Therapeutic alliance perceptions and medication adherence in patients with bipolar disorder

Journal of Affective Disorders 107 (2008) 53 – 62 www.elsevier.com/locate/jad

Research report

Therapeutic alliance perceptions and medication adherence in patients with bipolar disorder John E. Zeber a,b,⁎, Laurel A. Copeland a,b , Chester B. Good c,d , Michael J. Fine c,d , Mark S. Bauer e,f , Amy M. Kilbourne g,h a

Veterans Affairs HSR&D: South Texas Veterans Health Care System (VERDICT), San Antonio, TX, United States University of Texas Health Science Center at San Antonio, Department of Psychiatry, San Antonio, TX, United States c Center for Health Equity Research and Promotion (CHERP), VA Pittsburgh Healthcare System, Pittsburgh, PA, United States d Division of General Internal Medicine, Department of Medicine, University of Pittsburgh, United States e Department of Psychiatry, Brown University, Providence, RI, United States f Veterans Affairs Medical Center, Providence, RI, United States VA Ann Arbor Healthcare System, Serious Mental Illness Treatment Research and Evaluation Center, Ann Arbor, MI, United States h University of Michigan, Department of Psychiatry, Ann Arbor, MI, United States b

g

Received 6 February 2007; received in revised form 25 July 2007; accepted 31 July 2007 Available online 5 September 2007

Abstract Background: Despite the dissemination of practice guidelines for bipolar disorder, outcomes remain suboptimal, largely due to poor treatment adherence. The episodic nature of bipolar disorder disrupts appropriate patient-provider dynamics, interfering with appropriate care. Maintaining a beneficial therapeutic alliance with providers is one important strategy for improving adherence. We examine the association between adherence and therapeutic environment perceptions among veterans with bipolar disorder. Methods: Participants were recruited from the Continuous Improvement for Veterans in Care — Mood Disorders (CIVIC-MD) study (N = 435). Individual items and a summary score from the Health Care Climate Questionnaire (HCCQ) for bipolar disorder solicited patient evaluations of their therapeutic environment. Multivariable logistic analyses examined the association between therapeutic alliance and two measures of adherence (missed medication days and intrapersonal barriers), adjusting for relevant patient characteristics. Results: Adherence difficulty was reported on both measures, with substantial differences between perceived barriers and actual medication behavior. Significantly fewer minority veterans endorsed good adherence than white patients (59% versus 77%), although no ethnic differences were noted in therapeutic environment perceptions. Multivariable results indicated that positive therapeutic alliance was associated with better adherence (HCCQ effect sizes 13–20%). Notably, patients reporting providers encouraged “staying in regular contact” were more likely to be adherent, as were patients whose “providers regularly review their progress”. Limitations: Generalizability from observational study; adherence defined by cross-sectional patient self-report. Conclusions: The observed association between medication adherence and therapeutic alliance with bipolar treatment supports intervention efforts to strengthen the patient–provider relationship, a bond likely to yield positive clinical outcomes. Published by Elsevier B.V. Keywords: Therapeutic alliance; Medication adherence; Treatment barriers; Ethnicity; Veterans

⁎ Corresponding author. South Texas Veterans Health Care System, 7400 Merton Minter Boulevard (Verdict 11c6), San Antonio, TX 78229-4404, United States. Tel.: +210 617 5300x16666; fax: +210 567 4423. E-mail address: [email protected] (J.E. Zeber). 0165-0327/$ - see front matter. Published by Elsevier B.V. doi:10.1016/j.jad.2007.07.026

54

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

1. Introduction Bipolar disorder is a chronic mental illness associated with substantial functional impairment, tremendous health care costs, and premature mortality (Murray and Lopez 1996; Bauer et al., 2002). Its significance, which the World Health Organization currently lists as #7 in terms of overall disease burden, may even be considerably underestimated given the number of atendant physical and mental health comorbidities, social and occupational instability, and substantial care-taker burden (Manning, 2005). A psychiatric illness uniquely characterized by alternating periods of mania, psychosis and depression, patients afflicted with bipolar disorder require intensive pharmacological and psychosocial management (Gitlin et al., 1989; Keck et al., 1997). The episodic nature of this disease frequently disrupts appropriate patient–provider dynamics, further complicating prognosis and outcomes. Despite the dissemination, if not wide implementation, of practice guidelines, bipolar disorder outcomes remain suboptimal, largely due to poor treatment adherence. Lingam and Scott reported non-adherence rates among bipolar patients ranged from 20–66% across 22 studies, with a median of 41% (Lingam and Scott, 2002). The problem is even less sanguine for antipsychotic drugs, increasingly prescribed for this population; even accounting for the newer atypical medications, with arguably better side-effect profiles, 48% of all veterans with bipolar disorder are poorly adherent (Sajatovic et al., 2006). Numerous adverse clinical and behavioral outcomes are directly attributable to inadequate adherence (Salloum et al., 2005). In addition to treatment disruption, ramifications such as worsening symptoms, deteriorating functional status, higher treatment costs, and increased risks for psychiatric admission and relapse are quite common (Post et al., 2003; Scott and Pope, 2002). The documented risk factors include demographic characteristics (e.g., younger age and male gender), lower cognitive functioning (Danion et al., 1987), manic episodes (Keck et al., 1996), and poor illness insight (Adams and Scott, 2000). Comorbid substance abuse is especially problematic, exacerbating symptoms in an already complex and challenging disorder (Swartz et al., 1998). Despite the severity of this problem, patient and provider explanations as to the primary reasons for medication failure are weakly correlated (Pope and Scott, 2003), hindering potential remedies for improving adherence. Maintaining a beneficial therapeutic alliance between patients and providers is one effective strategy for improving treatment retention, adherence, and subse-

