Journal of Substance Abuse Treatment 39 (2010) 132 140

Regular article

The 10-year course of remission, abstinence, and recovery in

dual diagnosis
Haiyi Xie, (Ph.D.) a,b,, Robert E. Drake, (M.D., Ph.D.) a,b , Gregory J. McHugo, (Ph.D.) a,b ,
Lynn Xie c , Anita Mohandas, (B.A.) d
a

Department of Psychiatry, Dartmouth Medical School, Rivermill Commercial Center, 85 Mechanic St., Suite B4-1, Lebanon, NH 03766, USA
b
Department of Community and Family Medicine, Dartmouth Medical School, Rivermill Commercial Center, 85 Mechanic St., Suite B4-1,
Lebanon, NH 03766, USA
c
Class 11, Dartmouth College, Hanover, NH 03755, USA
d
Cravath, Swaine and Moore, LLP. 825 8th Avenue, New York, NY 10019-7475, USA
Received 28 October 2009; received in revised form 13 April 2010; accepted 16 May 2010

Abstract
This study examined the frequency, stability, predictors, and long-term outcomes of 6-month remissions of alcohol use disorders among
116 adults with co-occurring severe mental illnesses followed up prospectively for 10 years. Remission was defined as 6 months without
meeting syndromal criteria for alcohol abuse or dependence. Most participants (86%) experienced at least one 6-month remission, and these
remissions were relatively durable. One third did not relapse during follow-up, and two thirds relapsed on average 3 years after remission.
Six-month remissions were preceded by increased participation in substance abuse treatments, reductions in alcohol and drug use, decreases
in psychiatric symptoms, increases in competitive employment, and increases in life satisfaction. Following remissions, participants
improved in multiple domains of adjustment: reductions of psychiatric symptoms, decreases in alcohol and drug use, increases in work and
social contacts with nonabusers, decreases in hospitalizations and incarcerations, increases in independent living, and increases in life
satisfaction. Participants with alcohol dependence rather than alcohol abuse were less likely to attain 6-month remissions and more likely to
relapse after attaining remissions. 2010 Elsevier Inc. All rights reserved.
Keywords: Remission; Abstinence; Recovery; Relapse; Dual diagnosis

1. Introduction
Individuals with severe mental disorders such as
schizophrenia and bipolar disorder have high rates of alcohol
abuse and dependencea condition commonly termed cooccurring disorders or dual diagnoses (Kessler et al., 1996;
Regier et al., 1990). Further, co-occurrence of this type is
associated with serious negative outcomes, such as relapses
of mental illness, homelessness, and episodes of violence, at
least over short-term follow-up (Drake & Brunette, 1998;
Drake & Mueser, 1996; Drake, Osher, & Wallach, 1989).
Few studies, however, have examined the long-term course
Corresponding author. Tel.: +1 603 448 0263; fax: +1 603 448 3976.
E-mail addresses: haiyi.xie@dartmouth.edu, hyx55@yahoo.com
(H. Xie).
0740-5472/$ see front matter 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jsat.2010.05.011

of alcohol use disorder in this population, as has been done in

the general population (Vaillant, 1995).
To study the course of alcohol use, abuse, and dependence, the co-occurring disorders field has generally used
definitions from successive versions of the Diagnostic and
Statistical Manual (e.g., American Psychiatric Association,
1987, 1994). The criteria for alcohol abuse specify at least
one functional deficit, such as failure to fulfill role
obligations or legal, vocational, or social problems, which
is caused or exacerbated by use. People with severe mental
disorders who use alcohol easily meet the abuse criteria
because they are sensitive to small amounts of alcohol, also
have these functional problems due to mental illnesses, and
experience problems that are often exacerbated by alcohol
use (Mueser, Drake, & Wallach, 1998). Further, people with
severe mental illnesses are monitored more closely than

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

others in the community by treatment providers, and their

use of alcohol and alcohol-related problems are therefore
easily detected. Thus, case managers and other clinicians are
able to rate with high reliability and validity when clients
have crossed the syndromal threshold to abuse and when
they no longer have problems related to use (Drake et al.,
1990). In this study and elsewhere, we have therefore
defined remission of alcohol use disorder as not meeting
syndromal criteria for abuse or dependence for at least
6 months. Note that having a problem in even one functional
area related to alcohol use qualifies as abuse (American
Psychiatric Association, 1994).
In this study, we used 10-year data from the New
Hampshire Dual Diagnosis Study (Drake et al., 2006) to
address several questions related to remission, abstinence,
and recovery from alcohol abuse and dependence, which is
the most common form of substance abuse or addiction
among people with severe mental disorders (Kessler et al.,
1996; Regier et al., 1990). Based on theory and previous
research, we addressed several questions. First, because brief
intervals of remission do not predict long-term recovery for
alcohol use disorder (Vaillant, 1995), we examined the rate
and stability of 6-month remissions. Our clinical and
research experience with this population have suggested
that 6-month remissions may be more predictive of longterm recovery. Second, because cross-sectional analyses do
not address temporal patterns of change, we examined trends
in adjustment both before and after initial 6-month remissions. Temporal patterns are theoretically important because
changes that occur before remission may have causal
significance in relation to remission, whereas changes that
follow remission may be a consequence of the remission. For
example, anthropologists have argued that people with
mental illnesses need to have stable housing, relationships
with nonusers, constructive activities, and treatment relationships before they can be stably in remission (Alverson,
Alverson, and Drake 2000). Third, because alcohol abuse
and alcohol dependence are theoretically different entities
(American Psychiatric Association, 1994), we compared the
two groups at each step of recovery. Fourth, because people
who are in recovery from alcoholism often endorse the
importance of abstinence rather than remission without
abstinence, but researchers often document the pattern of
remission with moderate drinking (Vaillant, 1995), we
compared participants who were in remission with abstinence and those who were in remission without abstinence
during their first 6-month remissions.

