In the United States, HIV incidence among injection drug users (IDUs) seems to be declining, most likely as a result of comprehensive harm reduction measures, including needle exchange and substance abuse treatment.1-10 There are few cohort data beyond 1998, however, with more current studies having been cross-sectional, using the serologic testing algorithm for recent HIV seroconversion (STARHS) to estimate incidence.7,11,12 Although the STARHS represents a reliable method for ascertaining the prevalence of recent infection, data from cohort studies can substantiate observed trends and provide information on risk factors. We previously reported a declining incidence of HIV from 1988 to 1998 among IDUs in Baltimore, Maryland, which was partially explained by parallel declines in drug-related and sexual risk behavior.10 The objective of this report was to examine recent trends in incidence (1999-2004) in this same cohort of IDUs in Baltimore.
Between 1988 and 1989, 2946 IDUs from Baltimore were enrolled into the AIDS Linked to the Intravenous Experience (ALIVE) study through community outreach and followed at 6-month intervals, as previously described.13 All participants acknowledged nonmedical injection drug use within the prior 11 years, were 18 years of age or older, and were free of AIDS at entry into the study. In 1994 and 1998, 436 and 245 participants were recruited, respectively. The study was approved by the Johns Hopkins University Institution Review Board (IRB), and all subjects provided written informed consent.
Subjects were eligible for this analysis if they were HIV-seronegative at enrollment and returned for at least 1 follow-up visit (n = 1983). HIV incidence was prospectively ascertained among these individuals by testing for HIV-1 antibodies at each visit using a commercially available enzyme-linked immunosorbent assay (Genetic Systems, Seattle, WA) and was confirmed by Western blot analysis (Dupont, Wilmington, DE). All subjects were counseled by trained personnel at each study visit about the risks for HIV and other infections and were given referrals to drug treatment programs. Of 331 observed HIV seroconversions, 27 had an interval >3 years between their last HIV-negative visit and first HIV-positive visit and were excluded because the date of seroconversion could not be accurately determined. Of the 304 remaining HIV seroconversions, 277 were previously described,10 5 occurred before 1998 but were not included in the earlier analysis, and 22 occurred between 1999 and 2004.
At enrollment and at semiannual follow-up visits, a standardized interviewer-administered questionnaire collected information on demographics, medical care (including types and numbers of health care encounters), illicit drug use (including types and frequency of drug use, needle sharing, and shooting gallery use), and sexual risk behaviors (including types and numbers of partners, condom use, anal intercourse, and trading sex for drugs or money) as well as drug treatment, including methadone maintenance and use of needle exchange programs in the prior 6 months. In August of 1998, audio computer-assisted self-interview (ACASI) for collection of risk behavior information was introduced beginning in randomized groups through 2001 and across the full cohort afterward.14
Person-time techniques were used to calculate HIV incidence density between February 1988 and December 2004. The date of HIV seroconversion was estimated as the midpoint between the last antibody-negative and first antibody-positive HIV tests. HIV incidence was further calculated by active drug injection status. Strata for active drug injection were defined by collecting all time intervals for which drug injection was reported in the prior 6 months. For each stratum, Poisson regression models were fitted separately using data from 1988 through 1998. The models assume a linear decline in the logarithm of the annual incidence rate as previously described.10 Based on these models, expected population incidence rates with 95% confidence intervals (CIs) and the number of expected HIV seroconversions for 1999 through 2004 were calculated. The assumption of linear temporal trend was tested by fitting a piecewise linear Poisson model allowing for different slopes in calendar periods 1988 through 1992, 1993 through 1994, and 1994 through 1998. No significant differences between the values of adjacent slopes were found; thus, the assumption of linearity was confirmed to be valid. As previously described, multivariate Poisson regression models were fit to determine independent contributions of covariates of interest.10 Interactions between calendar period (1988-1992, 1993-1998, and 1999-2004) and age, gender, education, and sexual and drug use risk variables were examined to determine if the relative contribution of any particular risk factor changed over time. Analyses were performed using SAS version 8.2 (SAS Institute, Cary, NC).
Of the 1983 initially HIV-seronegative participants, 1505 (76%) were male and 1801 (91%) were African American. The median age at enrollment was 35 years (interquartile range [IQR]: 30-40 years), and the median duration of injection drug use was 14 years (IQR: 6-20 years). The median follow-up time was 6.6 years (IQR: 1.9-12.5 years).
