Journal of Affective Disorders 112 (2009) 256 – 261
Is the collaborative chronic care model effective for patients with
bipolar disorder and co-occurring conditions? ☆
Amy M. Kilbourne a,⁎, Kousick Biswas b , Paul A. Pirraglia c , Martha Sajatovic d ,
William O. Williford b , Mark S. Bauer e
VA Ann Arbor Serious Mental Illness Treatment Research and Evaluation Center and Department of Psychiatry,
University of Michigan; Ann Arbor, MI, USA
Cooperative Studies Program Coordinating Center at the VA Maryland Health Care System at Perry Point, MD, USA
Providence VA Medical Center and Brown University; Providence, RI, USA
Case Western Reserve University School of Medicine, University Hospitals of Cleveland; Cleveland, OH, USA
VA Boston Healthcare System; Boston, MA, USA
Received 8 January 2008; received in revised form 6 April 2008; accepted 22 April 2008
Available online 27 May 2008
Background: The effectiveness of bipolar collaborative chronic care models (B-CCMs) among those with co-occurring substance
use, psychiatric, and/or medical conditions has not specifically been assessed. We assessed whether B-CCM effects are equivalent
comparing those with and without co-occurring conditions.
Methods: We reanalyzed data from the VA Cooperative Study #430 (n = 290), an 11-site randomized controlled trial of the B-CCM
compared to usual care. Moderators included common co-occurring conditions observed in patients with bipolar disorder, including
substance use disorders (SUD), anxiety, psychosis; medical comorbidities (total number), and cardiovascular disease-related
conditions (CVD). Mixed-effects regression models were used to determine interactive effects between moderators and 3-year
Results: Treatment effects were comparable for those with and without co-occurring substance use and psychiatric conditions,
although possibly less effective in improving physical quality of life in those with CVD-related conditions (Beta = −6.11; p = 0.04).
Limitations: Limitations included multiple comparisons and underpowered analyses of moderator effects.
Conclusions: B-CCM effects were comparable in patients with co-occurring conditions, indicating that the intervention may be
generally applied. Specific attention to physical quality of life in those with CVD maybe warranted.
Published by Elsevier B.V.
Keywords: Mood disorders-bipolar; Cardiovascular disease; Substance use disorders; Collaborative care; Interaction
Data from this paper were presented at the International
Conference on Mental Health Policy and Economics (ICMPE) of the 1. Introduction
World Psychiatric Association. March 14, 2007, Venice, Italy.
⁎ Corresponding author. VA Ann Arbor SMITREC (11H), 2215
Fuller Road, Ann Arbor, MI 48105, USA. Tel.: +1 734 845 5046;
Bipolar disorder is common and associated with
fax: +1 734 845 3249. substantial functional impairment, health care costs,
E-mail address: email@example.com (A.M. Kilbourne). and premature mortality (Bauer, 2003). Co-occurring
0165-0327/$ - see front matter. Published by Elsevier B.V.
A.M. Kilbourne et al. / Journal of Affective Disorders 112 (2009) 256–261 257
conditions (i.e., medical and/or substance use disorders) intervention to clinical practice and/or for further inter-
are the rule rather than the exception in this group. vention adaptation, since it confirms whether the inter-
Prevalence rates for co-occurring psychiatric and vention is equally effective across different subgroups,
substance use disorders ranged from 57.3% to 74.3%, and also identifies subgroups in which intervention
with 30% have three or more conditions (Bauer et al., effects may be blunted so that it can be modified.
2005; Otto et al., 2006). Co-occurring substance use Analysis of whether co-occurring conditions might
disorders can occur in over 60% of some clinical moderate effects of medical interventions has not been
samples (Bauer et al., 1997, 2005). Medical comorbidity available in part due to limited numbers of patients with
occurs frequently, with over 80% of some samples highly severe or comorbid illness represented in clinical
having at least one active medical disorder, 19–23% trials (Glasgow et al., 2003). Effectiveness-oriented cli-
having two such conditions, and 35–50% have three or nical trials, which seek to enroll a heterogeneous sample
more (Kilbourne et al., 2004; Fenn et al., 2005). Patients of participants with severe or comorbid illness can pro-
with co-occurring conditions are more likely to ex- vide the opportunity to identify potential moderators.