2. Methods

133

schizoaffective disorder, or bipolar disorder) and cooccurring substance use disorders (alcohol or other drugs).
Schizophrenia was the most common mental illness, and
alcohol use disorder was the most common substance use
disorder. Details of recruitment and long-term follow-up are
provided in several previous reports (Drake et al., 1998;
Drake et al., 2006; Drake, McHugo, Xie, & Shumway, 2004;
Xie, McHugo, Sengupta, & Drake, 2003). Participants were
recruited from the outpatient rolls of 7 of New Hampshire's
10 community mental health centers between 1989 and
1992. They were assessed at baseline and reassessed yearly
using standardized instruments. The first 3 years involved a
randomized controlled trial of case management, and
participants were subsequently released from their experimental conditions and followed naturalistically. Standard
services in New Hampshire during this interval combined
mental health and substance abuse treatments. The study was
approved and monitored by the Dartmouth and the New
Hampshire Institutional Review Boards. Participants signed
informed consent each year.
For the current analysis, we adopted the following
definitions. Alcohol use disorder referred to a standardized
diagnosis of alcohol abuse or dependence according to
Diagnostic and Statistical Manual of Mental Disorders,
Revised, Third Edition (DSM-III-R) criteria (American
Psychiatric Association, 1987). Remission of alcohol use
disorder indicated the absence of full criteria for abuse or
dependence. Six-month remission denoted the absence of
alcohol abuse or dependence for at least six consecutive
months. Recovery was defined as long-term remission
(1 year or more; Vaillant, 1995) accompanied by improvements in other areas of psychosocial adjustment.
2.2. Study group
Among the original 223 participants in the New
Hampshire Dual Diagnosis Study, 166 were diagnosed
with alcohol use disorders (50 with alcohol abuse and 116
with alcohol dependence at baseline). Severe mental illness
diagnoses included schizophrenia (119, 53%), schizoaffective disorder (50, 23%), and bipolar disorder (54, 24%). At
the 10-year follow-up, 17 (10%) of the 166 original
participants with alcoholism had died; 33 (20%) were lost
to follow up, had dropped out, or missed the interview; and
116 (70% of total or 78% of surviving participants) were
reassessed. There were no significant differences in deaths or
dropouts between those with abuse or dependence at baseline
(2 = 0.031, p =.86). Thus, the index study group for our
analyses of remissions included 116 participants who had
active alcohol use disorders at baseline and were followed up
for 10 years.

2.1. Overview
2.3. Measures
The New Hampshire Dual Diagnosis Study was a
prospective, long-term follow-up study of clients with severe
and persistent mental illnesses (chronic schizophrenia,

To assess active alcohol use disorder and remission, we

used participants' scores on the Alcohol Use Scale (AUS)

