Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jul 14.
Published in final edited form as: Subst Use Misuse. 2019 Jul 18;54(13):2134–2143. doi: 10.1080/10826084.2019.1638405

Predicting opioid and cocaine treatment response

Robert A Gardner 1, David H Epstein 2, Kenzie L Preston 2, Karran A Phillips 2
PMCID: PMC8278306  NIHMSID: NIHMS1533998  PMID: 31315479

Abstract

Background:

Treatment with methadone is effective in reducing heroin use, HIV risk, and death; however, not all patients respond to treatment. Better outcomes may emerge with personalized treatment based on factors that influence treatment courses.

Objectives:

to investigate psychosocial variables contributing to treatment response, using a comprehensive definition of treatment response.

Methods:

Seventy participants seeking treatment for heroin and cocaine addiction completed up to 40 weeks of daily methadone. At week 22, we administered a semi-structured interview for DSM-IV symptoms. We defined opioid treatment responders as people still enrolled at 22 weeks, not meeting past-30-day criteria for DSM-IV opioid abuse or dependence or DSM-5 opioid-use disorder, and providing ≥ 75% opioid-negative urine samples in the 30 days prior to week 22. The same criteria were applied to assess cocaine treatment response.

Results:

Sample was 71% male, 41% White, and averaged 39.4 ± 7.9 years old. Opioid treatment response was more likely in participants who had been employed over the past 3 years (OR 8.1, 95% CI 1.2–55) and less likely in those who spent more time on hobbies (OR 0.45, 95% CI 0.23–0.88). Cocaine treatment response was more likely in participants who had a good relationship with their father (OR 5.3, 95% CI 1.2–24) and less likely if positive for hepatitis C (OR 0.15, 95% CI 0.03–0.75).

Conclusions:

Pretreatment characteristics differentially predict treatment response for heroin and cocaine use. Similar research in diverse patient groups may aid in the development of personalized treatment combining biologic treatment with targeted psychosocial interventions.

1. INTRODUCTION

1.1. Background

Precision medicine is an initiative incorporating prevention and treatment strategies that considers individual differences in genes, environment, and lifestyle. While the proposed initiative has a near-term focus on cancers, there is a longer-term aim to generate knowledge applicable to the whole range of health and disease. The latter requires better assessment of disease risk, understanding of disease mechanisms, and prediction of optimal therapy (Collins and Varmus, 2015).

Prediction of optimal therapy in addiction would be more easily accomplished if we consistently and effectively defined treatment response and better understood the factors that predicted treatment response. Treatment response is often defined by objective markers of cessation of use and retention in treatment. These are both valuable indicators, but they may not adequately assess improvement in impulse control, social function, risky behavior, and pharmacological indictors (tolerance and withdrawal). These are relevant to clinical diagnosis as specified in the last several editions of the Diagnostic and Statistical Manual of Mental Disorders (DSM). If we use them for diagnosis, it is logical that we also use them to assess degree of response.

Treatment for addiction remains challenging because of the dynamic biopsychosocial factors underlying addictive behaviors. Even though the biological aspects of opioid addiction can be managed with opioid agonist treatment such as methadone, which reduces heroin use and craving (Ball et al., 1988; Barthwell et al., 1989; Gerstein and Lewin, 1990; McLellan et al., 1993; Caplehorn et al 1994; Goldstein and Herrera, 1995; Bell et al., 1997), outcome studies show that at the one-year mark, only 35% of patients treated with methadone are likely to remain adherent to their treatment program (Roux et al., 2014). These low treatment adherence rates may be secondary to the social systems where the individual resides and complicated by limited treatment options, antiquated drug policies, and a societal approach that focuses on criminalization over treatment. To develop an optimal therapeutic approach, it is imperative to additionally understand the biopsychosocial variables that influence treatment response.

Traditional prediction analyses in addiction focus on outcomes such as retention in treatment and cessation of use (Kelly et al., 2011; Roll, 2005; Hilhouse et al., 2009; Satre, 2011; Rowell et al., 2012; Peirce et al., 2009). In clinical settings, it may be more appropriate to identify predictors which determine a patient’s response to treatment defined in terms of control of use, social impairment, risky use, and tolerance and withdrawal (pharmacologic indicators). Most prior studies on that topic have emphasized the same outcome measure: substance use, as detected solely by urinalysis (Belding et al., 1998; Gerra et al., 2003; Lehman et al., 1993; Weiss et al., 2011). We believe in addition to objective urinalysis findings it is equally important to assess psychosocial functioning among our population of urban methadone maintenance clinic patients. Therefore, in our prediction analyses, in addition to substance use determined by urinalysis, we also include DSM symptoms in our definition of treatment response. To our knowledge, we are the only group to develop a treatment response measure that incorporates both urinalysis and whether the patient has changed from meeting to not meeting DSM-5 diagnostic criteria for a substance use disorder (see Methods section).

The analyses presented here were conducted on data we collected as part of a randomized, controlled clinical trial that evaluated fixed versus flexible (i.e., more than 100mg/day) methadone dosing combined with contingency management for treatment of opioid/cocaine dependence, in which we counterintuitively found that under double-blind conditions, dosages of methadone over 100 mg/day, even when prescribed based on specific signs and symptoms, were not better than 100 mg/day (Kennedy et al., 2013). Of note, in our primary analysis for that trial, we had defined treatment response as cessation of use, determined by urine toxicology. The purpose of this exploratory secondary analysis is to suggest a definition of treatment response that incorporates improvement in impaired control, decreased social impairment, decreased risky use, and a reduction in pharmacological indictors in addition to cessation of use and retention in treatment and then apply this definition to an exploration of the biopsychosocial factors that influence treatment response to better develop optimal therapeutic approaches and provide personalized addiction treatment.

