Polypharmacy and bipolar disorder: what's personality got to do with it?

Abstract

The majority of patients treated for bipolar disorder receive multiple psychotropic medications concurrently (polypharmacy), despite a lack of empirical evidence for any combination of three or more medications. Some patients benefit from the skillful management of a complex medication regimen, but iterative additions to a treatment regimen often do not lead to clinical improvement, are expensive, and can confound assessment of the underlying mood disorder. Given these potential problems of polypharmacy, this paper reviews the evidence supporting the use of multiple medications and seeks to identify patient personality traits that may put patients at a greater risk for ineffective complex chronic care. Patients with bipolar disorder (n = 89), ages 18 and older, were assessed on the Montgomery Asberg Depression Rating Scale (MADRS), Young Mania Rating Scale (YMRS), and the NEO Five Factor Inventory (NEO-FFI), and completed a treatment history questionnaire to report psychotropic medication use. We found that patients with lower scores on openness had significantly more current psychotropic medications than patients with higher scores on openness (3.7 ± 1.9 vs. 2.8 ± 1.8, p < 0.05). Patients with the highest lifetime medication use had significantly lower extraversion (21.8 ± 8.9 vs. 25.4 ± 7.6, p < 0.05) and lower conscientiousness (21.9 ± 8.2 vs. 27.9 ± 8.2, p < 0.01) than those reporting lower lifetime medication use. Low levels of openness, extraversion, and conscientiousness may be associated with increased psychotropic medication use. Investigating the role of individual differences, such as patient personality traits, in moderating effective polypharmacy warrants future research.

Bipolar disorder, medication, personality, polypharmacy

Introduction

The majority of patients treated for bipolar disorder receive multiple psychotropic medications concurrently (Post et al.1996; Freeman & Stoll, 1998; Frye et al.2000; Zarate & Quiroz, 2003; Lin et al.2006; Baldessarini et al.2008; Goldberg et al.2009). This practice, often referred to as polypharmacy, can have both positive and negative consequences. On the one hand, combining treatments can enhance efficacy; however, ineffective complex regimens can be expensive and have adverse effects that confound assessment of the primary mood disorder. Clinicians can better serve their patients by formulating treatment plans with an awareness of the positive and negative aspects of polypharmacy. This paper attempts to help clinicians parse the promise of polypharmacy from the problematic by reviewing the evidence-base comprised of randomized trials of combined medication regimens and examining the association of personality factors with the progression to potentially dangerous polypharmacy.

Poylpharmacy in the context of treatment for bipolar disorders

Polypharmacy is recognized as the rule rather than the exception in the treatment of bipolar disorder (Mojtabai & Olfson, 2010). In contrast to the 20th century view that lithium was a sufficient treatment for most patients with bipolar disorder, recent data show few patients receive a single psychotropic agent. Review of treatment utilization by patients in the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) found that less than 20% received monotherapy (Goldberg et al.2009; Sachs et al.2011a). Among inpatients treated for bipolar disorder at the National Institute of Mental Health (NIMH), Frye et al. showed a 10-fold increase in the percentage of patients receiving three or more psychotropic medications at discharge between 1974 and 1995 (Frye et al.2000). The practice of polypharmacy is not limited to tertiary treatment centres. A query of the Medco claims database revealed that among 8073 subscribers with a bipolar I diagnosis, 33% were exposed to at least four concurrent psychotropic medications (Private communication, Jon Morris). Despite recent quality improvement initiatives (Goren et al.2008) and physician training programmes (Thompson et al.2008) intended to reduce the polypharmacy, the rate of polypharmacy in outpatient psychiatric practice continues to increase (Mojtabai & Olfson, 2010).

The practice of polypharmacy can be understood as consisting of three distinct segments: (1) evidence-based combinations; (2) effective complex care; and (3) ineffective complex chronic care. Evidence-based combinations are those multidrug combinations that have positive results in adequately powered well-designed clinical trials. The literature (see Fig. 1) does support the notion that in some circumstances combining agents with proven efficacy for one or more phases of bipolar disorder may be more effective than monotherapy.

Fig. 1

Summary of atypical antipsychotics as adjunct to mood stabilizer monotherapy. (1) Risperidone + MS vs. MS. (2) Risperidone + MS vs. MS. (3) Olanzapine + MS vs. MS. (4) Quetiapine + MS vs. MS. (5) Quetiapine + MS vs. MS. (6) Aripiprazole + MS vs. MS. * ⩾50% reduction YMRS total score, except in Sachs et al. (2002), where data refers to patients much improved/very much improved on the CGI score ** Randomizing patients partially non-responsive to lithium or valproate.

