Suggested Citation: Ward D, Hess R, Lefebvre RC. Key Components in Planning, Implementing and Monitoring a Behavior Change Communication Campaign that Increased Condom Use Among Male Clients of Sex Workers in Southern India. Cases in Public Health Communication & Marketing. 2008; 2:105-125. Available from: www.casesjournal.org/volume2.

---

Abstract

India is home to nearly 10% of all people living with HIV worldwide or approximately 2.5 million people. The primary mode of HIV transmission in India is heterosexual commercial sex. In conjunction with the Avahan India AIDS Initiative, funded by the Bill and Melinda Gates Foundation, Population Services International (PSI) developed and implemented an intervention to reduce HIV incidence, in part by increasing consistent condom use among heterosexual male commercial sex clients in southern India. The project focused on 100 "high priority" towns in the provinces of Andhra Pradesh, Karnataka, Maharashtra and Tamil Nadu. The intervention targeted heterosexual male commercial sex clients of low socio-economic status in the high priority towns. Communication activities focused on interpersonal methods, such as street plays, contests, and group discussions. The project was largely successful in meeting its objective. Clients of commercial sex workers increased their reported consistent condom use during commercial sex from a baseline of 63% in May 2006 to 81% in May 2007 following the intervention. This 18% increase is strongly correlated with exposure to PSI's communication activities (p<0.01), and indicates a high level of achievement as compared to other voluntary behavior change communications campaigns.

Background

In India, more than 2.5 million people are infected with HIV. This is nearly 10% of all HIV cases in the world.¹ The majority of people infected with HIV in India (about 65%) live in six provinces: the southern states of Tamil Nadu, Andhra Pradesh, Maharashtra and Karnataka; and the northeast states of Manipur and Nagaland (see Figure 1, next page).

India's HIV situation is complex. The national HIV prevalence of less than 0.4% masks high prevalence rates in selected areas and among key populations. For example, among the population of more than 4 million in the Belgaum district of northern Karnataka, the prevalence of HIV infection is over 3%.²

While HIV affects people from all sectors of Indian society, it has most greatly affected four groups: sex workers, their clients and partners, men who engage in sex with men, and injection drug users. In the states of Manipur and Nagaland the epidemic is primarily driven through injection drug use; while in the other four high prevalence states, primary transmission occurs through heterosexual commercial sex.¹

At the start of the program, most male commercial sex clients were using condoms. A May 2006 PSI baseline survey on condom use found that 62% of male clients of female sex workers reported using a condom consistently in the previous 12 months. Only 2% reported never using a condom and 3% reported not using condoms at all in the past 12 months (PSI, unpublished data, May 2006). Based on this information, the challenge was to convert occasional and fairly regular condom users into consistent users.

Beginning in 2004, PSI implemented an evidence-based HIV prevention program in India as part of the Bill & Melinda Gates Foundation's Avahan Indian AIDS Initiative. The program made extensive use of qualitative and quantitative research in project design, monitoring and evaluation. This included identification of key determinants of consistent condom use, creation of messaging based on these determinants, and development of communication activities to deliver these messages effectively. The intervention demonstrated that it is possible to increase consistent condom use among groups at high risk of contracting HIV.

PSI used the following principles to frame its strategy:

1. Choose and define behavioral determinants based on evidence;
2. Focus messages on one simple idea at a time;
3. Decentralize creative process and implementation management to ensure cultural relevance and scale;
4. Use several media at high frequencies of exposure to communicate the message;
5. Pretest all activities thoroughly;
6. Change themes every four months to reduce message fatigue;
7. Monitor activities bi-annually to measure the following components:
   • a. Reach/recall - are the target planning strategies achieving desired reach?
   • b. Impact - is exposure to activities correlated with increases in desired behavior and determinants? At what frequency or in what combination? What is the most cost effective way forward?
   • c. Which determinants are likely to have most impact as these change over time?