2. METHODS

2.1. Participants

This study was a secondary analysis of a published randomized control trial (Kennedy et al., 2013). Participants were selected from 140 outpatients admitted for methadone treatment at a research clinic in Baltimore, MD. Screening included medical, psychiatric, and substance-use histories, physical examination, standard laboratory tests, and a battery of assessment instruments, including the Addiction Severity Index (ASI; McLellan et al., 1985), Social Adjustment Scale-Self Report (SAS-SR; Weissman and Bothwell, 1976), Substance Dependence Severity Scale (SDSS; Miele et al., 2000), and Diagnostic Interview Schedule (DIS-IV; Robins et al., 1995). Eligibility criteria for initial enrollment were: age 18–65, cocaine and opiate use (by self-report and urine screen), and physical dependence on opiates (by SDSS). Exclusion criteria were: current psychotic, bipolar, or major depressive disorders; current physical dependence on alcohol or sedatives; unstable serious medical illness; estimated IQ below 80, per the Shipley Institute of Living Scale (Zachary, 1986); and conditions precluding urine collection. Eligibility for data analyses included completion of at least 22 weeks in methadone treatment and a DSM interview at week 22. Participants were administered a DSM-IV-based tool, the Substance Dependence Severity Scale (SDSS), as the DSM-IV was current at the time of study start. Participants were also retroactively classified by DSM-5 criteria, because all the relevant information was collected during the course of the study (i.e. DSM-IV plus information on craving). Of 140 patients enrolled, 63 dropped out before completing 22 weeks and 7 did not complete the DSM interview. The remaining 70 were classified as treatment responders and nonresponders (details below). This study was approved by the Institutional Review Board of the NIDA Intramural Research Program; each participant gave written informed consent.

2.2. Standard Treatment

All participants received daily methadone and weekly individual counseling for up to 40 weeks, of which the last 10 weeks were a scheduled methadone dose taper (before and during which, our clinic staff helped participants transfer to community methadone clinics). In weekly individual-counseling sessions, counselors completed a semi-structured psychosocial assessment and treatment plan for each participant. Reduction of substance use was the primary goal. Methadone HCl (Mallinckrodt, Inc., St. Louis, MO) was administered orally in a fixed volume of 95 ml of cherry-flavored solution throughout the study. Dose was stabilized at 70 mg/day within seven days and continued for up to 28 weeks (target dose, 100 mg/day) (Kennedy, 2013).

2.3. Urine and breath testing

Mondays, Wednesdays, and Fridays, urine specimens were collected under observation. Testing was conducted with an Enzyme Multiplied Immunoassay Technique (EMIT; Syva Corp., Palo Alto, CA) system that provided qualitative results for cocaine (benzoylecgonine equivalents; BZE), opiates (morphine), marijuana, and benzodiazepines (oxazepam). Cutoffs were 300 ng/ml for cocaine, opiates, and benzodiazepines, and 50 ng/ml for marijuana. Breath alcohol level was determined with an Alco-Sensor III (Intoximeters, Inc., St. Louis, MO). Use of alcohol, benzodiazepines, and non-heroin opiates was rarely detected or reported; use of cannabis was detected in approximately 19% of urine screens.

2.3. Measures of treatment response

We defined treatment responders as individuals who 1) completed at least 22 weeks in the methadone program (retention in treatment), 2) provided ≥ 75% opioid-negative urine samples in the 30 days prior to week 22 (cessation of use), and 3) did not meet past-30-day criteria for DSM-IV opioid abuse or dependence based on the SDSS or DSM-5 opioid-use disorder (SDSS + craving) at week 22 (impairment and risk impact). Week 22 was used because it was the latest time point in the protocol with a reassessment and similar to the time frame of previous literature looking at treatment response (Weiss et al., 2011; Belding et al, 1998). Due to attrition, a follow-up analysis after Week 22 was not possible. Our main objective in defining a treatment responder was to have good clinical outcomes supported by both objective and subjective data. Criterion 2 above represents an objective measure that would be appreciated by a clinician caring for patients and parallels to the “treatment responder” definition used by Belding et al., 1998. Criterion 3 was created to assess which interventions played an integral part of progressing from a diagnosed diseased state to a healthy state. We wanted a standardized definition that allowed us to be confident the disease for treatment responders was well controlled. In the US, the most widely accepted methodology for mental health diagnoses is the DSM, whose current version was DSM-IV during our study and is DSM-5 now. Therefore, our definition of treatment responder uses standard definitions set by the DSM with added objective measures to assess progression from disorder to remission. Any participant who did not meet criteria for remission was categorized as a nonresponder. To retroactively assess DSM-5 criteria, which include craving, we used data from the SDSS plus a weekly questionnaire asking how much participants had “wanted” heroin or cocaine in the past week. The scale ranged from 0–4 with response anchors “not at all,” “a little,” “moderately,” “quite a bit,” and “extremely.” A DSM-5 craving score was calculated by taking the average of the completed questionnaires within the past 30 days of week 22, where an average past 30-day craving score > 0 was considered positive. The same procedures were completed for symptoms of cocaine-use disorders.

