Incarceration and Risk for HIV Infection Among Injection Drug Users in Bangkok

Objective: To assess potential multiple relationships between incarceration and HIV infection among injecting drug users (IDUs) in Bangkok. Previous cross-sectional studies have shown strong relationships between incarceration and HIV infection but have not been able to assess potential causal pathways.

Cited Here...: Injection drug users seen at methadone treatment programs in Bangkok were screened during 1995 to 1996 for enrollment into the study. With informed consent, 1,209 seronegative IDUs were enrolled in a cohort study to determine HIV incidence and identify factors associated with incident infections. Follow-up visits were conducted every 4 months, with HIV testing and assessment of risk behaviors.

Cited Here...: Overall incidence rate was 5.8 per 100 person-years (95% confidence interval [CI], 4.8-6.8) of follow-up. A four-step injection risk scale was constructed that included less frequent than daily injection, daily injection, daily injection with reported sharing of injection equipment, and injection while incarcerated. This scale was strongly related to HIV incidence, with incidence approximately doubling for each step in the scale. Incidence rate for follow-up periods that contained drug injection while incarcerated was 35/100 person-years at risk. In multivariate analyses, incarceration was related to incident HIV infection in multiple ways: previous incarceration and recent incarceration without drug injection, and the injection risk scale were all independently predictors of incident HIV infection.

Conclusions: Incarceration is related to incident HIV infection through multiple pathways. Previous incarcerations are likely to serve as markers for unmeasured high-risk behaviors, and it is also highly likely that HIV is transmitted during periods of incarceration. Programs to reduce HIV transmission in jails and prisons, including drug abuse treatment of inmates and programs to reduce the likelihood of incarceration of IDUs, are needed urgently. Given the current diffusion of injecting drug use, of HIV infection among drug injectors, and of the common policy of incarcerating drug users, it is very likely that the problem of HIV transmission in jails and prisons is increasing in many countries throughout the world.

The practice of injection of illicit psychoactive drugs has been spreading rapidly throughout the world over the last 2 decades ^(1,2). Injecting drug users (IDUs) are at risk for both incarceration and for infection with HIV. They are at risk for incarceration because the possession of the drugs is illegal, and they also may engage in illegal activities to obtain funds for their drugs. The actual periods of incarceration, however, can vary greatly from a few hours or days of detention to lifetime incarceration, or even to execution for persons convicted of drug dealing. Because incarceration is not an effective treatment of addiction, IDUs are usually at risk for multiple episodes of incarceration. IDUs are also at high risk for HIV infection because the multi-person use (sharing) of their drug injection equipment is a relatively effective method for transmitting HIV. They are also at risk for becoming infected with HIV through sexual activity and for transmitting HIV to their non-injecting sexual partners.

There is reason for concern that incarceration may actually increase the risk that an IDU will become infected with HIV. Most jails and prisons do not provide any form of drug abuse treatment, so that an addicted user will continue to need his or her drugs when incarcerated. The concentration of large numbers of drug users in an institution creates strong incentives to smuggle drugs into the institution, and a high likelihood that drug use will occur inside the institution. For example, Gaughwin et al. ⁽³⁾ estimated that 30% of inmates in Australian prisons inject drugs while incarcerated. Access to sterile injection equipment is usually very restricted within jails and prisons, so that if drug injection does occur, it is likely to be with shared injection equipment. Estimates of HIV incidence among prisoners in the United States ^(4-6) and Australia ⁽⁷⁾ suggest relatively low incidence, but with the potential for rapid outbreaks of HIV infection ⁽⁸⁾.

Considerable evidence suggests that incarceration of IDUs may be an important component of the HIV epidemic among IDUs in Bangkok, Thailand. It has been hypothesized that a high proportion of the initial transmission of HIV among IDUs occurred in jails ⁽⁹⁾. A series of cross-sectional HIV seroprevalence studies of IDUs in Bangkok have all shown significant associations between having been incarcerated and being infected with HIV:

In a 1989 study, the odds ratio (OR) for having been incarcerated and being HIV seropositive was 2.11 (95% confidence interval [CI], 1.38-3.23) ⁽¹⁰⁾.

In a 1993 study, the OR for having been incarcerated and being HIV seropositive was 4.14 (95% CI, 2.51-6.85), and the OR for having injected drugs while incarcerated and being HIV seropositive was 4.93 (95% CI, 2.93-8.34) ⁽¹¹⁾.

