Evidence-based medicine is "the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients." This model of medical practice has been extended to public health practice, such as the Centers for Disease Control and Prevention's (CDC) recent community preventive services guidelines project. Central to this practice is the identification of the best evidence to answer specific clinical questions, and the critical appraisal of that evidence. One such approach is to conduct a systematic review. Systematic reviews cull from the methods of meta-analysis and combine a comprehensive and detailed search for relevant studies, a critical appraisal of the quality of included studies, a qualitative synthesis of study findings and, if appropriate, a meta-analysis of the data to determine the combined effect size of similar interventions. Using this methodology, we have prepared the present report, which aims to identify the best evidence of effective HIV prevention interventions that are of use in African-American populations throughout the United States.
| HIV Infection and AIDS Among African Americans|
Although we are entering the third decade of the HIV pandemic, there is yet no cure or vaccine. At this time, our principal means for deterring the further spread of HIV remain behavioral risk prevention interventions. Thus, developing and implementing interventions that focus on behavioral prevention are of utmost importance.
Although the scope of the epidemic has reached into most segments of society, its disparity continues to be striking. Recent statistics have demonstrated that there is a significant over-representation of people of color among new HIV infections. With regard to African Americans specifically, this disparity has been evident since the first reports of what we now call HIV/AIDS. As early as 1982, African Americans were found to comprise 23% of the cases reported, while comprising only 12% of the entire U.S. population.(1,2) This trend has continued, and worsened. Recent statistics show that over 50% of new HIV infections are occurring among African Americans(3) although still only comprising approximately 12% of the U.S. population.(2)
This overrepresentation of African Americans is consistent across the major behavioral risk groups for HIV infection (men who have sex with men (MSM), injection drug users (IDU), heterosexuals, and youth/adolescents). For example, a recent Centers for Disease Control and Prevention (CDC) study of MSM in New York found that 33% of men testing positive for HIV were African American.(4) This same figure was also reported from a national study of young MSM.(5) In addition, of the over 25% of AIDS cases in men in the U.S. from drug use and heterosexual transmission, more than half were in African-American men. Among women and adolescents African Americans have the highest number of cases of HIV infection and AIDS compared to other racial groups.(3) These figures underscore the necessity for examining evidence for prevention interventions aimed at African Americans, focusing on the primary risk groups.
| Men Who Have Sex With Men|
MSM represent the largest risk category for HIV infection in the United States.(3) African Americans comprise 31% of the cases of HIV infections among MSM since the beginning of the epidemic.
Engaging in anal intercourse increases the risk of HIV transmission, when condoms are not used. During the late 1980s and early 1990s, dramatic increases in condom use for anal intercourse among MSM were observed.(6) For example, statistics from the San Francisco Men's Health Study demonstrated that between 1984 and 1988 the rates of unprotected anal intercourse declined from 69% to 10% (for insertive partner) and from 65% to 9% (for receptive partner).
However, those promising changes have not continued. Recent statistics have documented a significant increase in unprotected anal intercourse among MSM, and an increase in HIV infections has followed. These increases have been increasingly evident among African-American MSM. Statistics through June 2000 show that infection rates for African-American MSM nationally are almost double those of Whites, and more than triple those of Latinos.(3)
A recent study among MSM in New York City found that 33% of young African-American MSM surveyed and tested were found to be HIV-positive-compared to 2% of White men, and 14% of Latinos.(4) The results were especially puzzling because many behavioral factors associated with HIV risk were found to be lower in the African Americans surveyed (eg, history of syphilis or gonorrhea, lack of knowledge about HIV, having large numbers of sex partners). The findings were also surprising given that in the past two years, several groups (including the Congressional Black Caucus) have placed a special emphasis on ensuring that prevention efforts are targeted toward MSM of color, including African Americans. However, some have suggested that these numbers are a result of more longstanding deficits in adequate prevention measures aimed towards people of color, and that current efforts will not show significant impact on HIV infection rates for some time to come.
A study conducted by CDC found similar results in 6 U.S. cities (Baltimore, Dallas, Los Angeles, Miami, New York City, and Seattle). Valleroy and colleagues reported that among their sample of young MSM, there was a prevalence of 30% HIV infection among African Americans (compared to 7% among Whites, and 15% among Latinos).(5)
During most of the epidemic, African Americans have been notably lacking as participants in most of the studies investigating HIV risk behavior among MSM. It has been proposed that one reason for this is that many African-American MSM do not identify as being gay or homosexual, and, thus, prevention messages aimed toward the gay community do not reach them. Very few African-American MSM become part of the mainstream (ie, White) gay community.(7) In addition, specifically African-American gay communities are also fewer in number. Many African-American MSM are identified with their racial identity, rather than their sexual identity,(8) and have been likely unaffected by the myriad of prevention efforts aimed toward the traditional (ie, White) gay community. Furthermore, most African-American MSM associate the term "gay" with the White community.(9) Other issues that serve to complicate prevention efforts geared toward African-American MSM include racial discrimination, poverty, and homophobia. Given the statistics specifically regarding HIV infections, combined with other, more longstanding, societal problems, prevention efforts specifically targeting African-American MSM are all the more imperative.
| Injection Drug Users|
Drug use presents particular risks for HIV transmission. Indeed, injection is a far more efficient mode of transmitting HIV than sexual intercourse.(10) The practice of sharing drug injection equipment, and an increased likelihood of engaging in unprotected vaginal or anal sex with multiple sexual partners are both possible outcomes of injection drug use. Some studies have reported that during the initial period of the epidemic, needles and syringes were used up to an average of nine times before they were discarded.(11,12) Drug use, especially crack cocaine use, has also been associated with high rates of reported sexual activity.(13) In addition, drug use is often practiced among networks of people, which can facilitate the sharing of needles. Due to factors such as these, drug and alcohol use in general and injection drug use in particular have been major risk factors for HIV infection in the United States.(14) Currently, injection drug use accounts for approximately 25% of annual new infections.(3) This percentage has been previously reported as being between 33% and 50%.(14)
Among African Americans, injection drug use is a major risk factor for HIV transmission. Through June 2000 for African-American men, 34% of cumulative AIDS cases and 18% of cumulative HIV infections were among injection drug users. For the same time period, for African-American women, 42% of the cumulative AIDS cases and 17% of the cumulative HIV infections were among injection drug users.(3) The cumulative cases of HIV infection for African-American men who were classified as injection drug users are higher than for White men (9%) and comparable to rates for Latinos (19%).(3) For African-American women, cumulative cases of HIV infection among injection drug users are lower than Whites (27%) or Latinas (20%). However, national HIV reporting does not include data from several large states with significant HIV and injection drug use epidemics in African Americans (such as New York) and therefore, underestimates the number of infections. Examining the number of cumulative AIDS cases by exposure category demonstrates more similarity between these three ethnic groups for female injection drug users: White, 42%, African American, 42%, and Latino 40%.(3)
However, many researchers believe that the true percentage of HIV infections due to injection drug use is difficult to estimate given the illegal nature of drug use, and the hesitancy of injection drug users to admit to this behavior. Injection drug users, like other marginalized groups, seem to have responded better to treatment programs managed by non-governmental organizations.(10) Thus, evidence regarding the relative efficacy of behavioral interventions targeting drug abuse among African Americans is important to evaluate, in order to continue to positively impact rates of HIV infection in this group, and to ensure that all injection drug users have access to prevention programs.
Although heterosexual contact is the primary mode of HIV transmission around the world, in the United States it accounts for approximately 33% of new infections.(3) Among African Americans, a gender difference is evident. For African-American men, heterosexual contact accounts for 11% of the cumulative cases of HIV infection and 8% of cumulative AIDS cases. However, for African-American women, heterosexual contact was identified as the primary source of infection in 40% of the cumulative cases of HIV infection and 38% of cumulative AIDS cases.(3) When compared to Whites and Latinos, the rates for African-American men are higher for heterosexual contact, while the high rates for African-American women are very comparable to those for White and Latino women.
In addition to the rates of infection and AIDS cases described above, several survey studies have found increased HIV risk behavior in heterosexually active African Americans, such as the National Survey of Men,(15) and the National AIDS Behavioral Surveys.(16)
Thus, as heterosexual contact is the source of approximately one third of the new cases of HIV infection and is the largest mode of transmission of HIV infection for African-American women, evaluating prevention programs that address this population is a priority.
The challenge of how to improve prevention programs targeted toward sexually active youth/adolescents is a very important one, as this group exhibits high levels of sexual risk behavior. A large proportion of the young adults currently infected with HIV or diagnosed with AIDS were most likely infected during their adolescence.
Psychological factors unique to this age group place adolescents at increased risk due to their lack of perceived vulnerability. Changing risk behavior inherently involves identifying oneself as being at risk, and most surveys of adolescents have found that this age group does not perceive themselves to be at risk for most negative outcomes (eg, car accidents, HIV/STD infection, pregnancy, etc).(17,18) Adolescents are also at risk through several pathways, thus, interventions need to be tailored to the specific population of youth that is being targeted (eg, MSM, heterosexually active youth, etc).