2.4. Predictor variables

Addiction Severity Index (ASI).

The ASI is a semistructured interview (Kosten et al., 1985; McLellan et al., 1992; McLellan et al., 1985) that all participants completed during pretreatment screening. It assesses seven potential problem areas: medical status, employment and support, drug use, alcohol use, legal status, family/social status, and psychiatric status. Within these categories, the specific questions evaluated in our study included employment for the last 3 years, number of times in treatment, and lifetime and recent history of heroin and cocaine use.

Social Adjustment Scale Self-Report (SAS-SR).

The SAS-SR is a questionnaire on quality of life; it was administered within the first week of admission. It has 54 items, most of which assess past-2-week social and employment status. Examples include time spent on hobbies and frequency of feeling bored.

Psychosocial Evaluation.

Within the first week of admission, participants underwent an in-depth psychosocial evaluation with a masters-level counselor. Counselors collected data on 13 variables, including the participant’s relationship with each immediate family member, family support available, family history of medical and addiction problems, relationship status, quantity and quality of friends and whether these friends used illicit substances, chronic disease status, age of first use of each substance tried, and any previous diagnoses such as PTSD or alcohol dependence.

2.5. Data Analysis

The distribution of each variable was evaluated by descriptive statistics. Although the ASI provides composite scores, we were interested in specific items, an approach like that used by Belding and colleagues (1998). To avoid small cells, we dichotomized marital status (separated, divorced, never married, and widowed) to married versus not married. Likert items from the SAS-SR were analyzed as continuous. For continuous measures with normal distributions, we performed student t-tests. Non-normally distributed variables were either log transformed and analyzed by a student t-test or grouped into natural categories and analyzed by categorical methods. For example, the ASI question, “How many days did you use heroin in the past 30 days?” revealed a left-skewed data set that was unable to normalize on log transformation. We then analyzed the variable by groups, not daily heroin users versus daily heroin users, which was defined as using heroin for 30 of the past 30 days. Categorical measures were evaluated by chi-square tests or, for those with low cell counts, Fisher’s exact tests.

We performed two multiple logistic regression models to investigate which variables were related to treatment response. To be included in a logistic model, a predictor had to have been bivariately associated with the outcome measure at a p-value < 0.20. The bivariate cutoff for the cocaine model was lowered to 0.15 because too many predictors were being retained. This method for including variables in our model is supported by Hosmer & Lemeshow, 2000. Before accepting a model, we ran tests for multicollinearity to ensure no predictors were highly intercorrelated. The alpha level was < 0.05, two-tailed.

3. RESULTS

3.1. Participant Characteristics

Participant characteristics are presented in Table 1. Overall our sample was 71% male, 41% White, and had an average age of 39.4 years (SD 7.9, range 21–53). Participants had a mean educational level of 11.9 years (SD 1.5, range 7–15). Prior to admission, the average heroin use was 10.9 years (SD 7.7, range 1–35) and average cocaine use was 4.1 years (SD 3.2, range 0–20). Fifty-two participants (74%) reported payment for work within the past 30 days.

Table 1.

Bivariate and multivariate predictors of opioid treatment response

Bivariate Multivariate
Characteristic Nonresp (n = 36) Resp (n = 34) P-value OR 95% CI P-value
Demographics
 Age (yrs), mean (SD) 39.1 (7.7) 39.6 (8.1) 0.790
 White, n (%) 17 (59) 12 (41) 0.311
 Male, n (%) 24 (48) 26 (52) 0.364
 Married, n (%) 1 (17) 5 (83) 0.102* 7.54 (0.64 – 89.5) 0.109
 Education (yrs), mean (SD) 11.8 (1.3) 11.9 (1.7) 0.775
 Paid for work past 30, n (%) 25 (48) 27 (52) 0.340
ASI
 Employed last 3 years, n, (%) 25 (44) 32 (56) 0.008* 8.07 (1.19 – 54.7) 0.033**
 Times in treatment, mean (SD) 2.4 (1.80) 1.4 (1.40) 0.021* 0.77 (0.52 – 1.15) 0.208
 Heroin
  Lifetime use (yrs), mean (SD) 11.8 (7.3) 9.8 (8.1) 0.280
  Daily users, n (%) 29 (49) 30 (51) 0.378
  IV use, n (%) 19 (50) 19 (50) 0.794
 Cocaine
  Lifetime use (yrs), mean (SD) 5.1 (3.2) 3.2 (3.1) 0.101* 0.35 (0.10 – 1.23) 0.100
  Used past 30 (days), mean (SD) 12.1 (9.3) 13.4 (8.6) 0.554
  IV use, n (%) 11 (55) 9 (45) 0.705
SAS-SR
 Time in hobbies, mean (SD) 3.00 (1.2) 2.65 (0.92) 0.178* 0.45 (0.23 – 0.88) 0.020**
 Feeling bored, mean (SD) 2.39 (1.1) 2.44 (1.0) 0.840
Psychosocial Eval
 First heroin use (yrs), mean (SD) 25.5 (6.5) 25.0 (7.0) 0.752
 First cocaine use (yrs), mean (SD) 21.9 (5.7) 24.6 (7.8) 0.111* 1.08 (0.98 – 1.20) 0.129
 Good father relation, n (%) 17 (56.7) 13 (43.3) 0.387
 Good mother relation, n (%) 28 (50.9) 27 (49.1) 0.952
 Family support, n (%) 30 (53) 27 (47) 0.663
 Partner illicit drug use, n (%) 12 (63) 7 (37) 0.203
 Hepatitis C, n (%) 15 (50) 15 (50) 0.916
 Illicit drug categories, mean (SD) 4.2 (1.2) 3.6 (0.95) 0.020* 0.64 (0.33 – 1.25) 0.193
Open in a new tab
*

P < 0.20 in bivariate χ2 and t tests, variable included in multiple logistic model.

