Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2007 Aug 17.
Published in final edited form as: AIDS Care. 2006 Feb;18(2):158–165. doi: 10.1080/09540120500162114

Risks for HIV infection among users and sellers of crack, powder cocaine and heroin in central Harlem: Implications for interventions

W REES DAVIS 1, B D JOHNSON 1, D RANDOLPH 1, H J LIBERTY 1
PMCID: PMC1950088  NIHMSID: NIHMS26560  PMID: 16338774

Abstract

This article investigates behaviours that may be associated HIV infection among users and sellers of crack, powder cocaine and heroin in central Harlem. Chain referral sampling and other strategies were combined to acquire a sample of 637 (Users = 546; Sellers = 91) who provided urine specimens that were tested for the presence of drugs and HIV. Nearly a quarter (23.9%) of all respondents were HIV positive. Drug injectors were more than 2.5 times more likely to have HIV infections than other respondents (OR = 2.66; 95% CI 1.66–4.26). Those involved in frauds/cons were almost as likely to be HIV positive (OR = 2.58; 95% CI 1.64–4.06). Those with a marital status of being separated, divorced or widowed were twice as likely to be HIV infected (OR 2.16; 95% CI 1.43–3.25). Respondents currently having multiple partner sex (OR = 1.66; 95% CI 1.1–2.51) or who were female (OR = 1.66; 95% CI 1.12–2.45) were more than 1.5 times more likely to be HIV positive. Thus, controlling for lifetime drug injection and current multiple partner sex, other factors, such as participating in frauds/cons, as well as relationship status and being female, were also associated with HIV infection.

Introduction

This article focuses on factors directly involved in HIV transmission, and a wide variety of sociodemographic, crime, and drug-use factors, to determine their relative impact on HIV infection among a sample of crack, powder cocaine, and heroin users and sellers in central Harlem. The clustering and relative associations of risk factors with HIV infection may describe the contexts of HIV transmission among users/sellers of these drugs. Patterns of factors associated with HIV infection may be of importance to HIV prevention and treatment interventions, especially those associated with criminal justice systems and programmes providing drug treatment and prevention services.

At the outset of HIV epidemic in New York City (NYC), HIV positive persons were more likely to belong to specific groups already stigmatized for engaging in hidden or criminalized behaviours. Men who had sex with men (MSM) and injection drug users (IDUs) had accounted for the large majority of AIDS cases in 1999, the final year of data collection in this research. NYC AIDS surveillance data indicated that 50% of adult cases with a reported transmission category were IDUs, 35% MSM, and 14% heterosexual (NYC Department of Health–NYCDOH, 1999). Current data (through 2002) indicate several changes since 1999, as only 35.7% of adult cases with a reported transmission category were IDUs, 35.1% MSM, and 24.4% heterosexual (NYCDOH, 2004). Research and prevention efforts may have been responsible for the reduction in seroprevalence and transmission behaviours among IDUs in the last decade (Maslow, et al., 2002; Des Jarlais et al., 1998, 1999).

Factors not directly involved in HIV transmission, hereafter referred to as ‘lifestyle risk factors’, have emerged from a variety of research efforts as important to reducing the spread of HIV. For instance, the proportion of those meeting criteria for alcohol abuse among HIV+ persons has ranged from 29% to 60% (Brown et al., 1992; Summers et al., 1995; Dew et al., 1997). Those who have multiple sex partners, particularly crack users engaging in sex work, have also been found to have higher proportions of AIDS cases, HIV seroprevalence, and HIV risk factors (Elwood et al., 1997; Inciardi, 1995; Boyd et al., 1994).

More than one-fifth (22%) of NYC AIDS-diagnosed cases were women in 1999 (NYCDOH, 1999). In the last few years, women of color have experienced the most rapid increase in HIV infections (Pulerwitz et al., 2000; CDC, 1999; Chu and Curran, 1997). Several other factors have possibly significant associations with HIV infection, including age, education, housing, criminal activities, and marital status. For instance, low education level may interfere with understanding or acceptance of prevention efforts (Deren et al., 1997).

