Trends in HIV Seroprevalence and Needle Sharing Among Puerto Rican Drug Injectors in Puerto Rico and New York: 1992-1999

This study assessed trends in HIV seroprevalence and needle-sharing behaviors among Puerto Rican injection drug users (IDUs) in Puerto Rico and New York. Data from two studies of IDUs conducted from 1992 through 1995 and 1998 through 1999 in Bayamón, Puerto Rico, and East Harlem, New York, were examined to assess trends over this period. Separate analyses were conducted for IDUs who were current crack smokers. Significant decreasing trends in seroprevalence were found among IDUs in the New York and Puerto Rico samples (p < .001). Significant decreasing trends in receptive and distributive needle sharing were found in the New York sample, and a significant decline in receptive sharing was found in the Puerto Rico sample. Overall, higher levels of needle-sharing behaviors were reported in Puerto Rico compared with New York. Decreasing trends in needle sharing and seroprevalence in both communities are an encouraging finding. Ongoing epidemiologic studies to monitor the epidemic and continued prevention efforts to help maintain or further these declines are needed, particularly to address the higher rates of needle sharing in Puerto Rico.

Puerto Rican injection drug users (IDUs) are at high risk for HIV ^(1-3). Studies of Puerto Rican IDUs in Puerto Rico and New York City from the late 1980s through early 1990s have found high HIV seroprevalence levels. Seroprevalence for IDUs in New York City has declined from a high of approximately 50% in the late 1980s to approximately 25% by 1996 ⁽⁴⁾. This decline has been reported for Hispanics (primarily Puerto Ricans) in New York City as well as for other ethnic groups ⁽⁵⁾. A seroprevalence among IDUs in Puerto Rico of approximately 45% was reported in data collected in the late 1980s ⁽⁶⁾; however, no reports of trends in seroprevalence among IDUs in Puerto Rico since that time have been published. Needle-sharing behaviors have been associated with HIV among IDUs since early in the epidemic ⁽⁷⁾, and there is some evidence that these behaviors have been declining in New York ⁽⁸⁾, although no trend data comparing Puerto Rican IDUs in New York City and Puerto Rico have been published.

Many changes that have an impact on injection-related risks in both communities have occurred since the late 1980s. These include the institution of legal needle exchange in New York in 1992 ⁽⁹⁾ and in Puerto Rico in 1995 ⁽¹⁰⁾ as well as the presence of outreach intervention projects. It is difficult to assess the impact of any single intervention over time in environments where multiple educational/prevention efforts are underway. In addition, the impact of the large number of AIDS-related illnesses and deaths on communities in general and on those individuals who are themselves at high risk of being infected is also difficult to determine. Nevertheless, it is important to examine changes over time in both seroprevalence and needle sharing so as to assess the possible cumulative effects of multiple prevention efforts. This article, drawing from research efforts in East Harlem, New York, and Bayamón, Puerto Rico, conducted from 1992 through 1995 and from 1998 through 1999 was undertaken to assess changes in HIV seroprevalence and needle-sharing behaviors among Puerto Rican IDUs. The higher rates of risk behaviors among Puerto Ricans who reside in Puerto Rico as compared with those who reside in New York have also been reported ^(11-14), and this study also assesses whether similar trends in seroprevalence occur in different risk behavior environments. These data thus can extend the findings in New York in terms of assessing whether the downward trend in seroprevalence has continued for Puerto Rican IDUs through the end of the 1990s, provide data on whether a similar trend has occurred in Puerto Rico during the 1990s, and provide information on trends in needle-sharing behaviors.

METHODS

Cooperative Agreement Project

East Harlem, New York, and Bayamón, Puerto Rico, were both participating sites in the cooperative agreement project funded by the National Institute on Drug Abuse to recruit IDUs and crack smokers for an intervention study from 1992 through 1995 (principal investigators: S. Deren and R. Robles). This article focuses only on those subjects who reported injection drug use during the 30 days before being interviewed.

