Proceedings--Workshop on Needle Exchange and Bleach Distribution Programs (1994)

PETER LURIE, JAMES G. KAHN, BENJAMIN BOWSER, and DONNA CHEN, JILL FOLEY, JOSEPH GUYDISH, T. STEPHEN JONES (Centers for Disease Control and Prevention), SANDRA LANE, ARTHUR L. REINGOLD, and JAMES SORENSON, Needle Exchange Program Evaluation Project, The University of California

In this paper we describe the multidisciplinary methods used in our report on needle exchange programs (NEPs) in the US and abroad^1,2,3,* and list our conclusions and recommendations. The project was funded through a cooperative agreement between the Centers for Disease Control and Prevention (CDC) and the Association of Schools of Public Health, which contracted with the University of California, Berkeley to conduct the study. A significant amount of the work was subcontracted to the Institute for Health Policy Studies at the University of California, San Francisco. The project began in April 1992 and was funded through October 1993. Draft/Interim Reports were completed and submitted to the CDC on April 1 and July 16, 1993.

We assembled a 12-person research team in which no investigator was identified in his or her writing as either in favor of or opposed to NEPs. In order to better understand the diverse epidemiological, social, and policy aspects of NEPs, and to design a research methodology appropriate to their evaluation, investigators representing a variety of disciplines were involved in the project. The research team included members with training in clinical medicine, nursing, psychology, anthropology, sociology, cost-benefit modeling, and epidemiology. In addition, project investigators were accompanied on site visits to seven cities by one of four CDC staff, including the Project Officer.

The goal of the project was: ''To assess the public health impact of needle exchange programs." The central research questions surrounding NEPs were identified by team members. Revisions to the list of research questions were made based on comments solicited from members of the project Advisory Committee (see below), public health officials, NEP staff members, researchers and experts in drug treatment and injection drug use, and community leaders either favoring or opposing the programs. Three team members also made two trips to CDC headquarters in Atlanta, GA early in the project to seek further input on the research questions and to delineate methods for answering them.

This process yielded a list of 14 research questions, grouped into seven categories: (1) NEP Descriptions; (2) Populations Reached by NEPs; (3) Community Responses to NEPs; (4) NEP Effect on HIV Risk Behavior; (5) Studies of Syringes; (6) NEP Effect on Disease Rates; and (7) NEP Cost-effectiveness (see Appendix 1). Project members were aware that in several cases available data might prove insufficient to answer the research question definitively. In each such case we intended to review available data, indicate gaps in existing information, and suggest research methods that might prove fruitful in answering that research question.

Due to the strong community reactions to NEPs in some settings, it was crucial to ensure adequate community and expert input. This was sought in three distinct manners: (1) Advisory Committee; (2) expert reviewers; and (3) Kaiser Family Foundation Forum.

A 13-member Advisory Committee based in the San Francisco Bay Area was formed. Members included injection drug use researchers, staff of the San Francisco NEP, a representative of the San Francisco Department of Public Health, and members of ethnic minority and religious groups. Care was taken to include members who supported and opposed NEPs.

Two Advisory Committee meetings were held: shortly after the initiation of the project (June 26, 1992) and prior to submission of the first Draft/Interim Report (March 5, 1993). At each meeting, project investigators reported on progress made to date and sought input from the Advisory Committee.

Letters describing the project's research questions and proposed methods were sent to 31 leading researchers, public health officials, and community members, and comments from 10 respondents were received and integrated into the project approach.

We also solicited comments from the staff of the approximately 20 NEPs known to the project investigators at the beginning of the project; five responses were obtained. To promote accuracy in the descriptions of individual NEPs, staff at all sites visited were sent copies of sections of the report describing their NEP and given the opportunity to suggest corrections. Finally, five expert reviewers were contracted to review the full report and submit their comments. Two mathematical modeling experts also reviewed the chapter on cost-effectiveness.** Several additional chapters were reviewed by individuals at CDC and the National Institute on Drug Abuse who had specific expertise in the material covered in that chapter.

On December 10 and 11, 1992 the Kaiser Family Foundation sponsored a forum entitled "Needle and Syringe Availability and Exchange for HIV Prevention" at the Foundation headquarters in Menlo Park, California. The approximately 60 attendees included researchers, NEP staff, law enforcement and public health officials, injection drug users (IDUs), and community members. Project members assisted in designing the Forum program, identifying speakers, developing a bibliography on NEPs, and producing a list of active NEPs for dissemination to Forum attendees. At the Forum, project members described the project methods, summarized the history of NEPs in North America, and offered preliminary descriptions of the NEPs visited.^4,5 Feedback obtained was integrated into the Draft/Interim and Final Reports.

The research protocol was submitted to and approved by the Committee for Protection of Human Subjects at the University of California, Berkeley. Focus groups with IDUs were conducted anonymously, while other key informants were offered the opportunity to remain anonymous in the Final Report; only one accepted. IDUs participating in the focus groups were provided with a participant information sheet. NEPs were only listed in the Final Report if they provided permission.

Data analysis in this portion of the report included published and unpublished materials from the US and foreign countries.

In order to assess the data necessary to answer the research questions, a Data Collection Outline was drawn up. For each research question, the outcome variables of interest were listed and likely data sources identified. This document formed the basis for interview protocols (see below) and the chapters in the Final Report addressing each research question.

To assemble a data file on NEPs, AIDSline and Medline computer searches were undertaken. In the articles identified by these searches, references were examined for additional articles related to the research questions. Abstracts from the annual International AIDS Conferences from 1988 to 1993 were examined. Authors of abstracts on NEPs were contacted by mail for original poster or oral presentation materials as well as any published articles based on the abstract or otherwise germane to NEPs. Conference abstracts from the annual American Public Health Association conferences (1987-1992) were also obtained and the authors similarly contacted.

To identify additional unpublished materials, NEP staff were contacted and asked about internal program reports or other unpublished materials they were willing to share. Newspaper clippings, magazine articles, government and institutional reports, and book chapters dealing with NEPs were also collected. The American Public Opinion Index (1985-1991) and Gallup Index (1987-1991) were searched to identify local and national public opinion poll questions addressing NEPs.

Per capita city AIDS case rates and selected HIV seroprevalence data were obtained from the CDC and demographic information on the cities visited came from the Bureau of the Census⁶and Statistics Canada.⁷ Information on drug treatment availability in the US cities where site visits were conducted was obtained from the National Drug and Alcoholism Treatment Unit Survey (NDATUS).⁸ Data sources used in an effort to track drug use trends over time were the Drug Abuse Warning Network (DAWN),⁹ Drug Use Forecasting (DUF),^{10,11,12,13,14,15,16,17,18,19} and the Uniformed Crime Reports (UCR).²⁰

All materials assembled were reviewed by the Project Director or the Research Assistants. Documents were coded according to which research question(s) they could help answer and filed with each NEP site to which they referred. All data sources were entered into a bibliography software program called End Note Plus²¹ and coded according to research question and NEP site. Table 1 describes the data sources reviewed.

Project members were assigned responsibility for synthesizing information on one or more of the 14 research questions. This process included data collected in the manner described above as well as the interview and observational information collected during the site visits (see below).

Drafts of all chapters were circulated within the group and modified accordingly. Six of the chapters addressing the research questions and seven additional chapters on the site visits were part of the first Draft/Interim Report. The second Draft/Interim Report contained the remaining chapters. Both Draft/Interim Reports were reviewed by both paid and unpaid outside reviewers, whose comments were then integrated into the Final Report.

Between May and September 1992, site visits to 15 cities were conducted: 10 in the US, three in Canada, and two in Europe (see Table 2). Except for the site visits to Santa Cruz, Boulder, and Amsterdam, which only lasted one day, all site visits were at least three days in duration. All site visits involved either two or three project investigators or CDC personnel, except the brief visit to Boulder, which was conducted by a single investigator.

Programs were identified from a list of NEPs in a published article²² and through the community and expert input process described above. They were selected in consultation with the CDC to reflect the range of existing NEPs with respect to size, legal status, geographical location, IDU HIV seroprevalence, and extent of prior evaluation research.

The Project Director initially called the NEP staff to explain the purpose of the project and to obtain permission to visit the NEP. In no case was permission denied. NEP staff members were also used as initial sources for identifying other key informants. We supplemented these sources by adding individuals known to the investigators through their reading or discussions with other key informants as important figures in the

development of NEPs in that city. Key informants were selected in order to represent a diversity of views both in support of and opposition to NEPs.

