Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Mar 15.
Published in final edited form as: Addiction. 2025 Jan 9;120(4):756–769. doi: 10.1111/add.16731

Substance use and psychiatric outcomes following substance use disorder treatment: An 18-month prospective cohort study in Chile

Ignacio Bórquez 1,2,3, Noa Krawczyk 1, Ellicott C Matthay 1, Rafael Charris 1, Sofía Dupré 3, Mariel Mateo 2,4, Pablo Carvacho 3, Magdalena Cerdá 1, Álvaro Castillo-Carniglia 2,5, Eduardo Valenzuela 3
PMCID: PMC12988487  NIHMSID: NIHMS2138219  PMID: 39789832

Abstract

Background and aims:

Evidence from high-income countries has linked duration and compliance with treatment for substance use disorders (SUDs) with reductions in substance use and improvements in mental health. Generalizing these findings to other regions like South America, where opioid and injection drug use is uncommon, is not straightforward. We examined if length of time in treatment and compliance with treatment reduced subsequent substance use and presence of psychiatric comorbidities.

Design:

Prospective cohort analysis (3 assessments, 18 months) using inverse probability weighting to account for confounding and loss to follow-up.

Settings:

Outpatient/inpatient programs in Región Metropolitana, Chile.

Participants:

Individuals initiating publicly funded treatment (n = 399).

Measurements:

Exposures included length of time in (0–3, 4–7, 8 + months, currently in) and compliance with treatment (not completed, completed, currently in) measured in the intermediate assessment (12 months). Primary outcomes were past-month use of primary substance (most problematic) and current psychiatric comorbidities (major depressive episode, panic, anxiety or post-traumatic stress disorders) measured 6 months later (18 months). Secondary outcomes included past month use of alcohol, cannabis, cocaine powder and cocaine paste.

Findings:

18.3% [95% confidence interval (CI) = 14.7%–22.6%] of individuals participated for 3 or fewer months in treatment and 50.1% (95% CI = 45.2%–55.1%) did not complete their treatment plan at 12 months. Participating for 8 + months in treatment was associated with lower risk of past month use of primary substance at 18 months [vs. 0–3 months, risk ratio (RR) = 0.62, 95% CI = 0.38–1.00] and completion of treatment (vs. not completed, RR = 0.49, 95% CI = 0.30–0.80). Neither participating 8 + months (vs. 0–3 months, RR = 0.83, 95% CI = 0.57–1.22) nor treatment completion (vs. not completed, RR = 1.02, 95% CI = 0.72–1.46) were associated with lower risk of psychiatric comorbidity at 18 months.

Conclusions:

Longer time in treatment and compliance with treatment for substance use disorders in Chile appears to be associated with lower risk of substance use but not current comorbid psychiatric conditions 18 months after treatment initiation.

Keywords: alcohol, cocaine, inverse probability weighting, South America, substance use disorder, treatment outcomes

INTRODUCTION

Longitudinal observational studies conducted in the global north have linked length of time in treatment and compliance with treatment with reductions in substance use and improvements in mental health, among other positive outcomes [1–10]. There is extended documentation on the high prevalence of co-morbid psychiatric disorders among people with substance use disorders (SUDs), especially mood, anxiety and personality disorders [11]. SUD treatment systems usually lack integration with other mental health services and the specialization needed to address them [12], contributing to worse outcomes [13]. Particularly relevant has been minimal retention (>90 days) as a milestone associated with better long-term treatment outcomes [4, 9, 10]. Other studies suggest an inverted ‘U’ pattern between time in treatment and improvements, as individuals who are exposed to longer periods (e.g. 18 months) in intensive treatment modalities (outpatient or inpatient) usually have more risk factors, and experience more difficulties after discharge [14]. Compliance with treatment (e.g. finishing a treatment plan or program) has also been consistently associated with better outcomes. However, critiques around the idea of ‘completion’ of a treatment episode have emerged, giving the chronicity of SUDs for some individuals, and advocating for health systems that also address current use and maintenance phases [15, 16].

Transporting these findings into other settings is challenging, because most studies involve people who use opioids and/or inject drugs. South America has a unique supply and demand of substances, with higher rates of cocaine, alcohol and cannabis use disorder, and lower rates of opioid and amphetamine use compared to other regions [17–19]. Smokable cocaine, such as cocaine paste and crack cocaine, disproportionally harms vulnerable populations [20–22] because of their low cost given the high rates of cocaine production in the region [23]. Polysubstance use, including alcohol and cocaine, is relatively common in South America because of their synergic effects [24]. Differences in the substance of use also limit therapeutic resources. For instance, medication for opioid use disorder (e.g. methadone) is highly effective [25], whereas regretfully there is no standard medication for cocaine use disorder [26]. Stimulant use has also been consistently associated with worse retention rates in psychosocial interventions [27], which might lead to worse outcomes and unique challenges for the already unfunded healthcare systems in the region.

In the last three decades, Latin America has increasingly implemented evidence-based interventions to address SUDs, including reforms to integrate treatment into primary care [28, 29] along with national standards and epidemiological surveillance [30]. Chile has one of the most robust treatment systems in the region and has experienced a movement toward promoting evidence-based interventions designed and evaluated in the global north (the United Kingdom mainly) and professionalization of the treatment staff (i.e. requiring psychologists and social workers, among others) [31]. Publicly funded treatment is provided both in public and private institutions using therapeutic approaches (e.g. abstinence- or/and harm reduction-oriented, cognitive behavioral therapy or religion-focused, among others) that vary greatly depending on the treatment program [31]. There are still major difficulties in the production, recollection and analysis of information to evaluate treatments for SUDs [32]. Longitudinal prospective studies assessing treatment outcomes using rich survey data in Latin America are scarce [30]. In the case of Chile, for instance, all peer-reviewed studies to date have used administrative data focusing on readmissions [33, 34] and criminal legal system contact [35]. To our knowledge, this is the first longitudinal cohort study examining mid-term SUD treatment outcomes in South America under routine conditions of care.

This study aims to assess if length of time in treatment and compliance with treatment reduce substance use and mental health comorbidities. We do this by analyzing a prospective cohort of individuals initiating publicly funded treatment for SUDs in Región Metropolitana, Chile, who were followed for 18 months (2018–2020).

METHODS

Study design, settings and participants

We conducted a prospective cohort study among adults initiating publicly funded SUD treatment in Región Metropolitana, Chile. Approximately 80% of the population in Chile had public insurance in 2018. At baseline, 82 of 94 programs (87.2%) in the National Service of Drug and Alcohol Use (SENDA) network in the region decided to participate. We conducted three assessments: the baseline interview was assessed from July to October 2018 (n = 612, see Table S1 including missing values in covariates), with subsequent follow-ups at 12 (July–October 2019, n = 465) and 18 months (January–April 2020, n = 399). We prioritized enrolling people who were less than 30 days in the current treatment episode (87.5%) and those declaring 3 or fewer months of treatment in the past 12 months (84.8%) before initiating treatment. The analysis for this study was not pre-registered, therefore, the results should be considered exploratory. The STROBE statement is presented in Supplementary Material 6.

Treatment options included outpatient high-intensity (5 days a week, 8 months recommended) and inpatient treatment plans (12 months recommended), which were assigned based on clinical needs (e.g. individuals walked in or were referred from other services) [36]. SENDA sets treatment standards and oversees implementation, but programs have freedom in their therapeutic approach. Programs funded by SENDA must provide: biopsychosocial needs assessment, psychological therapy, group meetings, psychiatric and medical consultations, medication and laboratory tests and home visits (e.g. a re-engagement strategy). Continuum care after discharge is encouraged but not financed, and therefore, scarcely implemented.

Data collection

Licensed psychologists conducted the fieldwork. Participants were identified by the program staff and provided consent in a private setting. A standardized questionnaire was completed face-to-face and on paper. The baseline survey asked for the time before initiating the index treatment episode. The second and third assessments were applied using computer-assisted methods (SurveyToGo software) in a setting chosen with each participant. Participants were compensated with vouchers (5000, 10 000 and 20 000 Chilean pesos increasing by assessment). The study was approved by the Ethics Committee of the Pontificia Universidad Católica de Chile (protocol no. 171 211 011) and the Ethics Committee of the South Metropolitan Health Service (protocol no. 37/2018).

Variables

Outcomes

The two main outcomes were past month use of the primary substance—declared as most problematic at baseline—(alcohol, cannabis, cocaine powder or cocaine paste) and any current psychiatric comorbidities (including major depressive episodes, panic, anxiety and post-traumatic stress disorders [PTSD]), which were assessed using the Spanish version of the Mini-International Neuropsychiatric Interview (MINI) version 6 [37]. The MINI encompasses diagnostics from the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). As secondary outcomes, we used past month use of alcohol, cannabis, cocaine powder and cocaine paste (one variable for each), irrespective of primary substance. All outcomes were measured in the last assessment (18 months).

