Addiction

Polysubstance Use Among Veterans in Intensive PTSD Programs: Association With Symptoms and Outcomes Following Treatment

Ish P. Bhallaa,b , Elina A. Stefanovicsa,c, and Robert A. Rosenhecka,c,d

aDepartment of Psychiatry, Yale School of Medicine, New Haven, Connecticut, USA; bNational Clinician Scholars Program, University of California Los Angeles, Los Angeles, California, USA; cMental Illness Research, Education, and Clinical Center, Veterans Affairs New England, West Haven, Connecticut, USA; dSchool of Public Health, Yale University, New Haven, Connecticut, USA.

ABSTRACT Objective: A distinct group of patients has recently been described who experience poly- substance use disorder characterized by use of multiple addictive substances. This study examines baseline characteristics and longitudinal outcomes of a group of such patients in specialized intensive Veterans Health Administration posttraumatic stress disorder (PTSD) programs and followed 4 months after discharge. Methods: Patients with diagnosed PTSD or subsyndromal PTSD and who used a single substance at baseline were compared to those who used two or three and more than three different addictive substances on meas- ures of PTSD symptom severity and functioning. Comparisons were also adjusted for differ- ences in total days of any substance use and other potentially confounding factors. Patients were reclassified according to the number of substances used at follow-up and again com- pared on symptoms and functioning. Results: Bivariate analysis of baseline data (N¼ 8,240) showed frequent polysubstance use (n¼ 3,695, 44.8% of the sample) and that use of greater numbers of substances was associated with more severe PTSD symptoms as well as more total days of substance use. At follow-up after treatment, 58.2% of the original sample (n¼ 4,797) was assessed. Polysubstance use was less frequent (n¼ 756, 15.8% of the follow- up sample), but showed a similar association with more severe symptoms, although differ- ences were attenuated after adjusting for total days of substance use. Conclusions: Polysubstance use, conceptualized within the multimorbidity perspective, is associated with increased severity of PTSD symptoms among veterans with dual diagnoses requiring com- plex interventions, the evaluation of which will require innovative trial designs.

KEYWORDS Polysubstance use; PTSD; multimorbidity; dual diagnosis; psychiatric comorbidity

Introduction

Substance use disorders are common comorbidities among veterans with posttraumatic stress disorder (PTSD) and are associated with more severe symptoms and impaired functioning (McCauley, Killeen, Gros, Brady, & Back, 2012). Many such patients with dual diagnoses use multiple substances, potentially leading to an even greater degree of clinical and therapeutic complexity. Despite evidence suggesting that polysub- stance use is increasing (Connor, Gullo, White, & Kelly, 2014), outcome research to date has primarily focused on patients with a single substance use disorder, pri- marily alcohol, opiate, or stimulant use disorders. However, some recent studies suggest that compared to a single substance use disorder, use of multiple substan- ces is associated with additional medical and psychiatric comorbidities as well as a greater risk of homelessness,

legal problems, and other manifestations of social dys- function (Hedden et al., 2010).

Interest has also grown in the issue of multimor- bidity more broadly, including the recognition that far more patients present with multiple chronic medical, psychiatric, and substance use disorders than present with a single disorder. These multiple morbidities compound each other, often leading to increasingly severe psychosocial dysfunction (North, Brown, & Pollio, 2016), along with higher levels of service use and related costs (Yoon, Zulman, Scott, & Maciejewski, 2014) and extensive and complex poly- pharmacy (Bhalla & Rosenheck, 2018; Doos, Roberts, Corp, & Kadam, 2014). Attention to multimorbidity is emerging as an important elaboration and expansion in the clinical assessment of psychiatric illness as the limits of focusing on single “primary” disorders is

CONTACT Ish P. Bhalla ibhalla@ucla.edu Department of Psychiatry, Yale School of Medicine, 10940 Wilshire Blvd., Suite 710, New Haven, Connecticut, USA Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/wjdd. � 2018 Taylor & Francis Group, LLC

JOURNAL OF DUAL DIAGNOSIS 2019, VOL. 15, NO. 1, 36–45 https://doi.org/10.1080/15504263.2018.1535150

increasingly recognized (Bhalla & Rosenheck, 2018; Langan, Mercer, & Smith, 2013; North et al., 2016). Drawing on the multimorbidity perspective, polysub- stance use disorder has been identified as a distinctive clinical phenomenon posing particularly severe clinical challenges that warrant further research (Bhalla, Stefanovics, & Rosenheck, 2017). A national study from the Veteran’s Health Administration (VHA) sug- gested that the use of multiple substances is associated with sharply increased risk of homelessness, serious mental illness, and medical problems, along with greater use of various psychiatric and medical services and more numerous psychotropic prescription fills (Bhalla et al., 2017). However, previous studies have only examined the number of substance use diagno- ses, as documented in administrative databases, and have not determined whether multiple substance use is associated with a greater number of total days of substance use or whether the greater number of total days of use, by itself, accounted for the greater sever- ity of associated problems. In addition, previous stud- ies were based on a cross-sectional analysis without evaluation of the longitudinal prevalence of polysub- stance use disorder and related outcomes. It has been recently demonstrated that while the use of a single substance of abuse is associated with poorer PTSD outcomes than use of no substances, no specific sub- stance (i.e., alcohol or any single illicit drug) has more adverse effects than any other (Manhapra, Stefanovics, & Rosenheck, 2015). The differential effect of multiple substance use, however, has not been examined.