In a 1995 study, the OR for having been incarcerated and being HIV seropositive was 1.7 (95% CI, 1.2-2.3), and the OR for having injected drugs while incarcerated and being HIV seropositive was 2.6 (95% CI, 1.8-3.7) ⁽¹²⁾.

In a 1997 study, the OR for having been incarcerated and being HIV seropositive was 2.32 (95% CI, 1.26-4.31), and the OR for having injected drugs while incarcerated and being HIV seropositive was 3.51 (95% CI, 1.75-7.07) (K. Choopanya, unpublished data).

These studies show consistent relationships between incarceration and HIV infection among IDUs in Bangkok, but the cross-sectional design of the studies does not provide much insight into possible causal mechanisms for the relationship. Substantial HIV transmission may be occurring during incarceration or incarceration may be a marker for other HIV risk behaviors, or both these processes may be occurring. We have recently conducted a longitudinal cohort study among IDUs in Bangkok in which the overall HIV incidence rate was 5.8/100 person-years at risk. Data from this cohort have shown low income, drug injection frequency, sharing of injection equipment, incarceration, and drug injection while incarcerated were statistically independent risk factors for incident HIV infection ⁽¹³⁾. The present report is an extended analysis of these cohort-based data. This extended analysis permits us to 1) place drug injection while incarcerated within a broader context of injection risk behaviors, 2) assess multiple causal pathways linking incarceration to incident HIV infection, 3) identify factors that are protective against injection risk behavior, and 4) identify factors protective against incarceration.

METHODS

A full description of the methods has been presented previously ^(12,13); an abbreviated description of the methods is presented here.

Study Subject Recruitment

Potential subjects were recruited from people seen at any of 15 Bangkok Metropolitan Administration (BMA) drug abuse treatment clinics. A full informed consent was obtained, a blood sample for HIV testing was obtained, and trained interviewers administered a brief, standardized questionnaire. After the HIV test result was known, each volunteer was counseled confidentially about his or her HIV status. HIV-seronegative persons were offered enrollment with informed consent into a prospective cohort study. On each study visit, participants received 200 Thai baht (about US$8 in 1996). Screening and enrollment were conducted from May through November 1995 and May through December 1996. A description of data from the 1995 screening phase has been published previously ⁽¹²⁾.

A second HIV test was administered to exclude persons who might have been in the process of seroconversion at the initial test. Participants were asked to return for follow-up visits every 4 months. At each visit, participants were interviewed using a standard questionnaire, had blood collected for serologic testing, and received HIV risk-reduction counseling and health education. Bleach for cleaning injection equipment and male condoms were provided at no charge. Sterile syringes and needles are available at pharmacies and convenience stores throughout Bangkok without prescription for about 5 to 8 Thai baht (about US$0.20 to $0.32 in 1996).

Participants who became HIV seropositive were offered follow-up enrollment under a separate protocol on clinical, immunologic, and virologic evaluation of HIV seroconverters. They received social and medical care according to BMA guidelines.

Laboratory Testing

Serum specimens were tested for HIV antibodies by enzyme immunoassay (EIA) at the BMA laboratory as described previously ⁽¹³⁾. During cohort follow-up, HIV-seropositive specimens identified during EIA were tested at the HIV/AIDS collaboration laboratory with another EIA (Genetic Systems, Redmond, WA, U.S.A.) and Western blot test (Novapath HIV-1 Immunoblot, BioRad, Hercules, CA, U.S.A.); specimens positive on EIA and Western blot were considered HIV seropositive.

Data Management and Statistical Analyses

We used Epi-Info (v. 6.04c; Centers for Disease Control and Prevention [CDC], Atlanta, GA, U.S.A.), SAS (v. 6.12; SAS Institute, Cary, NC, U.S.A.) and Stata (v. 6; Stata Corporation, College Station, TX, U.S.A.) for data management and analyses. Incidence rates for HIV-1 infection were calculated per 100 person-years (py) of HIV-negative observation; infections were assumed to have occurred at the midpoint of the interval between the last HIV-seronegative and the first HIV-seropositive test results. Exact 95% Poisson CIs were calculated for the estimated HIV-1 incidence rates.

To estimate relative hazard rates (RHs), we used a proportional hazard regression model. The discrete or grouped time analog of the continuous time proportional hazards model was employed; the grouped time points of infection being each 4-month follow-up visit. All models were stratified by year of enrollment. For time-dependent factors (e.g., drug injection risk scale, recent incarceration) the person-time calculated was based on the status reported between consecutive 4-month visits.

