A SOCIAL NETWORK CONSISTS OF a set of individuals or groups connected by links that represent relationships, such as friendship, or interactions, such as sex (sexual networks).1 Social network research focuses on these links and how they can affect individuals and group dynamics.2 Network concepts have been very important in research on human immunodeficiency virus/sexually transmitted infection (HIV/STI) and also used in infection control through partner notification (contact tracing).3,4 The usefulness of social networks is based on the quality and completeness of linkage information and, therefore, is highly dependent on individuals' abilities to identify all other persons with whom they have had some interaction.
Several studies have shown that respondents, when asked to list all people having different relationships with them, tend to forget a great proportion of those people.5 Even for intimate relationships, such as sexual contacts, it has been shown that individuals at high risk for STI recalled only about 40% of their sexual partners when asked to recall all sexual contacts in the 2-year period preceding the interview.6 This is particularly true for people having multiple sexual partners. Therefore, a list of recalled partners is not a list of all network members but a nonrandom subset of network members.7
Brewer et al.8–10 have been working on techniques to increase partner recall and have developed a methodology based on the cognitive structures people use to organize items in memory. They were able to show that people cluster their sexual and injection partners in recall according to location of subject–partner interaction (such as house or hotel), role relationship (such as wife/husband, friend, and coworkers), and social proximity (degree of social interaction between partners). They developed supplementary cueing techniques, to be administered after subjects had freely listed all possible recalled partners, which correspond to the described way people organize their memories. These techniques have been used and evaluated in the United States, but have never been applied in other cultural contexts.
Materials and Methods
The study was conducted between June, 2005, and August, 2006, at the HIV Counseling and Testing Center (CTC) located at the Hospital Escola São Francisco (Universidade Federal do Rio de Janeiro), Rio de Janeiro, Brazil. The CTC is currently an anonymous testing center, free of charge and open to anybody. Approximately 600 individuals are tested monthly in the CTC. It has been described as attracting a high-risk heterosexual population with an HIV prevalence of approximately 8%.11 This prevalence rate is much higher than the overall Brazilian prevalence, approximately 0.5%.12,13 Based on routinely collected data, roughly half of the people attending the CTC are women (51.2%) and approximately 11% are men who have sex with men. Approximately 43% of the CTC population are married and 51.2% have 8 to 11 years of education. The main reasons that individuals report for going to the CTC for HIV testing are concern about a possible exposure (56.5%) and prenatal care (11.9%). The most important risk exposure is a sexual relationship (90%), and approximately 85% of the CTC clients reported fewer than 4 partners in the past year.
The current study is part of a larger network research study carried out at the CTC. The purpose of the main study was to evaluate whether an individual's personal network is an important predictor of HIV status. We used the cue methodology to increase partners' recall and wanted to investigate, as a secondary objective, the effectiveness of the technique in our population. We piloted the cues in 20 people and identified some problems in the location and role cues. For example, Carnaval is a major national festivity and is a well-known time of the year where several people in Brazil meet sexual partners. Carnaval was not part of the location cue and we decided to include it after our pilot study. We made all pertinent changes before applying the cues to our volunteers.
The study was cross-sectional in its design. All individuals older than 18 years and visiting the CTC for the first time were invited to participate in our study during regular group counseling activities. We decided to exclude pregnant women under the assumption that, on average, their sexual behavior in the year before the interview would not correlate with their lifetime sexual behavior. Because of lack of physical spaces for more than 2 interviews at the same time and lack of availability of some volunteers for staying at the CTC for longer than a regular visit, only a portion of all potential volunteers were interviewed. Every volunteer was interviewed only once and tested for HIV. Voluntary recruitment methods were used for the program, including education about the project and informed consent before screening. Only consenting individuals were enrolled in the study. The study was approved by both the Institutional Review Board at University of Maryland, and the Ethics Committee at Universidade Federal do Rio de Janeiro, Brazil.
All interviews were face-to-face and used standardized, prevalidated questionnaires administered verbally in Portuguese. Two interviewers were available during the entire day to apply the questionnaire, 1 man and 1 woman. The main questionnaire contained questions about each person's risk factors, and a second part addressed recall of regular and casual sex partners and questions about partners' risk factors.
