Direct comparison of the relationship between clinical outcome and change in CD4+ lymphocytes in human immunodeficiency virus-positive homosexual men and injecting drug users.

BACKGROUND AND METHODS: To compare rates of decline of CD4+ lymphocytes among human immunodeficiency virus-positive homosexual men and injecting drug users, we followed up prevalent human immunodeficiency virus-positive homosexual men and current or former injecting drug users from February 1988 through August 1991. Subjects were free of acquired immunodeficiency syndrome at study entry and had semiannual clinical and laboratory evaluation, including measurement of T-cell subsets, under common protocols. Initial levels and rates of change of CD4+ lymphocyte counts were compared according to cohort membership and clinical progression, defined by the development of thrush or an acquired immunodeficiency syndrome--defining illness. Median follow-up was 30 months for both cohorts. RESULTS: At study entry, homosexual men had lower absolute numbers of circulating CD4+ lymphocytes than did injecting drug users (459/microL [0.46 x 10(9)/L] vs 509/microL, respectively). During follow-up, homosexual men exhibited a faster decline in CD4+ lymphocyte counts as well as more frequent development of HIV-related symptoms (thrush or acquired immunodeficiency syndrome). In both cohorts, initial levels of CD4+ lymphocytes and rates of decline in these cells were strongly associated with progression of disease, defined as remaining asymptomatic, onset of thrush, or onset of acquired immunodeficiency syndrome. Once homosexual men and injecting drug users were stratified by disease progression, their initial levels and rates of decline of CD4+ lymphocyte counts were similar. Thus, crude differences between the two study groups largely resulted from differences in development of clinical symptoms. CONCLUSIONS: In these cohorts of homosexual men and injecting drug users, clinical outcome was much more important than risk group membership in determining changes in CD4+ lymphocyte numbers. The close similarity between the groups also suggests that drug use, ethnicity, and socioeconomic status play a minor role in the progression of human immunodeficiency virus infection.

Effect of isoniazid chemoprophylaxis on HIV-related mycobacterial disease.

OBJECTIVES: The aims of this study were to (1) identify trends and risk factors for mycobacterial disease and (2) determine the effect of expanded access to isoniazid chemoprophylaxis on tuberculosis incidence. METHODS: A prospective observational cohort study was conducted among community-based injecting drug users (IDUs); 2960 IDUs (942 human immunodeficiency virus [HIV] seropositive) were followed up from January 1988 to June 1994. Directly observed chemoprophylaxis with twice-weekly isoniazid (10 to 15 mg/kg) was offered to purified protein derivative (PPD) tuberculin-positive (> or = 5-mm induration diameter in HIV-seropositive subjects and > or = 10-mm diameter in HIV-seronegative subjects) individuals but not to those with cutaneous anergy. Overall and annual incidence rates of disease due Mycobacterium tuberculosis, Mycobacterium avium complex, and other atypical mycobacteria were estimated using Poisson regression. RESULTS: HIV seropositivity was the strongest risk factor for tuberculosis, M avium complex, and other mycobacterial disease (relative risk [RR], 3.8, 17.2, and 6.9, respectively). Median CD4 lymphocyte cell counts for the three groups of mycobacterial disease were 0.17, 0.03, and 0.02 x 10(9)/L (167/microL, 30/microL, 18/microL) within 6 months of diagnosis (before or after). Overall incidence rates of tuberculosis, M avium complex disease, and other mycobacterial disease were 1.9, 8.8, and 2.7 per 1000 person-years, respectively. Tuberculosis incidence peaked in 1991 at six per 1000 person-years. However, after access to directly observed preventive therapy was expanded for tuberculin-positive subjects, incidence fell to only one case in 1992 and zero cases for 24 months from mid-1992 to mid-1994. During this period the number of PPD-positive patients who completed at least 26 weeks of therapy (or were still receiving isoniazid) more than tripled (from 21 to 70). None of the 12 patients with tuberculosis diagnosed during follow-up had received any preventive therapy. In addition, no tuberculosis developed among participants with cutaneous anergy. Calendar trends in risk for M avium complex and tuberculosis diverged after expanded access to isoniazid prophylaxis. Compared with 1988-1989, risk of M avium complex increased sevenfold. Tuberculosis risk fell 83% from the peak risk in 1990-1991. CONCLUSIONS: Expanded access to directly observed isonazid therapy for tuberculin-positive IDUs with and without HIV infection was associated with an 83% drop in tuberculosis incidence, while in the same period M avium complex incidence significantly increased. These population-based data are consistent with those obtained from clinical trials of isoniazid prophylaxis and were obtained without offering chemoprophylaxis to HIV-infected patients with cutaneous energy.