From 1992 to 2011, the Northeast Program Evaluation Center (NEPEC) of the VHA monitored baseline characteristics and outcomes 4 months after discharge of almost 50,000 veterans who participated in specialized intensive treatment programs for PTSD, more than 8,000 of whom reported using at least one addictive substance (alcohol or illicit drugs) in the 30 days prior to admission to the program. Data from this program evaluation have consistently shown that substance use comorbidities complicate PTSD treat- ment and that baseline and persistent post-discharge substance use are associated with worse PTSD out- comes (Manhapra et al., 2015; Wilkinson, Stefanovics, & Rosenheck, 2015). In this study, we describe base- line characteristics, symptomatology, and measures of functioning, as well as post-discharge clinical out- comes among veterans with diagnosed PTSD or sub- syndromal PTSD who used one addictive substance and compare them to veterans who used two or three substances and those who used more than three sub- stances during the 30 days prior to admission. In

addition, we examine the correlation of the number of different substances used and index of the total days of reported substance use and, furthermore, control for this index in evaluating the impact of the use of multiple substances. These analyses were repeated on data gathered four months after discharge from the program. We thus seek to further understand the impact of polysubstance use on symptomatic and functional status, net of total days of substance use, at the time of program entry and 4 months after discharge.

Methods

Sample

This study used data from an administrative program evaluation conducted by NEPEC to evaluate treatment outcomes for veterans admitted to specialized intensive PTSD programs between 1992 and 2011. The programs primarily provided inpatient and residential treatment services in addition to day programs, where there was an expectation of abstinence from substance use during the program. All patients were admitted for treatment of PTSD with documentation of psychiatric comorbid- ities, including alcohol and drug abuse or dependence diagnoses, following DSM-IV, current at the time. Patients also reported the number of days of substance used for each of five classes of substances (alcohol, opioids, sedative/hypnotics, cocaine/amphetamines, and cannabis) during the 30 days prior to program entry, which formed the basis of the polysubstance use classification. Patients were evaluated upon entry and 4 months after discharge with a standard set of symptom and functional measures allowing for evaluation of changes in the frequency of polysubstance use and its continued association with adverse outcomes.

Measures

Sociodemographic data including clinical and military history were derived from structured clinical inter- views administered by program clinicians not involved in the care of the individual patient using a standar- dized set of forms. These measures documented age, sex, marital status, race, employment, possession of a valid driver’s license, lifetime history of incarceration for greater or less than two weeks, comorbid psychi- atric and substance use diagnoses in addition to PTSD, personality disorders, and Veteran’s Affairs (VA) service connected disability status (Rosenheck & Fontana, 2001). The year of program entry was added

JOURNAL OF DUAL DIAGNOSIS 37

as a measure in the analysis to account for program changes over the years of data collection.

Psychiatric symptom severity, alcohol or other drug use, employment, and medical problems were meas- ured with composite subscales of the Addiction Severity Index (ASI; McLellan et al., 1985). A measure of violent behavior was based on the National Vietnam Veterans Readjustment Study (Kulka, 1990). The measure was an aggregate of four items that assessed whether in the past 4 months the participant had destroyed property, made threats of violence against another person with or without a weapon, or got into a physical fight with another person (Buchanan, Stefanovics, & Rosenheck, 2018).

Indicators of psychiatric diagnoses, PTSD symptoms, and days of each substance used including alcohol, opioids, sedative/hypnotics (i.e., “downers” such as unprescribed barbiturates or benzodiazepines), cocaine, and cannabis in the past 30 days were derived from vet- eran self-report data upon entry to the program and 4 months post-discharge and allowed for construction of a classification of polysubstance use disorder as well as an index of total days of substance use based on the sum of reported days of each individual substance of abuse. Since a veteran who used multiple substances could report use of each of them for up to 30 days, this index had a potential range from 1 to 150.

The Short Form of the Mississippi Scale and four supplemental questions concerning key PTSD symp- toms both based on DSM-III criteria were administered to measure PTSD symptom severity and addressed symptoms such as nightmares, hypervigilance, the use of drugs to avoid feelings or help with sleep, irritability, suicidal thoughts, sleep disturbances, and difficulty expressing feelings (Fontana & Rosenheck, 1994). PTSD symptoms from these 15 items were combined to repre- sent total PTSD symptoms as well as subscales reflecting key features of clinical importance in the treatment of PTSD (i.e., numbness, hyperarousal, reexperiencing trauma, avoidance, suicidality, and violence; Buchanan et al., 2018). The aggregate measure of PTSD symptom- atology was based on the total severity rating of all symptoms (Fontana & Rosenheck, 1997). Subscales were conceptually derived by combining items from two scales based on face value concepts of clinical importance in treating PTSD in veterans.

Data analysis

Participants, all of whom had diagnoses of either an alcohol or drug use disorder, were first divided into three polysubstance use disorder categories based on

the number of types of substances they reported using in the past 30 days: one substance, two or three sub- stances, and four or five substances. This polysub- stance use disorder grouping was based on frequencies in the data and intuitive clinical meaning: 4,545 (55.1%) reported using one substance; 3,118 (37.8%) two or three substances; and 577 (7.0%) four or five substances. The choice of three groups allowed us to observe monotonic trends across the groups while preserving a degree of parsimony and respecting the frequencies in the data.

Next, bivariate analysis of variance and chi-square tests were used to compare these three groups on socio- demographic characteristics, comorbid psychiatric diag- noses, and indicators of functioning. Significance testing using a p value of less than .05 was used to iden- tify measures that differentiated the groups for inclu- sion in subsequent multivariate analyses.