The intervention was implemented in 100 high priority towns in Andhra Pradesh, Karnataka, Maharashtra and Tamil Nadu, all located in southern India. These towns were selected based on HIV prevalence data and number of sex workers. The overall goals of the project were to reduce HIV incidence by increasing access to condoms, increase consistent condom use among the target audience, improve access to proper treatment for sexually transmitted infections, and create demand for proper STI treatment. For condom use, which is the focus of this article, the objective was to increase the percentage of the target population reporting consistent condom use with commercial sex partners in the previous 12 months from 62% to 69% (a mark which was quickly surpassed). This was to be achieved by:

• Increasing access to condoms, as measured by:
  • Increasing the percentage of areas of commercial sex and solicitation in target towns or "hot zones" meeting minimum coverage standards for condom availability from 40% in 2005 to 95% by 2008 (annual PSI coverage surveys) and
  • Selling 129 million condoms over five years (PSI internal sales figures);
• Increasing demand for condoms by addressing key behavioral determinants; i.e., among the target population, increasing the mean score of positive beliefs about condoms from 0.73 to 0.78 on a scale of 0 to 1 (Biannual PSI tracking surveys).

Target Audience
To complement the work of other Avahan partners whose programs were targeting commercial sex workers, PSI focused on the male clients of sex workers. Men are recognized as an essential target group because they can bridge high risk groups to the general population; furthermore, men often control the decision process regarding condom use in encounters with sex workers. Currently, researchers estimate the male clientele of sex workers in the 100 priority towns of the PSI project area at between 4 and 5.5 million men (PSI estimates). PSI's research indicates that 27% of men in these towns have paid for sex in the past 12 months (PSI, unpublished data, May 2006). Due to PSI's mission to meet the needs of low-income and vulnerable people, PSI focused on clients in Social Economic Classes C, D, and E. These social economic classes are defined by the Indian government based on education levels and head of household income; classes C, D, and E are among the lowest classes.

Description of Intervention:
Increasing Consistent Condom Demand

In order to increase consistent condom use, PSI undertook integrated behavior change initiatives. An iterative process of planning, implementation, monitoring, and evaluation (as shown in Figure 2, next page) guided these efforts. The planning phase lasted six months, followed by a year of implementation. The present case study will describe in detail the process of message development and determining the target's ideal frequency of exposure. It should be noted that the program was fortunate to have the substantial resources necessary to research and monitor the program every six months. The process of evidenced-based message design and evaluation is the subject of this article.

Planning
At the time PSI launched the current program, it had little (if any) presence in most of the 100 priority towns. Therefore, much of the planning phase involved establishing offices and hiring staff. Although other components of the program started in 2003, such as subsidized distribution of condoms, the initial baseline study for condom use was conducted in May 2006 (described below) and other activities were launched shortly thereafter.

Message Development -
Qualitative Hypothesis Development
Qualitative research was first used to develop the quantitative surveys. The PSI research methodology uses an individual behavior change framework to design messaging. The framework draws on a variety of theories, including Social Learning Theory, 6 Theory of Reasoned Action,^7,8 Theory of Planned Behavior,⁹ and the Health Belief Model10 that explain why people undertake behaviors like consistent condom use. The first step is to define each behavioral determinant as it applies to this target audience and behavior. Each behavioral determinant is a multi-item scaled construct, meaning the concept is defined by several consumer statements. Constructs (hypotheses) were developed based on a secondary review of validated instruments to evaluate condom use from multiple sources; hypothetical statements were then generated on the topic of why a person might or might not use condoms. Next, focus group discussions were held among representatives of the target audience to identify and evaluate perceptions, beliefs and attitudes for using or not using condoms consistently.

Based on the findings, researchers developed simple unambiguous statements about condom use and categorized them under various determinants, per guidance from PSI concept paper on Behavior Change Framework.¹¹ These statements were then used to develop the quantitative survey.