Exposures

We defined two primary exposures measured via self-report to describe participants’ treatment status at 12 months after baseline. Length of time in the index treatment was self-reported in months and categorized into: (i) 0 to 3; (ii) 4 to 7; (iii) 8 or more months; and (iv) currently in the index or posterior treatment episode. We decided on the cut-offs based on the known importance of minimal retention (>90 days) in SUD treatment [4, 10] and because 7 months was the median time in treatment. The last group included both people currently in the index treatment and those who were enrolled in a posterior treatment episode after finishing or discontinuing the index treatment (6.5%, data not shown) at 12 months of follow-up. Treatment compliance was measured through the self-reported discharge status from the index treatment: (i) not completed (e.g. voluntary, expelled because of administrative reasons, closure of treatment center, among others); (ii) completed (e.g. received therapeutic discharge from the treatment staff); and (iii) currently in the index treatment or posterior treatment.

Baseline and childhood confounders

The International Guidelines for the Evaluation of Treatment Services and Systems for Psychoactive Substance Use Disorders informed the selection of confounders [38]. These guidelines address client and clinic-level characteristics to be considered in SUD outcomes studies. For a full description of the variables see Table S2. All confounders, except for adverse childhood experiences (ACEs), were measured at baseline. To characterize the current treatment episode, we included a variable describing if participants were <30 days in the index treatment episode before the baseline interview, the modality, ratio of patients to providers and participation in a program that requires abstinence. Participants’ socio-demographics included: age, gender, educational level, past week employment and insecure housing in the past 12 months. We included past month prevalence of alcohol, cannabis, cocaine powder, cocaine paste and non-prescribed benzodiazepines, past month polysubstance use, primary substance and early initiation in substance use (≤15 years old). We assessed a readiness to change score by administrating the University of Rhode Island Change Assessment (URICA) scale [39]. The scale was validated in Spanish among people with SUD [40] and showed moderate internal consistency (Cronbach’s α = 0.72). Higher scores can be interpreted as being more motivated to change substance use patterns. Treatment exposure in the 12 months before baseline interview and first-lifetime SUD treatment were measured. We constructed a family quality index summing the perceived quality individuals attributed to their family relationships using a five-point Likert scale (range, 0–25; less score can be interpreted as having a more deficient social network) and current cohabitation with someone with an SUD. We also included variables for having a close friend with SUD and utilization of mutual support groups (e.g. alcoholics anonymous/narcotics anonymous) in the last 6 months. Any self-reported physical comorbidities in the last 4 years, including 13 conditions, were measured, as well as any current psychiatric comorbidities (following the same definition as in the outcome section). Lifetime antisocial personality disorder was assessed (using the MINI). We included disruptive events (e.g. divorce) and criminal involvement in for-profit offenses (e.g. theft) in the last 12 months. Finally, we measured the prevalence of 11 ACEs and created a continuous variable by summing them.

Statistical methods

We compared means and SDs for continuous variables and percentages and counts for categorical variables comparing length of time in treatment and compliance with treatment groups. To assess bivariate differences between exposure groups we used analysis of variance for continuous variables and χ2 tests for categorical variables.

Inverse probability weighting

We used a marginal structural model (MSM) approach [41] and used a target trial emulation to inform the analytic plan (Table S3) [42, 43]. The directed acyclic graph (DAG) that guided our analysis is presented in Figure S1. To address potential selection bias because of loss to follow-up and potential confounding bias because of non-random treatment assignment (in our case, the exposures of interest: length of time in treatment and compliance with treatment), we used stabilized inverse probability of censoring weighting (IPCW) and stabilized inverse probability of treatment weighting (IPTW), respectively [44–46]. Inverse probability weighting (IPW) procedures usually involve two stages [45]. First, the probability of not being censored (IPCW) at 18 months and the probability of receiving the treatment actually received (IPTW) at 12 months after baseline were estimated, known as propensity models (one for each). Second, weights for each individual are estimated as the inverse of their predicted probability. For stabilized weights, the numerator of the weights (which is 1 in the case of unstabilized weights) is replaced with the probability of not censorship/treatment of the propensity model without covariates [45]. The intuition behind IPW procedures is to use weights to generate a pseudo-population in which the distribution of observed covariates is balanced across different levels of treatment and censoring status [45].

We used the ipw [47] package in R to the IPCW and IPTW. For the IPCW we used a logistic regression with baseline confounders (except ACEs, not measured at baseline). For the IPTW, we used a multinomial logistic regression including all confounders. Models’ specifications are shown in Supplementary Material. To handle missing data in the propensity models (~1%–10% in covariates) (Table S4–S5) we used multiple imputations by chained equations (MICE) using the mice [48] package in R. In each imputed dataset (m = 30), an IPCW and IPTW were calculated and the average by individual was extracted to use [49, 50]. The final weights were obtained by multiplying IPCW*IPTW, which was later trimmed to the 2 to 98 percentiles to limit imprecision caused by extreme weights (Table S6). Standardized mean differences (SMD) in covariate values across censoring (Table S7) and treatment groups were assessed before and after weighting (Tables S8–S9). Distributions of the propensity scores and weights are presented in Figures S2–S11.

Changes in the outcomes over time and regression models

We first described changes in outcomes over time using paired t tests. We, then, performed log-binomial regression models for our primary and secondary outcomes and presented the unadjusted risk ratio (RR) and the marginal RR adjusted using the final weights. We report robust standard errors to account for the weighting procedure using the geeglm [51] package in R.

Sensitivity analysis

We conducted six sensitivity analyses. First, we excluded treatment-experienced individuals (declaring >30 days in index treatment or >3 months of treatment in the last 12 months at baseline). Second, we considered exposures as measured in the last assessment (18 months) to check for consistency. Third, we included only individuals whose propensity scores (i.e. probability of treatment) were in the area of common support [i.e. within the maximum (or minimum) of observed propensity scores across all treatment groups]. Fourth, we excluded participants who had experienced a posterior treatment episode after the index treatment. Fifth, we ran a traditional outcome regression estimating the conditional effect of our exposures using MICE. Last, because there might be unmeasured confounders, such as source of referral [38], we calculated E-values to evaluate how strong an unmeasured confounder would need to be to ‘explain away’ our observed effect estimates for the primary outcomes [52]. All analyses were conducted in R (version 4.1.1). The code is available at https://github.com/idborquez/sbstncpsych_outcome_sud.

RESULTS

Retention

A total of 606 individuals were eligible at 18 months (6 deaths). Our retention was 76.4% for the second (n = 465) and 65.8% for the third assessment (n = 399). Figure S12 depicts a flow diagram of retention over time. To see participants’ final disposition and survey response rates please see Table S10. Individuals lost to follow-up were more likely to be men, younger participants, people experiencing housing insecurity, self-reporting committing crimes, those with an early initiation in substance use and people who declared cocaine paste as their primary substance (Table S1). After applying the IPCW weights the SMD across censoring groups were below 0.1 for all covariates (Table S7). SMDs reflect the difference in the mean of each covariate between groups, but are standardized by their pooled SD. This translates into <10% of a SD difference between our censoring groups in our weighted pseudo-population.

Sample description

Among the analytic sample (n = 399), the mean age was 37.7 (SD = 10.6) years at baseline (2018), a third were women (32.3%) and 63.1% were employed the week before initiating index treatment (Table 1). A third received an inpatient treatment plan (33.6%), a fifth received care in programs that required abstinence to participate (20.7%) and 43% declared no prior lifetime treatment. At the 12-month exposure assessment, 18.4% self-reported participating for 3 or fewer months in treatment, 30.7% between 4 and 7 months and 23.7% for 8 or more months. A total of 22.6% completed their treatment plan, whereas half did not (50.1%). A total of 27.3% were currently in the index treatment or a subsequent treatment.

TABLE 1.

Participants characteristics according to length of time in treatment and treatment compliance.

