Comprehensive, up-to-date information on HIV/AIDS treatment, prevention, and policy from the University of California San Francisco

Injection drug use-related HIV infection accounts for a substantial and growing fraction of the total cases of HIV in the United States. Injection drug use has spread HIV rapidly also in parts of Europe and Asia. Risk factors for HIV infection among injection drug users (IDUs) include demographic characteristics and practices that increase sexual transmission and parenteral exposure to infected blood. This chapter describes the epidemiology, prevalence, and incidence of HIV infection among IDUs in the United States and abroad, and preventive interventions, including addiction treatment, risk-reduction education, the use of disinfectants, and the provision of sterile injection equipment.

In 1981, early in the HIV epidemic, it became apparent that the disease was spreading among IDUs in the metropolitan New York City area. Opportunistic infections identical to those in homosexual men with the newly recognized syndrome termed AIDS were being diagnosed in IDUs with no history of male-to-male sex.(1) Shortly thereafter, the same opportunistic infections began appearing in non-drug-using female sexual partners of male IDUs.(2) This evidence suggested that the immunodeficiency syndrome was caused by an infectious agent similar or identical to that afflicting homosexual men with AIDS, and that the agent was initially acquired by the male partner through parenteral drug use and then transmitted sexually. Investigators eventually identified HIV as the etiologic agent, and quickly determined that HIV could be isolated from blood-contaminated needles, syringes, and injection paraphernalia, providing a biologic rationale for transmission of HIV among IDUs.(3)

Injection drug use is reported as a risk factor in a substantial and steadily increasing proportion of Americans diagnosed with HIV and is largely responsible for heterosexual and perinatal transmission of HIV (Figure 1). Through 1985, 17% of AIDS cases reported to the Centers for Disease Control and Prevention (CDC) had been among IDUs.(4) During 1995, reports described injection drug use among at least 30% of the 73,380 adults and adolescents diagnosed with CDC-defined AIDS in the United States: 26% were female or heterosexual male IDUs, and 5% reported both male-to-male sex and injection drug use.(5) An additional 3.5%, most of whom were women, reported heterosexual activity with an injection drug-using partner as their only risk factor.

In the United States, injection drug use is directly or indirectly responsible for the majority of HIV infections among women and children. Of women reported to the CDC as having AIDS,47% have been IDUs themselves and another 42% have acquired the disease heterosexually. At least 88% of pediatric cases of HIV disease have occurred in children of women with or at risk for HIV.(5) Injection drug use is also a more prevalent risk factor among racial and ethnic minorities than among American whites. In comparison with their representation in the population at large, African-Americans and Hispanics are over represented by factors of two to threefold among American IDUs with HIV disease.(6) By 1990, while only 11.5% of Americans were African-Americans, they comprised 28.8% of those diagnosed with CDC-defined AIDS; similarly, Hispanics accounted for 6.5% of the American population, but 16.2% of cumulative CDC-defined AIDS cases. Further, more than 36% of African-Americans and Hispanics diagnosed with CDC-defined AIDS had been heterosexual IDUs, as compared with only 8% of non-Hispanic whites diagnosed with CDC-defined AIDS.(5) Finally, in the United States and other industrialized countries, where cases to date have occurred primarily among homosexual men and IDUs, transmission of HIV among and from IDUs represents the most threatening conduit for spread of HIV to non-injecting, heterosexually active populations.

HIV seroprevalence surveys among at-risk populations of asymptomatic IDUs are subject to the serious methodologic challenges inherent in all descriptive studies of IDUs. These challenges include establishing the overall prevalence of injection drug use in a given community as the population denominator for disease incidence and prevalence calculations, and recruiting a representative sample of that population. Furthermore, while clinical AIDS data are more readily obtainable than HIV seroprevalence data, the median time from HIV infection until HIV disease produces CDC-reportable AIDS is estimated to be approximately 10 years.(7) Changes in the demographics of AIDS cases reported to the CDC thus lag far behind changes in HIV disease prevalence. Nevertheless, the epidemiology and determinants of HIV transmission among IDUs are reasonably well characterized in the United States and abroad.