A total of 304 individuals seroconverted to HIV in this analysis over 14,770 person-years (PYs) of follow-up (overall incidence rate of 2.06 per 100 PYs, 95% CI: 1.83 to 2.31). HIV incidence per 100 PYs declined over time from 4.57 (95% CI: 3.12 to 6.71) in 1988 to 0.53 (95% CI: 0.16 to 1.59) in 2004. Overall trends of declining HIV incidence were similar for those recruited into the initial cohort compared with those recruited in later periods and for men compared with women, except for 2003 through 2004, when incidence rates were higher in the later enrollees and women experienced an increased incidence per 100 PYs from 0 in 2002 to 1.61 in 2003 and to 1.20 in 2004. HIV incidence was highest among individuals <30 years of age (5.30 per 100 PYs) and declined steadily with age through 54 years (0.49 per 100 PYs), after which it increased to 1.28 per 100 PYs among individuals 60 years of age and older.
HIV incidence rates were examined separately for time intervals when injection drug use was and was not reported. In both strata, a general decline in incidence was observed through 2002. Among active drug injectors, however, the seroconversion rate increased in recent years from 0 in 2002 to 2.59 per 100 PYs (95% CI: 1.17 to 5.77) in 2003. This trend did not continue in 2004, when no seroconversions were reported among those who reported injecting drugs in the 6 months before their first positive visit. The incidence of 2.59 per 100 PYs in 2003 was significantly higher than what would have been expected if the linear decline that was observed in our previous analysis (1988-1998) had continued (expected incidence of 1.08, 95% CI: 0.51 to 1.65; Fig. 1). By contrast, the observed incidence was significantly lower than expected from 1999 to 2002 (all observed incidence rates were below the lower 95% confidence limit).
In univariate analysis, risk factors for HIV infection in the recent period (1999-2004) were not substantially different from those previously reported in 1988 through 1992 and 1993 through 199810; however, multivariate analysis for the current period alone was not performed because only 22 cases were observed from 1999 through 2004. In a multivariate model that included all seroconverters between 1988 and 2004, there was a suggestion of an interaction between calendar period and gender, with women having greater risk than men in the latest period (1999-2004; Table 1). Estimates for other covariates were similar to those previously published from 1988 through 1998.10
Because of the increase in 2003, we further characterized subjects who seroconverted in 2003 through 2004 (n = 9). All were African American, 4 were male, 5 were female, and they ranged in age from 33 to 59 years. Six (all from 2003) reported injection drug use in the interval when they seroconverted, of whom 4 (67%) reported sharing needles with multiple partners (range: 2-12). Only 1 subject reported attending a shooting gallery. Overall, 3 reported sexual activity in the interval (only 1 of whom had also reported injecting), but only 1 reported risky sexual behavior (eg, inconsistent condom use, a sexually transmitted disease [STD], and crack cocaine use) in the interval before seroconversion. Two (both from 2004) reported no drug-related or sexual risk factors.
We also examined trends in drug-related and sexual risk behavior over the 16 years of the study, which were, for the most part, consistent with the trends of the previous report through 1998.10 Overall, there was a slight decline over time in reported sexual risk behaviors (eg, number of sexual partners, sex with an IDU, and trading sex for drugs; data not shown). More substantial declines were seen for the proportion of persons reporting injection drug use. Among those who reported injecting, the frequency of sharing needles also declined steadily between 1988 and 1998 from 60% to 30%; however, a slight increase to 40% in 2004 was subsequently observed. Among those who reported injecting, reported use of needles from the needle exchange increased over time from 16% in 1994 to 49% in 2004. No substantial changes were noted in the proportion of individuals accessing drug treatment, including methadone maintenance (data not shown).
It is encouraging that data from this longitudinal study continue to support the previously described trend of declining HIV incidence among IDUs.7,10,12 HIV seroconversions continued to occur in this cohort, however, despite advancing age and substantial declines in reported risk behavior. Moreover, we observed an unexpected increase in HIV incidence in 2003 that was most likely attributable to drug-related risk behavior. Of further concern is that we observed trends of increasing HIV incidence in recent years among women and older IDUs.
