Dysthymia

INTRODUCTION

Depressive disorders are common among opiate abusers with rates ranging from 14% to 55% (1-3). The impact of depression on opiate users is extensive. Comorbid depression may be associated with premature drop-out from treatment, may lead to continued drug use or drug relapse, and has been associated with poorer prognosis after drug treatment (2, 4-9).

Injection drug users constitute the major risk group for HIV infection in many countries. We have previously documented that high levels of depressive symptoms are associated with increased injection-related risk behaviors which place them at risk of HIV infection (8). Specifically, depression severity was significantly associated with injection-related HIV risk (odds ratio 1.5; 95% CI 1.1-2.3) after controlling for injection frequency, cocaine use, and demographic variables. Depression also has been implicated as a risk factor for needle sharing among drug injectors in treatment for substance abuse (10, 11). Possible explanations for these findings include that depression may increase HIV risk by affecting drug users' sense of fatalism, with hopelessness overwhelming the perceived benefits of behavior change; depression may lead to increased drug use, secondarily increasing HIV risk; depression may also affect risk avoidance planning as injectors need to acquire clean needles before the time of injection.

Based on the aforementioned association, we hypothesized that successful treatment of depression would reduce HIV drug risk behaviors. We designed an intervention targeted at a high risk subpopulation of depressed active injectors--the MINERVA (Multidisciplinary Intervention for Needle Users to Reduce Viral Acquisition) study. In MINERVA, we recruited injection drug users for a study testing the efficacy of combined cognitive behavior therapy and pharmacotherapy for the treatment of depression. We previously reported that overall depression remission rates were 19.2%, 17.0%, and 29.9% at 3-, 6-, and 9-month follow-ups, respectively (12). Significant treatment effects were observed at 3-months when 26.2% of participants randomized to treatment were in remission, compared with about 12% of controls. Highly adherent participants were estimated to be about 5.96 (p=006) times more likely than controls and 4.04 (p=059) times more likely than participants with low adherence to treatment to be in depression remission at 3-months; however this short-term effect was not observed at 6 and 9 months (13).

Our primary objective in the current analysis was to determine if treatment of the depression within the MINERVA protocol significantly reduced HIV drug risk behaviors compared to an assessment-only condition in out-of-treatment drug injectors. Our secondary objective was to examine the relationship of injection-related risk and depressive symptom improvement over a nine-month period.

METHODS

Description of the Study Site and Participant Population

Between March 2000 and June 2002 we recruited participants in Rhode Island for a study of combined psychotherapy (cognitive behavioral therapy) and pharmacotherapy for the treatment of depression in out-of-treatment drug injectors. Combination treatment was selected because it may be the most powerful intervention for depression (14, 15). We advertised the study as a "Quality of Life" research project for drug users with a financial incentive at various community agencies, through newspaper ads and on the street with flyers. Those interested were directed to call the study phone number to be screened by study research assistants (RA) who inquired about the majority of the inclusion criteria (13). If eligible after a phone screening, which included demographics, recent drug use, and needle sharing behaviors, and nonuse of prescribed psychotropic medications, individuals were invited to come to the research site at Rhode Island Hospital for a more detailed assessment (12).

Inclusion criteria were:

1) age between 18 and 70 years;

2) injection of opiate or cocaine during the prior 30 days;

3) needle sharing during the prior 30 days;

4) meeting criteria for a DSM-IV diagnosis (SCID-P) (16) of major depression, dysthymia, substance-induced mood disorder with depressive symptoms persisting at least 3 months, or major depression plus dysthymia;

5) having a scale score of > 13 on the 17-item Modified Hamilton Rating Scale for Depression (MHRSD) (17);

6) not currently enrolled in drug abuse treatment (methadone, residential, outpatient);

7) absence of current or past diagnosis of bipolar disorder, schizophrenia, schizo-affective disorder, schizophreniform disorder, or delusional disorder;

8) able to speak and read English sufficiently to complete the procedures of the study;

9) having a verifiable mailing address;

10) providing contact names for follow-up assessments; and

11) providing written informed consent for this study which was approved by the Rhode Island Hospital Institutional Review Board.

At the baseline visit, participants underwent urine toxicological testing for cocaine metabolites and opiates to confirm active drug use status, and women received urine pregnancy testing. Participants were then administered a 90-minute structured interview that included sections on demographics, SCID-P mood, psychotic screen modules, drug and alcohol modules, and HIV risk behaviors. At the conclusion of the interview, participants were escorted to the blood lab for HIV antibody testing.