Next, an analysis of covariance (ANCOVA) adjusted for significant variables from the prior analysis was applied to the baseline measures collected from the ASI including employment and medical composite indices in addition to the measure of violent behavior and PTSD symptomatology. If the overall type III sum of squares for the polysubstance use disorder classification term was significant at p < .005, the difference in least square means among polysubstance use disorder groups were compared using effect sizes as measured by Cohen’s d, the difference in means divided by the pooled standard deviation. In order to account for mul- tiple comparisons, we used a Bonferroni correction for the significance of the polysubstance use disorder classi- fication term. The standard p value of .05 was divided by 10 (the number of comparisons), which yields a sig- nificance p value threshold of .005. An effect size threshold of Cohens d greater than 0.20 or less than �0.20 was used to assess small effects (Cohen, 1988).

This analysis was then repeated with additional adjustment for the index measure of total reported days of substance use in the past 30 days, again with comparisons of least square means using Cohen’s d if the overall type III sum of squares analysis of the pol- ysubstance use disorder classification term was signifi- cant at p < .005.

Finally, follow-up data were used to reclassify the follow-up sample on the polysubstance use classifica- tion and to identify changes in polysubstance use after treatment. An additional set of ANCOVAs were used to compare outcomes 4 months after discharge from the program of the re-classified groups. The analysis was first conducted with adjustment only for baseline measures and then repeated with adjustment for the

38 I. P. BHALLA ET AL.

index of the total days of substance use assessed four months after discharge. As previously, a Bonferroni corrected p value < .005 was used for the overall test and followed by descriptive paired comparisons using Cohen’s d with an effect size difference of 0.20 used as the criterion for small effects. The study was approved by the Institutional Review Board committee of the VA Connecticut Healthcare System. A waiver of informed consent was obtained, as the study used administrative data and there were no patient identi- fiers included.

Results

Altogether, 8,240 veterans with diagnosed PTSD or subsyndromal PTSD plus a concurrent alcohol or

drug use disorder and who reported use of at least one drug or alcohol in the past 30 days were included in the analysis: 89.2% with alcohol use, 66.2% with drug use, and 55.2% with both. The average index measure of days of substance use was 25.2 (SD¼ 25.5) and was strongly correlated with the number of sub- stances used (r ¼ .69, p < .0001).

Bivariate analysis at baseline showed that use of greater numbers of substances was associated with younger age, program entry during earlier years of the program evaluation effort, being unmarried, being non-White, and having several problematic behaviors including a history of incarceration, greater likelihood of having a diagnosis of affective and personality dis- orders, recent hospitalization for a suicide attempt, not having a valid driver’s license, and self-reported

Table 1. Bivariate analysis of baseline characteristics by substance use group. Group 1 Group 2 Group 3

1 substance 2 or 3 substances 4 or 5 substances

n¼ 4,545 n¼ 3,118 n¼ 577

LS Mean SE LS Mean SE LS Mean SE p value

Age 47.653 0.159 46.260 0.181 44.104 0.492 <.0001 n % n % n % v2

Female 101 2.83 76 3.07 9 1.97 .4233 Marital status Married 1,271 27.96 654 20.97 79 13.69 <.0001 Separated/divorced 2,283 50.23 1,698 54.46 352 61.01 <.0001 Widow 102 2.24 61 1.96 11 1.91 .648

Race White 3,069 67.52 1,728 55.42 321 55.63 <.0001 Black 898 19.76 1,100 35.28 192 33.28 <.0001 Hispanic 297 6.53 164 5.26 31 5.37 .0564 History of incarceration 1,701 54.4 1,298 63.88 266 72.68 <.0001

Psychiatric diagnoses PTSD 4,298 94.59 2,946 94.48 541 93.76 .7152 Subsyndromal PTSD 210 4.62 142 4.56 32 5.57 .5658 Alcohol use/dependence 4,070 89.55 2,754 88.33 526 91.16 .0689 Drug use/dependence 1,954 42.99 2,475 79.38 523 90.64 <.0001 Anxiety disorder 559 12.3 404 12.97 90 15.63 .0743 Affective disorder 1,264 27.85 964 30.98 177 30.73 .0092 Bipolar disorder 233 5.13 183 5.88 35 6.07 .2993 Schizophrenia 46 1.01 33 1.06 10 1.73 .2861 Personality disorder 497 10.95 371 11.94 101 17.5 <.0001

Hospitalized for suicide attempt 240 5.28 190 6.09 54 9.36 .0004 Attempted suicide in the past 4 months 409 15.26 350 18.13 86 22.93 .0002 Participated in atrocities 784 17.27 626 20.09 125 21.66 .0012 Witnessed atrocities 1,054 23.22 715 22.95 148 25.65 .3644 Service use (type of PTSD program) Residential treatment 2,030 44.66 1,442 46.25 293 50.78 .0154 Day treatment 367 8.6 198 6.77 36 6.7 .0106 Evaluation/brief treatment 908 19.98 612 19.63 93 16.12 .0883 Specialized intensive (long-term inpatient) 958 22.46 671 22.93 115 21.42 .7196

Discharge status Employed 1,139 25.06 809 25.95 163 28.25 .2215 Living independently 2,873 64.94 1,620 53.77 227 41.2 <.0001 Living elsewhere 719 16.25 563 18.69 121 21.96 .0005 Transferred to another program 199 4.5 248 8.24 71 12.91 <.0001 Too sick to complete program 19 0.43 12 0.4 3 0.55 .8869 Completed program 3,560 80.52 2,219 73.7 366 66.55 <.0001 Chose to leave prematurely 280 6.33 191 6.34 34 6.18 .9894 Asked to leave for rule violation 345 7.8 331 10.99 72 13.09 <.0001

Note. PTSD¼ posttraumatic stress disorder; LS ¼ least squares; SE ¼ standard error.