Mycobacterium tuberculosis infection and disease are not associated with protection against subsequent disseminated M. avium complex disease.

OBJECTIVE: To determine the relationship between Mycobacterium tuberculosis infection and disease and subsequent disseminated M. avium complex (MAC) disease in HIV-infected persons. DESIGN: A prospective observational cohort study. SETTING: The AIDS Linked to the Intravenous Experience (ALIVE) cohort of injecting drug users and the Johns Hopkins Hospital Adult HIV Clinic (JHHAHC). PARTICIPANTS: HIV-infected persons aged > 18 years with CD4 lymphocytes < 100 x 10(6)/l were followed between July 1989 and 31 October 1996. There were 182 persons in the ALIVE cohort and 1129 persons in JHHAHC who met these criteria. MAIN OUTCOME MEASURE: The relative risk of disseminated MAC was determined according to a history of prior opportunistic infection, MAC prophylaxis, antiretroviral therapy, M. tuberculosis infection or disease, race, sex, and injecting drug use. RESULTS: Amongst the 30 patients with active tuberculosis, eight developed disseminated MAC, compared with 208 cases of disseminated MAC amongst 1148 patients without prior M. tuberculosis infection or disease [relative risk (RR), 1.5; 95% confidence interval (CI), 0.8-2.7; P=0.2]. Amongst the 10 patients with extrapulmonary tuberculosis, five developed disseminated MAC (RR, 2.8; 95% CI, 1.5-5.2; P=0.02). Injecting drug use was associated with a decreased risk of disseminated MAC (RR, 0.7; 95% CI, 0.6-0.9; P=0.007). In a logistic regression analysis, disseminated MAC was significantly associated with extrapulmonary tuberculosis and other opportunistic disease, whereas antibiotic prophylaxis and injecting drug use were protective. CONCLUSIONS: A history of M. tuberculosis infection or disease was not associated with protection against subsequent disseminated MAC disease in HIV-infected persons. However, persons with extrapulmonary tuberculosis were at increased risk for disseminated MAC, particularly at low CD4 cell levels.

Comparison of clinical manifestations of HIV infection between male and female injecting drug users.

OBJECTIVE: To compare occurrence of clinical symptoms, physical examination findings and hematologic variables in male and female HIV-seropositive injecting drug users (IDU) with similar CD4+ lymphocyte counts. METHOD: We interviewed and examined 118 female and 444 male AIDS-free HIV-seropositive IDU for clinical signs and symptoms. HIV serology and T-lymphocyte subset evaluations were performed. Comparisons were analyzed by Mantel-Haenszel procedures. RESULTS: In this population, median age for men was 35 years versus 33 years for women; median CD4 cell count was 490 x 10(6)/l for men versus 480 x 10(6)/l for women. The overall frequency of oral candidiasis increased as CD4 cell count decreased, but did not vary by sex. Recent history of genital herpes was more frequent (P < 0.05) in women than men, but this difference was not significant on physical examination. Symptoms of diarrhea, fatigue, weight loss, shortness of breath, presence of enlarged posterior cervical lymph nodes did not vary by CD4 cell count or sex, and no strong interactions were evident. Although absolute values of hematocrit were higher (P < 0.001) and platelet count lower (P < 0.001) in HIV-seropositive men than women, distributions of hematocrit and platelet count by sex were similar for HIV-seropositive participants and HIV-seronegative controls. CONCLUSION: Our data on IDU prior to a diagnosis of AIDS suggest that constitutional signs and symptoms are generally similar among men and women early in HIV infection. Additional follow-up is needed to determine whether differential rates of signs and symptoms by sex appear with progression of HIV infection.

The influence of drug use patterns on the rate of CD4+ lymphocyte decline among HIV-1-infected injecting drug users.