In order to achieve internal and external validity for our study we utilized methodological triangulation:²³ multiple data collection methods with multiple iterations. This cross-checking helps to ensure a level of confidence in the results that would otherwise be lacking.^24,25,26 Interviews, focus groups, and NEP site observations used the Rapid Assessment Procedure (RAP) method as a general framework. RAP is a collection of qualitative research strategies used for quick evaluation of health interventions.²⁴

A training manual for conducting interviews, focus groups, and NEP site observations was produced and used as a guide by project members. In addition, a two hour training session was organized in which salient elements of the training manual were reviewed and interview role-playing conducted. Members of the CDC who participated in the site visits received a one hour training session by telephone. At least one project member who had attended the more comprehensive training session was present at all site visits.

Observations of NEP sites were conducted in all cities visited (see Table 3). These observations lasted from one to four hours and occurred in the presence of NEP staff. In the 13 US and Canadian cities visited, we visited 18 NEPs that provided services at 102 different sites. Six sites operated by five programs in London and Amsterdam were also observed.

In order to create as little intrusion into NEP services as possible, investigators followed the direction of the NEP staff in terms of dress, location from which to observe the NEP, and number of observers. In several cases, IDUs had been told that researchers would be visiting the site on that particular day. A list of items to be observed at each site guided the observation process.²⁴ These included physical characteristics of the site, the exchange process itself, and interactions with the local community. The observation guidelines were pre-tested at two sites and the results compared qualitatively for inter-rater reliability before adopting the final guidelines.

Interview protocols for four distinct types of interviews were developed: (1) NEP directors and staff; (2) public health officials; (3) IDU researchers; and (4) community leaders, including leaders from the following groups: law enforcement, ethnic minority groups, religious groups, local businesspeople, elected officials, and neighborhood groups. The protocols were designed as guides for the interviews; investigators were not expected to follow the questions in exact form or sequence. Rather, emphasis was placed on

making the interviewee comfortable and allowing answers to emerge during the course of conversation.

For each interview, one project member was assigned responsibility for guiding the interview, although all project members had the opportunity to clarify respondent answers or add questions of their own; another investigator was responsible for recording respondent answers and writing up the interview. Except for the interviews of NEP directors and staff, which lasted up to three hours, most interviews were completed in one hour. A total of 231 interviews were conducted during the site visits (see Table 3). In eight cases, key informants were unavailable at the time of the site visit and were subsequently interviewed by telephone using the same procedures.

The purpose of the study was explained to key informants who were asked for permission for direct attribution of their comments. Except for one individual in the Boston area who had operated an illegal NEP, such permission was always granted, although some respondents provided additional "off the record" information which was not included in the Final Report. Permission to record the interviews on audio tape was sought; in only two cases was permission denied. The interviews also provided the opportunity to augment the data collection effort and to clarify local investigator's research findings.

IDU focus group guidelines were prepared and reviewed in a fashion similar to the interview protocols. These included inquiries into their use of the NEP, factors promoting or deterring use of the NEP, enforcement of exchange rules, and suggestions for program improvement. Separate interview protocols were designed for clients of the NEP and those not using the program.

In initial telephone contacts with NEP staff, the Project Director discussed conducting anonymous focus groups with IDUs who were clients of the NEP and with those not using the NEP. NEP staff provided guidance on how best to recruit client focus group members. In most circumstances, fliers or sign-up sheets were provided to the NEP staff, who then identified potential subjects from those volunteering, paying attention to achieving an appropriate mix of gender, ethnicity, and opinion of the NEP. Eighty-two clients from 11 NEPs were interviewed.

IDUs rarely or never using the NEP were recruited from the clientele of drug treatment programs and ongoing street-based research projects. Again, recruiters were asked to seek a representative group with respect to gender, ethnicity, and opinion of the NEP. Forty-seven non-clients from seven NEPs were interviewed.

One investigator took responsibility for leading the focus group session and explaining the purpose of the study. This was facilitated by providing focus group attendees with a participant information sheet. Individuals attending the focus groups were provided with refreshments and reimbursed between $10 and $15 for their time. The focus groups were attended by between two and 10 IDUs, lasted approximately one hour and were not recorded on tape, except in Montreal where the focus groups were conducted in

French and recorded for later translation. NEP staff were not present during the client focus groups.

Project members were provided with blank computer templates of the interview and focus group protocols and recorded the interviews by reporting interviewee responses in spaces on the template below the appropriate questions. Information provided by subjects that was not directly responsive to any protocol question was noted as an addendum to the interview write-up. These write-ups were completed using notes taken during the interviews and/or tape recordings of the interviews. Each interview and observation write-up was circulated to all investigators present at the interview to confirm the accuracy of recorded information and to identify additional information for inclusion. A binder with hard copies of all interview write-ups was provided to each investigator.

Interview write-ups were also entered into a computerized qualitative database software program called askSam.²⁷ This software is frequently used in ethnographic research and permits the categorization and indexing of data so that sections of separate interviews addressing the same research question can be easily assembled. The database was used in the synthesis of the site visit findings.

In order to obtain as complete a description of US NEPs as possible, the site visits were supplemented by a mail survey of 20 NEPs not visited by project members. After the surveys were completed in May 1993, we identified four additional US NEPs. A five-page questionnaire seeking basic descriptive information was sent to the 20 NEPs. Phone follow-up was used to increase the response rate and to clarify any uncertainties regarding the responses received. The information obtained through the mail survey was integrated into the description of NEPs.

Cost-effectiveness modeling involved a four-step process: (1) an assessment of NEP budgets; (2) the systematic evaluation of studies of NEP impact on HIV risk behaviors; (3) the assessment and further development of existing mathematical models of NEP impact on HIV transmission; and (4) the use of mathematical models to estimate the cost-effectiveness of NEPs, defined as cost per HIV infection averted. In this section, we describe only methods used to evaluate existing research.

We summarized the methods and findings of all available studies addressing the effects of NEPs on HIV risk behaviors and needle-borne disease rates using a similar

format. Only information on behavior change was utilized in the mathematical modeling. The assessment of the overall usefulness of each study for answering the particular research question was ranked on a scale from 1 to 5: (1) completely inadequate or not relevant to research question; (2) unacceptable: contains flaws in design or reporting that make interpretation unreliable; (3) acceptable: provides credible evidence but has limited detail, precision, or generalizability; (4) well done: provides detailed, precise, and persuasive evidence; and (5) excellent: compelling and complete. The final ranking was determined by agreement of at least two project members.

NEPs have continued to increase in number in the US and by September 1, 1993 at least 37 active programs existed. The evolution of NEPs in the US has been characterized by growing efforts to accommodate the concerns of local communities, increasing likelihood of being legal, growing institutionalization, and increasing federal funding of research, although a ban on federal funding for program services remains in effect.

About one-half of US NEPs are legal, but funding is often unstable and most programs rely on volunteer services to operate. All but six US NEPs require one-for-one exchanges and rules governing the exchange of syringes are generally well enforced. In addition to having distributed over 5.4 million syringes, US NEPs provide a variety of services ranging from condom and bleach distribution to drug treatment referrals.

Some NEPs have made significant numbers of referrals to drug abuse treatment and other public health services, but referrals are limited by the paucity of drug treatment slots. Integrating NEPs into the existing public health system is a likely future direction for these programs.

The median annual budget of US and Canadian NEPs visited is relatively low at $169,000, with government-run programs tending to be more expensive. Some NEPs are more expensive because they also provide substantial non-exchange services such as drug treatment referrals. The annual cost of funding an average NEP would support about 60 methadone maintenance slots for one year.

Although NEP clients vary from location to location, the programs generally reach a group of IDUs with long histories of drug injection who remain at significant risk for HIV infection. NEP clients in the US have had less exposure to drug abuse treatment than IDUs not using the programs.

Studies of adequately-funded NEPs suggest that the programs do have the potential to serve significant proportions of the local IDU population. While some NEPs appear to have reached large proportions of local drug injectors at least once, others are reaching only a small fraction of them. Consequently, other methods of increasing sterile needle availability must be explored.

Unlike in many foreign countries, including Canada, proposals to establish NEPs in the US have often encountered strong opposition from a variety of different communities. Consultation with affected communities can address many of the concerns raised.

Although quantitative data are difficult to obtain, those available provide no evidence that NEPs increase the amount of drug use by NEP clients or change overall community levels of non-injection and injection drug use. This conclusion is supported by interviews with NEP clients and by IDUs not using the programs, who did not believe that increased needle availability would increase drug use.

NEPs in the US have not been shown to increase the total number of discarded syringes and can be expected to result in fewer discarded syringes.

The majority of studies of NEP clients demonstrate decreased rates of HIV drug risk behavior, but not decreased rates of HIV sex risk behavior.

The limitations of using the testing of syringes as a measure of IDUs' behavior or behavior change can be minimized by following syringe characteristics over time, or by comparing characteristics of syringes returned by NEP clients with those obtained from non-clients of the program.