Proportional odds logistic regression ⁽¹⁴⁾ was used to investigate possible predictors of a four-point ordinal drug-injection risk scale (described in Results). Because the data for this analysis were longitudinal, we used robust variance estimates that accounted for repeated measures on the same person over time, in calculating CIs and tests of significance. This approach is equivalent to the generalized estimating equation (GEE) methodology using a working independence correlation structure in the analysis of longitudinal data ⁽¹⁵⁾. GEE logistic regression was used to investigate possible predictors of incarceration.

With only a few exceptions (specified below), each risk factor status was measured based on the questionnaire answers regarding exposures preceding the current visit.

Ethics Review

The study protocols were approved by the Ethical Review of Research Committee, Ministry of Public Health, Nonthaburi, Thailand; the Institutional Review Board, CDC, Atlanta, GA, U.S.A.; and the Global Program on AIDS, World Health Organization, Geneva, Switzerland.

Results

Subject Characteristics

In all, 1209 persons were enrolled in the study. All subjects were of Thai ethnicity, men (93.5%) between 20 and 39 years of age (80.2%), and more than 70% had some form of employment. Table 1 presents other demographic and drug use history data for these subjects.

Table 1

Table 1

Most subjects had begun injecting drugs before age 25 and about half had been injecting for >6 years. Heroin was the most commonly used drug and injecting was the most common route of drug administration. In the 6 months before the screening interview, 96.4% reported heroin use, and almost all heroin users had injected; about one third of stimulant users had injected; and 78.5% reported at least daily injection. Almost all (95.9%) IDUs reported they had obtained new injection equipment at pharmacies or drug stores. About one third reported using a syringe or needle used by someone else in the past 6 months. About two thirds of them had been arrested during the time they had been injecting drugs.

The level of sexual activity was modest; 44.3% reported no sex with someone of the opposite sex in the previous 6 months and only 11.2% reported sex with a casual partner.

Follow-Up Rate

Overall, 1124 (93.0%) participants made at least one visit after enrollment and contributed time to follow-up. The median total follow-up time was 29 months (range, 4-40 months). Return for follow-up was 81.9% at 12 months, 62.2% at 24 months, and 47.6% at 36 months.

HIV-1 Incidence

During follow-up, 133 participants seroconverted to HIV-1, yielding an incidence rate of 5.8 per 100 py at risk (95% CI, 4.8-6.8).

Risk Behavior and Incarceration

Incarceration of study participants was fairly frequent during follow-up. In total, 43.1% (of the 1124 subjects who made at least 1 follow-up visit) reported they were incarcerated at least once during follow-up. We did not collect data directly on the lengths of time subjects were incarcerated. From our experience in this study and previous studies in Bangkok, however, we believe that most of the incarcerations were for relatively short periods, often only for several days.

As shown in Table 2, substantial percentages engaged in drug injection and different types of sharing of drug injection equipment during follow-up, and these behaviors were associated with seroconversion in bivariate analyses. The drug injection frequencies, sharing behaviors, and drug injecting while incarcerated were all highly intercorrelated, so that we considered them as a continuum of risk rather than independent risk factors for becoming infected with HIV. We constructed a four-point scale of increasing drug injection risk behavior that combined injection frequency, sharing behaviors, and injecting while incarcerated:

Table 2

Less than daily drug injection.

Daily or more frequent drug injection, without any reported sharing behaviors.

Daily or more frequent drug injection, with one or more reported sharing behaviors. The sharing behaviors included 1) injecting drugs with a needle or syringe that had been used by someone else, 2) sharing a drug solution, and 3) dividing a drug solution with a needle and syringe that had already been used.

Drug injection while incarcerated. This was considered to be the highest risk drug injection behavior, in that: 1) access to injection equipment would be extremely limited in this setting, 2) drug users would probably not be able to select or limit the number of their sharing partners in this setting, and 3) a high percentage of sharing partners would be infected with HIV.

As shown in Table 3, this drug injection risk scale was strongly related to HIV incidence among subjects in this study, with incidence doubling for each step in the risk scale.