OBJECTIVES: To assess the relationship between various injecting drug use patterns and the rate of CD4+ lymphocyte decline in HIV-1-infected injecting drug users in Baltimore, Maryland, USA. METHODS: A cohort of 605 HIV-1-infected injecting drug users was recruited between 1988 and early 1989 in East Baltimore using extensive community outreach techniques. The participants were interviewed semi-annually to collect information on drug use practices. The outcome measure of interest was the rate of CD4+ lymphocyte decline between pairs of CD4+ lymphocyte counts. A mixed model was used to evaluate the relationship between the change in CD4+ lymphocyte count per month and previous CD4+ lymphocyte count and various drug use variables. RESULTS: The 605 HIV-infected injecting drug users had a median initial CD4+ lymphocyte count of 513 cells x 10(6)/l. Using 3209 paired observations, the mean change in CD4+ lymphocyte count was -3.2 cells x 10(6)/l per month. The rate of decline was higher in those with a higher level of CD4+ lymphocytes (P < 0.01) and length of drug use (P < 0.01), but did not vary by injection frequency or injection intensity of specific drug types. Although animal studies have suggested that the pattern of drug administration (continuous versus intermittent) and episodes of withdrawal or overdose might impact the rate of CD4+ lymphocyte decline, this was not observed in the present study. CONCLUSION: Patterns of injecting drug use, based on self-report, were not associated with the rate of decline in CD4+ lymphocytes.

Field effectiveness of needle disinfection among injecting drug users.

To examine the putative protective effect of disinfectant use on HIV seroconversion among injecting drug users, we conducted a nested case-control study of black heterosexuals comparing 34 HIV seroconverters with 154 persistent seronegatives matched on gender, cocaine injection (yes/no), date of study entry, and duration of follow-up. Injecting drug users who reported using disinfectant all the time had an odds ratio of seroconversion of 0.87, as compared with those who reported no use of disinfectants; the corresponding odds ratio was 1.00 for those who used disinfectants less than all the time. We examined the effect of drug use and sex practice variables, and responses to a socially desirable responding scale as possible confounders for the effect of needle disinfection on HIV seroconversion; the adjusted odds ratios for disinfectant use and HIV seroconversion were unchanged in this analysis. Despite limited statistical power and the potential for residual confounding, these data suggest that disinfection of injection equipment is not a substitute for abstinence from drugs or use of sterile injection equipment.

Prognostic indicators for development of AIDS among intravenous drug users.

A cohort of 544 human immunodeficiency virus-1 (HIV-1) seropositive intravenous drug users (IVDUs) was recruited in 1988 and early 1989; data on laboratory markers, clinical symptoms, intravenous drug use, and demographics were collected. Forty-one IVDUs developed AIDS within 2 years of enrollment. Data were analyzed using methods of survival analysis. None of the individuals reported use of antiviral agents or Pneumocystis carinii prophylaxis at baseline. A very strong (p less than 0.001) dose-response relationship was identified between CD4 cell count at baseline and the subsequent development of AIDS. In multivariate analysis, both the presence of more than one clinical HIV-1-related symptom and serum neopterin greater than 12 nmol/L showed significant associations with the relative hazard (95% confidence interval) of AIDS after controlling for CD4 of 2.9 (1.6, 5.6) and 2.0 (1.0, 3.7), respectively. In these IVDUs, serum beta 2-microglobulin did not add predictive power for progression to AIDS. The effect of clinical symptoms was stronger for high CD4 cell counts, indicating the need to monitor HIV seropositive IVDUs with both laboratory studies and clinical evaluation.

Preparations for AIDS vaccine trials. Incident human immunodeficiency virus (HIV) infections in a cohort of injection drug users in Baltimore, Maryland.

In order to prepare for a possible trial of a preventive human immunodeficiency virus (HIV) vaccine in a population of injection drug users (IDUs) we followed a cohort to determine their HIV incidence, compliance with follow-up visits at 3-month intervals (i.e., quarterly) and their attitudes toward HIV vaccine trial participation. A population of 671 HIV-seronegative subjects were recruited from a cohort of IDUs already in follow-up in Baltimore, MD (the ALIVE study). We detected 19 seroconverters in 1677.80 person-quarters of follow-up, an annual incidence of 4.52/100 person-years. Compliance with quarterly follow-up was 93% at 6 months and 89% at 9 months. Although many subjects (n = 104) are not yet due for their 12-month visit, compliance to date has been 76%. The most sensitive risk behaviors associated with HIV seroconversion have been continued injection of illicit drugs and the frequency of drug use. Other reported drug associated risk behaviors, e.g., needle sharing and shooting gallery use, have decreased despite a high HIV incidence; we believe some of the reported reductions in high-risk behavior represent socially desirable responses by the study subjects. Although the difficulties in successfully doing a trial of an HIV vaccine in the population should not be underestimated, our data suggest that such a trial would be feasible.

Interest in HIV vaccines among injection drug users in Baltimore, Maryland.

