Sie sind auf Seite 1von 6

Special Section on Relapse Prevention

Substance Abuse Relapse in a Ten-Year

Prospective Follow-up of Clients With
Mental and Substance Use Disorders
Haiyi Xie, Ph.D.
Gregory J. McHugo, Ph.D.
Melinda B. Fox, M.A.
Robert E. Drake, M.D., Ph.D.

Objectives: This study addressed the rate and predictors of substance stance abuse, clinical and research at-
abuse relapse among clients with severe mental illness who had attained tention should logically focus on the
full remission from substance abuse. Methods: In a ten-year prospective goal of relapse prevention. Yet we
follow-up study of clients with co-occurring severe mental and sub- know remarkably little about this
stance use disorders, 169 clients who had attained full remission, de- area. Several recent reviews of inter-
fined according to DSM-III-R as at least six months without evidence of ventions for people with dual disor-
abuse or dependence, were identified. The Kaplan-Meier survival curve ders note that existing studies empha-
was developed to show the pattern of relapse, and a discrete-time sur- size engagement in treatment, moti-
vival analysis was used to identify predictors of relapse. Results: Ap- vation for remission, or initiation of
proximately one-third of clients who were in full remission relapsed in remission rather than emphasizing
the first year, and two-thirds relapsed over the full follow-up period. relapse prevention (8–10).
Predictors of relapse included male sex, less than a high school educa- To develop interventions and sup-
tion, living independently, and lack of continued substance abuse treat- ports for relapse prevention, one crit-
ment. Conclusions: After attaining full remission, clients with severe ical step is to understand the timing
mental disorders continue to be at risk of substance abuse relapse for and predictors of relapse (11). Thus
many years. Relapse prevention efforts should concentrate on helping the purpose of the study reported
clients to continue with substance abuse treatment as well as on devel- here was to analyze the pattern and
oping housing programs that promote recovery. (Psychiatric Services predictors of substance abuse relapse
56:1282–1287, 2005) in a ten-year prospective follow-up of
clients with co-occurring severe men-
tal and substance use disorders.

ubstance use disorder, or sub- (5–7). That is, although many clients
stance abuse, is a chronic, relaps- who enter outpatient mental health or Methods
ing illness. Cross-sectional stud- dual disorder programs achieve full re- The New Hampshire Dual Diagnosis
ies of clients with dual diagnoses con- mission within months, most of these Study, which receives funds from fed-
sistently show that their substance use clients relapse to active substance eral and state sources, is a longitudi-
disorders are often in remission (1–6). abuse. For example, two separate nal, prospective follow-up of clients
Although this finding could be taken as prospective longitudinal follow-up with dual diagnoses throughout New
evidence of recovery from substance studies (5,6) showed that the overall Hampshire. Participants in the study
use disorder, prospective longitudinal rate of active substance abuse did not have been assessed yearly for more
studies—some of which are presented change because as many individuals re- than ten years.
in this special section of Psychiatric lapsed as achieved remission.