Length of time in treatment, n (%) or mean (SD)
 Treatment compliance, n (%) or mean (SD)
Total (n = 399) 3 or less months (n = 73) 4–7 months (n = 122) 8 or more months (n = 94) Currently in treatment (n = 109) P-value Not completed (n = 200) Completed (n = 90) Currently in treatment (n = 109) P-value
Treatment characteristicsa
 <30 days in index treatment (% yes) 346 (87.2) 70 (95.9) 99 (81.1) 84 (89.4) 92 (87.6) 0.023 178 (89.0) 75 (83.3) 93 (86.9) 0.409
 Inpatient 134 (33.6%) 29 (39.7%) 40 (32.8%) 33 (35.1%) 32 (29.9%) 0.573 69 (34.5%) 33 (36.7%) 32 (29.4%) 0.514
 Ratio patients: providers 8.7 (7.7) 9.2 (9.1) 8.7 (8.1) 8.4 (5.7) 8.5 (8.0) 0.906 8.8 (7.9) 8.7 (7.1) 8.4 (8.0) 0.93
 Abstinence needed for treatment (% yes) 81 (20.7) 20 (27.8) 20 (16.7) 18 (19.4) 23 (22.1) 0.308 47 (24.0) 11 (12.4) 23 (21.7) 0.077
Participant characteristicsa
 Age, y 37.7 (10.6) 36.5 (9.4) 36.2 (10.6) 39.4 (10.5) 38.8 (11.4) 0.068 36.3 (9.9) 39.5 (11.0) 38.6 (11.4) 0.036
 Female 129 (32.3%) 25 (34.2%) 32 (26.2%) 29 (30.9%) 43 (40.2%) 0.152 54 (27.0%) 32 (35.6%) 43 (39.4%) 0.062
 Educational level 0.355 0.34
 Less than high school 155 (38.9%) 34 (46.6%) 45 (36.9%) 38 (40.9%) 37 (34.6%) 78 (39.0%) 39 (43.8%) 38 (34.9%)
 Completed high school 129 (32.4%) 24 (32.9%) 45 (36.9%) 28 (30.1%) 32 (29.9%) 70 (35.0%) 27 (30.3%) 32 (29.4%)
 More than high school 114 (28.6%) 15 (20.5%) 32 (26.2%) 27 (29.0%) 38 (35.5%) 52 (26.0%) 23 (25.8%) 39 (35.8%)
 Employment (% yes) 250 (63.1) 39 (53.4) 76 (62.3) 63 (69.2) 69 (64.5) 0.212 123 (61.8) 57 (64.8) 70 (64.2) 0.858
 Insecure housing in the past 12 months (% yes) 29 (7.3) 9 (12.3) 9 (7.4) 8 (8.5) 3 (2.8) 0.106 17 (8.5) 9 (10.0) 3 (2.8) 0.093
 Primary substance 0.147 0.031
  Alcohol 100 (25.1%) 16 (21.9%) 24 (19.7%) 28 (29.8%) 32 (29.9%) 40 (20.0%) 28 (31.1%) 32 (29.4%)
  Cocaine 147 (36.8%) 28 (38.4%) 45 (36.9%) 30 (31.9%) 43 (40.2%) 75 (37.5%) 29 (32.2%) 43 (39.4%)
  Cocaine paste 124 (31.1%) 27 (37.0%) 43 (35.2%) 31 (33.0%) 22 (20.6%) 75 (37.5%) 26 (28.9%) 23 (21.1%)
  Other 28 (7.0%) 2 (2.7%) 10 (8.2%) 5 (5.3%) 10 (9.3%) 10 (5.0%) 7 (7.8%) 11 (10.1%)
 Past month benzodiazepine use (% yes)a 92 (57.1) 14 (45.2) 30 (60.0) 15 (50.0) 32 (65.3) 0.269 43 (53.8) 16 (51.6) 33 (66.0) 0.307
 Past month polysubstance use (%) 277 (69.4) 52 (71.2) 89 (73.0) 62 (66.0) 72 (67.3) 0.662 151 (75.5) 54 (60.0) 72 (66.1) 0.02
 Early initiation in substance use (% yes) 231 (58.2) 47 (65.3) 73 (59.8) 50 (53.2) 59 (55.7) 0.413 127 (63.8) 44 (48.9) 60 (55.6) 0.047
 Readiness to change score (URICA) 11.3 (1.2) 11.3 (1.3) 11.4 (1.1) 11.4 (1.1) 11.2 (1.3) 0.412 11.5 (1.1) 11.2 (1.1) 11.1 (1.3) 0.066
Exposure to treatment past 12 months 0.026 0.039
  No 202 (51.0%) 37 (50.7%) 65 (54.2%) 53 (56.4%) 45 (42.5%) 106 (53.5%) 50 (55.6%) 46 (42.6%)
  3 or less months 134 (33.8%) 30 (41.1%) 36 (30.0%) 32 (34.0%) 35 (33.0%) 69 (34.8%) 29 (32.2%) 36 (33.3%)
  4 or more months 60 (15.2%) 6 (8.2%) 19 (15.8%) 9 (9.6%) 26 (24.5%) 23 (11.6%) 11 (12.2%) 26 (24.1%)
 First SUD treatment (% yes) 171 (43.0) 31 (42.5) 56 (45.9) 39 (41.5) 43 (40.6) 0.858 87 (43.5) 41 (45.6) 43 (39.8) 0.702
 Network quality index 13.5 (4.9) 14.7 (5.2) 13.7 (4.7) 13.4 (4.8) 12.6 (4.7) 0.037 13.9 (5.0) 13.9 (4.6) 12.6 (4.7) 0.072
89 (22.4) 13 (18.1) 26 (21.5) 21 (22.3) 28 (26.2) 0.633 42 (21.2) 19 (21.1) 28 (25.7) 0.63
 Cohabitation with someone with a SUD (% yes)
 Has a close friend with SUD (% yes) 181 (45.4) 33 (45.2) 55 (45.1) 39 (41.5) 52 (48.6) 0.796 88 (44.0) 41 (45.6) 52 (47.7) 0.822
 Participated in self-support groups past 6 months (% yes) 148 (37.1) 22 (30.1) 44 (36.1) 34 (36.2) 47 (43.9) 0.291 65 (32.5) 35 (38.9) 48 (44.0) 0.123
 Physical comorbidity (% yes) 99 (24.8) 22 (30.1) 26 (21.3) 25 (26.6) 25 (23.4) 0.533 54 (27.0) 20 (22.2) 25 (22.9) 0.594
 Antisocial personality disorder (% yes) 110 (27.6) 26 (35.6) 39 (32.0) 20 (21.3%) 25 (23.6) 0.102 60 (30.0) 24 (26.7) 26 (24.1) 0.526
 Disruptive event past 12 months (% yes) 313 (78.4) 57 (78.1) 98 (80.3) 62 (66.0) 94 (87.9) 0.002 149 (74.5) 68 (75.6) 96 (88.1) 0.016
 For-profit crimes past 12 months (% yes) 117 (29.3) 19 (26.0) 40 (32.8) 25 (26.6) 31 (29.0) 0.7 61 (30.5) 24 (26.7) 32 (29.4) 0.802
 Adverse childhood experiences 3.8 (2.4) 3.9 (2.4) 3.6 (2.4) 3.5 (2.4) 4.3 (2.3) 0.13 3.8 (2.5) 3.4 (2.3) 4.3 (2.3) 0.057
Open in a new tab

Note: P-values were obtained using ANOVA for continuous variables and χ2 test for categorical variables. Tests were conducted excluding missing data.

Abbreviations: ANOVA, analysis of variance; SD, standard deviation; SUD, substance use disorder; URICA, University of Rhode Island Change Assessment.

a

Missing values (~1%–10%) are not shown (see Table S5–S6). The data shown is not weighted.

Changes in substance use and psychiatric outcomes from baseline to 18 months of follow-up

Table 2 shows weighted outcomes throughout time, to account for loss to follow-up (IPCW), hence, the analytic sample ‘stands in’ for those who were censored. A total of 90.7% declared past month use of their primary substance at baseline, decreasing to 35.3% at 18 months of follow-up (P < 0.001). Reductions were also observed for past month use of alcohol (73.5% to 50.5%, P < 0.001), cannabis (41.5% to 24.1%, P < 0.001), cocaine powder (51.7% to 16.1%, P < 0.001) and cocaine paste (43.6% to 16.1%, P < 0.001). Approximately half (51.4%) screened positive for at least one of the current psychiatric diagnoses included at baseline, which decreased to 39.1% at 18 months (P < 0.001). The most prevalent current psychiatric comorbidities included major depressive episodes (40.9% baseline, 30.7% 18 months, P = 0.002), generalized anxiety (27.2% baseline, 21.7% 18 months, P = 0.009), panic (10.6% baseline, 5.5% 18 months, P = 0.021) and PTSD (7.4% baseline, 8.3% 18 months, P = 0.587).

TABLE 2.

Changes in drug use and psychiatric comorbidities according to length of time in treatment and treatment compliance.

