Between-City Variation in Frequency of Injection Among Puerto Rican Injection Drug Users: East Harlem, New York, and Bayamon, Puerto Rico

Frequency of injection has been consistently found to be higher among Puerto Rican Injection drug users (IDUs) than among other groups of IDUs. Several explanations have been suggested, but an empirical explanation has yet to be presented. This study compares the frequency of injection of Puerto Rican IDUs in East Harlem, New York, with that of IDUs in Bayamon, Puerto Rico. Study subjects comprised 521 Puerto Rican IDUs from East Harlem and 303 IDUs from Bayamon. The mean frequency of injection among IDUs in East Harlem was 2.8, the corresponding mean in Bayamon was 5.4. Younger IDUs reported a higher number of daily injection episodes than older IDUs, and the IDU group in Bayamon was 5 years younger than the group in East Harlem. The drug use variables accounted for a greater portion of the between-city difference than the demographic and psychosocial variables. Use of noninjected drugs, as well as the use of prescribed methadone, were found to be associated with a lower number of daily injections. Conversely, injection of cocaine, injection of cocaine mixed with heroin (speedball), and injection of larger amounts of drug solution were found to be associated with a higher number of daily injections.

Puerto Rican injection drug users (IDUs) have been shown to differ from other groups of IDUs in several behaviors that put them at increased risk of infection with HIV and other blood-borne pathogens. In particular, frequency of injection has been consistently found to be higher among Puerto Rican IDUs than among other groups of IDUs ^(1-8). Several explanations for the high frequency of injection among Puerto Rican IDUs have been suggested ⁽⁹⁾, but an empirical examination of this pattern has yet to be presented.

Understanding the behaviors that place Puerto Rican IDUs at heightened risk of infection with HIV is of critical concern because Puerto Rico is one of the HIV epicenters of the United States, and the HIV seropositivity rates of Puerto Rican IDUs are among the highest rates reported in all the U.S. states and territories ^(4,10). Puerto Rico ranks fifth in AIDS incidence rates after Washington, D.C., New York State, Florida, and the U.S. Virgin Islands ⁽¹¹⁾. In addition, San Juan has the ninth highest AIDS incidence rate among large metropolitan areas in the U.S. ⁽¹¹⁾. Puerto Ricans have also been identified as the Latino group with the highest prevalence of AIDS in the U.S. mainland ⁽¹²⁾, and injection drug use is the most common AIDS risk category among Puerto Ricans both in the island of Puerto Rico and in the U.S. mainland ^(13,14).

Ascertaining the sources of ethnic and geographic variation in the frequency of injection is also of public health importance in the design of preventive interventions. Several researchers have called attention to the need to understand geographic and ethnic variations in HIV-related behaviors to be able to design interventions that are sensitive to the unique characteristics of particular populations and effective in changing or modifying the targeted behaviors ^(15-17). Moreover, frequency of drug injection has received only limited comparative or analytic attention ⁽⁵⁾.

Previous studies have examined several potential determinants of frequency of drug injection. Frequency of injection has been found to vary significantly by several demographic and psychosocial factors such as age, gender, city of study, ethnicity, and presence of depression ^(6,17-21). Drug use factors, such as types of drugs injected and enrollment in methadone maintenance programs, have also been found to affect frequency of injection ^(18-25). Several researchers have also proposed that variations in the frequency of injection across cities and ethnic groups might be indicative of site-or ethnic group-specific traditions and customs of drug consumption passed on as part of the socialization process among drug users ^(5,26).

This study compares the frequency of injection of Puerto Rican IDUs in East Harlem (EH), New York with that of IDUs in Bayamon (BY), Puerto Rico. The aim of this study is to identify the factors that account for the difference in the injection frequency of Puerto Rican IDUs between these two cities. In addition to demographic and psychosocial factors, we examined the use of injected as well as the use of noninjected drugs and their influence on the between-city variation in injection frequency. We also examined the amounts of drug solution injected and whether the amounts affected the injection frequencies. The well-known pattern of back-and-forth mobility of Puerto Ricans between mainland and island cities ^(10,27) also offered us the opportunity to assess site of drug use socialization as another factor potentially influencing between-city variations in injection frequency.