HIV was probably introduced around 1976 into the New York metropolitan area, the North American epicenter for HIV infection in the IDU risk group. Sera collected from New York City IDUs entering drug treatment, and stored until a radioimmunoprecipitation assay made possible cross-sectional serosurveys of HIV prevalence, showed that in 1978, 9% of specimens (1 of 11) were already seropositive.(8) Once introduced into this community of IDUs, HIV spread with alarming rapidity, infecting a majority of local IDUs in a period of a few years (Figure 2).(8) This pattern suggested that HIV transmission via infected injection paraphernalia may be highly efficient, particularly compared to sexual transmission in the developed world.(9) While there was considerable uncertainty about such early point estimates, later samplings were more precise. Seroprevalence in New York rose quickly to 26% in 1979, 38% in 1980, and 50% during the period 1981 to 1983, reaching a plateau by 1984 at 55 to 60% (see Figure 2).

Reports described similar patterns of rapid, localized spread of HIV among IDUs in Europe. In Edinburgh, Scotland, seroprevalence rose rapidly to 51% in the early 1980s, with most of the seroconversions occurring within a 2-year period.(10) In Milan, Italy, seroprevalence increased from 5% to almost 50% of IDUs in less than 3 years and had reached 62% by early 1987;(11) the incidence of new HIV infection was estimated to have stabilized at 4% per year between 1987 and 1991.(12)

HIV has spread with startling speed among IDUs in Asia, where seroincidence may be up to an order of magnitude higher even than it was in New York in the early 1980s. In Bangkok, Thailand, seroprevalence among IDUs was estimated at only 1% in late 1987 but had risen to 15% by March 1988 and to 43% by December of the same year, for an average seroconversion rate of 4% per month.(13) In the northeastern Indian province of Manipur, estimated seroprevalence among the region's 15,000 IDUs rose from 0 to 50% over 6 months from 1989 to 1990.(14)

In some American metropolitan areas, the spread of the virus has been much slower, particularly since its recognition as the causative agent of AIDS. In San Francisco, Moss et al. measured seroconversion rates in IDUs who had repeat HIV testing during studies at methadone maintenance sites between 1985 and 1990.(15) The seroconversion rate was 1.9% per person-year in 681 heterosexual IDUs. Seroconversion occurred at a higher rate in African-Americans (4.0%) than in whites (1%). In a multivariate model, seroconversion was associated with more than five sex partners, injection drug use in a "shooting gallery," and short duration of methadone maintenance.

In 1996, after reviewing 350 published and unpublished documents and several large national and state datasets, Holmberg estimated HIV prevalence and incidence among various risk groups in 96 large U.S. metropolitan areas, and then extrapolated these figures to derive national estimates of HIV prevalence and incidence.(16) Holmberg estimated approximately 1.46 million injection drug users in the United States; calculated that perhaps 204,000 of them are HIV seropositive; and concluded that approximately 19,000 more become infected annually (range: 7,600 to 28,500). This method depends heavily on notoriously difficult estimates of the size of the at-risk populations.(17) Nevertheless, Holmberg estimated an overall national seroprevalence of 14% HIV infection among IDUs and an incidence of 1.5%, or 1.5 new infections per 100 person years, among IDUs.

HIV seroprevalence among IDUs in the United States varies markedly by geographic region, and by proximity to major metropolitan centers. The highest prevalence (41%) and incidence (5%) rates are found in the metropolitan corridor stretching from Boston to Washington D.C., in Miami, and in San Juan, Puerto Rico.(16) Prevots et al. also described wide geographic variation in seroprevalence rates among IDUs,(18) with the highest rates noted in the Northeast (27%) and lower rates in the South (12%), Midwest (7%), and West (3%). In 1984, 58% of 86 IDUs entering drug detoxification programs in New York were seropositive.(19) In neighboring northern New Jersey, seroincidence and seroprevalence have varied dramatically and directly with proximity to New York City. A 1984 study of more than 900 IDUs in drug treatment in New Jersey demonstrated a 50% HIV seroprevalence among those living within 5 miles of New York, versus 2% among those living more than 100 miles away.(20)