The bulk of this literature involves the use of a dopamine-blocking agent in combination with lithium or valproate for acute mania (Sachs et al.2002, 2004; Tohen et al.2002; Yatham et al.2003, 2007; Vieta et al.2008a,b). This provides the evidence base for several national and international guidelines that recommend combining treatments when there is inadequate response to monotherapy (See Table 1). For bipolar depression, we are aware of only three placebo-controlled studies that reported results for combination treatments compared to monotherapy. STEP-BD did not find any advantage of adding standard antidepressants (bupropion or fluoxetine) over placebo for bipolar depression (Sachs et al.2007)). The combination of olanzapine and fluoxetine was, however, found to be significantly more efficacious than either placebo or fluoxetine monotherapy (Tohen et al.2003). The addition of lamotrigine was found to be superior to adding placebo for lithium treated patients with acute bipolar depression (van der Loos et al.2009).

Table 1

Summary of guideline recommendations for acute mania

	WFSBP (Grunze et al., 2009)	BAP (Goodwin et al., 2009)	CANMAT (Yatham et al., 2009)	ATALAN (Vieta et al., 2010)
Recommendations for initial therapy	Choose monotherapy with a recommended grade 1A agent – ARI, RISP, ZIP* or LI#	For severe mania or for mixed patients: oral atypical antipsychotic or valproate For less severe patients, can consider lithium or carbamazepine	Either monotherapy or combination therapy with Li/Val plus ARI, OLZ, QTP or RISP recommended as first-line treatment	Either monotherapy or combination therapy recommended, depending on patient's previous history of response
If partial response to initial therapy	Optimize dosage and, if no improvement, add a second grade 1A agent	*Optimize dose of first-line agent If symptoms are inadequately controlled and/or mania is severe, consider combination of lithium or valproate with an antipsychotic	Optimize dose and check compliance *If insufficient response to lithium/valproate, add or switch to AP *If insufficient response to AP, add or switch to MS *If insufficient response to AP+MS in combination, replace one or both agents with other recommended first-line agents	*Switch therapy to another Level 1 or 2 medication Treatment refractory or highly complex mania should be treated by specialist teams who are likely to use a complex combination medication regimen
If no response to initial therapy	After 2 wk, switch to an alternative grade 1A agent. If still no response, add a second grade 1 agent

ZIP has grade I recommendations in some countries due to potential cardiotoxicity concerns. Li is considered a grade 1 agent if considering long-term treatment from the outset; otherwise has grade 2 recommendation for acute treatment of mania.

We did not find any placebo-controlled relapse prevention studies comparing monotherapy to combination therapies. However, the best available evidence comes from the BALANCE study (Geddes et al.2010). In this open, randomized trial all subjects started open treatment with lithium and valproate. Subjects who tolerated this treatment were randomized to continue this combination treatment, discontinue lithium or discontinue valproate. This design produces a sample enriched for the ability to tolerate the combination of lithium and valproate rather than necessarily response to combination treatment. The investigators found a significant advantage of combination therapy over valproate monotherapy, but not for lithium monotherapy. Since this study employed a discontinuation paradigm, the results must be viewed with caution. It is also possible that maintaining therapeutic lithium levels protects against recurrence due to valproate discontinuation, but valproate does not have this same effect when lithium is discontinued. While the fund of knowledge supporting these evidence-based combinations is limited, the effect size associated with two-drug combinations compared to monotherapy plus placebo is of a similar magnitude to that of active monotherapy over placebo (Sachs & Gardner-Schuster, 2007; Sachs et al.2011a).

Although there are no adequate clinical trials demonstrating the efficacy of combinations of three or more medications for any bipolar indication, case reports suggest that some patients may benefit from complex regimens resulting from iterative additions and subtractions based on clinical empiricism (Conforti et al.1999; Post, 2007). Admittedly, this form of personalized treatment produces uncontrolled individual outcomes, which do not meet the standards of evidence-based medicine. Results emerging from individualized care are, however, the mainstay of clinical care. Attibution of benefit from this type of polypharamcy may not be confident, but it is an acceptable guide for clinical management, especially when such treatment-emergent outcomes result from a systematic measure-based process. We may, therefore, leave aside any quibble regarding the merits of ‘effective complex care’: polypharmacy administered to patients for whom excellent results might conceivably be attributed to a skillful treatment regimen of four or more psychotropic medications.

The main concern of this paper is the path that generally begins with inadequate response to proven treatments, proceeds to polypharmacy via iterative additions to the treatment regimen, but does not lead to clinical improvement and, thus, evolves into ineffective complex chronic care (ICCC). This group can be defined as those patients who remain ill despite receiving five or more medications for six or more months. ICCC is quite common in the treatment of bipolar disorder and describes the majority of bipolar patients seen for consultation in our programme.