METHODS AND PROCEDURES

Sampling and Study Subjects

Data for this study were collected as part of the ongoing Alliance for Research in El Barrio and Bayamon (ARIBBA) project. The catchment areas of ARIBBA comprise the EH section of New York City, and the urban section of the Municipality of Bayamon in Puerto Rico, which is part of the San Juan metropolitan area. Ethnographic mapping procedures were conducted to locate the areas where drugs are procured (copping areas), shooting galleries (premises where drugs are purchased and injected), venues where streetwalking is common (prostitution strolls), and other locations drug users frequent (hang-outs) in both catchment areas ⁽²⁸⁾. Before initiating study subject recruitment, the geographic distributions of these locales were used to divide the catchment areas into sectors. Each sector covered a similar number of city blocks that could be walked through in less than an hour. The sectors were continuously monitored during the recruitment phase and new locales mapped. Monthly plans of random visits to the sectors were developed for daily recruitment that covered daylight hours, Mondays through Fridays. Concerns about the personal security of researchers impeded recruitment during evening hours.

On the predetermined day, sector, and time, outreach workers approached a drug user who had not been previously recruited in the study, determined eligibility, explained the study objectives and activities, and invited him or her to participate. To be eligible, study subjects had to self-define as being of Puerto Rican ethnicity, had to have injected drugs or smoked crack cocaine during the last 30 days, be ≧18 years old or more, and have not been in an in-patient drug treatment program in the previous 30 days. Consenting and eligible study subjects were taken to an assessment facility. At the assessment site, an informed consent previously approved by the institutional review boards (IRBs) of each research group was verbally explained and study subjects were asked to agree to participate and sign the informed consent. The OnTrak urinalysis TestStik system (Roche Diagnostic Systems, Inc., Branchburg, NJ, U.S.A.) was used to confirm recent use of heroin or cocaine. On consent, participants were assessed with a structured interview and offered HIV testing and counseling.

Sampling and recruitment of participants were conducted between January 1998 and August of 1999 and 1,200 drug users completed the baseline assessment: 800 in New York City and 400 in Puerto Rico.

Measures

Preliminary ethnographic field research provided the basis for designing the behavioral questions of the measurement instrument. The interview protocol was designed in English by both teams of researchers and was then translated into Spanish. The accuracy of the translation was verified through back translation. The pilot protocol was tested in both sites and the feedback provided by the interviewers was used to rephrase some of the questions. Field personnel conducted cross-site visits to ensure the comparability of the recruitment procedures and the interviewing process.

Frequency of Injection

Respondents were asked when had they injected last and how many times had they injected on that day. Respondents who had injected the last time more than seven days before the interview day were excluded from the analyzes to minimize recall errors. In EH, 555 participants reported injecting during the previous 30 days; 521 of these had injected last during the 7 days before the study interview. In BY, 308 participants had injected during the previous 30 days, and 303 of these had injected last within the previous 7 days.

Demographic and Psychosocial Factors

Consistent with previous studies of IDUs, schooling was recoded for less than a high school education, and high school education or more. Homelessness was operationally defined as living in the street, or in a shelter. Depression was measured using the Center for Epidemiological Studies Depression scale (CES-D) ⁽²⁹⁾, a 20-item instrument that screens for depression symptoms occurring during the last 2 weeks. The CES-D has been shown to be a valid screening measure of depression ⁽³⁰⁾ and has been normed to the population of Puerto Rico ⁽³¹⁾. Three questions ascertained city of drug use socialization: Where did you first used drugs, Where did you first inject, and Where did you first inject on a regular basis. The responses to all three questions were highly intercorrelated and their associations with injection frequency were very similar. For this study, we used the response to the question about city of first drug injection as a proxy of city of injection drug use socialization.

Drug Use Factors

Respondents were asked about their noninjection use of 13 types of drugs. Use of barbiturates, amphetamines, hallucinogens, and inhalants were reported by fewer than 10 IDUs and were not included in the analyses. Respondents were also asked whether they were currently taking methadone as part of a methadone maintenance program (i.e., prescribed methadone).

