysis. Twenty respondents reported no sexual activity during the 3 months before the survey. Another 6 persons who reported some sexual activity during the previous 3 months did not complete the condom use measures. Thus, the analysis included 292 recently sexually active clients (148 male and 144 female) who had previously received an HIV antibody test (at a specified time) and who provided information on their use of condoms during the 3 months before the clinic visit. As Table 1 indicates, most of these respondents identified themselves as exclusively heterosexual (94.5%) and never married (81.4%), although nearly half (47.9%) reported current involvement in a monogamous sexual relationship. The mean age of the study sample was 29.4 (SD = 9.3) years, and almost half (44%) of the respondents were less than 26 years old. African Americans comprised 79.1% of the respondents; 10% of the sample was white, 5% was Hispanic, and 6% was of other or mixed race/ethnicity. Approximately 36% of the respondents had not completed high school, and almost 48% earned less than $10,000 annually. Eleven (3.8%) clients reported testing HIV-positive on their most recent serologic test.
Comparisons between the study sample and the 70 respondents who reported that they had never been tested for HIV revealed no significant intergroup differences on the basis of sociodemographic variables (see Table 1). Additional comparisons of the study sample with 63 sexually active of the 70 nontesters (not shown) revealed no significant differences on condom use. Conversely, study respondents reported significantly greater exposure to radio programs about AIDS (P < 0.01), were more likely to have engaged in discussions with family and friends about HIV/AIDS (P < 0.03), and were more likely to have previously participated in an AIDS prevention program (P < 0.05) than those who never been tested for HIV.
Trends in Sexual Safety After HIV Testing
The mean length of time between the most recent HIV test and survey assessment was 17 months (SD = 29.6, range = 0-178 months). Half of the respondents had received an HIV antibody test within 6 months of the survey, and 105 (36%) individuals reported testing less than 3 months before the assessment visit. To mitigate the effects of this skewed distribution as well as to ensure adequate cases in each period to detect changes in the prevalence of safer sex, we divided respondents who had been tested for HIV less than 13 months before their visit to the clinic into 3 3-month and 1 4-month testing-to-assessment time intervals: 0 to 2 months (n = 105), 3 to 5 months (n = 34), 6 to 8 months (n = 36), and 9 to 12 months (n = 23); we expanded the last interval to 4 months to include an additional 9 cases in the category for analysis. A fifth group of respondents (the reference category) consisted of men and women who had received an HIV test more than 1 year (actually, 13 or more months) before completing the survey (n = 94).
Figure 1 displays the percentages of respondents reporting condom use at last intercourse and frequent condom use within each testing-to-assessment time interval. Almost half (48.1%) of the respondents who had been tested for HIV less than 3 months before assessment reported using a condom on the occasion of their last intercourse. For those who had been tested less recently, the percentage practicing safer sex according to this criterion decreased to between 31% and 34%, except for a slight nonsignificant increase among respondents in the 9- to 12-month HIV testing-to-assessment interval. Univariate logistic regression indicated that clients who had been tested recently (<3 months before the survey) were significantly more likely to have used a condom at last intercourse than respondents who had been tested more than 1 year before completing the survey (OR = 2.04, 95% CI: 1.14, 3.65). Respondents who had been tested 3 to 5 months, 6 to 8 months, or 9 to 12 months before the assessment did not differ significantly on condom use at last intercourse from those who had been tested more than 1 year previously. Post hoc comparisons of those who had been tested for HIV less than 3 months before assessment with respondents in the 3- to 5-month and 6- to 8-month intervals revealed a trend (P < 0.07) toward decreased condom use beyond the third month after HIV testing, however.
The percentages of respondents who reported using condoms “most of the time” or “every time” during the previous 3 months increased from 36.2% among those who had been tested less than 3 months before the survey to 44.1% for those testing 3 to 5 months previously and then decreased to between 27% and 30% among respondents in subsequent periods (see Fig. 1). Univariate logistic regression analysis suggested that the odds of frequent condom use were more than twice as high (OR = 2.18; P < 0.06) among respondents who had been tested 3 to 5 months before the assessment than among those who had been tested for HIV more than a year before completing the survey.