Table 3

To assess possible independence of incarceration from injection risk behavior (including injecting while incarcerated), we created two incarceration variables: 1) recent incarceration without reported drug injection during the immediate follow-up period, and 2) previous incarceration. Previous incarceration was the sum of reported incarcerations from the time of first injection (taken from the baseline questionnaire) through the follow-up period preceding the immediate follow-up period. We used a square-root transformation of this sum to avoid undue influence of unusually high values.

We then entered the injection risk scale, recent incarceration without drug injection and previous incarceration into a multivariate model to predict seroconversion. (This model also included monthly income, the only other independent risk factor for seroconversion in this cohort [13]). As shown in Table 4, injection risk scale, recent incarceration without injection, and previous incarceration were all independent predictors of seroconversion. (Note also that there were minimal changes in the relative risks from the univariate analyses to the multivariate analysis.)

Table 4

Population Attributable Risk Related to Incarceration

Although only a modest proportion of follow-up periods included incarceration, HIV incidence related to incarceration constituted a meaningful proportion of total HIV incidence. In only 2% of the follow-up time periods did subjects report injection while incarcerated, but 14% (19 of 133) of the incident HIV infections occurred during follow-up periods during which the subjects reported drug injection while incarcerated. Only 11% of the follow-up time periods involved incarceration (with or without drug injection while incarcerated), but these were associated with 31% of the incident HIV infections.

Factors Associated With Injection Risk Behavior

Given the very strong relationship between the four-point injection risk behavior scale and HIV seroconversion in this cohort, we then examined potential risk factors for engaging in injection risk behavior. Table 5 presents factors associated with the injection risk scale in univariate and multivariate analyses. There were two factors significantly associated with injection risk behavior in the multivariate analyses: younger age and being in methadone detoxification treatment rather than methadone maintenance treatment.

Table 5

Factors Associated With Incarceration

We also examined factors associated with incarceration during the follow-up period. Forty-three percent of these study subjects were incarcerated at least once during the study follow-up. Table 6 presents those factors that were statistically significant predictors of incarceration during follow-up. Increased age and being in methadone maintenance treatment were protective against incarceration because they were protective against higher values on the injection risk scale that includes injecting while incarcerated. Less education, heroin use, and previous incarcerations were also positively associated with recent incarceration.

Table 6

DISCUSSION

Drug injection while incarcerated, recent incarceration without drug injection, and previous incarcerations all were independent risk factors for HIV seroconversion in this study. This suggests multiple causal pathways. There was a very high seroconversion rate associated with injection while incarcerated-35/100 py at risk. This rate is more than double the rate among subjects who reported high injection frequency and sharing of injection equipment outside of incarceration-17/100 py at risk. Although we cannot document the exact times or locations in which the new HIV infections occurred-they might occur just before incarceration or immediately after release-it would appear extremely likely that some injection-related HIV transmission is occurring in the jails. This injection setting clearly has many characteristics (limited access to sterile injection equipment, limited ability to select or limit the number of sharing partners, and a high probability of HIV-infected sharing partners) that would lead to high rates of HIV transmission. As already noted above, in-prison outbreaks of HIV transmission have been reported in Scotland ⁽⁸⁾ and Australia ⁽¹⁶⁾.

Incarceration without concurrent drug injection was also an independent risk factor for HIV seroconversion in this study. This suggests that recent incarceration is serving as a marker for risks that were not fully measured in the study. These may be standard risk behaviors not reported by some subjects, such as sharing injection equipment. Recent incarceration may also be serving as a marker for unreported sexual risk behavior while incarcerated. (Although sexual risk behavior was measured in this study, it was not associated with higher HIV incidence, and we did not ask about specific sexual behaviors while incarcerated.) Recent incarceration may also be a marker for having risk partners who are very likely to be infected with HIV. As noted in the introduction, IDUs in Bangkok with histories of incarceration are particularly likely to be HIV seropositive.

Previous incarceration (before the period in which seroconversion occurred) was also independently associated with HIV seroconversion. Given the time lag between previous incarceration and seroconversion, previous incarceration must be seen as a marker for unreported risk behavior and/or for engaging in risk behavior with HIV-positive partners.

Bangkok may provide an important example of multiple associations between incarceration and increased risk for HIV infection; the problem is very likely to exist in many countries throughout the world as well. In recognition of this problem, a few countries have developed HIV prevention programs that target incarcerated individuals. Several different types of programs have been implemented:

Education about AIDS, including peer education ^(17,18).

Provision of drug abuse treatment, including medically supervised detoxification to relieve withdrawal symptoms, long-term methadone maintenance, and long-term drug-free counseling treatment ⁽¹⁹⁾.