The injected drugs included in the analyses were also those reported by at least 10 respondents: cocaine (52.9%), heroin (78.7%), and speedballs (an injected mixture of heroin and cocaine) (65.9%). Respondents were also asked what had been the minimum and maximum amount of drug solution injected on the last day of injection. Responses to these questions were measured with the syringe calibrations in tenths of a milliliter. Interviewers had 1-ml and 0.5-ml syringes for the respondents to indicate the minimum and maximum amounts injected.

Analyses

We first examined the distributions of the frequencies of injection. Measures of central tendency and dispersion were computed and the distributions were graphed. The two study groups were then compared in terms of their demographic and psychosocial characteristics, and their drug use profiles. χ ² Tests of homogeneity independence were used to assess the significance of the differences. The marked departure from normality of the injection frequency distributions and their right skewness showed that the assumption of normality of ordinary least squares regression would be violated. Thus, to assess the effect of each independent variable on the frequency of injection, we employed Poisson regression as recommended by Gardner et al. ⁽³²⁾. The independent variables were entered into the Poisson model in blocks, beginning with the demographic and psychosocial factors, followed by the drug use factors. Variance estimates were adjusted for overdispersion by multiplying the variance matrix by the deviance of the model divided by its degrees of freedom. The SAS procedure GENMOD was employed to estimate the parameters of the Poisson regression (SAS Institute Inc., Cary, NC, U.S.A.).

Finally, the effect of site of drug injection socialization on the drug use profiles was examined by the use of logistic regression. For each drug use variable, a logistic model was fitted with drug use socialization site as the independent variable of interest, and study site, age, gender, education, and depression as covariates.

RESULTS

Table 1 shows the measures of central tendency and dispersion of the frequencies of injection reported by both study groups, based on the last day of injection. The mean frequency of injection among Puerto Rican IDUs in EH was 2.8, the corresponding mean in BY was almost twice as high, 5.4 (p < .001). Figure 1 depicts the frequency distribution of the number of injections reported by the two study groups. As can be seen in Figure 1, the distributions are markedly skewed to the right. The difference in the median number of injection episodes was also statistically significant (2 in EH vs. 4 in BY, p < .001); and the ranges in both distributions were very similar (25 in EH vs. 24 in BY).

Table 1

Fig. 1

Table 2 compares both groups of IDUs in terms of their demographic and psychosocial characteristics. The sample in EH had a larger proportion of females (21.1% in EH vs. 14.8% in BY;p = .026). IDUs in EH were also older, mean age was 38.4 in EH versus 33.2 in BY (data not shown, p < .001). There was no significant difference in the level of schooling with 42.3% of the IDUs in EH having completed high school and 45.5% in BY (p = .382). A higher proportion of study participants in East Harlem reported homelessness than in BY (34.1% in EH vs. 23.2% in BY, p = .001). In both study sites, more than half these IDUs scored above the cutoff point for depression in the CES-D scale, although the proportion was higher in Bayamon (54.7% in EH vs. 63.0% in BY;p = .020). Nearly a third of the IDUs recruited in East Harlem had initiated drug injection in Puerto Rico, but only 10% of the IDUs in Bayamon had initiated drug injection in New York City or in another U.S. city (p < .001).

Table 2

Table 3 shows the proportion of participants that used each type of drug during the prior 30 days. IDUs in Bayamon reported significantly lower levels of use of all non-injected drugs. The only exception was the use of sedatives/hypnotics which was reported by similar proportions of IDUs in both sites (31.9% vs. 32.0%, p = .977). IDUs in Bayamon were also significantly less likely to be currently taking either prescribed or non-prescribed methadone than IDUs in East Harlem (prescribed methadone, 54.4% in EH vs. 10.4% in BY;p < .001; non-prescribed, 21.3% in EH vs. 3.7% in BY, p < .001). IDUs in EH were more likely to inject heroin alone than their counterparts in BY (87.5% vs. 64.2;p < .001). compared with IDUs in EH, IDUs in BY were more likely to report injection of cocaine alone and of heroin and cocaine together (cocaine alone, 46.3% in EH vs. 66.6% in BY;p < .001; heroin and cocaine together, 53.9% in EH vs. 91.1% in BY;p < .001). The maximum amount of drug solution injected also differed in the two groups of IDUs. IDUs in BY reported higher maximum amounts of drug solution than IDUs in EH. The mean maximum amount of drug solution injected was 0.03 ml in EH and 0.035 ml in BY (p < .001, data not shown).

