Secondary Logo

Journal Logo


People and Places: Behavioral Settings and Personal Network Characteristics as Correlates of Needle Sharing

Latkin, Carl*; Mandell, Wallace*; Vlahov, David; Oziemkowska, Maria*; Celentano, David

Author Information
Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology: November 1, 1996 - Volume 13 - Issue 3 - p 273-280
  • Free


Sharing HIV-contaminated injection equipment remains a major mode of HIV transmission in injection drug users (1). Several cross-sectional studies in the U.S. and Great Britain have examined the characteristics associated with sharing injecting equipment, finding sharing to be positively associated with unemployment, history of arrest, receiving public assistance, low levels of education, heavy drug involvement, availability of syringes, and injecting in shooting galleries (2-5). One of the few longitudinal studies on needle sharing in the U.S. found that in Seattle, Washington, needle sharing was associated with younger age, having a sex partner who was an injection drug user, and using psychoactive substances (6). Frequency of needle sharing has also been reported to be associated with perceptions of peers sharing needles and with specific social roles within the drug-using community (7,8).

Although most injection drug users are reported to possess basic knowledge of the modes of HIV transmission, many continue to share needles and continue to experience HIV seroconversion (9-11). Maintenance of high-risk drug-related behaviors and continued seroconversion suggests a need to examine social and environmental factors that may inhibit the reduction of HIV-related behaviors.

Analysis of personal networks provides one method of examining social context. The data on relationships between social context, particularly social networks, and HIV infection and risk behaviors are sparse. The research to date on social networks and HIV has effectively used network methodology to map the transmission of HIV through and among social networks (12). There are few extant studies on HIV risk behaviors and personal networks. Trotter et al. (13), in one of the few extant studies on HIV risk behaviors and personal networks, identified four types of drug sharing subnetworks based on composition and accessibility to new members: closed, kinship, long-standing friendship, and open. In this ethnographically oriented network study, closed networks reported the highest drug use and kinship networks the second highest. In New York City, Friedman and his colleagues have found that for women, higher frequency of network turnover is associated with HIV seropositive status (14). Neaigus et al. (15) found a high degree of concordance among drug-sharing dyads in reports of HIV risk behaviors.

One approach to analyzing the environmental context of needle sharing is to delineate the settings where individuals inject drugs and the relationship between settings and risk behaviors. Although settings of injection behaviors have received little attention in the literature, Barker and his colleagues have undertaken pioneering research on behavior and settings (16,17). Behavioral settings have been reliably measured, and several studies have found systematic relationships between settings and quality of life, drug treatment outcome, psychiatric symptoms, and quantity of alcohol consumed (18-22). In the studies of injection drug users, attendance in one setting—shooting galleries (defined as locations where injection equipment is available for rent)—has been associated with higher risk for HIV infection (1,23,24). In an analysis of 15 U.S. cities, Friedman and his colleagues found that the low-seroprevalence cities, HIV seroconversion was associated with self-reports of injecting in outdoor settings or abandoned buildings (25).

This study examined the relationship between needle sharing and the social and environmental context of drug behavior in a sample of inner-city injection drug users, a group that continues to be a high risk for contracting and transmitting HIV through the sharing of unhygienic injection equipment. The social context of risk behavior was assessed through analysis of respondents' personal networks; the environmental context of behavior was defined as the settings where individuals inject. Sharing of injecting equipment was examined in two sets of analyses. Individuals who reported sharing injecting equipment after cleaning it with bleach were compared with those who reported not sharing needles. A second comparison was made between those who reported sharing injection equipment without first cleaning it with bleach and those who reported never sharing needles.



Participants were a nontreatment sample recruited from the AIDS Links to Intravenous Experiences (ALIVE) study, a prospective study of the natural history of HIV infection in injection drug users in Baltimore, Maryland (26). The primary means of recruitment for the ALIVE study were community outreach and word of mouth. Every six months, participants in the ALIVE study were reinterviewed, and those who were seronegative were rescreened for HIV. Antibody to HIV-1 was screened by licensed enzyme immunoassay (Genetic Systems, Seattle, WA, U.S.A.) with confirmatory Western blot (DuPont, Wilmington, DE, U.S.A.), using standard procedures. At the ALIVE clinic, participants were asked to participate in the Stop AIDS For Everyone (SAFE) study if they had reported at their regular 6-month followup visit that they had injected drugs in the preceding 6 months and had shared drugs. Eligibility requirements included attaining a minimum age of 18 and injecting and sharing drugs within the prior 6 months. Because of the concern that some injectors may not readily report sharing needles, the broader criterion of sharing drugs was chosen. At the SAFE clinic, all eligible participants were administered a detailed interview on their socio-demographic background, HIV-related behaviors, and names of persons in their personal networks. The data for the present study were collected in 1991 and 1992. Participants were randomly assigned to either a control of an experimental condition. The experimental condition was a cognitive-behavioral HIV/AIDS preventive intervention. Individuals who were assigned to the intervention condition were considered eligible to participate if they could bring at least two individuals with whom they had shared drugs within the prior 6 months to the clinic for the intervention. These individuals were called “drug network” members and were also interviewed.