Outcomes Total (n = 399) Length of time in treatment
 Treatment compliance
3 or less months (n = 73) 4–7 months (n = 122) 8 or more months (n = 94) Currently in treatment (n = 109) Not completed (n = 200) Completed (n = 90) Currently in treatment (n = 109)
Primary substance month prevalence (% yes)
 Baseline 90.7 86.8 94.6 88.9 90.5 93.7 83.5 90.6
 18 months 35.3 39.9 42.5 32.8 25.2 44.6 24.7 24.9
 n Miss 14 1 6 2 5 5 3 6
Alcohol month prevalence (% yes)
 Baseline 73.5 72.6 73.2 77.0 71.3 73.7 77.1 70.0
 18 months 50.5 63.3 62.7 40.8 34.3 63.5 37.8 34.6
Cannabis month prevalence (% yes)
 Baseline 41.5 48.3 41.2 37.2 40.7 47.6 28.5 39.9
 18 months 24.1 39.5 27.1 16.1 15.6 34.2 10.1 15.3
Cocaine month prevalence (% yes)
 Baseline 51.7 57.5 50.0 50.0 52.8 54.3 43.9 53.0
 18 months 16.0 16.4 20.5 13.8 13.1 18.8 12.3 13.7
Cocaine paste month prevalence (% yes)
 Baseline 43.6 49.1 44.4 49.0 33.7 52.4 33.8 34.1
 18 months 16.1 26.5 17.2 14.3 8.9 23.4 7.2 8.7
Any psychiatric comorbidity (% yes)
 Baseline 51.4 56.8 49.5 40.7 60.4 50.3 44.8 59.2
 18 months 39.1 46.9 35.9 42.5 35.1 41.4 38.9 34.4
Mayor depressive episode (% yes)
 Baseline 40.9 48.5 36.7 33.7 47.3 39.1 38.6 46.3
 18 months 30.7 40.3 28.9 32.0 25.3 34.9 27.4 24.8
Panic disorder (% yes)
 Baseline 10.6 18.4 8.2 8.7 9.5 12.8 6.7 9.3
 18 months 5.5 7.2 3.4 7.1 5.3 5.9 4.7 5.2
Generalized anxiety disorder (% yes)
 Baseline 27.2 31.0 29.3 17.1 32.0 27.4 22.0 31.3
 18 months 21.7 28.9 19.1 22.3 19.7 23.9 19.4 19.3
Post-traumatic stress disorder (% yes)
 Baseline 7.4 9.9 7.1 5.7 7.5 7.6 6.8 7.4
 18 months 8.3 10.0 6.0 5.8 12.4 8.1 4.1 12.2
Open in a new tab

Note: The data shown is weighted using inverse probability of censoring weighting to account for loss to follow-up.

Unadjusted and adjusted effects of length of time in treatment and compliance with treatment on substance use and psychiatric outcomes

After trimming the weights at the 2nd and 98th percentiles, covariate balance was generally improved, most with SMD <10% (Table S8–S9). Figure 1 summarizes the main results and Table 3 shows the unadjusted RR and the adjusted marginal RR using IPCW and IPTW for our primary outcomes.

FIGURE 1.

FIGURE 1

Open in a new tab

Marginal risk ratio (RR) and 95% CI (LL, lower limit; UL, upper limit) using the IPTW*IPCW trimmed at 2 to 98 percentiles for length of time in treatment and treatment compliance at 12 months on our primary and secondary outcomes. Adjustment for confounding and selection bias because of censoring on observed covariates is achieved by using the weights. IPCW, inverse probability of censoring weighting; IPTW, inverse probability of treatment weighting.

TABLE 3.

Length of time in treatment and treatment compliance at 12 months effect measure for last month’s primary substance use and any psychiatric comorbidities at 18 months.

Unadjusted RR
 Marginal RR weights (trimmed 2–98 percentiles)
RRa 95% CI P-value RRb 95% CI P-value
Past month’s use: primary substancec
Length of time in treatment (Ref. 0–3 months)
 4–7 months 1.19 0.83, 1.77 0.4 0.96 0.66, 1.41 0.9
 8 or more months 0.84 0.54, 1.31 0.4 0.62 0.38, 1.00 0.049
 Currently in treatment 0.71 0.45, 1.11 0.13 0.47 0.28, 0.81 0.006
Treatment compliance (Ref. not completed)
 Completed 0.56 0.36, 0.82 0.005 0.49 0.30, 0.80 0.004
 Currently in treatment 0.59 0.39, 0.83 0.005 0.5 0.31, 0.80 0.004
 n 382 368
Any psychiatric comorbiditiesd
Length of time in treatment (Ref. 0–3 months)
 4–7 months 0.8 0.57, 1.14 0.2 0.72 0.50, 1.03 0.074
 8 or more months 0.89 0.63, 1.28 0.5 0.83 0.57, 1.22 0.4
 Currently in treatment 0.76 0.53, 1.11 0.15 0.54 0.35, 0.84 0.006
Treatment compliance (Ref. not completed)
 Completed 0.97 0.70, 1.31 0.9 1.02 0.72, 1.46 0.9
 Currently in treatment 0.85 0.61, 1.15 0.3 0.76 0.51, 1.12 0.2
 n 399 383
Open in a new tab

Abbreviations: IPCW, inverse probability of censoring weighting; IPTW, inverse probability of treatment weighting; RR, risk ratio.

a

Unadjusted RR.

b

Marginal RR using trimmed (2–98 percentiles) IPCW*IPTW. Adjustment for confounding and selection bias is achieved by using the weights.

c

n = 382 for the unadjusted model, n = 368 for the marginal model. Fourteen participants declared ‘other’ as their primary substance and were not considered.

d

n = 399 for the unadjusted model, n = 383 for the marginal model.

Participating for 8+ months in treatment (vs. 0–3 months, measured at 12 months) was associated with lower risk of past month primary substance use at 18 months (RR = 0.62, 95% CI = 0.38–1.00), although this result was borderline significant, as did completing the treatment plan (vs. not completed, RR = 0.49, 95% CI = 0.30–0.80). Being currently in treatment at 12 months of follow-up was associated with lower risk of past month primary substance use at 18 months (vs. participating for 0–3 months: RR = 0.47, 95% CI = 0.28–0.81; vs. not completed treatment plan: RR = 0.5, 95% CI = 0.31–0.80).

Participating for 4 to 7 or 8+ months in treatment (vs. 0–3 months, measured at 12 months) were not associated with the risk of presenting any current psychiatric comorbidity at 18 months of follow-up (4–7 months: RR = 0.72, 95% CI = 0.50–1.03; 8+ months: RR = 0.83, 95% CI = 0.57–1.22), nor was treatment compliance (vs. not completed, RR = 1.02, 95% CI = 0.72–1.46). Being in treatment at 12 months of follow-up (vs. participating for 0–3 months: RR = 0.36, 95% CI = 0.17–0.75; vs. not completed treatment plan: RR = 0.66, 95% CI = 0.37–1.17) was associated with lower risk of current psychiatric comorbidities at 18 months.

Participating for 8+ months in treatment (vs. 0–3 months, measured at 12 months) was associated with a decrease in the risk of past month alcohol use at 18 months of follow-up (RR = 0.55, 95% CI = 0.38–0.79), as did completion of the treatment plan (vs. not completed, RR = 0.60, 95% CI = 0.43–0.84) (Table 4). In the case of cannabis, participating for more than 3 months (vs. 0–3 months, measured at 12 months) was associated with lower risk of past month use at 18 months (4–7 months: RR = 0.60, 95% CI = 0.38–0.94; 8+ months: RR = 0.31, 95% CI = 0.16–0.58) as did completion of treatment (vs. not completed, RR = 0.29, 95% CI = 0.12–0.71). For the past month’s use of cocaine powder at 18 months of follow-up all estimates include the null. Looking at the descriptive results, all groups presented noteworthy decreases in cocaine powder use from baseline to 18 months regardless of their treatment status at 12 months. Last, participating for 8+ months in treatment (vs. 0–3 months, measured at 12 months) was associated with a decrease in the risk of past month cocaine paste use at 18 months by a factor of 0.39 (95% CI = 0.17–0.90).

TABLE 4.

Length of time in treatment and treatment compliance at 12 months effect measure for last month’s alcohol, cannabis, cocaine powder and cocaine paste prevalence at 18 months.