Table 3

Results of the Poisson regression are shown in Table 4. The β coefficients have been exponentiated to facilitate their interpretation. In the Poisson regression, the exponential of a β coefficient is the factor by which a unit increase (or decrease) in the independent variable multiplies (or divides) the expected frequency of injection. In the first model, study site (BY vs. EH) was estimated to have the effect of doubling the expected frequency of injection (exponential of β = 1.91), a similar effect to that observed using bivariate analysis in Table 1. This effect was reduced to 1.70 when the demographic and psychosocial factors were entered in the model. Age was negatively associated with the expected frequency of injection with younger IDUs injecting more frequently than their older counterparts. Homelessness was also found to significantly increase the expected frequency of injection by 14% (p = .019). A similar effect of a 14% increase in expected frequency of injection was estimated among IDUs who started injecting drugs in Puerto Rico, although the effect was only marginally significant (p = .051). After entering the drug use factors, the between site difference was reduced from 1.70 to 1.19, although it remained statistically significant (p = .018). Controlling for the drug use factors, the only demographic/psychosocial factor that remained significantly associated with frequency of injection was age. Of noninjected drugs, the use of marijuana, heroin, and prescribed methadone were significantly associated with frequency of injection; all these factors had the effect of reducing frequency of injection. Use of marijuana had the effect of reducing the expected frequency of injection by 12% (p = .015), and both heroin and prescribed methadone by 17% (p = .010 and .006, respectively). Injection of cocaine alone and injection of speedballs were both found to increase the expected frequency of injection by about 30% (p < .001 in both cases). Every increase of 0.1 ml in the maximum drug solution injected was also estimated to increase the expected frequency of injection by 10% (p < .001).

Table 4

Given that many of the drug use variables were found to be significantly associated with frequency of injection, we surmised that site of drug injection socialization might not affect frequency of drug injection directly, but indirectly through affecting the drug use factors. Each drug use variable was regressed against site of drug injection socialization. Study site, age, gender, education, and depression were entered into logistic regression models as covariates. Site of drug injection socialization was not found to be associated with the drug use variables in any of the models (data not shown).

DISCUSSION

Injection drug users in Puerto Rico reported close to twice as many injections per day as those reported by their counterparts in New York City. The two groups of IDUs differed considerably in their demographic and psychosocial profiles. However, the only demographic or psychosocial variable that helped to account for a significant portion of the between-city differences in the frequency of injection was age. Younger IDUs reported a higher number of daily injection episodes than older IDUs, and the IDU group in Puerto Rico was, on average, 5 years younger than the group in New York City.

The two groups of IDUs also differed considerably in their drug use profiles. The drug use variables accounted for a greater portion of the between-city differences in injection frequency than demographic and psychosocial variables. Use of noninjected drugs, specifically marijuana and noninjected heroin, were found to be associated with a lower number of daily injections. Participation in methadone maintenance treatment programs (MMTPs, i.e., with use of prescribed methadone) was also found to be associated with a lower injection frequency. Conversely, injection of cocaine, injection of heroin and cocaine mixed together (speedballs), and injection of larger amounts of drug solution were found to be associated with a higher number of daily injections.

Several studies have found that newly initiated or young IDUs are at higher risk of infection with blood-borne pathogens than those with more experience, or older IDUs ^(33-38). In our study, the number of daily injections was negatively associated with age, that is, younger IDUs were found to inject more frequently than older IDUs. Higher frequency of injection of young IDUs may in part explain earlier findings of higher risks HIV among more recent IDUs. Sensitization, or an increase in the effect of psychostimulants with repeated administration, could provide a possible explanation for the negative age trend in the number of daily injections ^(39,40). It is conceivable that individuals who have been sensitized to a drug for years need to administer the drug less frequently to achieve a similar action. Nevertheless, the reasons why younger IDUs might be injecting more frequently than older IDUs have yet to investigated.