Both control subjects and experimental subjects in the SAFE study were interviewed every 6 months as part of the ALIVE study protocol. Those who were HIV seronegative were given HIV antibody testing and counseling, and HIV seropositive participants were provided sura samples for assays of immune function. The SAFE study control subjects did not receive any additional education. After completion of the study, the control subjects were offered the experimental intervention.

Of the 369 individuals who met the criteria for enrollment, 89% (330) completed the followup interview; 58% (193) of the 330 were ALIVE participants. The other SAFE participants were recruited by the ALIVE participants. The mean length of time between the two interviews was 5.2 months (SD = 1.6), and the median interval was 5 months.

Dependent Variables

At baseline, respondents were queried about their HIV-related behaviors, including frequency of sharing needles. At followup, participants were asked about the frequency of sharing needles that they had first cleaned with bleach and the frequency of sharing needles without prior cleaning with bleach. The survey included the following two questions: (a) “During the last 6 months, how often did you use a needle right after somebody else had used it but you first cleaned it with bleach?” (b) “During the last 6 months, how often did you use a needle right after somebody else had used it without first cleaning it with bleach?”

Network Instrument

According to a significant body of research, personal network characteristics can be reliably measured (27-29). Methodological studies indicate that individuals are more accurate in recollecting the people with whom they typically interact than in reporting events of a specific time frame of interaction. Hammer (30) reports that the names of network members recalled are more likely to be frequent, intense, and recent contacts. Barrera (31) reports test-retest reliability of 74% for recall of social contacts in the past month and 80% for persons whom respondents reported they typically encountered.

In the baseline interview, all study participants delineated their personal networks. Using a network instrument based on the work of Barrera (31-32), the inventory asked participants to list, by giving names or pseudonyms, members of their personal support network whom they had known for at least 1 month. Respondents were asked to list individuals whom they felt could provide support in each of six domains: material assistance, socializing, intimate interaction, physical assistance, health information, and positive feedback. Together, these six network domains were defined as the individuals' personal social support network. After naming the members of their support network, participants were asked to list individuals with whom they “shared drugs” in the last 6 months. The network's two structural characteristics, density and multiplexity, were assessed. Density is defined as the actual number of network ties among individuals in a personal network, divided by the number of possible ties. Respondents were given a matrix with the names of their network members shown on both axes, and then were asked to indicate the members of their personal network who knew each other. Density thus represents the percentage of network members who interact with each other. Multiplexity is defined as the number of functional relationships between the participant and the network members, and was indexed in this study by the number of network members named in two or more support domains (e.g., material aid and physical assistance).

Behavioral Settings

These were particular places where individuals inject drugs. A list of injection settings was first derived from open-ended questions. The subsequent selection of categories for the structured survey was based on the frequency of settings named in the open-ended questions. In the baseline interview, participants were asked whether they had injected in the following places within the last 6 months: (a) “At your own place” (b) “At a friend's place” (c) “In a shooting gallery, that is, a place where people go to rent, borrow or buy a needle” (d) “Mother's place” (e) semipublic areas: “On the street, in a park, in a car, on rooftops, public bathrooms, or an abandoned building.”


Two principle sets of comparisons were made: (a) participants who had reported never sharing a needle in the prior 6 months were compared with participants who reported, at least once in the prior 6 months, using a needle right after it had been used by somebody else after cleaning it with bleach (i.e., sharing with cleaning); and (b) participants who had reported never sharing a needle in the prior 6 months were compared with participants who reported, at least once in the prior 6 months, using a needle right after it had been used by somebody else without first cleaning it with bleach (i.e., sharing without cleaning).