Based on the phone screen, 161 individuals came to the study office for a full screening interview; 52 did not meet full eligibility criteria. One hundred nine persons were eligible for the study and enrolled, with 53 randomly assigned to the combined therapy group and 56 to the assessment-only control group. Participants were given a 2-week follow-up appointment to receive the results of their HIV tests (including posttest counseling) and 3-, 6- and 9-month follow-up appointments for face-to-face research assessments by study staff blind to treatment group assignment. At the 9-month assessment, repeat HIV testing was performed. At each assessment visit participants also received any medical or mental health referrals requested. Control group participants received neither psychotherapy nor pharmacotherapy from study personnel.

Participants assigned to the intervention group were scheduled to receive a total of 8 individual psychotherapy (CBT) visits (over the course of 3 months) and 3 pharmacotherapy visits (monthly over the course of 3 months). Treatment group participants were scheduled for their first psychotherapy and pharmacotherapy appointment within a week of the baseline assessment. After this acute, combined 3-month phase, participants were offered continuation pharmacotherapy alone for study months 3 to 6. Participants wanting to continue medications beyond 6 months were referred for treatment.

Further details of the assessment protocol have been described previously (12).

Treatment

Cognitive-Behavioral Therapy

This treatment was aimed at reducing the participant's level of depression and increasing his/her coping skills regarding depression. This 8-session therapy was a modified version of a "Coping with Depression" protocol treatment intervention used by Brown et al. (18) for treatment of depression in alcoholic patients. The first sessions were devoted to presenting a social learning view of depression and guiding the participant in learning how to identify and differentiate mood states. The next sessions addressed acquiring skills in three specific areas: increasing pleasant activities, changing negative cognitions, and improving social skills/increasing positive social interactions (19). Each participant was encouraged to develop a personalized plan to work on problem areas using a participant workbook. Each session consisted of a review of material from the previous session, presentation of new material, discussion and exercises related to the new material, and a homework assignment. No sessions were specifically devoted to cessation of drug use. The treatment was administered in 60-minute individual sessions by Ph.D.-level clinical psychologists trained and supervised in the protocol by two of the co-investigators (SR, DH) at weekly trainings.

Pharmacotherapy

All intervention group participants were scheduled to meet with a psychiatrist monthly over the course of 3 months. Whenever possible, psychiatric visits took place immediately after CBT sessions. Psychiatrists followed the clinical management guidelines developed by Fawcett and colleagues (20). Participants were first prescribed citalopram 20 mg. The dose and dose schedule was determined by the clinical judgment of the treating psychiatrist (up to 80 mg) at the monthly visits. If citalopram was not effective or produced adverse effects, the psychiatrist could prescribe venlafaxine, 37.5 mg (up to 375 mg), or bupropion, 150 mg (up to 300 mg). Once a satisfactory response was achieved with any of these medications, participants were maintained on that dose for the remainder of the acute combined treatment period, that is, up to 3 months. At each medication visit, participants received 1 month of study medication. Medication study visits were limited to 15 minutes to minimize the possibility that the study psychiatrist was providing other active psychotherapy ingredients to the treatment group.

Confirmation of study medication use was assessed by pill self-report. Participants who reported citalopram use at the 3-month assessment had serum levels verifying self-report, and all were concordant with self-report.

CBT Monitoring

The training for the two Ph.D.-level clinical psychologists began with a study of the project CBT manual. They also received two hours of didactic training in HIV risk reduction, injection behaviors, and local substance abuse treatment programs. Therapist manuals developed for the project were used in both initial training and continuing supervision. Each therapist then performed the intervention with three pilot training cases under the supervision of Dr. Ramsey (Brown University School of Medicine) prior to study initiation. Monitoring adherence to the intervention protocol was accomplished by the therapists completing a checklist after each intervention session. All intervention sessions were audio taped, and 10% were randomly reviewed by the therapy supervisor for internal quality control to prevent deviation from the protocol.