perience worse outcomes than those without comorbid- Using outcome data from a long-term multi-site effec-
ity (Strakowski et al., 1998; Tsai et al., 2005). tiveness trial of a B-CCM in a sample of veterans with
Not surprisingly, outcomes for patients with bipolar bipolar disorder which was characterized by high rates
disorder are suboptimal despite the availability of effi- of psychiatric (e.g. anxiety, psychosis), substance use,
cacious medications (APA, 2002). Adjunctive psycho- and medical comorbidity, we determined whether such
social treatments, such as cognitive-behavioral therapy putative moderators blunted the effect of the B-CCM on
(CBT) (Lam et al., 2005), family therapy (Clarkin et al., the primary trial outcome and select secondary outcomes.
1998; Miklowitz et al., 2000), and psychoeducation We hypothesized that B-CCM would be less effective for
(Perry et al., 1999; Colom et al., 2003) trials have yielded patients with bipolar disorder and co-occurring psychia-
promising results. tric or medical comorbidity compared to those without co-
Nonetheless, in the few trials that have that included occurring conditions.
patients with co-occurring conditions, psychosocial
treatments have only minimal effect (Scott et al., 2. Methods
2006; Nierenberg et al., 2006) while nothing is known
about the effects of medical comorbidities on such 2.1. Study design
treatment effects. In a recent trial of CBT in a sample of
patients with co-occurring psychiatric conditions (Scott VA Cooperative Study Program (CSP) #430 was a
et al., 2006), lack of significant effect was associated 3-year, 11-site, randomized controlled trial in 306
with psychiatric severity in the sample, though others veterans randomized of B-CCM vs. usual VA care.
have questioned that explanation (Lam, 2006). Study design and main results have been reported else-
Recently, two long-term trials of over 700 partici- where (Bauer et al., 2006a,b). Briefly, participants were
pants (Bauer et al., 2006a,b; Simon et al., 2006) de- randomized at discharge from index hospitalization for
monstrated that collaborative chronic care models for bipolar disorder to 3 years of follow-up under B-CCM
bipolar disorder (B-CCMs) can improve outcome in this treatment or usual care. B-CCM treatment consisted of
chronic mental illness much as they do in chronic three components: evidence-based pharmacotherapy via
medical illnesses and depression treated in primary care simplified practice guidelines, augmented patient self-
(Badamgarav et al., 2003). Notably, B-CCMs had signi- management skills through group-based psychoeduca-
ficant effects despite the trials enrolling patients with co- tion, and enhanced access to and continuity of care using
occurring conditions. nurse care managers. Usual care included the dissemi-
Kraemer and Wilson (2002) have devised a frame- nation of practice guidelines to providers but no active
work to address this issue by assessing the effect of management or self-management sessions for the
moderators on treatment effect. This framework defines patient. Original trial primary analysis demonstrated
moderators as those characteristics present at the time of significant reduction in weeks in affective episode with
treatment initiation (and uncorrelated with treatment the B-CCM compared to usual care, subserved primarily
assignment) that have an interactive effect on the by significant reduction in weeks manic; secondary
intervention to affect its impact on outcome. They can analyses showed significant improvement in social role
also represent different subgroups of patients within the function and mental health quality of life but no
population who have different effect sizes. Moderators difference in physical health-related quality of life
analysis is critical both for optimal dissemination of the (Bauer et al., 2006b). All participants provided informed
258 A.M. Kilbourne et al. / Journal of Affective Disorders 112 (2009) 256–261
consent, and the study was approved by the Institutional 2.3. Analyses
Human Subjects Review Boards at each site.
Moderators can be defined by determining the inter-
2.2. Measures active effect between the treatment and a baseline
variable of interest (Kraemer and Wilson, 2002). For
Potential moderators were identified prior to this each candidate moderator we reapplied the same mixed-
reanalysis based on the prior literature, which identified effect regression models used in the original CSP #430
characteristics associated with poor outcome in bipolar analyses for each outcome, but added the candidate
disorder (Scott et al., 2006; Bauer., 2003; Fenn et al., moderator and the interaction term for the moderator
2005). We focused on co-occurring psychiatric, sub- and B-CCM treatment effect as described below. We
stance use, and medical conditions, which have been applied this strategy separately for each of the following
associated with non-response to other types of treatment characteristics, which we hypothesized a priori to be
(Nierenberg et al., 2006). potential moderators: current substance use disorder,
Data on co-occurring baseline substance use dis- current anxiety disorder, psychosis, three or more active
orders (SUD; excluding nicotine), anxiety disorders, and medical comorbidities, and CVD-related comorbidity.
