Site no longer supporterd

This site is no longer supported and will not be updated with new content. You are welcome to browse and download all content already included in the site. Please note you will have to register your email address to access the site. We expect to close the site down by December 2016.

You are here

Depression outcome in alcohol dependent patients: An evaluation of the role of independent and substance-induced depression and other predictors

Journal of Affective Disorders 2015, Volume 174, pages 503–510

Editorial comment

The authors conducted a trial in which 138 outpatients with alcohol dependence and major depression participated. Patients were prescribed naltrexone and randomized to citalopram or placebo for 12 weeks, followed by a 12-week naturalistic outcome phase. Alcohol dependent patients whose depression was categorized as substance-induced showed greater improvement in depression during treatment compared to those with independent depression. However this appears to be mainly because they reduced their drinking more. The authors conclude that regardless of whether depression is believed to be independent or substance-induced, alcohol reduction should be prioritized as the first step in treating depression in all actively drinking, alcohol dependent patients

Abstract

Background

Depression commonly co-occurs with alcohol use disorders but predictors of depression treatment outcome in patients with both conditions are not well established.

Methods

Outpatients with alcohol dependence and major depression (n=138) were prescribed naltrexone and randomized to citalopram or placebo for 12 weeks, followed by a 12-week naturalistic outcome phase. General linear mixed models examined predictors of Montgomery Asberg Depression Rating Scale (MADRS) score over 24 weeks. Predictors included whether depression was independent or substance-induced, and demographic, alcohol use, and personality variables (Temperament and Character Inventory subscales).

Results

Most improvement in drinking and depression occurred between baseline and week 3. During follow-up, patients with substance-induced depression reduced their drinking more and they had better depression outcomes than those with independent depression. However, greater reduction in drinking was associated with better depression outcomes for both independent and substance-induced groups, while antidepressant therapy had no effect for either group.

Baseline demographic and alcohol use variables did not predict depression outcomes. Among personality variables, high self-directedness was a strong predictor of better depression outcomes.

Limitations

Subjects were not abstinent at baseline. The influence of naltrexone on depression outcomes could not be tested.

Conclusion

Alcohol dependent patients with substance-induced depression have better short term depression outcomes than those with independent depression, but this is largely because they reduce their drinking more during treatment.

Highlights

 

  • We examined predictors of six-month depression outcome in depressed, alcohol dependent patients.
  • Patients with substance-induced depression were compared to those with independent depression
  • Substance-induced depression patients had more improvement in both drinking and depression.
  • Reduced drinking improves depression even if it is categorized as independent.
  • High self-directedness strongly predicted better depression outcomes.

Keywords: Alcohol-induced disorder, Alcohol-related disorders, Depressive disorder, Depression, Treatment outcome, Comorbidity.

1. Introduction

Patients with alcohol use disorders often have high levels of depressive symptoms when they enter treatment. Depression on average improves substantially in the first six weeks of treatment and more gradually thereafter ( Pettinati et al., 2010 ). However a significant minority of patients experiences enduring depressive symptoms. It would be clinically useful to be able to identify this group prospectively, as this would help treatment allocation and might predict who is at risk of other outcomes such as suicide, ( Henriksson et al., 1993 ) alcohol relapse ( Hasin et al., 2002 ) and increased healthcare utilization ( Parthasarathy and Weisner, 2005 ).

Categorizing depression asindependentorsubstance-inducedat baseline is a common way of attempting to predict the course of depression in patients with an alcohol use disorder.Substance-induced depression(SID) by definition improves when patients stop drinking, and is thought to account for a substantial proportion of major depressive episodes among patients with an alcohol use disorder ( Schuckit et al., 2007 ). Converselyindependent depression(ID) should not necessarily remit with abstinence, but is believed to be more responsive to specific depression treatments such as antidepressant medication ( Pettinati, 2013 ).

The established method to differentiate between ID and SID is to examine the timing of onset of heavy drinking and depression in the past, and to determine whether depression was ever present during a sustained period of abstinence ( Schuckit et al., 2007 ). This approach has been operationalized in instruments such as the Structured Clinical Interview for DSM (SCID) ( Spitzer et al., 1992 ) and the Psychiatric Research Interview for Substance and Mental Disorders (PRISM) ( Hasin et al., 1996a ).

Despite the wide use of this terminology, the size of any difference in depression outcome between ID and SID has not been well characterized ( Nunes, 2006 ). Furthermore patients with SID appear to show a high incidence of ID when followed up over 12 months (Nunes, 2006 and Brown et al, 2011) suggesting SID may not be a stable diagnosis.

Little is known about more general predictors of depression outcome in patients with co-occurring alcohol dependence and major depression. In patients without an alcohol use disorder, early improvement in depression and some personality characteristics appear to be stronger predictors of 6-month outcome than either demographics or depression characteristics ( Mulder et al., 2006 ). However these findings have not been replicated in comorbid populations.

The objective of this study was to investigate whether patients with ID had a different treatment outcome compared to those with SID during 24 weeks of treatment for depression and alcohol dependence. We also examined a range of other candidate variables as predictors of depression outcome, including baseline depression and drinking measures, demographic variables and personality measures.

2. Methods

2.1. Overview of study design

Study methods and key results have been previously reported ( Adamson et al., in press ). In brief, this was a 12-week trial comparing naltrexone plus citalopram to naltrexone plus placebo followed by a 12-week naturalistic continuation phase. The trial was conducted in 8 outpatient alcohol treatment centers in New Zealand from 2007 to 2011. It received approval from local ethics committees at each site and was registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12606000413527).