For African Americans this is a group at increased risk. African Americans between the ages of 13-24 comprise approximately 35% of reported AIDS cases in this age group.(19) Across all ethnic groups the majority of both recent and cumulative HIV infection for 13-19 year old males were among MSM (50%). For 13-19 year old females, however, heterosexual contact accounted for 50% of the new and cumulative cases of HIV infection.(3) Additionally, for both males and females, a high percentage of both HIV infection and AIDS cases are of unknown risk category (between 28-55%). In order to clarify those cases of unknown risk, one study of adults in six states sought to follow up on cases that fell into this category. The authors reported that they were able to then identify the risk category for 44% of those previously unidentified cases, and that 34% were MSM, 37% were injection drug users, and 24% were infected by heterosexual contact.(20) While this study was not specifically focused on adolescents, it may offer some information regarding the possible breakdown of unknown infections in adolescents. For example, all of those risk groups are represented as known sources of HIV infections in adolescents, with MSM and heterosexual contact accounting for the bulk of the infections in males and females, respectively. In short, the numbers given above for the percentages per risk category for adolescents could be significantly underestimated.
Several prior reports have reviewed the literature regarding the effectiveness of interventions to reduce HIV for various risk groups (eg, for adolescents(21,22,23); for injection drug users(14,24,25); and women(26,27)). However, fewer reviews have focused specifically on African Americans.(28,29)
For all of the above factors, an evaluation of effective behavioral interventions aimed towards African-American adolescents is urgently needed to identify best practices to prevent further increases in HIV infection.
In sum, African Americans, compared to other racial groups in the U.S., have the highest HIV prevalence, the highest incidence of HIV/AIDS, the highest HIV mortality, and the greatest number of years of potential life lost.(28) African Americans are at risk of HIV infection through all the major modes of transmission. Thus, it is imperative to review systematically our current knowledge of prevention interventions for the major behavioral risk groups, in order that future research efforts can be implemented in the most effective manner, thereby preventing additional negative consequences from HIV in the African-American community.
| Search Results and Description of Studies|
Overall, 137 relevant studies with ethnic minority participants were identified, 82 (60%) from the Cochrane HIV/AIDS register; 33 (24%) from AIDSLINE; 32 (23%) from MEDLINE; 27 (20%) from psycINFO; and 24 (18%) through contacts or scanning reference lists [as some studies were identified through more than one source, figures do not add up to 100%).
Of the 137 studies of ethnic minority participants, 36 (26%) focused exclusively on African Americans. Ninety-three (68%) studies had participants of mixed race/ethnicity including African Americans (93 studies [68%]), Latino/Hispanic populations (73 [53%]); White populations (71 [52%]); Asian/Pacific Islander populations (16 [12%]); Native American populations (11 [8%]); and other racial/ethnic groups (52 [38%]).
| Description of Studies Including African-American Participants|
Of the 129 studies including African-American participants, 121 (94%) tested one or more interventions that targeted individual behavior only; 6 (4%) tested one or more interventions that included an explicit attempt to change peer or social norms; and 2 studies (<2%) tested an intervention aimed to change individual behavior through the introduction of a policy or administrative decision. The vast majority of studies (106 [82%]) included populations whose primary risk for HIV transmission was heterosexual sex, 59 (46%) studies included populations with drug-related risk behavior, and only 8 (6%) included populations whose primary risk was through same-sex behavior. Again, some studies included more than one population, so the numbers exceed 100%.
Seventy-three (53%) studies had study populations of mixed age; 46 (34%) focused on youth (less than 21 years old), 11 (8%) on adult populations, and 7 (5%) studies did not specify the age of the study population. In terms of gender, the majority of studies (80 [62%]) were conducted with mixed gender populations; 36 (28%) studies were conducted with women only, 12 (9%) with men only, and 1 study did not specify the gender of the study population.
Forty-one studies (32%) were conducted in a health setting, 30 (23%) in a community setting, 25 (19%) in an education setting, one in the workplace, 24 (19%) in other settings (including outreach, household), and 27 (21%) did not specify the setting were the intervention was delivered (as some studies identified more than one setting, figures do not add up to 100%).
The interventions tested included: information (118 studies [91%]), skills training (74 [57%]), distribution of risk reduction materials such as condoms and drug paraphernalia (38 [29%]), HIV testing and counseling (26 [23%]), referrals to health and/or social services (25 [19%]), peer education (22 [17%]), personalized advice or counseling (22 [17%]), mass media components (5 [4%]), provision of health and/or social services (4 [3%]), and other intervention components (63 [49%]).
The majority (86 [67%]) of the 129 studies were randomized controlled trials; 30 (23%) were non-randomized trials; and 13 (10%) used another study design (including pre- and post-test or post-test only). The effect of the tested intervention was assessed using sexual behavior outcomes (97 studies [75%]), knowledge (66 [51%]), drug behavior outcomes (47 [36%]), attitudes (37 [29%]), intentions (31 [24%]), skills outcome (24 [19%]), health outcome (20 [16%]), and other outcomes (71 [55%]).
Thirty-six studies were identified as including 100% African-American participants in their samples. Of these, 14 studies were excluded from our sample due to not meeting our methodological criteria for study design, inadequate analysis reported or multiple published reports from one study sample. Thus we report on 22 studies with 100% African-American participants. Nineteen studies were identified as including African-American participants as well as other racial/ethnic groups and reported separate analyses for the African-American participants. Of these, 6 studies were excluded from our sample due to not meeting our methodological criteria for study design, inadequate analysis reported, or multiple published reports from one study sample. Therefore, we report on 13 studies with separate analyses for the African-American participants. African-American participants comprised at least 50% of the sample in an additional 43 studies. However, no separate analyses were completed for these participants. Of these, a total of 17 studies included over 80% African-American participants within the sample and were thus included in our sample. The remaining 31 studies were excluded from this review either due to not meeting our methodological criteria, or the sample was comprised of less than 50% African-American participants, and there had been no separate analyses conducted for those participants. Thus, we had a sample of 52 potentially relevant studies.
Below we will review those 52 studies categorized by percentage of African-American participants, by risk group, and by quality. All studies described in text below are randomized control trials; the studies representing the best evidence are presented first, then following in descending quality, within each section. These studies are presented in Table
5 and Table
Three included studies were not randomized controlled trials; therefore, they are not presented in the text of this review, but are described in Table
7. Seventeen studies were randomized controlled trials with at least 80% African-American participants with no separate analyses. Because of the high percentage of African-American participants included in these 17 studies, we describe in the text below the studies. The 26 studies containing at least 50% African-American participants, but which did not conduct separate analyses, are presented in Table
| Studies with 100% African-American Participants|
| Men Who Have Sex With Men|
| 'Good' Studies|
We were able to identify only one study that was specifically aimed toward preventing HIV infection in African-American MSM. Although 7 other studies identified in our searches contained some percentage of MSM participants, either there were no separate analyses for the African-American participants, and/or no separate analyses for MSM in studies that included participants from more than one risk group. Peterson and colleagues conducted a randomized controlled trial, investigating the effects of interventions of varying lengths on the risk behavior of the participants.(34) The sample consisted of 318 African-American MSM recruited from San Francisco and Alameda Counties in California. The study was randomized through a blocked randomization process, in order to ensure sufficient recruitment for the groups, and the randomization itself was generated by a computer. The intervention was theoretically derived from the AIDS Risk Reduction Model (ARRM),(35) and its components included information, skills training, and a cultural component, which addressed issues of self-identity that may arise from being African-American men who have sex with men.
The intervention compared a wait list control group to two intervention groups, which differed in lengths of time. One intervention group attended three 3-hour weekly group meetings, while the other group attended only one 3-hour group meeting. The authors reported that compared to the control groups, both intervention groups reported decreased unprotected anal intercourse at both 12- and 18-month follow-up. The actual percentage decrease for the three-session intervention group was 50% at the 12-month follow-up, and this was also observed at the 18-month follow-up. When comparing the single-session intervention to the 3-session intervention, the 3-session group reported significantly less HIV risk behavior than those who participated in the single-session intervention, at both 12 (from 46% to 20% for the three-session intervention; and from 47% to 38% for the single-session intervention) and 18 month follow-up (from 45% to 20% for the 3-session intervention; and from 50% to 38% for single-session intervention). Thus, the dose of the intervention had a significant relationship to the desired outcome (less HIV risk behavior).
Some limitations of this study exist, however, in that there were significant differences between groups at entry, and the authors reported fairly high attrition (47% for the 3-session intervention, and 55% for the single session). Nonetheless, it is a well-designed study, and the results are positive. Given that this is the only published intervention study for African-American MSM, it demonstrated that this is a population that can be recruited successfully, and can show significant reductions in risk behavior following a tailored intervention. Of great importance, is that significant reductions in HIV risk behavior continued to be measured, and were observed, at a fairly long follow-up (18 months).
| Injection Drug Users|
We were able to identify only 2 studies that focused on injection drug users in which 100% of participants were African American and that met our inclusion criteria. Many of the studies were part of the large multi-site studies, conducted by the National Institute on Drug Abuse (NIDA) and the National AIDS Demonstration Research Program (NADR) which focused on reducing the HIV risk and drug use of IDU through community outreach utilizing education and counseling models. Most studies' interventions compared an enhanced intervention to a standard intervention. The various sites usually developed their own enhanced interventions, and the standard intervention usually consisted of HIV testing and counseling. The interventions typically focused on reducing HIV risk by changing high-risk drug and sexual practices.
| 'Fair' Studies|
Latkin and colleagues investigated the effects of an intervention on 178 injection drug users (85% male) recruited from Baltimore, Maryland, in a randomized control trial.(36) The intervention was based in social influence theory,(37) and had several components. It emphasized information and the influence of social norms, and attempted to change social norms through peer education, role-playing, and skills training. Both intervention and control participants received HIV testing and counseling. The intervention was composed of six sessions. At 18-month follow-up, the authors reported that the intervention group had engaged in significantly less injection drug use of cocaine and heroin, as well as decreased needle sharing for HIV-negative persons. The intervention was delivered by former heroin users and achieved positive results with reductions in drug use. This suggests that peers can have a significant impact on injection drug users. However, there were no sexual behavior outcomes measured. In addition, attrition was fairly high. Nonetheless, the follow-up was sufficiently long to detect significant maintenance of behavior change, and decreases in needle sharing to the degree this study achieved could be an important factor in reducing the incidence of HIV infection. Thus, this result is significant.
