Skip directly to search Skip directly to A to Z list Skip directly to page options Skip directly to site content


This online archive of the CDC Prevention Guidelines Database is being maintained for historical purposes, and has had no new entries since October 1998. To find more recent guidelines, please visit the following:

Community-Level Prevention of Human Immunodeficiency Virus Infection Among High-Risk Populations: The AIDS Community Demonstration Projects

MMWR 45(RR-6);1-16

Publication date: 05/10/1996

Table of Contents




Method for assigning stage of change for consistent condom use
Intervention exposure by city, intervention community, and phase



The AIDS Community Demonstration Projects (ACDPs) were community-level human immunodeficiency virus-prevention programs targeting high-risk populations in five U.S. cities. For the intervention design, researchers developed a common study protocol based on behavior-change theories and models. This report describes the common study protocol used in the ACDPs, the preliminary findings, and the conclusions regarding the design, implementation, and evaluation of a community-level intervention; specific case studies from each project site are also described.


The AIDS Community Demonstration Projects (ACDPs) began in 1989 as community-level human immunodeficiency virus (HIV)-intervention projects targeting high-risk, hard-to-reach groups in five U.S. cities: Dallas, Denver, Long Beach, New York City, and Seattle. All project sites, except Dallas, developed interventions for 1-3 of the following high-risk groups: street-recruited injecting-drug users (IDUs); female sex partners of male IDUs; women who trade sex for money or drugs (female sex traders); men who have sex with men (MSM) but do not self-identify as homosexual (nonhomosexually identifying MSM); and youth in high-risk situations (e.g., street youth who spend most nights away from home). The Dallas project developed interventions for persons residing in two separate census tracts that had high rates of both sexually transmitted diseases (STDs) and injecting-drug use.

The goals of this research were to determine the efficacy of a specific community-level intervention for groups considered hard to reach and, if effective, to apply the strategies used in the intervention to subsequent HIV prevention programs. The objectives of the projects were to a) increase the prevalence of consistent condom use among persons in targeted groups in all sites and b) increase the use of bleach to clean injection equipment * among IDUs at sites directing interventions to this population. To achieve these objectives, project researchers developed a common study protocol based on behavior-change theories and models that allowed for flexibility in creating interventions for specific at-risk populations and communities in each city (1).

This report describes a) the common study protocol used by the ACDPs; b) preliminary data regarding community exposure to the interventions and self-reported behavior changes among targeted populations; and c) conclusions from the design and implementation of a community-level, multisite intervention designed to facilitate behavior change. Case studies describing the implementation and adaptation of the protocol for a specific population or community in each city also are included (Appendix).


Researchers from the selected project sites and CDC, as well as expert consultants, collaborated to design a common study protocol for community-level HIV prevention interventions. The protocol was based on previously designed and implemented methods (2-4). The key components of the protocol included a) use of behavior-change models and theories to design the intervention; b) formative research within the project communities before implementing the intervention; c) development of print materials (e.g., pamphlets, brochures, flyers) containing stories of persons in the targeted populations who had changed their HIV-risk behaviors (i.e., role-model stories); d) distribution of these materials with condoms and bleach kits ** by community networks; and e) use of an evaluation protocol that measured both implementation and outcome.

This protocol was systematically implemented and adapted to specific populations in each city. Implementation of the protocol varied across cities because of the unique characteristics and needs of the at-risk populations and the availability of local resources (5). The implementation and adaptation process for the ACDPs included a) defining the target population or community; b) using data from formative research to determine community characteristics, project and community resources, and community preferences for the print materials; c) recruiting community members to deliver these materials and build community support for the intervention; and d) using local data collected during the study period to develop role-model stories and intervention materials.
Theoretical Foundation of the Intervention

The design of the ACDPs' intervention was based on published behavioral research and incorporated elements of the following behavioral theories and models: Health Belief Model (6), Theory of Reasoned Action (7), Social Cognitive Theory (8), and Stages of Change (SOC) continuum of the Transtheoretical Model (9-11). These theories and models indicate that several factors can influence a person's intentions and behaviors. Such factors include the person's a) perception of susceptibility to a given disease or illness; b) attitude toward performing the behavior; c) normative beliefs, including the perception that others in the community are also changing behaviors; d) self-efficacy, or belief that one can perform the recommended behavior under various circumstances; e) acquisition of the social and physical skills necessary to perform the behavior; and f) readiness to change behavior (12). The project interventions also encouraged behavior change by focusing the intervention messages on one or more of these underlying cognitive factors that the empirical data indicated were important in each community. The relative importance of these factors as determinants of intention and behavior was expected to vary across behaviors and populations (National Institute of Mental Health, unpublished report, 1991). Because environmental constraints can also influence attempts at behavior change, the projects sought to create an environment that facilitated the use of condoms and bleach kits by making these items readily available to persons at risk.