2.2. Participants and measures

Screening was conducted on 474 potential subjects, of which 138 were included in the intent-to-treat sample. Eligible participants were outpatients aged 17–65 years seeking treatment for a drinking problem and low mood. They were recruited from addiction treatment centers, psychiatric clinics and via advertising. Participants met current DSM-IV criteria for alcohol dependence and major depressive episode in the past four weeks rated using the Structured Clinical Interview for DSM-IV (SCID), ( Spitzer et al., 1992 ). In addition they had a Montgomery Asberg Depression Rating Scale (MADRS) score of 20 or higher ( Montgomery and Asberg, 1979 ). Abstinence at baseline was encouraged but not enforced. Further baseline characteristics of the sample are shown in Table 1 .

Table 1 Baseline characteristics of sample for 24-week completers and dropouts.

  Range Value or % (sd)
Whole sample (n=138) Study completers to 24 weeks (n=106) Dropouts before 24 weeks (n=32) t or χ2; p
Demographics
Female   59.4% 60.4% 56.3% χ2=0.17, p=0.68
Age 20–64 43.6 (5.1) 43.9 (9.3) 42.5 (8.3) t=0.77, p=0.44
Employment          
 Full-time   34.1% 34.9% 31.3% χ2=1.30, p=0.52
 Part-time   21.0% 22.6% 15.6%
 Unemployed   44.9% 42.5% 53.1%
Relationship status          
 Single   31.9% 28.3% 43.8% χ2=2.85, p=0.24
 Married/de facto   46.4% 48.1% 40.6%
 Divorced/widowed   21.7% 23.6% 15.6%
Years of education   13.5 (3.1) 13.6 (3.1) 13.4 (3.2) t=0.24, p=0.81
 
Depression variables
MADRS score 20–48 31.0 (5.8) 31.1 (5.8) 30.6 (5.9) t=0.46, p=0.65
SCL-90 depression score 0.3–3.8 2.0 (0.7) 2.0 (0.7) 2.0 (0.8) t=−0.04, p=0.97
Independent depression   70.3% 67.0% 81.3% χ2=2.4, p=0.12
Assigned to citalopram   52.9% 50.9% 56.3% χ2=0.28, p=0.60
 
Alcohol variables
Percent days abstinent a 0–96% 25.8% (27.4) 25.2% (27.6) 27.9% (27.1) t=−0.48, p=0.63
Drinks per drinking day a 3.0–52.5 14.3 (8.0) 13.9 (7.3) 15.7 (9.9) t=−1.13, p=0.26
Readiness to change questionnaire 0–24 16.0 (5.5) 15.7 (5.4) 16.8 (5.5) t=−1.02, p=0.31
 
Comorbidity
Current tobacco smoker   57.2% 53.8% 68.8% χ2=2.25, p=0.13
Current cannabis disorder   15.9% 13.2% 25.0% χ2=2.55, p=0.11
Current anxiety disorder   47.8% 42.5% 65.6% χ2=5.29, p=0.02
Lifetime antisocial personality disorder   6.5% 5.7% 9.4% χ2=0.56, p=0.46
 
Temperament and character inventory subscales
Novelty seeking 2–20 10.7 (3.7) 10.7 (3.7) 10.9 (3.6) t=−0.27, p=0.79
Harm avoidance 2–19 11.4 (4.6) 11.6 (4.6) 10.6 (4.8) t=0.75, p=0.45
Reward dependence 3–20 11.7 (3.9) 11.7 (3.9) 11.9 (4.0) t=0.01, p=0.93
Persistence 0–20 10.9 (4.9) 10.9 (4.7) 11.2 (5.9) t=−0.23, p=0.82
Self directedness 1–20 10.1 (4.6) 10.2 (4.3) 9.7 (5.8) t=0.93, p=0.36
Cooperativeness 5–20 16.2 (3.2) 16.5 (3.1) 15.3 (3.5) t=1.9, p=0.06
Self transcendence 0–20 6.2 (5.2) 6.1 (5.2) 6.6 (5.2) t=−0.45, p=0.66

a Average in 12 weeks before treatment commenced.

Using an algorithm adapted from the SCID, the current episode of depression was assumed to be independent if at least one of the following criteria were met:

  • a) First episode of depression occurred before the onset of heavy drinking.
  • b) Past history of depression present for at least one month during a period of abstinence.
  • c) A past history of depressive symptoms substantially in excess of what would be expected given the duration or amount of alcohol use.

If none of the above criteria were met, the depression was labeled as substance-induced. The categorization as independent or substance-induced is, for convenience, referred to hereafter asdepression type. Participants were not explicitly informed whether they were categorized as having ID or SID, but their belief about the causal relationship between their drinking and depression was recorded at baseline.

Study exclusion criteria were: past or current intravenous drug use; current opioid use; psychosis; mania or hypomania; suicidality or homicidality; psychiatric hospitalization required; unstable medical illness; recent disulfiram, naltrexone, antidepressant or mood stabilizing medication; significantly elevated transaminases or raised bilirubin; pregnancy, breastfeeding or inadequate contraception; and current or pending imprisonment.

2.3. Interventions

The citalopram/placebo dose started at one capsule (20 mg) daily and it could be increased to three capsules during the trial if depression persisted. The naltrexone dose was 25 mg daily for one week, then 50 mg daily as tolerated. The dose could be further increased to 75 or 100 mg after 6 weeks. Citalopram/placebo and naltrexone were given for 12 weeks, after which both medications were scheduled to be withdrawn. However weeks 12–24 were designed as a naturalistic follow-up phase and any medications for mood or drinking could be freely prescribed following discussion between the patient and their treating clinicians. All participants received manualised clinical case management throughout the 24 weeks. This comprised: enhancing motivation and commitment to change drinking behavior; co-ordinating care; involving family and significant others; encouraging the development of coping strategies; promoting adherence; providing education about depression and alcohol dependence; and promoting involvement in self-help groups.

