Evolução do consumo de crack em coorte com histórico de tratamento Evolution of drug use in a cohort of treated crack cocaine users

MÉTODOS: Uma coorte de, originalmente, 131 dependentes de crack admitidos em uma enfermaria de desintoxicação em São Paulo, SP, entre 1992 e 1994, foi re-entrevistada em três ocasiões: 1995-1996, 1998-1999 e 20052006. As variáveis averiguadas foram: dados demográfi cos, comportamento sexual de risco, padrões de consumo de crack e outras substâncias, prisões, desaparecimentos e óbitos. Na análise estatística empregou-se o teste de quiquadrado, a regressão logística multinomial e regressão de Cox.


INTRODUCTION
Longitudinal studies of trends in substance abuse are an essential tool for planning public health interventions.Notwithstanding, long-term follow-up studies of the consequences of crack cocaine use are still scarce. 3,7topic of interest in recent years has been the longevity of crack cocaine use among treated and nontreated users, especially given its intensity, recurrence, and, in many cases, persistence. 13These studies indicate that the use of crack cocaine is no longer an essentially short-term practice.Such a pattern raises the need for methodological and conceptual tools that allow for a better understanding of the complexity and evolution of phenomena related to crack cocaine use.
The documentation of different drug use trajectories 24 allows us to characterize heterogeneous groups, as well as to defi ne groups that are more vulnerable to health-related disorders and factors that are associated with stable abstinence. 14e aim of the present study was to investigate the evolution of crack cocaine use among treated users.

METHODS
We carried out a prospective study within a larger research project following a cohort of 131 crack cocaine users admitted to a detoxifi cation unit in a general hospital in the North region of the city of Sao Paulo, Southeastern Brazil.Subjects were followed up two (1995-96), fi ve (1998-99), and twelve (2005-06) years after discharge. 6,15,21is convenience sample was obtained from the admissions registry of the detoxifi cation unit, and included patients consecutively admitted between May 1992 and December 1994.The criterion for inclusion was addiction to crack cocaine, diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV).Diagnosis was established during a clinical interview with the team of psychiatrists responsible for admissions at the time.
We chose a study design that allowed us to follow a cohort of subjects for a long time so as to identify potential outcomes, such as death, disappearance, imprisonment (information obtained from relatives), and intake of or abstinence from crack cocaine (usage patterns in the last 12 years).
The hospital we investigated pioneered the specialized treatment of crack cocaine addiction within the public health care network in Sao Paulo in the early 1990s, at the onset of the crack cocaine epidemic in the city.Admissions were scrutinized by an interdisciplinary team (comprising psychiatrists, a psychologist, nurses, a social worker, and an occupational therapist), and lasted for two to three weeks.Data were obtained from patient charts fi lled out at the time of admission and from in-person interviews using structured questionnaires. 22The information obtained from patient charts included demographics, HIV/AIDS (tested upon admission), history of usage and prior admissions, intravenous drug use, and family history of chemical addiction.During the interviews, we investigated the following indicators: employment situation, schooling, risky sexual behavior, imprisonment, disappearance, death, occurrence and usage patterns for multiple substances (legal and illegal), lifetime use of crack cocaine and other substances, change of route of cocaine administration, peak period of crack cocaine use, and crack cocaine use trajectory (change in patterns throughout the 12-year period).
Intake data were self-reported.We considered use as sporadic when occurring up to two times per week; frequent when occurring three to four times a week; and heavy when occurring fi ve to seven times a week.
The interviewer for the third follow-up was the same person that conducted the prior evaluations and was therefore known to the subjects.To improve the quality of these encounters, questionnaires were administered during home visits.
We confi rmed the data on mortality and imprisonment by consulting offi cial government records (State Data Analysis System Foundation [Fundação Sistema Estadual de Análise de Dados] and Program for the Improvement of Mortality Information [Programa de Aprimoramento das Informações de Mortalidade]).
For statistical analysis, the characteristics of the sample were described using simple frequencies and percentages.For quantitative variables, we calculated mean and standard deviation.To compare categorical variables, we used the chi-squared and Fisher's exact tests, with a signifi cance threshold of 5%.For contingency tables larger than 2x2 and in which there was evidence of statistical association (p<0.05),we used residual analysis to identify categories that were different within the groups (standardized residuals > 1.96).
Mortality evaluation was carried out in two steps.In univariate analysis, stratifi ed survival curves were tested and compared using the log-rank test, taking into account variables collected during admission for treatment.Three of these variables showed signifi cant associations (p<0.10) and were tested using Cox regression including variables that could independently lead to increased risk of death.For the signifi cant variable in the fi nal model, we calculated relative risk and 95% confi dence interval.
Using multinomial logistic regression (with stepwise forward inclusion), none of the variables in the admission chart remained as a predictive factor of stable abstinence or persistent intake.
Analyses were carried out using SPSS for Windows, version 13.0.
All procedures were approved by the Research Ethics Committee of the Universidade Federal de São Paulo (Process no.1420/05).