Services—indicate that remission and Because clients with dual disorders Participants
relapse represent dynamic processes are highly prone to relapse of sub- A total of 223 outpatients with co-oc-
curring severe mental and substance
use disorders began the longitudinal
The authors are affiliated with the New Hampshire–Dartmouth Psychiatric Research study between 1989 and 1992. The
Center, 2 Whipple Place, Suite 202, Lebanon, New Hampshire 03766 (e-mail, 223 original participants were pre- This article is part of a special section on relapse prevention dominantly male (166 participants, or
among persons with co-occurring substance abuse and other mental disorders. Robert E. 74 percent), Caucasian (215 partici-
Drake, M.D., Ph.D., served as guest editor of the section. pants, or 96 percent), young
1282 PSYCHIATRIC SERVICES ♦ ♦ October 2005 Vol. 56 No. 10
(mean±SD age of 34±8.5 years), and ipants were assessed by research psy- Addiction Severity Index (ASI) (16);
unmarried (199 participants, or 89 chiatrists to confirm that they met the detailed chronological assessment of
percent). In terms of diagnoses, 119 study criteria of co-occurring severe housing history and institutional stays
(53 percent) were given a DSM-III-R mental illness (defined as schizophre- as determined by a self-report calen-
diagnosis of schizophrenia, 50 (22 per- nia, schizoaffective disorder, or bipo- dar supplemented by outpatient
cent) a diagnosis of schizoaffective dis- lar disorder plus long-term disability) records and hospital records (17); the
order, and 54 (24 percent) a diagnosis and active substance use disorder Quality of Life Interview (QOLI) (18)
of bipolar disorder. All had diagnoses (defined as abuse or dependence on to assess objective and subjective di-
of co-occurring substance use disor- alcohol or other drugs in the previous mensions of quality of life; the Ex-
ders; 166 (75 percent) had alcohol use six months). Participants were then panded Brief Psychiatric Rating Scale
disorder and 91 (42 percent) had drug assessed in terms of symptoms, func- (BPRS) (19) to assess current psychi-
use disorder. Cannabis was the most tional status, living situation, and atric symptoms; and management in-
commonly abused drug, followed by quality of life by research interview- formation systems data and the Ser-
cocaine. Other specific drugs were ers. The same two interviewers re- vice Utilization Interview (20) to as-
abused by fewer than 11 participants assessed the participants yearly for sess service use. In addition, we con-
(5 percent). At baseline, the partici- the ten-year follow-up. The partici- ducted urine toxicology screens in our
pants had high rates of recent hospital- pants were paid for their time and for laboratory by using EMIT enzyme
ization (121 participants, or 58 per- providing a urine sample at each in- immunoassay (Syva-Behring) to as-
cent), homelessness (57 participants, sess drugs of abuse. Follow-up inter-
or 26 percent), unemployment (204 views contained the same instru-
participants, or 91 percent), and other ments, without reassessment of de-
manifestations of clinical and psy- mographic and lifetime variables. Re-
chosocial instability. In all these re- liability on all scales was satisfactory,
spects, the study participants were To develop with intraclass correlation coeffi-
similar to other clients with dual diag- cients ranging from .94 to 1.00 for in-
noses in the New Hampshire mental interventions and terrater reliability and from .41 to .94
health system at that time (12). for two-week test-retest reliability.
During ten years of follow-up, 46 supports for relapse To supplement the substance abuse
participants (21 percent) were lost to self-reports, clinicians (case man-
attrition, 19 (9 percent) because they prevention, one critical agers) rated patients every six months
died and 27 (12 percent) because on three rating scales: the Alcohol
they dropped out of the study or step is to understand Use Scale (AUS), the Drug Use Scale
could not be located. Thus 177 partic- (DUS), and the Substance Abuse
ipants (86 percent of those still living) the timing and Treatment Scale (SATS). The AUS
were actively participating in the and the DUS are 5-point scales based
study at ten years. Time trends for predictors of on DSM-III-R criteria for severity of
longitudinal outcomes were similar disorder: 1, abstinence; 2, use without
for the cohort of 177 and the cohort relapse. impairment; 3, abuse; 4, dependence;
of 223, suggesting minimal bias due and 5, severe dependence (2). The
to attrition. SATS (21) is an 8-point scale that
For the analysis reported here, we rates progressive movement toward
identified 169 participants who expe- recovery from a substance use disor-
rienced a full remission from sub- terview. Clinicians (case managers) der according to Osher and Kofoed’s
stance use disorder during the ten- also assessed the participants for sub- (22) model of treatment and recov-
year follow-up. As many participants stance use disorder yearly. ery: 1 and 2 represent the early and
as possible were followed in each sub- late stages of engagement, defined as
sequent year. Measures developing a regular treatment rela-
Research psychiatrists established co- tionship; 3 and 4 represent the stages
Procedures occurring diagnoses of severe mental of persuasion, defined as developing
Participants with dual disorders were illness and substance use disorder by motivation for abstinence; 5 and 6
recruited from seven of New Hamp- using the Structured Clinical Inter- represent the stages of active treat-
shire’s ten comprehensive mental view for DSM-III-R (13). At baseline ment, defined as developing skills and
health centers through advertise- the research interview included items supports for and achieving absti-
ments, public meetings, and clinical from the Uniform Client Data Inven- nence; and 7 and 8 represent the
programs. The study was approved by tory (14) to assess demographic infor- stages of relapse prevention, defined
the New Hampshire and Dartmouth mation; the Time-Line Follow-Back as developing skills and supports to
institutional review boards. Partici- (TLFB) (15) to assess the number of maintain abstinence.