Studies of the effect of NEPs on injection-related infectious diseases other than HIV provide limited evidence that NEPs are associated with reductions in subcutaneous abscesses and hepatitis B among IDUs.

Studies of the effect of NEPs on HIV infection rates do not and, in part due to the need for large sample sizes and the multiple impediments to randomization, probably cannot provide clear evidence that NEPs decrease HIV infection rates. However, NEPs do not appear to be associated with increased rates of HIV infection.

Multiple mathematical models of NEP impact support the findings of the New Haven model. These models suggest that NEPs can prevent significant numbers of infections among clients of the programs, their drug and sex partners, and their offspring. In almost all cases, the cost per HIV infection averted is far below the $119,000 lifetime cost of treating an HIV-infected person.

These conclusions demonstrate that NEPs can provide a variety of public health services to significant numbers of IDUs who continue to inject drugs and who may otherwise not receive these services. Such services can be provided cost-effectively and are associated with diminutions in HIV drug (but not sex) risk behavior. Although the data available are limited, they provide no evidence that NEPs increase the amount of drug use by their clients or change overall community levels of injection and non-injection drug use. Any controversy in local communities can be minimized by involving all interested communities in the planning of needle exchange services, both prior to opening the NEP and after it is implemented, to address concerns such as program sites and hours of operation. NEPs should be conceptualized as an integral part of public health efforts to stem HIV infection among drug users and should be part of a comprehensive approach to drug use, that should also emphasize expanded access to drug treatment and school-and community-based interventions to prevent the initiation and continuation of drug use.

NEPs should be supplemented by the expanded sale of syringes by pharmacists, an approach that has the advantage of protecting client confidentiality while still guaranteeing the client that the syringe obtained is sterile. This is in marked contrast to the situation on the street, where syringes are often repackaged by unscrupulous dealers so as to appear new.²⁸ However, pharmacy schemes provide a weaker link to other public health services and pharmacists may be reluctant to participate because of concerns about syringe disposal and the effect of IDUs' presence on their businesses.

Although the research studies upon which this report's conclusions and recommendations are based cannot definitively prove that NEPs decrease HIV infection rates, four lines of evidence suggest that this is likely:

• Needle exchange is an intervention based on the sound theoretical principle of eliminating the vector (a contaminated syringe) that transmits infection from one person to another.²⁹ This is analogous to reducing the number of mosquitoes in an attempt to prevent malaria.

• There is clear evidence of decreases in HIV drug risk behavior among NEP clients, which should translate into decreased HIV infection rates.

• Hepatitis B infections appear to be reduced by NEPs.

• Mathematical modeling by this project and other researchers consistently estimate substantial decreases in HIV transmission rates.

We believe that the data reviewed in this report meet the two criteria established by the US Congress for lifting the ban on the use of federal funds for NEP services. Federal law currently requires that ''the Surgeon General of the United States [determine] that such programs are effective in preventing the spread of HIV and do not

encourage the use of illegal drugs."³⁰ With new HIV infections in IDUs and their offspring occurring daily, the time has arrived for federal, state, and local governments to remove the legal and administrative barriers to increased needle availability and to facilitate the expansion of NEPs in the US.

• The federal government should repeal the ban on the use of federal funds for needle exchange services and substantial federal funds should be committed both to providing needle exchange services and to expanding research into these programs.

• State governments in the ten states that have prescription laws should repeal these laws.***

• States should repeal the paraphernalia laws as they apply to syringes.**** Recommendations for Local Governments and Communities

• Local governments should enter into discussions with local community groups to develop a comprehensive approach to preventing HIV in IDUs, their sex partners, and their offspring. This approach should include NEPs and the expansion of drug treatment services.

• Local communities should seek to further increase sterile syringe availability by encouraging the sale, distribution, or exchange of syringes by pharmacists.

• Descriptions of the "kinetics" and determinants of needle use patterns: IDUs' sources of needles, methods of disposal of needles, frequency of needle re-use, and needle-sharing patterns. How do these change when an NEP or other changes in needle availability are implemented?

• Evaluations of "natural experiments" in which needle availability laws change or pharmacists expand the over-the-counter sales of syringes.

• Surveys of pharmacists to determine their willingness to participate in pharmacy-based syringe sale, distribution, or exchange and to identify the barriers to their participation.

• Assessments of the effects of design features of NEPs (e.g., administering bodies, site characteristics, opening hours, and program rules) upon process measures of NEP outcome (e.g., needles distributed, drug treatment referrals, discarding of needles).

• Ethnographic and other qualitative research to assess the factors involved in drug use initiation and in transitions between various routes of drug use.

• Case-control studies of the relationship between use of the NEP and acute hepatitis B, particularly in cities with active surveillance for the infection.

• Large, multicenter case-control studies within existing cohorts of IDUs to assess whether use of the NEP is associated with hepatitis B or HIV seroconversion.

• Mathematical modeling using program data and behavior change evaluations to determine which aspects of program design determine effectiveness and cost-effectiveness.

1. NEP Descriptions

   1. How and why did NEPs develop?

   2. How do NEPs operate?

   3. Do NEPs act as bridges to public health services?

   4. How much does it cost to operate NEPs?

2. Populations Reached by NEPs

   1. Who are the IDUs who use NEPs?

   2. What proportion of all IDUs in a community uses the NEP?

3. Community Responses to NEPs

   1. What are the community responses to NEPs?

   2. Do NEPs result in changes in community levels of drug use?

   3. Do NEPs affect the number of discarded syringes?

4. NEP Effect on HIV Risk Behavior

   Do NEPs affect rates of HIV drug and/or sex risk behaviors?

5. Studies of Syringes

   What is the role of studies of syringes in IDU research?

6. NEP Effect on Disease Rates

   1. Do NEPs affect rates of diseases related to injection drug use other than HIV?

   2. Do NEPs affect HIV infection rates?

7. NEP Cost-effectiveness

   Are NEPs cost-effective in preventing HIV infection?

HOLLY HAGAN, Tacoma-Pierce County Health Department, Washington; DON C. DES JARLAIS, Chemical Dependency Institute, New York; SAMUEL R. FRIEDMAN, National Development and Research Institute, New York; and DAVID PURCHASE, Point Defiance AIDS Projects, Tacoma, Washington

In almost all developed countries, syringe availability programs have become a major component of public health programs to reduce transmission of HIV and other blood-borne infections via needle-sharing.^1,2,3 In the U.S., the controversy surrounding syringe exchange has limited opportunities for conducting research^4,5,6, and no large-scale controlled studies have been carried out to date. The approach taken by many researchers has been to collect different indicators of the effects of syringe exchange, including HIV risk behavior information, HIV and hepatitis B serology, and community incidence trends, and to determine whether the relationships between exchange programs and these indicators suggest reduction in blood-borne viral transmission^{7,8,9,10,11,12,13,14,15,16}. The ongoing study of the Tacoma syringe exchange is an example of such a research strategy.

The current approach of the Tacoma study is to examine the association between exchange use and the incidence of hepatitis B. Because HIV and hepatitis B virus (HBV) have similar routes of transmission, and the incidence of HBV infection is higher than that of HIV^17,18 studies of hepatitis B will have greater statistical power to detect a difference in risk of infection attributable to exchange use¹⁹. Studies conducted in Europe and the U.S. have found that 65-95% of IDUs have serologic evidence of previous hepatitis B infection^{20,21,22,23,24,25,} and the Centers for Disease Control and Prevention (CDC) estimates that 10-20% of susceptible IDUs acquire hepatitis B every year¹⁸. In favor of studying hepatitis B as an outcome in Tacoma are that HIV seroprevalence in local IDUs is low, and a comprehensive hepatitis surveillance system has existed in the county since 1979.

In this paper, we summarize the methods and findings of the Tacoma study and examine the limitations. The preliminary results of the first analysis of the association between hepatitis B and the syringe exchange at the individual-level are also described in this paper.

The Tacoma syringe exchange began operating in August, 1988, and is the longest-running program in North America. It is estimated that the program serves 8001000 clients per week—approximately 30% of an estimated 3000 IDUs residing in the county. Between 50,000 and 60,000 syringes are exchanged each month in four locations: two stationary street-based sites, a pharmacy-based exchange, and a mobile delivery program. The staff of the program have sought to limit barriers to participation, and users are not required to show evidence of legal age or recent injection, or to register or participate in any research.

The syringe exchange has been the primary HIV prevention strategy for IDUs in Pierce County. Some HIV education is given to drug injectors when they enter drug treatment, but there is no other outreach education program for this segment of the population in the community.