| 'Limited' Studies|
McCoy and colleagues reported on results from one of the NADR sites.(13) This particular intervention focused on 185 male injection drug users recruited from high drug-use neighborhoods in Miami, Florida. The intervention combined aspects of the health belief model,(38) the theory of reasoned action,(39) and conflict theory.(40) Components of the intervention included information, referrals, risk reduction materials, skills training, and helping the participants to learn to identify their individual barriers to behavior change and positive supports. The intervention group received the components in one group and one individual session. Both the intervention and control group received HIV testing and counseling. The authors reported that at follow up (no follow-up time was specified), the intervention group and the control group did not differ in their levels of HIV risk behavior-but both groups had significantly decreased their risk behaviors relative to baseline. In addition, crack cocaine use had significantly decreased for both groups, but again, there was no significant difference between the enhanced intervention and the standard intervention (control group). However, it is extremely difficult to evaluate the quality of this study as no information was given about attrition during the study, nor were any possible group differences at baseline presented (it was rated as having significant methodological limitations). In addition, although the design of a randomized controlled trial is strong, there was no information given as to the method of randomization, and detracts from any ability to generalize these results.
In sum, we identified two studies that met our inclusion criteria conducted among African-American injection drug users. The higher quality study used a six-session harm-reduction model-based educational intervention that resulted in decreased needle-associated risk behavior through 18 months of follow-up.(36)
We were able to identify several studies that focused on the prevention of HIV infection through heterosexual risk. Approximately half of the studies focused exclusively on women, while the other half focused exclusively on men. This distribution emphasized the necessity of tailoring interventions by gender, as heterosexual men and women have different prevention needs, which will be discussed at more length later in the report.
| 'Good' Studies|
A recent study by Dancy and colleagues(41) investigated the effects of how an intervention grounded in social cognitive theory,(42) aspects of the health belief model,(38) and the theory of reasoned action(39) influenced the HIV risk behavior of a sample of low-income women in Chicago. Two hundred eighty women were recruited from two geographically separate but environmentally and demographically similar communities, and the community was the unit of randomization. The intervention combined information, skills training, role-playing, and modeling in an attempt to increase self-efficacy and improve positive social norms. The intervention consisted of six 90-minute sessions, followed by three booster sessions at 3, 6, and 9 months. The women who were in the treatment community significantly increased protected HIV sexual behavior at the 6-month follow-up. However, by the 9-month follow-up they had returned to their level of risk at the 3-month follow-up. This pattern of regression towards baseline was also observed in the women's level of self-efficacy, which significantly increased at the 6-month follow-up, but also returned to 3-month-follow-up levels at the 9-month follow-up. Limitations of this study include significant differences between groups at entry and more than 20% attrition. However, the initial results are promising, and the inclusion of booster sessions served to prolong the positive effects of the intervention.
An earlier study of 128 low-income women was conducted by DiClemente et al, in San Francisco, California.(43) The intervention was based upon social cognitive theory,(42) as well as including important constructs from gender and power theory.(44) The intervention provided the women with information, peer education, skills training, and discussions regarding gender and racial pride. There were three groups: an enhanced intervention group (five sessions containing all four elements), a single HIV education session, and a delayed HIV education control. All of the sessions were two hours in length. The authors reported that at 3-month follow-up, the women in the social skills group (the enhanced intervention) had reported increased condom use, greater sexual self-control, greater sexual communication, greater sexual assertiveness, and an increase in partners' adoption of norms supporting consistent condom use. The social skills intervention was delivered by African-American women peer health educators. The authors posit that the success of the social skills intervention was due to their influence. They suggest that the peer educators may be seen as a more trusted source of information, communicate in a way that was easily understood by the participants, and provide a salient source for modeling social skills.(43) Some limitations include differences between groups at entry, as well as differences between groups in attrition at follow-up. In addition, a follow-up period of three months is relatively short. However, the study was well designed, and the results are encouraging. In addition, this intervention was included in the "HAPPA" list of effective and scientifically rigorous studies described above.(33)
Kalichman and colleagues have conducted several studies focusing on heterosexual risk, in both men and women, in various parts of the country. One of the strongest studies reported on the effects of a video-based intervention on HIV risk behavior.(45) Utilizing a sample of 117 men, recruited from sexually transmitted disease clinics, the authors investigated whether a video-based intervention, grounded in the information-motivation-behavioral skills model,(46) would increase condom use. The two groups both used videos to present the material; however, the intervention group's video was theory-based, while the comparison group simply provided basic educational information. There were no differences between the groups in knowledge, condom attitude, or behavioral intentions. However, there were significant behavioral differences between the groups with the intervention group reporting decreased use of both alcohol and drugs with sex at 3- and 6-month follow-up. In addition, at 3-month follow-up (but not 6-month follow-up), the intervention group reported an increase in using condoms during every episode of intercourse, an increase in talking with their partner about AIDS, decreased rates of intercourse, increased condom use, and an increase in carrying of condoms. These results are very encouraging, as this intervention technique would be useful and feasible in sexually transmitted disease clinics. The limitations are that there was 30% attrition at the 6-month follow-up and a relatively small sample size, which could account for some of the lack of findings due to inadequate power.
Few studies have examined the role that female condoms might play in improving the ability of women to protect themselves from HIV infection. Kalichman and colleagues investigated the effects of an intervention aimed at increasing female condom use in a sample of 105 women recruited from a sexually transmitted disease clinic.(47) The intervention compared a general women's health intervention (control) with a skills-building and culturally sensitive intervention targeted towards increasing use of female condoms. Both groups were exposed to a single two-and-one-half-hour session. At a three month follow-up, women in the intervention group were significantly more likely to have redeemed coupons for female condoms. At the one-month follow-up the intervention participants reported greater use of the female condom. However, this effect was strongest for those women who had only one partner. The authors had hypothesized this effect based on prior research findings that had found that the female condom was preferred by women living with their male partner.(48) Indeed this finding was true across groups, so that those women with a single partner reported significantly greater use of the female condom regardless of intervention group. The authors pointed out that the overall rates of female condom use remained quite low (20%), even in the group receiving the intervention. Although this method might be most relevant to women with only one partner, it is important to investigate female-controlled prevention options, and this well-designed study offers a good foundation for future research.
In an earlier study, Kalichman and his colleagues investigated how interventions that were similar in their time frame, but differed in their content, differentially impacted HIV risk behavior in a sample of 128 low-income women in Milwaukee.(49) The different interventions all consisted of information, skills training, and education, and were based on cognitive-behavioral theory.(42) There were four groups: (1) one education session and three sessions of sexual communication skills; (2) one session covering basic HIV education and three sessions of behavioral self-management skills; (3) one session of education, 1.5 sessions of behavioral self management, and 1.5 sessions of sexual communication skills; and (4) four sessions of risk education without skills training. All groups had four total sessions that met twice weekly. At 3-month follow-up, in all groups intentions to change risk behavior increased, as did condom use. In addition, groups with communication skills showed increased rates of talking to partners about sex and refusing unprotected sex. Groups with communication skills also had significant increases in the percentage of participants who had engaged in "any use of condoms." The lowest risk was for women who received both sexual communication and behavioral self-management skills. Some limitations of this study include the relatively small sample size, the relatively short follow-up period, and the fact that those women who completed the intervention were more likely to have a sexually transmitted disease and an injection-drug user partner at study entry than those who dropped out. However, the finding that the group that received the intervention containing both sexual communication and behavioral self-management is of interest and shows that a combination of skills training may broaden the impact of an intervention.
Kalichman and colleagues next conducted a randomized control trial aimed towards the prevention of HIV risk behavior in 81 heterosexual men recruited from social service agencies and housing developments in Milwaukee.(50) The intervention, grounded in cognitive behavioral theory,(42) was comprised of providing information and skills training and had a particular emphasis on identifying triggers/barriers to high-risk behavior. There were two intervention groups: one group was provided with cognitive-behavioral skills training, and the other was provided with risk education sensitization. Both groups were four sessions in length, and met two times per week for two weeks. The authors reported that there were no differences between groups on any outcome and that neither intervention increased condom use. However, across groups, there were increases in knowledge, intentions, and reductions in vaginal intercourse compared to baseline. The authors stipulated that possible reasons for the lack of findings included the small sample size and differences between the groups that were present at study entry. In addition, 45% of the participants dropped out during the intervention.
Interventions conducted in inpatient settings can be helpful in establishing positive baseline behaviors to be continued upon release from various types of inpatient programs (eg, substance abuse treatment). Malow and colleagues investigated whether or not such an intervention had an effect on a sample of 152 male patients who were participating in an inpatient drug dependence treatment program at a Veterans Affairs hospital in New Orleans.(51) The intervention was theoretically based in the AIDS risk reduction model (ARRM),(35) and contained information, skills training, and explanations of the benefits of safer sex and attempted to increase the patients' perceived susceptibility to HIV infection, and to help them identify barriers to changing high-risk behavior. There were two groups, an information control and an enhanced psycho-educational group that contained the elements described above. Both groups were conducted in 2-hour sessions on three consecutive days. At the 3-month follow-up assessment, the intervention group reported an increase in their communication skills, an increase in their condom use skills, and a decrease in their risk behavior. The participants who were classified as "high-risk" according to their behavior dropped from 75% at baseline, to 32% at 3-month follow-up. Across both groups, the number of sexual partners decreased.