Other Predictors of Condom Use
Statistical analyses were performed to test the univariate associations of each condom use outcome measure with the other predictors. The following groups were more likely to have used condoms at their last intercourse: African Americans (OR = 1.93, 95% CI: 1.03, 3.62), men and women who were not in monogamous relationships (OR = 1.70, 95% CI: 1.05, 2.75), persons who tested seropositive for HIV (OR = 4.68, 95% CI: 1.21, 18.03), and those who had received HIV/AIDS prevention messages from schoolteachers (OR = 1.91, 95% CI: 1.18, 3.11). Conversely, those respondents who indicated that they were “somewhat worried” or “very worried” about getting AIDS were less likely to have practiced safer sex (OR = 0.49, 95% CI: 0.3, 0.8) at last intercourse than those more sanguine concerning their exposure. Participating in a formal HIV prevention program (OR = 1.66) was marginally associated (P < 0.10) with condom use at last intercourse.
Significant univariate predictors of higher rates of frequent condom use over the past 3 months included homosexual or bisexual orientation (OR = 5.03, 95% CI: 1.69, 14.92), HIV-seropositive status (OR = 5.95, 95% CI: 1.54, 22.96), discussions with family and friends about AIDS-related topics (OR = 2.17, 95% CI: 1.03, 4.55), and participation in an HIV prevention program (OR = 2.05, 95% CI: 1.14, 3.68). Respondents who believed that AIDS would be “the worst thing that could happen to someone” were significantly less likely to use condoms frequently than those who indicated a lower severity rating for AIDS (OR = 0.54, 95% CI: 0.3, 0.97). Marginally significant (P < 0.10) results suggested higher rates of frequent condom use among male respondents, African Americans, repeated HIV testers (>3 times), and those exposed to HIV prevention messages on the radio.
Variables that were not significantly, or even marginally, associated with either condom use measure in the univariate analysis included age, education, income, history of STD infection (either within the previous 3 months or in the more distant past), self-perceived risk of HIV infection, concerns that one's own (or a partner's) sexual behavior had placed the respondent at risk, HIV/AIDS knowledge, and having encountered HIV prevention messages in news articles or television programs. Consequently, these variables were not considered as predictors in the multivariate regression analyses.
Multivariate Models for Predicting Safer Sex Practices
The results of the multiple logistic regression predicting condom use at last intercourse are presented in Table 2. The odds of using a condom the last time one had intercourse were more than 2 times greater among recently tested respondents (<3 months) than for those who had been tested more than a year before completing the survey (OR = 2.18, 95% CI: 1.18, 4.01). For respondents in subsequent time intervals (3-5 months, 6-8 months, and 9-12 months), condom use rates did not differ significantly from those among individuals in the reference category (>1 year). Three of the other predictors tested were significantly associated with the outcome measure. Individuals who were not monogamous as well as those who had received some type of HIV/AIDS instruction from schoolteachers were significantly more likely to have used a condom during their last intercourse. However, respondents who indicated more acute worries about becoming infected were less likely to practice safer sex than did those who reported little, if any, concern (OR = 0.46, 95% CI: 0.27, 0.77).
Results from the regression model predicting frequent condom use (“most of the time” or “every time”) in the past 3 months are presented in Table 3. The odds of reporting frequent condom use were almost 2.5 times greater among respondents in the 3- to 5-month testing-to-assessment time interval than for those individuals who had been tested for HIV more than 1 year before their clinic visit (OR = 2.48, 95% CI: 1.03, 5.95). Findings also revealed a marginal tendency (P < 0.07) for higher rates of condom use among recently tested individuals (<3 months) versus respondents in the >1-year category. Rates of reported safer sex among clients in the testing-to-assessment intervals beyond 5 months did not differ significantly from those in the reference category. The regression results also indicated significantly higher rates of frequent condom use among male respondents, gay or bisexual respondents, and HIV-seropositive persons. Respondents who indicated that AIDS would be the “worst thing that could happen to someone” reported lower levels of frequent condom use.