We failed to find an association between depression and frequency of injection whereas several other studies have found such an association ⁽¹⁹⁾. The measure of depression used in this study, the CES-D scale, has been found to be a valid screening measure for Puerto Rican populations and has been normed to the population of Puerto Rico ⁽³¹⁾. However, in both study sites, more than half the IDUs scored above the cut-off point for depression in the CES-D scale. There might not have been a sufficient number of respondents with low scores in the CES-D to provide the scale the ability to discriminate between low and high frequency injectors.

Homeless IDUs have been found to report higher numbers of injections than nonhomeless IDUs ^(41,42). In this study, homelessness was significantly associated with frequency of injection. However, after controlling for the drug use variables in the Poisson regression, the association was no longer statistically significant. Thus, the effect of homelessness on frequency of injection seems to have been due, in our study, to differences in the patterns of drug use.

Several investigators have proposed that variations in the frequency of injection across cities and ethnic groups might be indicative of site-or ethnic-specific traditions and customs of drug consumption passed on as part of the socialization process of drug users ^(5,26). We used geographic site of first injection as a proxy for site of socialization into drug injection and no association was found between this variable and frequency of injection. We also tested the responses to other questions that could indicate differential drug use socialization such as Where did you first used drugs, and Where did you first inject on a regular basis. After controlling for study site, none of these variables showed a significant association with frequency of injection. We further tested the possibility that site of drug injection socialization might not affect frequency of drug injection directly, but indirectly through affecting the types of injectable and noninjectable drugs used. Site of drug injection socialization was not found to be associated with any drug use variables. Thus, it appears that site of drug injection socialization had no noticeable influence on the differences in frequency of injection between the two groups of IDUs. Practices learned during early injection might not be able to persist when faced with changes in the drug injection conditions or when IDUs move to another community with different injection customs. In the case of Puerto Ricans, however, frequent travels back and forth between island and mainland communities ^(10,27) might have an additional effect of attenuating site differences in drug use customs.

Consistent with other previously published results, we found that IDUs enrolled in MMTPs reported fewer daily injections than IDUs not in MMTP programs ^(24,25). We should note, however, that our study involved only currently active IDUs and did not include MMTP participants that had completely stopped injecting drugs. Thus, our test of the potential effects of MMTPs in reducing the frequency of drug injection is rather a more stringent one than those that have been published before. During the period of study there were five MMTPs operating in the EH area and only one in BY. Not surprisingly, more than half of the IDUs recruited in EH reported being in an MMTP and only 10% of the IDUs in BY reported current participation in an MMTP.

We had reported elsewhere that, compared with IDUs in several mainland sites, IDUs in Puerto Rico used a narrower array of drugs and were less likely to use noninjected drugs ⁽⁷⁾. This previous finding was replicated in the current study. The findings of the current study further suggest that this pattern of use of fewer types of drugs and less use of noninjected drugs also accounted in part for the higher number of injections of IDUs in Puerto Rico. More specifically, marijuana and noninjected heroin were found to be associated with fewer number of daily injections and to be more commonly used among IDUs in New York.

Injection of cocaine and injection of speedballs had the largest estimated effect over frequency of injection. Injection of cocaine alone and injection of speedballs were found to significantly increase the expected frequency of injection by about 30% each. Moreover, injection of both cocaine and speedballs were found to be quite common in Puerto Rico; Nearly two thirds of the IDUs in BY reported cocaine injection and virtually all reported injection of speedballs (91.1%). Cocaine is known to have a shorter half-life than heroin, and cocaine-addicted individuals have been found to consume cocaine with greater frequence and are more likely to report binge episodes than users addicted to other drugs ^(22,23,43). Moreover it has been demonstrated that speedballs have synergistic effects in the elevations of important neurotransmitters in the brain, which are related to drug self-administration and reinforcement ⁽⁴⁴⁾. This synergistic effect could explain why speedball use was found to increase the frequency of injection after controlling for the effect of both heroin and cocaine injection.