**

P < 0.05 in multivariate model.

3.2. Treatment Response - Opioids

Out of 70 participants, 34 (49%) met our criteria for opioid treatment response. Responders and nonresponders tested positive for opioids in 7.1% (SD 8.2%) and 65% (SD 35%) of specimens collected in the last month of treatment, respectively. Responders had an average DSM-5 opioid severity of 0.47 (SD 0.51); nonresponders, 3.2 (SD 1.6). Responders had an average methadone dose of 84 mg/day and an average maximum methadone dose of 112 mg/day; nonresponders had an average methadone dose of 81 mg/day and an average maximum methadone dose of 119 mg/day.

The results of bivariate and multivariate analyses of opioid-response predictors are shown in Table 1.

3.2.1. Bivariate variables included in regression - Opioids

Seven variables met the criterion of having a bivariate p-value < 0.20. Opioid treatment responders were more likely to be married (χ2 = 3.17, df = 1 p = 0.102) and to have been employed for the past three years (χ2 = 7.0, df = 1, p = 0.008). They had shorter durations of lifetime cocaine use (t = 1.66, p = 0.101), fewer times in treatment (t = 2.37, p = 0.021), and spent less time in hobbies (t = 1.36, p = 0.178). In the psychosocial interviews, responders reported having started cocaine use at an older age (t = 1.62, p = 0.111) and had tried fewer categories of illicit drugs (t = 2.38, p = 0.020).

3.2.2. Multivariate predictors - Opioids

Of the 7 variables with bivariate p-values < 0.20, only 2 remained significant in our multivariate model. Participants who had been typically employed for the past 3 years were 8.07 times more likely to be treatment responders (95% CI 1.19–54.7). People spending more time on hobbies were 2.2 times less likely to be responders (95% CI 0.23–0.88). Based on the psychosocial evaluations conducted by counselors at the beginning of treatment, treatment responders were more likely to have highly active hobbies, e.g. hunting, fishing, or playing sports, whereas nonresponders were more likely to have passive hobbies, e.g. watching movies and television or listening to music.

3.3. Treatment Response - Cocaine

Out of 70 participants, 28 (40%) met our criteria for cocaine treatment response. Responders and nonresponders tested positive for cocaine in 3.8% (SD 6%) and 82% (SD 28%) of specimens collected in the last month of treatment, respectively. Responders had an average DSM-5 cocaine severity of 0.39 (SD 0.50); nonresponders, 4.2 (SD 2.5) Cocaine responders’ average methadone dose was 86 mg/day; cocaine nonresponders’ average methadone dose was 82 mg/day. Both cocaine responders and cocaine nonresponders had an average maximum methadone dose of 116 mg/day.

The results of bivariate and multivariate analyses of cocaine-response predictors are shown in Table 2.

Table 2.

Bivariate and multivariate predictors of cocaine treatment response

Bivariate Multivariate
Characteristic Nonresp (n = 42) Resp (n = 28) P-value OR 95% CI P-value
Demographics
 Age (yrs), mean (SD) 39.3 (8.0) 39.4 (7.9) 0.976
 White, n (%) 17 (59) 12 (41) 0.843
 Male, n (%) 27 (54) 23 (46) 0.105* 3.76 (0.72 – 19.5) 0.115
 Married, n (%) 3 (50) 3 (50) 0.677
 Education (yrs), mean (SD) 11.6 (1.6) 12.2 (1.5) 0.154
 Paid for work past 30, n (%) 30 (58) 22 (42) 0.503
ASI
 Employed last 3 years, n, (%) 33 (58) 24 (42) 0.452
 Times in treatment, mean (SD) 2.2 (1.62) 1.5 (1.69) 0.066* 0.60 (0.34 – 1.05) 0.076
 Heroin
  Lifetime use (yrs), mean (SD) 11.5 (8.1) 9.9 (7.2) 0.416
  Daily users, n (%) 38 (64) 21 (36) 0.102* 0.37 (0.06 – 2.17) 0.272
  IV use, n (%) 21 (55) 17 (45) 0.378
 Cocaine
  Lifetime use (yrs), mean (SD) 4.69 (3.1) 3.4 (3.2) 0.238
  Used past 30 (days), mean (SD) 12.5 (9.2) 13.2 (8.7) 0.729
  IV use, n (%) 11 (55) 9 (45) 0.589
SAS-SR
 Time in hobbies, mean (SD) 2.70 (1.1) 3.04 (1.1) 0.196
 Feeling bored, mean (SD) 2.64 (1.1) 2.1 (0.94) 0.028* 0.61 (0.30 – 1.24) 0.175
Psychosocial Eval
 First heroin use (yrs), mean (SD) 24.7 (0.94) 26.0 (7.6) 0.457
 First cocaine use (yrs), mean (SD) 23.2 (6.2) 23.2 (6.9) 0.998
 Good father relation, n (%) 14 (46.7) 16 (53.3) 0.059* 5.33 (1.17 – 24.5) 0.031**
 Good mother relation, n (%) 33 (60) 22 (40) 0.846
 Family support, n (%) 37 (65) 20 (35) 0.099* 0.27 (0.05 – 1.60) 0.150
 Partner illicit drug use, n (%) 14 (74) 5 (26) 0.140* 0.33 (0.07 – 1.61) 0.169
 Hepatitis C, n (%) 21 (70) 9 (30) 0.117* 0.15 (0.03 – 0.75) 0.021**
 Illicit drug categories, mean (SD) 4.1 (1.1) 3.6 (0.99) 0.045* 0.91 (0.46 – 1.80) 0.787
Open in a new tab
*

P < 0.15 in bivariate χ2 and t tests, variable included in multiple logistic model.