psychosis at baseline were collected via Structured Because the B-CCM was designed to improve mental
Clinical Interview for DSM-IV (SCID) (First et al., health-related outcomes, we focused our moderator
1996). Co-occurring active medical conditions at base- analyses on the three main outcomes that were improved
line were identified from a structured chart review among patients in the B-CCM intervention compared to
method of established validity and reliability (Fenn usual care overall: percent weeks in affective episodes,
et al., 2005). Using this method, we identified par- percent weeks in manic episodes, and mental health-
ticipants with three or more general medical conditions related quality of life. We also assessed moderator
given that the modal number of conditions was one and effects for physical health-related quality of life given
because having three or more signified substantial that we also chose to examine medical comorbidities as
medical burden (Fenn et al., 2005). In addition, we potential moderators.
identified those with a current diagnosis of any of the For each mixed-effects regression model we included
following CVD-related conditions present in the the interaction effect between B-CCM and the potential
patient's medical record: type 2 diabetes, hypertension, moderator of interest, adjusting for the main effect of the
hyperlipidemia, coronary artery disease, or angina. B-CCM, the moderator, time, time⁎B-CCM effect, and
Prospective trial outcome data were collected by site. For all analyses, F-statistics were used and two-
Survey Coordinators who were blinded to randomiza- sided tests were applied for significance (p b 0.05). A
tion assignment. Outcomes included the Longitudinal significant overall interaction coefficient indicates that
Interval Follow-Up Examination (LIFE) every 8 weeks, the co-occurring condition is likely a moderator of
with primary outcome variable for the trial being weeks treatment effect, as it is enhanced (or diminished)
in affective episode (Keller et al., 1987). The LIFE beyond would be expected from an additive effect of
utilizes time-line follow-back methodology to provide treatment and the moderator alone. In the presence of a
weekly Psychiatric Symptom Ratings (PSRs) for mania significant moderator effect with the B-CCM ( p b 0.05),
and depression based on the number of DSM-IV criteria the direction of the interaction was determined by the
endorsed: no/minimal symptoms (PSR = 1–2), subthres- sign of the beta coefficient.
hold symptoms (PSR = 3–4), or episode (PSR = 5–6).
Prior to trial outcome we summarized symptom out- 3. Results
comes as the mean percentage of weeks each of the
3 years in manic, depressive, or any episode (PSR = Of the 306, 290 had complete data on potential
5–6). The secondary outcomes of mental and physical moderator characteristics. The mean age was 46 years
quality of life data were collected every 6 months (SD = 10), 104 (34%) were diagnosed with a current
using, respectively, the mental (MCS) and physical SUD, 116 (38%) with current anxiety disorder, and 157
(PCS) component scores of the SF-36 (Stewart et al., (52%) with psychosis. In addition, 144 (51%) had three
1988), both of which have a range 0–100 and popula- or more medical comorbidities, and 52 (18%) were
tion mean ± SD of 50 ± 10. All original trial analyses diagnosed with a CVD-related condition. Those missing
utilized intention-to-treat mixed-effects regression and not missing medical comorbidity data did not differ
models with weeks-in-episode square root-transformed in demographic or clinical characteristics, and comor-
to stabilize variance. bidity frequencies were similar across treatment arms.