The SOC continuum provides a framework for understanding a person's readiness to change behavior. This continuum describes a series of steps or stages on a behavior-change continuum: a) precontemplation; b) contemplation; c) ready-for-action; d) action; and e) maintenance. At the first stage (precontemplation), an at-risk person may have no intention of changing the high-risk behavior or of adopting a given risk-reduction behavior. Any one of several events may then lead the person to form intentions to adopt the behavior in the distant future (contemplation). These intentions may be followed by an intention to adopt the behavior in the immediate or foreseeable future and may be accompanied by initial, perhaps exploratory, attempts to adopt the behavior (ready-for-action). The new behavior is then adopted (action) and ultimately becomes a routine part of the person's life (maintenance). Effective movement through these stages is assumed to be sequential, although persons may relapse at any stage and cycle back through the stages repeatedly before achieving long-term maintenance (9-11).

Based on these assumptions, an intervention designed to help persons change their behavior should first determine their stage on the SOC continuum and then help them move to subsequent stages. By identifying such immediate objectives for adopting a specific risk-reduction behavior among the at-risk populations in each city, researchers were able to target the interventions to the needs of persons in these specific communities. For example, persons who had not even thought about adopting a risk-reduction behavior received different messages to encourage behavior change than those trying to adopt that behavior.

Earlier research in the study communities had indicated that the behavioral goals of "condom use" and "cleaning of injection equipment" were too broad. For precise measurement, the behavioral goals of the intervention were defined as a) consistent condom use in four situations (vaginal intercourse with a main partner, vaginal intercourse with nonmain partners, anal intercourse with a main partner, and anal intercourse with nonmain partners) and b) consistent use of bleach to clean injection equipment. Consistent condom use was considered necessary to protect oneself from STDs, including HIV infection (13-14). In addition, condom use in each of these situations was considered a distinct behavior because the cognitive factors influencing condom use can vary by situation (e.g., intercourse with a main partner versus intercourse with a nonmain partner). For cleaning injection equipment, consistent and correct use of bleach was considered necessary for persons sharing injection equipment. Behavioral outcomes for each individual were measured along the SOC continuum from the precontemplation stage to maintenance stage (Table 1). Progress along these behavioral-change stages was considered success, although the consistent practice of risk-reduction behaviors (i.e., the action and maintenance stages) is necessary to prevent transmission of HIV.

Formative Research

By implementing interventions that facilitate behavior change directly in their communities, project investigators attempted to reach at-risk persons who might not use facility-based programs. The projects defined the specific communities to receive the interventions based on local needs. The Dallas project defined community in terms of geography and developed a community-level intervention in two separate census tracts that had high rates of STDs. The other four projects defined community in terms of specific populations at risk (e.g., IDUs or female sex traders) and developed interventions directed to those populations.

Before implementing the intervention, researchers in each project site developed a thorough knowledge base about their targeted populations that included a) the geographic areas in each city where at-risk populations congregated and where the intervention and evaluation could be conducted; b) the specific subpopulations within each at-risk population; c) the existing risk behaviors; and d) barriers (e.g., dislike of condoms, desire to become pregnant) and facilitators (e.g., belief that condoms protect against acquired immunodeficiency syndrome {AIDS}) to risk reduction among persons in the risk populations.

To collect this information, project staff in each city conducted 6 months of formative, ethnographic research (15-16). This research included interviews with health department professionals and AIDS researchers to determine their knowledge of the risk populations as well as interviews with other professionals who had contact with the targeted populations, including staff at drug-treatment and mental-health facilities, family-care services, police departments, and community-based organizations.

Using a standard protocol that was adapted to the local needs of the sites, project staff also conducted qualitative, semistructured interviews with persons identified by health department and other professionals as potential "gatekeepers" in the community. These persons often served as the link between the at-risk populations and the larger community and could be either members or nonmembers of these populations. In addition to providing additional information about the targeted populations from the perspective of persons closer to the community, the gatekeepers helped to build support for the project among community members and to facilitate access to target population members.

Project staff also initiated field observations in areas where target population members lived or gathered. These observations documented the demographics, activity patterns, and physical environment of the target population. The observations were useful in supplementing information from the individual interviews with professionals and gatekeepers, mapping potential intervention sites, establishing the visibility of project staff, and creating contacts with community members.

Through the gatekeeper interviews and field observations, project staff were able to identify persons in the at-risk populations and recruit them to participate in interviews and focus groups concerning the design and implementation of the intervention. Data obtained from these interviews and focus groups were integrated with information collected in earlier phases of the formative evaluation to provide a broader perspective of the life circumstances and HIV risk of persons in these communities.

These data also were used locally to adapt the common study protocol. Project staff in each city used the data to develop intervention materials that more appropriately addressed the attitudes, norms, barriers, and facilitators regarding risk reduction among persons in the targeted populations. Through gatekeepers and members of the target populations, project staff recruited peers and other community members to deliver and build further community support for the intervention. Data from the formative research were also used to develop evaluation instruments.

Print Materials

The intervention materials were designed to facilitate behavior change; they included brochures, pamphlets, flyers, and baseball trading cards containing role-model stories (17). These stories presented first-hand accounts of persons within the target communities who had changed their behaviors. Each story described the person's motivation for initiating or considering a behavior change, the type of change initiated, how barriers to change were overcome, and the reinforcing consequences of the change. Through the role-model stories, an at-risk person could learn to perform or begin thinking about performing the new behavior and receive positive reinforcement (8).

