BRIEF REPORT
The Role of the Family and Improvement in Treatment
Maintenance, Adherence, and Outcome for Schizophrenia
Ira D. Glick, MD,* Anya H. Stekoll, MS, and Spencer Hays, MA
Introduction: In the context of a large, random assignment, controlled
study evaluating the relative effectiveness and safety of antipsychotic
medication (CATIE), we examined the relationship between treatment
outcome and 2 family variables: their presence and their ability to sup-
port treatment adherence.
Methods: Post hoc, we assessed the 50 study patients (40 of whom had
families) and their families by dividing them into 2 groups. The rst had
a family/signicant other, available and mostly supportive, to work col-
laboratively on adherence with the treatment team (n = 27). The second
group either did not have the family/signicant other or, if they did,
lacked support for long-term maintenance (n = 23). Next, we examined
outcome on 2 measures: study completion (vs discontinuation) and
global outcome.
Results: Of 27 patients with available/supportive families, 23 remained
in treatment for the full study course. In contrast, 13 of 23 patients, who
were discontinued or dropped out, either did not have families or, if they
had them, were unable to support adherence (P G 0.01). As to global
outcome, 24 of the 27 patients who had supportive families improved,
compared with only 9 of the 23 of the other group (P G 0.001).
Discussion: In summary, in the context of a large medication efcacy-
effectiveness trial, we present data suggesting that having a family
available and supportive (regardless of the interpersonal issues between
patient and family) improves outcome mediated by improving long-term
adherence.
Key Words: schizophrenia, family, treatment maintenance, adherence
(J Clin Psychopharmacol 2011;31: 82Y85)
C urrent practice guidelines for the treatment of patients
with schizophrenia suggest a combination of antipsychotic
medication plus individual and family interventions, especially
psychoeducation.1 Psychoeducation has a component that is
aimed at reducing expressed emotion (i.e., the level of a
family_s criticism of and involvement with or support for the
identied patient) and, in combination with other family in-
terventions, are associated with better long-term outcomes. 2Y4
The crucial issue, of course, is that, in part, these family variables
are thought to be mediating variables related to antipsychotic
medication adherence, and degree of adherence is positively
correlated to outcome.5 The difculty of working with a chronic
psychotic patient is well known, especially those patients with
From the *Psychiatry and Behavioral Sciences, Pacific Graduate School of
Psychology, Stanford Psy.D. Consortium, and Department of Statistics and
Operations Research, University of North Carolina at Chapel Hill, Schizo-
phrenia Research Program, Stanford University School of Medicine, Stan-
ford, CA.
Received May 10, 2010; accepted after revision November 08, 2010.
Reprints: Ira D. Glick, MD, Psychiatry and Behavioral Sciences,
Schizophrenia Research Program, Stanford University School of
Medicine, Stanford, CA (e-mail: iraglick@standford.edu).
This study was supported in part by a National Institute of Mental Health
contract MH90001.
Copyright * 2011 by Lippincott Williams & Wilkins
ISSN: 0271-0749
DOI: 10.1097/JCP.0b013e31820597fa
82 www.psychopharmacology.com Journal of Clinical Psychopharmacology
Copyright 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
sporadic substance abuse/dependence. However, what is not
clear is whether medication can be effective if there is no family
involved or, if present, the patient-family interaction prevents
them from being supportive. In fact, in practice, this issue is
usually not appreciated or even commonly part of the clinical
calculation of whether effective treatment is even possible.
Unexpectedly, CATIEVthe recent, large-scale, multi-site
National Institute of Mental Health trial of effectiveness of
antipsychotics, provided the opportunity to examine the issue of
the relationship of family support to drug compliance and out-
come.6 The CATIE study was designed to keep families involved
throughout treatment using an individually tailored education
plan. The study focused on all cause discontinuation as the
single, best outcome measure of effectiveness6 but had other
secondary outcome measures as well. As one of the CATIE sites
with a large number of enrolled subjects, as we progressed, we
noted 2 patterns of patient discontinuation or dropout. Patients
either had families with low support and/or high conict or had
no family at all, and both patterns seemed to be associated with
either early dropout or worse outcomes compared with the rest of
our sample. Accordingly, after the study ended, we decided to
systematically examine these family variables at our site to de-
termine the correlation, if any, with outcome.
Post hoc, we hypothesized that the presence of a family
and/or a relatively higher level of family support especially for
medication adherence correlated positively with better out-
comes and vice versa. Conversely, not having a family or having
lower levels of availability/support were predicted to correlate
with early dropout and/or worse outcomes. Although the crucial
role of the family in improving outcome of patients with schizo-
phrenia has been well studied, described, and included in most
American Psychiatric Association and other guidelines, com-
monly patients on antipsychotic medication are prescribed drugs
and managed without regular and continuous family involve-
ment. We recognize, of course, that there are many practical rea-
sons for this lack of involvement. A patient may not have any
family (broadly dened), or if they do, the patient-family inter-
action patternsVespecially during the acute phases of the illness
when patients may be fearful and paranoidVmay prevent the
family from being supportive or working collaboratively.
The central question is whether it is possible to achieve
reasonable outcomes without a family, specically a supportive
family, regardless of good medication practices. This study ex-
amines this question.
METHODOLOGY
Study Setting and Design
The CATIE study was initiated by the National Institute
of Mental Health to compare the relative effectiveness of anti-
psychotic drugs over 18 months in 3 phases with all-clause
discontinuation as the primary outcome measure.6 Of note to this
report, the CATIE design specically included regular, ongoing
family education and psychoeducation using a specially de-
signed, detailed manual for the study. Further design details can
be found in the 2003 publication.6
& Volume 31, Number 1, February 2011
Journal of Clinical Psychopharmacology & Volume 31, Number 1, February 2011 Treatment Outcome and Family Variables