Interviews with injecting drug users in the county have been carried out since 1988. The details of the methods used in this study have been described elsewhere²⁶. Exchange users were systematically sampled at exchange locations; IDUs who had never used the exchange (non-exchangers) were recruited from health and social service agencies and street locations where the exchange didn't operate. Risk behavior interviews and blood draws for HIV and HBV serology were conducted in the field. Variables included in the standardized instrument included demographic characteristics and injection risk behavior during the previous 30 days while participating (exchange users) or not participating (non-exchangers) in the program. Exchange users were also asked about behavior before first use of the program (pre-exchange), and this was compared to post-exchange behavior (while participating).

For variables with a continuous distribution (age, number of injections per month, number of unsafe injections per month), the median for the entire sample was used to create dichotomous categories. Prevalence odds ratios (P.O.R.) were used to estimate the relationship between syringe exchange use and injection behavior, and these P.O.R. were adjusted for duration of injection, a potential confounding variable, by use of the Mantel-Haenszel statistic²⁷.

Pierce County is one of four U.S. sentinel counties for hepatitis surveillance conducted by the Hepatitis Branch of the CDC. Stimulated reporting of hepatitis by the

sentinel surveillance system has resulted in approximately 50% of all symptomatic cases of hepatitis B infections in Pierce County being reported, compared to 17% nationwide²⁸.

Since 1988, the Tacoma-Pierce County Health Department's drug-treatment program has enrolled newly admitted opiate users in the CDC Family of HIV Seroprevalence Surveys²⁹. HIV risk behavior was recorded for each client, and HIV testing was done on sera remaining after routine blood tests were performed.

Cases of acute hepatitis B infection reported to the sentinel surveillance system from January 1991 to December 1992 who were IDUs were included in the case series. Demographic data and information relating to potential source of exposure were collected according to the sentinel county protocol³⁰. Additional information obtained from cases included syringe exchange use and duration of drug injection (less than or five or more years). We excluded IDUs who also reported male-with-male sexual activity or sexual contact with a confirmed hepatitis B case, as HBV transmission may have occurred via a causal pathway other than the one of interest in this study. No other risk factors for hepatitis B were reported by the cases.

The control series was assembled from among IDUs enrolled in the HIV Family of Seroprevalence Surveys during the study period. Potential controls were screened for antibody to hepatitis B core antigen (anti-HBc), and hepatitis B surface antigen (HBsAg). Heterosexual IDUs who had neither of these serologic markers of immunity or infection were eligible to serve as controls.

IDUs with acute hepatitis B were compared with their controls for differences in demographic characteristics and injection behavior, and the independence of factors that were significant on univariate analysis was examined using multiple logistic regression with backward elimination.

In the interview study comparing 426 exchange users and 159 non-exchangers, there were no differences in the gender or racial/ethnic composition of the two IDU samples. Exchange users were significantly older than non-exchangers, and had been injecting longer (Table 1). One percent of exchange users and 4% of non-exchangers were new injectors, having begun injecting in the previous year (p=0.12).

Changes in injection practices among exchange users pre-and post-first use of the exchange are described in Table 2. There were no changes in the rate of injection by exchange users, but there were significant declines in unsafe injections. Fifty-eight percent of exchangers reported any unsafe injections pre-exchange, compared to 33% while participating (P.O.R.=0.36, 95% CI 0.26 to 0.49). Exchange users also reported fewer occasions when they passed a used syringe onto another injector.

Injection practices at baseline reported by exchange users and non-exchangers are compared in Table 3. Exchange users were more frequent injectors, with a median of 37 injections per month, compared to 17 for non-exchangers. However, fewer exchange users reported any injections with a syringe used by another injector. For both samples, heroin and speedball (an injection of heroin and cocaine together) were the primary injected drugs. The differences in injection frequency and unsafe injection persisted when the data were adjusted for duration of injection (Table 3). There were no differences in the frequency of passing on a used syringe to another injector.

The presence, at time of interview, of antibody to HIV, HBV and HCV is compared in Table 4. In both IDU groups, a high proportion of subjects had evidence of previous HBV or HCV infection. Fewer exchange users (2%) than non-exchangers (8%) were HIV-antibody positive. There have been no HIV seroconversions detected in either exchange users or non-exchangers in follow-up testing (66.9 person-years followup for non-exchangers, 223 person-years for exchange users).

Figure 1 shows HBV incidence trends in Pierce County from January, 1985 to December, 1992. An outbreak of drug-related hepatitis B infection began in 1985 that persisted until several months following the opening of the exchange program, and then rapidly declined. Hepatitis B in persons whose source of infection was not identified has followed a similar trend. Incidence in persons with other sources of infection, primarily heterosexual and homosexual transmission, have been relatively stable during this time period. HIV seroprevalence in opiate users entering methadone treatment in Pierce County has remained between two and five percent from 1988 to 1992³¹.

In the case-control study, eligible cases included 34 heterosexual IDUs with acute hepatitis B, and 25 eligible controls with no hepatitis B markers. Control subjects were older than cases, but there were no differences with regard to gender or race. Cases of incident hepatitis B were eight times more likely to have injected for less than five years (95% CI 2.3 to 35.3), and five times more likely to never have used the syringe exchange (95% CI 1.4 to 20.0). In multiple logistic regression analysis, two factors, not using the syringe exchange (AOR 2.1, 95% CI 1.1, 4.2) and injecting for less than five years (AOR 2.5, 95% CI 1.1, 4.2) were independently related to acquisition of hepatitis B³².

These studies have examined the relationship between the Tacoma syringe exchange and self-reported injection behavior, individual-level HIV-, HBV-, and HCV-seroprevalence, community-level HIV-seroprevalence trends and HBV-incidence trends, and acquisition of hepatitis B. All of these measures suggest that syringe exchange protects against blood-borne infection; each has limitations, which will be summarized.

Self-reported safer injection may have been influenced by subject-bias whereby exchange users would be motivated to report cessation of injection risk behavior. All interviews have been carried out by well-trained interviewers who are not affiliated with the exchange program, so subjects are unlikely to expect to gain by giving specific response. We examined the influence of social desirability on report of safer injection

in a previous analysis³³ by re-analyzing the data after excluding subjects who reported certain behaviors that could be viewed as socially desirable. The results were consistent with those reported for the entire sample, so it is unlikely that social desirability of respondents substantially biased the results.

Differences in injection practices between exchange users and non-exchangers may be due to volunteer bias, whereby IDUs motivated to inject safely would be attracted to the exchange. Klee et al.³⁴ have reported that IDUs may inject for several years before adopting safe injection practices. Because older IDUs may be safer injectors as a result of their longer experience, controlling for duration of injection may have eliminated some of this form of bias. Additionally, exchange users did report safer injection after beginning to participate in the program, so there was a change beyond that which may be attributable to baseline differences in the two groups.

Ecologic correlations between HIV prevalence and hepatitis B incidence trends and the opening of the syringe exchange program cannot be interpreted as an individual-level association. No "control" community with similar demographic and disease incidence features has been compared to the Pierce County trend. Other factors that may have contributed to the observed trends include community-wide HIV education campaigns, legal sale of injection equipment by pharmacists, use of disinfectant bleach, and saturation of susceptibles during the hepatitis B outbreak.

Separating the education and bleach distribution effect of the exchange from the single effect of syringe availability was not possible in these studies. Syringe exchange programs have always been "more than just needles", and work to achieve multiple public health objectives in the course of their interactions with IDUs. Condom and bleach distribution, screening for tuberculosis and other infectious disease, health and social service referrals, and facilitation of drug treatment admission have been elements in the majority of syringe exchange programs in the U.S.² It is also difficult to draw the line between the effects on individual users and community norms. It is conceivable that, by increasing awareness of injection-related HIV risk and establishing new norms of needle-hygiene, syringe exchange's influence extends far beyond the individual user. Within-community studies may be hampered by this "contamination" of potential controls.

With respect to the case-control study, the small sample size led to wide 95% confidence limits about the estimates. Additionally, there was limited data on covariates, and some residual confounding could be present in the final estimates. However, duration of injection was probably the most important potential confounder and was included in the analysis. There was little detail in quantification of use of the syringe exchange, but measuring exchange use as ever vs. never would tend to underestimate the protective effect. Similar studies should be carried out to address these methodologic concerns.

While these individual studies of the Tacoma syringe exchange each had their limitations, there was strong consistency among their findings, all indicating reduction of transmission of blood-borne viruses among participants in the exchange program.

KATHY OLIVER, Outside In; H. Maynard, Portland State University; SAMUEL R. FRIEDMAN, National Development and Research Institutes, Inc.; and DON C. DES JARLAIS, Beth Israel Medical Center

The Syringe Exchange in Portland, Oregon began November 1, 1989. The opening of the Exchange was delayed for two years because our insurance company refused to cover syringe exchange and threatened to withdraw coverage from our clinics and other programs if we started an exchange. Outside In operates medical and prenatal clinics and housing and emergency services programs for homeless youth.