pants gave written informed consent days of alcohol and drug use over the Self-report of substance use among
at baseline and continue to do so at previous six months; the medical, le- persons with severe mental illness is
yearly follow-ups. At baseline, partic- gal, and substance use sections of the problematic because of denial; mini-
PSYCHIATRIC SERVICES ♦ ♦ October 2005 Vol. 56 No. 10 1283
Table 1 dependently rated a randomly select-
Characteristics of 169 clients with substance abuse and mental disorders who ed subgroup of 32 percent of the pa-
attained remission from substance abusea tients (433 observations of 65 pa-
tients). Intraclass correlation coeffi-
Variable N % cients were high for all three scales:
.94 on the AUS, .94 on the DUS, and
Sex, female 49 29
Education at baseline, less than high school 60 36
.93 on the SATS.
Age (mean±SD years) 36.1±8.5
Marital status at remission, married 11 7 Statistical analyses
Bipolar disorder 46 27 To examine the distribution of relapse
Antisocial personality disorder 29 19 over time, we plotted a Kaplan-Meier
Age at first psychiatric encounter (mean±SD years) 21.4±6.7
Number of psychiatric hospitalizations at baseline 3.81±1.6
survival curve (25). To predict dura-
Alcohol use disorder at baseline 120 71 tion of remission, we modeled sur-
Drug use disorder at baseline 65 39 vival analysis by using discrete-time
Global Assessment Scaleb 47.4±12.5 survival analysis methods. Survival
BPRS total scorec 40.7±11.8 analysis is a technique for studying
Lived in a group home for more than 60 days in the past year 43 26
Lived independently at least 80 percent of the time in the past year 57 34
time to an event. However, because
Had contact with friends who did not use substances in the past year 53 34 duration cannot be modeled directly,
Currently working 34 22 hazard or risk of an event of interest
Received outpatient substance abuse treatment in the past two weeks 62 44 over time, which is a mathematical
transformation of duration, is usually
At the time of first full remission, except where “baseline” is indicated
b Possible scores range from 1 to 100, with higher scores indicating a higher level of functioning. modeled (26).
c Possible scores range from 24 to 168, with higher scores indicating greater severity of symptoms. The discrete-time survival model is
a modified logistic regression in which
hazard is defined as the conditional
mization; failure to perceive that sub- all available data on substance use probability that an event will occur in
stance use disorder is related to poor disorder (from the ASI, the TLFB, a particular time interval given that it
adjustment; distortions due to cogni- clinician ratings, and urine drug has not yet occurred (27,28). In our
tive, psychotic, and affective factors; screens) to establish separate ratings analysis, outcome is a logit hazard (lo-
and the inappropriateness of tradi- on the AUS, the DUS, and the SATS. gistic transformation of hazard) for re-
tional measures for this population Our basic rule was that all three re- lapse to occur. We examined several
(23). Self-report data are therefore sources had to be consistent to estab- predictors, drawn from theory and
often supplemented with clinical rat- lish a positive rating, such as remis- empirical research: time to first full
ings, laboratory measures, or multiple sion (24). Thus any evidence of abuse remission, sex, education, age, marital
instruments to attain more valid as- or dependence was taken as an indi- status, psychiatric diagnosis, type of
sessments. In this study, a team of cator of relapse. To determine the in- substance abuse (alcohol, drug, or
three independent raters considered terrater reliabilities, researchers in- both), antisocial personality disorder,
number of days of independent living,
number of days living in a group
home, frequency of contact with indi-
Figure 1 viduals who do not use substances,
Proportion still in remission (survival function) employment, and participation in out-
patient substance abuse treatment
1.0 (individual, group, or self-help). We
.9 considered relationships for which p
was .05 or less to be significant and
those for which p was between .05 and
.7 .10 to be marginally significant.