Although additional follow-up is needed to see if the trends are sustained, they are noteworthy for several reasons. A resurgence in HIV incidence among IDUs in Baltimore would be different from the findings of most other reports, including our own cohort, which have noted declining incidence.6,7,10,12 These earlier declines were partially attributed to programs that emphasized reductions in needle-sharing risks and included syringe exchange programs, bleach distribution to disinfect needles, and drug abuse treatment. Residual risk was attributed to sexual behaviors,15,16 and calls were made for improved efforts at sexual risk reduction programs.
Recent increases in HIV incidence have been observed in other groups of IDUs, however. Despite an overall declining HIV incidence in New York City, a recent report noted a higher HIV incidence in older minority injectors,11 which departs from observations from other researchers in the same city over the prior 6 years.5,12 A report among IDUs in England and Wales documented an increasing HIV incidence from 2000 through 2003 after a steady decline from 1990 through 199817 and suggested that the increase might be related to suboptimal dosing of methadone maintenance prescriptions.18
Periodic increases in HIV incidence have also been observed in other risk groups. Studies of men who have sex with men have shown recent increases in sexual risk behaviors and subsequent HIV incidence. The increase has been attributed to HIV prevention “fatigue” among other factors.19,20 More recently, with the introduction of highly active antiretroviral therapy (HAART), some studies have shown that reduced HIV concern or beliefs about reduced HIV transmissibility with the use of HAART are associated with relapse to high-risk behavior that has contributed to increasing HIV transmission.21,22 Although access to HAART in our cohort has lagged behind other groups, the proportion receiving HAART has risen.23 Nevertheless, data from our cohort showed no significant HAART-associated increase in drug-related risk, and relapse to sexual risk was not substantial;24 moreover, a recent meta-analysis concluded that the impact on HIV-related risk with the introduction of HAART was modest.25
Although HIV programs such as needle exchange and bleach distribution continue, the unexpected increase in HIV incidence we observed in 2003 was accompanied by a slight increase in needle-sharing behavior, an increase that was not attributable to the introduction of ACASI for capturing information on risk behaviors (data not shown). Although data to explain this increase further are not available, they do suggest that risky drug-related behaviors are continuing in this cohort despite earlier declines. Potential explanations for continued high-risk behavior include limited access to prevention programs or “safe injection fatigue.” The fact that needle exchange program use increased and methadone treatment remained stable over time argue against the first hypothesis. Prior data from this cohort in 2000 suggested that 40% of this cohort reported difficulty in practicing safer injection behaviors on a consistent basis, however, which was associated with an increased likelihood of needle sharing.22
Several limitations of our study need to be acknowledged. Because so few new HIV cases were observed in the current period, we were not able to perform a separate multivariate analysis for this period to confirm associations with risk behaviors that were observed in prior periods. The model that included all follow-up time did not differ substantially from that including data only through 1998, however. Although follow-up rates in our cohort are high for a drug-using population, losses did occur and could have influenced our results. Finally, our population represents a convenience sample of IDUs in Baltimore. The median age of our cohort at the last follow-up visit was 47 years and thus may not be entirely representative, especially of younger IDUs. Other data suggest that there is a core group of older African-American injectors in Baltimore, however, and our data may at least be representative of this group.26
Moreover, given that previous studies have consistently demonstrated HIV incidence rates are higher among IDUs <30 years of age,7,10 our figures may underestimate HIV incidence rates among younger IDUs. Our data highlight the fact that HIV seroconversions continue to occur even among older IDUs. Additional data on incidence trends and risk behaviors among younger injectors are needed to obtain a more complete picture of the current epidemic in Baltimore. In the meantime, renewed efforts are urgently needed to reach older and younger IDUs and to provide HIV prevention interventions aggressively.
1. Fleming PL, Ward JW, Karon JM, et al. Declines in AIDS incidence and deaths in the USA: a signal change in the epidemic. AIDS
. 1998;12(Suppl A):S55-S61.