TABLE 1. Sample Demographics, Family Variables, and Outcome of Unimproved Study Patients

Participant Number Age Sex Marital Status Race Family Variables Treatment Outcome
1 42 Male Married Asian Very supportive Dropout, worse
2 51 Male Single Asian No family available Dropout, no change
3 36 Male Single White No family available Dropout, no change
4 39 Male Single White High conflict Dropout, worse
5 20 Male Single White Very supportive Dropout, no change
6 36 Female Single White No family available Dropout, no change
7 44 Male Single White High conflict Completed, no change
8 31 Male Single White No family available Dropout, no change
9 18 Female Single White High conflict Dropout, no change
10 29 Male Single White High conflict Dropout, worse
11 50 Male Single White High conflict Dropout, no change
12 32 Female Single White No family available Dropout, no change
13 38 Male Married White Very supportive Completed, no change
14 33 Female Married White Neutral Dropout, worse
15 23 Male Single White High conflict Dropout, worse
16 46 Female Single African American High conflict Dropout, no change
17 23 Male Single White High conflict Dropout, worse

Subjects assessment was dichotomized into improved versus no change or

Inclusion criteria required all participants to have a formal
diagnosis of schizophrenia as determined by the Structured Clin-
ical Interview for Diagnostic and Statistical Manual of Mental
Disorders, Fourth Edition, and to be between the ages of 18 and
65 years. As one of the largest enrollers in CATIE, our study
group consisted of the 50 patients randomized at our site. De-
mographically, our sample was representative of the multisite
CATIE sample. Table 1 lists the demographics of those patients
who were discontinued/dropped out and were unimproved/worse.
Most had chronic schizophrenia. Most were white male sub-
jects, single or divorced, with a mean age of 35 years, mode 22.
Two of the patients regularly abused marijuana. Forty of the 50
had families, or signicant others involved to varying degrees.
Thirty-three of 50 patients completed either Phase I or II of the
study; the other 17 were early dropouts or discontinued.
Evaluation
There were a wide range of efcacy, effectiveness, and
safety ratings embedded in the CATIE study. For this report,
post hoc our treatment team, by group consensus, used the data
we collected about treatment course and family patterns at
the time of study enrollment and over the course of study visits,
to divide the 50 families into 2 groups. The rst group had a
family/signicant other, available and mostly supportive, to work
collaboratively to facilitate adherence with the treatment team
(n = 27). The second group was more heterogeneousVeither
they did not have the family/signicant other support because
there was no one available (the patient was essentially on his
own, n = 10), or more commonly, the patient-family unit were in
a highly conicted struggle (usually because the patient-
usually in his 20s or early 30sVwas psychotic and irritable), or
the family would not/could not come in regularly for treatment
visits. By way of case description, we refer to this group as
nonsupportive, but obviously, this phrase does not entirely
capture the complexity of the patient-family interaction (or lack
of it). Nevertheless, the distinction between groups was marked.
Also, at the conclusion of the study, for patients who were
randomized, we made a global rating of outcome at the time of
their dropout or at the time they nished with the study. This
worsened at the end of their duration in the study. Here, we used
not only our treatment team ratings of outcome but also, because
our rating may have been biased, the outcome ratings in the
CATIE database (see below) which come from blinded raters.
We recognize the rating of unimproved or worse was partially
articially inated by those patients who dropped out as well as
those who stayed in the study and did not improve.
Finally, we correlated study completion and outcome with
the 2 patient-family groupings.
Although our focus was on family presence and sup-
portVthe CATIE evaluative measures included a global measure
(the Clinical Global Impression [CGI]) and a Quality of Life
(QOL) assessment derived from information from the patient.
We examined QOL data in the context of understanding our
results. The QOL is a 31-item scale with multiple questions
assessing an individual_s perception of their degree of enjoyment
and satisfaction experienced in everyday life. The QOL assess-
ment included several questions that indirectly spoke to the level
of family support. One of the questions was: In the past month,
how often did you spend time with one or more family mem-
bers? Patient choices included being reminded of appoint-
ments, being taken to appointments, being reminded to take
medication, and being given their medication. These questions
were not originally intended to be used as a measure for level
of family support; however, they seem to capture a signicant
TABLE 2. Family Presence and Support Versus Patient
Completion
Patient
Family Completed Discontinued/Dropout Total
Present and 23 4 27
supportive
Not present or 10 13 23
nonsupportive
Total 33 17 50
W 2
1 = 9.63, P G 0.01.