During the course of ten-year follow-
.4 up, 169 participants obtained at least
.3 one full remission of substance use
disorder, according to the re-
searchers’ ratings described above
.1 and DSM-III-R criteria of at least six
0 months without evidence of abuse or
0 1 2 3 4 5 6 7 8 9 10 dependence. Because these assess-
N=169 N=117 N=89 N=74 N=61 N=57 N=49 N=39 N=34 N=23 ments were done yearly, participants
Year and number of clients in remission at follow-up were in these first remissions for a pe-
1284 PSYCHIATRIC SERVICES ♦ ♦ October 2005 Vol. 56 No. 10
riod of six to 12 months. Characteris- Table 2
tics of the 169 participants who at- Discrete-time survival analysis model of hazard of the first relapse in a sample of
tained full remission are summarized 128 clients with substance abuse and mental disorders
in Table 1. The data represent char-
acteristics at the time of first full re- Logit
mission rather than at baseline for the Variable hazard SE p OR
larger study, because the date of the Intercept –1.96 .92 .033 .14
first full remission was used as the Time to first remission (years) .16 .08 .060 1.17
starting point for the survival analysis. Sex, female –1.19 .32 <.001 .30
Of the 169 participants who achieved Education, less then high school .60 .30 .049 1.82
remission, we had at least one year of Age (years) .03 .02 .099 1.03
Marital status (not married) –.72 .63 .253 .49
follow-up data for 158. Bipolar disorder –.34 .34 .326 .71
No diagnosis of antisocial personality disorder .58 .36 .107 1.78
Sustainability of remission Lived in a group home for more than 60 days
Figure 1 shows the proportion of in the past year .07 .37 .847 1.07
clients with dual diagnoses who re- Lived independently at least 80 percent
of the time in the past year .99 .33 .002 2.70
mained in full remission as a survival Had contact with friends who did not use
function. Relapse was especially com- substances .02 .30 .952 1.02
mon in the first year (almost one- Currently working –.64 .38 .095 .53
third). By three years, approximately Received outpatient substance abuse treatment
half had relapsed, and by nine years, in the past two weeks –.68 .28 .015 .51
Type of substance abuse –.04 .15 .805 .96
more than two-thirds had relapsed.
Because of these relapses as well as
variable durations of follow-up and
missing interviews, the number of re-
maining participants was lower at ently was attenuated, which suggests and those who have not participated
each assessment. that the effect on relapse is significant recently in substance abuse treat-
only when other variables are con- ment. Time to remission may indicate
Predictors of trolled for. The plots also suggested more severe substance abuse, less in-
duration of remission that there might be interactions ternal capacity for remission, or fewer
For the discrete-time survival analysis among these covariates. For example, external supports for remission. The
of first relapse, the number of partici- the difference due to residential set- marginal relationship with employ-
pants was reduced to 128 because of ting appeared larger for females than ment suggests that having a construc-
missing data for several covariates. for males. However intriguing these tive daily activity is a critical support,
Table 2 shows several factors that pre- refinements may be, we reported as many clients report (29). Similarly,
dicted the hazard of relapse. Clients only the main effect model, because clients who have less education may
who were male, who had less educa- the sample was not large enough for have fewer internal or external re-
tion, who lived independently, and who us to explore them rigorously. sources to help them sustain remis-
did not participate in outpatient sub- sion. Unemployment and low levels
stance abuse treatment of some type Discussion of education could also be markers
were more likely to relapse. Longer The primary finding of this study was for more severe neuropsychological
time to first remission, older age, and that we confirmed the chronic, fluc- deficits. Men with severe mental ill-
lack of employment were marginally tuating nature of substance use disor- ness generally have worse outcomes
related to relapse. Type of mental ill- der among people with severe mental than women (30), and men with co-
ness, including co-occurring antisocial illness. More than 75 percent of the occurring disorders also appear to do
personality disorder, and type of sub- study group achieved full remission of worse than women.
stance use disorder (alcohol, other six to 12 months. However, relapse More interesting, because of their
drugs, or both) did not predict relapse. was also very likely, with the great ma- possibilities for intervention, are the
To understand the effect of predic- jority of participants relapsing at findings related to living settings and
tors of relapse (sex, residential set- some point, often within the first year ongoing substance abuse treatment.