To examine whether the associations between condom use and time since testing were moderated by other factors, we reran the regression models to include interactions between intervals of time since testing and each predictor. None of these interactions was significantly associated with either condom use outcome measure.
Next, we reanalyzed the multivariate regressions for each outcome measure using alternate cutoff points for time since HIV testing. For condom use at last intercourse, respondents who had been tested for HIV less than 3 months before their clinic visit were compared with all clients who had been tested 3 or more months earlier. Results indicated that recently tested clients had significantly higher rates of condom use than those respondents who had been tested for HIV 3 or more months before assessment (OR = 1.92, 95% CI: 1.15, 3.21). Because previous results had indicated that frequent condom use peaked at 3 to 5 months after HIV testing, the reference group for this reanalysis consisted of all clients who had received their HIV test 6 or more months before visiting the clinic. Regression findings revealed a marginally significant (P < 0.06) contrast between respondents in the 3- to 5-month testing-to-assessment interval versus those in the (6 or more months) reference group.
Previous research has not demonstrated that HIV C&T is effective at reducing sexual risk behavior among those who test negative for the virus. We are not aware of any study that focused specifically on length of time since testing as a possible mediating variable in the association between HIV test results and condom use, however. From our survey of male and female STD clients, we found evidence that those who had recently been tested for HIV were more likely to practice safer sex at last intercourse than clients who had been tested 3 or more months before their clinic visit. This association between recent HIV testing and risk behavior remained significant even after we adjusted for the effects of other predictors of condom use, including client serostatus.
Multivariate analysis also revealed that clients who had been tested 3 to 5 months before assessment were significantly more likely to use condoms every time or almost every time they had intercourse than respondents who had been tested for HIV more than a year before the survey. Recent testers (<3 months) manifested a moderate rate of frequent condom use (36.4%) compared with respondents who had been tested 3 to 5 months before assessment (44.1%). This finding that frequent condom use peaked among respondents in the 3- to 5-month interval would ostensibly suggest that higher rates of condom use persist for relatively longer periods after HIV testing than the results for condom use at last intercourse would indicate. Nevertheless, it must be noted that frequent condom use was a measure of the respondents' behavior for a 3-month retrospective period. Thus, the actual time during which much of this reported safer sex took place was essentially within the initial months after HIV testing.
Our findings suggest the existence of a brief “surge” in condom use among these high-risk STD clinic clients, possibly lasting for a few months after HIV testing. After this 1- to 3-month window, the prevalence of safer sex declines to levels that we would presume to be more typical within this population based on reported levels of condom use among those who had been tested some time previously as well as among those who had never been tested for HIV.
Several other covariates predicted higher rates of condom use among the respondents, although none of these other predictors was significantly associated with both outcome measures. Results from our predominantly heterosexual sample indicated that male respondents were more likely than female respondents to report condom use on a regular basis. The data also showed higher rates of frequent condom use among small numbers of clients who were gay or bisexual or HIV-seropositive. Respondents who were not in a monogamous sexual relationship and those who recalled receiving HIV/AIDS prevention messages from schoolteachers were more likely to report using a condom during their last intercourse.