* 2011 Lippincott Williams & Wilkins www.psychopharmacology.com 83

Copyright 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Glick et al Journal of Clinical Psychopharmacology
TABLE 3. Family Presence and Support Versus Patient
Outcome
Patient Outcome
Family Improved No Change/Worse Total
Present and supportive 24 3 27
Not present or 9 14 23
nonsupportive
Total 33 17 50
W21 = 13.7, P G 0.001.
amount of information about what kind of support was provided
and the nature of the patient-family interaction. In addition, we
also examined family involvement as it effected compliance, that
is, did higher involvement correlate with better compliance?
Data Analysis
The Pearson W2 test of independence was used to measure
the association between the family variables and the patient
outcome variables, that is, study completion and improvement.
RESULTS
Remaining in Treatment
The rst question addressed was whether having a sup-
portive family (versus either not having a family or the family not
being available or the interaction being conicted) is associated
with treatment completion. Table 2 reveals that of the 27 patients
with supportive families, 23 (85%) remained in treatment for the
entire study. In contrast, 13 (56%) of the 23 patients of the
second group dropped out (P G 0.01).
Treatment Outcome
The next issue is whether having a supportive family is
associated with better global outcome. Table 3 reveals that 24
(89%) of the 27 patients with such families improved. On the
other hand, 14 (61%) of the 23 patients of the nonavailable/
nonsupportive group showed no change or worsened postbase-
line (P G 0.001). We also ran this analysis using the all-cause
discontinuation rates for the 48 patients from the CATIE data
base. Here, the results were the sameVW21 was 11.7, P G 0.001.
Patient and Family Contact and Compliance
To better understand these ndings, we examined the cor-
relation of family support with CGI and QOL scales. Interest-
ingly, the less the patient improved (as shown on the CGI), the
more the patient-family contact. Our interpretation was that
when patients were doing badly, that is, psychotic or more
symptomatic or irritable, they required more family help. On the
QOL scale, we found, in fact, that when the patients were doing
well, they had (as expected) less contact with the family. The
QOL item on family contact and medication compliance showed
that patient improvement correlated with the family_s help in
taking medication.
DISCUSSION
Using the patient sample from our site, a subsample from a
large-scale, multisite study of the effectiveness of antipsychotics,
we found having both a family (and even more importantly, their
being supportive of adherence), positively was associated not
only with study completion but also with good outcome. These
ndings are consistent with the literature suggesting the impor-
84 www.psychopharmacology.com
Copyright 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
& Volume 31, Number 1, February 2011
tance of a supportive family environment for patients with
schizophrenia.5 In fact, family interventions have been found to
improve medication compliance and reduce relapse rates by
approximately 20%.5 Likewise, medication compliance is im-
proved when patients have help from family caregivers.7 On the
other hand, not having a family8 or having a nonsupportive/
available/conicted family situation was associated with early
termination and poorer outcome. Examining these family units,
we found that usually the problem was that either the patient
(usually in his 20s or early 30s) was psychotic, paranoid, and/or
irritable when interacting with the family, or if older, the fam-
ily stayed away, that is, worn-out from trying to help their
chronically ill child or spouse. These outcomes seemed true
regardless of which antipsychotic the patient was taking (al-
though post hoc we did not, and could not, because of lack of
statistical power to examine this question).
There are a number of limitations to our study. First, this
was not a prospective, controlled, double-blind, random assign-
ment study focusing on the family. It was a correlative study;
therefore, it is not clear what is the cause and what is the ef-
fect. For example, nonsupportive families could have seemed to
be so because of the patient_s problematic, symptomatic beha-
viors (being irritable or psychotic) or nonresponse to treatment.
In fact, all of the patient-family units, who were judged
antagonistic or nonsupportive with high expressed emotion,
had identied patient spouses or children who were psychotic,
denied their illness, and/or refused medication. Our impression
was that the family_s behavior remained relatively consistent
from the study onset, and as we noted (above), the more the
family contact, usually the worse the patient was doing. Second,
the number of subjects was small (compared with a prospective,
random assignment efcacy study), although relatively large
for studies involving families of patients with schizophrenia.
Furthermore, our results are very consistent with the family lit-
erature suggesting that the results are valid. Third, although
we were able to accurately determine dropout/discontinuation,
there may have been bias in our nonblinded judgment of each
family_s support level and of patient outcome as we rated
retrospectively (although to us the judgments seemed relatively
straightforward and were consistent with the CATIE data recor-
ded from blind raters). Finally, which antipsychotic a patient was
taking might have confounded these ndingsValthough this
seemed unlikely, given the efcacy ndings from CATIE.
In summary, in the context of a large medication trial,
we have found data suggesting that, regardless of the patient
symptoms, having a supportive family, able to work collabo-
ratively with the treatment team from the onset of treatment
through stabilization, improves outcome. Simply put, without a
supportive family, patients will not stay on their medication or
remain in treatment even in well-staffed settings. These data are
consistent with the recent schizophrenia guidelines for family
education and psychoeducation.1 Although this nding may not
be totally new, in the context of medication adherence, it is at
the very least, an important replication.
Further research on these issues is needed because it might
be that family presence and support are even more important
for medication adherence and outcome than what is currently
believed (i.e., adequate medication is usually considered as
(by far) the most crucial element in the treatment equation).
Although this statement is still true, the family provides the
context for the medication maintenance. The message for those
prescribing antipsychotic medication isVit may be that it is
necessary (almost mandatory) for patients with schizophrenia to
have a supportive family or friend or case worker with relatively
low expressed emotion or conict in the family to achieve
* 2011 Lippincott Williams & Wilkins
Journal of Clinical Psychopharmacology & Volume 31, Number 1, February 2011 Treatment Outcome and Family Variables