The technique has been extensively described elsewhere.8,9 Briefly, we first defined sexual contact to the volunteers as any kind of oral, vaginal, or anal sex that involved 2 or more people, regardless of the situation in which the sexual contact occurred or the type of relationship the volunteer had with the other person. Free recall was prompted with the following sentence: “Quando eu perguntar, me diga os nomes de todas as pessoas com quem você teve sexo no último ano.” (Translation: “When I ask you to, tell me the names of all the people you had sex with in the past year.”). Volunteers were asked to report first names, nicknames or anything that reminded them of a specific partner. If they did not know the name of a partner they had sex with, they were asked to give some word or brief description to refer to this person. No positive identification of any partner was done in our study, and no partner notification was ever attempted during the study period. When participants indicated that they did not have or could not recall any more partners, we used the nonspecific prompt: “Com quem mais você teve sexo no último ano?” (translation: “Who else have you had sex with in the past year?”), until participants insisted they could not recall any additional partners. After that, we read back the list of partners to the volunteer to ensure that all recalled partners were correctly recorded. We then used the nonspecific prompt again.
The supplementary technique was used only after the volunteers were sure they did not recall any other sexual partner in the prior year. Also, the technique was applied only to volunteers who reported more than 1 regular (stable relationship such as husband/wife and boyfriend/girlfriend) and/or casual partner (no intention of keeping a long-lasting relationship). Previous research has shown no benefit when using the technique for people reporting only 1 partner.10 The proposed order for the cues was based on its efficacy on partners' recall: location, alphabetic, role, and network.8 After a short introduction to the technique, the location cues were given to volunteers. The method uses a list of different locations (total of 32 cues) where people have sex with other people or meet people with whom they later have sex. Volunteers were asked to report any other sexual partner they had met in those locations but had forgotten to mention during free recall. The alphabetic cues are based on reading all the letters in the alphabet for the participants (including k, w, and y that are not officially used in Portuguese) and asking them whether they have any other partner not mentioned before with a name that starts with a specific letter. During the administration of the role cues (total of 16 cues), interviewers read a list of different kinds of relationships people have with the persons they have sex with and asked volunteers to think of all the persons with whom they had this kind of relationship and report whether they had sex with these persons. Finally, for the network cues, interviewers read back the list of people the participants had mentioned before. The interviewer repeated the name of each person, and participants were asked to think about all other people that this person may know, hang out or interact with, and to report whether they had sex with any of those people.
Our aim was to evaluate if the number of recalled sexual partners could be augmented by using supplementary cueing techniques after free recall. After an initial description of our sample, we calculated the proportion of participants listing additional partners in response to each sequential cue and in response to all cues. We also calculated the proportion of recalled partners that was elicited by a specific type of cue and by all cues combined. We calculated the mean number of additional partners elicited and the mean proportional increase in number of partners elicited by each set of cues and all of them combined, among volunteers who named at least 1 additional partner during the cueing process. We calculated these measures for the study group as a whole and then for a number of demographic subgroups.
We analyzed associations between reporting any additional partners and the following interviewee variables: gender, age, race, sexual orientation, and HIV serostatus. We did that by using 5 logistic regression models. We were not able to do a more in depth analysis with other variables because of the low number of people recalling additional partners.
All data were entered by scanning using the Teleform software version 6.1 Standard (Cardiff Software, Inc., San Marcos, CA). For data management and quality control we used Microsoft Office Access 2003 and Data Access Objects/Visual Basic for Applications (Microsoft Corporation). SAS (V.9.1 Cary, NC: SAS Institute, Inc.; 2002–2003) was used to analyze the data.
During the 15-month study period, 1290 volunteers were approached for inclusion. Forty individuals (3.1%) were not eligible for the study for various reasons. Nineteen (1.5%) of the 1250 volunteers included in our research were excluded, mainly because of refusal to give blood later in the CTC (n = 13/68%). A total of 1231 individuals volunteered to participate in our study. Of these, 590 reported more than 1 regular or casual partner in the previous year and were included in the cueing analysis. Subjects were predominantly male (mostly because of exclusion of pregnant women) and heterosexual (Table 1). The median age of our study population was 28 years (ranging from 18 to 67). Reported family income ranged from R$0.00 to R$6000.00 per month, with a median of R$375.00. The median number of regular, casual, and anonymous sex partners in the previous year was 3 (ranging from 2 to 700). We found the same number when we considered only regular and casual sexual partners (ranging from 2 to 33). Only 11 volunteers (1.9%) reported more than 10 regular and casual partners in the year before the interview.