**

P < 0.05 in the multivariate model.

3.3.1. Bivariate variables included in regression - Cocaine

Nine variables met the criterion of having a bivariate p-value < 0.20. Cocaine treatment responders were more likely to be male (χ2 = 2.63, df = 1, p = 0.105) and were less likely to be daily users of heroin (χ2 = 3.038, df = 1, p = 0.081). Responders reported fewer times in treatment (t = 1.87, p = 0.066) and, in SAS-SR responses, reported lower frequencies of boredom in the past two weeks (t = 2.25, p = 0.028). They were less likely to have tested positive for hepatitis C (χ2 = 2.46, df = 1, p = 0.117).

In the psychosocial interviews, they were more likely to report having had good relationships with their fathers (χ2 = 3.58, df = 1, p = 0.059), yet less likely to report availability of family support (χ2 = 3.14, df = 1, p = 0.076). They were less likely to report having a partner who uses illicit drugs (χ2 = 2.21, df = 1, p = 0.137), and they had tried fewer categories of illicit drugs (t = 2.04, p = 0.045).

3.3.2. Multivariate predictors - Cocaine

Of the nine variables with bivariate p values < 0.20, only two remained significant in our multivariate model. Participants who reported good relationships with their fathers were 5.3 times more likely to be treatment responders (95% CI 1.17–24.5), and participants with hepatitis C were 6.75 times less likely to be responders (95% CI 0.029–0.747).

3.4. Treatment response – Opioids and Cocaine

By our study design, participants could be placed into 4 groups. Sixteen (23%) were solely opioid treatment responders, 10 (14%) were solely cocaine treatment responders, and 18 (26%) were both opioid and cocaine treatment responders. The remaining 26 (37%) were nonresponders for both opioids and cocaine.

4. DISCUSSION

4.1. Main Findings

Because the addiction literature does not provide a standard definition of treatment response, we developed a novel approach focusing on good clinical outcomes. Unique to our definition of treatment responder was that, in addition to accounting for retention in treatment and cessation of use, we incorporated DSM diagnostic criteria at baseline versus week 22 of treatment, which allowed us to include participants for whom the negative consequences of drug use were decreasing but who had not yet obtained abstinence. Our definition of treatment response integrated elements of 1) retention in treatment (enrolled at week 22), 2) cessation of use (≥ 75% drug-free urines), and 3) reduction in DSM symptomatology. We determined 22 weeks to be an appropriate length to evaluate outcome and maintain consistency with previous research (Belding 1998 and McLellan 1993 at 24 weeks each, Weiss 2011 at 16 weeks).

Once our comprehensive definition of treatment response was established, we then sought to identify some of its predictors. For example, retention in methadone treatment at 90 days has been predicted by female sex, greater baseline treatment readiness, and lower desire for help, whereas retention at 365 days may be predicted by higher baseline ASI medical composite, lower ASI legal composite, higher 3-month treatment-satisfaction scores, and higher methadone dose (Kelly et al., 2011). In a study by Roll et al., individuals who were employed at study enrollment were almost 14 times more likely to complete the treatment program, and those with a history of IV drug use were 5.5 times less likely to complete the program (Roll et al., 2005). Cessation of use during treatment for alcoholism and other substance-use disorders has been predicted by female sex, not losing a partner to separation, divorce, or death, not experiencing a decline in health, having any close friends supportive of recovery, and not having any close friends who encourage drug use (Satre et al., 2011). Additional predictors of drug use post-treatment include cocaine-positive urine samples at study intake (Peirce et al., 2009; Preston et al., 1998). In the studies cited here, treatment response was evaluated in terms of retention and verified abstinence. We added DSM criteria for substance-use disorders so our definition of treatment responder would more fully reflect participants’ clinical status and quality of life. The addition of DSM symptomatology to the definition of treatment responder provides insight beyond urine toxicology and retention as it explores how a patient’s day-to-day life is improved when treatment results in reduced use but not yet abstinence. This might be likened to using improved functioning on a patient activity scale as a marker of treatment response in a chronic condition such as rheumatoid arthritis in addition to little or no elevation in C reactive protein, a marker of inflammation as defined by the American College of Rheumatology and the European League Against Rheumatism (Felson et al., 2011).

Utilizing our comprehensive definition of treatment response, we had a mixture of expected and unexpected findings when we applied it to individuals with opioid and cocaine use. Expected findings included the positive association of past-three-year employment status with opioid treatment response. Previous research suggests that patients with a history of employment have better retention, higher abstinence, and are more likely to complete a treatment program (Magura et al., 2004; Roll et al., 2005). Our finding that time spent in hobbies predicted poorer opioid treatment response was not expected, but made more sense when we looked at the types of hobbies being reported: responders favored active hobbies such as hunting, fishing, and playing sports, while nonresponders favored passive hobbies such as watching movies and television and listening to music. This finding is consistent with results from a separate cohort of buprenorphine-maintained individuals in our clinic (Kowalczyk et al., 2015).