To gauge interest about participation in human immunodeficiency virus type 1 (HIV vaccines, we interviewed 375 HIV-seronegative injecting drug users who are participants in an ongoing longitudinal study of HIV infection. Nearly all (93%) responded that they thought it likely that an HIV vaccine would become available, and 85% expressed interest in participating in a study of vaccine effectiveness. However, levels of interest decreased to 47% when respondents were informed that the vaccine might result in a positive HIV test and to 27% when offered to be in a study where the vaccine might contain a piece of the virus. Factors that increased interest in trial participation included assurances of confidentiality, being fully informed about the protocol and remuneration. Most respondents (78%) felt that injecting drug users would maintain other risk reduction activities (e.g., condom use) if they participated in a vaccine study. These data suggest a high level of interest for participation in HIV vaccine trials, but that more education about vaccines and the risks involved is needed. Ongoing communication with the community and responsiveness to community concerns is crucial to achieve a successful vaccine study.

Variance components analysis of forced expiration in families.

Familial aggregation of forced expiration (as measured by forced expiratory volume in 1 sec (FEV1) and the ratio of this to total forced vital capacity (FEV1/FVC) was analyzed in 439 adult members of 108 families ascertained through control patients who had participated in a genetic and epidemiologic study of chronic obstructive pulmonary disease. Residual values for both FEV1 and FEV1/FVC obtained from regression on age, sex, race, and cigarette smoking (and height for FEV1) were used in a variance components analysis to assess the relative importance of genetic and nongenetic factors influencing familial aggregation of pulmonary function among adults. For both residual FEV1 and residual FEV1/FVC, the "best" model among a series of genetic and nongenetic models was a simple additive genetic model. A modified score test, which is robust to the assumption of multivariate normality, was used to test the significance of these estimated components. Under the most parsimonious model, additive genetic variation accounted for 28% of the variation in residual FEV1 in 108 families and 24% of the variation in residual FEV1/FVC. After outlying individuals were identified by examining goodness-of-fit statistics, the simple genetic model still gave the best fit to these data. There was little indication of non-normality in FEV1 in these families; however, FEV1/FVC did show evidence of non-normality when examining goodness-of-fit statistics. This genetic component contributing to the distribution of forced expiration may be a factor in the familial aggregation of certain respiratory diseases.

Familial risks of congenital heart defect assessed in a population-based epidemiologic study.

Congenital heart defects (CHD) represent a heterogeneous group of disorders caused by chromosome abnormalities, mendelian disorders, teratogenic exposures, and unknown etiologic mechanisms. A large group of various isolated defects is presumably multifactorial in origin. Previous studies of familial risks for specific anatomic defects obtained from clinical series may include significant biases and obscured pathogenic relationships. In this population-based study we analyzed all cases of CHD in infants and a control birth cohort in the Baltimore-Washington area. The rates of CHD were defined for first-degree relatives of cases with isolated defects, grouped by a pathogenic classification scheme. Precurrence risks were found to vary among the groups, and risks for flow lesions were higher than previously reported. The sibling precurrence risk for hypoplastic left heart syndrome (13.5%) was not significantly different from that expected for an autosomal recessive mechanism; the risks for different types of ventricular septal defects (VSD) varied among mechanistic groups. The results indicate that the additive multifactorial model does not adequately account for the risks in all forms of isolated CHD of unknown etiology.

The effect of low-level 60-Hz electromagnetic fields on human lymphoid cells. II. Sister-chromatid exchanges in peripheral lymphocytes and lymphoblastoid cell lines.

Dividing human peripheral lymphocytes from 10 normal adults (5 males and 5 females) as well as lymphoid cell lines from patients with the chromosomal instability syndromes were exposed to low-level 60-Hz sinusoidal electromagnetic fields (EMF). The current density of the electrical field was 30 microA/cm2 while the strength of the magnetic field was either 1 or 2 gauss. The cytological endpoints measured included the frequency of sister-chromatid exchanges per chromosome; the distribution of first-, second-, and third-division cells and chromosome breakage (lymphoblastoid cells only). No statistically significant differences, indicative of EMF effects were observed between the treated and control cells regarding SCE frequency, cell cycle progression or chromosome breakage.

Repeated measurement of spontaneous and clastogen-induced sister-chromatid exchange.