JOURNAL OF DUAL DIAGNOSIS 39

participation in atrocities during military service (Table 1). These measures were used as covariates in subsequent analyses.

ANCOVA of baseline clinical measures, adjusting for these potentially confounding variables, showed a clear trend demonstrating that more extensive poly- substance use was associated with more severe violent behavior prior to admission, more severe total PTSD symptoms, and more severe symptoms of hyperar- ousal (Table 2).

After further adjusting these analyses for the index of the days of substance use before admission, severity trends remained with statistically significant differen- ces between polysubstance use groups on violent behavior and suicidality (Table 3).

Altogether, 4,797 veterans (58.2%) were reevaluated 4 months after discharge. Veterans who were success- fully followed after discharge were older than those not followed, were more likely to be married and to have access to an automobile, had lower scores on the ASI drug use composite index, and were less likely to have a diagnosis of a personality disorder (data avail- able on request).

As shown in Figure 1, at follow-up assessment the proportion of veterans with polysubstance use was far lower than at admission, with 2,745 (57.7%) reporting no use of any substances, 1,257 (26.4%) reporting using one substance, 684 (14.4%) using two or three substances, and 72 (1.5%) reported using 4 or 5 substances. Chi-square of the frequency of polysubstance use at baseline compared with the fre- quency at follow-up was highly significant, m2(6)¼ 265.4, p< .0001.

ANCOVA of follow-up data adjusting for baseline levels (Table 4) found a similar yet even more robust

trend associating higher levels of polysubstance use with more severe medical problems, greater violence at follow-up, and more severe PTSD symptomology affecting the total score and all subscales, although differences on the reexperiencing and avoidance sub- scale showed differences that were at least small by Cohen’s d only between those with any substance use and those with none. After adjusting for the index of the total number of days of substance use at follow- up, the trend persisted but was less robust, with at least small differences on violent behavior, the PTSD total scale, and all PTSD subscales only observed between veterans who used any substances as com- pared to those who used none (Table 5).

Discussion

This study used data from the national evaluation of VHA specialized intensive PTSD programs to evaluate the psychosocial burdens associated with comorbid polysubstance use. Higher levels of polysubstance use were associated with more severe PTSD symptoms at the time of program entry. After controlling for the number of total days of substance use, higher levels of polysubstance use remained significantly associated with two of these subscales: violent behavior and sui- cidality, even with Bonferroni adjustment for multiple comparisons.

Longitudinally, although polysubstance use remained a persistent problem for some veterans, fol- lowing treatment, it became far less prevalent. However, among those who used multiple substances at follow-up, symptoms of PTSD remained more severe after treatment than among those who use none or only one.

Table 2. Least squares means at baseline by substance use group adjusted for potentially confounding factors. Group 1 Group 2 Group 3

1 substance 2 or 3 substances 4 or 5 substances

LS Mean SE LS Mean SD LS Mean SD p value Comparisons� Employment 0.448 0.003 0.446 0.003 0.439 0.008 .517 ns Medical problems 0.556 0.005 0.577 0.007 0.596 0.015 .009 ns Violence at admission 1.779 0.021 2.024 0.025 2.416 0.058 <.0001 3> 2 > 1 PTSD total 3.604 0.007 3.656 0.008 3.709 0.018 <.0001 3> 1 PTSD subscales Numbness 3.675 0.007 3.714 0.009 3.730 0.020 .001 ns Hyperarousal 3.669 0.007 3.741 0.009 3.801 0.020 <.0001 3> 1 Reexperiencing 3.814 0.011 3.834 0.014 3.911 0.031 .014 ns Avoidance 4.086 0.014 4.130 0.017 4.162 0.039 .071 ns Suicidality 2.734 0.015 2.826 0.018 2.920 0.042 <.0001 ns Agitation 3.688 0.015 3.731 0.018 3.771 0.042 .081 ns

Note. Adjusted for age, year of program entry, White race, married, diagnosis of drug abuse/dependence or personality disorder, hospitalization for suicide attempt, possession of a valid driver’s license, and participation in war atrocities.

PTSD¼ posttraumatic stress disorder; ns¼ not significant; LS ¼ least squares; SD ¼ standard deviation.�Significance threshold of p <.005 and effect size of Cohen’s d> 0.20 or < �0.20.

40 I. P. BHALLA ET AL.

Table 3. Least squares means at baseline by substance use groups adjusted for the index of the total days of substance use in the past 30 days.

Group 1 Group 2 Group 3

1 substance 2 or 3 substances 4 or 5 substances

LS Mean SE LS Mean SD LS Mean SD p value Comparisons� Employment 0.444 0.003 0.449 0.004 0.456 0.010 .526 ns Medical problems 0.569 0.006 0.567 0.007 0.539 0.018 .268 ns Violence at admission 1.817 0.023 1.997 0.026 2.256 0.069 <.0001 3> 2,1 PTSD total 3.620 0.007 3.645 0.008 3.642 0.022 .103 ns PTSD subscales Numbness 3.685 0.008 3.707 0.009 3.688 0.024 .187 ns Hyperarousal 3.700 0.008 3.720 0.009 3.670 0.024 .043 ns Reexperiencing 3.828 0.012 3.824 0.014 3.848 0.037 .792 ns Avoidance 4.104 0.015 4.118 0.018 4.090 0.047 .709 ns Suicidality 2.730 0.016 2.829 0.019 2.938 0.050 <.0001 3> 1 Agitation 3.708 0.016 3.718 0.019 3.690 0.050 .808 ns

Note. Adjusted for age, year of program entry, White race, married, diagnosis of drug abuse/dependence or personality disorder, hospitalization for suicide attempt, possession of a valid driver’s license, and participation in war atrocities.