134

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

at baseline and yearly assessments. The AUS is a 5-point

severity scale based on DSM-III-R criteria applied to the
previous 6 months: 1 = abstinence, 2 = use without
impairment (remission without abstinence), 3 = abuse, 4 =
dependence, and 5 = severe dependence (Drake, Mueser, &
McHugo, 1996). A team of three independent raters (blind
to original study condition) established consensus ratings
considering all available data on alcohol use and consequences (from self-report, other interview scales, clinician ratings, and urine toxicology; Drake et al., 1996).
Self-reports concerning alcohol use were collected using
the Addiction Severity Index (ASI; McLellan, Luborsky,
O'Brien, & Woody, 1980), which yields an alcohol
problem severity scale, and the Time-Line Follow-Back
(Sobell, Maisto, Sobell, Copper, & Sanders, 1980), which
yields days of use and number of drinks per drinking day.
The yearly ratings refer to 6-month point prevalence.
Researchers independently rated a randomly selected
subgroup of 29% of the participants (433 observations on
65 individuals) to determine reliability. The interrater
reliability (intraclass correlation coefficient) for AUS
ratings was 0.94. Validity of the AUS has been confirmed
by several studies (Drake et al., 1996).
In addition to alcohol use, we examined several other
outcomes: drug use (substances of abuse other than alcohol),
participation in substance abuse treatment, psychiatric
symptoms, residential situation, legal issues, functional
status, and quality of life. The Drug Use Scale (DUS) and
days of drug use in past 6 months (from the Time-Line
Follow-Back) were used to assess drug abuse. Like the AUS,
the DUS is a 5-point severity scale based on DSM-III-R
criteria (Drake et al., 1996). The Substance Abuse Treatment
Scale (SATS) assesses involvement in substance abuse
treatment, motivation for recovery, and current substance use
on an 8-point scale: 12 = early and late stages of
engagement, 34 = stages of persuasion, 56 = stages of
active treatment, and 78 = stages of relapse prevention and
recovery (McHugo, Drake, Burton, & Ackerson, 1995). Like
consensus AUS ratings, DUS and SATS ratings were made
by three independent raters using all data sources. Intraclass
correlation coefficients of reliability were 0.94 on the DUS
and 0.93 on the SATS. Psychiatric symptoms were assessed
as the total score on the Expanded Brief Psychiatric Rating
Scale (Lukoff, Nuechterlein, & Ventura, 1986), which
ranges from 24 to 168, with the higher score indicating
greater symptom severity. Criminal justice system involvement was assessed using the legal section of ASI (McLellan
et al., 1980). Residential history and institutional stays were
assessed using a self-report calendar supplemented by
outpatient and hospital records (Clark, Ricketts, & McHugo,
1996). Social functioning was assessed according to selfreports of days of competitive work (in log scale due to right
skewness) and amount of social contacts with nonsubstance-abusing friends. General life satisfaction (scores
range from 1 to 7) was measured using the Quality of Life
Interview (Lehman, 1988).

2.4. Data analysis

We examined six research questions: (1) Which
participants attained at least one 6-month remission
during 10 years of follow-up? (2) How stable were the
initial 6-month remissions? (3) Among those achieving
6-month remissions, was abstinence more stable than
remission without abstinence? (4) What changes preceded
the initial 6-month remissions? (5) Did 6-month remissions lead to changes in other life domains consistent
with recovery? (6) Were the patterns for participants with
alcohol abuse different from those for participants with
alcohol dependence?
To address Question 1, we first identified the
participants with alcohol use disorder who attained at
least one 6-month remission and then compared baseline
characteristics between those who attained remission and
those who did not, using either t-tests (for continuous
measures) or chi-square tests (for dichotomous measures).
We used survival analysis to address Questions 2 and 3.
We used mixed-effects regression models (for continuous
outcomes) and mixed-effects logistic regression models
(for dichotomous outcomes) to address Questions 4 and 5.
To address Question 6, we repeated the analyses for
Questions 2 to 5 adding abuse versus dependence status at
baseline to the models. We used LIFETEST and PHREG
procedures in SAS 9.2 (SAS Institute Inc., 2008) for
survival analyses and GLIMMIX procedure for mixedeffects model analyses. Because individuals obtained the
remissions at different time points over 10 years, after we
identified those participants who attained initial 6-month
remissions, we aligned them according to their time of first
remission for the subsequent analyses. This resulted in
different numbers of participants at the different time
points with fewer participants in the years that were further
(before or after) from their remission.
Specifically, every participant has 11 data (time) points,
a baseline plus 10 years of follow-up. The year of initial
remission ranged from 1 to 10 for the 100 participants in
this study. This resulted in a different number of data points
before and after the first remission for different individuals.
For example, if Participant A experienced a remission in
Year 10, he or she would have 10 data points before
remission and none afterward. If Participant B experienced
remission in Year 1, he or she would have 1 data point
before remission (baseline) and 9 data points afterward. If
Participant C achieved remission in Year 5, he would have
5 data points before remission and 5 afterward (this is
shown in Appendix A). Because few participants took as
long as 710 years to achieve remission, the number of
cases for the first 4 assessment points (that is, t10, t9,
t8, and t7) was very small (from 2 to 12). To ensure
adequate sample size for each time point for the
longitudinal analysis concerning Questions 4 and 5, we
therefore excluded the first 4 data points before initial
remission from the analyses. In other words, we included 6

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

data points before the initial 6-month remission and 9 data

points after the initial remission.

135

than dependence and had fewer days of alcohol use over the
previous 6 months. In other words, clients with more severe
alcohol use disorders were less likely to attain a 6-month
remission over 10 years.