The role-model stories were developed from interviews with members of the target populations and communities; new stories were produced approximately once a month. Ongoing feedback from collected data guided the selection and development of new role-model stories. Stories were written to reflect the behavioral stages beyond the most common or median SOC among persons in the community. The stories highlighted the specific cognitive factors that correlated most strongly with movement to that next behavioral stage, based on the current data from that community.

In addition to the role-model stories, the written materials contained basic AIDS information, instructions on the use of condoms or bleach to clean injection equipment, biographies of community members participating in the project, notices of community events, and information on other health and social services (e.g., locations of homeless shelters or needle-exchange settings, schedules for free meals, mammogram screening, or drug- and alcohol-treatment services). By including other relevant community concerns identified from the formative and ongoing qualitative research, the investigators hoped to promote community interest in and support of the program.

Local data also were used to design the print materials that appealed to the targeted populations and reflected their culture and language. For example, in Seattle, the materials produced for nonhomosexually identifying MSM were folded so that men could discreetly store the materials in a shirt pocket. The Long Beach project developed a "Road Dogs" flyer for IDUs; the term "road dog" referred to persons, generally drug users, who associated with each other on the streets. Several of the projects recruited local persons to serve as models for photos accompanying the stories.

Distribution of Materials by Community Network Members

The print materials, along with condoms and bleach kits, were distributed to persons in the at-risk populations by networks of community members. These networks were created by the projects and comprised members or peers of the target populations and "interactors," defined as merchants, community leaders, or other nonpeers who interacted regularly with the target population. Some materials also were distributed by project staff.

Network members were recruited through contact with a project outreach worker, referral from a service organization, or referral from current or former network members (18-20). Training sessions for the network members were held at each site and included the following elements: a) an introduction to the project; b) basic education about HIV infection and AIDS; c) an explanation of the role-model stories; d) a discussion of methods of street approach, nonthreatening conversation, positive reinforcement, and responding to persons who refuse materials; and e) role-playing interactions between the network members and the recipients of the materials.

During distribution of these materials, network members emphasized the role-model stories to the recipients and reinforced the social acceptability of behavior change. Peer network members who had tried to reduce their high-risk behavior were encouraged to share their personal stories and experiences with other community members.

In the second year of the intervention, the average number of peers distributing materials per month for each specific intervention varied across cities from 4 to 85. In many sites where peer networks were small, interactors were instrumental in distributing materials to the at-risk populations. Also during this period, the average number of print materials distributed per month for each specific intervention varied across cities from 800 to 6,350 (Appendix). These differences were caused by several factors, including: a) size of the distribution networks; b) number of populations at which the materials were directed; and c) estimated size of the at-risk population in each city. In addition to the print materials, thousands of condoms and bleach kits were distributed to provide the targeted populations with easy access to materials that would enable them to practice safer behaviors.

Several strategies were used to help retain network members. These included a) offering material incentives (e.g., small amounts of cash, food or movie coupons, and T-shirts or buttons with the project logo); b) providing recognition of the achievements of the network members through awards or certificates; and c) maintaining frequent contact between outreach workers and network members for encouragement and reinforcement of their roles (18,20).

Three of the project sites (Dallas, Denver, and New York) maintained storefronts either within or near the intervention neighborhoods. These served as focal points for project activities (e.g., meetings to assemble intervention materials, support-group meetings for peer network members or for HIV-infected persons, health-screening programs, and community events). Project efforts provided the community with intensive prevention messages by maximizing the size and effectiveness of the peer network, the cooperation of community businesses, and the appeal of the print materials.

Evaluation Protocol

Process measures.

To determine if the intervention was reaching persons in the targeted groups, project sites used the following process measures: a) records of monthly averages of the numbers of network members and distributed print materials; b) interviews with key observers and persons in the target populations regarding changes in the community; c) records of daily outreach activities; and d) in the Dallas project, unobtrusive observations of discarded condoms, bleach bottles, or print materials.

Outcome evaluation design.

The common outcome evaluation design was linked to the underlying behavioral theory of the intervention and involved three basic elements: a) each intervention was implemented in one geographic area, while a paired area served as a comparison; b) the data-collection schedule included multiple waves of cross-sectional data collection in both areas before and after the intervention began; and c) the data-collection instrument measured a common set of behavioral and cognitive variables across all communities and survey periods. In Dallas, intervention and comparison status was assigned randomly in each of the two pairs of communities; in the other sites, status was assigned based on practical considerations (e.g., the availability of a storefront and geographic proximity). Because assignment to intervention or comparison status was random only in Dallas, the outcome evaluation design was considered a quasi-experimental design (21).

Information from the formative evaluation process was used to identify areas that were comparable in terms of environmental conditions and the characteristics linked to the risk behaviors under study. For eight of the 10 intervention communities, a comparison community was identified in a nearby geographic area. For nonhomosexually identifying MSM in Seattle, a community in Long Beach was selected as the comparison area. Because population shifts disrupted the community that was originally selected to serve as a comparison area for the Denver intervention targeting IDUs, data from comparison areas used for the Long Beach intervention targeting IDUs were duplicated for use in the Denver project.

Data collection instrument.