ting, and participation in outpatient of attaining full remission. Relapse Living situation is of great interest,
substance abuse treatment), separate became less likely over time, particu- because housing is such a common
survival curves were plotted for each larly after two years of remission. problem for clients with co-occurring
variable and for combinations of the The predictors of relapse suggest disorders. At this point, the literature
three variables. The plots were con- that some clients are more vulnerable shows that the great majority of per-
sistent with the modeling results re- to relapse than others: those who are sons with severe mental illness—per-
ported in Table 2, but several refine- males, those who take longer to haps 85 percent or more—do well in
ments were suggested. For example, achieve remission, those who have independent housing with supports,
the difference between living inde- less education, those who are unem- which is called supported housing
pendently and not living independ- ployed, those who live independently, (31,32). At the same time, this litera-
PSYCHIATRIC SERVICES ♦ ♦ October 2005 Vol. 56 No. 10 1285
6. Cuffel BJ, Chase P: Remission and relapse
ture shows that clients with dual dis- expelled from these critical supports. of substance use disorder in schizophrenia:
orders tend to do poorly in supported Several other caveats warrant at- results from a one-year prospective study.
housing programs. Our study con- tention. This study was conducted in Journal of Nervous and Mental Disease
182:342–348, 1994
firms that these clients have difficul- a predominantly rural state, with rel-
ties in independent housing, even if atively low availability of illicit drugs 7. Rollins AL, O’Neill SJ, Davis KE, et al:
Substance abuse relapse and factors associ-
they have already achieved full remis- other than cannabis, with little racial ated with relapse in an inner-city sample of
sion from their substance use disor- and cultural diversity, with a relative- patients with dual diagnoses. Psychiatric
ders. The issue may be that these ly competent mental health system, Services 56:1274–1281, 2005
clients are forced by poverty and and under other conditions that may 8. Brunette MF, Mueser KT, Drake RE: A re-
housing policies to live in high-risk have limited the generalizability of view of research on residential programs
for people with severe mental illness and
neighborhoods, where they remain our results. Relapse was assessed co-occurring substance use disorders. Drug
extremely vulnerable to substance yearly at six-month intervals, and and Alcohol Review 23:471–481, 2004
abuse and other endemic problems. some relapses may have been missed. 9. Drake RE, Mueser KT, Brunette M, et al:
Research indicates that some clients Nevertheless, the study also has nu- A review of treatments for people with se-
are aware of the dangers of independ- merous strengths, including the high vere mental illness and co-occurring sub-
stance use disorder. Psychiatric Rehabilita-
ent living in high-risk neighborhoods rate of long-term follow-up, the mul- tion Journal 27:360–374, 2004
and seek out residential programs and timodal assessments of substance
10. Mueser KT, Drake RE, Sigmon S, et al:
other protective environments abuse, and the interdisciplinary re- Psychosocial interventions for adults with
(33,34). Other active substance search team. The results obtained severe mental illnesses and co-occurring
abusers seek out independent living here need to be tested among clients substance use disorders: a review of specif-
ic interventions. Journal of Dual Disorders
and do poorly (35). More recently, a with dual diagnoses who are living in 1:57–82, 2005
series of controlled studies estab- urban areas, where there are much
11. Marlatt GA, Gordon JR: Relapse Preven-
lished that clients with co-occurring higher rates of illicit drug use, racial tion: Maintenance Strategies in the Treat-
disorders tend to do well in long-term diversity, and involvement with the ment of Addictive Behaviors. New York,
residential programs (8). Clearly, criminal justice system (40). Guilford, 1985
there is a need to provide residential 12. Drake RE, McHugo GJ, Clark RE, et al:
programs that foster recovery from Conclusions Assertive community treatment for patients
with co-occurring severe mental illness and
substance abuse. This study provided several initial substance use disorder: a clinical trial.