PTSD¼ posttraumatic stress disorder; ns¼ not significant; LS ¼ least squares; SD ¼ standard deviation.�Significance threshold of p< .005 and effect size of Cohen’s d> 0.20 or < �0.20.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Baseline 4 Months A�er Discharge

0 1 2-3 Substances 4-5 Substances

Figure 1. Proportion of veterans using substances at baseline compared to 4months after discharge, grouped by the number of substances used.

Table 4. Least squares means 4 months after program discharge by substance use group adjusted for potentially confounding factors. Group 0 Group 1 Group 2 Group 3

0 substances 1 substance 2 or 3 substances 4 or 5 substances

n¼ 2,745 n¼ 1,257 n¼ 684 n¼ 72

LS Mean SE LS Mean SE LS Mean SD LS Mean SD p value Comparisons� Employment 0.487 0.004 0.487 0.006 0.454 0.009 0.440 0.027 .0019 ns Medical problems 0.469 0.007 0.538 0.010 0.547 0.013 0.658 0.041 <.0001 3,2> 0; 3> 2,1 Violence after discharge 0.774 0.022 1.269 0.032 1.613 0.044 2.007 0.133 <.0001 3> 2 > 1> 0 PTSD total 3.207 0.009 3.457 0.014 3.561 0.018 3.687 0.056 <.0001 3> 2 > 1> 0 PTSD subscales Numbness 3.410 0.010 3.580 0.014 3.664 0.019 3.744 0.059 <.0001 3,2,1> 0; 3> 1 Hyperarousal 3.177 0.011 3.478 0.016 3.580 0.021 3.739 0.065 <.0001 3,2,1> 0; 3> 2,1 Reexperiencing 3.463 0.013 3.670 0.020 3.765 0.027 3.824 0.081 <.0001 3,2,1> 0 Avoidance 3.718 0.018 3.978 0.027 4.055 0.037 4.150 0.114 <.0001 3,2,1> 0 Suicidality 2.199 0.016 2.578 0.024 2.738 0.033 2.999 0.100 <.0001 3,2,1> 0; 3> 2,1 Agitation 3.224 0.017 3.518 0.026 3.645 0.035 3.867 0.107 <.0001 3,2,1> 0; 3> 2,1

Note. Adjusted for age, year of program entry, White race, married, diagnosis of drug abuse/dependence or personality disorder, hospitalization for suicide attempt, possession of a valid driver’s license, and participation in war atrocities.

PTSD¼ posttraumatic stress disorder; ns¼ not significant; LS ¼ least squares; SD ¼ standard deviation.�Significance threshold of p< .005 and effect size of Cohen’s d> 0.20 or < �0.20.

JOURNAL OF DUAL DIAGNOSIS 41

The association of polysubstance use disorder with more severe clinical and functional problems has been described previously. A study based on national VHA administrative data found that veterans with more numerous substance use disorders were more likely to have more numerous psychiatric and medical comor- bidities in addition to homelessness, greater use of inpatient psychiatric and residential treatment, and more numerous psychotropic prescription drug fills (Bhalla et al., 2017). The data presented here, while focused on the comorbidity of PTSD and polysub- stance use, are consistent with several studies showing that polysubstance use is associated with sociodemo- graphic characteristics, including younger age (Hedden et al., 2010; Midanik, Tam, & Weisner, 2007; Quek et al., 2013), greater medical (De Alba, Samet, & Saitz, 2004) and psychiatric disease burden (e.g., depressive disorders; Midanik et al., 2007) and a greater likelihood of lifetime suicide attempts (Smith, Farrell, Bunting, Houston, & Shevlin, 2011), more numerous legal problems including incarceration (Hedden et al., 2010) and arrests (Herbeck et al., 2013), and other indicators of psychosocial dysfunc- tion such as homelessness (Greenberg & Rosenheck, 2010; Tsai, Kasprow, & Rosenheck, 2014) and finan- cial problems (Herbeck et al., 2013).

In addition, a study of participants in the National Institute of Mental Health–funded Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) schizophrenia trial found a significant associ- ation of severe symptoms and poorer quality of life with either a comorbid substance use disorder or an additional psychiatric disorder. Thus, the adverse effect of multimorbidity seems to emerge regardless of whether the additional comorbidity is a psychiatric

disorder or one, or even more than one, substance use disorder (Bhalla, Stefanovics, & Rosenheck, 2018).

This study adds to this developing literature on multimorbidity in psychiatry by examining whether the number of discrete substances used by patients with dual diagnoses has adverse clinical effects even after accounting for an index of the number of days of any substance use. In addition, it is the first to our knowledge that investigated the effect of the use of multiple substances in a “dual diagnosis” sample, as all participants in this study had diagnoses of either PTSD or subsyndromal PTSD. Finally, the current study demonstrated the adverse effects of polysub- stance use on PTSD symptom severity both at baseline and longitudinally, finding attenuation of polysub- stance use following intensive treatment but also per- sistently increased symptoms with increased numbers of substances use following treatment.

The observed reduction in rates of polysubstance use after the intensive PTSD program could be explained in any of three ways. First, effective psy- chosocial and pharmacologic treatment targeted at PTSD symptoms may reduce the need for patients to self-medicate with addictive substances. Second, substance use may have declined either from specific treatments or from the restricted access to alcohol and drugs in supervised residential programs, thereby minimizing their exacerbating effect on PTSD symp- toms. Alternatively, patients may have entered resi- dential treatment at a time of extreme exacerbation of both substance use and PTSD symptoms, which may have abated as a result of regression to the mean. Most likely all three processes are at work, although their effects cannot be differentiated in this study.