Although IDUs may account for a growing proportion of the HIV epidemic, a number of investigators have observed a stabilization or small decline in seroprevalence. Prevots et al. found that HIV seroprevalence has stabilized in most urban centers, with moderate declines among white IDUs.(18) Watters also found that San Francisco seroprevalence stabilized from 1987 to 1992 after an initial doubling from 1986 to 1987.(21) There are reports of modest declines in seroprevalence in the largest drug-using communities such as New York, New Jersey, and Los Angeles.(22-24) Furthermore, Wiebel et al. measured a decline in seroconversions in out-of-treatment IDUs serially tested between 1988 and 1992.(25) In this study, seroconversions decreased from 8.4 to 2.4 per 100 person-years. Collectively, these studies indicate a wide geographic variation of seroprevalence of HIV in IDUs in the United States, and that rates have stabilized in the highest seroprevalence areas and declined in other areas.

Transmission of HIV among IDUs occurs primarily through HIV-infected blood contamination of injection paraphernalia, which is re-used by an uninfected IDU. Behaviors that increase the likelihood, frequency, and magnitude of exposure to infected blood increase the risk of infection. Among IDUs, several demographic and behavioral characteristics are associated with greater risk of acquiring HIV. Residence in or near an area of high prevalence is a powerful risk factor, both domestically as just described and worldwide. Within the United States, other consistent independent predictors of HIV seroprevalence among IDUs include minority race or ethnicity, low income, male gender, and a diagnosis of antisocial personality disorder.(8,26) African-American and Hispanic IDUs have significantly higher seroprevalence than do white, non-Hispanic IDUs in New York, Connecticut, and San Francisco.(27-29) Gender differences in HIV infection vary geographically, from New York, where seroprevalence is 13% greater among male than among female IDUs, to San Francisco, where the seroprevalence among men and women IDUs is identical.(27,30) Antisocial personality disorder, which may be prevalent in 40% of IDUs as compared with 3% of the general population, is associated with more frequent injection, more frequent needle sharing with a greater number of partners, and greater use of alcohol and cocaine.(26)

Until recently, no biologic differences in susceptibility to acquiring HIV had been demonstrated to explain the different rates of infection among different demographic, racial, ethnic, and gender groupings. Recent reports describe a nucleotide deletion with the beta-chemokine receptor 5 (CCR5) gene that is strongly associated with remaining uninfected despite frequent sexual exposure to HIV.(31) This allele was found in 8% of white gay men enrolled at the Chicago site of the Multicenter AIDS Cohort Study, but in no participants of African or Asian descent. It is likely that other, as yet undiscovered genetic correlates of race, ethnicity, and gender that confer resistance or susceptibility will explain some of the variability in infection rates.

Behaviors related to drug use that are associated with increased risk of acquiring HIV in the United States and Europe include frequency of drug injection, sharing injection equipment, and high-risk sex behavior. The frequency of drug injection per se is a significant risk factor. An early study in New York showed a seroprevalence of 21% among IDUs who denied needle use in the past month, 31% among those injecting one to five times per month, 47% in IDUs injecting 6 to 45 times per month, and 61% in those injecting more than 45 times per month.(32)

Foremost among risk factors is the sharing of needles, syringes, and other injection equipment. Sharing is a common practice among IDUs worldwide. Although some researchers have emphasized the cultural or social value attached to sharing by communities of IDUs,(33) studies suggest that needle sharing is largely driven by legal and economic necessity. For example, diabetic IDUs in Baltimore are only half as likely to be infected with HIV as are comparable nondiabetic IDUs, probably because diabetics have ready access to clean injection equipment.(34) Possession of injection equipment is a criminal offense in virtually all jurisdictions in the United States, and sharing equipment reduces the risk of being apprehended with "works" in their possession.(35) When available, reliably sterile equipment is often prohibitively expensive. A study of 110 IDUs in New York found that peer-group behavior, attitudes conducive to sharing, economic motivation, not owning injection equipment, and fatalism about acquiring HIV disease predicted needle sharing.(36)

Seroprevalence correlates with reported needle-sharing episodes. A study in Edinburgh demonstrated a seroprevalence of 30% among IDUs who denied currently sharing equipment, compared with 56% among those who reported sharing sometimes, and 75% among those who said they usually shared.(10) Among 1,363 Italian IDUs, HIV seroprevalence was 22% among those who reported never sharing needles as compared to 67% among those who said they always did so.(37)