Measures

We used the drug-risk component of the RAB to assess drug risk behaviors at baseline and at each follow-up assessment (21). This measure was comprised of 8 items; the scoring protocol generates scale scores ranging from 0 to 1.0 with high scores indicating relatively higher levels of HIV drug-risk behavior. The observed distributions at follow-ups were distinctly non-normal. We used the LADDER and GLADDER procedures as implemented in Stata Version 7 to identify possible transformations. Before using these procedures, we reexpressed the scales to range from 0 to 100 (multiplied by 100) and arbitrarily added 1 (since many transformations are undefined at 0). At all time periods, a square-root transformation provided the best approximation to the normal distribution. The potential range of scores on the transformed scale is 0 to 10.05. The statistical tests reported were based on the transformed RAB HIV drug-risk scale. It should be noted that parallel models estimated using the untransformed measure yielded completely comparable statistical results. For descriptive purposes, we have also presented the proportion of participants reporting no drug-related HIV risk behaviors (RAB drug-risk scale score was 0 at follow-up).

The Modified Hamilton Rating Scale for Depression (MHRSD) scale was used to assess depression severity at both baseline and follow-up (17). Depression remission was defined as a MHRSD score [less than or equal to] 8 (22). We used the SCID-P to define depression type. Participants were classified as meeting symptomatic criteria for a DSM-IV diagnosis of major depression, dysthymia, "substance-induced mood disorder" with depressive symptoms persisting for at least the last 3 months, or major depression plus dysthymia. For multivariate analyses we included the two participants diagnosed with dysthymia in the major depression plus dysthymia group.

Measures of background characteristics included years of age and education, gender, and race. At baseline and at each follow-up, participants were asked to report the number of days on which they had used heroin and the number of days on which they had used cocaine, in the 30-days prior to the interview according to the Addiction Severity Index (23).

In another report, we had described participants' adherence to the treatment protocol in this cohort (24). In the current study, we conducted an intent-to-treat analysis, and also compared HIV drug-risk behaviors by treatment adherence. We defined participants as "high adherence" if they attended 6 or more of the 8 scheduled CBT sessions, or if the participant adhered to the medical regimen on 75% or more of the days on which depression medications were prescribed.

Analysis Plan

We present descriptive statistics for the full intent-to-treat sample and test for between group differences on background characteristics and possible confounding variables using t-tests (F-tests for comparing high-adherence, low adherence, and controls) and [chi square] tests. Paired sample t-tests were used to test for differences in HIV drug-risk behaviors between assessments. We used generalized estimating equations (GEE) to estimate population averaged treatment effects over time. The models were estimated using the XTGEE procedure in Stata Version 7. Tests of significance for individual parameter estimates are reported as z-statistics; the Wald chi-square test ([W.sup.2]) is reported for tests involving multiple parameter estimates. We analyzed the square-root transformed RAB HIV drug-risk scale as a normally distributed response variable. We specified the within participant correlation matrix as unstructured because we anticipated the correlations would not be constant across time. Baseline RAB drug-risk scores were included as a covariate in the GEE models, and reported tests of significance are based on robust standard error estimators. Models comparing participants with high adherence to the treatment protocol with low-adherent participants and controls included the reported number of days on which heroin was used prior to baseline assessment and depression type as covariates; prior analyses found that both measures were significantly associated with treatment adherence.

GEE models were also used to assess the association between depression remission status and HIV drug-risk behavior at each follow-up. Age, gender, baseline depression severity, baseline RAB drug-risk score, and depression type were included as covariates.

Finally, to test if treatment was differentially effective in participants with specific characteristics, we tested the interaction of age, gender and depression type with treatment on HIV drug risk scores, three variable associated with HIV risk (8).

RESULTS

Characteristics of the study sample are presented in Table 1. Randomization appeared effective in creating equivalent intent-to-treat comparison groups. Overall, most participants (82%) were White and a majority (64.2%) were male. Treatment groups did not differ significantly with respect to HIV status (p=743); 6 (10.7%) controls and 4 (7.6%) participants randomized to treatment were HIV positive at baseline. Participants were considerably more likely to meet diagnostic criteria for current opiate abuse/dependence than for cocaine abuse/dependence. Ninety-nine (90.8%) met criteria for current opiate dependence and 2 additional participants met criteria for current abuse but not dependence. Participants were considerably less likely to meet criteria for current cocaine abuse or dependence; 35 (32.1%) were currently dependent on cocaine, and 9 (8.3%) met criteria for abuse but not dependence. Most (39 of 44) of the participants who met criteria for current cocaine abuse/ dependence also met criteria for current opiate abuse/dependence. Treatment groups did not differ significantly with respect to either current opiate abuse/ dependence (p=27) or current cocaine abuse/dependence (p=24).