Continued involvement in sub- findings regarding factors that may American Journal of Orthopsychiatry
stance abuse treatment has been enable clients with dual diagnoses to 68:201–215, 1998
identified many times as an important sustain remission from substance use 13. Spitzer R, Williams J, Gibbon M, et al:
factor in stable remission, abstinence, disorders. Individuals who are Structured Clinical Interview for DSM-III-
R–Patient Version (SCID-P). New York,
and recovery (36–39). As McLellan amenable to intervention, such as New York State Psychiatric Institute, Bio-
(36) describes it, substance abuse is employment, safe housing, and con- metrics Research Department, 1988
like other chronic illnesses in that tinued substance abuse treatment, 14. Tessler R, Goldman H: The Chronically
people need treatment and other sup- should be considered for relapse pre- Mentally Ill: Assessing Community Sup-
ports to manage their illnesses over a vention planning. ♦ port Programs. Cambridge, Mass, Harper
and Rowe, 1982
lifetime, not just during episodes of
symptoms. Thus long-term involve- References 15. Sobell MB, Maisto SA, Sobell LC, et al:
Developing a prototype for evaluating alco-
ment with self-help and treatment 1. Dixon L, McNary S, Lehman AF: Remis-
hol treatment effectiveness, in Evaluating
should be part of the expectation of sion of substance use disorder among psy-
Alcohol and Drug Abuse Treatment Effec-
chiatric inpatients with mental illness.
care, just as it is for diabetes, hyper- American Journal of Psychiatry 155:239–
tiveness. Edited by Sobell LC, Sobell MB,
Ward E. New York, Pergamon, 1980
tension, and many other chronic 243, 1998
medical conditions. For clients with 2. Drake RE, Osher FC, Noordsy DL, et al:
16. McLellan AT, Luborsky L, O’Brien CP, et
al: An improved diagnostic instrument for
dual disorders, we need more infor- Diagnosis of alcohol use disorders in schiz-
substance abuse patients: the Addiction
mation about what kinds of long-term ophrenia. Schizophrenia Bulletin 16:57–67,
Severity Index. Journal of Nervous and
treatments and supports are effec- Mental Disease 168:26–33, 1980
tive—for example, peer groups, dual 3. Graham HL, Maslin J, Copello A, et al:
17. Clark RE, Ricketts SK, McHugo GJ: Mea-
Drug and alcohol problems amongst indi-
recovery groups, case management, viduals with severe mental health problems
suring hospital use without claims: a com-
parison of patient and provider reports.
and family help. in an inner city area of the UK. Social Psy-
Health Services Research 31:153–169,
One important caveat regarding chiatry and Psychiatric Epidemiology
36:448–455, 2001
housing and ongoing substance abuse 18. Lehman AF: A Quality of Life Interview for
treatment is that the relationships 4. Maisto SA, Carey MP, Carey KB, et al: Use
the chronically mentally ill. Evaluation and
of the AUDIT and the DAST-10 to identify
with outcomes could be circular. alcohol and drug use disorders among
Program Planning 51:51–62, 1988
Clients who are more motivated to adults with a severe and persistent mental 19. Lukoff D, Nuechterlein KH, Ventura J:
pursue and maintain abstinence may illness. Psychological Assessment 12:186– Manual for expanded brief psychiatric rat-
192, 2000 ing scale (BPRS). Schizophrenia Bulletin
well seek residential protection and 12:594–602, 1986
ongoing treatments. At the same time, 5. Bartels SJ, Drake RE, Wallach MA: Long-
term course of substance use disorders in 20. Clark R, Teague G, Ricketts S, et al: Mea-
those who are unmotivated, who re- severe mental illness. Psychiatric Services suring resource use in economic evalua-
lapse, or who otherwise falter may be 46:248–251, 1995 tions: determining the social costs of men-

1286 PSYCHIATRIC SERVICES ♦ ♦ October 2005 Vol. 56 No. 10

tal illness. Journal of Mental Health Ad- chological Association, 1991 27:140–150, 2003
ministration 21:32–41, 1994
27. Singer J, Willett J: It’s about time: using 34. Drake RE, Wallach MA: Mental patients’
21. McHugo GJ, Drake RE, Burton HL, et al: discrete-time survival analysis to study du- attraction to the hospital: correlates of liv-
A scale for assessing the stage of substance ration and the timing of events. Journal of ing preference. Community Mental Health