There were 19 research clinicians involved in the overall TEAM study, of which demographic data were available for 18. Clinicians were aged 34–61 (mean 48), 10/18 were female and 14/18 were of New Zealand European ethnicity. Clinicians had worked in the addiction field for between 9 and 30 years and had trained in psychiatry, psychiatric nursing, social work, clinical psychology or counseling.

2.4. Main outcome

The primary outcome was the MADRS score at 3-weekly intervals from week 3 to week 24.

2.5. Predictors

Alcohol measureswere percentage of days abstinent (PDA) and drinks per drinking day (DDD) obtained using the timeline follow back procedure ( Sobell, 1986 ), for the 12 weeks prior to baseline and then 3-weekly until 24 weeks.

Personality characteristicswere assessed using the 144-item Temperament and Character Inventory (TCI) ( Cloninger et al., 1993 ). This self-report instrument measures personality on four temperament dimensions: novelty seeking (NS), harm avoidance (HA), reward dependence (RD) and persistence (PS); and three character dimensions: cooperativeness (C), self-directedness (SD) and self-transcendence (ST).

Stage of changewas assessed with the Readiness to Change Questionnaire ( Rollnick et al., 1992 ). Alcohol addiction severitywas measured with the Leeds Dependence Questionnaire ( Raistrick et al., 1994 ).Therapeutic alliancewas assessed with the Working Alliance Inventory ( Horvath and Greenberg 1989 ) client and therapist versions, measured between weeks 3 and 6. This is a self-report instrument that assesses the quality of the relationship between therapist and client in relation to bond, tasks and goals. An independent research clinician administered the instruments. Clinicians and patients were blind to each other׳s responses.

2.6. Power calculation

The study was powered to be able to detect a clinically significant difference (effect size 0.5) in change in depression score from baseline to 12 weeks between citalopram and placebo. The study was also designed to detect a difference in citalopram effect according todepression type, but as the ratio of independent to substance-induced depression was not known a priori, a power calculation was not possible. The approximately 2.5–1 ratio of independent to substance-induced depression that eventuated provided less power for this subgroup effect (relative to the power for a main citalopram effect) because of unbalanced group sizes but this loss of power was mitigated through the use of repeated measures analyses for the present study ( Gueorguieva and Krystal, 2004 ).

2.7. Statistical methods

Analyses were conducted with general linear mixed models using PROC MIXED in SAS version 9.3 (SAS Institute Inc., Cary, NC). These methods have advantages over the last observation carried forward method in alcohol studies, as they allow more realistic handling of missing data ( Hallgren and Witkiewitz, 2013 ). Point estimates are least squares means from linear mixed models, expressed with 95% confidence intervals, unless otherwise stated.

There were three series of models fitted to the data: the first series comprised univariate analyses of baseline predictors; the second comprised multivariate analyses on baseline predictors identified as relevant from the first series; and the third investigated potential mediators of depression outcome during treatment. All models were repeated measures analyses using the MADRS scores in follow up (weeks 3, 6, 9, 12, 15, 18, 21 and 24) as the dependent variable and the baseline MADRS score, predictor(s) and predictor×time interaction term(s) as independent variables. The time variable was treated as continuous, and was centered due to the multilevel structure of models and the presence of interaction terms. There was no evidence of higher-order polynomial trends therefore these terms were omitted from models. An autoregressive covariance structure was specified and Kenward-Roger adjustment of degrees of freedom was applied for estimates of fixed effects. Model assumptions were checked using diagnostic plots and models were formulated on the assumption that data were missing at random. Model fit was assessed using the Bayesian Information Criterion.

In the first series of models, the effect of individual baseline predictors was separately examined in univariate models. The baseline predictors were grouped into demographic, depression, alcohol, comorbidity and personality (TCI) variables.

In the second series of models, baseline predictors that were significant at the 0.10 level in the first series were investigated by combining them in multivariate models. The first model in this series included TCI variables alone; the second added depression type and the final model dropped variables that were not significant at the 0.10 level in the previous models.

In the third series of models, dynamic mediators of outcome during treatment were investigated. These were: Working Alliance Inventory therapist and patient scores (WAI-T and WAI-P respectively, measured between weeks 3 and 6); adherence (number of treatment sessions attended); and drinking measures (PDA and DDD), modeled as time-varying covariates in linear mixed models. Finally baseline and mediator variables were combined in a full model. The purpose of this final model was to determine whether there was evidence the potential mediator variables lay on a causal pathway between selected baseline predictors (SDI and depression type) and depression outcome.

3. Results

3.1. Subject characteristics

Follow up data were available at 24 weeks for 106 subjects (77%). Baseline characteristics of the sample are shown in Table 1 , broken into 24-week completers and non-completers. As shown in Table 1 , non-completers had a significantly higher prevalence of comorbid anxiety disorders but did not differ from completers on other baseline measures.

3.2. Overall depression outcomes

The mean MADRS score was 31.0 (sd 5.8, range 20–48) at baseline. Estimates of the mean MADRS score were 16.1 (14.4, 17.8) at week 3, 10.9 (9.1, 12.7) at week 12 and 9.3 (7.5, 11.1) at week 24. At week 24, 15/106 subjects (14.2%) who were still in follow up had a MADRS score of more than 19, indicating at least moderate depression ( Snaith et al., 1986 ). There was no evidence of significant heterogeneity in depression response across treatment venues. As previously reported, there was no evidence that citalopram had an effect on depression outcome (Adamson et al., in press).

3.3. Drinking outcomes

As shown in Table 1 , at baseline subjects were abstinent for a mean of 25.8% of days (sd 27.4) and their mean daily alcohol consumption was 14.3 drinks (sd 8.0). The correlation between PDA and DDD was −0.20 at baseline and the correlation at each time point ranged from −0.27 to −0.43 in weeks 3–24.