quent outcomes. Proactive patients, autonomous individuals who feel respected as a treatment partner and expressing confidence in their healthcare environment, are hypothesized to fare better than patients lacking a voice in their own treatment course. Investigations into the association between psychiatric disorders, therapeutic alliance, and adherence can be traced back to at least 1970 (Howard et al., 1970). A rich body of literature has since developed, comprising numerous descriptive studies, incorporating an array of conceptual frameworks borrowed from medical sociology (Blackwell, 1997). The link between patient–provider relationships and adherence is observed across several mental health conditions, including substance abuse, schizophrenia, and depression outcomes (Meier et al., 2005; Fenton et al., 1997; Klein et al., 2003). However, despite the apparent conceptual applications, until recently comparatively few studies connecting clinical relationships and adherence has been translated to bipolar disorder research, at least in explicit quantitative efforts. The excellent qualitative study by Sajatovic and colleagues recently explored one significant dimension of therapeutic alliance, the collaborative care model. They found a consistently positive association between alliance and treatment adherence as expressed through the Drug Attitudes Inventory (a multi-question survey of medication beliefs), self-reported adherence, and primary care attendance (Sajatovic et al., 2005). In another smaller study (n = 61), Gaudiano and Miller observed that better alliance perceptions improved treatment outcomes in patients with bipolar disorder, such as psychotherapy retention (Gaudiano and Miller, 2006). Berk et al. conducted a comprehensive literature review exploring the practical and theoretical issues surrounding the therapeutic environment. Psychosocial interventions, including efforts to develop stronger clinical relationships, offered tremendous optimism for sustained adherence (Berk et al., 2004). Patients entering treatment with greater therapy expectations and trust in their providers achieve better adherence. The beneficial nature of appropriate clinical relationships has itself been described as a “mood stabilizer” (Havens and Ghaemi, 2005). Certain patients may garner particular advantages from a strong patient–provider alliance. Demographics, health beliefs, and cultural values are frequently expressed in attitudes surrounding adherence to psychotropic medications. Fleck et al. observed that African– American and white patients endorsed different sets of culturally-based values and treatment perceptions as explanations for poor adherence (Fleck at al., 2005). While some studies revealed a minimal association between ethnicity and either the strength of clinical

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

relationships or the influence of alliance upon outcomes (Ricker et al., 1999), others indicate that minority patients might place greater value on closer therapeutic relationships than white patients (Tonigan, 2003). The role of ethnicity and healthcare environment perceptions are quite relevant, given consistent findings of significantly worse adherence among minority patients with serious mental illness (Valenstein et al., 2004; Opolka et al., 2003). 1.1. Objective Per the National Psychosis Registry (Blow et al., 2005), the Veterans Health Administration (VA) provided care to 79,000 veterans diagnosed with bipolar disorder in 2004. As the nation's largest integrated healthcare system, the VA provides preferential care to its more disadvantaged patients, prioritizing treatment to veterans who are poorer and more disabled. In addition, access and cost barriers have historically been less restrictive than in other health systems; this includes pharmacy benefits and drug costs, the latter a significant adherence barrier (Zeber et al., 2007). As a result, the VA represents an excellent setting to examine our primary objective, the association between perceptions of the therapeutic environment and medication adherence in patients diagnosed with bipolar disorder. 2. Methods 2.1. Study population and sample Participants were recruited from the Continuous Improvement for Veterans in Care — Mood Disorders (CIVIC-MD) population-based study of VA patients with bipolar disorder (additional details provided elsewhere (Kilbourne et al., 2007)). This naturalistic cohort study examined patient and provider factors associated with treatment quality and outcomes, along with important mediators of these outcomes (e.g., adherence). Eligible patients were currently receiving inpatient or outpatient treatment for bipolar disorder at a large urban VA mental health facility in Western Pennsylvania from July 2004 through July 2006. This medical center serves as the catchment area for the vast majority of VA psychiatric care in the region. Upon providing written informed consent, individuals completed a baseline survey with a trained interviewer. Inclusion criteria included a current diagnosis of bipolar disorder (I, II, NOS), cyclothymia, or schizoaffective disorder-bipolar subtype based upon chart review and a confirmatory diagnosis from their provider. The primary exclusion