All cueing methods identified additional partners (Table 2). The overall efficiency of cueing was low with 7% of those volunteers reporting 2 or more sexual partners recalling additional partners using the supplementary technique. This increased the total number of partners recalled by about 5% in the aggregate. Among those remembering 4 or more partners during free recall, 18% recalled additional partners, yielding an 8% increase in the number of partners listed (Table 3). For those volunteers reporting 2 or more sexual partners, the mean number of partners recalled after free listing was 3.36 (SD 2.01) and this increased to 3.54 (SD 2.69) when partners recalled by cueing were added. Although the network cue was used last, it seemed to be more effective than the alphabetic and social role cues for this Brazilian population.
The efficacy of eliciting additional partners was not found to differ across most demographic categories examined. However, men were significantly less likely to report an additional partner compared with women when prompted with the social role cues (OR 0.09; 95% CI 0.02–0.42). For the combined technique, those who were less likely to report additional partners were older than those who were more likely (OR 0.95 for 1 year older; 95% CI 0.91–0.99).
In our study, we evaluated whether the methodology developed by Brewer et al. was effective in augmenting the number of recalled partners at an STI center in Brazil. Among volunteers reporting 2 or more sexual partners (n = 590), 41 (7%) recalled 1 or more additional partners by using the supplementary technique, with 105 partners of 2090 (5%) recalled only after using the cues. For volunteers reporting 4 or more sexual partners (n = 193), 34 (18%) recalled 1 or more additional partners by using the supplementary technique, and 98 of 1177 (8%) of all sexual partners were recalled after using the cues. Men were less likely than women to report sex partners after prompting with the social role cues (OR 0.09), and overall the combined techniques were slightly less effective for older individuals (OR 0.95).
Network research has been identified as pivotal to understanding the rate and extent of spread of STI/HIV and the risk to acquire and transmit an infectious agent.14,15 Social network research depends greatly on the ability of people to recall all interactions they have had in a prespecified period of time. However, recalling all sexual interactions is not an easy task for most people. Brewer made an extensive review of the issue and showed that forgetting can vary from 1% to 90% of all the people a person is asked to remember.5 Forgetting in network research can distort network measures and bias network-based conclusions.
Name generators help people to list other people with whom they have a direct interaction or relationship. Several name generators are available that can be used for different relationships.16–19 However, we were able to find only one technique that was designed to deal with sex interactions and this has only been evaluated in small studies in the United States.8,10 In one of these studies, cueing was found to enhance recall for 17% of 35 subjects with more than 1 sexual partner.10 In other study, for 5 subjects participating in a study in which 5 techniques (location, alphabetic, timeline, role, and network) were administered as a set to each participant, the mean proportional increases in the number of partners elicited was 40% for sex partners.9 In yet another study, the number of partners a participant recalled before the cues predicted the number of additional partners elicited by the cues.8
While cueing improved recall in our study as well, several factors could have contributed to the smaller effect we found. Our volunteers reported low numbers of sex partners in the prior year, and cueing is less effective in this situation, perhaps because the total number of contacts has been elicited even before starting the cues. Brewer and Garrett8 used the same recall period as we did, but their volunteers recalled on average a higher number of sex partners even before the administration of the cues. Although Brewer et al.10 reported a higher number of recalled partners, it is important to note that they used a shorter (3 month) recall period. Brewer et al.10 study was conducted in a sample that was predominantly young, heterosexual, and female. Our population was predominantly male, older, and heterosexual. Our data suggest that gender and sexual orientation had no effect on the likelihood of recall of additional partners with the combined technique. However, according to our data, the cueing technique was less likely to elicit additional partners among older volunteers. Third, although we have made some changes in the methodology for our social context, it is unlikely that we have been able to completely adapt the technique to our reality. Maybe the way our population members organize their memories differs from the same logic described by Brewer et al.9 We also may not have yet found the full range of locations and social roles that would be appropriate for Brazil. Fourth, we performed our study among volunteers seeking HIV testing. In Brazil, contact tracing is not a standard public health practice. The stigma related to HIV may have prevented volunteers from reporting more sexual partners.