At the time of the study, the Exchange operated out of a fixed site at Outside In. It now operates out of two fixed sites and one van.

At the time of the study, the Exchange operated 3-7 pm Monday-Friday. Clients were age 18 or over-it is illegal to give minors syringes in Oregon unless by order of a physician for an authorized use. Clients were asked to exhibit needle tracks at the first visit to ensure that only IV drug users were provided syringes. At the first visit clients were given three syringes-whether or not they brought any in. Thereafter, syringes were exchanged on a one-to-one basis. Outside In also gave out rinse water, cotton, alcohol swabs, condoms and information handouts.

Of over 1,000 participants in the Exchange, 753 (67%) were enrolled in the research component. Subjects were interviewed at three month intervals using the AIA (for intake) and AFA (for follow-up) Questionnaires designed by NIDA for use in its national survey of drug injectors. Consenting subjects were also tested for HIV and Hepatitis B antibodies at three month intervals. To encourage use of the exchange, subjects were allowed to remain anonymous, which prevented active tracking for follow-up, and thus limited follow-up at the three and six month data collection points.

Behavior of Exchange clients was examined at intake and at six months to determine the extent to which clients changed their behavior. To determine differences between a syringe exchange program and an outreach program, Exchange clients were compared with clients of the Portland NIDA-funded NADR outreach project for IV drug users. Clients of both groups were those who were still shooting up at six-month follow-up. The same questionnaires were used with both groups. Change in risk behavior over time was compared between groups.

A substudy was conducted to determine the extent to which the existence of the Exchange would lead injectors to return syringes to the Exchange rather than discarding them in public places. Prior to the opening of the Exchange, staff began to count (and collect) syringes in the vicinity on a daily basis. The data on syringes found before opening were compared to data on syringes found after the exchange opened.¹

In the first two years since the Syringe Exchange began, 1,145 clients made 6,368 visits. Nearly 49,000 syringes were given out, with over 45,000 returned. The syringe return rate for the two years was 93% (see Table 1).

A description of the first 700 injectors using the Exchange and participating in the study is provided in Table 2.

The HIV infection rate among clients at intake was 3.9%. There was only one HIV seroconversion in 162 person-years at risk, for an HIV seroconversion rate of 0.619 per 100 person-years at risk (95% confidence intervals 0.109, 3.506). The single seroconversion was noted at the six month follow-up, so it is possible that this person may have become infected prior to beginning to use the Exchange.

The infection rate for hepatitis B core antibody at intake was much higher-54%. Thirteen subjects seroconverted in 63.3 person-years at risk, for an HBV seroconversion rate of 20.5 per 100 person-years at risk (95% confidence intervals 12.0, 35.1).

Syringe exchange clients showed a considerable reduction in risk behavior over a range of risk behaviors measured at intake and at six months. Significant and meaningful declines were reported in sharing of syringes, as well as renting works, borrowing works, and cleaning works. (See Table 3).

Of the 117 subjects for whom follow-up data were obtained at six months, 34 attended the Exchange three or fewer times, and 83 attended it four or more times. These two groups were distinguished to try to separate out clients who came in only to collect interview fees and not otherwise using the Exchange from those really using the Exchange. The frequent attenders report significantly greater risk reduction on borrowing and on throwing away used syringes. This latter result indicates that injectors are using the Exchange as intended, that is returning syringes for safe disposal rather than simply discarding them. (Table 4)

On the other hand, the analysis indicate that while all clients reduced drug injection frequency, frequent attenders reduced frequency of injection less than infrequent attenders. While this did not reach significance (P < .099), it is worthy of note. Further analysis is needed to determine whether this is a methodological artifact analogous to regression to the mean (since those who come to inject only rarely will have little reason to frequently attend the Exchange to get new syringes.); a result of personal characteristics of these subjects (such as the stage of their addiction); or is an unintended effect of the program.

The respondents had long histories of IV drug use; there is little evidence in our data to support the idea that syringe exchanges recruit new users. Less than 2% of our

respondents had histories of injecting of less than a year. The average duration of IV drug use was 14 years, and over 75% of respondents had been injecting for 5 years or more. Those with the longest histories of injecting were likely to be the heaviest current IV drug users (p <.004).

Drug injectors using the Syringe Exchange were then compared to drug injectors not using the Exchange. Clients for the comparison group were drawn from the Portland NADR outreach project. This was not an ideal comparison group in that drug injectors in the project did receive interventions. All NADR clients received bleach and were provided with HIV education. A subset either participated in groups or received one-on-one counseling. They were also encouraged to buy and use sterile syringes (legally available over-the-counter in Oregon).

Seventy-seven Exchange clients who attended the Exchange four or more times were compared to 355 NADR clients. Demographic information for both groups is shown in Table 5.

The comparison between these projects in complicated by potentially different subject populations. While there were no differences between samples on most variables, there were some differences. For example, NADR subjects were considerably more likely to be African American and less likely to be Latino; more likely to live in their own place, and were engaged in less risky injection practices at intake (e.g., they were more likely to clean syringes after use). Perhaps because we were studying frequent attenders, exchange users at six months follow up injected more frequently than the NADR subjects. These differences between samples may limit the conclusions that can be drawn from the comparison.

Subjects in both projects report significantly lower levels of risk over a wide range of risk behaviors at follow-up than at intake. On most of these measures no statistically significant differences in the amount of risk reduction were found between projects. However, syringes exchange subjects were significantly better on two variable involving risky injection behavior (Table 6): reduction in the extent to which subjects re-used syringes without cleaning-a key goal of the project, and the extent to which they no longer threw away used syringes (but returned them to the Exchange).

Differences are of particular significance in that syringes are legally available in Oregon, and the bleach outreach program encourages drug injectors to buy and use sterile syringes. Differences between the Exchange and the outreach program are likely smaller than they would be between an Exchange and an outreach program in a state where purchase of syringes was illegal.

Of considerable importance is the fact that syringe exchange and bleach outreach projects seem to recruit different clienteles. There was little overlap between the samples, with the Portland NADR project finding that only 11% of its sample had ever used the syringe exchange. Thus, it appears that syringe exchanges and outreach programs might best be seen as complementary strategies that recruit and produce risk reduction among different sub-populations of drug injectors, rather than as competing options that should be chosen among to find which is the best approach to HIV prevention.

The substudy on discarded syringes assessed the impact of the Syringe Exchange on the community in terms of the number of potentially infectious syringes thrown away on the streets.¹ In order to provide baseline data, a daily syringe search was begun three-and-one-half months prior to the opening of the Exchange. Discarded syringes were collected, counted and disposed of safely. Using consistent search patterns, this syringe count was continued until June 30, 1991.

Table 7 presents data comparing the number of days on which syringes were and were not found prior to and subsequent to the opening of the Exchange. Syringes were significantly more likely to be found prior to its opening (chi-square = 4.048; p < .05).

In spite of the fact that syringe exchanges aim to increase the availability of (sterile) syringes for drug injectors, this exchange has not led to an increase in the number of discarded syringes with which children or others might stick themselves. Instead, it reduced the number of discarded—and possibly contaminated—syringes in the streets.

The data presented here support the growing body of evidence that exchanges produce behavioral risk reduction. They also provide evidence that the number of potentially infected syringes in public places can be reduced by opening syringe exchanges. Comparisons between Exchange subjects and NADR are preliminary, with further analyses needed to control for possible differences between samples. Syringe exchange and bleach outreach programs are best seen as strategies that complement each other. They recruit different populations of drug injectors, and both lead to risk reduction. Differences between programs are likely smaller than in a city where syringes were illegal and the bleach outreach programs could not encourage people to buy sterile syringes.

These data are of course not definitive evidence that syringe exchanges reduce the spread of HIV or other pathogens. Indeed, it is not likely that a truly definitive study can ever be conducted of syringe exchanges, any more than this is feasible with evaluations of drug abuse treatment, drug interdiction, or laws against drug use. The cumulative weight of the research, however, and the fact that no studies have found any indication that the exchanges are doing any damage, clearly puts the burden of proof on opponents of syringe exchanges. In the interim, given the dangers from HIV spread, syringe exchanges should become an important part of the public health response to AIDS.

We would like to acknowledge the support by the American Foundation for AIDS Research and the Multnomah County Health Division.

We would also like to acknowledge the staff of Outside In and the volunteers from Portland State University and the Portland community who assisted in making the Portland, Oregon syringe exchange and this research possible.