The proportion of participants assessed at each follow-up was highly consistent across treatment conditions. Eighty-six participants (78.9%) completed all 3 follow-ups, 12 (11.0%) completed 2 of the 3 follow-ups, 9 (8.3%) completed only 1 follow-up, and only 2 (1.8%) participants were lost at all 3 follow-up interviews. At 9 months, no study participants had new HIV positive test results.

Ten (20.8%) controls and 12 (26.1%) participants randomized to treatment reported receiving mental-health care (defined as one or more therapy sessions or a hospitalization for emotional or psychological problems) outside of the current study during between baseline and 3-month follow-up. Between group differences did not approach statistically significant levels (p=63). The treatment groups also did not differ significantly with respect to the proportion of participants receiving mental-health care at either 6- (p=80) or 9-months (p= 1.00). At 9-months 38/97 (39.2%) reported at least 1 day of mental health care.

At 3 months the treatment groups did not differ significantly with respect to entering inpatient drug treatment (p= 1.00), outpatient treatment (p=317), or detoxification programs (p= 158). At 9 months, treatment groups did not differ significantly with respect to their likelihood of entering residential (p=260) or outpatient (p=639) treatment. However, a higher proportion of participants randomized to treatment (49.1% vs 27.8% for controls) entered a detoxification program (p=029).

Reported HIV drug-risk behavior decreased sharply between baseline and first follow-up assessment at 3 months (Table 2). Indeed, mean HIV drug-risk scores were significantly higher at baseline than at 3 (t= 11.82, p < 001), 6 (t=13.19, p < 001), or 9 months (t=17.51, p < 001). Though the substantive magnitude of differences were smaller, mean HIV drug-risk scores were significantly lower at 9-months than at 3-(t=-2.50, p=015) or 6-months (t=-2.35, p=021). Differences between risk scores at 3- and 6-months were not statistically reliable (t= 1.12, p = 266). While all participants reported HIV drug-risk behaviors at baseline, 13 (13.8%), 26 (26%), and 39 (40.2%) of the participants who completed assessments at 3-, 6-, and 9-months, respectively, had HIV drug-risk scores of 0.

Observed treatment group differences with respect to mean HIV drug-risk score (both raw and transformed) were very small at all 3-time points (Table 2). Additionally, the proportion of participants with zero HIV drug-risk was similar in both treatment groups at all follow-up assessments. Table 2 also gives population averaged between group differences adjusted for baseline drug-risk. The coefficients giving adjusted between group differences in mean drug risk (square root transformed RAB) at 3 (b=-.021, z=-.04, p=967), 6- (b=-.310, z=-.60, p=550), and 9 months (b=-.019, z=-.24, p=968) were very small and statistically insignificant. Analyses using the untransformed outcome yielded results that were both substantively and statistically consistent with those reported in Table 2.

Participants attended on average 4.04 (+3.15) of the 8 CBT appointments. On average, participants took medications on 39.4% of the days they were prescribed. Overall, 43.4% were "highly adherent" to treatment. Participants who adhered to CBT treatment tended to follow the prescribed medical regimen, and vice versa; the Spearman rank-order correlation between these indicators of treatment adherences was .74 (p < 001).

Table 3 displays the observed mean drug-risk scores at each assessment for participants with high and low adherence to the treatment protocol. GEE was used to estimate mean differences in HIV drug-risk at each follow-up; covariates included baseline HIV drug risk, frequency of heroin use prior to baseline, and depression type. Differences between highly adherent participants and controls, and between participants with high and low treatment adherence are reported in Table 3.

Participants with high and low adherence to treatment were not significantly different with respect to HIV drug-risk behaviors at baseline ([t.sub.51] =49, p=627), a result confirmed with the nonparametric Wilcoxon rank-sum test. Significant between group differences were observed at the 3-month assessment ([W.sup.2]=7.88, df=2, p=019). At 3 months, participants with high treatment adherence had significantly lower mean drug-related HIV risk behavior than either controls (b=- 1.03, p=049) or those with low adherence to the treatment protocol (b=-1.74, p=009); (Table 3). Controls were not significantly different than participants who were randomized to treatment and had low adherence to the protocol (b=-.25, p=650). At the 6- ([W.sup.2]=82, df=2, p=665) and 9-month ([W.sup.2]=69, df=2, p=707) assessments between group differences were small and not statistically significant.