3. Results
3.2. How stable were 6-month remissions?
3.1. Which participants attained a 6-month remission?
All of the 116 participants with alcohol use disorders
experienced brief episodes of remission (no data presented).
More than 5 of 6 (86%, n = 100) also attained at least one
6-month remission. Of these 100 6-month remissions, two
thirds (n = 66, 66%) were remissions without abstinence,
and one third (n = 34, 34%) were remissions with
abstinence.
Participants who attained 6-month remissions (n = 100)
were compared with those who did not (n = 16) with respect
to baseline characteristics: basic demographics, diagnosis
(bipolar disorder vs. schizophrenia spectrum disorder),
substance use dimensions, residential status (days of
hospitals, jails and independent livings), working status,
social contacts, and general life satisfaction. Table 1 shows
that only two baseline variables were significant predictors.
The remitters were more likely to have alcohol abuse rather

The time between initial 6-month remission and relapse

was examined using KaplanMeier's survival curves
(Fig. 1). Among the 100 clients who attained remission,
35% (n = 35) did not relapse during follow-up (they were
censored), and the time to relapse for those who relapsed
ranged from 1 to 9 years with median survival time of
3 years. Because participants attained initial 6-month
remission at different points after study entry, the time or
chance for relapse varied. Participants with alcohol
dependence rather than abuse showed a trend toward
relapsing sooner (2 = 2.84, p = .09).
3.3. Were remissions with abstinence more stable than
remissions without abstinence?
We next examined survival curves separately for the
remission groups with and without abstinence. The

Table 1
Baseline characteristics of participants who attained 6-month remission and those who did not
Remission group (n =100)

Nonremission group (n = 16)

Variables

M/Count

SD/%

M/Count

SD/%

Difference

Age (years)
Race (White)
Gender (male)
Marital status (ever married)
Education (high school or greater)
Diagnosis (bipolar)
BPRS a, total score
AUS b
Days of alcohol use
ASI alcohol composite c (yes)
Daily average number of drinks
DUS d
ASI drug composite e (yes)
Days of drug use
SATS f
Hospital stay past year (yes)
Jail/Prison past year (yes)
Proportion of days of independent living past year g
Competitive job past year (yes)
Social contact with nonabusers (yes)
General Life Satisfaction (17)

33.84
97
70
37
66
30
45.66
3.76
68.01
90
6.30
2.37
62
23.98
2.78
57
8
0.47
27
11
3.83

8.65
97%
70%
37%
66%
30%
14.95
0.62
58.67
96%
5.99
1.31
65%
40.93
0.84
59%
8%
0.42
30%
11%
1.48

35.88
15
15
6
11
3
48.94
4.06
113.90
16
9.40
2.09
9
32.56
2.44
4
2
0.67
3
1
3.80

6.71
94%
94%
38%
69%
19%
12.22
0.40
54.12
100%
8.70
1.27
56%
54.80
0.51
31%
13%
0.43
19%
6%
1.16

ns
ns
ns
ns
ns
ns
ns

ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns

Brief Psychiatric Rating Scale (24168).

Alcohol Use Scale (15).
c
Addition Severity IndexAlcohol Composite (dichotomized [zero vs. nonzero] due to many zeros).
d
Drug Use Scale (15).
e
Addition Severity IndexDrug Composite (dichotomized due to many zeros).
f
Substance Abuse Treatment Scale (18).
g
Independent Living consists of the following residential settings: house/trailer, apartment, rooming house.
p b .05.
p b .01.
b

136

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

relapse) in relation to type of remission (6-month remissions

with and without abstinence), we examined the survival rates
of four subgroups: 1 = abuse to abstinence, 2 = abuse to
remission without abstinence, 3 = dependence to abstinence,
and 4 = dependence to remission without abstinence. The
results showed a significant overall group difference (2 =
9.1, p = .03). The pair-wise comparisons showed that only
one pair had a significant difference in survival times: the
participants who went from abuse to abstinence had a
significantly longer survival time than those who went from
dependence to remission without abstinence (2 = 7.01, p =
.008). When the same covariates used above were included
in the model, this difference became marginally significant
(2 =3.65, p = .06).
Fig. 1. Time to relapse for 100 participants after initial six-month remission.

3.4. What changes preceded remissions?

abstinence group had a better survival rate, that is,
participants were more stable in remission. The median
survival times were 2 years for the remission without
abstinence group and 6 years for the abstinence group, and
the log-rank test for difference in the survival times is
significant (2 = 7.80, p = .005).
To examine whether this difference was confounded by
group nonequivalence, we conducted a multivariate survival
analysis with Cox regression that included a set of relevant
covariates: baseline AUS, SATS, social contact with
nonusers, working status, number of treatment episodes for
alcohol problems at the point of the first remission, ASI
alcohol composite score at the baseline, sex, and diagnosis
(bipolar vs. schizophrenia spectrum). This analysis showed
that working status was a marginally significant predictor for
risk of relapse with workers less likely to relapse (hazard
ratio = 0.47, p = .06) and that the effect of group (remission
with abstinence vs. without abstinence) was no longer
significant (hazard ratio = 0.59, p = .21). In other words,
group nonequivalence, rather than type of remission,
accounted for the difference in time to relapse.
To investigate further the effect of baseline abuse versus
dependence status on the difference in survival (time to