Methods

Sample

Several methods were combined to obtain a representative sample of crack, powder cocaine, and heroin users and sellers. The New York Police Department (NYPD) provided the researchers with indicator data in the form of a computerized map of Central Harlem, pinpointing the locations of all drug-related ‘allegations’/‘complaints’ over a two-year period; April, 1996, through March, 1998. The drug allegations were linked to 1990 census block maps, and allegations/capita were computed to subdivide Central Harlem into 45 primary sampling units (PSUs), 11 with a high number of drug allegations/capita, 13 with mid-level allegations/capita and 21 with low allegations/capita.

Chain referral (‘snowball’ – Goodman, 1961) sampling was modified by targeted sampling (Watters & Biernacki, 1989) and aspects of respondent driven sampling (RDS – Heckathorn, 1997, 2002) to recruit respondents in nine randomly selected PSUs. Project ethnographers (‘privileged access interviewers’– Griffiths et al., 1993) led by coauthor Doris Randolph, spent 2–4 weeks in each PSU locating where crack, powder cocaine, and heroin use/sales activities occurred. Locations were recorded on an ‘ethnographic map’ and initial respondents were chosen over the entire PSU in order to reduce bias by selecting individuals with diverse social networks and locations, approximating a random sample (Snijders, 1992).

Recruitment

Selection criteria excluded anyone who did not use or sell crack, powder cocaine, or heroin. Users were limited to those who only used these drugs and had no operative role; operatives could use these drugs as well as report their role(s) in the distribution of these drugs. Interviews were conducted in a variety of settings (parks, cars, the streets, restaurants, a room rented for two hours from someone known to interviewers, etc.).

The chain referral methods were refined in a pilot phase to avoid ethical pitfalls and to investigate various methods of steering the recruitment of respondents in order to minimize bias. Confidentiality was protected primarily by recording only the first and/or street names of respondents and excluding all police and/or government identifiers such as height, weight, eye color, home address, birthmarks/tattoos, etc. Further details concerning the sampling methods employed can be found in the existing literature (Davis & Johnson, 2000; Cross et al., 2001; Davis et al., 2003, 2004, 2005).

Interviews were approximately 1.5 hours long, excluding 15 to 30 minutes for the informed consent review and signature. Interviewers asked all potential respondents to review informed consent materials and agree to participate in the study in exchange for $20. Respondents who successfully recruited a new respondent received a $20 recruiter’s fee. Only three persons were approached but refused to participate.

Urine testing

In the informed consent process, respondents were told that they would remain anonymous; their full names and addresses were never obtained. Respondents were also told that they would be asked to provide urine specimens so that drug use could be confirmed by biological analyses before the interview began. They were also told that these specimens might be tested for diseases. The urine specimen was tested for cocaine and heroin metabolites immediately with a Testcup™ from Hoffman-LaRoche. Only ten of all respondents did not provide urine specimens, and results from Testcup analyses indicated that only 11 self-reported users may not have been users of crack, powder cocaine or heroin. However, cocaine and heroin metabolites in urine can only be detected within three days of use, and the 11 respondents may have used drugs four or more days prior to interviewing.

HIV testing

All but 20 respondents’ urine samples were tested for the presence of HIV with the Calypte method for a total N of 637. Specimens were tested at least three months after collection with the ELISA screening method and positive results were confirmed using the Western Blot method. Since the respondents were anonymous, without any contact information, no attempt was made to convey results to respondents.

Description of data

Most of the information regarding the focal behaviours of this study were elicited from respondents in terms of the recency of self-reported behaviour. The investigators defined ‘current’ behaviours as occurring within the last 30 days and ‘lifetime’ behaviours as those occurring at some other time in the person’s life but not in the last 30 days.

Pilot testing determined that questions concerning sexual activities were limited to the past 30 days, and questions regarding quality of life offences (Kelling & Coles, 1996) such as loitering, littering, drinking alcohol from an open container, transit fare beating, etc., were limited to the past year. Inquiries regarding violence were limited to the past two years. Results concerning most factors in this article use the lifetime criteria.