The bivariate associations of personal network characteristics and injection settings at baseline and sharing needles at followup were first examined. In these analyses, chi-square tests were used to examine the relationship between injection settings and sharing needles within the prior six months. Student's t tests were also calculated to assess differences in size of support networks. The subsequent analyses used multiple logistic regression to model the independent contribution of each network characteristic in predicting needle sharing at followup. In the final model, demographic and background characteristics, and prior reports of sharing needles, were added to the multiple logistic regression model. In an exploratory analysis, needle sharing was treated as a linear variable, and an ordinary least squares linear multiple regression model was used.


The 330 respondents were predominately unemployed (88%) inner-city African-American (98%) men (81%). At baseline, 46% of the respondents reported that they had been homeless and 85% reported being on public assistance at least once in the previous 10 years. The mean and median year of birth was 1954 (SD = 6.4). The majority (57%) of the participants were at least daily injectors of heroin or cocaine. At baseline, 71% of the participants reported sharing needles without first cleaning them with bleach or alcohol. The mean size of drug network was 5.0 individuals (SD = 3.0, median = 5), and the mean size of the support network (independent of the drug network) was 6.9 individuals (SD = 2.9, median = 7). The mean overlap between the drug network and the social participation network was 2.1 (SD = 1.7, median = 2).

We have previously reported on the intervention results (33). At followup there were significant differences between participants in the experimental and control conditions on self-reported needle sharing. Participants assigned to the experimental condition reported greater reduction in needle sharing with and without cleaning with bleach. On the followup survey, 64.7% reported that they had shared a needle after cleaning it with bleach within the prior 6 months, with 41% reporting that they shared more than once a month in that period. Over onethird (38.5%) of the participants reported sharing needles without cleaning within the prior 6 months, with 17% reporting that they shared without cleaning more than once a month in that period. For the subsequent analyses, participants were categorized into three mutually exclusive categories based on reports of needle sharing in the prior 6 months: (a) not sharing, (b) sharing after cleaning needles with bleach, and (c) sharing without cleaning needles with bleach.

The first univariate analysis examined the difference in the size of between personal network characteristics at baseline and sharing at followup. Participants who reported that they had not shared needles were compared with those who reported sharing needles with cleaning and those who reported sharing needles without cleaning. As seen in Table 1, participants who reported sharing needles after cleaning had significantly larger drug networks and denser networks than participants who reported that they did not share needles. Participants who reported that they shared needles without cleaning had significantly larger drug networks than individuals who reported not sharing needles.

The next set of analyses examined the relationship between sharing and injection settings. Overall, 92% of the participants reported injecting at their “own place,” 86% at a “friend's place,” 27% in a “shooting gallery,” and 30% at “mother's place”; and 35% reported injecting in one or more of the following semipublic places: “the street, a rooftop, a park, a car, a public bathroom, or an abandoned building.” Participants who reported sharing needles with cleaning in the prior 6 months were significantly more likely than participants who reported not sharing needles to state that they had injected at a friend's residence (Table 2). Respondents who reported sharing needles without cleaning in the prior 6 months, compared with respondents who reported that they had not shared in the prior 6 months, were significantly more likely to state that they injected in a semipublic place.

Multiple logistic regression models were used to examine correlates of sharing needles after adjusting for baseline reports of frequency of sharing needles; the demographic variables of age, gender, education; the background variables of frequency of drug use (injecting cocaine and/or heroin at least once a day vs. less than once a day); and receiving public assistance, experiencing homelessness, and history of arrest within the prior 10 years.

After controlling for baseline levels of needle sharing and for demographic characteristics and history of arrest, participants who reported sharing needles after cleaning them with bleach, compared with those who reported that they had not shared needles in the prior 6 months, continued to show significantly larger drug networks and denser personal networks (Table 3). Reports of injecting at a friend's residence remained associated with sharing after cleaning needles. As seen in Table 3, injecting at friend's residence increased the odds of sharing with cleaning by a factor of 3 (odds ratio [OR] = 3.2). An increase in the drug network by one individual was associated with a 19% increase in the probability of reporting sharing needles after cleaning, and an increase in density by 1% was associated with a 4.3 increase in the odds of sharing needles after cleaning.

After controlling for baseline levels of needle sharing, demographic characteristics, and history of arrest, sharing needles without cleaning with bleach continued to be associated with injecting in one or more of the following semipublic settings: on the street, on a rooftop, in a park, in a car, in a public bathroom, or in an abandoned building (Table 3). Participants who reported injecting in these semipublic areas were almost twice as likely to report sharing needles without cleaning.