To collect data from the study populations, the project sites developed a street interview instrument *** containing questions linked to the behavioral theories underlying the intervention. At the beginning of the interview, persons were screened to determine if they were eligible for the complete interview. Respondents were eligible if they a) reported having had sexual contact in the past 30 days or having shared injection equipment in the past 60 days and b) met specific criteria regarding the particular risk population (e.g., IDUs, female sex traders) targeted for the local intervention.

The interview questions assessed the person's stage on the SOC continuum with respect to consistent use of condoms during sexual contact and consistent use of bleach when sharing injection equipment and the theoretical factors underlying these risk-reduction behaviors. Specific sexual or drug-using behaviors were defined as a) vaginal intercourse with a main partner; b) vaginal intercourse with nonmain partners; c) anal intercourse with a main partner; d) anal intercourse with nonmain partners; and e) for IDUs, sharing injection equipment. Interview questions also assessed the person's exposure to HIV/AIDS information and to the project intervention as well as other indicators of risk-reduction behaviors (e.g., carrying a condom or having been tested for HIV). No personally identifying information was collected. Because the number of persons reporting having had anal intercourse with a main partner was too small to yield statistically meaningful results, these data are not presented in this report.
Data collection schedule.

Following a pilot period, periodic cross-sectional surveys were performed in 10 waves of data collection from February 1991 through June 1994. The timing of these waves was generally consistent across sites. In nine of the 10 pairs of communities, two waves of baseline data were collected from February through June 1991, before the intervention was implemented. Eight subsequent waves of data were collected during the development and implementation of the intervention. Each wave lasted 3-5 months and was followed by a 4-6-week period during which no data were collected. In the second pair of communities in Dallas, data collection began in August 1991 but thereafter was consistent with the other sites (Appendix).

A minimum number of respondents (sampling quota) was determined for one or more of the risk behaviors for each targeted community (e.g., vaginal intercourse with a main partner for female sex partners of IDUs). Locations and times for interviewing were selected to reflect the best available information about the targeted population in each community. In most communities, persons to be interviewed were selected by tables of randomly generated numbers; when populations were sparse (e.g., nonhomosexually identifying MSM), only times and locations for interview were randomized and all eligible respondents were interviewed. Similar sampling methods have been used elsewhere with other communities considered hard to reach (22).

Respondents were given small amounts of cash or food coupons for participating. Interviewers were not involved with distributing the intervention materials, although they did know which areas were receiving the intervention.


Preliminary assessments regarding the outcome of the intervention efforts were based on a) the respondents' reports of exposure to project materials, network members, and staff, and b) changes in self-reported behaviors across time.

Exposure to Project Materials, Network Members, and Staff

Exposure to project materials, network members, and staff was measured by a series of questions at the end of the street interview. All respondents were asked to describe HIV/AIDS-related materials they had seen in the community as well as persons who had spoken to them about AIDS during the last 3 months. Interviewers recorded information on up to four occasions of exposure to materials and three occasions of interpersonal contact. Responses were coded into specific categories for exposure to various project materials, project staff, network members, storefronts, and nonproject sources. Respondents who reported any exposure to project materials or talking about AIDS with someone associated with the ACDPs were classified as exposed. Respondents reporting only nonproject sources or no sources were classified as nonexposed.

The 10 waves of data collection were summarized into four phases: baseline, start-up, early implementation, and late implementation. For seven pairs of communities, the baseline phase included two waves of data collected from February through June 1991. However, because data collection began later for the second pair of communities in Dallas, the baseline phase for this pair included only one wave of data collection from July through October 1991. In addition, because the two designated waves of baseline data collection in the Seattle comparison areas for both female sex traders and nonhomosexually identifying MSM yielded few responses, the baseline phase for these two pairs was extended by 3 months (one wave) through October 1991. Therefore, the end of the baseline phase for these two pairs of communities coincided with the schedule for the second pair of communities in Dallas. Thereafter, the phases for these three pairs are the same as for the other seven pairs.

The start-up phase ran from July 1991 (or November 1991 for the three pairs of communities previously mentioned) through May 1992 and included data from the first three (or two) waves after the intervention began. The early implementation phase ran from June 1992 through August 1993 and included the next three data-collection waves. The late implementation phase began in September 1993 and included the two final waves of data collection.

Two methods were used to eliminate duplicate interviews. During the baseline phase, interviews were included only from respondents who reported no previous interview. During subsequent phases, interview items were added to allow exclusion of previous interviews from the same community within the same phase that matched on sex, race/ethnicity, and location and date of birth.

Exposure rates among eligible respondents ranged from 1% to 18% during the start-up phase and from 22% to 68% during the early implementation phase (Table 2). Characteristics of the populations (e.g., size, interest in the intervention, migration patterns) varied among the 10 intervention communities and are likely reflected in the proportion of respondents who indicated that they had been reached by project efforts. As of early implementation, female sex traders were most likely among respondents to report exposure to project materials and staff, and nonhomosexually identifying MSM were the least likely to report exposure. This result appears consistent with the relative accessibility and geographic mobility of these two populations. Overall, however, these data indicate that in several populations, high rates of exposure to this type of intervention can be achieved within 2 years of implementation of the intervention.