Based on our own previous work showing marriage as a protective factor (Heinz et al., 2009), we had expected marriage to predict positive opioid treatment response, but this did not hold true in our multivariate analysis. Our ability to assess that association in this dataset was limited by small cell sizes: only six of our participants were married (five of them were responders). Based on previous literature, we had also expected responders to have had more treatment episodes, but in our bivariate analysis the opposite was true (a finding that did not remain significant in the multivariate analysis).

There were no surprises in our bivariate analyses for predictors of cocaine treatment response. Our mixed findings on sex differences are consistent with prior literature (Poling et al., 2007; Satre et al., 2011). The results for number of prior treatment episodes were similar to what we found in our model for opioids. The bivariate association of past-two-week boredom with nonresponse was consistent with prior findings that boredom can trigger lapses and relapses (Marlatt and Gordon, 1985) and that boredom increases in the hours before cocaine-use episodes (Epstein and Preston, 2010). The high proportion of cocaine nonresponders with family support could reflect types of support that are more akin to enabling of drug use. Similarly, the high proportion of nonresponders with a partner using illicit drugs could reflect an influence on continued drug use both directly and indirectly if that partner was a source of family support. However, none of these variables was a significant predictor in the multivariate model.

The multivariate model showed that cocaine treatment response was more likely in participants who had reported good current or previous relationships with their fathers. Previous studies have shown that childhood adversities, including interpersonal loss (parental death, parental divorce, other parental loss), are associated with a substantial proportion of onsets of child/adolescent psychiatric disorders, including more than 40% of behavior-disorder onsets and one-third of substance-use-disorder onsets (McLaughlin et al., 2012). Roy et al. (2002) showed that in cocaine addiction, adverse childhood experiences, and particularly poor child–parent relationships, appear to negatively influence personality development and noted that emotional neglect during infancy predicts low adult levels of serotonin and dopamine metabolites.

The other multivariate predictor for cocaine treatment nonresponse was hepatitis C positivity. We and others have shown that a diagnosis of hepatitis C is associated with more risky behavior, including continued substance use (Murphy et al., 2015; Willner-Reid et al., 2008). We have also shown that people with substance-use disorders and hepatitis C have also been shown to experience more negative mood and to attribute their substance use to negative mood (Phillips et al., 2014).

4.2. Limitations

We note two important limitations. The first, which is not specific to our study, is that the field of addiction has no standard definition of treatment responders. We approached this problem by developing a definition of treatment response that was true to participants’ main goals (to cut down on substance misuse and resolve the problems associated with it) and consistent with what most clinicians would consider good outcomes (retention in treatment and negative urine toxicologies). A second limitation is inherent in our data set: we had no data beyond 22 weeks to see whether treatment response was maintained. Thus, our analyses provide a starting point, but need to be extended.

A minor limitation derives from the validity of asking sensitive questions about substance use, including our method for asking about craving. Although this limitation is inherent to self-report (Sayette et al., 2000), we minimized it by having trained professional maintain a nonjudgmental demeanor and by respondents with clear definitions (Del Boca & Noll, 2000). Another limitation is the small sample size of 70 participants that reduces the power of the study. Our sample size is similar to previous treatment response literature (Belding et al., 1998; Lehman et al., 1993) and likely secondary to high dropout rates in the field of addiction medicine.

4.3. Conclusion

We have sought to develop a comprehensive definition of treatment responder, integrating objective criteria such as retention in treatment and percent negative urine toxicologies with DSM symptomatology indicating improvement in important indicators of treatment response such as improved control of craving and use, decreased social impairment, decreased risky use, and diminished pharmacological indicators (tolerance and withdrawal). We then applied our definition of treatment response to a cohort of 70 individuals with opioid +/− cocaine use to determine which pretreatment factors might predict treatment response. Literature suggests that psychosocial interventions in addition to methadone treatment result in better outcomes in regards to compliance, completion of program, reduction of opioid use, and results at follow up (Amato et al., 2010; Amato et al., 2006). Our results build on these findings by identifying specific information from pretreatment assessments that might be considered for optimization of psychosocial intervention treatment approaches consistent with Precision Medicine initiatives. Our findings are correlational, but their degree of causality could be tested by using them for treatment matching. For patients with only an opioid-use disorder, treatment might focus on job assistance and on redirection toward more active types of hobbies. Patients with a cocaine-use disorder might benefit from information on how family relationships and chronic health conditions such as hepatitis C can influence behavior. Utilizing a comprehensive definition of treatment response and through the analyses of pretreatment psychological and psychosocial variables, we can optimize therapy and improve treatment response. With each identified treatment response predictor, we enhance our ability to personalize treatment and improve treatment response in keeping with Precision Medicine initiatives.

Acknowledgments

This research was supported by the National Institute on Drug Abuse, Intramural Research Program, National Institutes of Health.

Footnotes

The authors have no financial disclosures or conflicts of interest to report.