Change in drinking occurred early in treatment: the mean PDA was 63.8% (sd 33.6) at week 3 and there was little change in this figure from week 3 to week 24. The pattern was similar for DDD: the mean DDD was 6.4 (sd 7.0) at week 3 and it remained close to this level between week 3 and week 24.

3.4. Predictors of depression outcome

Table 2 shows the results of univariate analyses of predictors of MADRS score at weeks 3–24, adjusted for baseline MADRS. In these models, baseline demographic, drinking and depression variables did not have a main effect on depression outcome, with the exception of depression type. No predictor×time interactions were significant, indicating the rate of change in depression in weeks 3–24 was not altered by any of the predictors.

Table 2 Univariate predictors of depression score during treatment a .

  b SE p
Treatment variables
Antidepressant assignment
 Citalopram 0.35 1.23 0.78
 
Demographic variables
Gender      
 Male 0.43 1.25 0.73
Age −0.06 0.07 0.36
Current employment category      
 Unemployed 1.54 1.39 0.53
 Part time 1.24 1.68  
 Full time (reference group) 0    
Relationship status      
 Single 0.40 1.67 0.13
 Married or de facto −2.20 1.54  
 Widowed, separated or divorced (reference group) 0    
Years of education −0.28 0.20 0.17
 
Depression variables
Baseline MADRS score 0.45 0.11 <0.001
Melancholic depression 1.99 1.51 0.19
Depression type (independent vs substance-induced) 3.74 1.29 0.004
Number of previous episodes      
 Fewer than 3 −2.60 1.99 0.60
 3–5 −0.40 1.48  
 5 or more 0.08 1.80  
 Too many or too indistinct to count (reference group) 0    
Duration of current episode      
 6 months or less −0.63 1.40 0.58
 6–12 months 1.22 1.73  
 More than 12 months (reference group) 0    
 
Alcohol variables
Baseline percent days abstinent 3.24 2.22 0.15
Baseline drinks per drinking day 0.01 0.08 0.88
Leeds dependence questionnaire score 0.06 0.10 0.58
Readiness to change questionnaire score 0.02 0.11 0.85
 
Comorbidity variables
Current tobacco smoker 0.77 1.23 0.53
Current cannabis disorder 1.55 1.70 0.36
Current anxiety disorder 1.17 1.24 0.35
Lifetime antisocial personality disorder 2.14 2.63 0.42
 
Temperament and character subscales
Novelty seeking −0.05 0.11 0.76
Harm avoidance 0.39 0.13 0.005
Reward dependence −0.09 0.16 0.58
Persistence −0.13 0.13 0.32
Cooperativeness −0.44 0.20 0.03
Self-directedness −0.61 0.13 <0.001
Self-transcendence 0.04 0.12 0.71

a Results from linear mixed models of predictors of MADRS scores at 3-weekly intervals from week 3 to week 24, adjusted for baseline MADRS. Models included predictor×time interaction terms, however no significant interaction effects were found therefore these results have been omitted. For predictors with two categories (eg gender), only the results for the non-reference group category are shown.

3.5. Depression type: independent and substance-induced depression

The percentage of patients who had ID was 70.3%, and the remaining 29.7% had SID. There was no significant difference in mean MADRS score at baseline, which was 31.3 (sd 5.8, range 20–48) for ID subjects and 30.2 (sd 5.6, range 20–45) for SID subjects (t=0.90p=0.31). Compared to subjects with SID, those with ID had higher self-directedness (t=−2.63,p=0.01), their current episode of depression was longer (t=2.03,p=0.04) and they had more years of education (t=2.35p=0.02). The two groups were comparable on all other baseline measures.

At baseline patients with SID (16/39; 41.0%) were more likely than those with ID (9/95; 9.5%) to believe drinking caused their depression while more patients with ID (66/95; 69.4%) than SID (17/39; 35.9%) believed drinking did not cause, but worsened their depression (χ2=20.2,p<0.001).

At week 24, ID study completers had a mean MADRS score of 10.3 (sd 9.4, range 0–43) and for SID completers the mean score was 7.0 (sd 7.9, range 0–29). As shown in Table 2 , there was an overall difference in depression outcome between the two groups but no evidence of a group×time interaction effect, indicating the difference between the two groups was already apparent by week 3, and the groups followed a parallel path thereafter. This pattern is also shown in Fig. 1 . The estimate for the mean difference in MADRS score between ID and SID during follow up was 3.9 (1.3, 6.5), a standardized mean difference of 0.68.

gr1

Fig. 1 MADRS score from baseline to week 24, for subjects with independent and substance-induced depression;pvalue=0.004 for the overall difference between groups in repeated measures analysis. Estimates of MADRS score at timepoints from week 3 to week 24 are least squares means from linear mixed models including time as a categorical predictor. Standardised mean difference 0.68 for group difference across all timepoints (excluding baseline) and 0.54 for group difference at week 24.

Subjects with SID had significantly higher PDA during follow up compared to ID subjects, after adjusting for baseline PDA (75.6% compared to 62.7%;p=0.007), but there was no difference in DDD between the two groups (p=0.10). The difference in PDA between the groups was already present at week 3, and persisted throughout the 24 weeks of follow up. There was a significant correlation between change in PDA and change in depression from baseline to week 3 for both ID subjects (r=0.285,p=0.005) and SID subjects (r=0.442,p=0.004).

3.6. Temperament and character measures

TCI data were missing for 11 subjects. TCI subscales were not significantly correlated with baseline depression or drinking measures, with two exceptions: SD was negatively correlated with baseline PDA (r=−0.226) while HA was negatively correlated with baseline DDD (r=−0.224). TCI subscales were not correlated with change in drinking during the study, with the exception of SD and change in DDD (r=0.232) and C and change in DDD (r=0.182).