GEE models were used to estimate the association between depression remission status and HIV risk behavior at the follow-up assessments. Age, gender, baseline depression type, baseline depression severity as measured by the MHRSD, and baseline HIV drug-risk scores were included as covariates. We first estimated models to test the first-order interaction time by depression remission status interaction. This interaction term was not statistically significant for either the continuous ([W.sup.2]=2.76, df=2, p=252) or the dichotomous ([W.sup.2]=3.98, df=2, p= 137) drug-risk behavior outcomes, indicating the association between depression status and HIV drug-risk behaviors at the follow-up assessments could be summarized by a single parameter. A significant association between depression status and HIV drug-risk behaviors at follow-up was observed using both operationalizations of the outcome variable. Specifically, those in remission for depression at any follow-up had lower average HIV drug-risk scores (b=-.89, z=-2.92, p=004) than those not in remission. Additionally, those in remission for the depression were estimated to be less than half (OR=41, z=2.79, p=005) as likely to report any HIV drug-risk behaviors at follow-up than those not in remission for depression.

Finally, we estimated a GEE model that included the first-order treatment by gender interaction term. The treatment by gender interaction effect was statistically significant (b=2.14, z=2.91, p=004). We then estimated a GEE model to determine if gender differences in treatment effects were constant across the three follow-up assessments; the second-order treatment by gender by time interaction was not statistically significant (Wald [chi square] =46, df=2, p=793), indicating that gender differences in treatment effects could be summarized by a single parameter estimate.

For simplicity, we report treatment effects separately for females and males. Females randomized to treatment had significantly lower mean HIV drug-risk scores at follow-up (b=-1.37, z=-2.25, p=024) than females randomly assigned to the control condition. Treatment effects for males were trivially small and not statistically significant (b=58, z=1.49, p=136). No treatment interaction effects noted for age or depression type.

Using similar procedures, we also estimated the association between depression remission status and number of days participants reported using any heroin in the 30 days prior to each follow-up interview. Those in remission for depression at any follow-up reported an average of 5.9 fewer heroin use days (b=-5.92, z=-3.96, p=000) at the follow-ups. Frequency of heroin use was also associated positively with HIV drug-risk behavior (b=073, z=5.47, p=000).

DISCUSSION

Our primary finding was that combined pharmacotherapy and psychotherapy for depression did not result in a significant reduction in HIV drug risk behavior beyond that seen in an assessment-only control group by intent-to-treat analysis among active drug injectors. Overall, reported HIV drug risk behavior scores decreased sharply in the first three study months and continued to decline at the later follow-up assessments. Similar behavioral changes have been reported in other longitudinal studies of drug injectors (11).

At all follow-up points, those whose depression had remitted reported taking significantly fewer HIV drug risks. Our findings suggest that depressive symptoms are intimately associated with risk-taking. This may have been due in part to lower heroin use among those with remitted depression; with lower heroin use there may be less opportunity to share injection equipment. While this finding corroborates the reported association of mood disorders and use of illicit psychoactive drugs, the direction of causality is unclear here. Was it that participants were less depressed because they had halted drug use, or were they able to control their substance use because they were less depressed?

However, only a minority of our population experienced depression remission at any of the follow-ups, and this may in part be due to low treatment adherence. Those highly adherent to the combination depression treatment had significantly greater reduction in HIV drug risk behaviors than others at three months. These HIV drug risk findings parallel the depression outcome findings we have previously reported, with study participants who were fully adherent having superior depression outcomes after the acute phase of treatment (12). Lower HIV risk-taking among this subgroup may be a marker for persons more amenable to change or may be an indication of the effectiveness of treatment. The salutary effects of treatment diminished rapidly however, as those who were fully adherent had similar drug risk to others at the six and nine month assessments.

A persistent treatment effect on HIV drug risk behavior did exist for women. There was a treatment by gender interaction effect through the follow-up periods. We speculate that women may have been more amenable to maintaining treatment gains due to stronger alliances with female therapists, greater acquisition of coping skills, or more dramatic lifestyle changes following therapy. However, the exact reason or reasons for this interesting finding remain unexplained and worthy of further study.