Why is ICCC common in the management of bipolar disorder? Exploring this question, we consider the nature of bipolar disorder, the state of clinical art regarding treatment for bipolar disorder, prescriber characteristics and patient characteristics

ICCC may in part result from inadequacies of the monotherapy and combination treatments currently proven effective for bipolar disorder. While these treatments are clearly better than placebo, the rates of ICCC make it obvious that much clinical need remains unmet. ICCC may also be common based on the nature of the disorder and the medications used to treat the illness, as well as prescriber and patient characteristics.

Bipolar disorder is a complex chronic condition frequently accompanied by other psychiatric and medical conditions (Baldassano, 2006; Magalhaes et al.2011). Such a course requires shifting therapeutic priorities and dilemmas to treat the disorder due to uncertainties arising as patients make irregular transitions between acute episodes of varying polarity and maintenance phases. Clinicians must also take into account the lag time between administration of a medication and its effect (Preskorn & Lacey, 2007). This makes management of bipolar disorder inherently difficult. Therefore, it is not surprising that many psychiatrists regard bipolar disorder as the most challenging condition they treat (Bishop et al.2008), and that patients with bipolar disorder often receive multiple medications concomitantly.

Preskorn suggests that characteristics of the medications themselves may contribute to the tendency for polypharmacy (Preskorn, 2006). He speculates that increasing rates of polypharmacy seen in psychiatry might reflect a confluence of recent trends, including an increase in the number of approved medications available and a tendency for newer medications to have more specific actions and better tolerability, but at the price of being less effective than older medications.

Prescriber characteristics also likely influence the practice of polypharmacy. We were, however, unable to find any studies reporting data regarding prescriber characteristics. There are reports relating patient characteristics with the propensity for polypharmacy. For example, Goldberg et al. (2009) reported findings from STEP-BD that complex regimens are especially common in patients with substantial depressive illness burden and suicidality, since these conditions may require complex drug regimens to produce acceptable levels of response (Goldberg et al.2009). In addition to indicators of mood disorder severity, data suggest a plausible role of personality factors in bipolar disorder on the development of ICCC. Using the NEO Five Factor Personality Inventory (NEO-FFI), Barnett et al. (2011) examined personality characteristics of subjects enrolled in STEP-BD and reported that individuals with bipolar disorder tend to have high levels of neuroticism and openness on average, but low levels of extraversion, agreeableness, and conscientiousness, in comparison to population norms (Barnett et al.2011).

To investigate the role of personality factors in polypharmacy and identify personality factors that increase the risk of maintaining complex and ineffective medication regimens, we conducted a systematic review of our clinical database for patients seen for consultation at the Massachusetts General Hospital (MGH) Bipolar Clinic and Research Program. We hypothesized that high levels of neuroticism and low levels of conscientiousness would be associated with the use of a higher number of current and lifetime psychotropic medications.

Methods

With approval from the MGH Internal Review Board, we selected 100 consecutive records for patients seen for consultation between January 2009 and December 2010 in the MGH Bipolar Clinic and Research Program. Inclusion criteria for this analysis required that patients be over age 18, request consultation for treatment recommendations, have records that included computer administered assessments of current and lifetime treatment history, and have the Montgomery-Åsberg Depression Rating Scale (MADRS; Montgomery & Asberg, 1979), Young Mania Rating Scale (YMRS; Young et al.1978), and the NEO-FFI (Costa et al.1992) obtained within 7 d prior to their initial visit.

Personality assessment

The revised NEO-FFI Personality was collected and scored by computer using the NEO-FFI (Costa et al.1992), a 60-item self-report questionnaire. The NEO-FFI assesses each of five personality dimensions based on twelve items (raw scores for each items are obtained from a common five-point scale): neuroticism (N), the tendency to experience negative affect; extraversion (E), a tendency towards energy, positive emotions, and stimulation-seeking; conscientiousness (C), a tendency towards self-discipline and dutiful behaviour; openness to experience (O), sometimes described as intellectual curiosity; and agreeableness (A), a tendency to be cooperative and compassionate. Scores were converted to sex-adjusted t-scores with a (normative) mean of 50 and a standard deviation of 10 using normative data.