RESULTS
Twelve years after discharge, we were able to trace 107 (81.6%) of the 131 subjects in the initial sample.Of these, which 27 (20.6%)had died, two (1.5%) were missing, and 13 (10%) were imprisoned.The other 65 subjects (49.6%) were distributed among abstinent subjects and regular users of crack cocaine (for 12 months or more), with 43 (32.8%) and 22 (16.8%)subjects in each group, respectively.The 24 subjects (18.3%) lost to follow-up could not be located by the researchers due to successive changes in address.Of these, two were evaluated only in the fi rst follow-up (1995-96), seven only in the second (1998-99), 12 were interviewed in both follow-ups, and three could never be traced.There were no statistically signifi cant differences between traced subjects and losses in terms of admission variables such as demographics, history of use of crack cocaine and other substances, prior treatment, legal problems, and HIV seroprevalence.
Of the variables recorded during admission, three were related to decreased probability of survival: years of schooling (p=0.044);prior use of intravenous cocaine (p=0.002); and positive HIV status (p=0.000).According to the Cox regression model, prior use of intravenous cocaine increased by 2.5 fold the risk of death in 12 years (p-0.031)(95%CI: 1.08; 5.79) (Figure).
In the 12 year period, abstinent subjects and crack cocaine users were divided in terms of marital status, with 43% being single and 37%, married.Mean age was 35 years, with 49% of subjects aged between 31 and 38 years.Only 15% of subjects were unemployed.Almost one-half of subjects (49.2%) did not study beyond elementary schooling, and 77% did not attend school in the year preceding the interview.
Regarding use of legal and illegal drugs in the year preceding the interview, the wide majority of subjects (n=45; 74%) reported sporadic alcohol consumption.Of these, 28 were abstinent from crack cocaine, and 17 were users.The same percentages were found for heavy tobacco use (n=45; 74%), also divided among abstinent subjects (n=28) and crack cocaine users (n=17).Most subjects did not use marijuana in the preceding year (n=42; 69%), divided into 31 abstinent subjects and 11 crack cocaine users.Among marijuana users, 39% showed sporadic use and 33%, intensive use.There was no signifi cant difference between crack cocaine users and abstinent subjects in terms of use of alcohol, tobacco, and marijuana in the year preceding the interview (Table 2).
More than half the subjects (n=40; 64.5%) reported no use of snorted cocaine in the year preceding the interview.Among subjects who used snorted cocaine during this period, 68% reported frequent to heavy use.Former crack cocaine users (n=34) differed signifi cantly from current crack cocaine users (n=6) with respect to abstinence from snorted cocaine use (p=0.000).
Use of the remaining substances -including amphetamines, hallucinogens, opioids, and intravenous cocaine -in the last year was not reported by any of the subjects.
The mean age of onset of snorted cocaine use was 18 years.Mean age of onset of crack cocaine use was 22 years, with no signifi cant differences between users and abstinent subjects for both snorted and crack cocaine.There was a four-year interval between the onset of snorted cocaine and crack cocaine use.
Mean lifetime use of snorted cocaine and crack were 11 years, six months (140 months; SD=78.4) and eight years, ten months (106 months; SD=65.47),respectively.There was, therefore, a long-term simultaneous use of snorted and crack cocaine.Lifetime use of snorted cocaine ranged from frequent to heavy for 53% of subjects and use of crack cocaine was heavy for 74% of subjects.
Regarding the migration of routes of administration, snorted cocaine remained as the primary route employed by 88.5% (n=54) of subjects.The smoked route (crack) was the initial route of 9.8% (n=6) of subjects, and only 1.6% (n=1) began to use cocaine by the intravenous route.
Smoking (crack cocaine) remained as the second route of cocaine administration in 82% of subjects; only 12% used intravenous administration as their second route, and 7%, snorted cocaine.There was no transition to a third route in almost all subjects (81.5%).The alleged reason for moving from the fi rst (snorted) to the second (crack) route was preference for the intensity of the effect (51%).
Approximately 12 months after the fi rst use, users reached "binge" crack cocaine intake, with an individual average of 50 rocks (SD=5.3) in four consecutive days (SD=1.86).
Based on the information on the oscillation of use and/ or on the periods of drug use and abstinence among subjects (diversity of usage patterns), we were able to discern three groups with distinct post-discharge usage patterns; (group 1; n=31) stable abstinent subjects: subjects reporting abstinence for at least fi ve consecutive years; (group 2; n=20) intermediate: subjects alternating between periods of use and abstinence; and (group 3; n=14) prolonged users: subjects who sustained crack cocaine use throughout the 12-year period (Table 3).
Factors associated with sustained crack cocaine use in the 12-year follow-up period (group 3) included positive HIV test at admission (p=0.046);use of snorted cocaine in the previous year (p=0.001); and lifetime use of snorted cocaine > 132 months (p=0.000).
The only variable to differ signifi cantly between stable abstinent subjects (group 1) (67.7% abstinent for fi ve to ten consecutive years) and the intermediate (group 2) and prolonged use (group 3) groups was safe sexual behavior with condom use (p=0.001)(Table 4).
Age of onset of alcohol, tobacco, and marijuana use was distributed as follows: alcohol, 15 years (range = 5-34 years); tobacco, 15.5 years (range 10-32 years); and marijuana, 16 years (range 11-36 years).There was no statistically signifi cant difference between abstinent subjects and crack cocaine users in terms of age of onset of these other substances (p=0.773,p=0.930, p=0.705, respectively).
Lifetime duration of alcohol, tobacco, and marijuana use were 19 (SD=6.6),18 (SD=6.5),and 12 (SD=7) years, respectively, with no statistically signifi cant differences between crack cocaine users and abstinent subjects.Lifetime use of marijuana was sporadic for 47% of users, and use of alcohol was frequent for 75% of users.Lifetime use of tobacco was heavy for 95% of users.