Based on the Health Belief Model of behavior change, we anticipated that acute worries of infection and perceived severity of HIV/AIDS would predict higher rates of condom use among the respondents.20 The statistical findings revealed that these indicators were negatively associated with the condom use measures, however. We made several unsuccessful attempts to shed light on these contradictory results, for example, by testing whether those individuals with elevated concerns about HIV shared any unique demographic or other background attributes that might be correlated with lower rates of condom use. Perhaps some individuals with acute worries about AIDS may also have a heightened sense of fatalism regarding their chances of infection that interacts with their inability, or unwillingness, to avoid high-risk sex. We were not able to test this hypothesis with our data, however. As previous research has shown, several key constructs of the Health Belief Model are not clearly associated with HIV-related behavior change, including condom use.22-25 One investigation of individuals seeking HIV antibody tests revealed that those who reported higher levels of anxiety about AIDS were actually less likely to return for their results.26 Another study found that perceived severity of AIDS was inversely related to an individual's attendance at HIV prevention workshops.27
In interpreting these findings, one needs to be mindful of their limitations. Our cross-sectional analysis does not enable us to confirm actual behavior changes among the respondents. Even if we infer behavior change based on the observed peak in the rate of condom use among recently tested clients, we cannot establish precisely when this change occurred relative to HIV testing. In other words, we cannot rule out the possibility that many clients reduced their high-risk sexual practices before being tested for HIV. Thus, our study does not offer evidence on the larger question concerning the effectiveness of existing C&T prevention intervention programs.
Our sample was similar demographically to the samples obtained in 2 large multicity studies conducted at public STD clinics nationwide19,28 and seems to be fairly representative of the overall clientele of such clinics (predominantly African American, economically disadvantaged, and relatively young). The particular circumstances surrounding a client's visit to an STD clinic introduce potential bias in the data that complicates interpretation of the relation between HIV testing and retrospectively reported condom use, however. For example, individuals with recent STD infections who return for a check-up may be more likely to report using condoms for intercourse. One could argue that in this context, practicing safer sex has little to do with putative effects of HIV counseling and testing. Also, some clients who report using condoms may have visited the clinic to get retested after the latency period for the detection of HIV antibodies to confirm their seronegative status before initiating unprotected sex with a new partner. With respect to STDs, the data indicate that clients who had had a recent infection (<3 months) did not report significantly higher rates of condom use than other respondents. Because concerns about the clients' privacy prevented us from asking why they were visiting the clinic on the day they were surveyed, we were unable to determine whether these individuals were at the clinic for a follow-up appointment or concerning a possible new infection.
Despite these limitations, our findings do indicate that sizable proportions of seronegative men and women practice safer sex for some time after they learn their HIV test result. This suggests the existence of a brief “window of opportunity” after testing, during which some recent C&T clients may be amenable to additional HIV primary prevention education efforts that are aimed at achieving longer term reductions in risky sex among these individuals.
1. Ickovics JR, Druley JA, Grigorenko EL, et al. Long-term effects of HIV counseling and testing for women: behavioral and psychological consequences are limited at 18 months posttest. Health Psychol. 1998;17:395-402.
2. Weinhardt LS, Carey MP, Johnson BT, et al. Effects of HIV counseling and testing on sexual risk behavior: a meta-analytic review of published research, 1985-1997. Am J Public Health. 1999;89:1397-1405.
3. Centers for Disease Control and Prevention. HIV Counseling and Testing in Publicly Funded Sites: 1993-1994 Summary Report. Atlanta, GA: Department of Health and Human Services, Centers for Disease Control and Prevention; 1996.
4. Centers for Disease Control and Prevention. HIV Counseling and Testing in Publicly Funded Sites: 1996 Annual Report. Atlanta, GA: Department of Health and Human Services, Centers for Disease Control and Prevention; 1998.
5. Centers for Disease Control and Prevention. National HIV testing day at CDC-funded HIV counseling, testing, and referral sites-United States, 1994-1998. MMWR Morb Mortal Wkly Rep. 2000;49:529-532.
6. Holtgrave DR, Thomas CW, Chen H, et al. HIV prevention community planning and communities of color: do resources track the epidemic? AIDS Public Policy J. 2001;15:75-81.
7. Centers for Disease Control and Prevention. Advancing HIV prevention: new strategies for a changing epidemic-United States, 2003. MMWR Morb Mortal Wkly Rep. 2003;52:329-332.
8. Centers for Disease Control and Prevention. Technical guidance on HIV counseling. MMWR Morb Mortal Wkly Rep. 1993;42:11-17.
9. Coates TJ, Stall RD, Kegeles SM, et al. AIDS antibody testing: will it stop the epidemic? Will it help people infected with HIV? Am Psychol. 1988;43:859-864.