We also found that the maximum amount of drug solution injected on the last day of injection was associated with a higher injection frequency. We had presumed that the maximum amount injected indicated either the level of tolerance to the drug the respondent had developed, or the level of purity of the drug injected. That is, either higher levels of tolerance or drugs of lower purity would require greater amounts to achieve the desired effect. From our data, we cannot discern the extent to which differences in the amount of drug injected between the two study groups might be due to differences in the drugs available in the local drug markets (e.g., price, purity, availability). We do note, however, that according to the information available at the U.S. Drug Enforcement Administration, both the purity and price of street heroin in EH and BY during the study period were quite similar (average purity of 53.1% ± 4.3% in the Bayamon area and 53.9% ± 2.8% in the EH area, Michael Chapman, personal communication). Nevertheless, large differences in the drug use profiles of two groups of IDUs that are of the same ethnic origin and that are in frequent contact through travels, suggest important differences in other characteristics of the local drug markets such as the array of drugs available, particularly noninjectable drugs, such as marijuana. Few studies have examined how local drug scenes influence drug using behaviors.

There are some limitations in this study that merit comment. As is the case with most studies of IDUs, our samples cannot be considered to be representative of the IDUs in the study catchment areas. We implemented several sampling procedures to reduce selection biases and ensure the comparability of recruitment across the two study sites: ethnographic mapping of the street sites where IDUs could be found, random visit plans to the mapped locales, and cross-site visits of the field personnel. Notwithstanding these sampling procedures, recruitment of participants exclusively during daylight hours might have contributed to some selection bias. Another limitation of this study was that the data used in the analyses were based on self-reported information provided by the respondents and were not corroborated otherwise. Only reports of drug use in the 48 hours before the interview were confirmed with urinalysis. To reduce recall biases, the dependent variable in this study was defined as the number of injection episodes reported for the last day of injection and respondents whose last injection had been more than 7 days before the day of the interview were excluded from the analyses. Finally, we lacked resources to perform biologic assays to assess the quantities of drugs consumed. Such measures would have allowed us to determine whether study subjects reporting more frequent injections showed evidence of ingesting greater quantities of drugs.

Notwithstanding these limitations, we believe the results of this study are compelling and raise a number of important implications for the prevention of the transmission of blood-borne pathogens among IDUs and for future studies of IDU behaviors. Programs designed to help IDUs reduce their risks of transmission of blood-borne pathogens need to be cognizant of the large between-community differences in drug use profiles and in frequency of injection. In particular, programs aimed at Puerto Rican IDUs will require more intensive intervention strategies, including expanding access to MMTPs. Needle exchange programs serving Puerto Rican IDUs also need to be aware of the higher number of injections of their participants and the need to exchange greater number of syringes to achieve significant risk reduction ⁽⁴⁵⁾. Moreover, use of larger amounts of drugs of equivalent purity suggests higher levels of intoxication. If so, this could have important implications for the levels of risk faced by IDUs in Puerto Rico and the types of interventions needed to help in reducing such risks effectively. Further studies are also needed to deepen our understanding of younger-age IDUs and the factors-neuropharmacologic, psychological, and social-that can account for their increased risks of transmission of blood-borne pathogens, including increased frequency of injection.

In conclusion, we have been able to increase our understanding of the large difference in injection frequency between Puerto Rican IDUs residing in New York City and those in Puerto Rico. A lower mean age of injectors in Puerto Rico, a wider array of the types of drugs used in New York City, as well as the greater availability of MMTPs in New York, all significantly contributed to the differences in injection frequency. Nonetheless, a small but significant site difference remained after controlling for these variables, indicating that there are still some influences that remain unaccounted for. Environmental influences such as the effects of local drug market characteristics, should be examined in the future.

Acknowledgments:

This study was supported by grant DA 10425 from the National Institute on Drug Abuse. The authors are grateful to Dr. Carlos Jiménez for his assistance in the interpretation of the findings.