DISCUSSION
The present study describes the evolution of crack cocaine use in a cohort of treated users.This evolution is characterized by a high rate of mortality for violent causes, migration of administration routes, and lifetime use of legal (alcohol, tobacco) and illegal (marijuana, cocaine) substances.Periods of heavier consumption (50 rocks in four days) confi rm the presence of "binge" patterns in this cohort.We were also able to defi ne three groups with distinct consumption trajectories and their associated factors.
Use of other substances in addition to crack cocaine is frequent among users. 3,10In the present study, reported use of other substances, both ever in life and in the previous year, confi rmed this tendency, even though we were unable to clearly establish how these associations operate or their meaning.
It is estimated that subjects who seek treatment for cocaine use (in general) show high frequency of alcohol abuse. 7,1120The combination of snorted cocaine and alcohol or crack cocaine and alcohol can be very  different: while in the former case, alcohol generally tends to reinforce the positive effects of snorted cocaine, in the latter case, alcohol use takes place after crack cocaine use, as an attempt to "wet" the dryness of the mouth, "rebound" the intensity of the effects of crack cocaine, and/or counteract undesirable effects.Moreover, in late stages, alcohol tends to reduce the dose of crack cocaine used. 10r analysis of lifetime use of snorted or crack cocaine showed that transition from the snorted form of the drug, more common at the onset of cocaine use, 11 to the smoked form did not necessarily lead to forfeiting of the initial route.When this was the case, we observed concomitant use of snorted and crack cocaine for a period of many years, especially among patients in group 3 (prolonged use).According to Guindalini et al 11 (2006), the combination of the two routes would characterize a distinct class ("dual" users), associated with greater occurrence of legal problems and greater risk to health.On the other hand, there is doubt as to the role played by snorted cocaine in the long-term maintenance of crack cocaine use.Such a correlation was detected, but not elucidated, in the present study.We do not know whether dependency was aggravated by the combination of the two substances, and therefore whether there was greater diffi culty in interrupting the use and/or decreased consumption of crack secondary to snorted cocaine use (either in alternation or as a substitute), allowing the user to "stretch" crack cocaine use in the long term. 7e use of crack cocaine can follow a number of distinct trajectories, and its duration can vary from brief to prolonged periods. 12In the present study, we detected prolonged cycles of use (group 2) and uninterrupted use of crack cocaine for many years (group 3 and lifetime use).
In the international literature, Falck et al 8 (2007) evaluated a cohort of 430 crack cocaine users, mostly males, over 18 years of age, with fi xed residence and no pending criminal charges.The authors reported that, after an eight-year follow-up, 63% of subjects were still using the drug, indicating the viability of extended crack cocaine use for decades.
In Brazil, long term crack cocaine use tends to be stimulated by high availability (or profi tability) and ease of access.Moreover, the formulation of the drug has undergone changes (in color, consistency, effect, and size), which resulted in "impure" preparations (lower addictive potential, greater addition of diluents). 19nother aspect that plays a role in the longevity of crack norm rather than the exception, 7 there are studies (still incipient) that report moderate patterns of crack cocaine consumption 9 and its conciliation with daily activities.a Oliveira & Nappo 18 (2008) detected a trend among individuals that transitioned from compulsive patterns to more controlled use.According to these authors, the subjects themselves devised strategies for selfregulation such as, for instance, distancing themselves from the environment in which they used the drug, structuring daily and leisure activities, moderating the use of other "trigger" substances, or administering other drugs as substitutes for crack cocaine.Self-regulatory practices appear to indicate a dynamic process in which the user tests and accumulates experiences with regard to forms of use and their associated damage, relying on procedures aimed at reducing risks and actively building a culture of drug use.