For the temperament and character variables, low HA, high SD, and high C were univariate predictors of better depression outcome. The effect of SD is shown in Fig. 2 , which illustrates that the difference in outcome was most noticeable for subjects in the highest SD quartile, with the other 3 groups having a similar trajectory. The overall difference in MADRS score between the highest and lowest SD quartiles was 6.9 (3.5, 10.19; standardized mean difference 1.17) across all follow-up time-points.

gr2

Fig. 2 MADRS score from baseline to week 24 according to self-directedness quartile;pvalue<0.001 for overall group difference. Estimates of MADRS score at timepoints from week 3 to week 24 are least squares means from linear mixed models including time as a categorical predictor. Standardised mean difference 1.17 between quartile 1 (low self-directedness) and quartile 4 (high self-directedness) across all timepoints (excluding baseline) and 0.69 for difference between quartile 1 and quartile 4 at week 24.

3.7. Multivariate models of baseline predictors

Table 3 shows the results of the next step of the analyses, which was a series of multivariate models of baseline predictors that were significant at the 0.10 level in univariate models. In the final model, SD (p<0.001) and depression type (p<0.05) remained significant predictors of depression score. Time interaction effects were not significant for either variable.

Table 3 Multivariate models of baseline predictors of depression score during treatment a .

  Model 1: Temperament and character variables only BIC b =5460.3 Model 2: Temperament and character variables plus depression type BIC b =5454.6 Model 3: Best multivariate model BIC b =5423.4
b SE p b SE p b SE p
Harm avoidance 0.49 0.15 0.15 0.22 0.14 0.10 Not in model    
Cooperativeness −0.17 0.20 0.38 −0.18 0.20 0.36 Not in model    
Self−directedness −0.49 0.15 0.001 −0.41 0.15 0.01 −0.53 0.19 <0.001
Depression type c Not in model 2.62 1.32 0.05 2.47 1.26 0.05

a Results from linear mixed models of predictors of MADRS scores at 3−weekly intervals from week 3 to week 24, adjusted for baseline MADRS. Models included predictor×time interactions, however no significant interaction effects were found therefore these results have been omitted.

b Model fit, assessed using Bayesian Information Criterion.

c Independent or substance-induced depression.

3.8. Effect of mediator variables on change in depression

The effect of change in drinking on depression outcomes was initially explored by dividing drinkers into four categories according to the global change in PDA from baseline to week 24. The mean MADRS score at week 24 was 16.4 (11.4, 21.5) for subjects who did not increase their PDA during the study; 9.4 (5.6, 13.1) for subjects who increased their PDA but were still drinking >50% of days; 7.5 (4.7, 10.2) for those who increased their PDA and were still drinking but on <50% of days; and 8.3 (4.8, 11.9) for subjects who were abstinent at week 24. When depression type was added to this model, there was no evidence of an interaction effect of change in PDA×depression type. Illustrating this point, for patients with ID, the mean MADRS score at week 24 was 16.1 (10.4, 21.7) for those with no improvement in PDA and 9.9 (5.4, 14.4) for those who were abstinent. The corresponding figures for the SID group were 17.2 (6.7, 27.7) and 5.9 (0.1, 11.7) respectively, however these figures need to be viewed with caution as the standard errors are inflated due to small numbers in each cell.

These findings were further elucidated with models treating PDA and DDD as time-varying predictors. The results of these analyses are shown in Table 4 , along with results for therapeutic alliance and treatment adherence. In univariate models, DDD and PDA at weeks 3–24 (adjusted for their respective baseline values) and the therapist rating of therapeutic alliance (WAI-T) were significant predictors of depression outcome. In a model combining the two drinking measures and WAI-T, the drinking variables remained significant at the.001 level while therapeutic alliance was not significant (p=0.06). A final model added two baseline predictors- depression type and self-directedness- to the three mediators DDD, PDA and WAI-T. In this model depression type was no longer significant (p=0.25) suggesting differences in drinking patterns may have mediated the relationship between depression type and depression outcomes. SDI remained significant (p<0.001) suggesting it had an independent effect on depression outcome beyond any effect mediated via drinking.

Table 4 The role of mediator variables as predictors of MADRS score during follow up a .

  Univariate models of intermediate outcomes Multivariate model of intermediate outcomes. BIC b =4436.9 Combined model: baseline and intermediate predictors. BIC b =4303.3
  b SE p b SE p b SE p
Drinks per drinking day at weeks 3–24 0.40 0.10 <0.001 0.31 0.06 <0.001 0.29 0.06 <0.001
Percent days abstinent at weeks 3–24 −8.37 1.25 <0.001 −6.56 1.34 <0.001 −6.35 1.37 <0.001
Therapeutic alliance: patient rating −1.28 1.32 0.33 Not in models
Therapeutic alliance: clinician rating −2.49 1.15 0.03 −2.12 1.12 0.06 −1.50 1.10 0.18
Treatment adherence: number of sessions attended 0.01 0.08 0.78 Not in models
Self−directedness Not in models −0.53 0.14 <0.001
Depression type c Not in models 1.60 1.37 0.25

a Results from linear mixed models of predictors of MADRS scores during weeks 3–24 (repeated measures), adjusted for baseline MADRS. No significant predictor×time interactions were found in univariate models therefore interaction terms were dropped from multivariate models.

b Model fit, assessed using Bayesian Information Criterion.

c Independent or substance-induced depression.

4. Discussion

In this 24-week study of patients receiving comprehensive integrated treatment for alcohol dependence and depression, most of the change in depression and drinking occurred in the first three weeks of treatment. This pattern is consistent with other literature suggesting much of the reduction in symptom burden in comorbid patients occurs early in treatment, with as much as a third of the improvement even occurringbeforethe first treatment session ( Baker et al., 2014 ).