DENISE PAONE, DON C. DES JARLAIS, STEPHANIE CALOIR, PATRICIA FREIDMANN, and IMMANUEL NESS, Beth Israel Medical Center, New York; SAMUEL R. FRIEDMAN, National Development and Research Institutes, New York

New York City has experienced the largest HIV epidemic among injecting drug users (IDUs) of any city in the world. Approximately one-half of the estimated 200,000 IDUs in New York are already infected with HIV. 19,792 cases of AIDS among IDUs in New York City (including IDUs who also report male-sex-with-males) had been reported through December 1992 (NYC AIDS Surveillance Report). If one adds the 1,809 New York City cases of AIDS that resulted from heterosexual transmission from IDUs to persons who did not inject and the 815 cases of perinatal transmission of AIDS attributed to the mother's drug use or that of her sexual partner, then injecting drug use in NYC is associated with almost ten percent of the 250,000 total cases of AIDS reported in the U.S. through December 1992 (Centers for Disease Control, 1993).

Syringe exchange programs have become a primary method of preventing HIV infection among injecting drug users in almost all developed countries (Des Jarlais & Friedman, 1992), although they have remained quite controversial in the United States. Most syringe exchange evaluation studies have been conducted in areas with low HIV seroprevalence among the local population of IDUs. All syringe exchange evaluations conducted to date have shown reductions—but not elimination—of injection risk behavior and stability of the low HIV seroprevalence. In a high seroprevalence area, these residual levels of risk behavior may still leave an unacceptably high rate of HIV seroconversions. Only two studies of syringe exchanges have been conducted in moderate to high seroprevalence areas. The New Haven evaluation has estimated a one-third reduction in new HIV infections among IDUs attending the syringe exchange in that city, but it was not possible to directly measure the HIV seroconversion rate (Kaplan & O'Keefe, 1992). In Amsterdam, the HIV seroconversion rate was directly measured among participants in the exchanges, but the rate did not differ from that observed in IDUs not utilizing the exchanges (although many of the comparison subjects were probably obtaining sterile injection equipment from pharmacies or from exchangers).

Most syringe exchange programs that have been studied in the U.S. have also been relatively small exchanges, with a few exchange sites and open for a limited number of hours per week. If syringe exchanges are to become an important method of HIV prevention in the U.S., it will be necessary to operate them on a much larger scale. Given the importance of syringe exchanges being sensitive to the needs of the local population of IDUs, it is very unlikely that a single ''standard" model of syringe exchange

will be applicable to all locations. Instead, some form of "integrated system" of syringe exchanges that would be sufficiently flexible to meet the needs of the local mix of IDUs, but sufficiently uniform to meet the demands for accountability that will inevitably come with federal or state government funding. The present system of syringe exchanges in NYC, with five different exchanges currently operating, may serve as a model for how such a multi-exchange system might operate.

The New York City syringe exchange programs are required by the State Department of Health to participate in our evaluation study. This requirement is similar to the situation in Honolulu, HI, New Haven, CT, and Washington, DC. In all these cities, establishment of syringe exchange was controversial, and the requirement for an evaluation study was both part of the political process for obtaining legal authorization and part of a larger concern of public health officials with assessing the extent to which the multiple goals of syringe exchanges were being achieved. We are currently conducting an evaluation study of the recently legalized syringe exchange programs in New York City. This evaluation study will address two fundamental questions: the potential effectiveness of syringe exchange in a high HIV seroprevalence environment, and the New York exchanges as a prototype for an integrated system of syringe exchanges.

The design of the NYC needle exchange evaluation consists of two major components: (1) Ethnographic Study and (2) Behavioral Change and Seroconversion Study.

The ethnographic component has been designed to provide a description of the operations of the syringe exchanges and to study the organizational issues. The ethnographic methods in this evaluation study include direct observation of the syringe exchanges in operation, and semi-structured interviews to document the past and ongoing decision-making of the syringe exchange staff.

These ethnographic methods have produced the types of descriptions of the everyday operation of the programs that can be readily understood by decision makers who would otherwise have little familiarity with the "front line" work of preventing HIV infection among IDUs.

A key area for ethnographic inquiry has been an examination of how semi-legal syringe exchange programs make the transition from underground to legal status and how community-based organizations integrate syringe exchange into their pre-existing services.

Ethnographic research methods, with their great flexibility and ability to probe beneath "surface" events for underlying contradictions and hidden inconsistencies, are ideally suited for addressing the types of organizational questions faced by the syringe exchange programs in New York City. Ethnographic research methods are also highly likely to produce the types of scientific reports that will be useful for replication of effective syringe exchange programs in other areas, particularly other high seroprevalence areas.

The fundamental goal of syringe exchange programs is to reduce AIDS risk behavior and actual transmission of HIV. Changes in AIDS risk behaviors and HIV transmission among participants are examined through a series of cross-sectional studies with recapture of subjects who participated in previous cross-sectional studies.

Each cross-sectional study includes an interview concerning AIDS risk behavior and a saliva sample for HIV testing. Subjects for interviewing are randomly selected from those attending the exchanges within a given week. To be eligible for inclusion in the study, subjects must: (1) have been an active injecting drug user; (2) have used syringe exchange on at least one occasion; (3) have just made an exchange. Verbal informed consent is obtained in order to protect subjects' confidentiality. Two separate interview questionnaires are administered, one for the initial interview with a subject and one for subsequent interviews with the same subject. Subjects are paid a modest honorarium ($10) each time they participate in the study. A "unique identifier" anonymous coding system is utilized for tracking multiple interviews with the same subjects across the different cross-sectional studies.

In the initial interview, subjects are asked about their drug use and AIDS risk behavior during the month before they started to use the exchange, as well as for the past 30 days while using the exchange. Comparison of the "pre-exchange" and "during exchange" levels of illicit drug use and AIDS risk behavior has permitted inferences (Paone et al., 1993) about behavior change associated with use of the exchange.

Subsequent interviews of previously interviewed subjects include risk behavior since the last time interviewed. The questionnaire is also used to monitor any changes in the frequency of drug use and risk behavior over time among "regular" users of the exchanges. These questionnaires also permit comparison of AIDS risk behavior among participants in syringe exchange in the New York City exchanges with AIDS risk behavior among participants in syringe exchanges in cities of lower background seroprevalence.

In order to determine the effectiveness of syringe-exchanges to reduce HIV transmission, this study directly measures HIV seroprevalence and HIV seroconversions among regular participants of the exchanges.

Saliva samples are collected at the time of each research interview. The saliva samples are tested for HIV using GACELISA and Western blot assays. Although the saliva tests are not currently licensed in the United States, they have shown considerable accuracy. Compared with serum tests for HIV antibody, saliva tests are at least 95% sensitive and 100% specific. Saliva testing has been successfully utilized in the United Kingdom HIV seroprevalence studies (Johnson et al., 1988) and in United Kingdom syringe-exchange evaluations (Hart et al., 1989).

We have estimated that approximately 1000 (20%) of the first 5000 interviews and saliva samples will be "subsequent" interviews of persons who have already been interviewed and who have given a previous saliva sample. Using an estimate of 50% seroprevalence among persons using the syringe exchange, there will be an estimated 500 subjects with multiple saliva samples who are at risk for seroconversion. Assuming an average period of between 6 and 9 months between the first and last saliva sample among the matched saliva samples, there will be an estimated 250 to 375 person years at risk for determining the rate of HIV seroconversion among regular participants in the syringe exchanges.

Space limitations do not permit a full analysis of the complex issues involved in establishing a network of syringe exchanges within the politically complicated environment of New York City. It will be helpful, however, to start with brief descriptions of the different organizational players. There are four different types of organizations actively involved in syringe exchange work in New York City.

Both the New York State and the New York City Health Departments are involved in the syringe exchange system. The City Health Department provides moral and political support to the legal exchanges, but does not contribute financial resources to the programs. The State Department of Health, in contrast, occupies a dominant role within the syringe exchange system. New York State law permits the State Health Commissioner to waive the requirement for written prescriptions for dispensing of needles and syringes. This waiver was originally included in the law in order to reduce

the need for prescriptions in State operated hospitals, and was not intended to include syringe exchange programs. Operating under a public health emergency provision, this waiver has been extended to syringe exchange programs that file the appropriate applications with the State Health Department and agree to abide by the 27 page set of Department regulations for legal syringe exchanges.

The State Health Department also provides funding for syringe exchange work in New York. The funding is provided under an explicit comprehensive "harm reduction" (NYS Dept. of Health, 1993) philosophy which emphasizes using syringe exchanges not only to reduce HIV risk behavior but also as a linkage mechanism for providing additional health and social services to the drug injecting population.

Both Health Departments can be considered traditional bureaucracies, with clearly defined roles within the organization and well-articulated authority structures. Both need to be responsive to community concerns and the concerns of New York State Legislature, which must approve funding for the State Health Department and whose laws govern the operations of both departments. Both Health Departments are concerned that their activities be consistent with the relevant state laws and frequently rely upon the advice of their counsels.