Table 5. Least squares means 4 months after program discharge by substance use groups adjusted for the index of total days of substance use in the past 30 days.

Group 0 Group 1 Group2 Group 3

0 substances 1 substance 2 or 3 substances 4 or 5 substances

LS Mean SE LS Mean SE LS Mean SD LS Mean SD p value Comparisons� Employment 0.487 0.005 0.487 0.007 0.453 0.011 0.439 0.031 .035 ns Medical problems 0.482 0.008 0.532 0.010 0.516 0.017 0.579 0.048 .002 ns Violence after discharge 0.846 0.025 1.236 0.033 1.432 0.054 1.537 0.157 <.0001 3,2,1> 0; 3,2> 1 PTSD total 3.235 0.010 3.445 0.014 3.491 0.022 3.504 0.065 <.0001 3,2,1> 0 PTSD subscales Numbness 3.422 0.011 3.575 0.015 3.636 0.024 3.672 0.070 <.0001 3,2,1> 0 Hyperarousal 3.217 0.012 3.459 0.016 3.482 0.026 3.482 0.077 <.0001 3,2,1> 0 Reexperiencing 3.486 0.015 3.659 0.020 3.709 0.033 3.678 0.096 <.0001 3,2,1> 0 Avoidance 3.731 0.021 3.972 0.028 4.023 0.046 4.068 0.134 <.0001 3,2,1> 0 Suicidality 2.244 0.019 2.558 0.025 2.627 0.040 2.710 0.117 <.0001 3,2,1> 0 Agitation 3.279 0.020 3.493 0.026 3.509 0.043 3.513 0.125 <.0001 3,2,1> 0

Note. Adjusted for age, year of program entry, White race, married, diagnosis of drug abuse/dependence or personality disorder, hospitalization for suicide attempt, possession of a valid driver’s license, and participation in war atrocities.

PTSD¼ posttraumatic stress disorder; ns¼ not significant; LS ¼ least squares; SD ¼ standard deviation.�Significance threshold of p< .005 and effect size of Cohen’s d> 0.20 or <�0.20.

42 I. P. BHALLA ET AL.

During an evaluation of a patient with polysubstance use with or without psychiatric and medical disorders, clinicians are often placed in a difficult position of deciding which issue to address first, as the traditional approach is to evaluate, diagnose, and treat one prob- lem at a time. The multimorbidity perspective has emerged as a challenge to this paradigm, recognizing the additive and negative synergy of multiple chronic conditions. In this treatment approach, each of the multiple chronic conditions must be addressed simul- taneously and explicitly. Effectiveness research on treat- ment of polysubstance use within the multimorbidity framework is limited. For example, a review of clinical trials of PTSD treatment found the vast majority of randomized controlled trials (RCTs) (72%) excluded comorbid substance use disorders entirely, and a com- parable proportion (75%) excluded specific psychiatric disorders as well (Leeman et al., 2017). As a result, mental health professionals are left without evidence- based guidance in the treatment of veterans with poly- substance use. It is important to note that recently studies have begun to address co-occurring diagnoses with PTSD (Seal et al., 2011). For example, psychother- apy, particularly with a trauma focus, has been shown to improve PTSD symptoms in those with concurrent substance use disorder (Roberts, Roberts, Jones, & Bisson, 2015). There is also preliminary evidence from randomized controlled trials that pharmacological inter- ventions, for example, with antidepressants (Petrakis et al., 2012), topiramate (Batki et al., 2014), and prazo- sin (Simpson et al., 2015) can be effectively used for comorbid PTSD and alcohol use disorder, although unexpectedly naltrexone did not show additional sig- nificant benefit (Petrakis et al., 2012). Despite this recent progress, no trials have specifically addressed polysubstance use disorder in addition to PTSD.

The data presented here suggest that the number of different substances used by a patient is of particular clinical relevance. Two clinical scenarios are possible depending on the combination of substances used. The first scenario occurs when there are overlapping treatments for the multiple drugs. For example, a 12- step program is the treatment of choice for hazardous misuse of any addictive substance. There is also evidence supporting the use of naltrexone for both alcohol and opioid use disorders, so this medication in addition to a 12-step program could be suggested for patients who use these two substances concur- rently. In a second scenario, treatments do not overlap and may even conflict with one another. For example, patients withdrawing from opioids and alcohol con- currently may need opioid derivatives to be prescribed

for withdrawal from their opiate dependence, but these medications need to be used with caution in conjunction with benzodiazepines, the optimal stand- ard for treating alcohol withdrawal, since they may have adverse synergies. Future research is needed within the multimorbidity framework to address these real-world clinical challenges. Clinical trials are needed that specifically target patients with multiple co-occurring diagnoses. Such trials are likely to require complex designs involving multiple interven- tions, either simultaneously or sequentially, as in Sequential Multiple Assignment Randomized Trial (SMART) trials (Lei, Nahum-Shani, Lynch, Oslin, & Murphy, 2012), but only through such trials can our growing awareness of the adverse effects of multimor- bidity be matched by the required sophistication and complexity of appropriate evidence-based treatments.