Table 2 shows average annual changes prior to 6-month

remissions. Participation in substance abuse treatments
(SATS), competitive employment (in log scale), and general
life satisfaction increased the trend toward remissions; days
of alcohol use, average number of drinks per drinking day,
drug use severity (DUS), and total psychiatric symptoms
(Brief Psychiatric Rating Scale [BPRS]) decreased these
trends. In other words, participants who were headed toward
6-month remissions were participating more in substance
abuse treatments, were cutting down on alcohol and drug
use, were working more hours, and were experiencing
improvements in psychiatric symptoms and life satisfaction
before they attained initial 6-month remissions. There were
no significant differences between the alcohol abuse and
dependence groups in these trends. Both groups made
significant changes prior to developing 6-month remissions.
3.5. Did remissions lead to recovery?
Table 3 shows the overall trends following 6-month
remissions. Nearly all outcomes had significant and positive
time effects. Participation in treatment (SATS), proportion of

Table 2
Changes (or trends) prior to 6-month remissions (n =100)
Outcome variables

Estimates

SE

Significance

SATS
DUS
Days of alcohol use
Daily average number of drinks
BPRS, total score (24-168)
Proportion of days of independent living past year
Hospital stay past year (yes)
Jail/Prison past year (yes)
ASI legal composite (yes)
Competitive working hours past year (log scale)
Social contact with nonabusers (yes)
General Life Satisfaction (17)

0.419
0.094
13.386
0.748
0.822
0.020
0.059
0.024
0.001
0.129
0.055
0.085

0.039
0.027
1.410
0.169
0.332
0.010
0.065
0.093
0.078
0.060
0.071
0.037

ns
ns
ns
ns

ns

p b .05.
p b .01.

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

137

Table 3
Changes (or trends) following 6-month remissions (n =100)
Outcome variables

Estimates

SE

Significance

SATS
DUS
Days of alcohol use
Daily average number of drinks
BPRS, total score (24168)
Proportion of days of independent living past year
Hospital stay past year (yes)
Jail/Prison past year (yes)
ASI legal composite (yes)
Competitive working hours past year (log scale)
Social contact with nonabusers (yes)
General Life Satisfaction (17)

0.170
0.023
0.847
0.127
0.286
0.019
0.178
0.119
0.050
0.112
0.118
0.027

0.022
0.011
0.383
0.052
0.124
0.004
0.036
0.061
0.046
0.027
0.033
0.013

ns

p b .05.
p b .01.

days of independent living, social contacts with nonabusers,

competitive employment, and general life satisfaction had
significant increasing trends, whereas average number of
drinks per drinking day, drug use severity (DUS), total
psychiatric symptoms (BPRS), hospitalization, and incarceration had significant decreasing trends. Average days of
alcohol use, after a sharp decline in the preceding period,
tended to be stable, with a slight upward trend (p = .03)
following initial 6-month remissions. Comparing the alcohol
abuse and dependence groups, the dependence group had
higher scores on drug use disorder (DUS) at the time of

remission (p = .01), but there were no differences in the

trajectories of any outcomes following 6-month remissions.
Table 4 summarizes comparisons between the remission
groups with and without abstinence following initial
6-month remissions. Group effects reflect the average
difference between the two groups at the point of reaching
remissions. Four measures had significant group effects in
favor of the abstinence group: greater treatment participation (SATS), lower drug use severity (DUS), fewer days of
alcohol use, and fewer average drinks per drinking day.
The abstinence group also had a higher probability of

Table 4
Differences in recovery-related outcomes following remission with abstinence (n = 34) and remission without abstinence (n = 66)
Intercept a

Group b

Time c

G Td

Outcome variables

Estimate SE

Significance Estimate SE

Significance Estimate SE

Significance Estimate SE

Significance

SATS
DUS
Days of alcohol use
Daily average number
of drinks
BPRS, total score (24168)
Proportion of days of
independent living
past year
Hospital stay past year (yes)
Jail/Prison past year (yes)
ASI legal composite (yes)
Competitive working hours
past year (log scale)
Social contact with
non-abusers (yes)
General Life Satisfaction
(17)

4.890
1.937
19.481
3.685

0.204
0.102
3.591
0.398

0.241
0.054
0.072
0.214

0.027
0.013
0.472
0.065

ns

0.208
0.089
2.633
0.253

0.046
0.022
0.799
0.109

41.562
0.556

1.193
0.047

2.916 2.036 ns
0.093 0.080 ns

0.271
0.019

0.154 ns
0.005

0.036
0.002

0.260 ns
0.009 ns

0.410
3.090
2.330
1.494

0.218
0.388
0.304
0.298

0.658
1.387
0.770
0.053

ns

ns
ns

0.179
0.003
0.007
0.114

0.042
0.077
0.058
0.034

ns
ns

0.000
0.319
0.118
0.004

0.081
0.135
0.097
0.058

0.689

0.239

0.243 0.414 ns

0.110

0.040

0.024

0.071 ns

4.496

0.137

0.013 0.234 ns

0.032

0.016

0.014

0.028 ns

1.398
0.424
16.299
1.551

0.350
0.174
6.130
0.679

0.391
0.555
0.484
0.515

ns

ns
ns

Intercept is initial level of the outcome for remission without abstinence group (statistical test is against zero).
Group effect is the difference in initial level of the outcomes between remission without abstinence group and abstinence group.
c
Time effect is average rate of change (or slope) for remission without abstinence group.
d
G T interaction effect is the difference between the abstinence group and the remission without abstinence group in average rate of change.
p b .05.
p b .01.
b