Message Development -
Quantitative Research
After pre-testing and refining the statements developed from qualitative research, PSI commissioned a study of 2400 male clients of sex workers in project towns in May 2006. The goal of this study was to statistically examine what differentiates a consistent condom user from an inconsistent one. The theory is that if we can make inconsistent users more like consistent users in their motivation, ability and opportunity to behave, they will be more likely to adopt the behavior.

Two stage stratified sampling was used among men who reported visiting female commercial sex workers in the project sites in the preceding 12 months (typically about one-third of the adult male population in these areas, according to PSI unpublished research, May 2006, December 2006, and May 2007). Hot spots were selected using probability proportional to size sampling, meaning that areas with larger numbers of sex workers and clients were more likely to be included in the sample. Respondents were then recruited via a randomized selection procedure utilizing intercept interviews at the selected hot spots. A structured questionnaire collected information on demographics, behavior, hypothesized behavioral determinants, and levels of exposure to communication activities (interpersonal communication, street theatre and small group activities).

The above study grouped participants into one of two categories: "behavers," those who perform the desired behavior (in this case consistent condom use) or "non-behavers," those who do not use condoms consistently. These included self-efficacy (the self confidence needed to buy and use condoms), belief (that condoms can prevent AIDS), and subjective norms (perceived approval of friends and family for using condoms). Factor analysis and reliability tests were conducted with the collected data to develop operational definitions of these constructs. Analysis was conducted using SPSS statistical package version 13.0 (SPSS Inc., Chicago, IL). Positive beliefs about condom use were measured by participant agreement with any of 13 positive statements (see Appendix A); agreement with more of these statements indicated more positive beliefs.

After data collection and input, the STATA software package version 10.0 (StataCorp., College Station, TX) was used to conduct multivariate analysis, to estimate adjusted means and proportions, while controlling for confounding variables that could introduce bias. Logistic regression was used to examine relationships between the determinants and the dichotomous outcome variable (i.e., consistent or inconsistent condom use).

The primary objective of the analysis was to identify which determinants showed the most difference between behavers and nonbehavers. So while all determinants helped define behavior, project staff took no action on those which showed no statistical difference between the two groups. As such, social norms were not selected as a key determinant, since behavers and non-behavers had similar perceptions of the social norms of condom use in their community. Indeed, the behavior change framework lends support to this method, in stating that changing the behavior of the target audience (the non-behavers) on the determinants which most differentiate them from the behavers would be the easiest way to increase consistent condom use among the target group. This is illustrated in the following table.

The baseline study (see Table 1) allowed program planners to identify variables associated with consistent condom use. Social support, beliefs, and perceived severity were the variables most strongly associated with consistent condom use. For example, male clients of sex workers with positive beliefs about condoms were 5.12 times more likely to use condoms consistently in commercial sex than those with less positive beliefs.

Two key determinants identified by the research were consistent across all four states of the project area. One determinant, however (perceived severity of HIV) strongly correlated with consistent condom use only in the state of Maharashtra. In order to simplify message development and ensure better consistency across the project area, the design team opted not to focus on this state-specific determinant. Social support was also not selected as a key determinant because it correlated less with condom use than did beliefs; i.e., men in the target audience who rated social support highly were only 1.42 times more likely to use condoms consistently, substantially lower than the 5.12 times associated with beliefs.

Message Development -
Qualitative Research to Understand Beliefs
After analysis of the quantitative results, qualitative focus group discussions with the target audience helped PSI staff further understand the relationship between and meaning behind the 13 belief statements. Two focus groups and 16 in-depth interviews were held with "high believers" (those who answered "yes" to at least 10 belief questions), 2 focus groups and 16 in-depth interviews with "low believers" (those who answered "yes" to less than 9 of the belief questions), and 3 focus groups with a mix of high and low believers (called "conflict groups"). This research revealed that the 13 positive beliefs could be categorized according to 5 more general beliefs. Four of these were translated into communication themes (the fifth was determined to be most relevant to complete non-users; therefore it was not chosen as a theme since it would apply to less than 3% of the target audience). The four beliefs were:

1. Condoms won't burst if used correctly;
2. Condoms won't slip off if used correctly, therefore there is no need to worry;
3. No matter what choices you make, you always need a condom in commercial sex;
4. No matter which sex worker you choose, you always need a condom.