55

criterion was having an unstable medical condition or significant cognitive impairment precluding patients from completing the surveys or providing informed consent. The analysis for this study utilized baseline survey data, where assessments gathered information including demographics and other patient characteristics, symptomatology, substance use, behavioral factors, access issues, treatment adherence, plus perceptions of the therapeutic alliance and care received within the mental health care environment. The study was reviewed and approved by the medical center Institutional Review Board. 2.2. Measures 2.2.1. Dependent variables As the central outcome, we evaluated medication adherence using two complementary self-completed patient assessments. The first measure asked how many days within the past four the patient missed at least one medication dose for their bipolar disorder, a question strongly correlated with good adherence based upon electronic bottle cap data (Kilbourne et al., 2005). The second assessment utilized the validated Morisky scale (Morisky et al., 1986), a four-item yes/no instrument frequently used for adherence research across a variety of chronic medical and psychiatric conditions, including affective disorders (Shalansky et al., 2004; George et al., 2000). This scale aims at understanding commonly perceived intrapersonal barriers to adherence. We dichotomized good adherence as having only 0 or 1 of the following self-reported obstacles: ever forget to take medications, careless at times about taking medication, stop taking medications when feeling better, or stop taking medications when feeling worse. The scale demonstrates good test–retest reliability and our cut-point has been correlated with appropriate hypertension control (Morisky et al., 1986). 2.2.2. Independent variables Our primary predictor variable was patient perception of the therapeutic alliance, solicited from the Health Care Climate Questionnaire (HCCQ) (Ludman et al., 2002). The HCCQ is a 10-item instrument specifically developed for patients with bipolar disorder. It measures the degree of comfort that a patient expresses with mental health treatment according to statements about their health care environment. Examples of these statements include “I feel understood by my mental health team” and “I am encouraged to ask questions about my treatment”. Each question has a 0–6 Likert scale

56

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

response option (strongly disagree to strongly agree). There is a significant association between the HCCQ and the Patient Satisfaction Index (r = 0.63, p b 0.001), and with the Structured Clinical Interview for DSM-IV (SCID) patient confidence questionnaire. Covariates included age, gender, ethnicity, and homelessness within the past month. Self-reported ethnicity was collapsed into three categories: White, African– American (the predominate minority group in the study area), and Other. In addition to demographics, three other covariates incorporated clinical and substance abuse factors potentially influencing medication adherence. Hazardous drinking was assessed using one question from the Alcohol Use Disorders Identification Test (AUDIT), which inquired about having 5 or more drinks on a single occasion within the past month. This one question is strongly correlated with hazardous drinking as defined by the full AUDIT questionnaire (Saunders et al., 1993). Recognizing the important association between mood states and adherence, we controlled for the presence of any affective disorder Internal State Scale (ISS), validated for the identification of manic, depressive, mixed, and euthymic mood states in patients with bipolar disorder (Glick et al., 2003). The ISS generates a summary score from 15 current symptoms (past 4 weeks). A variable for any illicit drug use was created concerning past year use of marijuana, cocaine, stimulants, or other illicit drugs. Finally, we wished to control for the possibility that veterans might receive some medication outside the system, so a broad marker of any self-reported non-VA utilization (inpatient, outpatient, or ER) was also added to the models. 2.3. Study design and analysis The study used a cross-sectional prospective design from sequential patients seeking care and meeting inclusion criteria. Multivariable logistic regression models predicted the probability of good medication adherence per both outcomes (i.e., no missed doses and 0 or 1 Morisky barriers). Separate models were run for the HCCQ summary score and for each of the ten individual HCCQ items, adjusting for the aforementioned covariates. The sample size provided analytical power to detect statistically significant effects. In addition to the primary models, exploratory sub-analyses were conducted incorporating ethnicity⁎HCCQ interaction terms to determine if a stronger therapeutic alliance might be differentially associated with good adherence among minority veterans. All analyses were performed with SAS, version 9.0 (SAS Institute, Cary, NC).

3. Results The descriptive and bivariate characteristics of the study population (N = 435) are provided in Table 1. The mean age was 49.4 years (sd = 10.6), with 14% women and 23% ethnic minorities (including 13% African– Americans). This profile is well representative of all veterans diagnosed with bipolar disorder, with an average age of 51.4, 13.0% women, and 9.7% African–American (Blow et al., 2005). Over 2/3 of these patients had some post-high school education and 17% completed a college degree. Substance abuse was quite prevalent, with 28% reporting some drug use and 21% a hazardous drinking episode during the past year. Recent mania was noted in nearly 30% of this sample, while 55% experienced some affective disorder and 12% reported being homeless. A relatively small portion of these veterans reported significant access problems with medical or psychiatric services when needed, 11% and 16% respectively, while close to half utilized some non-VA care within the past year. In terms of general perceptions of the therapeutic environment, the mean HCCQ summary score was 39.4 (sd = 15.0), equivalent to a “slightly positive” overall view of alliance and comfort with mental health treatment. The average individual item scores were similar, ranging from 3.5 to 4.4; the full survey questions and mean values are provided in Table 2. Good medication adherence as defined by the two measures here indicated that 73% (n = 287) of patients claimed not to have missed any recent doses, while 54% reported experiencing minimal Morisky barriers (n = 232). 3.1. Association of therapeutic alliance and medication adherence Examining bivariate analyses by adherence status, no significant differences on either outcome measure existed for age, education, income, marital status, smoking, or use of non-VA services. Conversely, ethnicity and hazardous drinking were consistently associated with poor adherence: for example, approximately 60% of minorities reported missing no medication days and 40% with minimal Morisky barriers versus 77% and 58%, respectively, for white patients (p b .01). Hazardous drinkers were far less adherent on both measures than among veterans without such behavior (p b .001). Patients reporting poor access to mental health care services likewise had much lower adherence than those not perceiving access difficulty. Variations between the two adherence measures were also observed on several

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

57

Table 1 Study population — descriptive and bivariate statistics (N = 435) Patient characteristic

N

Overall population mean (sd)/%

Bivariate comparisons

Self-reported good adherence

Age, men

435

49.4 (10.6)