Analytic approach

Chi-square analyses were performed to test for significant associations between HIV status and direct transmission factors, sociodemographic factors, and lifestyle risk factors as independent factors. Statistically significant (P < .05) independent factors were dummy coded for simultaneous entry in multivariate logistic regression analyses to determine the relative risks of HIV infection for all independent factors. A subsequent analysis restricted the number of independent factors to those with P values less than .15 in the first logistic model. Finally, the number of risk factors for each respondent was computed to analyse whether the proportions of HIV infections increased directly with the number of risk factors.

Results

Demographic factors

Almost a quarter (23.9%; N = 152) of the 637 specimens tested positive for HIV. Women (28.5%) were more likely to be HIV positive than males (20.2% – Table I). Those who were HIV positive at various age levels were consistent with the overall percentage of 23.9% except for those aged 25 or younger (5.6%). Those few (8.5%) in the sample who were not African-American were more likely to be HIV+, especially the 38 Latinos, 42.1% of whom were HIV+.

Table I.

Demographic factors for 637 respondents with HIV data.

Factor N % of Total % HIV+ p value
Gender
 Male 356 55.9 20.2 0.016
 Female 281 44.1 28.5
Age
 < = 25 36 5.7 5.6 0.068
 26–30 61 9.6 26.2
 31–35 135 21.2 23.0
 36–40 179 28.1 22.3
 41–45 115 18.1 29.6
 46–50 69 10.8 27.5
 > = 51 42 6.6 23.8
Generation
 Heroin (born < = 1955) 175 27.6 28.6 0.071
 Cocaine (born 1956–1968) 399 62.8 23.1
 Blunt (born > = 1969) 61 9.6 14.8
Race
 African–American 583 91.5 22.5 0.027
 Latino 38 6.0 42.1
 Other 16 2.5 31.3
Education*
 <High school 267 42.1 28.1 0.032
 GED or high school grad. 367 57.9 20.7
Housing Status
 No address 56 8.8 23.2 0.905
 Address 581 91.2 23.9
Marital Status*
 Single 401 63.2 19.0 0.000
 Married 54 8.5 18.5
 Separated 87 13.7 35.6
 Divorced 65 10.3 30.8
 Widowed 27 4.3 51.9
Current Illegal Income**
 $0 334 52.6 17.1 0.000
 $1 to $499 118 18.6 25.4
 $5003–$999 86 13.5 36.0
 $1,0003–$1,999 55 8.7 43.6
 $2,000 or more 42 6.6 23.8
Open in a new tab
*

N = 634;

**

N = 635.

Those who had not earned a GED or graduated from high school were more likely to be HIV+ (28.1%) than those with a GED or who finished high school (20.7%; P < .032). Those widowed (51.9%), separated (35.6%), or divorced (30.8%) had proportionately more HIV+ persons that those who were single (19.0%) or married (18.5%–P < .000). Higher amounts of illegal income, particularly those having $500 to $1,999 in illegal income in the past 30 days, were more likely to be HIV+ (P < .000).

Factors directly involved in HIV transmission

Those who ever injected drugs (P < .000), participated in current multiple partner sex (P < .002), or sex work (P < .000) were proportionally more HIV+ than drug users/sellers who did not do so (Table II). Since only 17 respondents self-reported currently having same sex partners, large percentage differences between sexual minorities and heterosexuals were not statistically significant.

Table II.

Behaviors related to HIV risk for 637 respondents.

Behavioral Factor (Independent Variables–IV) N % of Total % HIV+ w/o IV % HIV+ w/IV p value
Direct Transmission Factors
 Current multiple partner sex (N = 634) 314 49.5 18.8 29.3 0.002
 Current/lifetime injection 113 17.7 20.6 38.9 0.000
 Current/lifetime sex work 126 19.8 20.2 38.9 0.000
 Current same sex partners 17 2.7 23.4 41.2 0.110
Current or Lifetime Prison and Crime Factors
 Arrested 444 69.7 18.7 26.1 0.039
 Prison 117 18.4 21.8 31.6 0.028
 Quality of Life Stop in last year (N = 635) 234 36.9 20.9 29.1 0.022
 Stealing 181 28.4 18.0 38.7 0.000
 Frauds and cons (N = 636) 123 19.3 18.9 44.7 0.000
 Drug dealing 166 26.1 21.9 29.5 0.050
 Drug sales assistance (N = 636) 151 23.7 21.6 31.1 0.019
 Robbery (N = 636) 53 8.3 22.8 35.8 0.041
 Attacker in last 2 years (N = 636) 152 23.9 22.7 27.0 0.288
 Attacked in last 2 years (N = 636) 239 37.6 20.4 29.3 0.011
Current or Lifetime Drug Use
 Crack (N = 636) 520 81.8 14.7 26.0 0.007
 Cocaine 481 75.5 17.3 26.0 0.023
 Heroin 241 37.8 19.7 30.7 0.002
 Marijuana 591 92.8 32.6 23.2 0.162
 Alcohol 602 94.5 28.6 23.6 0.510
Open in a new tab