Another randomized controlled trial that focused on the heterosexual risk of women who may have also had issues with drug dependence, was conducted by Nyamathi and colleagues among a sample of 300 women recruited from homeless shelters and drug recovery programs in Los Angeles.(52) The authors conducted additional analysis on a randomly generated sub-sample of a larger study population, in order to determine the effects of the original intervention two years later. The intervention included HIV testing and counseling, information, referrals, skills training, and risk reduction materials. In addition, it included aspects, derived from Nyamathi's comprehensive health-seeking and coping paradigm,(53) that aimed to improve the women's coping, self-efficacy, and ability to accurately appraise their own risk of HIV infection (threat appraisal). The standard intervention participants received one hour of basic AIDS education, while the enhanced intervention received two hours of "culturally competent" AIDS education and counseling. At a 2-year follow-up, (one of the longest follow-up periods included in this review), the authors found that the enhanced intervention participants reported less drug use behavior, increased social resources, less avoidant coping, and less emotional disturbance. Across both groups, the women reported less AIDS risk behavior compared to baseline, but there were no differences between groups. While these results are encouraging, many of the women were also in drug rehabilitation, which could have positively impacted their behavior, and it is difficult to discern the specific effects of the intervention.
Additional follow-up analyses from this same intervention were reported in a later paper.(54) These analyses focused on another computer-generated sub-sample of 410 women two years after the intervention. The authors reported that the women who had received the standard intervention were more likely to have multiple partners. Both intervention groups had reduced their injection drug use, but there were no significant differences between the groups. However, the enhanced intervention participants were significantly less likely to report non-injection drug use at both one- and two-year follow-up. As mentioned above, the length of follow-up (2 years) adds to the strength of these results, but it should be noted that this intervention appeared to have more positive effects on the drug use behavior of the participants, rather than on their sexual risk behavior.
Kalichman conducted one of the few studies that investigated whether or not men responded differently to different types of condoms (eg, latex vs. polyurethane) presented during an intervention.(45) A sample of 108 men, recruited from an urban sexually transmitted disease clinic in Georgia, was randomized into three groups. All of the groups met for one 3-hour session. Each of the groups received information, risk reduction materials (condoms), and skills training. One group focused on building condom skills with latex condoms, the second utilized polyurethane condoms, and the third was an HIV/AIDS education comparison condition. The theory for the intervention was derived from the information-motivation-behavioral skills model.(46) For both of the interventions, condom use at 1-month follow-up had increased, but this did not remain at 3-month follow-up. There were no differences between the groups in knowledge, attitudes, or intentions, and all groups improved in these areas. The polyurethane group was more likely to request condoms at follow-up. Limitations of this study include a relatively small sample size and the fact that all of the participants were recruited from a single site.
| 'Fair' Studies|
The earliest randomized controlled trial included in this review was by Kalichman and his colleagues who investigated the effects of an intervention among 106 low-income women in Chicago.(55) The trial consisted of three groups, each of which had a presentation of videotaped HIV information, that varied by ethnicity and cultural context of the presentation. The first group viewed a standard tape with a White broadcaster; the second group viewed the identical information presented by African-American women; and the third group viewed the same information delivered by African-American women, but the information was presented in a more culturally relevant manner. The intervention aimed to increase feelings of personal vulnerability and thus encourage participants to adopt more preventive behavior.(56) At a 2-week follow-up, there were no differences between the groups in seeking information about AIDS, attempting to use condoms more often, talking with sex partners about condoms, or purchasing condoms. However, women in the second and third groups were more likely to request condoms at follow-up, more likely to talk to friends about AIDS, and more likely to identify AIDS as a personal threat. Finally, women in the culturally relevant information group (Group 3) were more likely to be tested for HIV. The limitations of this study include a relatively small sample size, 28% attrition, and a very short follow-up period.
As mentioned above, women can be at increased risk of HIV infection through both injection drug use and heterosexual risk behavior. Harris and colleagues reported on a randomized controlled trial conducted among 204 methadone-dependent women.(57) The intervention consisted of counseling, information, and the "usual treatment" of methadone maintenance, which included support groups, compared to a "usual treatment" group. The intervention lasted for 16 weeks, and the women attended weekly 2- hour sessions for the first 8 weeks, followed by a weekly 1-hour session for the last 8 weeks. Across time, women in the treatment group engaged in safer sex practices (specifically condom use) than those in the usual-treatment group. In addition, the women in the treatment group reported being less depressed. However, there was more than 30% attrition at 3-month follow-up, and 28.5% of the women reported "not engaging in any type of sexual behavior" at the onset of the study.
In sum, we identified 12 studies among African-American heterosexuals that met our inclusion criteria. A variety of interventions were successful, both compared to baseline and control interventions. The most successful interventions included skills training, were tailored to the target group (eg, women), and were conducted over several sessions.
| 'Good' Studies|
Jemmott and his colleagues have conducted the interventions with the strongest methodological designs and the most promising results for African-American adolescents. The interventions have been conducted over a period of several years, and have improved their methodological strength over time. The methodological quality of these studies is consistently strong. Minor limitations will be discussed individually for each study.
Jemmot and colleagues conducted an intervention in a sample of 659 younger adolescents (6th and 7th graders) which contained both males (47%) and females (53%) in Philadelphia, Pennsylvania.(58) There were three groups-two intervention groups and a control. The theoretical basis for the intervention was social cognitive theory(42) and the theory of reasoned action.(39) The first intervention group focused on abstinence as the proper strategy to prevent HIV infection, whereas the second group was more focused on providing information regarding how to engage in safe sex. The control group was a health promotion intervention, not related to HIV, but rather to such issues as heart disease, diet, and exercise. All interventions were delivered in eight 1-hour modules divided equally over two consecutive Saturdays, in small groups. The interventions were delivered by both peer and adult facilitators.
The follow-up period extended through one year post-intervention, with interval follow-ups at 3 and 6 months. At 3-month follow-up, adolescents who had participated in the abstinence group were significantly less likely to have had sex than the control group and marginally less likely than the safe sex group. The safer sex group had more consistent condom use, higher frequency of condom use, and was more likely to have protected sex than either of the other two intervention groups. At the 6-month follow-up, the safer sex group again reported more frequent condom use. At the 12-month follow-up, both intervention groups reported more condom use than the control group. Only the safer sex group had significant and positive outcomes on frequency of unprotected sex. Across all of the follow-up periods, there were significant interactions between level of sexual experience and the effect of the intervention such that the intervention had the strongest effect on those students with previous sexual experience. There were little or no differences between the groups for students who were sexually inexperienced. In addition, there were no differences between adult and peer facilitators, which would facilitate the translation of the intervention.
Jemmott utilized this intervention design again with a sample of 496 older male (46%) and female (54%) adolescents who were recruited from public schools in Trenton, New Jersey.(59) For this particular study, the HIV intervention resembled those previously used, described above. The control condition received a general health promotion intervention, and both interventions were five hours in length. At the 3-month follow-up, the HIV group reported greater HIV knowledge, greater self-efficacy regarding condoms, more favorable prevention beliefs, and stronger intentions to use condoms. All of these results remained significant at the 6-month follow-up, with the exception of prevention beliefs. With regard to behavioral outcomes, there were no significant differences between the groups at 3-month follow-up. However, at 6-month follow-up, the HIV intervention participants reported less HIV risk behavior, and less anal intercourse. The original design had been strengthened by an additional follow-up, and retained the strengths of earlier studies with its design, low attrition, and extensive training of facilitators.
The origins of the above studies derive from the original randomized control trial, which was published in 1992. In this initial study, Jemmott and his colleagues(60) conducted an intervention based in cognitive-behavioral theory,(42) and the theory of reasoned action.(39) The sample was comprised of 157 adolescent males, who were recruited from medical clinics, high schools, and a YMCA in Philadelphia. The adolescents were randomized into two groups, both five hours in length. The intervention contained HIV risk reduction information, skills training, and role-playing. The control group received information on career training and opportunities. At a 3-month follow-up, the HIV intervention group reported greater AIDS knowledge, weaker intentions for risky sex, less risky behavior, fewer days of sex, and fewer episodes of sex without a condom. In addition, the participants reported that they had been involved with fewer women, and that they had been involved with fewer women who were involved with other men, when compared to the control participants. The study had a strong design, and very low attrition. One of the few shortcomings was a relatively short follow-up period (3 months). The short follow-up period was lengthened in later studies, as described above.
Stanton and colleagues have also done extensive testing of an intervention that they developed aimed at reducing HIV risk behavior in adolescents. They reported on a study(61) that utilized aspects of protection motivation theory(62) in the development of the intervention. The intervention also emphasized the importance of friends' influence for adolescents, and this was recognized by the use of delivering the intervention through naturally occurring friendship groups. The intervention group received information, risk reduction materials, and skills training, delivered through the mode of peer education, and delivered through groups of friends. The control group received HIV/AIDS information without the presence of friends and without a theoretical context. Both interventions consisted of 8 weekly meetings (seven 90-minute sessions and one day-long session). The authors reported that at 6-month follow-up the intervention participants reported increased condom use and increased condom use intentions. However, these promising results were no longer significant at a 12-month follow-up. The authors conducted analyses of contraceptive use that were presented in a separate paper.(63) These results showed that the intervention youths who used oral contraceptives also used condoms, thus demonstrating that they were acting to reduce their risk of HIV infection, as well as practicing contraception. This was a well-designed and executed study; however, significant attrition likely limited the strength of the findings.