Changes in Self-Reported Behaviors

The outcome objective of the ACDPs was to promote progress along the SOC continuum toward the goals of consistent use of condoms and bleach. The design of the ACDPs, along with the measurement of exposure in the intervention areas, enabled investigators first to compare this progress across time among intervention and comparison area respondents (intervention effect) and then to determine cross-sectionally whether intervention-area respondents who reported direct project exposure were at a more advanced stage on the SOC continuum than intervention-area respondents who did not (exposure effect).

General linear models were used both to evaluate the statistical significance of the intervention and exposure effects and to calculate adjusted mean SOC values for each phase (23). Analyses are presented briefly for four behaviors: consistent condom use for vaginal intercourse with a main partner, consistent condom use for vaginal intercourse with nonmain partners, consistent condom use for anal intercourse with nonmain partners, and consistent use of bleach by IDUs to clean injection equipment.

Behavioral outcomes for each respondent were measured along the SOC continuum; stages were coded from 1 (precontemplation) to 5 (maintenance) (Table 1). Independent variables used in these analyses included the following: a) time; b) interview area (intervention or comparison); and c) exposure to the project intervention. Analyses were stratified by city, type of community, sex, race or ethnicity, and age.

Consistent condom use for vaginal intercourse with a main partner.

During the first three phases of the study, 6,754 respondents reported vaginal intercourse with a main partner. More than half (n=3,884 {58%}) of the respondents were female, the largest subgroup of whom comprised sex partners of IDUs (n=1,541 {40%}). Among the 2,870 male respondents, approximately half were IDUs (n=1,382 {48%}). At baseline, the mean SOC value among intervention-area respondents (1.60) was similar to that among comparison-area respondents (1.57). These low values correspond to a high proportion of respondents in the precontemplation stage for this behavior (72% in both the intervention and comparison areas). Among intervention-area respondents who reported having had vaginal intercourse with a main partner in the past 30 days, 10.6% reported exposure to the project intervention during the start-up phase and 40.7% at the early implementation phase. As of early implementation, the mean SOC value among intervention-area respondents (2.02) was higher than that among comparison-area respondents (1.87). Although data indicated more progress toward consistent condom use in the intervention areas were more likely to consistently use condoms than in the comparison areas, the intervention effect was not significant. At the same time, however, the mean SOC value among intervention-area respondents who recalled exposure to project material (2.10) was significantly higher than that among persons in the intervention areas who did not recall exposure (1.96) (exposure effect, p less than 0.05).

Consistent condom use for vaginal intercourse with nonmain partners.

Vaginal intercourse with nonmain partners was reported by 5,532 respondents. Respondents were almost evenly divided between women (n=2,770 {50%}) and men (n=2,762 {50%}). Most male respondents were IDUs (n=1,410 {51%}), and most female respondents were sex traders (n=2,075 {75%}). At baseline, the mean SOC values were similar: 2.78 for intervention-area respondents and 2.74 for comparison-area respondents. Among intervention-area respondents who reported having had vaginal intercourse with nonmain partners during the previous 30 days, 9.0% reported project exposure during the start-up phase and 42.9% at early implementation. As of early implementation, the mean SOC value for this behavior had increased to 3.08 among intervention-area respondents, compared with 2.85 among comparison-area respondents, representing a statistically significant intervention effect for vaginal intercourse with nonmain partners (p less than 0.05). These results correspond to a 43.9% increase (from 23.0% to 33.1%) in the proportion of intervention-area respondents in the action or maintenance stages (always using condoms). In contrast, this proportion decreased slightly among respondents in the comparison areas. During the early implementation phase, the mean SOC value among intervention-area respondents reporting exposure to the project intervention (3.36) was higher than that for those reporting no exposure (2.87). A much higher proportion of exposed respondents (41.3%) than of nonexposed respondents (27.1%) reporting consistent condom use (action or maintenance stage) contributes to this highly significant exposure effect (p less than 0.001).

Consistent condom use for anal intercourse with nonmain partners.

Most (82%) of the 986 respondents who reported having had anal intercourse with nonmain partners were men (n=810), of whom 368 (45%) were nonhomosexually identifying MSM. Of the 176 female respondents, 72% were sex traders. Mean SOC values increased from 2.78 to 3.00 among comparison-area respondents but from 2.64 to 3.15 among intervention-area respondents. Although this increase was greater among respondents in the intervention areas than among those in the comparison areas, the difference did not reach statistical significance. However, at early implementation, the average SOC value of 3.78 among intervention-area respondents who recalled exposure to the project materials was significantly higher than the SOC value of 2.86 among the nonexposed group (exposure effect, p less than 0.001). In the intervention areas, 58% of respondents who were exposed to the project reported consistent condom use for anal intercourse compared with 27% of respondents who were not exposed to the intervention.

Consistent use of bleach to clean injection equipment.

Of the 3,441 IDUs responding to questions regarding consistent use of bleach to clean injection equipment, approximately two thirds (2,317) were male. At the early implementation phase, almost half (45.6%) of the intervention-area respondents recalled exposure to the intervention. The mean SOC value for this behavior decreased from 2.65 to 2.51 among comparison-area respondents, but increased from 2.94 to 3.12 among respondents in the intervention areas (intervention effect, p=0.002). As of early implementation, the mean SOC value among intervention-area respondents who recalled exposure to the project (3.33) was signficantly higher than that for the nonexposed group (2.95) (exposure effect, p=0.001).