138

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

incarceration during the 6 months in which they attained

remission. Time trends (for the remission without abstinence group, or reference group, only) again showed that
nearly all areas of adjustment improved following 6-month
remissions. For group differences after initial remission,
five outcomes showed significant group-by-time interactions: rate of incarceration was declining for the abstinence
group, whereas the remission without abstinence group was
changing more on treatment participation (SATS), drug use
severity (DUS), days of alcohol use, and average number of
drinks per drinking day. As a result, long-term follow-ups
showed that the abstinence and remission without abstinence groups tended to converge following initial 6-month
remissions. This pattern is illustrated in Fig. 2.
To examine interrelationships between the baseline
alcohol dependence and abuse groups and remission type
(remissions with and without abstinence) over time, we reran
the model with three-way interactions (baseline status by
remission type by time). There were no significant
differences for any outcomes.
4. Discussion
These longitudinal data illuminate several aspects of
recovery from alcohol use disorder among clients with
severe mental illnesses. Theoretical and practical implications emerge.
First, 6-month remissions appear to be relatively stable
indicators of recovery. One third of the participants who
attained 6-month remissions did not experience relapses of
alcoholism, and the others generally relapsed years after their
initial 6-month remissions. Thus, clinicians and programs are
justified in benchmarking on 6-month remissions.
Second, our definition of 6-month remission differs from
the DSM-IV (American Psychiatric Association, 1994)
definition, which specifies meeting no criteria for abuse or
dependence for 1 month for early full remission and no
criteria for abuse or dependence for 12 months for full
remission. Our data show predictive validity following
6 months without meeting the full syndrome of abuse or
dependence. The findings apply to people with severe mental
illness and alcohol use disorders, not to the general

Fig. 2. Days of alcohol use for two remission groups.

population of people with alcohol use disorders. Predictive

validity of the more lenient criteria implies that clients with
co-occurring disorders can be making significant progress,
perhaps being in recovery, without meeting more stringent
criteria. Time of stability rather than specific criteria appears
to be critical.
Third, the alcohol abuse versus dependence distinction
did have some predictive validity in this study. Participants
with dependence were less likely to attain 6-month
remissions and were more likely to relapse following
remissions. These findings regarding severity of alcoholism
and the abuse versus dependence distinction as predictors
are consistent with the literature on alcoholism in the
general population (Vaillant, 1995). Many of our participants met criteria for alcohol dependence without having
physiological dependence, and this distinction may be
worth pursuing further in studies of persons with severe
mental illnesses.
Fourth, participants experienced substantial changes
during the years prior to 6-month remissions: increased
participation in substance abuse treatment, decreased use of
alcohol and drugs, decreased psychiatric symptoms, increased competitive work hours, and increased quality of
life. These factors clearly indicate that clients were
developing skills and supports for stable remission by
changing their use behaviors, learning to manage their
illnesses, and substituting competitive employment for time
using substances for several months before they developed a
6-month remission. The trajectory therefore supports the
view that gradual change, consistent with harm reduction, is
the modal pathway toward recovery for this population
(Carey, 1993; Osher & Kofoed, 1989).
One clear implication of the preremission findings is
that achieving recovery involves much more than avoiding
alcohol and drugs. Participants changed many aspects of
their lives and experienced improvements in life satisfaction prior to attaining 6-month remissions. For these
clients, attaining remission thus appears to involve a
comparison of alternative lifestyles; people must believe
that life can be better without drinking, and they test the
alternatives before they adopt lengthy remissions. Studies
of supported employment demonstrate that competitive
employment and identity as a worker can substitute
effectively for substance use and mental patient identity
(Becker, Drake, & Naughton, 2005; Bush, Drake, Xie,
McHugo, & Haslett, 2009). Employment provides structure, relationships, and self-esteem. Our data confirm that
employment precedes remission for many people with
severe mental illnesses.
Fourth, the trend in virtually every area of adjustment
revealed further improvements following initial 6-month
remissions. Thus, 6-month remissions were definitely
harbingers of recovery. Programs can use these data as
well as clients' testimonials in educating clients that their
lives will be substantially better after they achieve 6-month
remissions of their alcohol use disorders. Clients need to

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

139

would not invalidate internal validity for a naturalistic

study, it did result in a smaller study group, with attendant
consequence for statistical power to detect significant
differences. Thus, our findings may be conservative,
especially for subgroup analyses. Conversely, we conducted numerous analyses and may be at risk of finding
significant results by a chance alone. Finally, our data
refer to 6-month point-prevalence, not continuous measures. Some episodes of relapse may have occurred
between our 6-month assessment intervals. Nevertheless,
time trends in recovery patterns based on 6-month
sampling every year are likely to be valid representations
of longitudinal trajectories.