The Planning Process
After identifying the key belief messages about condom use, described above, the next step was to incorporate the messages into a communication program and adapt it to scale. Messages were disseminated using interpersonal communication; through activities such as street theater, folk dramas performed on street corners, and small group contests and games. Project designers sought to learn where to reach the greatest numbers of male clients of sex workers, and what level of contact was needed to convert inconsistent condom users into consistent condom users. PSI found that that the male clients of commercial sex workers were not a cohesive group and thus, could not be easily segmented using the typical psychographic or demographic criteria (e.g., age, profession, marital status, religion, or drinking habits) to find appropriate media outlets. Researchers therefore decided that the best strategy would be to target clients by geography. All members of the priority audience paid for sex, and frequented a limited number of places to buy sex (a.k.a., hot spots). Cities such as Mumbai have known brothel areas; while in other cities, sex workers either work out of their homes or solicit clients in the streets and then visit a paid room in a lodge. This strategy of geographical targeting proved to be successful in reaching this audience, as only 6% of men in the project districts report paying for sex,4 compared to approximately 30% of men in these hot spots who report doing so (PSI, unpublished data, 2006).

Based on this information, PSI's communication teams conducted street plays, small group activities such as games and condom demonstrations, and one-on-one and group discussions with potential target group members. To maximize the chances of contact with actual clients, the program was carried out in neighborhoods known to be areas of solicitation for commercial sex. Since it was impossible to distinguish male clients from other males (non-clients), as mentioned above, any man in the target age range was invited to participate in activities.

Implementation

To ensure adaptation of the campaigns to the local culture and language unique to each of the four states, program implementation was decentralized. The state office teams developed creative briefs, scripts and materials for the four message themes, with technical assistance from a team of experts from the central office in New Delhi. Practitioners branded themes that were incorporated into all activities (street theater, interpersonal communication and small group activities) for four months. Due to limited resources, PSI was unable to use outside researchers to pre-test scripts and materials; therefore, the materials were pre-tested for persuasiveness and unintended consequences with relevant communities by in-state PSI teams. The acknowledged limitations of internally testing activities (especially potential confirmation bias, as the same people who design the activities are involved in their testing) were considered necessary in order to save time and money. To mitigate this potential concern, a technical advisor attended each pretest to oversee and help the pre-testing process. The advisors were skilled in both research and communications methodologies and had not been directly involved in crafting the messages, making them neutral parties. As such, they were unlikely to introduce bias in the process of material pretesting.

Activity pre-testing was performed with inconsistent condom users only, as they were identified to be the primary audience. Respondents were shown the communication activity and were then asked to complete a structured questionnaire that evaluated comprehension, relevance, believability, likeability, persuasiveness, and gross negatives. The findings were analyzed and the activities were refined where appropriate.

In 2007, as the project reached its peak of activities, it employed 457 communications staff (interpersonal communicators and supervisors) in addition to the more than 50 street theatre troupes who were on contract. As shown in Figure 3 (next page) the project reached over 1 million men per month through interpersonal communication activities (PSI internal data).