abuse treatment in persons with severe Educational Statistics 18:155–195, 1993 Journal 28:5–12, 1992
mental illness. Journal of Nervous and
28. Xie H, McHugo GJ, Sengupta A, et al: Us- 35. Goldfinger SM, Schutt RK, Seidman LJ, et
Mental Disease 183:762–767, 1995
ing discrete-time survival analysis to exam- al: Self-report and observer measures of
22. Osher FC, Kofoed LL: Treatment of pa- ine patterns of remission from substance substance abuse among homeless mentally
tients with psychiatric and psychoactive use disorder among persons with severe ill persons in the cross-section and over
substance use disorders. Hospital and mental illness. Mental Health Services Re- time. Journal of Nervous and Mental Dis-
Community Psychiatry 40:1025–1030, 1989 search 5:55–64, 2003 ease 184:667–672, 1996

23. Drake RE, Alterman AI, Rosenberg SR: 29. Alverson H, Alverson M, Drake RE: An 36. McLellan AT, Lewis DC, O’Brien CP, et al:
Detection of substance use disorders in se- ethnographic study of the longitudinal Drug dependence, a chronic medical ill-
verely mentally ill patients. Community course of substance abuse among people ness. JAMA 284:1689–1695, 2000
Mental Health Journal 29:175–192, 1993 with severe mental illness. Community
Mental Health Journal 36:557–569, 2000 37. Moos RH, Schaefer J, Andrassy J, et al:
Outpatient mental health care, self-help
24. Drake RE, Mueser KT, McHugo GJ: Using
30. Angermeyer MC, Kuhn L, Goldstein JM: groups, and patients’ one-year treatment
clinician rating scales to assess substance
Gender and the course of schizophrenia: outcomes. Journal of Clinical Psychology
use among persons with severe mental ill-
differences in treated outcomes. Schizo- 57:273–287, 2001
ness, in Outcomes Assessment in Clinical
phrenia Bulletin 16:293–307, 1990
Practice. Edited by Sederer LI, Dickey B. 38. Simpson DD, Joe GW, Brown BS: Treat-
Baltimore, Williams and Wilkins, 1995 31. Newman SJ: Housing attributes and serious ment retention and follow-up outcomes in
mental illness: implications for research the Drug Abuse Treatment Outcome Study
25. Kaplan EL, Meier P: Nonparametric esti- and practice. Psychiatric Services 52: (DATOS). Psychology of Addictive Behav-
mation from incomplete observations. 1309–1317, 2001 iors 11:294–307, 1996
Journal of the American Statistical Associa-
tion 53:457–481, 1958 32. Rog DJ: The evidence on supported hous- 39. Vaillant GE: Natural History of Alcoholism
ing. Psychiatric Rehabilitation Journal Revisited. Cambridge, Mass, Harvard Uni-
26. Willett J, Singer J: How long did it take? 27:334–344, 2004 versity Press, 1995
Using survival analysis in psychological re-
search, in Best Methods for the Analysis of 33. McCoy ML, Devitt T, Clay R, et al: Gaining 40. Mueser KT, Essock SM, Drake RE, et al:
Change: Recent Advances, Unanswered insight: who benefits from residential, inte- Rural and urban differences in patients
Questions, Future Directions. Edited by grated treatment for people with dual diag- with a dual diagnosis. Schizophrenia Re-
Collins L. Washington, DC, American Psy- noses? Psychiatric Rehabilitation Journal search 48:93–107, 2001

Free Subscription to Psychiatric Services

U.S. and Canadian members of the American Psychiatric Association can receive a
free subscription to Psychiatric Services as a benefit of their membership.
To take advantage of this benefit, simply visit the APA Web site at www. Print out and complete the one-page form, then fax or mail it as in-
structed on the form. Because of postal regulations, your signature on the form
is required. Thus requests cannot be taken over the telephone or by e-mail. The
first issue of your free subscription to Psychiatric Services will be mailed to you
in four to six weeks.
In addition, with your first issue of Psychiatric Services, you will receive in-
structions for activating your free online subscription at http://ps.psychiatry
Because of mailing costs, the free print subscription is not available to interna-
tional APA members. However, after requesting a free subscription (see above),
international members have online-only access (
Click on “Subscriptions” and on “Activate Your Member Subscription.” Members
can verify their member number or obtain help for activation problems by send-
ing an e-mail to

PSYCHIATRIC SERVICES ♦ ♦ October 2005 Vol. 56 No. 10 1287