Recruitment

Subjects were recruited using targeted sampling ⁽¹⁵⁾ from drug copping areas and other areas where drug users could be found. Criteria for recruitment included: (1) age of 18 years or older; (2) report of injection drug use in the prior 30 days; (3) recent use of heroin or cocaine as verified by urinalysis; and (4) not in drug treatment for the prior 30 days. After recruitment, subjects were brought to the field site at each location, where recent drug use was verified (using Roche ONTRAK, Indianapolis, IN), and informed consent was obtained before conducting a baseline interview. Subjects were compensated $15 in New York and $20 in Puerto Rico for participating in the interview. Recruitment occurred between 1992 and 1995 in both New York and Puerto Rico. Recruitment of individuals occurred throughout the East Harlem community in New York. In Puerto Rico, however, the copping areas in Bayamón where recruitment occurred were matched in pairs (according to geographic location and availability of crack, needles, and galleries), and pairs were selected for recruitment, after which individuals were randomly selected and assigned to either the standard or enhanced intervention ⁽¹⁶⁾.

Interviewing and HIV Testing

The primary data collection instrument was the risk behavior assessment (RBA) questionnaire. The RBA questionnaire was developed for the study by participating investigators in collaboration with investigators at the National Institute on Drug Abuse and has been reported to be reliable and valid ^(17,18). The RBA questionnaire focuses primarily on risk behavior data for the prior 30 days. After the baseline interview, HIV counseling and testing were provided. Oral testing (using Orasure; Epitope Inc., Beaverton, OR) was used in New York, and venipuncture was used in Puerto Rico. Subjects were subsequently assigned to standard and enhanced interventions ^(16,19), and 6-month follow-up interviews were conducted. HIV testing rates were approximately 54% in New York (those who reported having recently been tested or who refused testing were not tested in the project) and 96% in Puerto Rico.

Alliance for Research in El Barrio and Bayamón Project

The Alliance for Research in El Barrio and Bayamón (ARIBBA) project conducted recruitment in East Harlem, New York, and Bayamón, Puerto Rico, to study IDUs and crack smokers and the determinants of HIV risk behaviors at the two sites. This article focuses only on those who reported injection drug use in the 30 days before being interviewed.

Recruitment

Each of the two communities was divided into recruitment sectors (3 in Bayamón and 5 contiguous sectors in East Harlem) based on ethnographic mapping ⁽²⁰⁾ to ensure that recruitment occurred from a range of drug copping and using locations. These methods were similar to the targeted sampling used in the cooperative agreement project. Criteria for recruitment included: (1) age of 18 years or older; (2) self-report as Puerto Rican; and (3) injected drugs within the prior 30 days. As in the cooperative agreement project, verification of recent use of heroin or cocaine by urinalysis (using Roche ONTRAK) was required. Recruitment of subjects occurred between January 1998 and July 1999. After recruitment in the community, subjects were brought to the field site at each location, where recent drug use was verified, informed consent was obtained, and the baseline interview was conducted. Subjects were compensated $15 in New York and $20 in Puerto Rico for participating in the interview.

Interviewing and HIV Testing

The interview, based on revisions and an expansion of the RBA questionnaire, included sections on risk behaviors during the prior 30 days. After the interview, subjects received HIV pretest counseling and were offered HIV testing (using Orasure). Baseline HIV testing rates were approximately 95% at both sites.

Analyses Conducted

In both studies, although data on risk behaviors were based on the prior 30 days, there were important differences in how specific questions were asked, which precluded extensive direct comparisons across the two data sets. The specific questions regarding needle-sharing behaviors, however, were comparable across the two studies. At each site, analyses of trends were conducted for seroprevalence and needle-sharing behaviors, both distributive (i.e., gave needles to others after using them on self) and receptive (i.e., used needles after others had used them).