% women

62

14.3%

Women Men

Ethnicity White African–American Other Education: college graduate

336 58 41 75

77.3% 13.3% 9.4% 17.3%

White African–American Other College graduate bCollege graduate

Income b0,000 0,000–$20,000 $20,000–$30,000 $30,000–$40,000 N$40,000 % homeless, past 4 weeks

134 120 74 52 44 53

31.6% 28.3% 17.5% 12.3% 10.4% 12.2%

% married

131

30.2%

% manic episode

126

29.1%

% any affective disorder

240

55.2%

% binge drinking

93

21.4%

% any drug use

123

28.3%

% smoking

267

61.4%

% with some non-VA use

212

49.0%

% poor mental health care access when needed % poor medical care access when needed Medication Adherence No missed doses Morisky, 0 or 1 barrier

70

16.2%

48

11.1%

b0,000 0,000–$20,000 $20,000–$30,000 $30,000–$40,000 N$40,000 Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No

287 232

72.7% 53.8%

patient factors. That is, while no statistical differences existed on the missed doses outcome in terms of homelessness, income, mania, any drug use, or poor medical care access, these characteristics were significant regarding Morisky barriers. In the multivariable analyses, the overall HCCQ summary score demonstrated a significantly positive relationship with missed medication days but not Morisky barriers. The odds ratios (OR) predicting the association between alliance and no missed doses was 1.03 (95% CI 1.01–1.03, p = 0.022), suggesting a small improvement (i.e., approximately 3%) in adherence for

Morisky (0–1 barriers)

p-value

No missed doses

p-value

Good 51.2 Poor 48.7 41.9% 55.8% 57.8% 37.9% 43.9%

.0976

Good 50.9 Poor 49.2 54.1% 76.1% 76.7% 60.7% 59.0%

.1286

61.3% 52.3% 47.4% 57.1% 58.9% 52.9% 61.4%

.1522

79.4% 71.1% 49.7% 56.8% 58.1% 54.2% 60.8%

.1615

37.7% 56.2% 56.6% 52.5% 45.2% 57.2% 47.6% 67.4% 36.6% 58.8% 40.5% 59.0% 52.6% 60.1% 51.8% 55.7% 31.3% 56.7% 42.0% 56.4%

.0114

72.3% 72.6% 72.3% 72.7% 66.7% 74.9% 68.4% 81.8% 61.2% 75.7% 69.4% 73.9% 72.6% 72.8% 70.2% 75.4% 59.1% 74.4% 69.2% 74.4%

.9675

.0423 .0080

.3415

.4349 .0231 b.0001 b.0001 .0005 .1481 .3618 b0001 .0283

.0004 .0063

.3598

.9254 .0989 .0055 .0077 .3592 .9642 .2485 .0322 .5177

every point gain in the 0–60 HCCQ scale. Among covariates, hazardous drinking was strongly associated with lower adherence on both measures, with ORs of 0.51 (95% CI 0.29–0.90, p = 0.030) and 0.55 (95% CI 0.33–0.91, p = 0.023), meaning that heavy drinkers were twice as likely to report any missed doses or multiple barriers. African–Americans reported significantly more missed days and adherence barriers than white patients, with ORs of 0.39 (95% CI 0.20–0.78, p = 0.008) and 0.52 (95% CI 0.28–0.97, p = 0.041). While gender was highly significant for medication days, with women more frequently reporting missed

58

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

Table 2 Multivariate models predicting good medication adherence Probability of good adherence Morisky (0–1 barriers)

No missed doses

Model

Variable

Odds ratio (OR) (95% CI)

p-value

Odds ratio (OR) (95% CI)

p-value

HCCQ Summary (overall mean = 39.4; sd = 15.0)

HCCQ Age Women African–American Other Ethnicity Homeless Binge Drinking Affective Disorder Any Drug Use Non-VA utilization HCCQ

1.02 (0.99–1.04) 1.19 (0.95–1.44) 0.60 (0.34–1.08) 0.52 (0.28–0.97) 0.67 (0.33–1.34) 0.81 (0.42–1.56) 0.55 (0.33–0.91) 0.58 (0.37–0.91) 0.72 (0.45–1.18) 0.70 (0.46–1.09) 1.03 (0.98–1.07)

.0912 .1353 .0881 .0410 .2511 .5340 .0226 .0182 .0912 .0861 .0910

1.02 (1.01–1.03) 1.22 (0.95–1.57) 0.40 (0.22–0.73) 0.39 (0.20–0.78) 0.43 (0.21–0.91) 1.42 (0.67–3.04) 0.51 (0.29–0.90) 0.59 (0.34–1.05) 1.31 (0.74–2.32) 0.87 (0.53–1.40) 1.06 (0.93–1.20)

.0439 .1125 .0038 .0078 .0170 .3407 .0301 .0712 .2963 .5945 .4239

HCCQ

1.14 (0.98–1.30)

.0751

1.13 (0.98–1.31)

.0643

HCCQ

1.16 (1.03–1.29)

.0362

1.13 (1.02–1.28)

.0387

HCCQ

1.14 (1.02–1.26)

.0392

1.10 (0.97–1.25)

.1145

HCCQ

1.18 (1.05–1.29)

.0182

1.08 (0.94–1.22)

.3157

HCCQ

1.03 (0.92–1.15)

.5522

1.09 (0.95–1.23)

.2350

HCCQ

1.07 (0.96–1.20)

.2073

1.14 (1.01–1.28)

.0424

HCCQ

1.15 (1.02–1.28)

.0269

1.13 (1.01–1.28)