There are several methodological limitations of this study that deserve comment. First, since patients could not be randomly assigned, observed differences between the groups may be affected by unmeasured confounding factors other than the number of substan- ces used. Second, the threshold for differences between polysubstance use groups was a small effect size on Cohen’s d. In fact, the differences between groups, while meeting the threshold for small effects, were rarely of even moderate size. Third, data are based exclusively on VHA administrative data, so the general- izability to nonveterans and non-VHA settings is unknown. In addition, data on the substances of abuse gathered before and after treatment are based on self- report data the validity of which has not been demon- strated and was not validated by biochemical tests. The assessment of medical illness severity was based on an ASI subscales, which could limit the validity of assess- ment of medical multimorbidity. Finally, follow-up data were missing in approximately 42% of the sample.

Nevertheless, the data presented here indicate that while patients with dual diagnoses using multiple sub- stances appear to have more severe PTSD symptoms than other veterans with dual diagnoses, and while intensive treatment appears to effectively reduce the amount of polysubstance use, the adverse trends per- sist longitudinally among those with persistent poly- substance use. Treatment and research with these individuals, conducted within the multimorbidity framework, will require innovative trial designs cap- able of addressing complex interventions. Polysubstance use is only one of several types of mul- timorbidity that need complex intervention, but it can be hoped that therapeutic advances in the area of pol- ysubstance use will have wider relevance for the

JOURNAL OF DUAL DIAGNOSIS 43

development of treatments for patients with other forms of multimorbidity as well.

Acknowledgements

This article is the product of work with the Department of Veterans Affairs and the VA New England Mental Illness Research Education Clinical, Centers of Excellence, and the Yale School of Medicine Department of Psychiatry. IB would also like to acknowledge the UCLA National Clinician Scholars Program, the David Geffen School of Medicine at UCLA, and Cedars-Sinai Medical Center.

Disclosures

The authors declare no conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors report no financial relationships with commer- cial interests.

Funding

This work did not have any specific funding and was com- pleted as part of our employment with the VA Connecticut Healthcare System and Yale School of Medicine.

ORCID

Ish P. Bhalla http://orcid.org/0000-0002-9598-2689

References

Batki, S. L., Pennington, D. L., Lasher, B., Neylan, T. C., Metzler, T., Waldrop, A., … Herbst, E. (2014). Topiramate treatment of alcohol use disorder in veterans with posttraumatic stress disorder: A randomized con- trolled pilot trial. Alcoholism: Clinical and Experimental Research, 38(8), 2169–2177. doi:10.1111/acer.12496

Bhalla, I. P., & Rosenheck, R. A. (2018). A change in per- spective: From dual diagnosis to multimorbidity. Psychiatric Services, 69(1), 112–116. doi:10.1176/ appi.ps.201700194

Bhalla, I. P., Stefanovics, E. A., & Rosenheck, R. A. (2017). Clinical epidemiology of single versus multiple substance use disorders: Polysubstance use disorder. Medical Care, 55 (Suppl 9 Suppl 2), S24–S32. doi:10.1097/MLR. 0000000000000731

Bhalla, I. P., Stefanovics, E. A., & Rosenheck, R. A. (2018). Mental health multimorbidity and poor quality of life in patients with schizophrenia. Schizophrenia Research. Epub ahead of print. doi: 10.1016/j.schres.2018.04.035

Buchanan, A., Stefanovics, E., & Rosenheck, R. A. (2018). Correlates of reduced violent behavior among patients receiving intensive treatment for posttraumatic stress dis- order. Psychiatric Services, 69(4), 424–430. doi:10.1176/ appi.ps.201700253

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Connor, J. P., Gullo, M. J., White, A., & Kelly, A. B. (2014). Polysubstance use: Diagnostic challenges, patterns of use and health. Current Opinion in Psychiatry, 27(4), 269–275. doi:10.1097/YCO.0000000000000069

De Alba, I., Samet, J. H., & Saitz, R. (2004). Burden of med- ical illness in drug- and alcohol-dependent persons with- out primary care. American Journal on Addictions, 13(1), 33–45.

Doos, L., Roberts, E. O., Corp, N., & Kadam, U. T. (2014). Multi-drug therapy in chronic condition multimorbidity: A systematic review. Family Practice, 31(6), 654–663. doi: 10.1093/fampra/cmu056

Fontana, A., & Rosenheck, R. (1994). A short form of the Mississippi Scale for measuring change in combat-related PTSD. Journal of Traumatic Stress, 7(3), 407–414.

Fontana, A., & Rosenheck, R. (1997). Effectiveness and cost of the inpatient treatment of posttraumatic stress dis- order: Comparison of three models of treatment. American Journal of Psychiatry, 154(6), 758–765. doi: 10.1176/ajp.154.6.758

Greenberg, G. A., & Rosenheck, R. A. (2010). Mental health correlates of past homelessness in the National Comorbidity Study Replication. Journal of Health Care for the Poor and Underserved, 21(4), 1234–1249. doi: 10.1353/hpu.2010.0926

Hedden, S. L., Martins, S. S., Malcolm, R. J., Floyd, L., Cavanaugh, C. E., & Latimer, W. W. (2010). Patterns of illegal drug use among an adult alcohol dependent popu- lation: Results from the National Survey on Drug Use and Health. Drug and Alcohol Dependence, 106(2–3), 119–125. doi:10.1016/j.drugalcdep.2009.08.002

Herbeck, D. M., Brecht, M. L., Lovinger, K., Raihan, A., Christou, D., & Sheaff, P. (2013). Poly-drug and mari- juana use among adults who primarily used metham- phetamine. Journal of Psychoactive Drugs, 45(2), 132–140. doi:10.1080/02791072.2013.785824

Kulka, R. A. (1990). Trauma and the Vietnam War generation: Report of findings from the National Vietnam veterans readjustment study. New York: Brunner/Mazel.