10. Higgins DL, Galavotti C, O'Reilly KR, et al. Evidence for the effects of HIV antibody counseling and testing on risk behaviors. JAMA. 1991;266:2419-2429.
11. Rugg DL, MacGowan RJ, Stark KA, et al. Evaluating the CDC program for HIV counseling and testing. Public Health Rep. 1991;106:708-713.
12. Doll LS, Kennedy MB. HIV counseling and testing: what is it and how well does it work? In: Schochetman G, George JR, eds. AIDS Testing: A Comprehensive Guide to Technical, Medical, Social, Legal, and Management Issues. 2nd ed. New York: Springer-Verlag; 1994:302-319.
13. Phillips KA, Coates TJ. HIV counseling and testing: research and policy issues. AIDS Care. 1995;7:115-124.
14. Wolitski RJ, MacGowan RJ, Higgins DL, et al. The effects of HIV counseling and testing on risk-related practices and help-seeking behavior. AIDS Educ Prev. 1997;9(Suppl B):52-67.
15. Lupton D, McCarthy S, Chapman S. “Doing the right thing”: the symbolic meanings and experiences of having an HIV antibody test. Soc Sci Med. 1995;41:173-180.
16. Leaity S, Sherr L, Wells H, et al. Repeat HIV testing: high-risk behaviour or risk reduction strategy? AIDS. 2000;14:547-552.
17. Phillips KA, Paul J, Kegeles SM, et al. Predictors of repeat HIV testing among gay and bisexual men. AIDS. 1995;9:769-775.
18. Holtgrave DR, Reiser WJ, DiFranceisco W. The evaluation of HIV counseling-and-testing services: making the most of limited resources. AIDS Educ Prev. 1997;9(Suppl B):105-118.
19. Kamb ML, Fishbein M, Douglas JM, et al. Efficacy of risk-reduction counseling to prevent human immunodeficiency virus and sexually transmitted diseases: a randomized controlled trial. JAMA. 1998;280:1161-1167.
20. Rosenstock IM, Strecher VJ, Becker MH. The health belief model and HIV risk behavior change. In: DiClemente RJ, Peterson JL, eds. Preventing AIDS: Theories and Methods of Behavioral Interventions. New York: Plenum; 1994:5-24.
21. Pinkerton SD, Abramson PR. Occasional condom use and HIV risk reduction. J Acquir Immune Defic Syndr Hum Retrovirol. 1996;13:456-460.
22. Montgomery SB, Joseph JG, Becker MH, et al. The health belief model in understanding compliance with preventive recommendations for AIDS: how useful? AIDS Educ Prev. 1989;1:303-323.
23. McKusick L, Coates TJ, Morin SF, et al. Longitudinal predictors of reductions in unprotected anal intercourse among gay men in San Francisco: the AIDS Behavioral Research Project. Am J Public Health. 1990;80:978-983.
24. DiClemente RJ. Predictors of HIV-preventive sexual behavior in a high-risk adolescent population: the influences of perceived peer norms and sexual communication on incarcerated adolescents' consistent use of condoms. J Adolesc Health. 1991;12:385-390.
25. Walter H, Vaughn R, Gladis M, et al. Factors associated with AIDS risk behaviors among high school students in an AIDS epicenter. Am J Public Health. 1992;82:528-532.
26. Catania JA, Kegeles SM, Coates TJ. Psychosocial predictors of people who fail to return for their HIV test results. AIDS. 1990;4:261-262.
27. DiFranceisco W, Kelly JA, Sikkema KJ, et al. Differences between completers and early dropouts from two HIV intervention trials: a health belief approach to understanding prevention program attrition. Am J Public Health. 1998;88:1068-1073.
28. Irwin DE, Thomas JC, Spitters CE, et al. Self-treatment patterns among clients attending sexually transmitted disease clinics and the effect of self-treatment on STD symptom duration. Sex Transm Dis. 1997;24:372-377.
© 2005 Lippincott Williams & Wilkins, Inc.