This intervention was delivered through naturally occurring peer groups, and other reports from this project focused on this aspect of the intervention. Fang et al reported that at the 6-month follow-up the homogeneity of the intervention groups had increased from 31% to 46%, meaning that all members of the group exhibited similar risk behavior, and condom use was universal in the homogenous groups.(64) Also at 6-month follow-up, the percentage of groups in which all sexually active members of group used condoms increased from 33% to 69%, while the control group rates remained stable. This is one of the few studies that empirically investigated the influence of peers through an intervention delivered to groups of friends.
St. Lawrence and her colleagues conducted an extensive intervention in a sample of 246 adolescents recruited from a community health center.(65) Their theoretically-based(46) intervention consisted of 8 weekly education and behavioral skills sessions that were 90-to-120 minutes in length. The sessions included information, skills training, peer education, and role-plays and sought to increase self-efficacy. The comparison group received a single two-hour educational program. There were significant gender differences at baseline with males having much higher levels of risk behavior than females. Nonetheless, male intervention participants exhibited lower rates of unprotected sex compared to control participants at one-year follow-up. The female intervention participants maintained their low levels of risk behavior whereas the control group females increased their risk behavior. Intervention participants of both genders increased their condom use. In addition, the intervention participants who were not sexually active had significantly later rates of onset for sexual activity than non-sexually active control group participants. Finally, intervention participants demonstrated greater skills (eg, communication, negotiation) at follow-up. This study was methodologically strong, and demonstrated that effective change can be brought about in non-school-based studies with adolescents with relatively long-standing effects.
| 'Fair' Studies|
The intervention developed by Jemmott was used for a multisite study reported by Kennedy and colleagues.(66) Although the intervention was conducted across several cities, the Nashville, Tennessee site was the only site that met inclusion criteria for this review. There were 535 adolescent participants (59% female and 41% male) who were recruited from low-income housing. Again, using the intervention described by Jemmott and colleagues,(60,58,59) there were two groups--an HIV group and a wait list control. Again, the HIV group focused on presenting information, referrals, skills training, and role-plays. In addition, this intervention contained a separate intervention for parents. At 1-month follow-up, the intervention group reported increased skill, increased carrying of condoms, and decreased frequency of unprotected sex. However, there was very high attrition (44% for the intervention, 25% for the control), that could have limited the validity of the findings. Also, the 1-month follow-up period is quite short.
In sum, we identified 6 studies conducted in African-American adolescents that met our inclusion criteria. Similar to studies of adult heterosexual populations, these interventions included a variety of approaches. The highest quality studies were theoretically based, incorporated skills training, were culturally sensitive, and were comprised of multiple sessions. These studies demonstrated that adolescents' high-risk behavior could be modified successfully. Without exception, these studies demonstrated positive changes in HIV risk behavior of the adolescent participants. This evidence is promising for future interventions with this population.
| Studies in Which Some Percentage of the Sample is African American and There are Separate Analyses for African-American Participants|
| Injection Drug Users|
| 'Fair' Studies|
Booth et al reported on a multisite study (15 cities) conducted with out-of-treatment opiate injection drug users.(67) The sample was large (N=4,443), and there were both male and female participants (46% African American). The intervention provided the participants with HIV testing and counseling, information, risk reduction materials, and skills training. There were 2 groups. The standard intervention consisted of HIV testing and counseling. The enhanced intervention consisted of the standard intervention plus additional educational sessions. At a 6- month follow-up, significantly more participants from the enhanced intervention had entered into drug treatment, which, in turn, significantly predicted less drug use. Active referrals provided as part of the intervention significantly predicted drug treatment, leading the authors to conclude that this might be an important component of future interventions with this population. Limitations of this study are fairly high attrition (33%), although that is fairly typical of this population, and the interventions were not standardized across sites.
Investigating specific barriers to risk reduction was the aim of an intervention conducted by Copher and colleagues with a sample of adults recruited from housing projects in New Orleans, Louisiana.(68) The participants were 86% African American, and were both male and female. The intervention contained information, demonstrations and skills training, and coping training, aimed towards identifying individual obstacles to adopting less risky behavior. The control group only received 10 minutes of information regarding HIV/AIDS prevention, while the intervention group lasted for 3 hours. At 1-month follow-up, the enhanced intervention participants reported greater AIDS knowledge, and less needle risk through injection drug use. In addition, being African American predicted lower needle risk. However, overall, the intervention participants reported engaging in more sex risk behavior. Limitations that influenced the quality of this study included a small sample size and large attrition.
These studies demonstrate that interventions with injection drug users can effect positive change with regard to drug use. There is less evidence that they were successful in effecting risk due to sexual behavior.
Many of the studies investigating heterosexual behavior utilize samples recruited from sexually transmitted disease clinics. As an outcome measure, the authors often examine medical records in order to determine whether or not the patients returned to the clinic with a new sexually transmitted infection. This can be used as a proxy for HIV risk behavior, as condom use protects against most sexually transmitted infections as well as HIV infection.
| 'Good' Studies|
The National Institute of Mental Health reported on the results of a very well designed and executed study(69) utilizing similar methods to those described above. It was conducted in seven urban areas around the country and was comprised of 3,706 male and female patients recruited from community-based clinics (74% African American). There were two groups, a control group in which participants received one hour of HIV/AIDS education and an intervention group in which participants received seven sessions focused on HIV risk reduction. The sessions were 90 to 120 minutes in length and were conducted two times per week in small groups that were separated by gender. Follow-ups were conducted at 3, 6, and 12 months after the intervention. The enhanced intervention participants reported a reduced frequency of unprotected intercourse across each follow-up point, and in this group, unprotected intercourse decreased by 50% from baseline to 12-month follow-up. In addition, the intervention participants reported increased condom use across all time points, and they were more consistent condom users. With regard to medical outcomes, there was less gonorrhea in men from medical chart review. Of particular interest is that there was a significant association between how many sessions someone attended and the degree of behavior change later reported; thus, the dose of the intervention seemed to be a significant predictor of behavior. Analyses were conducted examining whether or not race was a significant predictor of behavioral outcomes, and it was found to be non-significant. Overall, this was a very well designed and executed study and includes the important aspect of separating men and women for the purposes of delivering an intervention. The intensity of the intervention could be a feasibility problem, but the positive results make a compelling argument for future interventions. One possible limitation is that there were fairly low rates of sexually transmitted diseases in the population. However, conducting this intervention in populations with higher sexually transmitted disease rates could only serve to strengthen the results, not lessen them. Finally, this study was included in the "HAPPA" list of well-designed, effective interventions described earlier.(33)
Another example of this type of study was conducted by Cohen and colleagues, who investigated the differential effects of several different types of interventions on a sample of adults recruited from a sexually transmitted disease clinic in Los Angeles, California.(70) The sample of 903 was 72% African American and was comprised of both men and women. There were four groups: a skills approach (including a condom demonstration), a social influence approach (derived from social learning theory), a condom access approach (cards were given out so that participants could obtain free condoms at businesses in the community), and a control group. All of the groups, with the exception of the control, received information, risk reduction materials, and skills training. The authors examined medical records in order to identify which participants returned to the clinic with a diagnosed sexually transmitted disease, approximately 6 to 9 months following the intervention. Across groups, the lowest rates of known sexually transmitted disease reinfection were found in young men between 20 and 30 years of age, and in African-American men exposed to the condom skills and social influence interventions. There were no differences between groups for women. This led the authors to conclude that the particular types of interventions given were more relevant to men, than to women. Overall, this was a well-designed and well-executed study. This intervention was also included in the "HAPPA" list of effective prevention programs (Card et al, 2001).
O'Donnell and colleagues investigated the influence of a video-based intervention on sexually transmitted disease clinic patients in New York City.(71) The sample of 2,004 was 62% African American and all male. There were three groups: one group that obtained all of their information through a video presentation, one group that received the video presentation along with participating in a group discussion, and a control group that received the regular sexually transmitted disease clinic services. All participants were offered free condoms. Records were examined for a diagnosis of sexually transmitted diseases approximately 18 months following the intervention. Those patients who received a video-based intervention had significantly lower rates of sexually transmitted infections when compared to those who received the regular clinic services; however, there were no differences between the two video-based groups. In addition, there was a significant reduction in sexually transmitted disease diagnosis for men who had previously reported multiple partners. Like the Kalichman et al study,(55) a video-presentation appears to have positive results in reducing risk behavior, and could be a feasible option for sexually transmitted disease clinics. This study was well designed and well executed, which serves to strengthen the power of the recommendations, one of which is that video presentation is a feasible and efficacious option for providing education to sexually transmitted disease clinic patients.
An earlier study by O'Donnell and colleagues utilized a similar theory-based intervention structure with a sample of 3,257 patients at a sexually transmitted disease clinic in New York City (62% African American).(72) The three groups consisted of a control, video-viewing only, and a video plus a group session. A differentiation between this and the later study described above was that the later study obtained a biological outcome (reinfection of sexually transmitted disease) whereas the 1995 study examined the redemption of coupons for condoms given to the participants. The video intervention participants were significantly more likely to redeem coupons for condoms than the control participants. Further, the video plus interactive group participants were more likely than video only participants to redeem coupons. Finally, African-American women were least likely to redeem coupons after exposure to the video alone. Because the intervention was designed to be culturally sensitive (all participants were either African American or Latino), this result could be more related to gender than ethnicity. Indeed, the authors described the video shown to African-American participants as being more geared toward men than women, and a differential impact by gender was observed in similar interventions (eg, Cohen(73)). The methodology of this study, like the one above, was strong but it was limited by the outcome of examining condom acquisition rather than report of actual condom use or a biological endpoint.