Langan, J., Mercer, S. W., & Smith, D. J. (2013). Multimorbidity and mental health: Can psychiatry rise to the challenge?. British Journal of Psychiatry, 202(06), 391–393. doi:10.1192/bjp.bp.112.123943

Leeman, R. F., Hefner, K., Frohe, T., Murray, A., Rosenheck, R. A., Watts, B. V., & Sofuoglu, M. (2017). Exclusion of participants based on substance use status: Findings from randomized controlled trials of treatments for PTSD. Behaviour Research and Therapy, 89, 33–40. doi:10.1016/j.brat.2016.10.006

Lei, H., Nahum-Shani, I., Lynch, K., Oslin, D., & Murphy, S. A. (2012). A “SMART” design for building individual- ized treatment sequences. Annual Review of Clinical Psychology, 8(1), 21–48. doi:10.1146/annurev-clinpsy- 032511-143152

Manhapra, A., Stefanovics, E., & Rosenheck, R. (2015). Treatment outcomes for veterans with PTSD and sub- stance use: Impact of specific substances and achievement of abstinence. Drug and Alcohol Dependence, 156, 70–77. doi:10.1016/j.drugalcdep.2015.08.036

McCauley, J. L., Killeen, T., Gros, D. F., Brady, K. T., & Back, S. E. (2012). Posttraumatic stress disorder and co- occurring substance use disorders: Advances in

44 I. P. BHALLA ET AL.

assessment and treatment. Clinical Psychology Science and Practice, 19(3), 283–304. doi:10.1111/cpsp.12006

McLellan, A. T., Luborsky, L., Cacciola, J., Griffith, J., Evans, F., Barr, H. L., & O’Brien, C. P. (1985). New data from the Addiction Severity Index. Reliability and validity in three centers. The Journal of Nervous and Mental Disease, 173(7), 412–423. doi:10.1097/00005053- 198507000-00005

Midanik, L. T., Tam, T. W., & Weisner, C. (2007). Concurrent and simultaneous drug and alcohol use: Results of the 2000 National Alcohol Survey. Drug and Alcohol Dependence, 90(1), 72–80. doi:10.1016/ j.drugalcdep.2007.02.024

North, C. S., Brown, E. S., & Pollio, D. E. (2016). Expanded conceptualization of multimorbidity to encompass sub- stance use disorders and other psychiatric illness. Annals of Clinical Psychiatry, 28(3), 182–188.

Petrakis, I. L., Ralevski, E., Desai, N., Trevisan, L., Gueorguieva, R., Rounsaville, B., & Krystal, J. H. (2012). Noradrenergic vs serotonergic antidepressant with or without naltrexone for veterans with PTSD and comorbid alcohol dependence. Neuropsychopharmacology, 37(4), 996–1004.

Quek, L. H., Chan, G. C., White, A., Connor, J. P., Baker, P. J., Saunders, J. B., & Kelly, A. B. (2013). Concurrent and simultaneous polydrug use: Latent class analysis of an Australian nationally representative sample of young adults. Frontiers in Public Health, 1, 61. doi:10.3389/ fpubh.2013.00061

Roberts, N. P., Roberts, P. A., Jones, N., & Bisson, J. I. (2015). Psychological interventions for post-traumatic stress disorder and comorbid substance use disorder: A systematic review and meta-analysis. Clinical Psychology Review, 38, 25–38. doi:10.1016/j.cpr.2015.02.007

Rosenheck, R., & Fontana, A. (2001). Impact of efforts to reduce inpatient costs on clinical effectiveness: Treatment of posttraumatic stress disorder in the Department of Veterans Affairs. Medical Care, 39(2), 168–180.

Seal, K. H., Cohen, G., Waldrop, A., Cohen, B. E., Maguen, S., & Ren, L. (2011). Substance use disorders in Iraq and Afghanistan veterans in VA healthcare, 2001–2010: Implications for screening, diagnosis and treatment. Drug and Alcohol Dependence, 116(1-3), 93–101. doi:10.1016/ j.drugalcdep.2010.11.027

Simpson, T. L., Malte, C. A., Dietel, B., Tell, D., Pocock, I., Lyons, R., … Saxon, A. J. (2015). A pilot trial of prazo- sin, an alpha-1 adrenergic antagonist, for comorbid alco- hol dependence and posttraumatic stress disorder. Alcoholism: Clinical and Experimental Research, 39(5), 808–817. doi:10.1111/acer.12703

Smith, G. W., Farrell, M., Bunting, B. P., Houston, J. E., & Shevlin, M. (2011). Patterns of polydrug use in Great Britain: Findings from a national household population survey. Drug and Alcohol Dependence, 113(2-3), 222–228. doi:10.1016/j.drugalcdep.2010.08.010

Tsai, J., Kasprow, W. J., & Rosenheck, R. A. (2014). Alcohol and drug use disorders among homeless veterans: Prevalence and association with supported housing out- comes. Addictive Behaviors, 39(2), 455–460. doi:10.1016/ j.addbeh.2013.02.002

Wilkinson, S. T., Stefanovics, E., & Rosenheck, R. A. (2015). Marijuana use is associated with worse outcomes in symptom severity and violent behavior in patients with posttraumatic stress disorder. The Journal of Clinical Psychiatry, 76(9), 1174–1180. doi:10.4088/JCP.14m09475

Yoon, J., Zulman, D., Scott, J. Y., & Maciejewski, M. L. (2014). Costs associated with multimorbidity among VA patients. Medical Care, 52(Suppl 3), S31–S36. doi: 10.1097/MLR.0000000000000061

JOURNAL OF DUAL DIAGNOSIS 45

Copyright of Journal of Dual Diagnosis is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.