Provision of bleach to disinfect injection equipment used in prisons ^(20,21).

Provision of sterile injection equipment through vending machines that exchange new needles and syringes for used needles and syringes ^(22,23).

Implementing HIV prevention programs for incarceration requires adopting a public health perspective on drug use rather than a sole reliance on a law enforcement perspective. Educational counseling programs certainly should be provided in all areas where incarcerated IDUs are not fully informed about HIV and AIDS. The subjects in this study did not lack necessary information, but providing information by itself should not be considered a sufficient intervention for controlling potential HIV transmission among IDUs. Providing either bleach for disinfecting drug injection equipment and/or sterile injection equipment in jails and prisons would face extremely difficult policy issues in almost all countries in the world. It clearly would be imprudent to rely solely on adopting these interventions as methods for addressing the problem of HIV among incarcerated IDUs.

Providing drug abuse treatment to incarcerated IDUs may offer the best combination of acceptability and effectiveness for addressing the problem of HIV transmission among incarcerated IDUs. Treatment to relieve withdrawal symptoms among inmates might be particularly effective in reducing the likelihood of high-risk injection behavior. Long-term treatment that continued after release should provide even greater advantages in reducing HIV transmission, because the period immediately after release may also be a time of very high risk for infection. Providing such treatment does, however, require substantial resources. Discussions are currently under way in Bangkok about providing drug abuse treatment to incarcerated IDUs, but the resource question is very real. We would hope that these Bangkok data would serve as an impetus to greater recognition of this public health problem and to providing more drug abuse treatment of IDUs in areas where greater resources are available.

Preventing HIV transmission related to incarceration should begin before incarceration. Being in methadone maintenance treatment (as opposed to methadone detoxification treatment) was associated with less drug injection risk behavior (as measured by the four-point injection risk scale) and with a lower probability of incarceration during the study follow-up period. These findings are consistent with the entire history of research on methadone ⁽²⁴⁾, including previous research in Thailand ⁽²⁵⁾, and a recent random assignment study conducted in the United States ⁽²⁶⁾. Although methadone maintenance treatment also faces issues of community acceptance, these Bangkok data reinforce the importance of providing methadone maintenance from a public health perspective. In situations such as Bangkok, IDUs who have been incarcerated or are at high risk for incarceration would be a logical group for receiving methadone maintenance treatment.

Finally, these Bangkok data also illustrate the desirability of an effective vaccine for preventing HIV infection. A phase III trial is currently being conducted with IDUs in Bangkok. There is a firm commitment from both the Thai government and the sponsoring company (VaxGen Inc., Brisbane, CA, U.S.A.) to providing the vaccine to IDUs if it should be effective.

The data in this report concern HIV transmission related to injection risk behavior and incarceration among IDUs. These data are also relevant to the transmission of other blood-borne viruses, such as hepatitis B virus (HBV) and hepatitis C virus (HCV) ⁽²⁷⁾. Because HBV and HCV are more efficiently transmitted than HIV, transmission of these viruses may be of greater concern than HIV in some incarceration settings.

As noted above, previous studies have shown low HIV incidence rates in the United States ^(4-6) and Australia ⁽¹⁶⁾, but with the potential for outbreaks of HIV infection in these settings ⁽⁸⁾. Additional research is needed to determine factors associated with high versus low HIV incidence among incarcerated drug users. Drug injection appears to be common among incarcerated IDUs. HIV transmission during incarceration may be most likely when there is a high background HIV prevalence in the local IDU population and when no drug abuse treatment is provided for incarcerated IDUs.

We do want to emphasize that we know of no reason to consider the situation regarding incarceration and increased risk of HIV infection in Bangkok to be unique. Given the continuing global diffusion of HIV among IDUs ⁽²⁸⁾ and the common policy of incarcerating IDUs, the conditions for HIV transmission associated with incarceration are probably increasing in many countries throughout of the world.

Acknowledgments:

This study was financially supported by the Global Program on AIDS (GPA) of the World Health Organization, the Joint United Nations Programme on HIV/AIDS (UNAIDS), and the U.S. Centers for Disease Control and Prevention as part of a bilateral HIV/AIDS research collaboration with the Thai Ministry of Public Health. Dr. Des Jarlais' work on the study was supported by grants DA RO1 03574 and P30 DA11041 from the U.S. National Institute on Drug Abuse (NIDA).