In sum, these studies demonstrate that patients at sexually transmitted disease clinics can be a receptive audience for messages aimed toward lessening their sexual risk behavior. Several methods produced positive results including videotape presentations, and small group sessions. An additional strength of these studies is utilizing a biological outcome (diagnosis of a sexually transmitted infection) by medical chart review. There appears to be evidence from the above studies that interventions differentially impact men and women and that intervention components (eg, video presentations) may need to be more specifically tailored for women. Finally, the interventions were also quite feasible, and could probably be implemented at relatively low cost by sexually transmitted disease clinics staff.
| 'Good' Studies|
Runaway adolescents in New York City have been found to exhibit relatively high rates of HIV risk behavior. Rotheram-Borus and colleagues have conducted several studies of this population, including interventions aimed towards changing the potentially high-risk behavior of this population. One such intervention was conducted with a sample of 312 runaways recruited from shelters for runaways (57% African American).(74) The unit of randomization was the shelter, thereby reducing the possibility of contamination between groups. The intervention included information, counseling, referrals, risk reduction materials, and skills training. There were 2 groups. The intervention group was conducted at two shelters, and the maximum dose that a participant could receive was 10 sessions. Two shelters were used as a control group. At 2-year follow-up, intervention participants reported greater reductions in unprotected sex and less substance use than control group participants. The intervention did not significantly decrease number of sex partners. With regard to ethnicity, African-American youths made greater reductions in substance use compared to Whites and Latinos. One of the strengths of this study was its length of follow-up, which was 2 years. Although there was a fair amount of attrition (30%), the fact that significant reductions in unprotected sex were still observed after 2 years is an important finding.
An important issue in evaluating interventions is their implementation and how feasible they are for use in community agencies. For example, investigating the amount of time that is needed in order to produce an effect is an important aspect of designing interventions. Rotheram-Borus and colleagues investigated this issue, again within a sample of 151 runaway youths in New York City (53% African American).(75) The intervention provided the adolescents with counseling, information, and skills training. There were 3 groups, which tested for differences in efficacy stemming from differences in the number of sessions, rather than the actual amount of time for the intervention. The first group consisted of 7 sessions of 90 minutes each (10.5 hours total), the second group consisted of three sessions of 3.5 hours each (10.5 hours total), and the third group was a no-intervention control. With regard to behavior change, at a 3-month follow-up, youths who had participated in the 7-session intervention had significantly fewer partners than those who attended 3 sessions. In addition, intervention participants reported engaging in fewer sexual risk acts. Finally, African-American participants reported engaging in fewer risky acts than other ethnic groups. With regard to the question of the effect of the number of sessions, the authors reported that there were no differences between the control group and the 3-session-intervention participants on any outcome variable of interest. Thus, it was the participants in the 7-session group that exhibited the highest degree of behavior change. Again, the actual amount of time spent was the same (10.5 hours); therefore, the number of sessions appears to have had a significant impact and better facilitated positive behavior change in this sample. This is an important finding, as it speaks to feasibility of interventions. Shorter interventions would be easier to implement in community settings. However, it may be that interventions need to be administered over longer periods of time in order for the mechanisms of behavior change to fully manifest. Though this may be more expensive, the money would be better spent if the chances of positive change are better. Money spent on shorter interventions that do not produce any effect might not be a wise use of funds.
Few interventions have focused on younger adolescents (eg, middle school students), and fewer still have followed those students in the crucial time period during which they become sexually active. Levy and colleagues conducted such a study, in which they examined a subset of a larger study of students who became sexually active between 7th and 8th grade in Chicago, Illinois.(76) The sample of 312 students was 63% African American. This randomized controlled trial utilized the school as the unit of randomization, again limiting the possibility of contamination between intervention groups. There were three groups: a parent interactive treatment group, a parent non-interactive treatment group, and a delayed treatment control. Each group consisted of 5 school districts. As there was very low participation in the parent-interactive group, the 2 intervention groups were combined into one for statistical analysis. The intervention consisted of 10 sessions over a 2-week period in 7th grade and 5 additional sessions over a week in 8th grade. The intervention attempted to increase self-efficacy regarding protective behavior and included information and skills training. At follow-up in the 8th grade, the intervention was found to have positively influenced the extent of condom use with foam in the past year and in the past month. It also positively influenced students' intentions to use condoms and foam in the next 12 months and to have sex less often. There were no differences between groups in the use of condoms alone. African Americans were the referent group for the analysis, and no significant results were found for ethnicity. Although this study has some limitations in terms of its high attrition (30%) and the intervention did not seem to have a great impact on actual condom use (without foam), the study still focused on an important group, younger adolescents who recently became sexually active. The study demonstrated that this age group could be positively influenced by an intervention aimed towards positively impacting their sexual behavior.
Ashworth and colleagues reported on the results of a study that examined how a brief intervention could influence adolescents' HIV risk behavior and knowledge.(77) The sample was comprised of 1,194 male and female high school students (65% African American) in Augusta, Georgia. The unit of randomization was the school, thereby lessening the probability of contamination between students. There were two groups-the intervention group received one hour of AIDS education, and they were compared to a no-treatment control. At a 2-week follow-up, the intervention group reported greater AIDS knowledge; however, there was no significant effect on drug use or behavior change. The authors reported that they were restricted by the school districts to not explicitly inquire about condom use, which limits the ability to interpret the effects of this intervention. African-American students expressed greater worry that they had been exposed to HIV. Although the study was well-designed, the short follow-up period, combined with the brief and limited nature of the intervention, severely limits strong conclusions and/or recommendations being made from this study.
Workman and colleagues conducted a study of 60 female adolescents who were attending an inner-city all-female parochial high school.(78) The intervention was conducted only with African-American and Latina students. The intervention group received 12 weekly 30-minute sessions on HIV prevention and the control group received 12 weekly group seminars on "womanhood development" (which did not emphasize sexual risk reduction). The intervention was based in cognitive-behavioral theory and included information, modeling and skills training. One week following the intervention, the students who received the HIV prevention intervention demonstrated greater knowledge. There were some differences due to ethnicity including an increase in sexual assertiveness for African Americans but not for Latinas, and significantly greater reported levels of comfort discussing AIDS preventive behaviors among African Americans than Latinas. The authors posit that cultural differences between these two groups could account for these differences. There were no significant changes between the groups with regard to AIDS preventive behaviors; however, the baseline levels across groups were already quite high, and a high percentage (68%) of the sample was not yet sexually active. Although this study was well-designed, and the intervention was intensive, the follow-up period was only one week following the intervention, which did not allow sufficient time to observe long-standing behavior change. Other limitations included a relatively small sample size, and the possibility that there could have been contamination between the two intervention groups. However, its focus on African-American and Latina adolescents is important, and future studies should conduct similar analyses that allow for comparisons to be made between ethnic groups.
In sum, the above studies demonstrate that adolescents can effect positive change that reduces their chances of HIV infection. These results were consistent across school-based and runaway populations. Skills training was, again, found to be an important component of the interventions. The interventions were theoretically based, and produced positive results for fairly heterogeneous populations (eg, mixed race and mixed gender). Especially promising are results obtained for younger adolescents who are just becoming sexually active,(76) as this is a population that could greatly benefit from obtaining skills to negotiate and engage in safer sexual practices both prior to, and coinciding with, their onset of sexual activity.
| Studies With At Least 80% of the Sample Comprised of African Americans|
As the purpose of this review is to describe the best evidence for prevention in African Americans we are including, in addition to the categories described above, studies in which at least 80% of the sample is comprised of African Americans (without separate analysis of the African-American participants). We believe that these results can contribute to our understanding of interventions that have been conducted with African-American participants.
| Injection Drug Users|
| 'Good' Studies|
Cottler and colleagues investigated whether or not a peer-delivered enhanced intervention was superior to a standard intervention.(79) The intervention participants were 725 out-of-treatment drug abusers in St. Louis, Missouri (93% African American, 61% male). The intervention included HIV testing and counseling, information, peer education, referrals, skills training, and the provision of risk-reduction materials. The standard intervention was comprised of 2 sessions and included HIV testing. The enhanced intervention included the standard protocol with 4 additional peer-delivered sessions (peers were former drug users in recovery). At 3-month follow-up, the enhanced intervention participants had significantly decreased their crack cocaine use compared to the standard group, although both groups had improved. When examining the results by gender, it was found that this result was true for men, but not for women. With regard to sexual behavior, there were no differences between groups for number of partners although, again, both groups demonstrated improvement in this area. Contrary to expectation, condom use worsened in the enhanced-intervention group, and there were no group differences. The authors describe these results as being in accord with prior findings that posited that drug using behaviors are easier to change than sexual behaviors.(11,12,80) This study was well-designed, had very low attrition and demonstrated that this population could effect positive change in their high-risk drug use behavior. It also demonstrated that a peer-delivered intervention can produce significant positive change in injection drug users risk behavior, and the authors recommend that utilizing peers as sources of positive social influence is a both feasible and effective method of reaching drug users. The authors also emphasize that future interventions should be tailored for women, as women did not demonstrate nearly the degree of positive change that the men did.