Both Health Departments also have a tradition of regulating and/or inspecting health care facilities. The State Health Department, for example, has the authority not only to license health care facilities but to set reimbursement rates for hospitals. The HIV/AIDS epidemic has extended the regulatory/inspection responsibilities to areas beyond traditional health care facilities. The State Health Department is currently regulating the syringe exchanges and the City Health Department is inspecting bath houses and sex clubs for unsafe sexual activities. The State Health Department has provided technical assistance to programs. The State Health Department's, AIDS Institute Staff, has worked closely with law enforcement community, providing education in order to gain acceptance for the programs.

AmFAR is a not-for-profit foundation that funds AIDS related research and that advocates for AIDS policies. AmFAR has been the most important source of funding for syringe exchange research in the United States. Until 1992, the federal government had not funded any research on syringe exchange programs. AmFAR is funding the current syringe exchanges in New York, with privately obtained money and money from the State Department of Health. AmFAR provides technical assistance to the syringe exchange programs and centralized purchasing of supplies for the programs. AmFAR is also funding the evaluation of the syringe exchange programs.

Two of the syringe exchange programs that received legal operation authority in the summer of 1992 had been operating previously as "underground" exchanges. A third program with a variation on the model noted below received authorization in (February) 1993. Prior to receiving this authority, these exchanges had no paid staff; necessary supplies were provided by private donations, mostly from the AIDS Coalition to Unleash Power (ACT UP or the volunteers themselves). Decision making was made primarily through developing a consensus among the volunteers who were present at the meeting when an issue was raised. This method of decision-making provides for maximum egalitarian participation among the totally volunteer staff. It also meant that no single person "representing" these two exchanges had authority to make commitments for the exchange.

With legal authorization to operate syringe exchanges and funding from AmFAR and the State, these exchanges were required to affiliate legally with a 501 (C) 3 not-for-profit corporation in the neighborhoods where they were to continue operations. Funding would go through the sponsoring organization, which would have legal control and legal responsibility for expenditures. These three exchanges had considerable difficulties in finding suitable sponsoring organizations and have moved towards incorporating as 501 (C) 3 organizations themselves, with their own boards of directors and formal organizational hierarchies.

These underground exchanges were able to expand greatly when they received legal authorization and funding support. They expanded from providing exchange services to 1,000 IDUs one year prior to legalization in June 1992, to 8,379 IDUs by June 1993. This expansion brought with it many of the problems that occur with rapid growth in any organization. (Space limitations do not permit examination of these problems and their solutions here.)

The three other syringe exchanges that were initially approved for funding in the summer of 1992 were to be operated by existing communitybased organizations that already held 501 (C) 3 status. These organizations were already providing AIDS related services to injecting drug users, but were not providing syringe exchange. Thus, they had to integrate syringe exchange into their existing services. (Space limitations do not permit presentation of the complicated issues raised during this integration process.) We should mention, however, that some staff in these organizations were philosophically committed to assisting IDUs to stop using drugs, and did not accept the harm reduction rationale for syringe exchange. Although these organizations already had 501 (C) 3 status, they also had undergone (or were still undergoing) problems associated with rapid growth, overdependence upon key personnel, and the staff burnout common in many CBOs in the AIDS field.

Not unexpectedly, these existing non-exchange service organizations encountered many difficulties in starting exchange services. One of them never officially obtained authorization to operate a legal syringe exchange, and subsequently AmFAR funding was withdrawn from that organization.

The evaluators must also be considered one of the organizational players within the syringe exchange system. As noted above, participation in the evaluation study was required for both AmFAR funding and for obtaining State Health Department authorization for operating a legal exchange. The evaluators have considerable experience in research on syringe exchanges, having consulted on several studies of European and American syringe exchanges and on the evaluation of the first legal exchange in New York City, and served as co-investigators on studies of the Portland, Tacoma and Boulder syringe exchanges in the U.S. This experience, as well as the research literature on syringe exchange, has led them to believe that syringe exchanges can be important components of AIDS prevention effort for IDUs, but that as a human service operation, syringe exchanges will not always be properly implemented, and even if properly implemented, may not necessarily be effective in all environments.

The purposes of the evaluation (as seen by the evaluators) were to advance scientific knowledge about the operations and potential effectiveness of syringe exchange, particularly in a high HIV seroprevalence environment; to utilize biological marker data as one outcome measure in this evaluation because of the greater perceived "objectivity" of such data; and to provide the exchanges with information that could be utilized to improve operations.

Another important aspect of the first year was the development of positive working relationships among the different organizations. Given the differences in organizational cultures, this in itself can be considered a major accomplishment (Broadhead & Margolis, 1993). Much of the development of positive working relationships occurred during monthly meetings hosted by AmFAR and attended by representatives of the syringe exchanges, the State Health Department and the evaluators. These meetings consisted primarily of problem identification followed by protracted, but usually successful, group problem solving.

The "tagging" of the syringes to be distributed by the exchanges is a good example of an identified problem with an eventual resolution. When the two underground exchanges were operating as underground exchanges, they usually "tagged" the syringes they distributed by painting a small mark on them. This tagging had several advantages. It permitted the exchanges to determine how many of their own syringes were being returned versus how many syringes from other sources were being brought in to the exchanges. The visible tag also served an educational/reminder function for the IDUs, indicating that it was possible to obtain HIV-free syringes from the exchanges.

When the exchanges were given legal authorization in the summer of 1992, it was expected that the tagging would continue. Tagging would have additional advantages for the legally authorized exchanges. Since it was now legal for an IDU to possess syringes obtained from an authorized exchange, tagging could assist law enforcement officials in

distinguishing between legal and illicit syringes. (Though this would obviously not be foolproof, since illicit syringes could also be marked by anyone who chose to do so.) When the regulations for legal exchanges were first drafted, tagging of syringes to be distributed was included, and State Health Department officials made several public statements that tagging would be included in the final regulations.

The rapid expansion of the legalized exchanges brought forth the difficulties in attempting to tag syringes. Discussions with syringe manufacturers led to a quick conclusion that it would not be economically feasible for manufacturers to modify their production processes to produce relatively small lots of "pre-tagged" syringes. The manufacturers also noted that post-production tagging compromised the sterility of the syringes. The two large-scale legal exchanges (formerly underground exchanges) found that the tagging operation rapidly became the rate limiting factor in their ability to deliver syringe exchange services. The funding level of the syringe exchanges required the exchanges to use volunteers for the task of tagging syringes. Large-scale tagging was creating a severe morale problem among the volunteers. The leaders of the two large exchanges reached a point where they believed it would be impossible to continue large-scale operations if the tagging was required.

As the exchanges proceeded during the first year, there were persistent reports of police harassment of IDUs participating in the exchanges, including confiscation of legally obtained tagged syringes and taking of the cards that identified the IDUs (by a code number) as participants in the legal exchanges. These reports undermined the rationale for having syringes that could be easily identified as legal. Concerns about liability issues if a tagged syringe should be involved in a needlestick injury of a person who did not inject illicit drugs were also raised, although there were no reports of needlestick injuries.

The need to know the percentage of distributed syringes that were returned to the exchanges was still considered important by the State Health Department and several of the exchanges themselves. The eventual resolution to the problem of continued tagging of syringes was to omit tagging from the regulations and to have the evaluators conduct a 'Tagging Alternative Study" (TAS). In this study, a sample of syringes will be tagged, with systematic measurement of the returned syringes over the next month. This study will provide an estimate of the percentage of syringes returned for each of the different exchanges over the one month measurement period. The study could be repeated for either all exchanges or for selected exchanges if a need should arise in the future.

The "tagging question" is an example of an issue where the perceived needs of the different organizations within the system were in sharp conflict. This issue reflects a more general conflict between having tightly regulated exchanges and the HIV prevention goal of providing sterile injection equipment to as many injecting drug users as possible. This more general conflict has been a dominant theme during the first year of operation, and we expect it to continue as a dominant theme for the indefinite future.

To date 2849 interviews have been conducted; this analysis is based on 1752 baseline interviews conducted between the months of October 1992 and June 1993. Less than one percent of the interviews have been excluded from this analysis due to unreliability of the responses (as determined by the interviewer who recorded reliability on a Likert scale based on consistency of respondents' answers and on face-to-face judgment of the validity of the data.

Of the 1752 subjects included in this analysis 70% were male and 30% were female. IDUs participating in this study were racially/ethnically diverse (38% Latino, 35% African American, 28% White). The mean age of participants was 36 (SD=8) and ranged from 18 to 67 years old (Table 1).