Summary and Limitations of Findings

For each of these four behaviors, the mean SOC value among persons in the intervention areas who reported exposure to the project intervention was greater than the mean SOC value among those who did not report exposure. For two of the behaviors (consistent condom use for vaginal intercourse with nonmain partners and consistent use of bleach by IDUs to clean injection equipment), a statistically significant divergence across the first three phases of the study was observed between all intervention-area respondents (including those who were not exposed to the intervention) when contrasted with respondents in the comparison areas.

Although the above analyses include adjustments for demographics, city, and type of community, they represent only an overview of the results of the various projects. The intervention and exposure effects may vary among sites and target populations.

Estimates of the project effects for some behaviors may be conservative because all respondents were included in the above analyses, regardless of whether the reported behavior was targeted by the intervention for a specific population. For example, most intervention efforts directed toward nonhomosexually identifying MSM involved condom use for both vaginal intercourse with a main partner and anal intercourse with nonmain partners because these were the most frequently reported behaviors. However, men in this population who reported either vaginal intercourse with nonmain partners or sharing injection equipment were interviewed regarding condom or bleach use during these behaviors and were included in these analyses. Similarly, although intervention messages for most populations did not address consistent condom use for anal intercourse with nonmain partners, all respondents who reported having had anal intercourse with nonmain partners were included in the analysis.

Persons who are at a higher stage on the SOC continuum may be more likely to seek information on preventive behaviors (selective exposure) or to remember and report project exposure (selective recall); this possibility should be considered when evaluating the extent and influence of community-level interventions (24-25). However, the trend across time for each of the four behaviors was positive, even among nonexposed respondents in the intervention areas. Furthermore, this trend was greater than that observed among respondents in the comparison areas for three of the four behaviors. These results do not reveal any substantial influence of selective exposure or selective recall on the apparent exposure effect.

Several methodological issues should be considered when making inferences from this research. First, the cities and geographic areas within each city where this research was conducted were chosen based on access to persons in at-risk populations and applicants' ability to follow a common study protocol. The study samples of persons at each site were used only to make inferences about changes in the study communities from which they were drawn; the study communities were not assumed to be a representative sample of a larger set of communities or populations at risk for HIV infection. Therefore, findings from the ACDPs should be viewed as indicative of what can occur with this type of community-level behavioral intervention. Data-collection methods did not include probability sampling to achieve a representative sample of the at-risk populations in the study communities. However, devices such as random number lists and minimal visual screening by interviewers were used to reduce some selection bias.

Finally, the private nature of sexual activity and drug use requires that this study rely on self-reported data for the primary outcome measures. However, externally measurable outcomes (e.g., carrying condoms) also were assessed and showed similar results.


The preliminary findings indicate that many persons who are at high risk for either sexually transmitted or needle-borne infections can be favorably influenced toward consistent risk-reducing behaviors through a community-level intervention. However, to maximize the influence of the intervention, efforts should be made to expose as many persons as possible in the intervention communities to the specific prevention messages. For each behavior, persons in the intervention areas who were exposed to the project intervention were more likely to be at a more advanced point on the SOC continuum than those who were not exposed. However, exposure levels of at least 20% were not reached in the project sites until approximately 1 year after the start-up phase began. Therefore, sufficient time should be allowed to establish a minimum level of exposure to the intervention among community members. In addition, because behavior change takes time and often occurs in discrete steps, use of the SOC continuum to examine these steps allows for documentation of important changes that could be overlooked if behavior were assessed as a dichotomous variable only.

As HIV-prevention programs shift toward a model of comprehensive intervention strategies, the potential contribution of a community-level component should be considered. Based on experience with this study protocol (Appendix), the following suggestions should be considered when designing and implementing a community-level HIV-prevention intervention targeted toward at-risk populations traditionally considered hard to reach:

  • Before implementing an intervention, health departments and other agencies should develop a thorough understanding of the target population. For the ACDPs, a health department was the lead agency in implementing the intervention in three of the five cities. Initially, project staff had little or no experience in conducting a community-level street intervention. The project sites needed to conduct formative research on the targeted risk groups as well as their communities. Formative research in the communities participating in the ACDPs was useful to identify the following: a) the behaviors that contributed to risk for HIV infection; b) the geographic areas that could serve as access points for the intervention and for evaluation interviews; c) subgroups within the risk groups; d) the attitudes, norms, barriers, and facilitators regarding risk reduction among persons in the risk groups; and e) potential community partners.
  • Concepts from behavioral theories and models should be used to develop and guide prevention activities (26-28). Important concepts from several behavioral science theories and models were integrated to form a framework for the ACDPs prevention activities. These concepts provided a basis for the specific prevention strategies used in the projects. For example, role-model stories were used to illustrate important determinants of intention and behavior change such as perceived norms and attitudes (7) or self-efficacy regarding condom use (8) in a way that was relevant to the target audience. In addition, the SOC continuum was used both to tailor intervention messages and to assess progress toward behavior change among participants.
  • An intervention protocol should be designed to allow for adaptation to different populations and communities. Although the ACDPs adhered to a common study protocol, each project was able to tailor the interventions to the specific targeted populations and to the environmental conditions in each city.
  • Ongoing qualitative research should be used to ensure that the intervention responds to changes in the community. After the project interventions began, the environment of the project sites continued to change. The information obtained through qualitative research enabled the project researchers to continue to appropriately tailor their interventions throughout the 3-year intervention period and provided insights regarding the effectiveness of the intervention.
  • Persons within the communities should be recruited to deliver intervention messages. Although the specific methods and relative success of the ACDPs differed, each project established a peer network and recruited a cadre of interactors (except New York, which sought to establish a peer network only). These networks enabled the interventions of the projects to reach greater numbers of persons and appeared to encourage behavior change among community members. Although network members received instruction in a core set of practices, the extent to which they assumed different roles in the intervention effort was determined by their personal strengths, beliefs, and level of comfort regarding their role in the intervention. The intervention strategy also encouraged community involvement with the project and enabled participants to acquire skills for helping others in their communities. To retain network members, project staff used both monetary and nonmonetary incentives.
  • Community-level HIV prevention interventions should be used to reach less accessible populations at risk for HIV infection. One of the primary goals of these projects was to target HIV-prevention interventions to at-risk populations who were not necessarily participating in established, facility-based prevention programs (29). The community-level activities were meant to initiate change in the community norms regarding protective behaviors to provide a more supportive environment for consistent condom and bleach use among community members. The research and prevention approach was designed to be as sensitive as possible to the populations concerned, mainly relying on the participation of persons who were part of that population (peers). Because the intervention was implemented in the environment of persons at risk for HIV infection, it could provide constant cues for adopting preventive behaviors.

The authors thank Mary Guinan, David Holtgrave, Laura Leviton, and Dan Schnell for their assistance with this report. The authors also recognize the contributions of Kevin O'Reilly, Christine Galavotti, and Dan Schnell in the design of these AIDS Community Demonstration Projects, and Carolyn Beeker and Linda Kay for their assistance with the projects. Finally, the authors thank the interviewers, outreach staff, and community volunteers without whom this research project could not have been completed.



  1. O'Reilly K, Higgins DL. AIDS community demonstration projects for HIV prevention among hard-to-reach groups. Public Health Rep 1991;106:714-20.
  2. McAlister A, Puska P, Koskela K, Pallonen U, Maccoby N. Mass communication and community organization for public health education. Am Psychol 1980;35:375-9.
  3. McAlister A. Population behavior change: a theory-based approach. J Public Health Policy 1991;12:345-61.
  4. Puska P, Salonen JT, Koskela K, et al. The community-based strategy to prevent coronary heart disease: conclusions from ten years of the North Karelia project. Annu Rev Public Health 1985;6:147-93.
  5. Cook TD, Leviton LC, Shadish WR. Program evaluation. In: Lindzey G, Aronson E, eds. The handbook of social psychology. Vol 1, 3rd ed. New York: Random House, 1985:699-777.
  6. Becker MH. The health belief model and personal health behavior. Health Education Monographs 1974;2:324-473.
  7. Fishbein M, Ajzen I. Belief, attitude, intention and behavior: an introduction to theory and research. Reading, MA: Addison-Wesley Publishing Company, 1975.
  8. Bandura A. Social foundations of thought and action. Englewood Cliffs, NJ: Prentice-Hall, 1986.
  9. Prochaska JO, DiClemente CC. Stages and processes of self-change of smoking: toward an integrative model of change. J Consult Clin Psychol 1983;51:390-5.
  10. Prochaska JO, DiClemente CC, Norcross JC. In search of how people change: applications to addictive behaviors. Am Psychol 1992;47:1102-14.
  11. Prochaska JO, DiClemente CC. The transtheoretical approach: crossing the traditional boundaries of therapy. Homewood, IL: Dow-Jones/Irwin, 1984.
  12. CDC. NCPS AIDS Community Demonstration Projects: what we have learned, 1985-1990. Atlanta: US Department of Health and Human Services, Public Health Service, June 1992.
  13. CDC. Condoms for prevention of sexually transmitted diseases. MMWR 1988;37:133-7.
  14. CDC. Update: barrier protection against HIV infection and other sexually transmitted diseases. MMWR 1993;42:589-91, 97.
  15. Higgins DL, O'Reilly KO, Tashima N, et al. Using formative research to lay the foundation for community-level HIV prevention efforts: an example from the AIDS Community Demonstration Projects. Public Health Rep (in press).
  16. Goldbaum G, Perdue T, Higgins D. Non-gay-identifying men who have sex with men: formative research results from Seattle, Washington. Public Health Rep (in press).
  17. Corby NH, Enguidanos SM, Kay LS. Development and use of role-model stories in a community-level risk-reduction intervention. Public Health Rep (in press).
  18. Guenther-Grey C, Noroian D, Fonseka J, Higgins D. Developing community networks to deliver HIV prevention interventions: the AIDS Community Demonstration Projects. Public Health Rep (in press).
  19. Simons PZ, Rietmeijer CA, Kane MS, Guenther-Grey C, Higgins DL, Cohn DL. Building a peer network for a community level HIV prevention program among injecting drug users in Denver. Public Health Rep (in press).
  20. CDC. AIDS community demonstration projects: implementation of volunteer networks for HIV-prevention programs -- selected sites, 1991-1992. MMWR 1992;41:868-9, 875-6.
  21. Cook TD, Campbell DT. Quasi-experimentation. Boston: Houghton-Mifflin, 1979.
  22. National Research Council. Evaluating AIDS prevention programs (expanded edition). Washington, DC: National Academy Press, 1991.
  23. Agresti A. Analysis of ordinal categorical data. New York: John Wiley & Sons, 1984.
  24. Judd CM, Kenny DA. Estimating the effects of social interventions. New York: Cambridge University Press, 1981.
  25. Milburn MA. A longitudinal test of the selective exposure hypothesis. Public Opinion Quarterly 1979; 43:507-17.
  26. Fisher JD, Fisher EA. Changing AIDS-risk behavior. Psychol Bull 1992;111:455-74.
  27. Coates TJ. Strategies for modifying sexual behavior for primary and secondary prevention of HIV disease. J Consul Clin Psychol 1990;58:57-69.
  28. Leviton LC. Theoretical foundations of AIDS-prevention programs. In: Valdiserri RO, ed. Preventing AIDS: the design of effective programs. New Brunswick, NJ: Rutgers University, 1989:42- 90.
  29. General Accounting Office. AIDS prevention programs: high-risk groups still prove hard to reach. Washington, DC: General Accounting Office, Human Resources Division, 1991; GAO/HRD-91-52.
* In 1989, because of available information and the unavailability of needle-exchange programs in the project areas, project researchers focused on the consistent use of bleach to clean injection equipment among IDUs who shared equipment.