Monitoring & Evaluation

Based on PSI's experience with previous communication programs in India, this project started with an understanding of the minimum number of exposures necessary to achieve behavior change. Therefore, a goal of reaching 40% of the target audience 6 times per year was deemed reasonable. However, two limitations hindered progress toward that goal. Lack of an exact number of male clients was the first (estimates in the selected cities varied widely from 1 million to 5 million men). The second related to monitoring methods of uncertain accuracy. Because the audience members were continually coming and going throughout the street theater performances and small group activities, it was impossible to achieve an exact count of attendees. Communication team leaders would typically count the number of adult men at a performance at several points during the play and reduce that average by a certain percentage, based on the percent of men in that hot spot who admitted to paying for sex (typically about 30%). Clearly, however, these were rough estimates. As the target audience size and PSI's targeting efficiency were considered unknowns PSI adopted a convergent (i.e., multiple methods) validation approach. After 6 months of activities, PSI compared the number of men estimated as having been reached through activities to the percentage of male clients surveyed who recalled seeing each activity. This resulted in the following equation for each activity in each state: If 100,000 men were reached, resulting in 80% recall, how many men must we reach to get 40% recall?

The design team hypothesized that this method would allow them to set appropriate communication reach targets, as long as "reach" was calculated in the same way (with all of its limitations), and the target audience size did not change drastically; thus achieving desired message exposure while remaining cost effective.

Results

Two additional surveys were conducted in December 2006 (n=1756 male clients) and May 2007 (n= 1741 male clients) to monitor recall of communication, and levels of reported condom use and behavioral determinants. Analyses examined whether recall of PSI's activities was correlated with any observed changes in the determinants of consistent condom use (see Table 1). Additional analyses were performed to determine whether the program should continue to focus on the same beliefs which were being targeted.

Results of the May 2007 survey (see Table 2) indicated that beliefs were still the determinant most correlated with consistent condom use. The research also showed the success of the communication activities, as both consistent condom use and positive beliefs increased. Further, the survey showed that 84% of male commercial sex clients in project areas recalled seeing at least 1 of the 3 PSI activities (street theatre, interpersonal communication and small group activities) in the previous 3 months (see Table 3, next page).

In determining efficacy of the behavior change communication activities, the correlation between activity recall and the increase in condom use and beliefs was examined. The data in Table 4 (below) indicates that exposure to each activity did indeed correlate with increased consistent condom use. Interestingly, no such correlation existed when examining activity exposure and subsequent reporting of positive beliefs (while positive beliefs did increase over time, exposure to PSI's activities in the previous three months was not associated with that increase). This was true except in Andhra Pradesh, where recall of street plays and small group activities were found to correlate with increased positive beliefs at the state level.

Discussion

A large-scale interpersonal and street theatre communication program demonstrated an 18% increase in the consistent use of condoms among men who frequently engage in sex with female sex workers. This level of success is beyond that of most health campaigns targeting behavior change.12 The December 2006 and May 2007 tracking surveys showed that recall of PSI's communication activities increased from 79% in December 2006 to 84% in May 2007. An increase in PSI's targeting efficiency is one possible explanation. For example, in May 2006, 60 out of 100 men at a street show may have been male clients of commercial sex; however, after a year of working in those communities, program planners may have been able to adjust performance times, locations, and attractiveness so that in May 2007, 90 out of 100 men at the street show were members of the target audience. Alternatively, the results could have been skewed by recall bias, meaning that men reported recalling activities they did not actually attend. To quantify this over-reporting and achieve greater reliability in its estimates, PSI has since implemented new research methods. New research methods include the use of photographs as visual aides. Respondents are asked to identify which of several street theater performances they attended, which are presented to them in photographs. In reality, only one of the photos depicts a performance taken in their town.

An on-going objective is to determine the optimal frequency and combination of activities to maximize and maintain consistent condom use behaviors. The Avahan-funded project was fortunate to have sufficient funding to implement this behavior change intervention on a large scale. Project staff are currently examining to what degree the activities can be scaled back while still achieving and maintaining behavior change.