A.M. Kilbourne et al. / Journal of Affective Disorders 112 (2009) 256–261 259
Interaction of treatment (B-CCM) and moderators by key outcomes
N=290 Percent weeks in any Percent weeks in manic SF-36 mental component SF-36 physical component
affective episode episode score score
Regression models: Beta* T score Beta⁎ T score Beta T score Beta T score
interaction term* (SE) (p-value) (SE) (p-value) (SE) (p-value) (SE) (p-value)
SUD⁎B-CCM (df = 238) 0.91 (0.56) 1.62 0.10) − 0.13 (0.44) − 0.29 (0.77) 0.28 (2.34) 0.12 (0.91) − 3.94 (2.36) − 1.67 (0.10)
Anxiety⁎B-CCM 0.07 (0.53) 0.13 (0.89) 0.46 (0.42) 1.08 (0.78) 1.48 (2.23) 0.66 (0.51) − 0.45 (2.24) − 0.20 (0.84)
(df = 234)
Psychosis⁎B-CCM −1.07 (0.53) − 2.02 a (0.04) − 0.85 (0.42) − 2.04 a (0.04) 1.33 (2.17) 0.61 (0.54) 2.06 (2.19) 0.94 (0.35)
(df = 232)
>=3 medical 0.67 (0.54) 1.25 (0.21) 0.22 (0.42) 0.52 (0.60) − 2.17 (2.18) −1.00 (0.32) − 0.88 (2.19) − 0.40 (0.69)
B-CCM (df = 230)
CVD Comorbidity* 0.47 (0.71) 0.66 (0.51) − 0.59 (0.56) − 1.06 (0.29) − 4.67 (2.96) −1.58 (0.12) − 6.11 a (2.97) − 2.06 (0.04)
B-CCM (df = 230)
*Beta reflects weeks in episode after square root-transformation to stabilize variance as in the original analyses. (Simon et al., 2006).
Negative beta for weeks in episode indicates augmented B-CCM effect since fewer weeks in episode reflects improved outcome; in contrast,
negative beta for physical quality of life indicates a possible blunted B-CCM effect since lower score reflects worse outcome.
Based on the mixed-effects regression models, in cases where treatment response is less optimal, for
treatment effects were comparable for those with and which subpopulations the intervention needs to be
without current SUD or anxiety disorders, including tailored. Such information is critical for planning future
weeks in affective episode, weeks manic, and mental dissemination and adaptation of interventions.
and physical health-related quality of life (Table 1). We found that B-CCM appears to be equally effec-
However, psychosis was associated with an augmented tive for patients with and without co-occurring psy-
B-CCM effect, with fewer weeks in affective episode chiatric and substance use disorders, and hence, the
(Beta = − 1.07, p = 0.04) and weeks manic episode intervention thus appears to be applicable to a broad
(Beta = − 0.85, p = 0.04); note that negative betas reflect population of individuals with bipolar disorder. This
better outcome. This translates into approximately one contrasts, for example, with the experience of Scott et al.
fewer week in affective or manic episode in the B-CCM (2006) who, using a different methodology, found that
treatment compared to usual care. CBT was less effective in patients with co-occurring
However, having a CVD-related condition might have conditions This may reflect the fact that the B-CCM
blunted the B-CCM effect on SF-36 physical health intervention was designed to assist a lower functioning
component score (Beta = −6.11, p = 0.04), with a negative population who may be experiencing acute manic or
beta as a lower score reflecting worse outcome. That is, depressive episodes, as opposed to CBT, which may not
those in B-CCM with CVD risk had their physical health- be completely effective for patients experiencing acute
related quality of life decline to a greater degree (i.e., ∼6 episodes given that such episodes may disrupt treatment
points) than those without CVD risk in usual care. engagement (Scott et al., 2006). Moreover, B-CCM
does much more to integrate services, apply evidence-
4. Discussion based pharmacotherapy, and manage patient needs
beyond bipolar symptoms. The lack of strong moderator
There has been little research examining moderators effects of B-CCM perhaps supports the ability of B-
of treatment effects in psychiatry clinical trial outcomes. CCM to respond to a broad group of patients with
Paucity of such analysis is likely due in part to the bipolar disorder, in part by enabling more flexible care
1) relative dearth until recently of effectiveness designs for those with co-occurring conditions than more pure
that attempt recruitment of samples with adequate psychological interventions.
representation of participants with co-occurring condi- It is intriguing to speculate regarding why the presence
tions, and 2) lack of RCTs that test interventions such as of psychosis or psychiatric comorbidity did not blunt the B-
the CCM that are specifically designed a priori for CCM effect, as appears to have been the case with
patients with co-occurring conditions. Such analyses are cognitive-behavioral therapy (Nierenberg et al., 2006; Scott
important for determining whether the intervention is et al., 2006). While we cannot be certain in multi-
equally effective across different subpopulations, and component interventions which specific component(s)
260 A.M. Kilbourne et al. / Journal of Affective Disorders 112 (2009) 256–261
were the most beneficial, the lack of difference in ral U.S. patient population or populations served in
pharmacotherapy intensity (Bauer et al., 2006b) and the community-based practices. Even so, the VA serves a
fact that the core elements of care reorganization and patient disproportionate number of patients with high rates of
self-management skill enhancement appear to be common relevant, active comorbidities which makes possible
to virtually all successful collaborative care models for such study of treatment moderators.