Probationers and parolees have often reported high rates of HIV risk behavior).(81) Mandell and colleagues investigated the effects of an HIV prevention intervention in a sample of 105 probationers and parolees (for drug-related charges) in Baltimore, Maryland (90% African American, 86% male).(82) In addition to their prior risk behavior, the authors surmised that recruiting from this population would reduce the likelihood of contamination in the intervention as they were instructed by their parole/probation officer not to associate with other injection drug users. The investigators sought to compare a standard intervention to an enhanced intervention with regard to HIV risk behavior. The standard intervention consisted of a 15-minute counseling session that included information regarding preventive sexual and drug-use behavior. All participants received this, and gave blood for an HIV test. When the participants returned, all seronegative participants who agreed to participate were randomized into groups. Those participants randomized into the enhanced group then received a theoretically based counseling session that was one hour in length. It was based in relapse prevention theory. At 6-month follow-up the participants in both groups had decreased their injection drug use (from 42% reporting daily injections to 17%), decreased their visits to shooting galleries, and increased their needle cleaning, but there were no significant differences between the groups. In addition, the intervention had no effect on cognitive mediators of risk (eg, perceived risk, intentions). Although this study was well designed, it is possible that the small sample size and the relative brevity of the enhanced intervention prohibited large effects from being found.
| 'Fair' Studies|
Andersen and colleagues conducted an intervention with a sample of 725 injection drug users in Detroit, Michigan (93% African American, 72% male).(83) The investigators randomized high-drug-use communities to standard- and enhanced-intervention communities. The standard intervention was derived from NIDA research protocols and consisted of 2 sessions, and HIV testing and counseling. The enhanced intervention included the standard intervention plus three additional sessions and attendance at a support group. It was based on the "LIGHT" model of nursing.(84) There was no specific information given on the length of the sessions. At 6-month follow-up the enhanced-intervention group demonstrated significantly more decreases in heroin and crack cocaine use, as well as in number of episodes of unprotected sex. In addition, those enhanced intervention participants who had participated in the support group had significantly more improvement with regard to crack cocaine use such that the more groups they attended the more improvement they showed. Although this intervention appeared to effect positive change, it is difficult to ascertain the specific components that had an effect. For example, the authors state that the support group was begun as a result of participants "dropping by for brief visits" to the project office. Thus, it is difficult to know how much this type of interaction influenced the participants' behavior and might be responsible for some of the positive change, above and beyond the effects of the intervention itself.
Lurigio and colleagues conducted a study with 99 probationers in Chicago, Illinois (86% African American, 90% male).(85) The study compared two HIV risk-reduction groups (an individual vs. a group presentation) against a heart disease education group (individual vs. group presentation). The HIV risk reduction group provided participants with information, referrals, and skills training. All interventions were provided in a single session, of unknown length. At 1-month follow-up participants in the HIV groups had significantly increased their knowledge and intentions to speak to others about bleach for syringes, but there were no significant differences between intervention groups (nor was there an effect of group size -- individual vs. group presentation). No other behavioral risk-reduction intentions were significant. When examining a scale comprised of several preventive behaviors together, there were no significant group differences. However, examining the behaviors separately, the HIV intervention group participants were significantly more likely to use condoms when compared to the heart disease group. The conclusions of this study are limited by high attrition (50%), but the authors argue that this is to be expected in this highly transient population. Nonetheless, the results also provide evidence of the equivalence of group presentations to individual sessions, and thus increase the feasibility of future interventions implementing group presentations.
Drug users who had entered treatment were the focus of a study conducted by Malow and colleagues.(86) They conducted an intervention with 127 inpatients (85% African American) in an inpatient unit at the Veterans Affairs Medical Center in New Orleans, Louisiana. The "psycho-education" group provided information regarding risk management and included modeling, goal setting, contracting for safer behavior, and other cognitive strategies to reduce risk. The "information" group provided the participants with comparable information, given through audiovisual and printed material with limited interpersonal interaction. (However, it is not clear whether the psycho-education group was presented in a group, or what percentage of time it was delivered by a facilitator). Both interventions were conducted in a 6-hour time frame over 3 days. At 1-month follow-up both groups had increased knowledge, had improved needle cleaning and condom use skills, and had improved their perceived susceptibility and response efficacy for their risk of HIV. The psycho-education group had significantly higher scores on condom demonstration test at post-test (immediately following the intervention), but this did not persist at 1-month follow-up. This study is limited in several ways. The follow-up time was very short, and the interventions were not equivalent in their delivery. In addition, the issue of contamination was not addressed in that it is not known what overlap there was between patients in different groups on the inpatient unit. If patients from two different intervention groups overlapped on the unit, it is possible that certain aspects of the intervention could have been shared between the patients. Insufficient information is provided to ascertain whether steps were taken to limit this possibility.
| 'Limited' Studies|
Kotranski and colleagues conducted a cognitive behavioral-based intervention in Philadelphia, Pennsylvania with a sample of 684 participants who had a history of drug abuse (85% African American, 63% male).(87) The standard intervention included HIV testing and counseling as well as information and individual counseling on risk reduction. The enhanced intervention afforded participants an opportunity to conduct a self-assessment with regard to their risk, and it emphasized STD prevention, additional information, and strategies to improve the risk from drug use and sexual practices (and included the standard intervention). At a 6-month follow-up, contrary to expectations, the standard intervention participants reported less unsafe vaginal sex and less use of cocaine, heroin, or speedballs during sex than the participants from the enhanced intervention. There were no other significant group differences. However, across groups participants were more likely to report safer behaviors with regard to both sexual and drug-use behaviors. The authors posited that perhaps their enhanced intervention was not sufficiently longer than the standard intervention to effect significant levels of change. An additional possibility is that because the standard intervention was delivered individually, the participants experienced a high degree of influence from this alone, and the additional sessions simply reinforced the original prevention messages they had already received. Limitations of this study include a randomization procedure that was accomplished by alternation rather than by methods that allow for true randomization and a fairly high attrition (30%).
Although most studies take place in urban communities, McCoy and colleagues(88) conducted an intervention in a small agricultural community in Florida (Belle Glade) that, at the time of the study, had the highest cumulative incidence rates of AIDS in the United States.(89) The sample consisted of 237 people recruited from neighborhoods with a high incidence of HIV infection (92% African American, 61% male). They tested the effect of a standard intervention that consisted of a 1-hour group counseling session with the standard intervention plus 2 intensive group counseling and skills-training sessions. At a 6-month follow-up, there were improvements across the groups in both sexual behavior and drug use, but there were no significant differences between the groups. The results were based on preliminary analyses, and the attrition for data collection at that point was 50%. In addition, the authors point out that there were different types of counselors (eg, HIV+ former injection drug users and professionally trained staff) and that these differences were not taken into account. The authors also posit that contamination could have occurred due to the small size of the community. Thus, conclusions that can be made from this study are quite limited.
In sum, we have identified 7 studies meeting our inclusion criteria in injection drug users that included a high percentage of African-American participants. The results demonstrate that injection drug users can successfully modify their high-risk behavior following interventions. There was more evidence of change in drug risk behavior than for sexual risk behavior. The most successful interventions were comprised of multiple sessions, were theoretically based, and included sufficient follow-up time to detect change. Future interventions should investigate how to tailor components to address sexual risk behavior as well as drug use behavior. Women were also underrepresented in the samples.
| 'Good' Studies|
Gollub and colleagues investigated the effects of different types of counseling aimed toward reducing sexually transmitted disease in a sample of 1,591 women in Philadelphia, Pennsylvania (91% African American).(90) The three groups provided a "hierarchical message" that included information on male condoms, female condoms, diaphragms, cervical caps and spermicides; a single message regarding female condoms; or a single message regarding male condoms. The primary outcome was later diagnosis of a sexually transmitted disease. The intervention was given in a single session. There were no differences between the groups in rates of reinfection. The authors reported on unpublished data regarding behavioral outcomes in which a subsample of 1,591 women were followed regarding their behavioral risk practices, and significant positive changes did occur (a doubling of rates of condom use and a reduction by two-thirds in the proportion of unprotected acts). Though behavioral changes have not always been linked to actual changes in disease rates, they still are a necessary component of prevention. The intervention did not have a "usual care" group, so it is also difficult to ascertain the degree of true degree of change of these particular interventions. The authors posit that the intervention may have been insufficient in length (1 session) to effect significant change, as prior successful interventions were multiple sessions (eg, Kamb et al(91)). Nonetheless, the study was well designed, and its focus on a tailored and multi-level approach to prevention of sexually transmitted disease (including HIV) in high-risk women is commendable.
O'Leary and colleagues conducted a multi-session intervention with a sample of 659 patients recruited from public sexually transmitted disease clinics in Georgia, Maryland, and New Jersey (91% African American, 59% male).(92) They compared a "usual care" intervention with a risk-reduction intervention comprised of seven 90-minute modules that incorporated information and skills training, and were based in cognitive behavioral theory.(42) At a 3-month follow-up, no significant differences were found between the groups with regard to sexual behavior outcomes. When both groups were examined together all participants were found to have reported fewer partners, fewer risky acts, a higher proportion of condom use, increased self-efficacy for risk behavior when compared to baseline rates. The lack of significant outcomes was contrary to expectation, and the authors offered several possible explanations. The participants who received the "usual care" intervention also participated in a 90-minute baseline interview (as opposed to just filling out a survey) regarding their risk behavior and this exposure may have prompted some examination of their behavior and influenced subsequent change. In addition, there was a high refusal rate from the patients they approached, and thus only those patients who kept their initial baseline interview appointments were randomized, which may have produced a somewhat biased sample. The authors recommend further study of brief interventions, which may produce sufficient motivation to change without a commitment to a long intervention.