Lessons Learned

Project staff learned a great deal concerning the design and implementation of evidence-based behavior change programs. One such lesson was not to underestimate the challenge of maintaining focus in a communications program based on interpersonal contacts. Project quality control personnel observed difficulties in maintaining campaign focus among the interpersonal communications teams, despite thorough training and supervision. Many of PSI's interpersonal communicators have an educational and/or professional background in social work; as such, they viewed each contact with target group members as an opportunity to disseminate as much information as possible, in order to help them. They easily digressed into areas not related to the direct messages of the campaign. Observers noted that the communicators were much more effective in sharing and ascertaining information about simple core beliefs, such as how easy it is to prevent condoms from breaking, when they did not talk about the broader issues of HIV, AIDS, STIs, risks, and symptoms.

Another lesson was that the effective campaign dose (i.e., intensity) needed for behavior change can be estimated even if the size of the target group is not known with certainty. Through quantitative research, project staff were aware of the percentage of the target group who had been exposed to PSI's on-ground communications teams. Thanks to the project's reporting system, staff were able to relate the number of contacts made through the various media (such as street theatre performances and group discussions) with actual percentage of the target group reached and the number of contacts achieved. If the percentage reached was below target, more communications teams could be mobilized. If reach was above target (meaning that the average number of contacts had gone beyond the threshold required for behavior change) on-ground activities could be cut back, thus increasing cost efficiency while still achieving the objective.

A final important lesson learned from the program was that multiple iterations accelerated the pace of change. Project researchers conducted quantitative monitoring studies on key behaviors approximately every six months. Results showed whether the given strategy carried the desired impact with the target group. Many mid-course corrections, including changing media and adjusting messaging were made possible by this research. Had this constantly evolving picture of the project not been available, it would have taken longer to discover successful campaign components and those that needed to be changed.

Conclusions

The most important lesson learned was that a carefully designed communications program-one that is evidence-based and applied consistently-can increase consistent condom use. Consistent condom use may be notoriously difficult to achieve, given the many obstacles to changing behaviors related to sexual health. However, when a program is focused on evidencebased messages, with adequate frequency and feedback loops, health communication campaigns can be an effective means of achieving behavior change.

---

Acknowledgments

Funding for this project was provided by Avahan: the India AIDS Initiative of the Bill & Melinda Gates Foundation. The views expressed herein do not necessarily reflect those of Avahan or of the Bill & Melinda Gates Foundation. The authors wish to thank the communication and research teams working on PSI's Avahan project for sharing their learnings and experiences.

---

References

1. Joint UN Programme on HIV/AIDS; National AIDS Control Organization, India; World Health Organization. 2.5 million people in India living with HIV, according to new estimates. 2007.
2. National AIDS Control Organization, India. Annual HIV sentinel surveillance country report. 2006.
3. National AIDS Control Organization, India. National baseline high risk and bridge population behavioural surveillance survey (BSS): Female sex workers and their clients. 2001.
4. Dandona L, Dandona R. Drop of HIV estimate for India to less than half. Lancet. 2007;370(9602):1811-1813.
5. Global HIV Prevention Working Group. Access to HIV prevention: closing the gap. 2003.
6. Bandura A. Social Learning Theory. Englewood Cliffs, NJ: Prentice-Hall; 1977.
7. Ajzen I, Fishbein M. Understanding Attitudes and Predicting Social Behavior. Prentice-Hall Englewood Cliffs, NJ; 1980.
8. Fishbein M, Ajzen I. Belief, attitude, intention, and behavior: An introduction to theory and research. 1975.
9. Godin G, Kok G. The theory of planned behavior: A review of its applications to health-related behaviors. Am J Health Promot. 1996;11(2):87-98.
10. Rosenstock I. The health belief model and preventive health behavior. Health Educ Monogr. 1974;2(4):355-385.
11. Patel DS, Chapman S. A tool for social marketing decision making. 2004.
12. Hornik RC. Public Health Communication: Evidence for Behavior Change. Lawrence Erlbaum Associates, Mahwah, NJ; 2002

---

Appendix