Nonetheless, one of the important findings here was a demonstration that it is possible to engage a substantial proportion of depressed active injection drug users in a multi-session, combination antidepressant therapy and that this dually-diagnosed population can be followed over time. Missing treatment appointments may have been due in part to drug users' erratic lifestyles, one of the ongoing challenges for caring for this population. Adherence also may have been low, of course, because study participants did not perceive benefits from treatment, did not engage well with therapists, or because appointments interfered with drug-seeking activities. Additionally, our findings speak to the severe, chronic nature of depression in this population and the limited generalizability of findings from many of the randomized controlled trials of antidepressant medications that typically exclude substance abusers (25).

The MINERVA study had several limitations. First, all data are self-reported. Second, we did not examine the therapeutic alliance between patients and therapists, which may impact adherence to treatment and outcome. Third, we defined 75% completion of either modality as an "high adherence" based on other clinical trials literature; we do not know the actual threshold for a treatment effect for either modality in our combined therapy regimen. Fourth, this study tested only specific medications, and a single form of psychotherapy. Fifth, our unblinded study design was unable to determine the active component of acute combined treatment that led to short-term group differences in HIV drug risk-taking.

Clearly there is a need for strategies to improve drug abusers' adherence with the range of outpatient treatment, including substance abuse treatment, mental health treatment, or a combination of both. Currently, there is a poor fit between dually diagnosed substance abusing patients and the available drug and mental health treatment systems (26). Developing and improving programs for dually diagnosed patients who do not enroll in drug treatment is important both in terms of HIV risk reduction and as a potential prelude to drug treatment entry.

ACKNOWLEDGMENTS

This research was supported by the National Institute of Mental Health (RO1MH 61141 and R01MH 62719). Dr. Stein is a recipient of a NIDA Mid-Career Investigator Award (K24 DA 00512).

REFERENCES

(1.) Croughan JL, Miller JP, Wagelin D, Whitman BY. Psychiatric illness in male and female narcotic addicts. J Clin Psychiatry 1982; 43:225-228.

(2.) Rounsaville BJ, Weissman MM, Crits-Christoph K, Wilber C, Kleber H. Diagnosis and symptoms of depression in opiate addicts. Arch Gen Psychiatry 1982; 39:151-156.

(3.) Brienza RS, Stein MD, Chen MH, Gogineni A, Sobota M, Maksad J, Hu P, Clarke J. Depression among needle exchange program and methadone maintenance clients. J Sub Abuse Treat 2000; 18:331-337.

(4.) Rounsaville B J, Kosten TR, Weisman MM, Kleber HD. Prognostic significance of psychopathology in treated opiate addicts: A 2.5 year follow-up study. Arch Gen Psychiatry 1986; 43:739-745.

(5.) Carroll KM, Power MED, Bryant K. One-year follow-up status of treatment-seeking cocaine abusers: Psychopathology and dependence severity as predictors of outcome. J Nerv Men Dis 1993; 181:71-79.

(6.) McLellan AT, Luborsky L, Woody GE, O'Brien CP, Druley KA. Predicting response to alcohol and drug abuse treatment: Role of psychiatric severity. Arch Gen Psychiatry 1983; 40:620-625.

(7.) Woody GE, Luborsky L, McLellan T, O'Brien CP, Beck AT, Blaine J, Herman I, Hole A. Psychotherapy for opiate addicts: does it help? Arch Gen Psychiatry 1983; 40:639-645.

(8.) Stein MD, Solomon DA, Herman DS, Anderson B J, Miller I. Depression severity and HIV injection risk behaviors. Am J Psychiatry 2003; 160:1659-1662.

(9.) Marlatt GA, Gordon JR. Relapse Prevention. New York: Guilford Press, 1985.

(10.) Kang S, De Leon G. Correlates of drug injection behaviors among methadone outpatients. Am J Drug Alc Abuse 1993; 19:107-118.

(11.) Mandell W, Kim J, Latkin C, Suh T. Depressive symptoms, drug network, and their synergistic effect on needle-sharing behavior among street injection drug users. Am J Drug Alc Abuse 1999; 25:117-127.

(12.) Stein MD, Solomon DA, Herman DS, Anthony JL, Ramsey SE, Anderson B J, Miller IW. Pharmacotherapy plus psychotherapy for treatment of depression in active injection drug users. Arch Gen Psychiatry 2004; 61:152-159.