** Bleach kits contained instructions (usually illustrated) for using bleach and water to clean injection equipment, a small bottle of bleach, and sometimes water and alcohol pads.

*** For further information regarding this instrument, contact the Behavioral Intervention Research Branch, Division of HIV/AIDS Prevention, National Center for HIV, STD, and TB Prevention (Proposed), CDC, MS E-44, 1600 Clifton Road, N.E., Atlanta, GA 30333.


To request a copy of this document or for questions concerning this document, please contact the person or office listed below. If requesting a document, please specify the complete name of the document as well as the address to which you would like it mailed. Note that if a name is listed with the address below, you may wish to contact this person via CDC WONDER/PC e-mail.
For single issue purchase 800-843-6356
write to: CDC, MMWR MS(C-08)
Atlanta, GA 30333

Table 1

TABLE 1. Method for assigning stage of change for consistent condom use *
                                             Stage of change
                    Stage 1         Stage 2        Stage 3       Stage 4         Stage 5
                     Pre-                         Ready for
Criterion        contemplation   Contemplation      action       Action        Maintenance
Frequency of
use +                ---              ---         Sometimes/     Every time     Every time
                                                  almost every

Duration of
"every time"
use &                ---              ---            ---         <6 months      >=6 months

intention @          ---              ---         Extremely/        ---            ---
                                                  slightly sure

 intention **        ---          Extremely/      Extremely/        ---            ---
                                  quite/          quite/
                                  slightly sure   slightly sure
                                  will            will
*  An individual's stage of change (SOC) for condom use is assigned by starting with the criteria
   necessary for Maintenance, then Action, etc. This method also is used for assigning SOC for
   consistent use of bleach to clean injection equipment.
+  Persons interviewed were asked, "When you have (vaginal/anal) intercourse with your
   (main/nonmain) partner, how often do you use a condom?" Respondents' choice of answers
   included the following: every time, almost every time, sometimes, almost never, never.
&  Persons interviewed who reported using a condom every time or almost every time were asked,
   "How long have you been using a condom (every time/almost every time) you have (vaginal/anal)
   intercourse with your (main/nonmain) partner?"
@  Persons interviewed were asked, "How likely do you think it is that from now on you will use
   a condom every time?" Respondents' choice of answers included the following:
   extremely/quite/slightly sure I will, undecided, slightly/quite/extremely sure I won't.
** Persons interviewed were asked, "In the next 6 months, how likely do you think it is that
   you will start using condoms every time?" Respondents' choice of answers included the
   following: extremely/quite/slightly sure I will, undecided, slightly/quite/extremely
   sure I won't.

Table 2

TABLE 2.  Intervention exposure by city, intervention community, and phase
                                                    Start-up       implementation
City            Intervention community           exposure rates    exposure rates
Dallas          Two census tracts                 33/329 = 10%      109/352 = 31%
Denver          IDUs *                             4/307 =  1%       48/217 = 22%
Long Beach      IDUs                              25/326 =  8%      168/305 = 55%
                Female sex traders                26/322 =  8%      187/274 = 68%
                Female sex partners of male IDUs  10/135 =  7%       55/121 = 45%
New York City   Female sex partners of male IDUs  32/176 = 18%       80/174 = 46%
Seattle         Nonhomosexually identifying men
                 who have sex with men             2/77  =  3%       24/108 = 22%
                Female sex traders                23/125 = 18%       79/217 = 36%
                Street youth                      32/214 = 15%       50/202 = 25%
* Injecting-drug users.

This page last reviewed: Wednesday, January 27, 2016
This information is provided as technical reference material. Please contact us at to request a simple text version of this document.