In the next set of analyses, the relationship between HIV status and sharing was examined. From serum assays for HIV status available for 193 (57%) of the 330 participants, 52 (27%) of the participants tested were HIV seropositive. No significant statistical association arose between sharing needles after cleaning and HIV status (chi-square = 2.22, n = 193, p =.14) or sharing needles without cleaning and HIV status (chi-square = 0.14, n = 193, p =.70). Neither the addition of HIV status nor group assignment, i.e., experimental or control, to the multiple logistic regression model significantly altered the association between needle sharing and injection settings or needle sharing and personal network characteristics.

In simultaneous entry of the independent variables presented in Table 3, using a linear multiple regression model, the dependent variable of frequency of sharing a needle after cleaning was found to be significantly associated with size of drug network (t = 3.11, p < 0.01) and total network density (t = 2.57, p < 0.02). The multiple R was 0.32. In a multiple regression analysis with frequency of sharing a needle without cleaning as the dependent variable, injecting in semipublic areas continued to be associated with sharing without cleaning (t = 2.09, p < 0.05). The multiple R for this regression model was 0.31.


The data from this study suggest that social and environmental context variables of personal network characteristics and injection settings are prospectively associated with drug-related HIV risk behaviors of sharing needles that have been cleaned with bleach and sharing needles without cleaning with bleach. The baseline network characteristics of size of drug network, total network density, and injecting at a friend's residence were correlated at followup with the item, sharing needles after cleaning with bleach. Reports at baseline of injecting in semipublic settings were found to be associated at followup with sharing needles without cleaning.

The association between needle sharing without cleaning and prior reports of injecting in semipublic areas may be an index of level of privacy and control over injection settings. The tendency to inject in semipublic areas may also be a result of purchasing drugs far away from one's residence. In these settings, individuals may have difficulties in acquiring and storing bleach in these areas and may rush to inject because they fear intrusion by the police or others. Larger drug networks at baseline may indicate greater frequency of injecting with others and greater involvement in a drug-using lifestyle. Consequently, these individuals evidence a higher probability of sharing of needles. Table 4

Another explanation for the relationship between baseline drug network size and needle sharing with cleaning is that injection drug users who have larger drug networks are more likely to have individuals in their networks who share needles and that they may place social pressures on their drug network members to share needles with them. Other investigators have found that reports of friends' HIV-related behaviors are associated with self-report of the risk behaviors (7) and that peer influence is associated with drug-related behaviors (34). The relationship between greater network density at baseline and sharing needles after cleaning at followup may indicate the frequency of interactions between social support network and drug network. If these two networks have frequent interactions, it may be difficult for injection drug users to remove themselves from their drug network when they inject. These high-risk dense networks may be similar to the category of closed networks discussed by Trotter et al (13), which demonstrated high levels of drug use. The relationship between network density and sharing needles after cleaning indicates that structural characteristics of personal networks may be important factors in understanding the social dynamics of HIV-related behaviors, and it demonstrates the utility of examining structural characteristics of the social environment.

Since the sample comprised volunteers from an HIV prevention study who were recruited by word of mouth, it is plausible that these volunteers, some of whom knew one another, differed from those who did not volunteer, did not meet the criteria for enrollment, withdrew from the study, or were not reinterviewed at followup. Thus, the generalizability of these findings may be limited. A further limitation may be presented by the bias inherent in self-report data; however, as the data on the independent and dependent variables were collected in two different interviews, the biases that are associated with cross-sectional surveys were limited. A third study limitation exists in that the injection settings and the network characteristics of study participants may have differed from those individuals who were not eligible or were not recruited for the study. The ability of participants to report on needle sharing constitutes another limitation. In this study, the definition of needle sharing is biased toward group injection settings, where individuals are more likely to be aware that they are sharing. The definition does not take into account individuals unknowingly injecting with contaminated syringes (35).

Several plausible explanations exist for the lack of a relationship between self-reports of injecting in shooting galleries at baseline and sharing needles at followup. Variance among injection drug users in their definition of a shooting gallery may partially account for this negative finding. Alternatively, as educational materials emphasizing the dangers in injecting in shooting galleries have reached injection drug users, there may arise a social desirability bias to underreport injecting in shooting galleries (36).