All analyses used only baseline data and included only those subjects who self-reported as being Puerto Rican. For comparability with the cooperative agreement data set, only those ARIBBA project participants who reported not being in drug treatment (during the prior 30 days) were selected for inclusion. Subjects were categorized by their recruitment year (i.e., 1992-1993, 1994-1995 [both from the cooperative agreement data set] and 1998-1999 [from the ARIBBA project]). Because some prior studies indicated that crack-using IDUs may have different seroprevalence levels than those not using crack (e.g., in one study of IDUs, crack use was associated with lower risk for HIV infection despite being associated with high levels of HIV risk behaviors ^[21]), seroprevalence analyses were conducted separately for injectors who reported crack use during the prior 30 days.

RESULTS

Analysis of sociodemographic characteristics and years of injection of the sample populations recruited during the three periods indicates some significant differences over time (Table 1) (data on years of injection were not available during the first period). In the New York sample, the percentage of the sample population that was male increased over time (79% to 89%;p < .01) as did the percentage reporting being homeless (21% to 30%;p < .05). In the sample from Puerto Rico, age significantly declined (34.2 to 32.7 years;p < .01) as did years of injection (14.4 to 11.6 years from time 2 to time 3;p < .001) and income (p < .001); in addition, there was an increase in the percentage who were homeless (12% to 24%;p < .001) and the percentage who were at least high school graduates (36% to 46%;p < .01). Data on place of birth (not shown) for the 1998 through 1999 cohorts indicated that in the New York sample, 57% were born in Puerto Rico and 39% in New York; for the sample recruited in Puerto Rico, the figures were 87% and 10%, respectively (data on place of birth were not collected for the earlier cohorts).

Table 1

A significant decline in HIV seroprevalence among IDUs in New York from 1992 through 1999 was found: 50% in 1992 through 1993 to 21% in 1998 through 1999 (χ² = 28.7;p < .001). This trend was significant for both subgroups-those who recently used or did not use crack (Table 2). In Puerto Rico, although there was an overall decline in HIV among IDUs from 46% to 21% (χ² = 46.9;p < .001), this was primarily the result of a decline among IDUs who were not also current crack users. Because of the differential age distributions in each period at each site, we grouped subjects by age categories and conducted multivariate analyses (multiple logistic regression), controlling for gender, homelessness, education, and income and comparing the sites for seroprevalence. There was no evidence of a differential decrease over time in seroprevalence at the two sites (i.e., there were similar significant declines at both sites).

Table 2

An examination of needle use risk behaviors for all IDUs (Table 3) indicated that significant declines occurred in New York City in both receptive and distributive needle sharing. Whereas 26% of subjects reported receptive sharing in 1992 through 1993, this declined to 9% in 1998 through 1999 (χ² = 19.20;p < .001). Distributive needle sharing declined among the New York City IDUs from 30% to 8% (χ² = 22.2;p < .001). In Puerto Rico, whereas receptive sharing declined from 39% to 20% (χ² = 20.4;p < .001), there was no significant decline in distributive needle sharing. In bivariate analysis, levels of receptive sharing and distributive sharing measures were significantly higher in Puerto Rico than in New York City for all time periods combined (36% vs. 18% and 37% vs. 13%, respectively; both at p < .001).

Table 3

CONCLUSIONS

Significant declines in seroprevalence were found among out-of-treatment IDUs at both locations, and similar levels of HIV prevalence (approximately 21%) were found at both locations in 1998 through 1999. Only IDUs in Puerto Rico who were also crack smokers did not seem to evidence a significant decline since the 1992 through 1993 period. Further research regarding this finding is needed; it may be related to increased risks for crack smokers in Puerto Rico through unprotected sexual activities. The crack epidemic emerged in Puerto Rico later than in the New York area ⁽²²⁾, and this may account for its possible influence in the maintenance of higher seroprevalence levels among crack-using IDUs throughout the study period ^[23]). Interventions with crack-smoking IDUs may be particularly important at this point in the epidemic in Puerto Rico. It is also interesting to note that the IDUs who smoked crack evidenced lower seroprevalence than other IDUs during the first two periods for both sites and higher seroprevalence during the last period. This may be related to the increasing importance of heterosexual transmission in the epidemic, which is another indication that IDUs who also smoke crack may need targeted intervention efforts.