Sister-chromatid exchanges (SCE), both spontaneous and chemically-induced [bleomycin (BLM), mitomycin-C (MMC), streptonigrin (SN), and 4-nitroquinoline-1-oxide (4NQO)], were studied in the lymphocytes of 24 normal individuals on 2 or 3 different occasions, separated by periods of up to 2 years. For all BLM-induced SCEs, the variation in SCE frequency among the samples from a single individual was significantly greater than the variation between replicate cultures on a given day. These results raise questions concerning the validity of conclusions based on a single observation of chemically-induced SCEs.

Clinical and laboratory characterization of early onset periodontitis.

Clinical and laboratory data were compared in 72 patients with localized periodontitis (LP) and 103 patients with generalized periodontitis (GP). Significantly more LP than GP cases had decreased neutrophil chemotaxis (CTX), and were seropositive for Actinobacillus actinomycetemcomitans (Aa). Significantly, more GP cases were seropositive for Bacteroides gingivalis (Bg). All clinical indices were similar on affected teeth in LP and GP, but the attachment loss was greater on clinically unaffected teeth in GP when compared with LP. LP cases with CTX defects had a significantly lower mean age, were more often seropositive for Aa antibodies, and were more often female than LP patients with normal CTX. Significantly more GP cases with CTX defects were seropositive for Aa antibody. GP patients with normal CTX had a higher plaque index on both affected and unaffected teeth than did GP patients with a CTX defect. Our data suggest that chemotaxis and/or specific bacteria may be contributory, but not always necessary, factors in these disorders. The overlap in clinical and laboratory profiles of LP and GP continues to cloud the distinction of these early onset forms of periodontitis.

Bacterial infections and skin cleaning prior to injection among intravenous drug users.

In a survey of 1,057 active intravenous drug users in Baltimore, MD, who were recruited through extensive community outreach, 12 reported endocarditis and 113 reported subcutaneous abscesses in the 6 months before being interviewed. Of all the persons surveyed, 556 reported cleaning their skin prior to injection at any time and 173 reported cleaning their skin all the time in the 6 months before the interview. The frequency of subcutaneous abscesses was lower among those who reported skin cleaning all the time; a similar trend was noted for frequency of endocarditis. The relatively simple procedure of encouraging intravenous drug users to clean their skin prior to injection will not eliminate but might reduce the frequency of these serious and expensive infectious complications of intravenous drug use.

Early immunologic and virologic responses to highly active antiretroviral therapy and subsequent disease progression among HIV-infected injection drug users.

We examined the prevalence and prognostic value of early responses to highly active antiretroviral therapy (HAART) among community-based injection drug users (IDUs) in Baltimore. Virologic (HIV RNA <1000 copies/ml) and immunologic (CD4 >500 cells/ul or increase of 50 cells/ul from the pre-HAART level) responses were examined in the 1st year of HAART initiation. Cox regression was used to examine the effect of early response on progression to new AIDS diagnosis or AIDS-related death. Among 258 HAART initiators, 75(29%) had no response, 53(21%) had a virologic response only, 38(15%) had an immunologic response only and 92(36%) had a combined immunologic and virologic response in the first year of therapy. Poorer responses were observed in those who were older, had been recently incarcerated, reported injecting drugs, had not had a recent outpatient visit and had some treatment interruption within the 1st year of HAART. In multiple Cox regression analysis, the risk of progression was lower in those with combined virologic and immunologic response than in non-responders, (relative hazard [RH], 0.32; 95% confidence interval [CI], 0.17-0.60). Those with discordant responses had reduced risk of progression compared to non-responders but experienced faster progression than those with a combined response, although none of these differences was statistically significant. Early discordant and non response to HAART was common, often occurred in the setting of injection drug use and treatment interruption and was associated with poorer survival. Interventions to reduce treatment interruptions and to provide continuity of HIV care during incarceration among IDUs are needed to improve responses and subsequent survival.

Utilization of health services in a cohort of intravenous drug users with known HIV-1 serostatus.

BACKGROUND: Intravenous drug users (IVDUs) with human immunodeficiency virus (HIV) infection and AIDS often have no health insurance or rely on public programs to finance their health care. We examined the independent contributions of HIV serostatus, clinical symptoms, CD4 cell counts, and health insurance to utilization of health services among 1881 intravenous drug users in Baltimore, Maryland. METHODS: Participants in an ongoing natural history study of HIV were informed of HIV serostatus and seropositives were informed of CD4 cell counts; 6 months later, participants were administered a questionnaire concerning self-reported use of health services, insurance coverage, and HIV-related symptoms. RESULTS: Of 1881 participants, 67% had health insurance (including Medicaid), 48% had at least one outpatient visit, and 12% had at least one inpatient visit within the prior 6 months. The proportion of the study population that was HIV-1 seropositive was 32%. In multivariate analysis, the single most important predictor of health care utilization was the presence of two or more HIV-related clinical symptoms. HIV positive serostatus alone or known low CD4 counts were not significantly associated with use of health care services. CONCLUSIONS: These data suggest that HIV seropositive IVDUs are not receiving recommended preventive care. Additional efforts will be needed to ensure that HIV-seropositive drug users participate in currently recommended protocols for early treatment of asymptomatic HIV-1 infection.