Irrespective of HIV status, over half (60%) of the respondents reported that they had shared needles in the prior 6 months, and one-third reported that they had shared without cleaning their needles. Although this sample was recruited by criteria of sharing drugs, reports of continued needle sharing have been documented in other nontreatment samples, including the ALIVE from which we recruited the present sample (10,37). Clearly many of these injection drug users continue to be at high risk for HIV infection and transmission of HIV. Even those injection drug users who reported cleaning needles with bleach continue to have high rates of seroconversion (38). It is not known whether injection drug users who report that they clean their needles with bleach are cleaning for the recommended amount of time (39,40). Because of their high risk of HIV infection, injection drug users who share needles should be a high priority for preventive interventions.

The results of this study suggest several methods of reducing HIV-related injecting behaviors in injection drug users. One approach is to promote strategies of reducing the size of individuals' drug networks. It may be unrealistic to expect most inner-city injection drug users to remove themselves completely from drug environments; however, counseling injection drug users on reducing the size of their drug networks may be a viable strategy for HIV risk reduction. Intervention programs should also address the composition of individuals' networks. For example, the data from the present study suggest that individuals with less dense networks may not have the same level of risk as those with denser networks. It is important to note that although size of the drug network was associated with less sharing with bleaching, it was not associated with sharing needles without bleaching.

Sharing with disinfection was associated with injecting in semipublic areas, where it is unlikely that bleach is readily available. In such places, outreach workers could leave a supply of bleach. Needle exchange programs and legalization of possession of drug paraphernalia may help to reduce the spread of HIV in those individuals who inject in these and other settings.

Acknowledgment: This research was supported by grants (DA04334, DA05911, & DA08985) from the National Institute on Drug Abuse. We wish to thank Amy Knowlton for her helpful comments on this manuscript.