Herein, we found that the cue technique was a simple way to improve sexual partner recall, mainly for individuals already reporting several partners before the cues. Those are exactly the individuals more likely to be central in HIV/STI transmission. If we had decided not to use any techniques to increase recall, we would have missed 105 additional sexual partners, 98 of them from the group reporting several partners. For network-based studies, missing partners could bias results and underestimate the potential spread of the infection in the studied populations because of the underestimation of network component sizes and cohesion.20 Although these techniques can help improve recall, they are unlike to completely eliminate forgetting and its consequences.
1. Jolly AM, Muth SQ, Wylie JL, et al. Sexual networks and sexually transmitted infections: A tale of two cities. J Urban Health 2001; 78:433–445.
2. Doherty IA, Padian NS, Marlow C, et al. Determinants and consequences of sexual networks as they affect the spread of sexually transmitted infections. J Infect Dis 2005; 191(suppl 1):S42–S54.
3. Parran T. Shadow on the Land. New York: Reynal & Hitchcock, 1937.
4. Potterat JJ. Contact tracing's price is not its value. Sex Transm Dis 1997; 24:519–521.
5. Brewer DD. Forgetting in the recall-based elicitation of personal and social networks. Soc Networks 2000; 22:29–43.
6. Brewer DD, Garrett SB, Kulasingam S. Forgetting as a cause of incomplete reporting of sexual and drug injection partners. Sex Transm Dis 1999; 26:166–176.
7. Hammer M. Explorations into the meaning of social network interview data. Soc Networks 1984; 6:341–371.
8. Brewer DD, Garrett SB. Evaluation of interviewing techniques to enhance recall of sexual and drug injection partners. Sex Transm Dis 2001; 28:666–677.
9. Brewer DD, Garrett SB, Rinaldi G. Patterns in the recall of sexual and drug injection partners. In: Levy JA, Pescosolido BA, eds. Advances in Medical Sociology: Social Networks and Health Elsevier Science, 2002:131–149.
10. Brewer DD, Potterat JJ, Muth SQ, et al. Randomized trial of supplementary interviewing techniques to enhance recall of sexual partners in contact interviews. Sex Transm Dis 2005; 32:189–193.
11. Cook RL, May S, Harrison LH, et al. High prevalence of sexually transmitted diseases in young women seeking HIV testing in Rio de Janeiro, Brazil. Sex Transm Dis 2004; 31:67–72.
12. Boletim Epidemiológico—Aids e DST. 2006, Ministério da Saúde—Secretaria de Vigilância em Saúde—Programa Nacional de DST e Aids.
13. Report on the global AIDS epidemic: Executive summary. “A UNAIDS 10th anniversary special edition”. 2006, UNAIDS: Joint United Nations Programme on HIV/AIDS: Geneva.
14. Klovdahl AS. Social networks and the spread of infectious diseases: The AIDS example. Soc Sci Med 1985; 21:1203–1216.
15. Potterat JJ, Muth SQ, Rothenberg RB, et al. Sexual network structure as an indicator of epidemic phase. Sex Transm Infect 2002; 78(suppl 1):i152–i158.
16. Campbell KE, Lee BA. Name generators in surveys of personal networks. Soc Networks 1991; 13:203–221.
17. Marin A. Are respondents more likely to list alters with certain characteristics? Implications for name generator data. Soc Networks 2004; 26:289–307.
18. Marsden P. Core discussion networks of Americans. Am Sociol Rev 1987; 52:122–131.
19. Marsden PV. Interviewer effects in measuring network size using a single name generator. Soc Networks 2003; 25:1–16.
© Copyright 2008 American Sexually Transmitted Diseases Association
20. Ghani AC, Donnelly CA, Garnett GP. Sampling biases and missing data in explorations of sexual partner networks for the spread of sexually transmitted diseases. Stat Med 1998; 17:2079–2097.