References

  1. Amato L, Minozzi S, Davoli M, Vecchi S, Ferri M, Mayet S. Psychosocial and pharmacological treatments versus treatments for opioid detoxification. Cochrane Database of Systematic Reviews 2010. [DOI] [PubMed]
  2. Amato L, Davoli M, Minozzi S, Vecchi S, Perucci C. Should psychosocial intervention be added to pharmacological treatment for opiate abuse/dependence? An overview of systematic reviews of the literature. Italian J. Public Health 2006; 3(2): 15–20. [Google Scholar]
  3. Ball JC, Lange WR, Myers CP, Friedman SR. Reducing the risk of AIDS through methadone maintenance treatment. J. Health Soc. Behav 1988; 29:214–226. [PubMed] [Google Scholar]
  4. Barthwell A, Senay E, Marks R, White R. Patients successfully maintained with methadone escaped human immunodeficiency virus infection. Arch. Gen. Psychiatry 1989; 46:957–958. [DOI] [PubMed] [Google Scholar]
  5. Belding MA, McLellan AT, Zanis DA, Incmikoski R. Characterizing “nonresponsive” methadone patients. J. Sub. Abuse Treat 1998; 15(6): 485–492. [DOI] [PubMed] [Google Scholar]
  6. Bell J, Mattick R, Hay A, Chan J, Hall W. Methadone maintenance and drug-related crime. J. Subst. Abuse 1997; 9:15–25. [DOI] [PubMed] [Google Scholar]
  7. Caplehorn J, Dalton M, Cluff M, Petrenas A. Retention in methadone maintenance and heroin addicts’ risk of death. Addiction 1994; 89:203–207. [DOI] [PubMed] [Google Scholar]
  8. Collins F and Varmus H. A New Initiative on Precision Medicine. N Engl J Med 2015; 372:793–795. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Epstein DH, Preston KL. Daily life hour by hour, with and without cocaine: an ecological momentary assessment study. Psychopharmacology 2010; 211: 223–232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Felson DT, Smolen JS, Wells G, Zhang B, van Tuyl LH, Funovits J, … Boers M. American College of Rheumatology/European League against Rheumatism provisional definition of remission in rheumatoid arthritis for clinical trials. Ann Rheum Dis 2011; 70:404–413. [DOI] [PubMed] [Google Scholar]
  11. Gerra G, Ferri M, Polidori E, Santoro G, Zaimovic A, Sternieri E. Long-term methadone maintenance effectiveness: psychosocial and pharmacological variables. J Sub. Abuse Treat 2003; 25: 1–8. [DOI] [PubMed] [Google Scholar]
  12. Gerstein D, Lewin L. Treating Drug Problems. New Engl. J. Med 1990; 323:844–848. [DOI] [PubMed] [Google Scholar]
  13. Goldstein A, Herrera J. Heroin addicts and methadone treatment in Albuquerque—a 22-year follow-up. Drug Alcohol Depend 1995; 40:139–150. [DOI] [PubMed] [Google Scholar]
  14. Heinz AJ, Wu J, Witkiewitz K, Epstein DH, Preston KL. Marriage and relationship closeness as predictors of cocaine and heroin use. Addict Behav 2009; 34: 258–263. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hilhouse M, Canamar CP, Ling W. Predictors of outcome after short-term stabilization with buprenorphine. J Sub. Abuse Treat 2009; 44: 336–342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hosmer D, & Lemeshow S. (2000). Model-building strategies and methods for logistic regression Applied Logistic Regression Second edition. (pp 91–135). New York, NY: John Wiley & Sons. [Google Scholar]
  17. Kelly SM, O’Grady KE, Mitchell SG, Brown BS, Schwartz RP. Predictors of methadone treatment retention from a multi-site study: A survival analysis. Drug Alcohol Depend 2011; 117: 170–175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kennedy A, Phillips KA, Epstein DH, Reamer D, Schmittner JP, and Preston KL. A randomized investigation of methadone doses at or over 100 mg/day, combined with contingency management. Drug and Alcohol Dependence 130(1–3):77–84, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kosten TR, Rounsaville BJ, & Kleber HD. (1985). Comparison of clinician ratings to self-reports of withdrawal during clonidine detoxification of opiate addicts. Am. J. Drug Alcohol Abuse. 1985; 11:1–10. [DOI] [PubMed] [Google Scholar]
  20. Kowalczyk WJ, Furnari MA, Phillips KA, Jobes ML, Ghitza U, Epstein DH, and Preston KL. Reducing the cost of free time: treatment success in a randomized trial of clonidine as adjunct to buprenorphine maintenance is associated with more leisure activities in the clonidine condition. 77th Annual Scientific Meeting of the College on Problems of Drug Dependence, Phoenix, AZ, 2015. [Google Scholar]
  21. Lehman F, Lauzon P, Amsel R. Methadone maintenance: Predictors of outcome in a Canadian milieu. J. Subst. Abuse Treat 1993; 10: 85–89. [DOI] [PubMed] [Google Scholar]
  22. Marlatt GA, Gordon JR. (Eds) (1985) Relapse Prevention: Maintenance Strategies in the Treatment of Addictive Behaviors (New York, Guilford Press; ). [Google Scholar]
  23. Magura S, Staines GL, Blankertz L, Madison EM. The effectiveness of vocational services for substance users in treatment. Subst Use Misuse 2004; 2165–213. [DOI] [PubMed]