Unadjusted RR (n = 399)
 Marginal RR weights (trimmed 2–98 percentiles) (n = 383)
RRa 95% CI P-value RRb 95% CI P-value
Past month’s use: alcohol
 Length of time in treatment (Ref. 0–3 months)
  4–7 months 1.04 0.83, 1.34 0.7 0.93 0.74, 1.17 0.5
  8 or more months 0.67 0.48, 0.91 0.012 0.55 0.38, 0.79 0.001
  Currently in treatment 0.59 0.42, 0.81 0.001 0.48 0.33, 0.70 <0.001
 Treatment compliance (Ref. not completed)
  Completed 0.6 0.43, 0.78 <0.001 0.6 0.43, 0.84 0.003
  Currently in treatment 0.57 0.42, 0.74 <0.001 0.56 0.40, 0.78 <0.001
Past month’s use: cannabis
 Length of time in treatment (Ref. 0–3 months)
  4–7 months 0.67 0.43, 1.05 0.074 0.6 0.38, 0.94 0.025
  8 or more months 0.39 0.21, 0.69 0.002 0.31 0.16, 0.58 <0.001
  Currently in treatment 0.39 0.22, 0.68 0.001 0.29 0.15, 0.57 <0.001
 Treatment compliance (Ref. not completed)
  Completed 0.29 0.13, 0.54 <0.001 0.29 0.12, 0.71 0.007
  Currently in treatment 0.45 0.26, 0.73 0.002 0.47 0.25, 0.89 0.021
Past month’s use: cocaine powder
 Length of time in treatment (Ref. 0–3 months)
  4–7 months 1.25 0.68, 2.43 0.5 0.9 0.45, 1.82 0.8
  8 or more months 0.84 0.41, 1.76 0.6 0.68 0.30, 1.51 0.3
  Currently in treatment 0.8 0.39, 1.65 0.5 0.5 0.21, 1.21 0.12
 Treatment compliance (Ref. not completed)
  Completed 0.56 0.27, 1.01 0.075 0.58 0.28, 1.24 0.2
  Currently in treatment 0.64 0.35, 1.09 0.12 0.61 0.29, 1.28 0.2
Past month’s use: cocaine paste
 Length of time in treatment (Ref. 0–3 months)
  4–7 months 0.64 0.34, 1.19 0.2 0.59 0.30, 1.17 0.13
  8 or more months 0.49 0.22, 0.99 0.052 0.39 0.17, 0.90 0.028
  Currently in treatment 0.38 0.17, 0.80 0.014 0.28 0.11, 0.71 0.007
 Treatment compliance (Ref. not completed)
  Completed 0.29 0.10, 0.65 0.007 0.43 0.17, 1.07 0.071
  Currently in treatment 0.43 0.20, 0.82 0.018 0.41 0.18, 0.94 0.035
Open in a new tab

Abbreviations: IPCW, inverse probability of censoring weighting; IPTW, inverse probability of treatment weighting; RR, risk ratio.

a

Unadjusted RR.

b

Marginal RR using trimmed (2–98 percentiles) IPCW*IPTW. Adjustment for confounding and selection bias because of censoring on observed covariates is achieved by using the weights.

Sensitivity analysis

When excluding treatment-experienced individuals (n = 90) the direction of the associations remained the same, although estimates lost precision (Table S11). When using an updated version of the exposures (measured at 18 months instead of 12 months) the protective association of being currently in treatment included the null (Table S12). There was little variation in the effect sizes when restricting the sample to those with propensity scores within the area of common support (Table S13) or those who had experienced another treatment episode between baseline and 12 months (Table S14). The traditional adjusted outcome regression approach, which estimates treatment effects conditional on covariates (as opposed to marginal effects), yielded similar directions yet slightly stronger effect sizes (Table S15). E-values evaluate how strong an unmeasured confounder associated with both treatment and outcome needs to be to ‘explain away’ the observed effect estimates beyond measured confounders. The strongest effect size for past month primary substance use was an RR of 0.49 (95% CI = 0.30–0.80), which could be explained by an unmeasured confounder that was associated with both treatment compliance and the outcome by an RR of at least 3.5-fold each. That is, there would have to be a very strong unmeasured confounder to ‘explain away’ the observed RR of 0.49. The weakest observed effect size was 0.62 (95% CI = 0.38–1.00), in which case the unmeasured confounder would need to have an RR of at least 2.61 under the same conditions.

DISCUSSION

Among a cohort of individuals in Región Metropolitana, Chile, we observed a reduction in substance use 18 months after initiating treatment for SUD. Specifically, we observed reductions in the past month’s primary substance use (from 91% to 35%), and for specific substances (alcohol, cannabis, cocaine powder and cocaine paste).

Our findings suggest an association between participating for longer times in treatment and compliance with the treatment plan and a reduction in substance use, consistent with evidence from higher-income countries [1–8]. Longer time in treatment (8+ months) was particularly relevant for reductions in alcohol use. Novel harm reduction approaches, such as managed alcohol programs, have been shown to improve retention among people with more severe alcohol use disorder [53], but have not been implemented in Chile. Although the follow-up considered in the study aligns with international standards [38], longer follow-up periods (e.g. 5 or 10 years) are critical to surveilling treatment outcomes for SUD given their chronicity for some individuals [54]. Longitudinal prospective cohort studies evaluating treatment under routine conditions are key to studying the outcomes of treatment-seeking individuals and have been around for over 20 years in high-income countries [4, 55, 56]. However, studies situated in low- and middle-income countries are scarce if not nonexistent. Two small (n = 65 and n = 98) randomized controlled trials (of a 12-week intervention) comparing contingency management to treatment as usual in Brazil among people who use crack cocaine showed favorable results [57, 58].

Reductions in cocaine powder use occurred regardless of length of time in or compliance with treatment. We might not observe a treatment effect because we lack a no-treatment comparison group, which is the norm in treatment outcome studies [38]. However, the ground-breaking United States Drug Abuse Treatment Outcome Study (DATOS) [59] also found decreases in cocaine powder use regardless of time in treatment at 1-year follow-up. In the case of Chile, representative national surveys show that cocaine powder use is more prevalent among those with low socio-economic status, nevertheless, cocaine paste use is almost nonexistent in higher-status groups [60]. Therefore, people who use cocaine powder might have additional unmeasured protective factors, such as income or social class, which could explain the reductions in cocaine use despite exposure to treatment. Not just in South America, cocaine powder is a more affluent drug, whereas smokable versions of cocaine (such as crack or cocaine paste) suffer from greater stigmatization and discrimination [61–63]. Notably, half of our sample (54.5%) used alcohol and some type of cocaine the month before index treatment. Alcohol and cocaine create a new psychoactive substance when taken together (cocaethylene) [64], which is more damaging [65, 66] but rarely acknowledged in clinical settings or research [67]. Encouraging awareness of the increased risks and harms of co-use of alcohol and cocaine among patients and healthcare providers in the region is critical given their synergetic effects and highly concurrent use among people who use drugs.

Half of the participants (50.6%) met the criteria for at least one of the current comorbid conditions considered in our analysis (major depressive episode, panic, anxiety and PTSD) at baseline. Other studies have documented between 20% and 70% for residential treatment programs [68], and 62.6% in a study in México [69]. Our estimates are conservative considering the entire MINI was not assessed. Most of the psychiatric comorbidities (except PTSD) showed a decrease from baseline to 18 months, however, their reduction could not be attributed to the length of time in treatment or compliance with the treatment. Interestingly, a third of those without psychiatric comorbidities at baseline (29.4%, data not shown) met criteria at 18 months, among whom two-thirds discontinued treatment (65.5%). Understanding the potentially harmful effects of ‘failing’ a treatment episode is critical, given how stigmatized treatment for SUD is in the general public and media [70, 71]. Still, fluctuations in mental health symptoms are common among this population, which could explain these results. Integrating SUD treatment with other mental health services to address the complex comorbid profiles of treatment-seeking individuals might help improve their outcomes, although historically they have been provided in separate systems in Chile [72, 73].

Limitations

This study has several limitations. First, because treatment conditions were not randomized, results cannot be attributed directly to treatment factors. We used a robust MSM approach to adjust for confounding and selection bias because of loss to follow-up and presented a traditional multivariate regression approach, which yielded similar results. Although covariate balance was improved after weighting, some variables remained over the recommended thresholds, pointing toward potential residual confounding. For example, other psychiatric comorbidities that were not measured could still be associated with both treatment and outcome. However, our E-value analysis suggests that a very strong degree of unmeasured confounding could fully ‘explain away’ our findings. Second, our consistency assumption is hard to defend entirely, because the received interventions are likely heterogeneous, which might reduce precision. As in other regions [38, 59, 72], non-governmental private organizations first implemented treatment under the model of therapeutical communities in Chile. This complicates defining the intervention as each program implements its therapeutic approach, some not being evidenced-based. Implementation science studies are needed to unveil what interventions are provided by SUD treatment programs in South America, which implement mostly behavioral health interventions. Furthermore, our small sample size prevented doing stratified analysis by treatment modality. Still, both outpatient and residential modalities have shown positive effects on alcohol and cocaine use [59] and we adjusted for treatment modality in the IPW procedures. Additional analysis showed consistency in the direction of most estimates when stratifying by modality (Table S16). Third, our results are not generalizable to the population receiving treatment for SUD in Región Metropolitana, because some programs refused to participate, and there are privately funded programs that are not part of SENDA’s network. Fourth, our retention is probably missing not at random (MNAR) and the assumptions for IPCW might not hold, although covariates were balanced when applying weights between censoring groups. Additionally, retention was slightly below 70%, which is recommended [38], still, our study shows similar retention to other longitudinal cohorts among treatment-seeking populations [10, 27, 74, 75]. Fifth, we were unable to use random effects to account for site-level variation because some sites contributed one or few observations. Sixth, our survey included sensitive topics, such as ACEs and substance use, which are prone to under-reporting because of social desirability bias.