Seventy-two percent (1256) of participants reported having been tested for HIV. Of these, 1112 reported knowing the results, 70% reported testing negative, 27% reported testing positive, 1.5% of the test results were inclusive and 1% of the participants refused to reveal their status.

The mean age of first injection was 19 (SD=5.5). Respondents had long histories of injecting drug use; the average length of injection was 16 years (SD = 9.0) (Table 2). Only 59 respondents (3%) had been injecting for 1 year or less. Participants had been using syringe exchange for an average of 5.8 (SD= 9.0) months, and reported an average use of 15.2 (SD= 13.1) syringes per week. Respondents reported obtaining 14 (93%) of their syringes from the exchange per week (Table 2).

Participants reported that in the 30 days prior to using the syringe exchange, they injected with previously used works an average of 11.6% of the time compared to 3.9% of the time in the last 30 days while using the needle exchange (p<.001). Prior to using the exchange, respondents reported injecting an average of 95.2 times per month compared with 85.6 times per month during the last 30 days (p<.0001). There was a significant decrease in the number of participants who reported injecting with syringes used by others in the last 30 days compared with 30 days prior to using the exchange: with rented or bought used syringes, 21.7% prior, 5.8% current; with borrowed works, 29.3% prior, 12.1% current. The number of IDUs who reported using alcohol pads to clean their skin when injecting increased significantly, from 33.0% prior to 80.0% current (Table 3).

A total of 1337 participants responded to questions about renting or buying used syringes both 30 days prior to using the exchange and during the past 30 days. Of these, 1047 (78%) did not rent or buy used works prior to using the exchange and 290 (22%) did. In the last 30 days since using the exchange, 1259 (94%) did not inject with works used by others and 78 (6%) did. Among the 1047 who did not practice this behavior previously, only 11 (1%) began to do so. In comparison, among the 290 who did rent or

buy used works prior to using the exchange, 223 (77%) no longer did (McNemar Chi square, p<.001).

A total of 1328 participants responded to questions about borrowing or using used works. Of these, 939 (71%) did not practice these behaviors in the 30 days prior to using the exchange and 389 (29%) did. Among those who did not, only 23 (2%) initiated this behavior. In the last 30 days while using the exchange, 1167 (88%) reported not borrowing or using used works and 161 (12%) did. In comparison, among those who previously borrowed or used works, 251 (65%) have not in the last 30 days (McNemar Chi square, p<.001).

A total of 1269 participants responded to queries about using alcohol pads while injecting. Of these, 851 (67%) did not use alcohol pads prior to using the exchange and 418 (33%) did. Among those who did not, only 9 (2%) still do not. During the last 30 days while using the exchange, 1015 (80%) reported using alcohol pads, and 254 (20%) reported not using them. Among those who did not previously use alcohol pads, 606 (71%) have begun since using the exchange (McNemar Chi square, p<.001).

Of the 1752 participants interviewed, 60% reported that they had anal, oral, or vaginal sex in the last 30 days. Table 4 presents the sex risk behaviors of these 1055 sexually active participants.

In contrast to questions about the use of bleach by IDUs to disinfect syringes (Center for Disease Control, 1982; Contoreggi et al., 1992; Vlahov et al., 1991), there is no need to conduct research on the efficacy of commercially manufactured needles and syringes in preventing HIV transmission. If only one person uses the needle and syringe, the chances of HIV transmission are as close to zero as anything in epidemiology. The effectiveness of syringe exchanges in reducing HIV transmission among injecting drug users is thus solely a question of the ability of the service providers to provide exchange services and the subsequent behavior of the persons utilizing the services.

The ability to provide services is largely determined by the specific operational procedures of the syringe exchange program, which in turn are substantially determined by the other organizations within the syringe exchange system and the availability of volunteers, and indirectly determined by a wide variety of supporters and opponents of syringe exchange. The New York syringe exchange system is the first attempt to establish a multi-exchange system within the United States. If syringe exchanges are to be an important factor in preventing HIV infection in the United States, it is very likely that they will operate within state-wide systems (or possibly a federal system). The systems issues that have arisen in New York are thus likely to be replayed in other states and possibly at a federal level.

The different organizations involved in the New York syringe exchange system have many differences in their organizational cultures. The potential for miscommunications and organizational cultural conflict, or serious conflicts of interest cannot be underestimated. That most of the problems have been successfully resolved so far is

something we attribute to the skills and good will of the individuals who have been devoting many long hours to the system issues.

There also have been recurrent problem themes, such as the conflict between having tightly regulated, legally compliant exchanges and reaching the largest number of IDUs per dollar of funding. This conflict also occurs for single, isolated exchanges, but is much more complicated within a system of exchanges. (The smallest exchange in the New York system currently has 85 enrolled participants, while the largest has 5,888 enrolled participants.)

The effectiveness of the exchange will also depend upon the behavior of the participants. Similar to other studies (Buning, 1989; Hagan et al., 1991; Hart et al., 1989), participants in the NYC syringe exchange have demonstrated a reduction, but not elimination, of some high risk behaviors. This is particularly evident for syringe sharing behaviors. Overall the data indicate that there has been a significant reduction in renting and buying used syringes and a marked decrease in borrowing used works. These findings are consistent with international studies which have shown that exposure to harm reduction strategies and access to clean equipment reduces some high risk behaviors associated with HIV infection. However, despite the evidence of drug-related risk reduction among participants of syringe exchanges in NYC, sexual risk reduction still lags behind (Abdul-Quader et al., 1990; Deren et al., 1993; Des Jarlais, 1992). Our data indicate that participants are having unprotected sex more than half of the time, whether with primary or casual partners. This suggests that syringe exchange programs may need to expand their efforts at sexual risk reduction education, but will more than likely require increased funding in order to accomplish this expansion. Research is still needed which will explore the determinants of consistent use of condoms among this population.

These data from the first year of the NYC syringe exchange evaluation, are consistent with data from other studies which have shown that syringe exchanges are not attracting new injectors (Des Jarlais et al., 1988). Our data indicate that the majority of IDUs attending syringe exchange are long time injectors (average length of injection 16 years) and that only 59 (3%) have been injecting for less than one year.

The most important single datum for assessing the New York syringe exchanges will obviously be the HIV seroconversion rate among the participants. Previous studies have shown seroconversion rates from 6 to 11 per 100 person-years at risk among persons continuing to inject illicit drugs in high seroprevalence areas. We will eventually be able to compare the rate observed among New York syringe exchangers to these other studies, with appropriate multivariate controls. The important question, however, will not be a statistically significant difference, but an epidemiologically substantial difference. It is quite possible that in an area like New York, both more extensive and more intensive prevention services will be needed in addition to syringe exchanges.

This study was funded by the American Foundation for AIDS Research (AMFAR). We would like to acknowledge all syringe exchange staff who have cooperated with this

evaluation and who have provided access to the participants, our interviewing staff, Judith Milliken for her work on this manuscript, and Lee Kochems for his comments.

ANDREW MOSS

Andrew Moss observed that almost a decade has passed since the first open operation of needle exchanges, yet the issue is still being hotly debated and the same arguments for and against needle exchanges are still being made. Now, however, much more is known about the operation and impact of individual needle exchange programs, and systematic studies of needle exchanges are beginning to emerge. He noted that, first, needle exchanges attract a large clientele. Second, they tend to attract older users, many of whom have long histories of injection drug use. Third, they do not seem to increase injection drug use in any way that can be observed or measured. Fourth, they may reduce needle sharing among injectors. Fifth, data are lacking—and may never be available—with which to isolate the impact of needle exchanges on HIV incidence from other factors. He remarked with interest on the finding that the Tacoma needle exchange had an individual-level protective effect against needle-borne transmission of hepatitis-B. Sixth, the data presented do not all point uniformly in the same direction: in particular, although needle exchanges attract older injectors, the data from the San Francisco presentation indicate a decline in sharing associated with needle exchange use by younger, not older, people. In addition, although many of the needle exchanges attracted a largely white clientele, African-Americans are known to be a high-risk group among injectors.

Given the data, Moss asked, should a push be made for large-scale federal and state funding of needle exchanges? His own response to the question was a qualified yes, for two reasons: first, the United States does not have a unified strategy for reducing HIV infection among injecting drug users, despite the fact that injection drug use is a major HIV risk factor in this country. Second, other serious diseases prevalent among the injector population, especially tuberculosis, are associated with HIV transmission, and reducing the incidence of HIV will have a positive impact on the incidence of those diseases as well. If the United States does move to the big model of federal and state funding of needle exchanges, Moss pointed out, the question then becomes how to do it in a way that does not destroy what has largely powered needle exchanges to date—social activism by people who have knowledge of and sympathy for people who inject illegal drugs.