Risk factors associated with injection initiation among drug users in Northern Thailand

A range of studies have examined the factors associated with transition from non-injection into injection drug use; however most of them are conducted in the developed world [1–5]. This transition is a critical point in a drug user's career for many reasons, particularly because it is characterized by a heightened risk of acquiring HIV and other blood-borne pathogens [6–8]. The process of initiating injection is complicated and informed by a myriad of individual, social-culture, and environmental factors. Demographic characteristics significantly associated with transition into injection in developed countries have included: being of younger age [5, 9]; being female [10]; ethnic minority [3, 11]; housing instability [5]; and being a high school dropout [3]. Having a current steady sexual relationship with a partner who is an IDU has long been observed as a facilitating factor for transition into drug injection [12]. Sexual abuse alone has put drug users at a higher likelihood of injection initiation [5, 13]. Many studies have recorded the independent effect of drug use characteristics on onset of injection, such as types of recently used drugs [5, 12]; intensiveness and/or length of drug use [9–12, 14, 15]; and current treatment status [1]. At the social-culture level, low socio-economic status has consistently been found to be one of the most important factors in initiating injection in the developed world [2, 4].

An exploration of the circumstances around the critical time of initiating injection practices in developing countries as related to medical outcomes has not been well addressed. Limited evidence from China [16, 17] indicated that risk factors of injection initiation in developing countries could be as similar as those in developed countries like younger age, ethnic minority and short duration of heroin use. However, Lai et al [17] reported that being male related to higher risk of injection initiation, which has not been shown in other studies [10, 18]. The situation in Thailand is of particular interest because injecting drug use continues to be a leading risk factor for HIV in the country that was previously characterized as having a heterosexual epidemic [19]. After implementing the 100% condom use program in the 1990s, by 2002 the national HIV prevalence was about 2% and prevalence among 21 year-old military conscripts was under 1%, having dropped from 4% in the mid-1990s [20]. Unfortunately, similar prevention efforts have not been waged to reduce HIV infection among injection drug users (IDUs), partly due to lack of knowledge about injection initiation.

Drug use, especially use of methamphetamines, has increased rapidly in recently years in Thailand, particularly among young people [21–23]. Sattah and colleagues [23] reported among adolescent students in northern Thailand, 41.3% males and 19.0% females ever used methamphetamines. The risk of HIV infection is high among drug users in Thailand. The reported HIV incidence was about 3%~19% [22, 24, 25]. As shown in other places, IDUs had a much higher risk of HIV infection than non-IDUs in northern Thailand [24–26]. Srirak et al. [21] reported that among female drug users in northern Thailand the HIV prevalence among IDUs was six more times than that among non-IDUs (25% vs. 4.1%). While it is feasible to deliver harm reduction programs in Thailand [27], they are currently very limited [26, 28, 29]

The purpose of this study was to identify demographic and drug use factors related to onset of injection among drug users, who were admitted to one large drug treatment center, in order to provide information to develop harm reduction strategies with the hope of preventing and controlling drug injection as well as HIV infection.

The study's findings indicate that demographic factors, sexual behaviors and drug use characteristics account for some of the risks in transition into injection. Drug users who were older or single, obtained more education, lived in urban areas, ever smoked or had an experience of incarceration were more likely to initiate injection. Multiple sex partners or a history of sex abuse related to a higher likelihood of starting injection; using amphetamines instead of heroin as the starting substance of abuse and later drug initiation were associated with a lower risk of injection initiation.

Contrary to a study conducted among street youths (14–25 years of age) in Montreal, Canada [5], our data showed that older age was associated with a higher risk of initiation into injection. The obvious difference in study settings and age structure between the two studies may account for some of the difference in the age effects. Our study participants were from a resource-limited country and older than those in the Canada study. Furthermore, all our participants were drug users, whose age was apparently associated with both drug use history and mode of drug administration. Old drug users might have involved in the drug culture for an extended period of time. Thus, it is not surprising that older drug users had great opportunity to transition into injection. In this analysis the effect of drug use history was partly controlled by including age of drug initiation into the final model; however, the impact of initial and sequential drug use modes on transition into injection was not extensively explored.

As a contextual factor, neighborhood (residence) has been shown as an importance determinant of starting injection [30]. Since most studies have been conducted in developed countries, the effect of rural residence on initiation of injection is unknown. Our data demonstrated that people living in rural areas had half the risk to initiate injection as those in urban areas after controlling for other confounders such as education. Ethnicity and employment were initially included into the model but were dropped due to statistical non-significance. In current analysis, postal code was employed to define the types of residence, and current residence was used as a surrogate for the neighborhood that influenced the occurrence of injection, which may have been inaccurate. If more detailed information is available, it would be helpful to explore how contextual factors influence and shape drug users' behavior and injection practices in Thailand. As Galea et al. [31] point out, to fully understand drug use behaviors, researchers should address not only individual risk behaviors but also contextual factors such as social norms and neighborhood socioeconomic status by means of statistical methods like multilevel analysis.

Not reported from some studies [5, 32], we found multiple sex partners (6 or more) to be a risk factor of transition into injection. However, in line with other studies [5, 13], the experience of sexual abuse increased the likelihood of injection initiation. Though not explored in this study, others [33, 34] have demonstrated that negative life events and risky behaviors following sex abuse could be possible mechanisms in the casual chain. In our study we cannot determine the temporal relationships between sex behaviors and injection initiation, which makes it hard to establish the causal pathway.