CONCLUSION

This study contributes valuable insights into substance use and psychiatric comorbidities outcomes among individuals with SUDs in Región Metropolitana, Chile, suggesting a positive association between longer time in and compliance with treatment and a subsequent reduction in substance use. The findings underscore the need to disclose the current interventions provided by treatment programs in South America, as well as the design and implementation of interventions tailored for its supply and demand, where cocaine and alcohol use are predominantly burdensome. Contingency management has been proven to increase treatment retention and decrease cocaine use [76], but is barely applied in research, clinical practice or encouraged through policy [77]. Improving treatment of co-morbid psychiatric disorders and integration with other mental health services is critical given how burdensome they are among this population.

Supplementary Material

Supplementary Material

SUPPORTING INFORMATION

Additional supporting information can be found online in the Supporting Information section at the end of this article.

ACKNOWLEDGEMENTS

We thank the Department of Budget of the Treasury of Chile (Direccion de Presupuesto del Ministerio de Hacienda, DIPRES) for funding this study. We thank Agencia Nacional de Investigacion y Desarrollo (ANID) for partially supporting this study. We acknowledge Andrea López, Luis Maldonado and Paloma Del Villar, who contributed to the broader evaluation. Ryan Cooper, Rodrigo Díaz and Claudia Chamorro from the DIPRES provided valuable feedback. Andrea Titus and Audrey Renson provided guidance and feedback on the analytic approach. We acknowledge Pablo Pinochet, Francisco Del Río and Elizabeth Pérez, for helping coordinate the fieldwork of the study. We also thank all interviewers, who dedicatedly helped us track and maintain contact with the participants and all the workers of SENDA and the treatment programs, for helping us throughout the whole process. Finally, we want to thank all participants of the study for providing us the opportunity to collect information about their lives.

Funding information

This study was funded by Dirección de Presupuesto, Ministerio de Hacienda de Chile. Agencia Nacional de Investigación y Desarrollo (ANID) partially funded this study—Millennium Science Initiative Program - NCS2021_003. N.K. was supported by the National Institute On Drug Abuse of the National Institutes of Health under Award Number K01DA055758. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

DECLARATIONS OF INTEREST

None.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study were from Dirección de Presupuesto, Ministerio de Hacienda. Restrictions apply to the availability of these data, which were used under license for this study. Data are available https://transparenciaactiva.presidencia.cl/Informacion_Publica.html with the permission of Dirección de Presupuesto, Ministerio de Hacienda and Transparencia, Presidencia de la República, Chile.