In the first three weeks, patients with substance-induced depression (SID) experienced more improvement in both depression and percentage of days abstinent than those with independent depression (ID). After that, the depression and drinking patterns of both groups followed approximately parallel paths from weeks 3 to 24, with alcohol outcomes remaining static and depression showing a small improvement over this period. The difference in depression outcomes between patients with ID and those with SID (an effect size of about 0.7) appeared to arise mainly because SID patients reduced their drinking more, rather than because they derived greater benefit from any reduction in drinking.

The greater increase in abstinent days among SID patients was an unexpected finding, as all patients regardless of depression type were encouraged to reduce their drinking. However, the assessment process did encourage patients to consider the relationship between their drinking and mood, and this may have provided a greater incentive for SID patients to address their drinking.

Although SID patients altered their drinking more, change in drinking was also associated with change in depression in ID patients; this runs counter to what would be expected if depression is truly “independent” of drinking in this group. For both groups of patients, a poor depression response was much more prevalent among subjects who did not improve their drinking at all or drank more; surprisingly, patients who achieved even modest improvement in drinking had similar depression outcomes to those who were abstinent at the end of the study.

The importance of categorizing depression as independent or substance-induced in patients with a concurrent substance disorder has been repeatedly emphasized (Schuckit, 1995, Schuckit et al, 1997, Grant et al, 2004, and Samet et al, 2013). In spite of this, few studies have prospectively compared the outcome of depression treatment between these groups ( Nunes, 2006 ) therefore little is known about the magnitude of any difference in outcome. The apparently high incidence of ID among patients originally diagnosed with SID (Brown et al, 2011 and Nunes, 2006) has raised further concerns about the face validity and predictive value of the concept as it is currently defined.

It is currently suggested that patients with ID benefit more from antidepressants and patients with SID benefit more from reduced drinking ( Schuckit, 2006 ). We found no evidence ID patients respond better to antidepressant medication, and limited support for the belief that reduced drinking benefits patients with SID more. In fact it seems likely reduced drinking benefits both groups. The limited antidepressant response among depressed, alcohol dependent patients coupled with evidence that the course of alcoholism affects the course of depression ( Hasin et al., 1996b ) suggests reduced drinking should be the primary treatment strategy for depression regardless of whether depression is believed to be independent or substance-induced.

We found baseline demographic and alcohol consumption variables, and depression variables other than depression type, did not predict the course of depression over 24 weeks. This is consistent with results from a study of depressed subjects without an alcohol disorder ( Mulder et al., 2006 ). In that study, Mulder et al. reported baseline demographic variables and depression characteristics did not generally predict depression outcome, while early depression response and personality variables did.

We identified three personality variables – high self-directedness, low harm avoidance and high cooperativeness – which were associated with better depression outcome in univariate models. Of these, only self-directedness remained significant in multivariate models. The effect of self-directedness was already apparent by week 3, after which group differences were maintained but did not increase. The findings for self-directedness are consistent with previous research in other populations. Low self-directedness is associated with poorer outcomes in non-comorbid depression populations (Sato et al, 1999 and Joyce et al, 2007) and worse 5-year outcome in women with bulimia nervosa ( Rowe et al., 2011 ). Self-directedness relates to an individual׳s ability to regulate and adapt their behavior to the environment ( Cloninger et al., 1993 ) and it appears to be a non-specific marker for personality disorders ( Cloninger, 2000 ). It also shares similarities with Bandura׳s conceptself-efficacy( Bandura, 1977 ), which features prominently in the alcohol literature ( Adamson et al., 2009 ). There are mixed views on whether personality disorders adversely influence treatment outcome in non-comorbid major depression (Newton-Howes et al, 2006 and Mulder, 2002). However the results from the present study suggest that personality variables are relevant to depression outcome in patients with an alcohol use disorder. This area has previously received little attention and warrants further study.

This study׳s strengths include its relatively high retention rate of 77% at 24 weeks, and its inclusion of several measures that have not been extensively reported in this population, particularly temperament and character. A potential limitation is that subjects were not required to be abstinent at baseline. However this reflects real-life outpatient practice whereby not all patients are able to immediately stop drinking. Recruitment for the study stopped before reaching the target sample size, reducing power for the main study outcomes. However in the present analyses we were able to mitigate the loss of power issue by using repeated measures analyses on up to nine time-points per subject. Lastly the potential influence of naltrexone on both drinking and depression outcomes in this study was not known, although it is now suggested naltrexone has little effect on depressive symptoms ( Petrakis et al., 2007 ).

In conclusion, alcohol dependent patients whose depression is categorized as substance-induced show greater improvement in depression during treatment compared to those with independent depression. However this appears to be mainly because they reduce their drinking more, possibly because of greater recognition of the benefit this will provide for their depression. Furthermore although patients with independent depression should not theoretically experience much, if any, improvement in depressive symptoms as a direct result of curtailing their drinking, this appears not to be the case. Our view is that regardless of whether depression is believed to be independent or substance-induced, alcohol reduction should be prioritized as the first step in treating depression in all actively drinking, alcohol dependent patients.

Role of funding source

The Health Research Council of New Zealand providing funding for the conduct of the clinical trial on which this research is based.

Conflict of interest

The authors have no potential conflicts of interest to declare.

Acknowledgments

This study has been funded by Grants from the New Zealand Ministry of Health and New Zealand Health Research Council (Grant HRC 07/138).