Criminal justice and drug use factors

Users and sellers of crack, powder cocaine, and/or heroin who had been arrested (26.1%), in prison (31.6%), or had quality of life offences in the past year (29.1%), had proportionately higher levels of HIV infections (P < .039, P < .028 and P < .022, respectively – Table II). Also, those engaging in property crimes, such as stealing (38.7%) and participating in frauds and cons (44.7%), had much higher percentages of HIV+ persons (P < .000). These property crimes were not associated with drug injection or the use of any specific drug (data not shown). Persons engaging in drug selling (29.5%) and helping with drug sales (31.1%) were also significantly more likely to have HIV infections. Although few participated in robbery, it was significantly associated with HIV infection (P < .041). Those who were attacked at least once in the past two years had higher proportions of HIV infection than those not attacked (P < .011). Those who ever used crack, powder cocaine, or heroin were more likely to be HIV+ than those never using these drugs (Table II). The 53.5% of the heroin users who used heroin without injecting it were not statistically associated with increased proportions of HIV infection (data not shown).

Logistic regression analyses

The first logistic regression model included the 20 independent factors statistically associated with HIV infections in chi-square analyses, but only two of these factors resulted in definitely increased odds of HIV infection, having lower bounds greater than one (Table III). Drug injectors were almost three times more likely to be HIV+ (OR = 2.7; P < .002) than non-injecting drug users and sellers of crack, powder cocaine, and heroin. Also, the respondents who were widowed, separated or divorced, compared with single and married respondents, were twice as likely to be HIV+ (OR = 2.03; P < .002). Five additional factors in model 1 had P values less than .15, including being female, not being African-American, currently having multiple sex partners, ever participating in frauds/cons, and ever using crack. These factors, drug injection, and marital status were used in a second logistic regression analysis, and only crack use and race did not yield statistically significant results. In model 2, odds ratios for injection and marital status remained relatively the same as in model 1, but those participating in frauds/cons were 2.5 times more likely to have HIV infection (OR = 2.58; P < .000). Female respondents and respondents currently having multiple sex partners were considerably more likely to have HIV infections (OR = 1.66; P < .02).

Table III.

Logistic regression of behaviors related to HIV+ status.

Model 1
Model 2
Lifetime Behavior (unless indicated) Odds ratio 95% CI p value Odds ratio 95% CI p value
Female 1.49 .89–2.47 0.127 1.66 1.12–2.45 0.011
Latino/other 1.77 .91–3.45 0.092
Not a high school graduate 1.20 .8–1.82 0.378
Separated/divorced/widowed 2.03 1.31–3.14 0.002 2.16 1.43–3.25 0.000
Some current illegal income 1.03 .57–1.86 0.925
Current multiple partner sex 1.48 .95–2.32 0.085 1.66 1.1–2.51 0.015
Ever injected 2.70 1.46–4.99 0.002 2.66 1.66–4.26 0.000
Ever a sex worker 1.11 .56–2.18 0.772
Ever arrested 0.91 .53–1.54 0.720
Ever in prison 1.18 .68–2.06 0.556
Quality of life in past year 1.00 .64–1.58 0.989
Ever stealing 1.24 .66–2.36 0.506
Ever drug dealing 1.38 .7–2.72 0.360
Ever assist in drug sales 0.94 .46–1.90 0.853
Ever frauds and cons 1.74 .91–3.31 0.093 2.58 1.64–4.06 0.000
Ever robbery 1.15 .55–2.41 0.706
Attacked in last 2 years 1.28 .81–2.02 0.293
Ever use crack 1.70 .91–3.19 0.099
Ever use powder cocaine 0.91 .53–1.58 0.742
Ever use heroin 1.10 .65–1.88 0.718
Open in a new tab