Existence and failure of T-cell homeostasis prior to AIDS onset in HIV-infected injection drug users.

Prior studies, based on populations of homosexual men, have shown that during HIV infection, levels of total circulating T-cells (CD3+ lymphocytes) remain constant for long periods of time after seroconversion. This suggested homeostatic phenomenon was observed to break down about 18 months prior to AIDS diagnosis with a quick loss of T-cells. The objective of this study was to determine whether (a) total T-cells are maintained at a constant level for long periods of time among HIV-infected injection drug users (IDUs) and (b) total T-cells decline before AIDS onset in this risk group and, if so, by how long. The design and setting was prospective follow-up, with semiannual clinic visits, of 646 HIV-infected IDUs who participate in the ALIVE study (Baltimore, MD). Among AIDS cases, T-cell levels remained quite stable at about 1500 cells/microl up to approximately 24 months prior to AIDS. However, a steep decline in CD3+ cell levels began approximately 24 months prior to AIDS diagnosis and was -17.5% per 6 months in the last 18 months before AIDS. Among seropositive IDUs without AIDS, a gradual decline of less than -4% per 6 months was observed. These trends remained virtually unchanged after accounting for current injection drug use, smoking, and HIV-related medications. IDUs, like homosexual men, exhibited T-cell homeostasis following HIV infection, as well as failure of this homeostasis about 2 years before AIDS. Although the mechanisms for the maintenance and later failure of the homeostasis of T-lymphocytes are not well understood, the observation has a potentially important prognostic value as well as biological interest.

Phenotypic assessment of early onset periodontitis in sibships.

Early onset periodontitis is a group of familial diseases that are not yet clearly defined by etiologic mechanisms, although some risk factors have been recognized. The disorders include a localized form of juvenile periodontitis (JP), and a more generalized form (GP). In a family study, 39 sibships (116 individuals, aged 13-48) were evaluated for clinical indices, neutrophil chemotaxis, and serum antibodies to A. actinomycetemcomitans (Aa). Of 77 siblings, 41 were healthy at examination. In 14 sibships, all affected persons had JP; 14 other sibships had all affected individuals with GP; and 11 had at least one sib with each form. For probands with decreased chemotaxis, 71% of affected sibs and 36% of clinically healthy sibs had decreased chemotaxis. For Aa seropositive probands, 83% of affected siblings and 65% of currently healthy sibs were also seropositive. The associations of disease with these risk factors were stronger in JP-only sibships. Some affected sibs had neither risk factor, while many currently healthy sibs had 1 or both. While these 2 factors demonstrate population association with disease, neither fits the pattern expected within families to clearly suggest a causal mechanism. The assessment of within and among family variability remains the best approach for recognition of possible causal mechanisms and sources of heterogeneity.

The trading of sex for drugs or money and HIV seropositivity among female intravenous drug users.

OBJECTIVES: Data from 538 women in a cohort study recruited in 1988-1989 were analyzed to determined whether trading sex for drugs or money was independently associated with human immunodeficiency virus (HIV) seroprevalence in a population of female intravenous drug users. METHODS: The women were grouped according to the number of partners with whom they reported trading sex for drugs or money during the previous 10 years: none, 1 through 49 (low), or 50 or more (high); the prevalence of HIV seropositivity in the three groups was 23.2%, 23.7%, and 47.6%, respectively. Logistic regression was used to compare the low- and high-trade groups separately with the group that reported no trading. RESULTS: Low trading was not associated with seroprevalent HIV infection. In a multivariate model, high trading (compared with no trading) was significantly associated with HIV seropositivity after adjustment for cocaine use, history of sexually transmitted diseases, and duration of intravenous drug use. CONCLUSIONS: These data indicate that, among intravenous drug-using women, high levels of trading sex for drugs or money were independently associated with HIV infection. This group needs to be targeted for further intensive intervention.