References

  • Adamson et al., 2015 S.J. Adamson, J.D. Sellman, J.A. Foulds, C. Crampton, D. Deering, A. Dunn, J. Berks, L. Nixon, G. Cape. A randomised trial of combined citalopram and naltrexone for non-abstinent outpatients with co-occurring alcohol dependence and major depression. J. Clin. Psychopharmacol.. 2015; (in press)
  • Adamson et al., 2009 S.J. Adamson, J.D. Sellman, C.M.A. Frampton. Patient predictors of alcohol treatment outcome: a systematic review. J. Subst. Abuse Treat.. 2009;36:75-86 Crossref
  • Baker et al., 2014 A.L. Baker, D.J. Kavanagh, F.J. Kay-Lambkin, S.A. Hunt, T.J. Lewin, V.J. Carr, P. Mcelduff. Randomized controlled trial of MICBT for co-existing alcohol misuse and depression: outcomes to 36-months. J. Subst. Abuse Treat.. 2014;46:281-290 Crossref
  • Bandura, 1977 A. Bandura. Self-efficacy: toward a unifying theory of behavioral change. Psychol. Rev.. 1977;84:191 Crossref
  • Brown et al., 2011 R.A. Brown, S.E. Ramsey, C.W. Kahler, K.M. Palm, P.M. Monti, D. Abrams, M. Dubreuil, A. Gordon, I.W. Miller. A randomized controlled trial of cognitive-behavioral treatment for depression versus relaxation training for alcohol-dependent individuals with elevated depressive symptoms. J. Stud. Alcohol Drugs. 2011;72:286
  • Cloninger, 2000 C.R. Cloninger. A practical way to diagnosis personality disorder: a proposal. J. Personal. Disord.. 2000;14:99-108 Crossref
  • Cloninger et al., 1993 C.R. Cloninger, D.M. Svrakic, T.R. Przybeck. A psychobiological model of temperament and character. Arch. Gen. Psychiatry. 1993;50:975 Crossref
  • Grant et al., 2004 B.F. Grant, F.S. Stinson, D.A. Dawson, S.P. Chou, M.C. Dufour, W. Compton, R.P. Pickering, K. Kaplan. Prevalence and co-occurrence of substance use disorders and independent mood and anxiety disorders: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Arch. Gen. Psychiatry. 2004;61:807-816 Crossref
  • Gueorguieva and Krystal, 2004 R. Gueorguieva, J.H. Krystal. Move over anova: progress in analyzing repeated-measures data andits reflection in papers published in the archives of general psychiatry. Arch. Gen. Psychiatry. 2004;61:310-317 Crossref
  • Hallgren and Witkiewitz, 2013 K.A. Hallgren, K. Witkiewitz. Missing data in alcohol clinical trials: a comparison of methods. Alcohol Clin. Exp. Res.. 2013;37:2152-2160 Crossref
  • Hasin et al., 2002 D. Hasin, X. Liu, E. Nunes, S. Mccloud, S. Samet, J. Endicott. Effects of major depression on remission and relapse of substance Ddependence. Arch. Gen. Psychiatry. 2002;59:375-380 Crossref
  • Hasin et al., 1996a D.S. Hasin, K.D. Trautman, G.M. Miele, S. Samet, M. Smith, J. Endicott. Psychiatric Research Interview for Substance and Mental Disorders (PRISM): reliability for substance abusers. Am. J. Psychiatry. 1996;153:1195-1201
  • Hasin et al., 1996b D.S. Hasin, W.-Y. Tsai, J. Endicott, T.I. Mueller, W. Coryell, M. Keller. Five-year course of major depression: effects of comorbid alcoholism. J. Affect. Disord.. 1996;41:63-70 Crossref
  • Henriksson et al., 1993 M.M. Henriksson, H.M. Aro, M.J. Marttunen, M.E. Heikkinen, E.T. Isometsa, K.I. Kuoppasalmi, J.K. Lonnqvist. Mental disorders and comorbidity in suicide. Am. J. Psychiatry. 1993;150:935-940
  • Horvath and Greenberg, 1989 A.O. Horvath, L.S. Greenberg. Development and validation of the Working Alliance Inventory. J. Couns. Psychol.. 1989;36:223 Crossref
  • Joyce et al., 2007 P.R. Joyce, J.M. Mckenzie, J.D. Carter, A.M. Rae, S.E. Luty, C.M. Frampton, R.T. Mulder. Temperament, character and personality disorders as predictors of response to interpersonal psychotherapy and cognitive-behavioural therapy for depression. Br. J. Psychiatry. 2007;190:503-508 Crossref
  • Montgomery and Asberg, 1979 S.A. Montgomery, M. Asberg. A new depression scale designed to be sensitive to change. Br. J. Psychiatry. 1979;134:382-389 Crossref
  • Mulder, 2002 R.T. Mulder. Personality pathology and treatment outcome in major depression: a review. Am. J. Psychiatry. 2002;159:359-371 Crossref
  • Mulder et al., 2006 R.T. Mulder, P.R. Joyce, C.M. Frampton, S.E. Luty, P.F. Sullivan. Six months of treatment for depression: outcome and predictors of the course of illness. Am. J. Psychiatry. 2006;163:95-100 Crossref
  • Newton-Howes et al., 2006 G. Newton-Howes, P. Tyrer, T. Johnson. Personality disorder and the outcome of depression: meta-analysis of published studies. Br. J. Psychiatry. 2006;188:13-20 Crossref
  • Nunes, 2006 E.V. Nunes. Independent versus substance-induced major depressive disorder in substance-dependent patients: observational study of course during follow-up. J. Clin. Psychiatry. 2006;67:1561 Crossref
  • Parthasarathy and Weisner, 2005 S. Parthasarathy, C.M. Weisner. Five-year trajectories of health care utilization and cost in a drug and alcohol treatment sample. Drug Alcohol Depend.. 2005;80:231-240 Crossref
  • Petrakis et al., 2007 I. Petrakis, E. Ralevski, C. Nich, C. Levinson, K. Carroll, J. Poling, B. Rounsaville. Naltrexone and disulfiram in patients with alcohol dependence and current depression. J. Clin. Psychopharmacol.. 2007;27:160-165 Crossref
  • Pettinati, 2013 H.M. Pettinati. Current status of co-occurring mood and substance use disorders: a new therapeutic target. Am. J. Psychiatry. 2013;170:23 Crossref
  • Pettinati et al., 2009 H.M. Pettinati, D.W. Oslin, K.M. Kampman, W.D. Dundon, H. XIE, T.L. Gallis, C.A. Dackis, C.P. O’Brien. A double-blind, placebo-controlled trial combining Sertraline and Naltrexone for treating co-occurring depression and alcohol dependence. Am. J. Psychiatry. 2009;167:668-675
  • Raistrick et al., 1994 D. Raistrick, B. Jackie, G. Tober, J. Weiner, J. Allison, C. Healey. Development of the Leeds Dependence Questionnaire (LDQ): a questionnaire to measure alcohol and opiate dependence in the context of a treatment evaluation package. Addiction. 1994;89:563-572 Crossref
  • Rollnick et al., 1992 S. Rollnick, N. Heather, R. Gold, W. Hall. Development of a short ‘readiness to change’ questionnaire for use in brief, opportunistic interventions among excessive drinkers. Br. J. Addict.. 1992;87:743-754 Crossref
  • Rowe et al., 2011 S. Rowe, J. Jordan, V. Mcintosh, F. Carter, C. Frampton, C. Bulik, P. Joyce. Dimensional measures of personality as a predictor of outcome at 5-year follow-up in women with bulimia nervosa. Psychiatry Res.. 2011;185:414-420 Crossref
  • Samet et al., 2013 S. Samet, M.C. Fenton, E. Nunes, E. Greenstein, E. Aharonovich, D. Hasin. Effects of independent and substance‐induced major depressive disorder on remission and relapse of alcohol, cocaine and heroin dependence. Addiction. 2013;108:115-123 Crossref
  • Sato et al., 1999 T. Sato, S. Hirano, T. Narita, K. Kusunoki, J. Kato, M. Goto, K. Sakado, T. Uehara. Temperament and character inventory dimensions as a predictor of response to antidepressant treatment in major depression. J. Affect. Disord.. 1999;56:153-161 Crossref
  • Schuckit et al., 1997 M. Schuckit, J. Tipp, M. Bergman, W. Reich, V. Hesselbrock, T. Smith. Comparison of induced and independent major depressive disorders in 2945 alcoholics. Am. J. Psychiatry. 1997;154:948-957
  • Schuckit, 1995 M.A. Schuckit. Alcoholism and affective disorder: clinical course of depressive symptoms. Am. J. Psychiatry. 1995;152:45
  • Schuckit, 2006 M.A. Schuckit. Comorbidity between substance use disorders and psychiatric conditions. Addiction. 2006;101:76-88 Crossref
  • Schuckit et al., 2007 M.A. Schuckit, T.L. Smith, G.P. Danko, J. Pierson, R. Trim, J.I. Nurnberger Jr., J. Kramer, S. Kuperman, L.J. Bierut, V. Hesselbrock. A comparison of factors associated with substance-induced versus independent depressions. J. Stud. Alcohol Drugs. 2007;(68):805-808
  • Snaith et al., 1986 R. Snaith, F. Harrop, D. Newby, C. Teale. Grade scores of the Montgomery–Åsberg Depression and the Clinical Anxiety Scales. Br. J. Psychiatry. 1986;148:599-601 Crossref
  • Sobell, 1986 M.B. Sobell. The reliability of a timeline method for assessing normal drinker college students׳ recent drinking history: utility for alcohol research. Addict. Behav.. 1986;11:149 Crossref
  • Spitzer et al., 1992 R.L. Spitzer, J.B.W. Williams, M. Gibbon, M.B. First. The Structured Clinical Interview for DSM-III-R (SCID) I: history, rationale, and description. Arch. Gen. Psychiatry. 1992;49:624-629 Crossref