Multiple risk factors

To investigate whether those having multiple risk factors increase their likelihood for having HIV infection, the four direct transmission factors (injection, sex work, multiple sex partners, and same sex partners), five crime factors (stealing, frauds/cons, drug dealing, drug sales assistance, and robbery) and three drug use factors (heroin, powder cocaine, and crack) were totalled and used in a chi-square analysis (data not shown). The proportion of those HIV positive increased directly with the number of direct transmission factors (1 factor, 25%; 2, 35.2%; 3, 75% – P < .000). Although over half the sample reported no participation in the five crimes examined, levels of HIV infections increased directly as the number of kinds of crime increased (P < .000). Those who used all of the three focal drugs were more likely to be infected than those using one or two (P < .000). The totalled 12 factors indicated that especially large increases in HIV+ status occurred in three places, between three (14.7% HIV+) and four factors (25.5% HIV+), between five (28.0% HIV+) and six factors (37.3% HIV+), and between eight (42.3% HIV+) and nine factors (75% HIV+).

Discussion

Factors associated with HIV infection

Nearly a quarter (23.9%) of 637 users and sellers of crack, powder cocaine, and heroin from Central Harlem were found to have HIV infections. Chi-square analyses (Tables I and II) indicated that HIV infection was associated with twenty statistically significant factors. There appeared to be three thresholds, or tipping points, where taking an additional behavioural risk substantially increased the proportions of respondents having HIV infections. Thus, our investigation confirms other studies’ (Wagner & Anthony, 2002; Boyer et al., 1999; Crum et al., 1996; Holmberg, 1997) documentation that risks cluster, and demonstrates that the number of risks increases the likelihood of HIV infection.

In logistic regression model 2 (Table III), only five of the factors significantly associated with HIV infections had not dropped out. The analysis confirmed previous data (NYCDOH, 1999, 2004) that the direct HIV transmission factor of drug injection was the most highly associated with HIV+ status (OR = 2.66). Also in line with other research studies (Elwood et al., 1997; Inciardi, 1995; Boyd et al., 1994), those having sex with multiple partners in the 30 days prior to interviewing had increased odds of HIV infection (OR = 1.66).

Remarkably, participating in frauds and cons (P < .000), a property crime, emerged as the second most predictive factor for HIV infection (OR = 2.58); but property crime was not associated with the heroin use or drug injection, as investigators had hypothesized on the basis of previous research (Johnson et al., 1985). Property crimes have few direct or obvious linkages to HIV transmission. Apparently, many HIV+ drug users/sellers support their lifestyles, at least in part, by obtaining income from frauds and cons, independent of their relationship status, drug use practices, or other behaviours examined. This finding demonstrates the importance of including measures for specific kinds of crime in future HIV research.

The model 2 logistic regression analysis also indicated that broken relationships (defined by separation, divorce or being widowed) were significantly associated with HIV infections (P < .000; Table III). Controlling for all other factors, this lifestyle factor at least doubled the odds of respondents having HIV infection (OR = 2.16). It is possible (and warrants further investigation) that broken relationships among those already at risk can result in a decline in risk reduction behaviours such as condom use or limiting injection equipment entirely to self-use.

Gender differences

In chi-square analysis, 28.5% of sampled women were HIV infected. Women were more likely than men to have HIV infections (OR = 1.66; model 2, Table III). The acceleration of HIV infections among women of colour has been attributed to a variety of factors that involve their relative social powerlessness relative to Caucasian women and men (Amaro and Raj, 2000; Amaro, 1995; Jenkins 2000; Jenkins & Coons, 1996). Also, the efficiency of male-to-female transmission of HIV is 2.3 times greater than that of female-to-male transmission (Nicolosi et al., 1994; Padian et al., 1997).