.0413

HCCQ

1.06 (0.95–1.18)

.2415

1.10 (0.97–1.24)

.0981

HCCQ

1.20 (1.05–1.33)

.0288

1.15 (1.03–1.28)

.0338

HCCQ #1 (mean = 4.1; sd = 1.7): “I feel that my mental health care provider team has provided me choices and options.” HCCQ #2 (mean = 4.0; sd = 1.6): “I feel understood by my mental health care provider team.” HCCQ #3 (mean = 4.0; sd = 1.6): “My mental health care provider team conveys confidence in my ability to make changes.” HCCQ #4 (mean = 4.2; sd = 1.8): “My mental health care provider team encourages me to ask questions.” HCCQ #5 (mean = 4.0; sd = 1.8): “My mental health care provider team tries to understand how I see things before suggesting a new way of doing things.” HCCQ #6 (mean = 4.0; sd = 1.7): “My mental health care provider team made me aware of what to expect from good bipolar disorder care.” HCCQ #7 (mean = 3.5; sd = 1.9): “My mental health care provider team has provided training in what I need to do to carry out good bipolar disorder care.” HCCQ #8 (mean = 3.7; sd = 1.8): “My mental health care provider team regularly reviews with me my progress in managing all aspects of my treatment plan.” HCCQ #9 (mean = 3.5; sd = 1.7): “My mental health care provider team has worked with me to develop a bipolar disorder care plan.” HCCQ #10 (mean = 4.4; sd = 1.9): “My mental health care provider team makes sure that we stay in regular contact.”

Note: Full multivariable model results (coefficients) for all HCCQ individual items are available from the authors.

days (OR 0.40 CI 0.22–0.73, p = 0.004), it was not associated with Morisky barriers. Conversely, having any affective disorder was predictive of experiencing adherence barriers (OR 0.58 CI 0.37–0.91 p = 0.018) but not missed days. Age, homelessness, drug use, and non-VA utilization were not significant in either primary model.

For the individual HCCQ items, several predicted good medication adherence: 5 of the items were statistically significant in relation to Morisky barriers and 4 in connection with no missed days, while a few others approached trend significance (p b .10). For example, the question about “Staying in regular contact with my providers” (HCCQ #10) was highly associated

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

with both outcomes. The ORs on this item, now relative to a single point increase on the 0–6 scale, were 1.15 (p = .034) for no missed days and 1.20 for Morisky barriers (p = .0102). “Regularly reviewing my treatment progress” (HCCQ#8) had ORs of 1.13 and 1.15, while “providers conveying confidence in my abilities” (HCCQ #3) yielded similar effect sizes. Overall, the general influence of alliance upon good adherence was slightly stronger for the Morisky outcome than for missed medication days (effect sizes from 14–20% versus 13–15%). Results for the other independent variables are similar to those in the HCCQ summary models. That is, binge drinking and ethnicity remained significantly associated with lower adherence on both outcomes, women for missed medication days but not barriers, while other factors did not reach statistical significance. Results for all 10 HCCQ items are detailed in Table 2; given consistent findings and coefficient ranges for the covariates across all 10 individual items, only the ORs for the primary predictor variable, therapeutic alliance, are presented here (full results available from the authors). The exploratory analyses examining possible interaction effects between ethnicity and alliance revealed no significant results in the HCCQ summary score or individual item models. A few of the individual items did approach a trend significance on the interaction terms, indicating that while minorities might experience some differential adherence benefits from improved therapeutic alliance, these effects were relatively weak and not statistically significant here. 4. Discussion Nearly 30% of patients acknowledged medication adherence difficulties, while close to half expressed significant barriers to taking their medications appropriately. The results of this quantitative study confirm and extend previous findings linking better clinical relationships with improved adherence, a critical goal in light of documented medication problems with this population. While a slightly favorable perception of trust and satisfaction with bipolar treatment currently exists in this study group, the quality of the therapeutic alliance is a mutable factor that can be addressed through ongoing provider training and other interventions (Weiden and Rao, 2005). We recognize that the odds ratios for the overall HCCQ score might imply a relatively small effect size; however, moving from a neutral to even a moderately positive impression on several of the individual items is associated with adherence gains of 15–20%. The potential benefits in

59

terms of clinical outcomes and quality of life from such improvement should not be underestimated. Findings concerning the individual alliance (HCCQ) items also provide important clinical insights, particularly the three questions significantly associated with both missed medication days and Morisky obstacles. Providers who “conveyed confidence” in patients' ability to participate in treatment and advocated “keeping in regular contact” proved to be highly influential upon good adherence. The latter reflects a standard concern over treatment consistency and retention, necessary elements for an optimistic therapeutic course of bipolar care. Not surprisingly, the perception of having a provider who “regularly reviews treatment plan progress” likewise supported an environment more conducive to better medication adherence. These elements of a positive clinical relationship clearly represent areas where ongoing efforts are likely to result in better patient satisfaction and outcomes. It is interesting, and quite troubling, that while relatively fewer patients here reported poor adherence (30%) in comparison to previous studies, almost half experienced at least two intrapersonal barriers. Furthermore, individuals encountering these problems were more likely to be minorities, individuals suffering from mood disorders, the homeless, those with substance abuse problems or lower perceived access to care — arguably more vulnerable patients. The lower prevalence of missed medication days we observed may be attributable to the short four-day reporting window. Yet the Morisky items are unbounded with respect to timeframe, therefore reflecting potentially serious ongoing obstacles to appropriate adherence behavior. Fortunately, results here imply that the influence of a solid therapeutic alliance appears slightly stronger in affecting stated intrapersonal barriers. This is an optimistic sign for addressing long-term obstacles to adherence and improving subsequent outcomes. In addition to hazardous drinking, ethnicity was strongly correlated with adherence, and offers a particularly intriguing discussion area. Although no differences in overall perceptions of the treatment environment existed across ethnic groups, nor were the interaction models significant, ethnicity is highly associated with poor adherence and other related factors. For example, in other bivariate analyses not reported here, minorities reported much higher rates of hazardous drinking (white = 19%, African–American = 28%, other = 37%), which in turn was directly related to lower alliance. Minorities were also far more likely to be homeless, use drugs, have lower income, and report access problems. Since minorities already face differential antipsychotic