Footnotes

a Department of Psychological Medicine, University of Otago, Christchurch, New Zealand

b Christchurch Health and Development Study, University of Otago, Christchurch, New Zealand

lowast Correspondence to : Department of Psychological Medicine, University of Otago, Christchurch, PO Box 4345, Christchurch 8140 New Zealand. Tel.: +64 3 3720 400; fax: +64 3 3720 407.


Related content

Relationships among independent major depressions, alcohol use, and other substance use and related problems over 30 years in 397 families

Editorial comment
There is ample evidence of important relationships between heavy drinking and mood changes as well as between...

Depression outcome in alcohol dependent patients: An evaluation of the role of independent and substance-induced depression and other predictors

Editorial comment

The authors conducted a trial in which 138 outpatients with alcohol dependence and major depression participated. Patients were...

Randomized controlled trial of MICBT for co-existing alcohol misuse and depression: Outcomes to 36-months

Editorial comment
This trial evaluates changes over 36 months following randomization of 284 outpatients to one of four motivational...

Investigating biases of attention and memory for alcohol-related and negative words in alcohol-dependents with and without major depression after day-clinic treatment

Editorial comment
Impaired cognitive control in alcohol use disorders contributes to the development and maintenance of addictive behaviour, and...

Brief Report: Excessive alcohol use negatively affects the course of adolescent depression: One year naturalistic follow-up study

Editorial comment
In this brief report, the authors present findings from a study conducted in Finland on the impact...

Featured presentations

Share

Notify your friends and contacts about this Progress in Mind Resource Centre, with free access to excellent content.

 

 

 

Suggestions

Please tell us what you like us to cover on the Resource Centre – we welcome your suggestions and comments

Search this site

Search form