and maintain stabilization and a reasonably good longer-term
outcome. It is possible that too much patient/family contact is
toxic, but this is not clear from our ndings. This may be
true, regardless of which antipsychotic is prescribed. This also
may be true for disorders other than schizophrenia such as bi-
polar disorder, in which patients are cognitively impaired with
mood elevation and/or grandiosity and would need family help
to adhere to treatment and to function.9,10
ACKNOWLEDGMENT
The authors thank Oxana Ivanova, MD, and Sudeepthi
Prasad, MD, for help in data collection and Lisa Dixon, MD,
and Donald Klein, MD, for reviewing the manuscript.
AUTHOR DISCLOSURE INFORMATION
The authors declare no conicts of interest.
REFERENCES
1. Lehman AF, Lieberman JA, Dixon LB, et al. Practice Guidelines for the
Treatment of Patients with Schizophrenia, Second Edition. Am J
Psychiatry. 2004;161(suppl):1Y56.
2. Dixon L, McFarlane WR, Lefley H, et al. Evidence-based practices
for services to families of people with psychiatric disabilities.
Psychiatr Serv. 2001;52:902Y910.
3. Murray-Swank AB, Dixon L. Family Psychoeducation as an
Evidence-Based Practice. CNS Spectr. 2004;9:905Y908.
4. Glick ID, Berman EM, Clarkin JF, et al. Marital and Family Therapy,
Fourth Edition. Arlington, VA: American Psychiatric Press Inc;
2000:479Y484.
5. Leucht S, Heres S. Epidemiology, clinical consequences, and
psychosocial treatment of nonadherence in schizophrenia. J Clin
Psychiatry. 2006;67(suppl 5):3Y8.
6. Stroup ST, McEvoy JP, Swartz MS, et al. The National Institute of
Mental Health Clinical Antipsychotic Trials of Intervention
Effectiveness (CATIE) Project: Schizophrenia Trial Design and
Protocol Development. Schizophr Bull. 2003;29:15Y31.
7. Ramirez G, Chang CL, Young JS, et al. Family support predicts
medication usage among Mexican American individuals with
schizophrenia. Soc Psychiatr Psychiatr Epidemiol. 2006;41:624Y631.
8. Dixon L, Goldberg R, Iannone V, et al. Use of a critical time intervention
to promote continuity of care after psychiatric inpatient
hospitalization. Psychiatr Serv. 2009;60:451Y458.
9. Cohen A, Patel V, Thara R, et al. Questioning an axiom: better
prognosis for schizophrenia in the developing world? Schizophr Bull.
2008;34:229Y244.
10. Drake RE, Goldman HH, Leff HS, et al. Implementing evidence-based
practices in routine mental health service settings. Psychiatr Serv.
2001;52:179Y182.

* 2011 Lippincott Williams & Wilkins www.psychopharmacology.com 85

Copyright 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.