Comparing to opium, taking heroin as the initiation drug was associated with a larger likelihood of transition into injection, while starting with amphetamines lowered the risk. Mode of drug administration and multiple drug use following initiation drug might account for their specific effect. In the past three months, 72.9% of heroin users injected the drug, while only 7.5% of amphetamine users were injectors (data not shown). Furthermore, those starting with amphetamines tended to stick with it; only 12.3% of them ever used heroin. Though behaviors in the past three months were not necessarily constant over time, they at least indicated some drug use pattern in our sample. Among individuals starting from marijuana or others (inhalants and tranquilizers/sedatives), 59.7% and 54.2% ever tried opium sometime late respectively, which might partly explain the non-significant difference in risk of injection between marijuana or others with opium. Multiple drug use and the intensity of drug use were not examined in the current analysis, though many other studies [10, 11, 14] have indicted that they are related to a higher risk of initiating injection.

In our study, younger age of drug initiation was significantly associated with a higher risk of injection initiation even after controlling for the effect of age and other demographics, which has not been reported in other studies [5, 32]. The interaction between age and age of drug initiation was examined and the effect was not statistically significant. As indicated by our results, those starting drug use at less than 16 years had four times the risk of transition into injection than those starting at the age 24 years or older. One implication for this finding is to focus health education programs on youths.

The HIV prevalence at study enrollment among drug users in our sample was 15.6%, which was comparable to the results from other studies in similar settings [22, 24, 25]. Compared to the estimated HIV prevalence of 1~2% among the adult population in Thailand [35], the risk of HIV infection among drug users was clearly quite high. Further analyses (data not shown) indicated that the risk of HIV infection varied by age, age at drug initiation, injection status and types of recently used drugs. Drug users 20 years old or over had a higher prevalence of HIV infection than those younger than 20 years old (18.4% vs. 3.7%, P < 0.0001), while those starting drugs at younger age (under 16 years old) had a higher risk of HIV infection than those starting drug use at older ages (age of 16 years old or over) (17.9% vs. 7.3%, P < 0.0001). Considering the correlations between age, length of drug use (age of drug initiation), and the accumulated risk of HIV infection during the entire course of drug use, the observed differences in the risk of HIV infection between different age groups and length of drug use were reasonable but unacceptable. HIV prevention programs and health education projects need to be enhanced and should target the population at risk, especially youth. The HIV prevalence among IDUs was 35.2% in our sample, which was 12.6 times higher than that of non-IDUs (2.8%) (P < 0.0001). In terms of drug use in the past 3 months, HIV prevalence was highest among users of other drugs (inhalants and tranquilizers/sedatives, 35.1%), followed by heroin users (29.9%), amphetamine users (16.0%), marijuana users (15.0%) and opium users (10.4%). A study from the same region [21] reported the HIV prevalence was 25% among heroin injectors and 4.1% among opium or methamphetamine smokers. Gender could be one reason to the observed difference in HIV risk since the sample for the above study only included females. However, in our sample 44.2% of participants were multi-drug users. In the past 3 months, 61.7 % of inhalants and tranquilizers/sedatives users, 46.4% of marijuana users, 40.8% of amphetamine users, and 21.9% opium users also reported using heroin, which matched well with the observed HIV risk for individual drugs. As shown in elsewhere [36], multiple drug use was associated with a higher risk of HIV infection, especially when heroin injection was involved. This finding indicates the importance of addressing the issue of multiple drug use in harm reduction programs, and not just focusing on injection of heroin or methamphetamines.

The interpretation of these results is subject to certain limitations. One concern is the representativeness of the sample. Recruited from one large drug treatment center, drug users in our study could be a highly selected group with relatively high risks. Data have shown that a history of drug treatment may reflect past severity of drug dependence, itself a predictor of transition into injection [12, 14]. Secondly, all behavioral data were from self-reports that might be inaccurate. Nevertheless, all interviewers were asked to evaluate the quality of the interview at the end of each interview and most reported good cooperation and judged the data to be moderately to very reliable. A study by McElrath and colleagues [37] also showed that retrospective self-reports from injection drug users had good reliability and consistency. The extent of under-reporting of socially unacceptable activities or over-reporting was unknown and inaccessible. Thirdly, some informative variables such as types, modes, and intensiveness of drug use prior to injection initiation were not assessed. Exploring these risk factors could shed additional understanding of the process of initiating injection, and might also benefit the design of appropriate harm reduction program to target the individuals at risk. Finally, the cross-sectional design of this study limited our ability to determine causality; some potential risk factors for transition into injection might result from the fact that the participant initiated injection and then identified those experiences a risk context, such as being in jail. Furthermore, the use of a cross-sectional design could induce recall bias and survival bias.

Despite these limitations, we have identified several correlates of injection initiation among drug users in this Southeast Asian context. Understanding the factors related to initiation into injection may be an important tool in guiding how we modify current harm reduction strategies or design new approaches to more specifically target high-risk populations and high-risk behaviors to delay or prevent the initiation of injection.

Injection initiation was related to several demographic factors, sexual behaviors and drug use characteristics in northern Thailand. The information from this study could be employed to improve current efforts in harm reduction in order to better serve the people at risk.