2. Centers for Disease Control and Prevention. HIV and AIDS-United States, 1981-2000. MMWR Morb Mortal Wkly Rep
3. Longshore D, Anglin MD. Seroprevalence stable and seroconversions rare among injection drug users in Los Angeles. J Acquir Immune Defic Syndr Hum Retrovirol
4. Watters JK. Trends in risk behavior and HIV seroprevalence in heterosexual injection drug users in San Francisco, 1986-1992. J Acquir Immune Defic Syndr Hum Retrovirol
5. Des J, Marmor M, Friedmann P, et al. HIV incidence among injection drug users in New York City, 1992-1997: evidence for a declining epidemic. Am J Public Health
6. Des JC, Perlis T, Friedman SR, et al. Behavioral risk reduction in a declining HIV epidemic: injection drug users in New York City, 1990-1997. Am J Public Health
7. Kral AH, Lorvick J, Gee L, et al. Trends in human immunodeficiency virus seroincidence among street-recruited injection drug users in San Francisco, 1987-1998. Am J Epidemiol
8. Moss AR, Vranizan K, Gorter R, et al. HIV seroconversion in intravenous drug users in San Francisco, 1985-1990. AIDS
9. Wiebel WW, Jimenez A, Johnson W, et al. Risk behavior and HIV seroincidence among out-of-treatment injection drug users: a 4-year prospective study. J Acquir Immune Defic Syndr Hum Retrovirol
10. Nelson KE, Galai N, Safaeian M, et al. Temporal trends in the incidence of human immunodeficiency virus infection and risk behavior among injection drug users in Baltimore, Maryland, 1988-1998. Am J Epidemiol
11. Nash D, Bennani Y, Ramaswamy C, et al. Estimates of HIV incidence among persons testing for HIV using the sensitive/less sensitive enzyme immunoassay, New York City, 2001. J Acquir Immune Defic Syndr
12. Des J, Perlis T, Arasteh K, et al. HIV incidence among injection drug users in New York City, 1990 to 2002: use of serologic test algorithm to assess expansion of HIV prevention services. Am J Public Health
13. Vlahov D, Anthony JC, Munoz A, et al. The ALIVE study, a longitudinal study of HIV-1 infection in intravenous drug users: description of methods and characteristics of participants. NIDA Res Monogr
14. Macalino GE, Celentano DD, Latkin C, et al. Risk behaviors by audio computer-assisted self-interviews among HIV-seropositive and HIV-seronegative injection drug users. AIDS Educ Prev
15. Strathdee SA, Galai N, Safaiean M, et al. Sex differences in risk factors for HIV seroconversion among injection drug users: a 10-year perspective. Arch Intern Med
16. Kral AH, Bluthenthal RN, Lorvick J, et al. Sexual transmission of HIV-1 among injection drug users in San Francisco, USA: risk-factor analysis. Lancet
17. Hope VD, Judd A, Hickman M, et al. HIV prevalence among injecting drug users in England and Wales 1990 to 2003: evidence for increased transmission in recent years. AIDS
18. Strang J, Sheridan J, Hunt C, et al. The prescribing of methadone and other opioids to addicts: national survey of GPs in England and Wales. Br J Gen Pract
19. Stolte IG, Dukers NH, Geskus RB, et al. Homosexual men change to risky sex when perceiving less threat of HIV/AIDS since availability of highly active antiretroviral therapy: a longitudinal study. AIDS
20. Katz MH, Schwarcz SK, Kellogg TA, et al. Impact of highly active antiretroviral treatment on HIV seroincidence among men who have sex with men: San Francisco. Am J Public Health
21. Tun W, Gange SJ, Vlahov D, et al. Increase in sexual risk behavior associated with immunologic response to highly active antiretroviral therapy among HIV-infected injection drug users. Clin Infect Dis
22. Tun W, Celentano DD, Vlahov D, et al. Attitudes toward HIV treatments influence unsafe sexual and injection practices among injecting drug users. AIDS
23. Celentano DD, Galai N, Sethi AK, et al. Time to initiating highly active antiretroviral therapy among HIV-infected injection drug users. AIDS
24. Vlahov D, Safaien M, Lai S, et al. Sexual and drug risk-related behaviours after initiating highly active antiretroviral therapy among injection drug users. AIDS
25. Crepaz N, Hart TA, Marks G. Highly active antiretroviral therapy and sexual risk behavior: a meta-analytic review. JAMA
26. Henderson LA, Walker DH. Drug use in the metropolitan area: epidemiology and trends, 1998 through the first half of 2002. National Institute on Drug Abuse, Epidemiologic Trends in Drug Abuse 2004;2:21-35.