make numerous changes en route to remission, and

remission in turn enhances further recovery.
Fifth, the findings regarding abstinence were mixed. The
naturalistic pattern of alcoholism recovery for these dually
diagnosed clients suggests a stepwise, motivational, harm
reduction approach rather than early abstinence. Remissions
with abstinence versus nonabstinence appeared to be
explained by group nonequivalence and to have minimal
predictive validity. Another factor may be that abstinent
participants were more likely to have been incarcerated and
to have had limited or no access to alcohol. Even when
participants developed 6-month remissions, many were
drinking occasionally rather than abstinent, and the difference did not strongly affect long-term outcomes. The
exception might be clients with alcohol dependence rather
than abuse, where the interaction with abstinence showed
some predictive validity. We previously found that moderate
drinking was not an effective recovery strategy for this
population over the long term (Drake & Wallach, 1993).
Further investigations of recovery patterns in this population
may clarify the role of abstinence, but our clinical impression
is that people with mental illnesses often continue to be
vulnerable to lapses for many years.
Finally, although only a minority of participants in this
study group abused other drugs (primarily cannabis and
cocaine), the longitudinal time trends showed that
participants reduced and stopped using these other drugs
gradually in parallel with their recoveries from alcoholism.
Other drug use did not strongly affect the course of
recovery from alcoholism.
Several caveats warrant mention. These data refer to a
relatively small study group of predominantly Caucasian
clients, mostly in their 30s at baseline, who lived in New
Hampshire and enrolled in a randomized controlled trial in
the early 1990s. The findings may not generalize to
people from ethno-racial minority groups, to other age
groups, to clients in urban areas, to those who are
disengaged from mental health services, to those who
predominantly use other drugs, or even to the drug and
treatment environments of the 2010s. Today people with
severe mental illnesses are more prone to polysubstance
use, more prone to involvement with the criminal justice
system, and also more likely to access integrated dual
diagnosis treatment programs.
In addition, our analysis was based on 70% of the study
group at 10 years of follow-up. Although 30% attrition

5. Conclusions
This longitudinal study of recovery from alcohol use
disorder among clients with severe mental disorders leads
to several conclusions. Six-month remissions of alcohol
abuse or dependence tend to predict lengthy remissions as
well as recovery in many other areas of adjustment. Clients
make a number of positive changes in their lives prior to
developing 6-month remissions, consistent with a stagewise approach to treatment and a philosophy of harm
reduction. Remissions with abstinence, as compared with
remissions without abstinence, were more stable for clients
with alcohol dependence rather than abuse. However, all
remission groups tended to have lapses to drinking but
nevertheless continued to recover over time. Finally, the
implications for policy and treatment related to this
population are several. We should include employment
or vocational services in treatment; counselors should
focus on helping clients to develop a healthier lifestyle that
is rich with nondrug reinforcement; and we should
recognize the importance of a continuing recovery
model, in which clients continue to be monitored and
receive supports for at least a few years after achieving
6-month remissions.
Acknowledgments
This work was supported by NIDRR Grant #H1336050181
from the Department of Education. The authors thank
Mark McGovern, PhD, for helpful suggestions on drafts
of the article.

Appendix A. Illustration of Data Re-ordered by Remission Time Point (Numbers are AUS scores)
t10
t9
Participant A
4
4
Participant B
Participant C
Remission point.