chronic medical illness (Badamgarav et al., 2003) suggest
that such psychosocial components deserve further 5. Conclusions
scrutiny. While it is difficult to imagine how psychosis
per se was associated with augmented B-CCM effects, this In summary, the B-CCM is a promising intervention
may be a proxy for greater overall illness severity. It is not even for patients with co-occurring psychiatric and
likely that regression to the mean was responsible for this substance use disorders, likely due in part to the fact
finding, as similar effects were not seen in those with that it was specifically designed to accommodate such
compared to those without psychosis treated in usual care. individuals, who represent the rule rather than the
Such more severely ill patients may benefit from more exception among individuals with bipolar disorder. Future
intensive services more than those with less severe illness. efforts should focus on the broader dissemination of
In contrast, the effect of the B-CCM on the secondary the B-CCM for bipolar disorder in community-based
outcome variable of physical health-related quality of practices, where many patients with bipolar disorder
life may have been possibly blunted for patients with suffer disproportionately from co-occurring conditions.
CVD comorbidities. In the original trial analyses there At the same time, the possible blunted effect on physical
was no overall difference in this outcome measure health-related quality of life in those with CVD risk
(Bauer et al., 2006b), perhaps because the B-CCM was indicates a need to tailor treatment interventions to
primarily designed to improve mental health-related address co-occurring medical conditions among indivi-
outcomes. CVD is the most common condition asso- duals with bipolar disorder, a process which is underway
ciated with premature death in patients with mental (Kilbourne et al., 2008). Moderator analyses can thus
disorders (Hennekens, 2007). Gaps in access, quality, yield important information from psychiatry clinical trials
and continuity of general medical care for patients with data both to optimize dissemination and to enhance future
bipolar and other chronic mental disorders are prevalent intervention development.
(Horvitz-Lennon et al., 2006). Competing demands on
B-CCM clinicians for attention to psychiatric needs may Role of funding source
have “out-competed” attention to more medical con- This research was supported by the Department of Veterans Affairs
cerns (Horvitz-Lennon et al., 2006). Regardless, our (VA), Veterans Health Administration Cooperative Studies Program
(CSP #430; MS Bauer, PI), the Health Services Research and
findings may suggest the need for the B-CCM to be
Development Service (IIR 02-283; AM Kilbourne, PI). The views
tailored to address physical health care, particularly in expressed in this article are those of the authors and do not necessarily
those with CVD-related risk (Kilbourne et al., 2008). represent the views of the Department of Veterans Affairs. The VA
NIMH had no further role in study design; in the collection, analysis
4.1. Limitations and interpretation of data; in the writing of the report; and in the
decision to submit the paper for publication.
Several limitations to this study warrant considera-
tion. First, we cannot be certain whether these results are Conflicts of interest
due to chance (i.e., from multiple comparisons). In The authors of this manuscript warrant that we have no actual or
addition, this study was not originally designed to perceived conflicts of interest — financial or non-financial — in the
procedures described in the enclosed manuscript.
examine moderators. For example, given that the other
medical illness moderator (3 or more illnesses) produced
no significant effects, the possible blunting effect of B- Acknowledgments
CCM on physical health outcomes among patients with
CVD needs to be interpreted with caution. Hence, our This research was supported by the Department
analysis might have been underpowered to detect signi- of Veterans Affairs, Veterans Health Administration
ficant differences in outcomes by subgroup, and our Cooperative Studies Program (CSP #430; MS Bauer, PI),
results might suggest a conservative estimate of the the Health Services Research and Development Service
differential effects of moderators on B-CCM outcomes. (IIR 02-283; AM Kilbourne, PI), and the VISN 4 Mental
Finally, the VA patient population, predominately male Illness Treatment Research and Evaluation Center. The
with military history, is not representative of the gene- views expressed in this article are those of the authors and
A.M. Kilbourne et al. / Journal of Affective Disorders 112 (2009) 256–261 261
do not necessarily represent the views of the Department of Horvitz-Lennon, M., Kilbourne, A.M., Pincus, H.A., 2006. From silos
Veterans Affairs. The funders did not have any role in the to bridges: Meeting the general health care needs of adults with
severe mental illnesses. Health Aff. 25, 659–669.