Implications for interventions

Data in this article have identified several lifestyle risk factors associated with HIV infection independent of direct transmission factors. Participation in one behavioural risk or lifestyle factor appears to enhance the probability of participating in others, even though some of these behaviours or factors, such as frauds/cons, or having been separated, divorced or widowed, have no clear linkages with HIV transmission. Interventions should be designed to address how taking one kind of HIV or criminal justice risk is highly likely to provide opportunities and/or increase willingness for taking other such risks. Preventive interventions should target women of colour who take these risks, especially as this and other reports indicate that women of colour are most at risk for HIV infection.

Study limitations and future research

Controlling for drug injection and sexual behaviours directly involved in HIV transmission, many of the lifestyle risk behaviours examined in this article dropped out of the logistic regression models (Table III). The interpretations of both the lifestyle factors that remain (such as being separated, divorced or widowed, or participating in frauds/cons) and the lifestyle factors that drop out of the models are sometimes difficult. Small cell sizes (e.g. in our analysis, for those having both a history of injection and fraud/con participation, there was an N of 10 for those who were HIV negative, and an N of 16 for those HIV+) also increase the likelihood that our results may have occurred by chance. Further research is required to replicate the behavioural lifestyle factors associated with HIV and to understand the complex links between lifestyle factors and direct transmission factors.

Another limitation was the lack of assessment regarding details of behaviours related to HIV infection, such as condom use and syringe cleaning and/or exchange. Also, the study did not collect data concerning many factors likely associated with elevated proportions of HIV infection, such as the ecological factors of impoverished neighbourhoods, domestic violence, or sexual abuse. Data were also not collected on the extent to which those in the sample had been contacted by outreach workers, attended HIV education sessions, or had knowledge of HIV transmission and prevention.

Acknowledgments

This research was primarily supported by the National Institute on Drug Abuse, “Estimating Current Hard Drug Users and Operatives: Lifestyles and Consequences” (NIDA–5R01 DA09339-05), and by other NIDA grants (5 T32 DA07233-19; RO1 DA14219-04, P30DA11041-07) and by National Development and Research Institutes. The authors acknowledge the many contributions to this research by Barry Randolph, Georgann Wright, Denecia Pearce, Phyllis Curry, James Pride, Howard Lune, Charles Small, and Fatos Kaba. Special thanks to Douglass Williamson and Sonia Statlock from City University of New York’s Center for Studies of the Environment, who linked mapped NYPD drug allegation data to 1990 census block maps and computed the number of drug allegations per housing unit for each census block.