t8
3

t7
4

t6
4

t5
3

t4
4

t3
4

t2
3

t1
3
3
3

t0
2
1
2

+t1

+t2

+t3

+t4

+t5

+t6

+t7

+t8

+t9

2
2

2
1

3
2

2
2

3
1

140

H. Xie et al. / Journal of Substance Abuse Treatment 39 (2010) 132140

References
American Psychiatric Association. (1987). Diagnostic and statisical manual
of mental disorders: DSM-III-R. Washington, DC: American Psychiatric
Association.
American Psychiatric Association. (1994). Diagnostic and Statisical
Manual of Mental Disorders (DSM-IV) fourth editionRevised ed.
Washington, DC: American Psychiatric Association.
Becker, D. R., Drake, R. E., & Naughton, W. (2005). Supported
employment for people with co-occurring disorders. Psychiatric
Rehabilitation Journal, 28, 332338.
Bush, P. W., Drake, R. E., Xie, H., McHugo, G. J., & Haslett, W. R. (2009).
The long-term impact of employment on mental health service use and
costs for persons with severe mental illness. Psychiatric Services, 60,
10241031.
Carey, K. (1993). Situational determinants of heavy drinking among college
students. Journal of Counseling Psychology, 40, 217220.
Clark, R. E., Ricketts, S. K., & McHugo, G. J. (1996). Measuring hospital
use without claims: A comparison of patient and provider reports.
Health Services Research, 31, 153169.
Drake, R. E., & Brunette, M. F. (1998). Complications of severe mental
illness related to alcohol and other drug use disorders. Recent
developments in alcoholism (Vol. XIV, Consequences of alcoholism
pp. 285299). New York: Plenum Publishing Company.
Drake, R. E., McHugo, G. J., Clark, R. E., Teague, G. B., Xie, H., Miles, K.,
et al. (1998). Assertive community treatment for patients with cooccurring severe mental illness and substance use disorder: A clinical
trial. American Journal of Orthopsychiatry, 68, 201215.
Drake, R. E., McHugo, G. J., Xie, H., Fox, L., Packard, J., & Helmstetter, B.
(2006). Ten-year recovery outcomes for clients with co-occurring
schizophrenia and substance use disorders. Schizophrenia Bulletin, 32,
464473.
Drake, R. E., McHugo, G. J., Xie, H., & Shumway, M. (2004). Three-year
outcomes of long-term patients with co-occurring bipolar and substance
use disorder. Biological Psychiatry, 56, 749756.
Drake, R. E., & Mueser, K. T. (1996). Alcohol-use disorder and severe
mental illness. Alcohol Health & Research World, 20, 8793.
Drake, R. E., Mueser, K. T., & McHugo, G. J. (1996). Clinician rating
scales: Alcohol Use Scale (AUS), Drug Use Scale (DUS), and Substance
Abuse Treatment Scale (SATS). In L. I. Sederer, & B. Dickey (Eds.),
Outcomes assessment in clinical practice (pp. 113116). Baltimore,
MD: Williams & Wilkins.
Drake, R. E., Osher, F. C., Noordsy, D. L., Hurlbut, S. C., Teague, G. B., &
Beaudett, M. S. (1990). Diagnosis of alcohol use disorders in
schizophrenia. Schizophrenia Bulletin, 16, 5767.

Drake, R. E., Osher, F. C., & Wallach, M. A. (1989). Alcohol use and abuse
in schizophrenia: A prospective community study. Journal of Nervous
and Mental Disease, 177, 408414.
Drake, R. E., & Wallach, M. A. (1993). Moderate drinking among people
with severe mental illness. Hospital and Community Psychiatry, 44,
780782.
Kessler, R. C., Nelson, C. B., McGonagle, K. A., Edlund, M. J., Frank, R.
G., & Leaf, P. J. (1996). The epidemiology of co-occurring addictive and
mental disorders: Implications for prevention and service utilization.
American Journal Orthopsychiatry, 66, 1731.
Lehman, A. F. (1988). A Quality of Life Interview for the chronically
mentally ill. Evaluation and Program Planning, 51, 5162.
Lukoff, D., Nuechterlein, K. H., & Ventura, J. (1986). Manual for expanded
Brief Psychiatric Rating Scale (BPRS). Schizophrenia Bulletin, 12,
594602.
McHugo, G. J., Drake, R. E., Burton, H. L., & Ackerson, T. H. (1995). A
scale for assessing the stage of substance abuse treatment in persons with
severe mental illness. Journal of Nervous and Mental Disease, 183,
762767.
McLellan, A. T., Luborsky, L., O'Brien, C. P., & Woody, G. E. (1980). An
improved diagnostic instrument for substance abuse patients: The
Addiction Severity Index. Journal of Nervous and Mental Disease, 168,
2633.
Mueser, K. T., Drake, R. E., & Wallach, M. A. (1998). Dual diagnosis: A
review of etiological theories. Addictive Behaviors, 23, 717734.
Osher, F. C., & Kofoed, L. L. (1989). Treatment of patients with psychiatric
and psycho-active substance use disorders. Hospital and Community
Psychiatry, 40, 10251030.
Regier, D. A., Farmer, M. E., Rae, D. S., Locke, B. Z., Keith, S. J., Judd, L.
L., et al. (1990). Comorbidity of mental disorders with alcohol and other
drug abuse. Results from the Epidemiologic Catchment Area (ECA)
Study. Journal of American Medical Association, 264, 25112518.
SAS Institute Inc. (2008). SAS/STAT User's Guide, Version 9.2. Cary, NC:
SAS Institute Inc.
Sobell, M. B., Maisto, S. A., Sobell, L. C., Copper, A. M., & Sanders, B.
(1980). Developing a prototype for evaluating alcohol treatment
effectiveness. In L. C. Sobell, M. B. Sobell, & E. Ward (Eds.), Evaluating alcohol and drug abuse treatment effectiveness (pp. 129150). New
York: Pergamon.
Vaillant, G. E. (1995). Natural history of alcoholism revisited, 2nd ed.
Cambridge, MA: Harvard University Press.
Xie, H., McHugo, G. J., Sengupta, A., & Drake, R. E. (2003). Using
discrete-time survival analysis to examine patterns of remission from
substance use disorder among persons with severe mental illness.
Mental Health Services Research, 5, 5564.