design and conduct of the study; collection, management, Keller, M., Lavori, P., Friedman, B., 1987. The Longitudinal Interval
analysis, and interpretation of the data; and preparation, Follow-Up Evaluation. Arch. Gen. Psychiatry 44, 540–548.
review, or approval of this manuscript. Kilbourne, A.M., Cornelius, J.R., Han, X., Pincus, H.A., Shad, M.,
Salloum, I., Conigliaro, J., Haas, G.L., 2004. Burden of general
medical conditions among individuals with bipolar disorder. Bipolar
Disord. 6, 368–373.
Kilbourne, A.M., Post, E.P., Nossek, A., Drill, L.J., Cooley, S., Bauer,
American Psychiatric Association, 2002. Practice guideline for the M.S., 2008. Improving Medical and Psychiatric Outcomes in
treatment of patients with bipolar disorder (revision). Am. J. Individuals with Bipolar Disorder: A Randomized, Controlled
Psychiatry 159, S1–S50. Effectiveness Trial. Psychiatr Serv.
Badamgarav, E., Weingarten, S.R., Henning, J.M., Knight, K., Kraemer, H.C., Wilson, G.T., 2002. Mediators and moderators of treat-
Hasselblad, V., Gano Jr, A., Ofman, J.J., 2003. Effectiveness of ment effects in randomized clinical trials. Arch. Gen. Psychiatry 59,
disease management programs in depression: A systematic review. 877–883.
Am. J. Psychiatry 160, 2080–2090. Lam, D., 2006. What can we conclude from studies on psychotherapy in
Bauer, M.S., 2003. Bipolar (manic-depressive) disorder, In: Tasman, bipolar disorder? Invited commentary on…Cognitive-behavioural
A., Kay, J., Lieberman, J. (Eds.), Psychiatry, 2nd ed. PA Saunders, therapy for severe and recurrent bipolar disorders. Br. J. Psychiatry
Philadelphia, pp. 1237–1270. 188, 321–322.
Bauer, M.S., McBride, L., Shea, N., Gavin, C., Holden, F., Kendall, S., Lam, D.H., Hayward, P., Watkins, E.R., Wright, K., Sham, P., 2005.
1997. Impact of an easy-access clinic-based program for bipolar Relapse prevention in patients with bipolar disorder: Cognitive
disorder. Psychiatr. Serv. 48, 491–496. therapy outcome after 2 years. Am. J. Psychiatry 162, 324–329.
Bauer, M.S., Altshuler, L., Evans, D.R., Beresford, T., Williford, W.O., Miklowitz, D.J., Simoneau, T.L., George, E.L., Richards, J.A., Kalbag,
Hauger, R., VA Cooperative Study #430 Team, 2005. Prevalence A., Sachs-Ericsson, N., Suddath, R., 2000. Family-focused treatment
and distinct correlates of anxiety, substance, and combined comor- of bipolar disorder: 1-year effects of a psychoeducational program in
bidity in a multi-site public sector sample with bipolar disorder. conjunction with pharmacotherapy. Biol. Psychiatry 48, 582–592.
J. Affect Disord. 85, 301–315. Nierenberg, A.A., Ostacher, M.J., Calabrese, J.R., Ketter, T.A.,
Bauer, M.S., McBride, L., Williford, W.O., Glick, H., Kinosian, B., Marangell, L.B., Miklowitz, D.J., Miyahara, S., Bauer, M.S.,
Altshuler, L., Beresford, T., Kilbourne, A.M., Sajatovic, M., Co- Thase, M.E., Wisniewski, S.R., Sachs, G.S., STEP-BD Investiga-
operative Studies Program 430 Study Team, 2006a. Collaborative tors, 2006. Treatment-resistant bipolar depression: A STEP-BD
care for bipolar disorder: Part I. Intervention and implementation in equipoise randomized effectiveness trial of antidepressant aug-
a randomized effectiveness trial. Psychiatr. Serv. 57, 927–936. mentation with lamotrigine, inositol, or risperidone. Am. J.