60

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

prescribing practices (Fleck et al., 2002), and that policies such as medication copayments can inequitably hurt these patients (Zeber et al., 2006), interventions to narrow the adherence gap represent an important priority to the VA and other systems. Efforts to develop stronger therapeutic relationships and culturally competent care, recognizing ethnic differences in health beliefs and treatment preferences, should work towards better engaging minorities and other vulnerable bipolar disorder patients in their own treatment. Notwithstanding the more proximate objective of targeting medication adherence, Swann believes there is tremendous hope for using solid therapeutic alliances to improve long-term outcomes for bipolar disorder (Swann, 2005). Psychosocial interventions, which include assisting the development of enhanced patient–provider relationship, have been found to be costeffective in reducing overall disease burden (Chisholm et al., 2005). This fact is intriguing, since the overall cost-effectiveness of medication adherence interventions has proven more difficult to assess (Elliott et al., 2005). However, few would claim such efforts are not well worth the investment to the patient or health system. Enhanced alliance or trust in providers has even help mediate adherence difficulty due to medication costs (Piette et al., 2005), not a trivial finding. We acknowledge possible limitations to these findings, beginning with the fact this was an observational study conducted in a naturalistic setting, with patients drawn from one urban clinic. Generalizability outside the VA population might also be difficult, since veterans are often considered different (older, predominantly male) and perhaps sicker than other patients. However, this group is quite similar to other indigent patient populations such as Medicaid recipients, except regarding gender, while medication adherence and appropriate therapeutic relationships extend to all patient groups. Adherence here was defined by a cross-sectional patient self-report of two validated measures. While memory issues and other subjectivity problems with such disclosure are recognized, this rich approach to primary data collection offers unique insights into patient perspectives and is a practical method well suited to naturalistic settings. Future reports from the CIVIC-MD project will incorporate medical chart reviews and administrative pharmacy data, as well as longitudinal follow-up to address issues of causal direction. Finally, we were unable to incorporate the influence of medication side effects. While some evidence suggests that attitudes and health beliefs are better predictors of nonadherence (Scott and Pope, 2002), side effects are consistently documented as one of the most significant

reasons for drug treatment failure (Matson et al., 2006; Johnson et al., 2007). Another benefit of improved patient–provider relationships is the enhanced opportunities for clinicians to cooperatively work with their patients to better manage drug side effects, such as switching medications if necessary or more openly discussing other treatment options. In conclusion, patient perceptions of the therapeutic alliance and mental healthcare setting are highly associated with medication adherence. Recognizing how veterans with debilitating conditions such as bipolar disorder view their treatment environment will enable the VA or other organizations to better target clinical relationships towards improving medication adherence, quality of life, and outcomes. These findings support intervention efforts aimed at strengthening patient– provider bonds, focusing upon appropriate psychopharmacology and treatment retention for individuals experiencing adherence problems. Such efforts are clearly justified when striving to improve the quality of care provided to this challenging population, and should likely yield positive clinical outcomes. Role of funding source As noted in the acknowledgment, funding for this study was provided by the Department of Veterans Affairs, Health Services Research and Development Service (Merit Review IIR 02-283-2, MREP Career Development Award HSR&D 02269). The VHA had no involvement in this research study. This includes data collection and analysis, the writing of this manuscript, and decision to submit for publication.

Conflict of interest The authors individually attest that they have no financial or personal relationships that would possibly interfere with this research study or potentially influence these results. As corresponding author, I have personally received direct confirmation of this fact with each co-author.

Acknowledgements This study was supported through funds provided by the Department of Veterans Affairs, Health Services Research and Development Service (Merit Review IIR 02-283-2, MREP Career Development Award HSR&D 02269). Dr. Copeland is funded by the Career Development Award Merit Review Entry Program from the VA Health Services Research and Development program (MREP 05-145). Dr. Fine was supported in part by a mid career development award (5K24AI1769) from the National Institute of Allergy and Infectious Diseases. In addition, the authors wish to acknowledge and appreciate resources provided by the VERDICT Research Program in San Antonio, the VA's Center for Health Equity Research and Promotion