References

  1. Amaro H, Raj A. On the margin: power and women’s HIV risk reduction strategies. Sex Roles. 2000;42(78):723–749. [Google Scholar]
  2. Amaro H. Love, sex, and power: Considering women’s realities in HIV prevention. American Psychologist. 1995;50:437–447. doi: 10.1037//0003-066x.50.6.437. [DOI] [PubMed] [Google Scholar]
  3. Boyd C, Guthrie B, Pohl J, Whitmarsh J, Henderson D. African American women who smoke crack cocaine: sexual trauma and the mother-daughter relationship. Journal of Psychoactive Drugs. 1994;26:243–247. doi: 10.1080/02791072.1994.10472437. [DOI] [PubMed] [Google Scholar]
  4. Boyer CB, Tschann JM, Shafer M. Predictors of risk for sexually transmitted diseases in ninth grade urban high school students. Journal of Adolescent Research. 1999;14:448–465. doi: 10.1177/0743558499144004. [DOI] [PubMed] [Google Scholar]
  5. Brown GR, Rundell JR, McManis SE, Kendall SN, Zachary R, Temoshok L. Prevalence of psychiatric disorders in early stages of HIV infection. Psychosomatic Medicine. 1992;54(5):588–601. doi: 10.1097/00006842-199209000-00006. [DOI] [PubMed] [Google Scholar]
  6. Centers For Disease Control & Prevention. HIV/AIDS surveillance report. 1999;11(1):1–43. [Google Scholar]
  7. Chu SY, Curran JW. Epidemiology of human immunodefiency virus infection in the United States. In: Devita VT, Hellman S, Rosenberg SA, editors. AIDS: Etiology, diagnosis, treatment and prevention. 4 Philadelphia: Lippincott-Raven Publishers; 1997. pp. 137–143. [Google Scholar]
  8. Cross J, Johnson BD, Davis WR, Liberty H. Supporting the habit: Income generation activities of frequent crack users compared with frequent users of other hard drugs. Drug and Alcohol Dependence. 2001;64(2):191–201. doi: 10.1016/s0376-8716(01)00121-1. [DOI] [PubMed] [Google Scholar]
  9. Crum RM, Lillie-Blanton M, Anthony JC. Neighborhood environment and opportunity to use cocaine and other drugs in late childhood and early adolescence. Drug and Alcohol Dependence. 1996;43(3):155–161. doi: 10.1016/s0376-8716(96)01298-7. [DOI] [PubMed] [Google Scholar]
  10. Davis WR, Johnson BD, Randolph D, Liberty H. Gender differences in the distribution of cocaine and heroin in Central Harlem. Drug and Alcohol Dependence. 2005;77:115–127. doi: 10.1016/j.drugalcdep.2004.07.010. [DOI] [PubMed] [Google Scholar]
  11. Davis WR, Johnson BD, Randolph D, Liberty H. Characteristics of hidden status among users of crack, powder cocaine and heroin in Central Harlem. Journal of Drug Issues. 2004;34(1):219–244. doi: 10.1177/002204260403400110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Davis WR, Johnson BD, Randolph D, Liberty H. An enumeration method of determining the prevalence of users and operatives of cocaine and heroin in Central Harlem. Drug and Alcohol Dependence. 2003;72:45–58. doi: 10.1016/s0376-8716(03)00173-x. [DOI] [PubMed] [Google Scholar]
  13. Davis WR, Johnson BD. Criminal justice contacts of users and sellers of hard drugs in Harlem. Albany Law Review. 2000;63(3):877–922. [Google Scholar]
  14. Deren S, Estrada A, Stark M, Williams M, Goldstein M. A multisite study of sexual orientation and injection drug use as predictors of HIV serostatus in out-of-treatment male drug users. Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology. 1997;15:289–295. doi: 10.1097/00042560-199708010-00007. [DOI] [PubMed] [Google Scholar]
  15. Des Jarlais DC, Friedman SR, Perlis TE, Chapman TF, Sotheran JL, Paone D, et al. Risk behavior and HIV infection among new drug injectors in the era of AIDS in New York City. Journal of Acquired Immune Deficiency Syndrome and Human Retroviology. 1999;20:67–72. doi: 10.1097/00042560-199901010-00010. [DOI] [PubMed] [Google Scholar]
  16. Des Jarlais DC, Perlis TE, Friedman SR, Chapman TF, Sotheran JL, Tortu S, et al. Declining seroprevalence in a very large HIV epidemic: Injecting drug users in New York City, 1991–1996. American Journal of Public Health. 1998;88:1801–1806. doi: 10.2105/ajph.88.12.1801. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dew MA, Becker JT, Sanchez J, Caldararo R, Lopez OL, Wess J, et al. Prevalence and predictors of depressive, anxiety and substance use disorders in HIV infected and uninfected men: A longitudinal evaluation. Psychological Medicine. 1997;27(2):395–409. doi: 10.1017/s0033291796004552. [DOI] [PubMed] [Google Scholar]