Bauer, M.S., McBride, L., Williford, W.O., Glick, H., Kinosian, B., Psychiatry 163, 210–216.
Altshuler, L., Beresford, T., Kilbourne, A.M., Sajatovic, M., Otto, M.W., Simon, N.M., Wisniewskim, S.R., Miklowitz, D.J., Kogan,
Cooperative Studies Program 430 Study Team, 2006b. Collabora- J.N., Reilly-Harrington, N.A., Frank, E., Nierenberg, A.A., Mar-
tive care for bipolar disorder: Part II. Impact on clinical outcome, angell, L.B., Sagduyu, K., Weiss, R.D., Miyahara, S., Thase, M.E.,
function, and costs. Psychiatr. Serv. 57, 937–945. Sachs, G.S., Pollack, M.H., STEP-BD Investigators, 2006. Prospec-
Clarkin, J.F., Carpenter, D., Hull, J., Wilner, P., Glick, I., 1998. Effects tive 12-month course of bipolar disorder in out-patients with and
of psychoeducational intervention for married patients with bipolar without comorbid anxiety disorders. Br. J. Psychiatry 189, 20–25.
disorder and their spouses. Psychiatr. Serv. 49, 531–533. Perry, A., Tarrier, N., Morriss, R., McCarthy, E., Limb, K., 1999.
Colom, F., Vieta, E., Martinez-Aran, A., Reinares, M., Goikolea, J.M., Randomised controlled trial of efficacy of teaching patients with
Benabarre, A., Torrent, C., Comes, M., Corbella, B., Parramon, G., bipolar disorder to identify early symptoms of relapse and obtain
Corominas, J., 2003. A randomized trial on the efficacy of group treatment. BMJ 318, 149–153.
psychoeducation in the prophylaxis of recurrences in bipolar patients Scott, J., Paykel, E., Morriss, R., Bentall, R., Kinderman, P., Johnson,
whose disease is in remission. Arch. Gen. Psychiatry 60, 402–407. T., Abbott, R., Hayhurst, H., 2006. Cognitive-behavioural therapy
Fenn, H.H., Bauer, M.S., Altshuler, L., Evans, D.R., Williford, W.O., for severe and recurrent bipolar disorders: Randomised controlled
Kilbourne, A.M., Beresford, T.P., Kirk, G., Stedman, M., Fiore, L., trial. Br. J. Psychiatry 188, 313–320.
VA Cooperative Study #430 Team, 2005. Medical comorbidity and Simon, G.E., Ludman, E.J., Bauer, M.S., Unützer, J., Operskalski, B.,
health-related quality of life in bipolar disorder across the adult age 2006. Long-term effectiveness and cost of a systematic care
span. J. Affect Disord. 86, 47–60. management program for bipolar disorder. Arch. Gen. Psychiatry
First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 1996. 63, 500–508.
Structured Clinical Interview for Axis I DSM-VI Disorders- Stewart, A., Hays, R., Ware, J., 1988. The MOS short-form general
(SCID), Patient Ed. Biometrics Research Department, New York health survey. Med. Care 26, 724–735.
Psychiatric. Institute, New York, NY. version 2.0. Strakowski, S.M., Sax, K.W., McElroy, S.L., Keck, P.E., Hawkins,
Glasgow, R.E., Lichtenstein, E., Marcus, A.C., 2003. Why don't we J.M., West, S.A., 1998. Course of psychiatric and substance
see more translation of health promotion research to practice? abuse syndromes co-occurring with bipolar disorder after a first
Rethinking the efficacy-to-effectiveness transition. Am. J. Publ. psychiatric hospitalization. J. Clin. Psychiatry 59, 465–471.
Health 93, 1261–1267. Tsai, S.Y., Lee, C.H., Kuo, C.J., 2005. A retrospective analysis of risk
Hennekens, C., 2007. Increasing global burden of cardiovascular and protective factors for natural death in bipolar disorder. J. Clin.
disease in general populations and patients with schizophrenia. Psychiatry 66, 1586–1591.
J. Clin. Psychiatry 68, 4–7.