Sharing with multiple injection partners is an additional risk factor. Having shared equipment with two or more people in the past 6 months was a significant predictor of HIV infection in Bangkok.(13) In San Francisco, HIV seroprevalence was 3% among IDUs who denied regular needle sharing, 9% among those who said they regularly shared with only one other person, and 15% among those regularly sharing with two or more people.(29) Studies consistently demonstrate that a significant risk factor for HIV infection is drug use at "shooting galleries," sites where drugs are sold and injection equipment is made available for rental or sharing.(28,32,33) Dozens or hundreds of IDUs may visit such a gallery in a single day, sharing a limited number of "works."

The frequency of needle sharing, the number of partners with whom paraphernalia are shared, the probability that those partners are HIV-infected, and the manner in which sharing occurs all influence HIV risk. Using a mathematical model based on seroprevalence rates in New Haven, and the prevalence of HIV in syringes collected in the local needle exchange program, Kaplan and Heimer estimated that the average risk of infection from an HIV-infected apparatus was 0.0067 per injection episode, or approximately one in 150.(9) For a seronegative IDU who always uses needles after a seropositive partner, this situation would translate to a 49% risk of becoming infected during the first 100 injection episodes.

Other specific drug-sharing practices increase the risk of acquiring HIV. "Frontloading" is a practice in which an IDU uses a syringe to measure out a quantity of drug for someone else's use and then squirts the drug directly into that person's apparatus.(38) At other times, IDUs may share the equipment in which their drugs are prepared, the water with which it is mixed, or the cotton through which it was filtered, without sharing the actual needle or syringe. HIV can be transmitted by contamination of these materials as well. Koester et al. found that 72% IDUs in Denver had practiced these "indirect" methods of apparatus sharing in the last 30 days,(39) twice the rate of direct needle sharing, which suggests that the potential risk of indirect sharing was unrecognized by most IDUs. Furthermore, contamination of injection equipment, and subsequent HIV transmission, can occur with any parenteral exposure and does not require intravenous puncture by either needle-sharing partner. Drug use by subcutaneous ("skin popping") or intramuscular ("muscling") injection can transmit HIV as well, although IDUs may erroneously believe they are safer if they avoid intravenous exposure.

In the United States, the use of cocaine, either in injectable form(27,30) or smoked as "crack," is associated with higher rates of HIV infection. A study of 2,597 IDUs in and around Baltimore showed that the odds of being HIV-seropositive were three times higher for cocaine injectors than for IDUs who did not use cocaine.(40) In a seroincidence study in San Francisco, all 11 HIV seroconversions in a population of 380 IDUs occurred among those who injected cocaine.(41) Cocaine users generally inject more frequently than do opiate users, and are more likely to engage in high-risk injection practices such as "booting," whereby blood is repeatedly drawn up into the syringe and mixed with drug.(26) Furthermore, in some cities, including San Francisco, noninjecting crack cocaine smokers have HIV seroprevalence rates comparable to those among opiate injectors,(42) which may reflect the common exchange of crack cocaine for sex.(43) Finally, no effective modalities have been developed to treat stimulant dependence, leaving little means by which health-care providers may assist cocaine addicts to cease or reduce their injecting.

Sexual transmission from or among IDUs also constitutes an important avenue for acquiring HIV. For female IDUs in the United States, having a male sexual partner who is also an IDU is an independent risk factor for being infected with HIV.(27) In Milan, IDUs with an HIV-seropositive sex partner are up to 4 times more likely to seroconvert after controlling for syringe sharing.(44) Male IDUs with a history of syphilis or recent sexually transmitted disease and female IDUs with more than 3 sex partners in the last 6 months are more likely to seroconvert.(45) Frequent trading of sex for drugs or money is an independent risk factor for HIV infection among female IDUs, after controlling for duration of drug use, use of intravenous cocaine, and sexually transmitted diseases.(46) Homosexual and bisexual male IDUs are at particularly high risk for HIV infection, not only via sexual transmission but also because of the high prevalence of HIV in their needle-sharing peer group.