A factor that was strongly correlated with long-term crack cocaine use was positive HIV testing upon admission.Maintenance of drug use by seropositive individuals raises the need for damage control strategies (consistent condom use, distribution of pipes, holders, and lip protectors) aimed at preventing HIV transmission and reinfection as well as other STDs (such as syphilis and gonorrhea).Crack cocaine abuse has been implicated not only as a risk factor for HIV infection, but also as a potential "catalyst" of disease progression among seropositive users. 1,4Combined interventions (effect maximization) are recommended as a treatment for both problems.Regarding HIV, it is recommended that users be referred to testing and follow-up facilities and undergo constant monitoring of their adherence to antiretroviral treatment and of their motivation to care for their health and body. 2 With the exception of a positive correlation with safe sexual behavior (condom use), we were unable to identify other factors associated with stability of abstinence trajectories.Notwithstanding, the composition of a group of long-abstinent subjects (group 1) indicated the rupture of skepticism with respect to the impossibility of overcoming crack cocaine addiction and sustaining abstinence for an extended period. 16In an attempt to understand the recovery process, studies in the literature seek to uncover indicators closely related to stable abstinence. 13,17These include the study by Siegal et al 24 (2002), which showed a strong correlation between stable abstinence and prolonged detoxifi cation treatment.
Regarding the cohort in the present study, effective forfeiting of crack cocaine use was seen to occur anywhere from early in treatment to much later in the process.Stable abstinence among treated crack users may establish itself both in the short term and after several years of treatment.A study conducted in the United States with a cohort of 1,271 drug users (64% of which used cocaine) estimated that the mean lag between onset of treatment and the last episode of drug use was nine years. 5Scott et al 23 (2005) reported a range of three to four treatment episodes spread across a long time period before stable abstinence patterns were achieved.The expectation of early interruption of crack use (high expectation), in addition to promoting the divestment of care when unsuccessful, also compromises the trajectory of change, which might take years to occur.
Future investigations should attempt to elucidate the so-called "turning points," which are signifi cant events in the life of users that tend to favor the interruption of crack cocaine intake.It will also be important to detect the moments in the life of individuals in which other mechanisms (related to health and social/cultural life) external to formal treatment play a supporting role in altering their exclusive relationship with the drug.
Limitations of the present study include the fact that our sample was obtained from a single treatment institution, the small proportion of women in our sample, and the absence of physical measurement of drug use.Other phenomena that are relevant to this fi eld of study were not explored, such as pipe or holder sharing, combined use of crack and tobacco or crack and marijuana, and occurrence of tuberculosis and hepatitis.
The design of the current study did not allow us to conduct an in-depth analysis of issues related to the motivation of subjects to seek formal treatment.Neither could we investigate the consequences of treatment to the users' lives, including its potential contribution to stable abstinence.Furthermore, the factors involved in prolonged crack cocaine use could not be explored, likely because the measurements we made were not appropriate for detecting positive correlations that would allow us to explore these factors.If this is the case, a qualitative approach would be useful to help elucidate these aspects in depth, providing consistent indicators and goals to aid in the planning of drug-user care initiatives.
a Absence of data is due to missing entries in admission fi les.

Table 2 .
Use of legal and illegal drugs among abstinent subjects and crack cocaine users (last 30 days and last 12 months).São Paulo, SoutheasternBrazil, 1992-2006.
a Absence of data for some variables is due to refusal to provide information or recall bias.* p-value <0.05.Figure.Survival curves stratifi ed according to signifi cant variables in the fi nal Cox model.São Paulo, Southeastern Brazil, 1992-2006.

Table 4 .
Trajectory of crack cocaine use and demographic and drug use profi le.São Paulo, Southeastern Brazil, 1992-2006.
cocaine use is the development of individual strategies for maintaining extended consumption patterns, albeit at lower intensity levels.Although heavy use is still the a Silva SL.Mulheres da luz: uma etnografi a dos usos e preservação no uso do crack [Tese de Doutorado].São Paulo: Faculdade de Saúde Pública da USP; 2000.