  24. McLaughlin KA, Green JG, Gruber MJ, Sampson NA, Zaslavsky AM, Kessler RC. Childhood Adversities and First Onset of Psychiatric Disorders in a National Sample of US Adolescents. Author Affiliations: Division of General Pediatrics, Children’s Hospital Boston (Dr McLaughlin), and Department of Health Care Policy (Mr Gruber, Ms Sampson, and Drs Zaslavsky and Kessler), Harvard Medical School, and School of Education, Boston University (Dr Greif Green), Boston, Massachusetts. Arch Gen Psychiatry 2012;69(11):1151–1160. doi: 10.1001/archgenpsychiatry.2011.2277. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. McLellan AT, Alterman AI, Cacciola J, Metzger D, O’Brien CP. A new measure of substance abuse treatment: Initial studies of the Treatment Services Review. J Nerv Ment Dis 1992; 180:101–110. [DOI] [PubMed] [Google Scholar]
  26. McLellan AT, Arndt I, Metzger D, Woody G, O’Brien P. The effects of psychological services in substance abuse treatment. JAMA 1993; 269:1953–1959. [PubMed] [Google Scholar]
  27. McLellan AT, Luborsky L, Cacciola JS, Griffith J, Evans F, Barr HL, O’Brien CP. New data from the Addiction Severity Index: reliability and validity in three centers. J. Nerv. Ment. Dis 1985; 173:412–423. [DOI] [PubMed] [Google Scholar]
  28. Miele GM, Carpenter KM, Smith Cockerham M, Trautman KD, Blaine J, Hasin DS. Substance Dependence Severity Scale (SDSS): reliability and validity of a clinician-administered interview for DSM-IV substance use disorders. Drug Alcohol Depend 2000. April 1;59(1):63–75. [DOI] [PubMed] [Google Scholar]
  29. Murphy SM, Dweik D, McPherson S, Roll JM. Association between hepatitis C and opioid use while in buprenorphine treatment: preliminary findings. Am. J. Drug Alcohol Abuse 2015; 41: 88–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Peirce JM, Petry NM, Roll JM, Kolodner K, Krasnansky J, Stabile PQ, Brown C, Stitzer ML. Correlates of stimulant treatment outcome across treatment modalities. Am. J. Drug Alcohol Abuse 2009; 35: 48–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Phillips KA, Epstein DH, Vahabzadeh M, Mezghanni M, Lin J-L, and Preston KL. Substance use and hepatitis C – an Ecological Momentary Assessment study. Health Psychology, 2014. July;33(7):710–9. doi: 10.1037/hea0000087. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Poling J, Kosten T, Sofuoglu M. Treatment outcome predictors for cocaine dependence. Am. J. Drug Alcohol Abuse 2007; 33: 191–206. [DOI] [PubMed] [Google Scholar]
  33. Preston KL, Silverman K, Higgins ST, Brooner RK, Montoya ID, Schuster CR, Cone EJ. Cocaine use early in treatment predicts outcome in a behavioral treatment program. Journal of Consulting and Clinical Psychology 1998; 66(4):691–6. [DOI] [PubMed] [Google Scholar]
  34. Robins LN; Cottler LB; Bucholz KK; Compton WM III. The Diagnostic Interview Schedule, version IV Washington University; St. Louis, MO: 1995. [Google Scholar]
  35. Roll JM, Prendergast M, Richardson K, Burdon W, Ramirez A. Identifying predictors of treatment outcome in a drug court program. Am. J. Drug Alcohol Abuse 2005; 31: 641–656. [DOI] [PubMed] [Google Scholar]
  36. Rowell TL, Wu E, Hart CL, Haile R, El-Bassel N. Predictors of drug use in prison among incarcerated black men. Am. J. Drug Alcohol Abuse 2012; 38: 593–597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Roux P, Lions C, Michel L, Cohen J, Mora M, Marcellin F, Spire B, Morel A, Carrieri P, Karila L, ANRS Methaville study group. Predictors of non-adherence to methadone maintenance treatment in opioid dependent individuals: implications for clinicians. Curr Pharm Design 2014; 20: 4097–4105. [DOI] [PubMed] [Google Scholar]
  38. Roy A. Self-rated childhood emotional neglect and CSF monoamine indices in abstinent cocaine-abusing adults: possible implications for suicidal behavior. Psychiatry Research, 112 (1) (2002), pp. 69–75. [DOI] [PubMed] [Google Scholar]
  39. Satre DD, Chi FW, Mertens JR, Weisner CM. Effects of age and life transitions on alcohol and drug treatment outcome over nine years. J. Stud. Alcohol Drugs 2012; 73: 459–468. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Weiss RD, Potter JS, Fiellin DA, Byrne M, Connery HS, Dickinson W, Gardin J, Griffin ML, Gourevitch MN, Haller DL, Hasson AL, Huang Z, Jacobs P, Selzer J, Somoza EC, Sonne SC, Ling W. Adjunctive counseling during brief and extended buprenorphine-naloxone treatment for prescription opioid dependence. Arch. Gen. Psychiatry 2011; 68: 1238–1246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Weissman MM, Bothwell S. Assessment of social adjustment by patient self-report. Arch Gen Psychiatry 1976. September;33(9):1111–5. [DOI] [PubMed] [Google Scholar]
  42. Willner-Reid J, Belendiuk KA, Schmittner JP, Epstein DH, and Preston KL Hepatitis C and HIV risk behaviors in heroin/cocaine users. Journal of Substance Abuse Treatment, 35, 78–86, 2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Zachary R. Shipley Institute of Living Scale. Revised Manual Western Psychological Services; Los Angeles: 1986. [Google Scholar]

RESOURCES