REFERENCES

  • 1.Eastwood B, Strang J, Marsden J. Effectiveness of treatment for opioid use disorder: a national, five-year, prospective, observational study in England. Drug Alcohol Depend. 2017;176:139–47. 10.1016/j.drugalcdep.2017.03.013 [DOI] [PubMed] [Google Scholar]
  • 2.Marsden J, Eastwood B, Bradbury C, Dale-Perera A, Farrell M, Hammond P, et al. Effectiveness of community treatments for heroin and crack cocaine addiction in England: a prospective, in-treatment cohort study. Lancet. 2009;374(9697):1262–70. 10.1016/S0140-6736(09)61420–3 [DOI] [PubMed] [Google Scholar]
  • 3.Stewart D, Gossop M, Marsden J. Reductions in non-fatal overdose after drug misuse treatment: results from the National Treatment Outcome Research Study (NTORS). J Subst Abuse Treat. 2002;22(1):1–9. 10.1016/S0740-5472(01)00206-9 [DOI] [PubMed] [Google Scholar]
  • 4.Hubbard RL, Craddock SG, Anderson J. Overview of 5-year followup outcomes in the drug abuse treatment outcome studies (DATOS). J Subst Abuse Treat. 2003;25(3):125–34. 10.1016/S0740-5472(03)00130-2 [DOI] [PubMed] [Google Scholar]
  • 5.Gossop M The National Treatment Outcomes Research Study (NTORS) and its influence on addiction treatment policy in the United Kingdom: NTORS and treatment policy. Addiction. 2015;110(S2):50–3. 10.1111/add.12906 [DOI] [PubMed] [Google Scholar]
  • 6.Fernández-Montalvo J, López-Goñi JJ, Illescas C, Landa N, Lorea I. Evaluation of a therapeutic community treatment program: a long-term follow-up study in Spain. Subst Use Misuse. 2008;43(10):1362–77. 10.1080/10826080801922231 [DOI] [PubMed] [Google Scholar]
  • 7.Fernández-Hermida JR, Secades-Villa R, Fernández-Ludeña JJ, Marina-González PA. Effectiveness of a therapeutic community treatment in Spain: a long-term follow-up study. Eur Addict Res. 2002;8(1):22–9. 10.1159/000049484 [DOI] [PubMed] [Google Scholar]
  • 8.Prendergast ML, Podus D, Chang E, Urada D. The effectiveness of drug abuse treatment: a meta-analysis of comparison group studies. Drug Alcohol Depend. 2002;67(1):53–72. 10.1016/S0376-8716(02)00014-5 [DOI] [PubMed] [Google Scholar]
  • 9.Simpson DD, Joe GW, Broome KM. A national 5-year follow-up of treatment outcomes for cocaine dependence. Arch Gen Psychiatry. 2002;59(6):538–44. 10.1001/archpsyc.59.6.538 [DOI] [PubMed] [Google Scholar]
  • 10.Jones A, Donmall M, Millar T, Weston S, Anderson T, Gittins M, et al. The Drug Treatment Outcomes Research Study (DTORS) : final outcomes report. London: Home Office; 2009. [Google Scholar]
  • 11.Lai HMX, Cleary M, Sitharthan T, Hunt GE. Prevalence of comorbid substance use, anxiety and mood disorders in epidemiological surveys, 1990–2014: a systematic review and meta-analysis. Drug Alcohol Depend. 2015;154:1–13. 10.1016/j.drugalcdep.2015.05.031 [DOI] [PubMed] [Google Scholar]
  • 12.Torrens M, Rossi PC, Martinez-Riera R, Martinez-Sanvisens D, Bulbena A. Psychiatric co-morbidity and substance use disorders: treatment in parallel systems or in one integrated system? Subst Use Misuse. 2012;47(8–9):1005–14. 10.3109/10826084.2012.663296 [DOI] [PubMed] [Google Scholar]
  • 13.Krawczyk N, Feder KA, Saloner B, Crum RM, Kealhofer M, Mojtabai R. The association of psychiatric comorbidity with treatment completion among clients admitted to substance use treatment programs in a U.S. national sample. Drug Alcohol Depend. 2017;1(175):157–63. 10.1016/j.drugalcdep.2017.02.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Zhang Z, Friedmann PD, Gerstein DR. Does retention matter? Treatment duration and improvement in drug use. Addiction. 2003;98(5):673–84. 10.1046/j.1360-0443.2003.00354.x [DOI] [PubMed] [Google Scholar]
  • 15.Best D, Hamer R. Addiction Recovery in Services and Policy: An International Overview. In: el Guebaly N, Carrà G, Galanter M, Baldacchino AM, editors. Textbook of addiction treatment [internet] Cham: Springer International Publishing; 2021[cited 2023 Nov 27]. p. 717–31. Available from: https://link.springer.com/10.1007/978-3-030-36391-8_50 [Google Scholar]
  • 16.McKay JR. Impact of continuing care on recovery from substance use disorder. ARCR. 2021;41(1):01. 10.35946/arcr.v41.1.01 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Castaldelli-Maia JM, Wang YP, Brunoni AR, Faro A, Guimarães RA, Lucchetti G, et al. Burden of disease due to amphetamines, cannabis, cocaine, and opioid use disorders in South America, 1990–2019: a systematic analysis of the global burden of disease study 2019. Lancet Psychiatry. 2023;10(2):85–97. 10.1016/S2215-0366(22)00339-X [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Degenhardt L, Webb P, Colledge-Frisby S, Ireland J, Wheeler A, Ottaviano S, et al. Epidemiology of injecting drug use, prevalence of injecting-related harm, and exposure to behavioural and environmental risks among people who inject drugs: a systematic review. Lancet Glob Health. 2023;11(5):e659–72. 10.1016/S2214-109X(23)00057-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Rehm J, Shield KD. Global burden of alcohol use disorders and alcohol liver disease. Biomedicine. 2019;7(4):99. 10.3390/biomedicines7040099 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Bastos FI, Caiaffa W, Rossi D, Vila M, Malta M. The children of mama coca: coca, cocaine and the fate of harm reduction in South America. Int J Drug Policy. 2007;18(2):99–106. 10.1016/j.drugpo.2006.11.017 [DOI] [PubMed] [Google Scholar]
  • 21.Epele ME. New toxics, new poverty: a social understanding of the freebase cocaine/Paco in Buenos Aires. Argentina Subst Use Misuse. 2011;46(12):1468–76. 10.3109/10826084.2011.576745 [DOI] [PubMed] [Google Scholar]
  • 22.Olivari CF, Gaete J, Rodriguez N, Pizarro E, Del Villar P, Calvo E, et al. Polydrug use and co-occurring substance use disorders in a respondent driven sampling of Cocaine Base paste users in Santiago. Chile J Psychoactive Drugs. 2022;54(4):348–57. 10.1080/02791072.2021.1976886 [DOI] [PubMed] [Google Scholar]
  • 23.United Nations Office on Drugs and Crime. World drug report 2023 [internet] United Nations; 2023 [cited 2023 Nov 9]. (World Drug Report). Available from: https://www.un-ilibrary.org/content/books/9789210028233 [Google Scholar]
  • 24.Magura S, Rosenblum A. Modulating effect of alcohol use on cocaine use. Addict Behav. 2000;25(1):117–22. 10.1016/S0306-4603(98)00128-2 [DOI] [PubMed] [Google Scholar]
  • 25.Santo T Jr, Clark B, Hickman M, Grebely J, Campbell G, Sordo L, et al. Association of Opioid Agonist Treatment with all-cause mortality and specific causes of death among people with opioid dependence: a systematic review and meta-analysis. JAMA Psychiatry. 2021;78(9):979–93. 10.1001/jamapsychiatry.2021.0976 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Kampman KM. The treatment of cocaine use disorder. Sci Adv. 2019;5(10):eaax1532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Lappan SN, Brown AW, Hendricks PS. Dropout rates of in-person psychosocial substance use disorder treatments: a systematic review and meta-analysis. Addiction. 2020;115(2):201–17. 10.1111/add.14793 [DOI] [PubMed] [Google Scholar]
  • 28.Levav I, Restrepo H, Guerra de Macedo C. The restructuring of psychiatric care in Latin America: a new policy for mental health services. J Public Health Policy. 1994;15(1):71–85. 10.2307/3342608 [DOI] [PubMed] [Google Scholar]
  • 29.Caldas de Almeida JM, Horvitz-Lennon M. Mental health care reforms in Latin America: an overview of mental health care reforms in Latin America and the Caribbean. Psychiatr Serv. 2010;61(3):218–21. 10.1176/ps.2010.61.3.218 [DOI] [PubMed] [Google Scholar]
  • 30.Marín-Navarrete R, Medina-Mora ME, Pérez-López A, Horigian VE. Development and evaluation of addiction treatment programs in Latin America. Curr Opin Psychiatry. 2018;31(4):306–14. 10.1097/YCO.0000000000000434 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Mateo Pinones M, González-Santa Cruz A, Portilla Huidobro R, Castillo-Carniglia A. Evidence-based policymaking: lessons from the Chilean substance use treatment policy. Int J Drug Policy. 2022;109:103860. 10.1016/j.drugpo.2022.103860 [DOI] [PubMed] [Google Scholar]
  • 32.Organización Panamericana de la Salud. Organización Mundial de la Salud. In: Sesión del Comité Ejecutivo: Estrategia y plan de Acción sobre El Consumo de Sustancias Psicoactivas y la Salud Pública: Informe final [internet] Washington, D.C.; 2022. [cited 2023 Oct 11]. p. 1–12 Report No.: CE170/INF/6. Available from: https://www.paho.org/sites/default/files/ce170-inf-6-s-sustancias-psicoactivas-salud-publica_0.pdf [Google Scholar]
  • 33.Ruiz-Tagle Maturana J, Cruz AGS, Rocha-Jiménez T, Castillo-Carniglia Á. Does substance use disorder treatment completion reduce the risk of treatment readmission in Chile? Drug Alcohol Depend. 2023;248:109907. 10.1016/j.drugalcdep.2023.109907 [DOI] [PubMed] [Google Scholar]
  • 34.Olivari CF, Gonzáles-Santa Cruz A, Mauro PM, Martins SS, Sapag J, Gaete J, et al. Treatment outcome and readmission risk among women in women-only versus mixed-gender drug treatment programs in Chile. J Subst Abuse Treat. 2022;134:108616. 10.1016/j.jsat.2021.108616 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Mateo Pinones M, González-Santa Cruz A, Castillo-Carniglia A, Bond C, Payne J, McGee TR. Substance use treatment completion and criminal justice system contact in Chile: a retrospective, linked data cohort study. Addiction. 2024;119(8):1440–52. [cited 2024 Mar 30]. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/add.16488 [DOI] [PubMed] [Google Scholar]
  • 36.Servicio Nacional Para la Prevención y Rehabilitación del Consumo de Drogas, Ministerio de Salud. Orientaciones técnicas de Programas de tratamiento y rehabilitación Para personas adultas con problemas derivados del consumo de drogas y alcohol en Convenio SENDA-MINSAL; 2020. [Google Scholar]
  • 37.Bobes J A Spanish validation study of the mini international neuropsychiatric interview. Eur Psychiatry. 1998;13(S4):198S–199S. 10.1016/S0924-9338(99)80240-5 [DOI] [Google Scholar]
  • 38.Marsden J, Ogborne A, Farrell M, Rush B. International guidelines for the evaluation of treatment services and systems for psychoactive substance use disorders [internet] Geneva: WHO/UNDC-P/EMCDDA; 2000. [cited 2023 Nov 9]. Available from: https://policycommons.net/artifacts/577688/international-guidelines-for-the-evaluation-of-treatment-services-and-systems-for-psychoactive-substance-use-disorders-john-marsden/1556468/ [Google Scholar]