1. Schoembaum EE, Hartel D, Selwyn PA, et al. Risk factors for HIV-1 infection in intravenous drug abusers. N Engl J Med 1989;321:874-9.
2. Banard, MA. Needle sharing in context: patterns of sharing among men and women injectors and HIV risks. Addiction 1993;88:805-12.
3. Gibson DR, Choi KH, Catania JA, Sorensen JL, Kegeles S. Psychosocial predictors of needle sharing among intravenous drug users. Int J Addictions 1993;28:973-81.
4. Magura S, Grossman JI, Lipton DS, et al. Determinants of needle sharing among intravenous drug users. Am J Public Health 1989;79:459-62.
5. Mandell W, Vlahov D, Cohn S, Latkin CA, Oziemkowska M. Correlates of needlle sharing among intravenous drug users. Am J Public Health 1994;84:920-3.
6. Saxon AJ, Calsyn DA, Jackson TR. Longitudinal changes in injection behaviors in a cohort of injection drug users. Addiction 1994;89:191-202.
7. Friedman SR, Des Jarlais DC, Sotheran JL, Garber J, Cohen H, Smith D. AIDS and self-organization among intravenous drug users. Int J Addictions 1987;23:201-19.
8. Johnson J, Williams ML. A preliminary ethnographic decision tree model of injection drug users' (IDUs) needle sharing. Int J Addictions 1993;28:997-1014.
9. Celentano DD, Vlahov D, Cohn S, Anthony JC. Risk factors for shooting gallery use and cessation among intravenous drug users. Am J Public Health 1991;81:1291-5.
10. Celentano DD, Munoz A, Cohn S, Nelson KE, Vlahov D. Drug-related behavior change for HIV transmission among drug users. Addiction 1994;89:1309-17.
11. Rosenberg PS, Levy ME, Brundage JF, et al. Population-based monitoring of an urban HIV/AIDS epidemic: magnitude and trends in the District of Columbia. JAMA 1992;268:495-503.
12. Klovdahl AS. Social networks and the spread of infectious diseases: the AIDS example. Soc Sci Med 1985;21:1203-1216.
13. Trotter RT, Bowen AM, Potter JM. Network models for HIV outreach and prevention programs for drug users. In: Needle R, Coyle S, Genser S, Trotter R, eds. NIDA technical review on social networks, drug abuse, and HIV transmission. Bethesda, August 1993.
14. Friedman SR, Jose B, Neaigus A, Goldstein M, Curtis R, Des Jarlais D. Female injecting drug users get infected with HIV sooner than males. Annual meeting of the American Public Health Association, San Francisco, 1993, NIH pub 95-3889, p. 144-81.
15. Neaigus A, Friedman SR, Curtis R, et al. The relevance of drug injectors' social and risk networks for understanding and preventing HIV infection. Soc Sci Med 1994;38:67-78.
16. Barker RG. Habits, environments, and human behavior. San Francisco: Jossey-Bass, 1978.
17. Barker RG, Wright HF. Midwest and its children. New York: Harper & Row, 1955.
18. Klassen D. Person, setting, and outcome in a drug abuse treatment program. Psychiatr Ann 1977;7:419-32.
19. Moos RH. Context and coping: toward a unifying conceptual framework. Am J Comm Psychol 1984;12:1-36.
20. Perkins DV. Individual differences and task structure in the performance of a behavior setting: an experimental evaluation of Barker's manning theory. Am J Comm Psychol 1982;10:617-34.
21. Schoggen P. Behaviroal settings and the quality of life. Am J Comm Psychol 1983;11:144-57.
22. Wigmore SW, Hinson RE. The influence of setting on consumption in the balanced placebo design. Br J Addiction 1991;86:205-15.
23. Page BJ, Smith PC, Kane N. Shooting galleries, their proprietors and implications for prevention of AIDS, in AIDS and alcohol/drug abuse: psychosocial research. Harrington Park Press, New York, 1991.
24. Vlahov D, Munoz A, Anthony JC, Cohn S, Celentano DD, Nelson KE. Association of drug injection patterns with antibody to human immunodeficiency virus type 1 among intravenous drug users in Baltimore, Maryland. Am J Epidemiol 1990;132:847-55.
25. Friedman SR, Jose B, Deren S, Jarlais DCD, Neaigus A. Risk factors for human immunodeficiency virus seroconversion among out-of-treatment drug injectors in high and low seroprevalence cities. Am J Epidemiol 1995;142(8):864-74.
26. Vlahov D, Anthony JC, Celentano DD, Solomon L, Chowdhury N. Trends of HIV-1 risk reduction among initiates into intravenous drug use 1982-1987. Am J Drug Alcohol Abuse 1991;17:39-48.
27. Jennings KD, Stagg V, Pallay A. Assessing support networks: stability and evidence for convergent and divergent validity. Am J Comm Psychol 1988;16:793-809.
28. Romney AK, Weller SC. Predicting informant accuracy from patterns of recall among individuals. Social Networks 1984;6:59-77.
29. Tardy CH. Social support measures. Am J Comm Psychol 1985;13:187-202.
30. Hammer M. Some comments on the validity of network data. Connections 1990;3:13-15.
31. Barrera M. A method for assessing social support networks in community survey research. Connections 1980;3:8-13.
32. Sandler IN, Barrera M. Towards a multimethod approach to assessing the effects of social support. Am J Comm Psychol 1984;12:37-52.
33. Latkin CA. Using social network methodology to study HIV risk behaviors in injecting drug users and to design a preventive intervention. In: Needle R, Coyle S, Genser S, Trotter R, eds. Social Networks, Drug Abuse, and HIV 1995, NIH pub. 95-3889, p. 181-95.
34. Elliot DS, Huizinga D, Ageton SS. Explaining deliquency and drug use. Beverly Hills: Sage, 1985.
35. Page BJ. Shooting scenarios and risk of HIV-1 infection. Am Behavioral Scientist 1990;33:478-90.
36. Latkin CA, Vlahov D, Anthony JC. Socially desirable responding and self-reported HIV infection risk behaviors among intravenous drug users. Addiction 1993;88:517-26.
37. McCusker J, Stoddard AM, Koblin BA, Sullivan J, Lewis BF, Sereti SM. Time trends in high-risk injection practices in a mutli-site study in Massachusetts: effects of enrollment site and residence. AIDS Education Prevention 1992;4:108-19.
38. Vlahov D, Munoz A, Celentano DD, et al. HIV seroconversion and disinfection of injection equipment among intravenous drug users in Baltimore. Epidemiology 1991;2:444-6.
39. McCoy CB, Rivers JE, McCoy HV, et al. Compliance to bleach disinfection protocols among injecting drug users in Miami. J Acquir Immun Defic Syndr 1994;7:773-6.
40. National Institute of Drug Abuse. Community Alert Bulletin, March 1993.

HIV; Social networks; Injection drug users; Needle sharing; Behavioral settings

© Lippincott-Raven Publishers.