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

(CHERP) in Pittsburgh, and the VISN 4 MIRECC. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs. References Adams, J., Scott, J., 2000. Predicting medication adherence in severe mental disorders. Acta Psychiatr. Scand. 101 (2), 119–124. Bauer, M., Unutzer, J., Pincus, H.A., Lawson, W.B., 2002. Bipolar disorder. Ment. Health Serv. Res. 4 (4), 225–229. Berk, M., Berk, L., Castle, D., 2004. A collaborative approach to the treatment alliance in bipolar disorder. Bipolar Disord. 6 (6), 504–518. Blackwell, B., 1997. Treatment compliance and the therapeutic alliance. Harwood Academic Publishers, Amsterdam. Blow, F.C., McCarthy, J.F., Valenstein, M., Austin, K., Gillon, L., 2005. Care for Veterans with Psychosis in the VHA, FY04: 6th Annual National Psychosis Registry Report. Ann Arbor, MI: VA National Serious Mental Illness Treatment Research and Evaluation Center (SMITREC). Chisholm, D., van Ommeren, M., Ayuso-Mateos, J.L., Saxena, S., 2005. Cost-effectiveness of clinical interventions for reducing the global burden of bipolar disorder. Br. J. Psychiatry 187, 559–567. Danion, J.M., Neunreuther, C., Krieger-Finance, F., Imbs, J.L., Singer, L., 1987. Compliance with long-term lithium treatment in major affective disorders. Pharmacopsychiatry 20 (5), 230–231. Elliott, R.A., Barber, N., Horne, R., 2005. Cost-effectiveness of adherence-enhancing interventions: a quality assessment of the evidence. Ann. Pharmacother. 39 (3), 508–515. Fenton, W.S., Blyler, C.R., Heinssen, R.K., 1997. Determinants of medication compliance in schizophrenia: empirical and clinical findings. Schizophr. Bull. 23 (4), 637–651. Fleck, D.E., Hendricks, W.L., DelBello, M.P., Strakowski, S.M., 2002. Differential prescription of maintenance antipsychotics to African–American and white patients with new-onset bipolar disorder. J. Clin. Psychiatry 63 (8), 658–664. Fleck, D.E., Keck Jr, P.E., Corey, K.B., Strakowski, S.M., 2005. Factors associated with medication adherence in African–American and white patients with bipolar disorder. J. Clin. Psychiatry 66 (5), 646–652. Gaudiano, B.A., Miller, I.W., 2006. Patients' expectancies, the alliance in pharmacotherapy, and treatment outcomes in bipolar disorder. J. Consult. Clin. Psychol. 74 (4), 671–676. George, C.F., Peveler, R.C., Heiliger, S., Thompson, C., 2000. Compliance with tricyclic antidepressants: the value of four different methods of assessment. Br. J. Clin. Pharmacol. 50 (2), 166–171. Gitlin, M.J., Cochran, S.D., Jamison, K.R., 1989. Maintenance lithium treatment: side effects and compliance. J. Clin. Psychiatry 50 (4), 127–131. Glick, H.A., McBride, L., Bauer, M.S., 2003. A manic-depressive symptom self-report in optical scanable format. Bipolar Disord. 5 (5), 366–369. Havens, L.L., Ghaemi, S.N., 2005. Existential despair and bipolar disorder: the therapeutic alliance as a mood stabilizer. Am. J. Psychother. 59 (2), 137–147. Howard, K., Rickels, K., Mock, J.E., Lipman, R.S., Covi, L., Baumm, N.C., 1970. Therapeutic style and attrition rate from psychiatric drug treatment. J. Nerv. Ment. Dis. 150 (2), 102–110. Johnson, F.R., Ozdemir, S., Manjunath, R., Hauber, A.B., Burch, S.P., Thompson, T.R., 2007. Factors that affect adherence to bipolar