Because HIV transmission requires both risk behavior and an infected partner, the risk of becoming infected depends on one's probability of encountering an HIV-infected injection or sexual partner. Not only risky sexual and injection behaviors, but also the risk status of injection and sexual partners is an important predictor of HIV infection. IDUs with a high risk personal network, defined as having sex or using drugs with someone who injects greater than once daily, or with someone more than 10 years older than the subject (and therefore at greater risk of past infection), are more than twice as likely (43% vs. 18%) to be HIV infected, even after controlling for syringe sharing, race, injection in shooting galleries, and sexual behaviors.(47)

Infection tends to cluster in isolated "pockets of infection." Particularly in relatively low seroprevalence areas, the likelihood of infection may be more strongly predicted by exposure to members of high prevalence clusters than by specific risk behaviors. Consecutive serum samples from 6882 IDUs in high and low seroprevalence cities were compared for HIV seroconversion.(48) In high prevalence cities, where the probability of encountering an HIV-infected partner is more uniformly high, only behavioral variables such as injection with contaminated needles and unprotected sex predicted seroconversion. In contrast, in low prevalence cities, the characteristics of one's immediate sexual partners and drug use contacts were better predictors of seroconversion than were actual risk behaviors. This finding was true even for contact characteristics that did not themselves confer high risk, such as injecting outdoors, receiving drug treatment, non-Latino ethnicity, and woman-to-woman sex. Therefore, interventions aimed at preventing HIV transmission in a low prevalence area should be tailored to relatively high seroprevalence local subpopulations, whose characteristics may be unique to that area.

Intervening to prevent or reduce the spread of HIV among IDUs is a major public health concern. Hampering prevention efforts are the social alienation, political disenfranchisement, and inaccessibility of those most at risk; the general limitations in funding for health promotion programs; and the illegality of and negative public attitudes toward drug use. Furthermore, determining the effectiveness of potential interventions is fraught with the difficulties of doing research among IDUs. Given the lethality of HIV infection and the high risk to which many IDUs are exposed, undertaking placebo-controlled trials of any intervention considered likely to be effective in reducing transmission raises serious ethical concerns.

Along with cessation of high-risk sexual behavior, cessation of drug injection is the most reliable way to eliminate the risk of transmission. As noted previously, no effective pharmacotherapy for stimulant addiction is known, but cocaine and amphetamine use, in some cases, can be treated by drug-free techniques that rely on counseling, group support, and social restructuring.(49) Opiate users in methadone maintenance treatment report greater rates of cessation of heroin use and up to 50% reduction in needle sharing, and have lower mortality rates, than IDUs not receiving methadone maintenance.(50-59) The proportion of people continuing to inject declines the longer users remain in treatment. Predictors of continued drug injection include cocaine use, inadequate methadone dosage, and the lack of a stable home life, job, and income.(60) In the United States and around the world, the demand for effective treatment for drug dependency far outstrips the supply of available treatment slots. Expansion of these services may help to reduce the spread of HIV.

One measure advocated to reduce HIV transmission in IDUs is the use of household bleach (5% sodium hypochlorite) to sterilize used injection paraphernalia. Laboratory studies have demonstrated that a minimum contact time of 30 seconds with undiluted bleach is necessary to reliably sterilize needles and syringes contaminated with HIV-infected blood.(61) Full-strength bleach is necessary because diluted bleach is ineffective.(62) Although bleach may be biologically effective under laboratory conditions, its efficacy in reducing HIV transmission has been difficult to demonstrate. When IDUs are videotaped during syringe disinfection, they maintain a median syringe contact time with bleach of 16 seconds, and only 20% had a contact time of more than 30 seconds.(63) Field studies have failed to demonstrate a statistically significant protective effect of using disinfectant bleach.(64,65) Although the field efficacy of bleach disinfection has not been clearly demonstrated, disinfection of contaminated injection drug paraphernalia ought to be encouraged when sterile syringes are unavailable.