  • 39.McConnaughy E, Prochaska J, Velicer W. Stages of change in psychotherapy: measurement and sample profiles. Psychotherapy: theory, research, and practice, 20, 368–375. Psychotherapy: theory. Dermatol Res Pract. 1983;20(3):368–75. 10.1037/h0090198 [DOI] [Google Scholar]
  • 40.Flores M, Faccio CO, Luque W. Escala de evaluación del cambio de la Universidad de Rhode Island (URICA): validez y confiabilidad en drogadependientes. Revista de Psicología Universidad César Vallejo. 2015;9(1):52–9. [Google Scholar]
  • 41.Robins JM, Hernán MÁ, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11(5):550–60. 10.1097/00001648-200009000-00011 [DOI] [PubMed] [Google Scholar]
  • 42.Moreno-Betancur M, Lynch JW, Pilkington RM, Schuch HS, Gialamas A, Sawyer MG, et al. Emulating a target trial of intensive nurse home visiting in the policy-relevant population using linked administrative data. Int J Epidemiol. 2023;52(1):119–31. 10.1093/ije/dyac092 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Hernán MA, Wang W, Leaf DE. Target trial emulation: a framework for causal inference from observational data. Jama. 2022;328(24):2446–7. 10.1001/jama.2022.21383 [DOI] [PubMed] [Google Scholar]
  • 44.Baidwan NK, Gerberich SG, Kim H, Ryan AD, Church T, Capistrant BD. A marginal structural model approach to analyse work-related injuries: an example using data from the health and retirement study. Inj Prev. 2020;26(3):248–53. 10.1136/injuryprev-2018-043124 [DOI] [PubMed] [Google Scholar]
  • 45.Chesnaye NC, Stel VS, Tripepi G, Dekker FW, Fu EL, Zoccali C, et al. An introduction to inverse probability of treatment weighting in observational research. Clin Kidney J. 2022;15(1):14–20. 10.1093/ckj/sfab158 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Hernan MA, Robins JM. Causal inference: what if [internet] Boca Raton: Chapman & Hall/CRC; 2020. [cited 2024 Mar 20]. Available from: https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/ [Google Scholar]
  • 47.van der Wal WM, Geskus RB. Ipw: an R package for inverse probability weighting. J Stat Softw. 2011;14(43):1–23. 10.18637/jss.v043.i13 [DOI] [Google Scholar]
  • 48.van Buuren S, Groothuis-Oudshoorn K, Vink G, Schouten R, Robitzsch A, Rockenschaub P, et al. Mice: multivariate imputation by chained equations [internet]; 2023. [cited 2023 Nov 12]. Available from: https://cran.r-project.org/web/packages/mice/index.html [Google Scholar]
  • 49.Mitra R, Reiter JP. A comparison of two methods of estimating propensity scores after multiple imputation. Stat Methods Med Res. 2016;25(1):188–204. 10.1177/0962280212445945 [DOI] [PubMed] [Google Scholar]
  • 50.Ségalas C, Leyrat C, Carpenter JR, Williamson E. Propensity score matching after multiple imputation when a confounder has missing data. Stat Med. 2023;42(7):1082–95. 10.1002/sim.9658 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Højsgaard S, Halekoh U, Yan J. The R package geepack for generalized estimating equations. J Stat Softw. 2006;15:1–11. [Google Scholar]
  • 52.Mathur MB, Ding P, Riddell CA, VanderWeele TJ. Website and R package for computing E-values. Epidemiology. 2018;29(5):e45–7. 10.1097/EDE.0000000000000864 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Smith-Bernardin SM, Suen LW, Barr-Walker J, Cuervo IA, Handley MA. Scoping review of managed alcohol programs. Harm Reduct J. 2022;19(1):82. 10.1186/s12954-022-00646-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Hser YI, Longshore D, Brecht ML, Anglin MD. Studying the Natural History of Drug Use. In: Sloboda Z, editorEpidemiology of drug abuse [internet] New York: Springer-Verlag; 2005[cited 2023 Jan 13]. p. 29–43. Available from: http://link.springer.com/10.1007/0-387-24416-6_3 [Google Scholar]
  • 55.Gossop M, Marsden J, Stewart D, Kidd T. The National Treatment Outcome Research Study (NTORS): 4–5 year follow-up results. Addiction. 2003;98(3):291–303. 10.1046/j.1360-0443.2003.00296.x [DOI] [PubMed] [Google Scholar]
  • 56.Teesson M, Ross J, Darke S, Lynskey M, Ali R, Ritter A, et al. One year outcomes for heroin dependence: findings from the Australian treatment outcome study (ATOS). Drug Alcohol Depend. 2006;83(2):174–80. 10.1016/j.drugalcdep.2005.11.009 [DOI] [PubMed] [Google Scholar]
  • 57.Miguel AQC, Madruga CS, Cogo-Moreira H, Yamauchi R, Simões V, da Silva CJ, et al. Contingency management is effective in promoting abstinence and retention in treatment among crack cocaine users in Brazil: a randomized controlled trial. Psychol Addict Behav. 2016;30(5):536–43. 10.1037/adb0000192 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Miguel AQC, McPherson SM, Simões V, Yamauchi R, Madruga CS, Smith CL, et al. Effectiveness of incorporating contingency management into a public treatment program for people who use crack cocaine in Brazil. A single-blind randomized controlled trial. Int J Drug Policy. 2022;1(99):103464. 10.1016/j.drugpo.2021.103464 [DOI] [PubMed] [Google Scholar]
  • 59.Hubbard RL, Craddock SG, Flynn PM, Anderson J, Etheridge RM. Overview of 1-year follow-up outcomes in the drug abuse treatment outcome study (DATOS). Psychol Addict Behav. 1997;11(4):261–78. 10.1037/0893-164X.11.4.261 [DOI] [Google Scholar]
  • 60.Ministerio del Interior y Seguridad Pública. SENDA. Décimo Tercer Estudio Nacional de Drogas en Población general; 2018. [Google Scholar]
  • 61.Goulian A, Jauffret-Roustide M, Dambélé S, Singh R, Fullilove RE. A cultural and political difference: comparing the racial and social framing of population crack cocaine use between the United States and France. Harm Reduct J. 2022;19(1):44. 10.1186/s12954-022-00625-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Nabben T, Korf DJ. Cocaine and crack in Amsterdam: diverging subcultures. J Drug Issues. 1999;29(3):627–51. 10.1177/002204269902900311 [DOI] [Google Scholar]
  • 63.Williams L, Parker H. Alcohol, cannabis, ecstasy and cocaine: drugs of reasoned choice amongst young adult recreational drug users in England. Int J Drug Policy. 2001;12(5):397–413. 10.1016/S0955-3959(01)00104-9 [DOI] [Google Scholar]
  • 64.Andrews P Cocaethylene toxicity. J Addict Dis. 1997;16(3):75–84. 10.1300/J069v16n03_08 [DOI] [PubMed] [Google Scholar]
  • 65.Henning RJ, Wilson LD. Cocaethylene is as cardiotoxic as cocaine but is less toxic than cocaine plus ethanol. Life Sci. 1996;59(8):615–27. 10.1016/0024-3205(96)00227-5 [DOI] [PubMed] [Google Scholar]
  • 66.Wilson LD, Jeromin J, Garvey L, Dorbandt A. Cocaine, ethanol, and Cocaethylene Cardiotoxity in an animal model of cocaine and ethanol abuse. Acad Emerg Med. 2001;8(3):211–22. 10.1111/j.1553-2712.2001.tb01296.x [DOI] [PubMed] [Google Scholar]
  • 67.Pergolizzi J, Breve F, Magnusson P, LeQuang JAK, Varrassi G Jr, et al. Cocaethylene: when cocaine and alcohol are taken together. Cureus. 2022;14(2):1–7. 10.7759/cureus.22498 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Hasin D, Nunes E, Meydan J. Comorbidity of Alcohol, Drug, and Psychiatric Disorders: Epidemiology. In: Kranzler HR, Tinsley JA, editors-Dual diagnosis and psychiatric treatment [internet] 0th ed. CRC Press; 2004[cited 2024 Feb 23]. p. 12–42. Available from: https://www.taylorfrancis.com/books/9781135539443/chapters/10.3109/9780203017616-7 [Google Scholar]
  • 69.Marín-Navarrete R, Medina-Mora ME, Horigian VE, Salloum IM, Villalobos-Gallegos L, Fernández-Mondragón J. Co-occurring disorders: a challenge for Mexican community-based residential care facilities for substance use. J Dual Diagn. 2016;12(3–4):261–70. 10.1080/15504263.2016.1220207 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.McGinty EE, Kennedy-Hendricks A, Barry CL. Stigma of Addiction in the Media. In: Avery JD, Avery JJ, editorsThe stigma of addiction: an essential guide [internet] Cham: Springer International Publishing; 2019[cited 2024 Mar 21]. p. 201–14. 10.1007/978-3-030-02580-9_11 [DOI] [Google Scholar]
  • 71.Corrigan PW, Kuwabara SA, O’Shaughnessy J. The public stigma of mental illness and drug addiction: findings from a stratified random sample. J Social Work. 2009;9(2):139–47. 10.1177/1468017308101818 [DOI] [Google Scholar]
  • 72.Hser YI, Anglin MD, Grella C, Longshore D, Prendergast ML. Drug treatment careers a conceptual framework and existing research findings. J Subst Abuse Treat. 1997;14(6):543–58. 10.1016/S0740-5472(97)00016-0 [DOI] [PubMed] [Google Scholar]
  • 73.McGovern MP, Lambert-Harris C, McHugo GJ, Giard J, Mangrum L. Improving the dual diagnosis capability of addiction and mental health treatment services: implementation factors associated with program level changes. J Dual Diagn. 2010;6(3–4):237–50. 10.1080/15504263.2010.537221 [DOI] [Google Scholar]
  • 74.Flynn PM, Craddock SG, Hubbard RL, Anderson J, Etheridge RM. Methodological overview and research Design for the Drug Abuse Treatment Outcome Study (DATOS). Psychol Addict Behav. 1997;11(4):230–43. 10.1037/0893-164X.11.4.230 [DOI] [Google Scholar]
  • 75.Comiskey C, Kelly P, Leckey Y, McCulloch L, O’Duill B, Stapleton R, et al. The ROSIE study: drug treatment outcomes in Ireland: a report for the National Advisory Committee on drugs [internet] Dublin: Stationery Office; 2009. [cited 2023 Oct 11]. Available from: https://www.drugsandalcohol.ie/11542/1/ROSIE3-YearReport.pdf [Google Scholar]
  • 76.Bentzley BS, Han SS, Neuner S, Humphreys K, Kampman KM, Halpern CH. Comparison of treatments for cocaine use disorder among adults. JAMA Netw Open. 2021;4(5):e218049. 10.1001/jamanetworkopen.2021.8049 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 77.Proctor SL. Rewarding recovery: the time is now for contingency management for opioid use disorder. Ann Med. 2022;54(1):1178–87. 10.1080/07853890.2022.2068805 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS2138219-supplement-Supplementary_Material.docx (743.6KB, docx)

Data Availability Statement

The data that support the findings of this study were from Dirección de Presupuesto, Ministerio de Hacienda. Restrictions apply to the availability of these data, which were used under license for this study. Data are available https://transparenciaactiva.presidencia.cl/Informacion_Publica.html with the permission of Dirección de Presupuesto, Ministerio de Hacienda and Transparencia, Presidencia de la República, Chile.

RESOURCES