The comparison of trends in needle-sharing behaviors indicates that declines have occurred in receptive needle sharing among Puerto Rican IDUs at both locations. Distributive needle sharing did not show a decline over the three periods in Puerto Rico. Further examination of needle-sharing behaviors, including assessment of their relation to serostatus, is being undertaken to assess the extent to which HIV transmission may be occurring.

Higher levels of needle-sharing behaviors were reported in Puerto Rico compared with New York. As noted previously, the ARIBBA project was developed to identify determinants of risk behaviors, including individual, social, and environmental factors. Preliminary analyses of a wide range of risk behaviors ⁽¹⁴⁾ as well as prior data collected at the two sites ^(11,12) have shown that risk behaviors are consistently higher in Puerto Rico; this may indicate that issues in the environment (in terms of availability of mechanisms to reduce risk) may be important factors influencing risk levels. Analyses of ARIBBA project data have found that variables having an impact on risk behaviors may include differences in needle exchange availability and use ⁽²⁴⁾ as well as differences in availability of methadone treatment ⁽²⁵⁾. Multivariate models are needed to incorporate individual as well as community variables to help understand the relative importance of multiple influences on risk behaviors.

In view of the higher risk behaviors in Puerto Rico, the similarity in seroprevalence at the two sites in 1998 through 1999 raises a question as to why seroprevalence is not higher in Puerto Rico. It may be that high mortality in Puerto Rico (H.M. Colón, Ph.D., personal communication, 1999) or factors related to migration of seropositive subjects from Puerto Rico to New York may account for this apparent anomaly. This is an important area for further analyses and research. The presence of higher risk behaviors in Puerto Rico indicates that unless additional efforts to address the epidemic and reduce risk behaviors are instituted, increases in seroincidence and a return to higher seroprevalence levels may result.

Several limitations of these data must be identified: the risk behavior data were based on self-report, and although targeted sampling helped to ensure that recruitment occurred throughout both communities, it is not possible to obtain a random sample of street-based drug users. The lower testing rate in the cooperative agreement data set from New York may also reduce the comparability of trend comparisons. Nevertheless, Ziek et al. ⁽²⁶⁾ found few differences in risk behaviors between those tested and those not tested (with the primary reasons for not testing including fear of testing and having recently been tested). In addition, other New York City data indicate that these cooperative agreement trends are consistent with other local seroprevalence trends during a similar period ⁽⁴⁾. Furthermore, three time points provide a small number of measures, and additional measurement points may provide greater confidence in detecting overall trends. Despite the limitations, some important implications can be drawn from these data.

The trends showing declining seroprevalence in both communities are encouraging. Many factors are likely to have contributed to these declines, including mortality among seropositive individuals as well as declining levels of risk behavior caused in part by the array of prevention efforts mounted against this epidemic. This finding, however, does not indicate that prevention efforts should be curtailed. Research on other high-risk populations has shown that return to high-risk behaviors after a period of decline in risk can occur ⁽²⁷⁾. Continued and increased efforts are needed to help maintain and continue declines in risk behaviors and seroprevalence. In addition, prevention efforts to target those behaviors and populations that do not seem to exhibit decreasing trends (e.g., IDUs who also use crack in Puerto Rico) are needed. Within a period of decline, continued monitoring of seroprevalence and risk behaviors is especially important so that any outbreaks or increases in seroprevalence in subgroups can quickly be identified and addressed. This is particularly important in communities that have had large HIV epidemics such as Puerto Rico and New York City; current seroprevalence levels of approximately 21% among IDUs at both locations and continued needle-sharing behaviors (albeit at reduced levels) indicate the potential for increased seroconversions and seroprevalence unless preventive efforts are maintained. These findings also indicate the importance of conducting multisite longitudinal analyses to assess trends and identify locations where interventions may be most needed.

Acknowledgments:

This research was supported by National Institute on Drug Abuse grants U01DA07286, R01DA10425, and P30DA11041. The authors thank Mark Beardsley, Rh.D., for his assistance in the preparation of the data sets and the anonymous journal reviewers for their helpful comments.