Decriminalizing the possession of injection equipment may reduce the perceived need to share needles. Gostin et al.(35) recently reviewed the laws and regulations governing the sale and possession of needles and syringes in the United States. They point out that existing legal restrictions deter physicians, pharmacists, and public health professionals from providing HIV prevention services to active drug users. After Connecticut partially deregulated the sale and possession of syringes, pharmacy sales of sterile syringes to IDUs increased and the multiperson use of contaminated syringes fell.(66) Gostin et al.(35) therefore propose: declaring that sterile syringe use serves the legitimate medical purpose of preventing infections; modifying drug paraphernalia laws to permit the purchase of syringes from reliable sources; repealing laws and regulations forbidding over-the-counter pharmacy sales of syringes; educating health and law enforcement professionals about the prevention of blood-borne infections; and permitting local jurisdictions to establish needle and syringe exchange programs.

The introduction of needle exchange programs in many cities has probably been at once the most effective and the most controversial intervention to reduce HIV transmission among IDUs in the United States. Needle exchange programs typically provide a one-for-one exchange of used syringes and needles for sterile equipment. The important potential consequences of needle exchange studied thus far include: changes in injection and sexual risk behavior, impact on rates of drug use, prevalence of contaminated syringes in the community, impact on hepatitis B seroprevalence, changes in HIV seroprevalence, and changes in HIV seroincidence. These studies are extensively reviewed elsewhere.(67)

The primary goal of needle exchange programs is to reduce needle sharing. In 1993, the Centers for Disease Control commissioned the University of California to evaluate the impact of needle exchange programs through a review of the published literature and a series of site visits to various programs.(68) The University of California report concluded that 10 of 14 credible studies demonstrated a reduction in needle sharing frequency; the other 4 showed no effect. A number of studies both in the United States and abroad are consistent with the overall University of California finding that needle exchange programs are associated with reduced high-risk injection behavior.(68) In San Francisco, Watters reported a 47% decline in needle sharing after introduction of a needle exchange program.(69) A decline in needle sharing from 12 to 4% of IDUs was associated with a needle exchange program in New York City, with comparable reductions in buying and borrowing used needles.(22,70) Similar results have been reported from needle exchange programs in Portland, Tacoma, and New Haven.(71-73)

In England and Scotland, 79% of needle-exchange clients reported reduced frequency of needle sharing and fewer sharing partners.(74) In Amsterdam, 82% of needle-exchange clients, but only 28% of non-exchanging IDUs, reported using their "works" only a single time, suggesting that the availability of exchanged needles may have reduced sharing.(75) Only one study has suggested an increase in needle sharing. Klee and colleagues found that needle exchange participants in Manchester, England were more likely then nonparticipants to give away, but not to receive, used needles.(53) Access to needle exchange was limited at the time of this study and there were reports of participants being pressured to give away their syringes. Except for this study, the evidence suggests that needle exchange programs are associated with consistent reductions in syringe sharing.

One important benefit of needle exchange programs is the opportunity they provide to contact and educate IDUs who are not otherwise reached by preventive interventions.(76) Because most needle exchange programs use multiple interventions, including safe sex information and counseling, many studies have addressed change in sexual risk behavior after introduction of a needle exchange program. These studies have generally found less consistent reductions in sexual risk behavior than in needle sharing. The University of California Report did not find consistent reductions in sexual risk behavior. Subsequent studies in New York have found similar results,(22,70) and rates of consistent condom use among needle exchange participants have been reported as low as 32%.(77)

Nevertheless, although substantial sexual risk behavior persists, there have been reports of reductions in sexual risk behavior associated with safer sex interventions, either as part of or separate from needle exchange programs. Condom use among street-recruited IDUs in San Francisco increased from 4.2 to 28.9% concomitant with a needle exchange-based intervention program.(21) In Padua, Italy, an intensive education campaign that began in 1985 was associated with an increase in condom use from 6 to 22%.(77) Finally, although overall use may be low, condom use may be more consistent among the highest risk individuals.(78) During a 2-year follow up of mostly out-of-treatment American IDUs, Celentano et al. described a progressive decline in self-reported sexual risk behavior, with a greater reduction among HIV-positive than among HIV-negative persons.(79)