(13.) Stein MD, Solomon DA, Anderson BJ, Herman DS, Anthony JL, Ramsey SE, Brown RA, Miller 1W. Persistence of antidepressant treatment effects in a pharmacotherapy plus psychotherapy trial for active injection drug users. Am J Addict. In press.

(14.) Keller M, McCullough J, Klein D, Arnow B, Dunner D, Gelenberg A, Markowitz J, Nemeroff C, Russell J, Thase M, Trivedi M, Zajecka J, Blalock J, Borian F, Fawcett J, Hirschfeld R, Jody D, Keitner G, Kocsis J, Koran L, Kornstein S, Manber R, Miller I, Ninan P, Rothbaum B, Rush A, Schatzberg A. The acute treatment of chronic major depression: a comparison of nefazodone, cognitive behavioral analysis system of psychotherapy, and their combination. N Engl J Med 2000; 342: 1462-1470.

(15.) Miller IW, Keitner GI. Combined medication and psychotherapy in the treatment of chronic mood disorders. Psychiatr Clin North Am 1996; 19:151-171.

(16.) First M, Spitzer R, Williams J, Gibbons M. Structured Clinical Interview for DSM-IV--Patient Version. New York: Biometrics Research Department, New York State Psychiatric Institute, 1995.

(17.) Miller IW, Bishop S, Norman WH, Maddever H. The modified hamilton rating scale for depression: reliability and validity. Psychiatry Res 1985; 14:131-142.

(18.) Brown RA, Evans DM, Miller IW, Burgess ES, Mueller TI. Cognitive-behavioral treatment for depression in alcoholics. J Consult Clin Psychol 1997; 65:715-726.

(19.) Lewinsohn PM, Graf M. Pleasant activities and depression. J Consult Clin Psychol 1973; 41:261-268.

(20.) Fawcett J, Epstein P, Fiester S, Elkin I, Autry J. Clinical management-imipramine/placebo administration manual: NIMH treatment of depression collaborative research program. Psychopharmacol Bull 1987; 23:309-324.

(21.) Navaline HA, Snider EC, Petro CJ, Tobin D, Metzger D, Alterman AI, Woody GE. An automated version of the Risk Assessment Battery (RAB): enhancing the assessment of risk behaviors. AIDS Res Hum Retrovir 1994; 10:S281-S283.

(22.) Montgomery SA, Dufour H, Brion S. The prophylactic efficacy of fluoxetine in unipolar depression. Br J Psychiatry 1988; 9:69-76.

(23.) McLellan AT, Kushner H, Metzger D, Peters R, Smith I, Grissom G, Pettinati H, Argeriou M. The fifth edition of the addiction severity index. J Subst Abuse Treat 1992; 9:199-213.

(24.) Stein MD, Herman DS, Solomon DA, Anthony JL, Anderson B J, Ramsey SE, Miller IW. Adherence to treatment of depression in active injection drug users: the MINERVA study. J Subst Abuse Treat 2004; 26:87-93.

(25.) Nelson JC. Overcoming treatment resistance in depression. J Clin Psychiatry 1998; 59:13-19.

(26.) Primm AB, Gomez MB, Tzolova-Iontchev I, Perry W, Vu HT, Crum RM. Mental health versus substance abuse treatment programs for dually diagnosed patients. J Subst Abuse Treat 2000; 19:285-290.

Michael D. Stein, M.D., (1) Bradley J. Anderson, Ph.D., (1) David A. Solomon, M.D., (2) Debra S. Herman, Ph.D., (1) Susan E. Ramsey, Ph.D., (1) Richard A. Brown, Ph.D., (2) and Ivan W. Miller, Ph.D. (2)

(1) Department of Medicine, Brown University School of Medicine, Providence, Rhode Island, USA

(2) Department of Psychiatry and Human Behavior, Brown University School of Medicine, Providence, Rhode Island, USA

Address correspondence to Michael D. Stein, M.D., Division of General Internal Medicine, Rhode Island Hospital, 593 Eddy St., Providence, RI 02903, USA; Fax: (401) 444-5040; E-mail: mstein@lifespan.org

COPYRIGHT 2005 Taylor & Francis Ltd.
COPYRIGHT 2005 Gale Group