61

disorder treatments: a stated-preference approach. Med. Care 45 (6), 545–552. Keck Jr, P.E., McElroy, S.L., Strakowski, S.M., Bourne, M.L., West, S.A., 1997. Compliance with maintenance treatment in bipolar disorder. Psychopharmacol. Bull. 33 (1), 87–91. Keck Jr, P.E., McElroy, S.L., Strakowski, S.M., Stanton, S.P., Kizer, D.L., Balistreri, T.M., Bennett, J.A., Tugrul, K.C., West, S.A., 1996. Factors associated with pharmacologic noncompliance in patients with mania. J. Clin. Psychiatry 57 (7), 292–297. Kilbourne, A.M., Good, C.B., Sereika, S.M., Justice, A.C., Fine, M.J., 2005. Algorithm for assessing patients' adherence to oral hypoglycemic medication. Am. J. Health-syst. Pharm. 62 (2), 198–204. Kilbourne, A.M., Post, E., Good, C.B., Zeber, J.E., Copeland, L.A., Fine, M., Pincus, H.A., 2007. Therapeutic drug and metabolic syndrome risk monitoring in patients with bipolar disorder. J. Affect. Disord. 102 (1–3), 145–151. Klein, D.N., Schwartz, J.E., Santiago, N.J., Vivian, D., Vocisano, C., Castonguay, L.G., Arnow, B., Blalock, J.A., Manber, R., Markowitz, J.C., Riso, L.P., Rothbaum, B., McCullough, J.P., Thase, M.E., Borian, F.E., Miller, I.W., Keller, M.B., 2003. Therapeutic alliance in depression treatment: controlling for prior change and patient characteristics. J. Consult. Clin. Psychol. 71 (6), 997–1006. Lingam, R., Scott, J., 2002. Treatment non-adherence in affective disorders. Acta Psychiatr. Scand. 105 (3), 164–172. Ludman, E.J., Simon, G.E., Rutter, C.M., Bauer, M.S., Unutzer, J., 2002. A measure for assessing patient perception of provider support for self-management of bipolar disorder. Bipolar Disord. 4 (4), 249–253. Manning, J.S., 2005. Burden of illness in bipolar depression. Prim. Care Companion J. Clin. Psychiat. 7 (6), 259–267. Matson, J.L., Gonzalez, M.L., Smith, K.R., Terlonge, C., Thorson, R.T., Dixon, D.R., 2006. Assessing side effects of pharmacotherapy treatment of bipolar disorder: a 20-year review of the literature. Res. Develop. Disabilities 27 (5), 467–500. Meier, P.S., Barrowclough, C., Donmall, M.C., 2005. The role of the therapeutic alliance in the treatment of substance misuse: a critical review of the literature. Addiction 100 (3), 304–316. Morisky, D.E., Green, L.W., Levine, D.M., 1986. Concurrent and predictive validity of a self-reported measure of medication adherence. Med. Care 24 (1), 67–74. Murray, C.J., Lopez, A.D., 1996. Evidence-based health policy-lessons from the Global Burden of Disease Study. Science 274 (5288), 740–743. Opolka, J.L., Rascati, K.L., Brown, C.M., Gibson, P.J., 2003. Role of ethnicity in predicting antipsychotic medication adherence. Ann. Pharmacother. 37 (5), 625–630. Piette, J.D., Heisler, M., Krein, S., Kerr, E.A., 2005. The role of patient–physician trust in moderating medication nonadherence due to cost pressures. Arch. Intern. Med. 165 (15), 1749–1755. Pope, M., Scott, J., 2003. Do clinicians understand why individuals stop taking lithium? J. Affect. Disord. 74 (3), 287–291. Post, R.M., Denicoff, K.D., Leverich, G.S., Altshuler, L.L., Frye, M.A., Suppes, T.M., Rush, A.J., Keck Jr, P.E., McElroy, S.L., Luckenbaugh, D.A., Pollio, C., Kupka, R., Nolen, W.A., 2003. Morbidity in 258 bipolar outpatients followed for 1 year with daily prospective ratings on the NIMH life chart method. J. Clin. Psychiatry 64 (6), 680–690. Ricker, M., Nystul, M., Waldo, M., 1999. Counselors' and clients' ethnic similarity and therapeutic alliance in time-limited outcomes of counseling. Psychol. Rep. 84 (2), 674–676. Sajatovic, M., Davies, M., Bauer, M.S., McBride, L., Hays, R.W., Safavi, R., Jenkins, J., 2005. Attitudes regarding the collaborative

62

J.E. Zeber et al. / Journal of Affective Disorders 107 (2008) 53–62

practice model and treatment adherence among individuals with bipolar disorder. Compr. Psychiatry 46 (4), 272–277. Sajatovic, M., Valenstein, M., Blow, F.C., Ganoczy, D., Ignacio, R.V., 2006. Treatment adherence with antipsychotic medications in bipolar disorder. Bipolar Disord. 8 (3), 232–241. Salloum, I.M., Cornelius, J.R., Daley, D.C., Kirisci, L., Himmelhoch, J.M., Thase, M.E., 2005. Efficacy of valproate maintenance in patients with bipolar disorder and alcoholism: a double-blind placebo-controlled study. Arch. Gen. Psychiatry 62 (1), 37–45. Saunders, J.B., Aasland, O.G., Babor, T.F., de la Fuente, J.R., Grant, M., 1993. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption—II. Addiction 88 (6), 791–804. Scott, J., Pope, M., 2002. Self-reported adherence to treatment with mood stabilizers, plasma levels, and psychiatric hospitalization. Am. J. Psychiatry 159 (11), 1927–1929. Shalansky, S.J., Levy, A.R., Ignaszewski, A.P., 2004. Self-reported Morisky score for identifying nonadherence with cardiovascular medications. Ann. Pharmacother. 38 (9), 1363–1368. Swann, A.C., 2005. Long-term treatment in bipolar disorder. J. Clin. Psychiatry 66 (Suppl 1), 7–12.

Swartz, M.S., Swanson, J.W., Hiday, V.A., Borum, R., Wagner, H.R., Burns, B.J., 1998. Violence and severe mental illness: the effects of substance abuse and nonadherence to medication. Am. J. Psychiatry 155 (2), 226–231. Tonigan, J.S., 2003. Project Match treatment participation and outcome by self-reported ethnicity. Alcohol Clin. Exp. Res. 27 (8), 1340–1344. Valenstein, M., Blow, F.C., Copeland, L.A., McCarthy, J.F., Gillon, L., Zeber, J.E., Bingham, C.R., Stavenger, T., 2004. Poor adherence with antipsychotic medication among patients with schizophrenia. Schizophr. Bull. 30 (2), 255–264. Weiden, P.J., Rao, N., 2005. Teaching medication compliance to psychiatric residents: Placing an orphan topic into a training curriculum. Acad. Psychiatry 29 (2), 203–210. Zeber, J.E., Grazier, K.L., Valenstein, M., Blow, F.C., Lantz, P.M., 2007. The effect of a medication copayment increase on veterans with schizophrenia. Am. J. Manag. Care 13 (6, part 2), 335–346. Zeber, J.E., Copeland, L.A., Miller, A.L., Kilbourne, A.M., Velligan, D., Mortensen, E.M., 2006. Ethnicity and the impact of higher medication copayments among veterans with schizophrenia (presentation at Veterans Affairs HSRD annual meeting, 2006).