The major concern expressed regarding needle exchange programs, apart from the appearance of condoning drug use, is the possibility that the prevalence of drug use might be increased by making syringes more accessible. In the eight credible studies reviewed in the University of California report that addressed rates of or recruitment to injection drug use, four found no impact, three found decreased drug use, and one found an increase in injection drug use that was not replicated in follow-up data by the same group. Studies found that the median age of participating users and their median duration of drug use increased over time, suggesting that there had been little or no increase in new initiates to injection drug use. Other studies have also failed to find increases in drug use. Recent American data show stable or declining levels of drug use and increasing ages of needle exchange participants.(69-71,79) Studies in Britain, France, Sweden, and the Netherlands also demonstrate no apparent increase in drug use with introduction of needle exchange programs.(76) Furthermore, the prevalence of improperly discarded needles in the community does not appear to increase with the introduction of needle exchange programs.(80)

Additional evidence of a protective effect of needle exchange programs comes from hepatitis B data. Needle-related hepatitis B transmission is similar to, but more efficient than, HIV transmission. In Tacoma, the incidence of IDU-related hepatitis B infection declined dramatically soon after implementation of a needle exchange program. Sexually acquired hepatitis B incidence remained stable over the same time period.(81) A later case-control study of hepatitis B in Tacoma revealed that failure to use syringe exchange was independently associated with hepatitis B infection.(72) The risk of hepatitis B, however, although important by itself, may not parallel the risk of HIV infection. Levine et al. found that incident hepatitis B infection predicted incident HIV infection among male but not among female IDUs.(82) Garfein et al. also found that hepatitis B and C seroprevalence did not strictly parallel HIV prevalence as a function of injection drug use duration.(83)

The ultimate goal of needle exchange programs is reducing new infections. Des Jarlais and colleagues used meta-analytic techniques to examine the relation between use of needle exchange and HIV incidence in three studies.(84) The studies included 1630 IDUs recruited from waiting in line at needle exchange sites, methadone maintenance programs, and street outreach. Seroincidence was compared in participants who reported no use of needle exchange versus those who reported regular use in the 3 months before HIV testing. After controlling for age, sex, race, and frequency of injection, Des Jarlais et al. found that not using needle exchange was associated with a 3.3-fold increased risk of HIV.

Controlled data concerning HIV transmission are difficult, if not impossible, to obtain. There have been some encouraging results, however, from cities such as Edinburgh and Sydney,(85) where HIV seroprevalence initially remained roughly constant, and then fell, during the first 6 months that clean needles were available. No conclusive prospective study has been reported, but the weight of evidence thus far suggests that clean needle and syringe exchange is likely to produce a net reduction in HIV transmission. These studies indicate that needle exchange programs are accompanied by reductions in HIV injection risk behavior, are associated with reduction of seroincidence of hepatitis B and of HIV, and are not associated with increased drug use. These findings, combined with the stabilization of seroprevalence in major metropolitan areas with needle exchange programs, suggest that such programs are likely to reduce HIV infections among IDUs.

An effective approach to preventing transmission of HIV among IDUs must encompass multiple interventions based on a realistic assessment of different levels of risk. The first goal must be to reduce the number of people injecting. Despite our best preventive efforts, however, some level of injection drug use will inevitably persist. The next goal must therefore be to reduce the frequency of needle use and of risky injection practices, such as needle sharing. Finally, as these practices will not be eliminated, a harm-reduction approach requires making each injection episode less risky and reducing the risks of sexual transmission. Prevention efforts must therefore begin with efforts that prevent the initiation of injection drug use, particularly among youth. Simultaneously, however, interventions must be aimed at those who are already injecting or who will begin to do so. One target must be the development and provision of adequate treatment for the addictions to all who need it. Another must be the implementation of education campaigns that are culturally appropriate and that inform high-risk and seropositive IDUs of their risks, with the goals of recruiting them into drug treatment, encouraging serologic testing and medical follow-up, and promoting behavioral changes. Changes should be considered to statutes that criminalize possession of drug paraphernalia. Finally, programs designed to facilitate and support safer behaviors must supplement education. These programs may include, but should not be limited to, the free provision or exchange of sterile injection equipment under controlled circumstances, and the distribution of bleach and condoms.