Nyamathi and her colleagues tested a brief intervention in a sample of 916 women recruited from homeless shelters and drug recovery programs (81% African American) in Los Angeles, California.(93) The "traditional" intervention was delivered in a 1-hour session and included HIV testing and counseling and HIV/AIDS education presented in a culturally sensitive manner. The "specialized" intervention incorporated the above components as well as an additional focus on tailoring information for the women's individual needs. The intervention was theoretically based on Nyamathi's comprehensive health seeking and coping paradigm.(53) At a 2-week follow-up, women in both groups had several positive outcomes including improved knowledge, decreased number of partners, and decreased injection and non-injection drug use. However, women in the traditional group had higher knowledge and fewer partners than the specialized group. Although this finding is contrary to expectation, the extremely brief follow-up period prohibits any strong conclusions being made from this result, and limit the quality of this otherwise well-designed and implemented study.
| 'Fair' Studies|
Kelly and colleagues conducted a well-designed intervention with 187 women recruited from an urban primary health care clinic in Milwaukee, Wisconsin (87% African American).(94) The HIV/AIDS intervention was grounded in cognitive-behavioral theory(46,35) and was delivered in four weekly 90-minute group sessions with a follow-up session one month later. The comparison group received three 90-minute group sessions on topics relevant to low-income women. At the 3-month follow-up, the women in the intervention group had reduced frequency of unprotected sex, increased condom use (doubled from baseline to follow-up), and used condoms with a larger percentage of their male partners than the women in the comparison group. There was no change in their substance use. With regard to cognitive outcomes, the intervention participants exhibited higher AIDS knowledge and better accuracy of personal estimation of risk. In addition, the intervention participants were rated higher by AIDS educators in their communication skills (eg,, requesting to postpone sex if a condom was not available). This study demonstrated extremely positive results, and its quality rating was only compromised by a high attrition rate. This study was also included in the "HAPPA" list of effective and well-designed prevention interventions.(33) However, there were no significant differences between the groups due to attrition, and, given the highly transient population, high attrition rates are not surprising. The results support interventions of moderate length, and show that significant changes in the behavior of high-risk women can occur following a theoretically-based, well-implemented intervention.
A similar approach to that used by Kelly et al(94) was utilized by Carey and colleagues with a sample of 102 women recruited from a community based organization known for its services to people of color.(95) The intervention was a replication of an earlier study(96) by Carey et al (not included here due to the percentage of African-American women being under 80%). Components of the intervention were derived from the information-motivation-behavioral skills model(46) and also incorporated aspects of motivational interviewing.(97) An HIV risk reduction intervention was compared to an intervention focused on stress, anger, nutrition, and breast health. Both groups were conducted during four 90-minute sessions over the course of 2 weeks. At a 3-month follow-up, women in the intervention group reported increased knowledge and increased intentions for safer sex. With regard to behavioral outcomes, there were interactions between intervention group and certain mediating variables. For example, the women in the HIV intervention who reported imperfect intentions regarding condom use significantly increased their condom use. In addition, when separately examining the women who talked about condom use and HIV testing with their partners (across groups), women in the HIV risk-reduction group did so on more frequently and were significantly more likely to refuse sex without a condom. The authors recommended that future research investigate the influence of cognitive factors such as intentions and that interventions that take the participants' intentions into account may be more successful. This was a well-designed and implemented study that, like the Kelly et al study described above,(94) suggests that interventions of moderate length and theoretically driven can produce meaningful change in the sexual behavior of high risk women. Its quality was only impacted by a fair amount of attrition.
Early in the epidemic, many studies were conducted to investigate whether having an HIV test influenced people's subsequent risk behavior. Wenger and colleagues conducted this type of study with a sample of 256 patients recruited from a sexually transmitted disease clinic in Los Angeles, California (88% African American, 67% male).(98) The standard group received HIV/AIDS education, while the intervention group received standard care in addition to undergoing HIV testing and counseling. At a 2-month follow-up, the intervention group participants were more likely to have asked partners about AIDS risk factors and had a decrease in unprotected vaginal and anal intercourse. There were no group differences in knowledge, AIDS worry, or mental health status. Both groups demonstrated a decrease in numbers of partners. As this study was conducted quite early in the epidemic (data were collected in 1988), it is difficult to know whether such interventions would continue to have a significant influence today. (At the time this study was conducted, HIV testing was not widely available). This study was also limited by a high attrition rate and a relatively short follow-up period.
Branson and colleagues sought to investigate differential impact of the length of interventions on subsequent risk behavior in a sample of 996 patients (90% African American, 57% male) recruited from a sexually transmitted disease clinic in Houston, Texas.(99) The standard intervention was comprised of two 20-minute sessions of client-centered HIV prevention counseling whereas the enhanced-group participants received 4 group sessions within 2 weeks followed by a booster group session 2 months after the intervention. The intervention's components were derived from the information-motivation-behavioral skills model.(46) At a 1-year follow-up both groups had decreased their risk behavior and had reductions in diagnosis of sexually transmitted disease, but there were no significant group differences. The authors concluded that two brief sessions were as effective as the more intensive intervention. However, they also point out that only half of the persons assigned to the enhanced intervention actually participated in either 4 or 5 of the group sessions and that attrition was also high for the comparison group. Although their primary outcome was biologically based (reinfection of sexually transmitted disease), it is possible that new infections were not diagnosed at follow-up visits due to an asymptomatic presentation. The lengthy follow-up period is a strength, as well as the examination of a biological outcome, but limitations such as attrition should prohibit strong recommendations regarding the length of effective interventions from this study alone. Several studies documented the efficacy of interventions of approximately 4 sessions as being successful in changing high-risk behavior (eg, Carey et al(95); Kelly et al(94)). However, there have not been many studies that clearly document a reduction in sexually transmitted disease diagnosis following documented reductions in high-risk behavior. Future studies should investigate this relationship.
| 'Limited' Studies|
Ashworth and colleagues investigated the differential effects of the method of delivery of information in a sample of 217 women (95% African American) recruited from WIC programs (federal women, infants, and children nutrition programs) in Augusta, Georgia.(100) A control group received a health department pamphlet about AIDS and could ask questions of staff. The two intervention groups received the same information regarding HIV risk prevention, but differed in their mode of delivery. One group received the information via videotape, while another received it personally from a nurse educator. The interventions were very brief (15 minutes). At a 2-month follow up, the 2 intervention groups had higher knowledge and less worry about dying if they got AIDS. There were no differences between the intervention groups (video vs. nurse educator). There were no significant effects on behavioral intentions. This study was limited by no behavioral outcomes (it was included due to its examination of intentions), and the authors provided no information regarding attrition. Therefore, any conclusions that can be drawn are severely limited.
Another study investigating the effects of an intervention on subsequent reinfection of sexually transmitted disease was conducted by Cohen and colleagues with a sample of 551 patients (93% African American, 71% male) recruited from a clinic in Los Angeles, California.(73) The intervention was delivered in the waiting room of the clinic, and was delivered on alternate times of day (morning vs. afternoon). The authors describe this procedure as randomization, but this method does not ensure that randomization has occurred. In addition, the delivery of the intervention in the waiting room resulted in the intervention not being uniform in exposure, as the authors describe that some patients were called to their appointments without hearing the entire presentation and that patients who did not agree to participate were exposed to the intervention (though they were not used as comparison subjects). Records were examined 7 to 9 months following the intervention, and men in the intervention group were found to have significantly lower rates of reinfection. However, given the limitations described above, as well as high attrition, the conclusions from this study are severely limited. Cohen and colleagues conducted a similar intervention that improved upon these limitations described earlier in this review.(70)
In sum, we reviewed 9 studies that investigated the effects of interventions on heterosexual risk behaviors. The most successful studies utilized interventions of moderate length (e.g, 4 sessions) and were theoretically based (most often in cognitive behavioral theory). Studies that utilized a biological outcome often did not find differential effects in rates of re-infection as a result of more intensive interventions (eg, Branson et al(99)), probably the result of being underpowered due to small sample sizes. However, studies that focused on behavioral outcomes did produce significant changes from more intensive interventions (eg, Carey et al(95); Kelly et al(94)). Finally, the most positive results for women were found in studies that focused on women exclusively, which points toward the need to tailor interventions for women.
| 'Fair' Studies|
Visits to a physician have been assumed to provide a good opportunity for intervention with adolescents. However, few studies have conducted a structured intervention to test whether or not this is actually the case. Mansfield and colleagues investigated whether an intervention delivered by physicians could have a positive impact on the HIV risk behavior of their adolescent patients.(101) They recruited a sample of 90 adolescents who were patients at an urban children's hospital (83% African American, 90% female). All participants had been diagnosed with at least one sexually transmitted disease. The intervention compared the standard care (10 minute discussion with physician) with an intervention consisted of standard care plus HIV testing and an individualized risk assessment (20 minutes). Both were delivered individually to subjects. At a 2-month follow-up, participants from both groups had decreased their number of partners, increased condom use, and decreased sexual activity, but there were no significant differences between the groups. Possible reasons given by the authors for the lack of findings included some significant differences between the groups at baseline that, although they were taken into account statistically, might still have biased the results. In addition, the small sample size might not have provided adequate power to detect the effects. Finally, the randomization strategy was poor (even or odd number). Future studies with larger samples could provide more evidence regarding the efficacy of providing interventions through primary care providers.