  18. El Wood WN, Williams ML, Bell DC, Richards AJ. Powerlessness and HIV prevention among people who trade sex for drugs (‘strawberries’) AIDS Care. 1997;9(3):273–284. doi: 10.1080/713613155. [DOI] [PubMed] [Google Scholar]
  19. Goodman LA. Snowball sampling. Annals of Mathematical Statistics. 1961;32:117–151. [Google Scholar]
  20. Griffiths P, Gossop M, Powis B, Strang J. Reaching hidden populations of drug users by privileged access interviewers: methodological and practical issues. Addiction. 1993;88(1):617–626. doi: 10.1111/j.1360-0443.1993.tb02036.x. [DOI] [PubMed] [Google Scholar]
  21. Heckathorn DD. Respondent-driven sampling II: Deriving valid population estimates from chain-referral samples of hidden populations. Social Problems. 2002;49(1):11–34. [Google Scholar]
  22. Heckathorn DD. Respondent-driven sampling: A new approach to the study of hidden populations. Social Problems. 1997;44(2):174–199. [Google Scholar]
  23. Holmberg SD. Risk factors for sexual transmission of human immunodefiency virus. In: Devita VT, Hellman S, Rosenberg SA, editors. AIDS: Etiology, diagnosis, treatment and prevention. 4 Philadelphia, PA: Lippincott-Raven Publishers; 1997. pp. 569–575. [Google Scholar]
  24. Inciardi JA. Crack, crack house sex, and HIV risk. Archives of Sexual Behavior. 1995;24:249–269. doi: 10.1007/BF01541599. [DOI] [PubMed] [Google Scholar]
  25. Jenkins SR. Introduction to the special issue: Defining gender, relationships and power. Sex Roles. 2000;42(78):467–493. [Google Scholar]
  26. Jenkins SR, Coons HL. Psychosocial stress and adaptation processes for women coping with HIV/AIDS. In: O’Leary A, Sweet-Jemmott LS, editors. Women and AIDS: Coping and care. New York: Plenum Press; 1996. [Google Scholar]
  27. Johnson BD, Goldstein PJ, Preble E, Schmeidler J, Lipton DS, Spunt B, et al. Taking care of business: The economics of crime by heroin abusers. Lexington, MA: Lexington Books; 1985. [Google Scholar]
  28. Kelling GL, Coles CM. Fixing Broken Windows: Restoring Order and Reducing Crime in our Communities. New York: Free Press; 1996. [Google Scholar]
  29. Maslow CB, Friedman SR, Perlis TE, Rockwell R, Des Jarlais DC. Changes in HIV seroprevalence and related behaviors among male injection drug users who do and do not have sex with men: New York City, 1990–1999. American Journal of Public Health. 2002;92(3):382–384. doi: 10.2105/ajph.92.3.382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. New York City Department of Health. AIDS New York City: Surveillence update, first quarter 1999. New York, NY: Department of Health, Office of AIDS Surveillance; 1999. [Google Scholar]
  31. New York City Department of Health. AIDS New York City: Surveillence update, first quarter 2002. New York, NY: Department of Health, Office of AIDS Surveillance; 2004. [Google Scholar]
  32. Nicolosi A, Correa-Leite ML, Musicco M, Arici C, Gavazzeni G, Lazzarin A. The efficiency of male-to-female and female-to-male transmission of the human immunodefiency virus: A study of 730 stable couples. Epidemiology. 1994;5(6):570–575. doi: 10.1097/00001648-199411000-00003. [DOI] [PubMed] [Google Scholar]
  33. Padian NS, Shiboski SC, Glass SO, Vittinghoff E. Heterosexual transmission of human immunodeficiency virus (HIV) in northern California: Results from a ten-year study. American Journal of Epidemiology. 1997;146:350–357. doi: 10.1093/oxfordjournals.aje.a009276. [DOI] [PubMed] [Google Scholar]
  34. Pulerwitz J, Gortmaker SL, DeJong W. Measuring sexual relationship power in HIV/STD research. Sex Roles. 2000;42(78):637–660. [Google Scholar]
  35. Snijders TAB. Estimation on the basis of snowball samples: How to weight? Bulletin de Methodologie Sociologique. 1992;36:59–70. [Google Scholar]
  36. Summers J, Zisook S, Atkinson JH, Sciolla A, Whitehall W, Brown S, et al. Psychiatric morbidity associated with acquired immune deficiency syndrome-related grief resolution. Journal of Nervous Mental Disorder. 1995;183:384–389. doi: 10.1097/00005053-199506000-00006. [DOI] [PubMed] [Google Scholar]
  37. Wagner FA, Anthony JC. Into the world of illegal drug use: exposure opportunity and other mechanisms linking the use of alcohol, tobacco, marijuana, and cocaine. American Journal of Epidemiology. 2002;155(10):918–25. doi: 10.1093/aje/155.10.918. [DOI] [PubMed] [Google Scholar]
  38. Watters JK, Biernacki P. Targeted sampling: options for the study of hidden populations. Social Problems. 1989;36(4):416–430. [Google Scholar]

RESOURCES