Symptom assessment

Interactive computer interviews (ICI) (provided by Concordant Rater Systems, USA) for YMRS and MADRS have demonstrated good agreement with the YMRS and MADRS administered by well-trained site-based raters (Reilly-Harrington et al.2010; Sachs et al.2011b). Unlike self-report scales, the ICI mimics clinician administration of rating scales by presenting a sequence of probes similar to the sequence of questions in the scripted interview guide site-based raters are trained to use. Subjects respond to the questions in multiple-choice format. Depending on the subject's response, the computer selects follow-up questions as necessary to map the subject's responses to the anchor points for each scale item.

Assessment of current and lifetime psychotropic medication use

The treatment history questionnaire administered by the computer collected information about use of 41 commonly prescribed psychotropic medications listed by brand and generic names grouped in six classes (mood stabilizers, antidepressants, dopamine blockers, anxiolytics/sedative hypnotics, stimulants, and other) and allowed free entry of treatments not listed. For each entry patients indicated the treatment as ‘never used’, 'current use, ‘past use’, or ‘unknown’). For the analyses, current medications were defined as those designated by the patient's report as ‘current use’ (prescribed during the week prior to the scheduled evaluation) unless specifically amended by the consulting clinician's report. Lifetime medications were counted as those reported by the patient as ‘past use’ plus those reported as ‘current use’.

Subgroups were defined by splitting the samples based on mean psychotropic medication use: current psychotropic medications (⩽3 = low vs. ⩾4 = high) and lifetime medications (⩽10 = low vs. ⩾11 = high). A group of highest lifetime medications utilizers was defined based on lifetime psychotropic use ⩾18 (mean + 1 s.d.) and compared to the subgroup reporting lifetime psychotropic use ⩽18. Student's t-test was used to compare the median scores on each of the five NEO-FFI dimensions. All analyses were considered exploratory and were carried out using Stata version 11.0 statistical software without adjustment for multiple comparisons.

Results

Electronic search identified 100 consecutive consultation records. Data sufficient for inclusion in the personality analysis were available for 48 females and 41 males (N = 89). Other sample characteristics are shown in Table 2. Of the cases (n = 56) that had treatment recommendations recorded, 86% (n = 48) included a recommendation for ‘management by subtraction’ and of these cases, 42% (n = 20) indicated a willingness to accept the recommendation to taper an existing medication.

Table 2

Sample characteristics

	N	Mean	Median	s.d.
Age	89	38.0	36	14.4
YMRS	87	18.9	19	9.5
MADRS	87	26.3	28	13.2
Neuroticism	89	28.2	28	5.6
Extraversion	89	24.5	24	8.0
Openness	89	30.7	31	6.9
Agreeableness	89	29.6	31	6.9
Conscientiousness	89	26.3	26	8.9
Current psychoactive medications	89	3.2	3	1.8
Lifetime psychoactive medications	89	11.5	12	7.5

Comparisons of subgroups defined by high vs. low NEO-FFI scores are summarized in Table 3. The subgroup with low openness had significantly more current psychotropic medications (3.7 ± 1.9) than patients in the higher distribution of openness scores (2.8 ± 1.8, p < 0.05). There were no other comparisons that reached statistical significance for either current or lifetime medications.

Table 3

Use of Medication in upper and lower 50% by NEO dimension

	Current medications mean	s.d.	Lifetime medications mean	s.d.
Neuroticism upper 50%	3.2	1.8	11.0	7.4
Neuroticism lower 50%	3.0	1.9	12.4	8.3
Extraversion upper 50%	3.0	1.9	11.1	7.7
Extraversion lower 50%	3.5	1.7	12.3	7.9
Openness upper 50%	2.8	1.8	11.2	8.0
Openness lower 50%	3.7*	1.9	12.2	7.0
Agreeableness upper 50%	3.2	1.8	10.9	6.4
Agreeableness lower 50%	2.9	2.0	12.6	10.1
Conscientiousness upper 50%	2.8	1.9	10.6	6.3
Conscientiousness lower 50%	3.5	1.7	13.3	9.9

*

significant at p < 0.05.

Comparisons of subgroups defined by current and lifetime medication usage are summarized in Table 4. For both current and lifetime psychotropic medication usage, there were no significant differences found between the high vs. low psychotropic medication users on any of the NEO-FFI personality dimensions, with the exception that the subgroup using 18 or more psychotropic medications had significantly lower extraversion (21.8 ± 8.9 vs. 25.4 ± 7.6, p < 0.05) and lower conscientiousness (21.9 ± 8.2 vs. 27.9 ± 8.2, p < 0.01) than those reporting lifetime usage of fewer than 18 psychotropic medications.

Table 4

Median personality scores split by utilization of psychotropic medications

Medication use	N	E	O	A	C
Current ⩾4 (n = 46)	30.8 ± 6.2	24.4 ± 8.1	30.3 ± 7.1	30.8 ± 6.2	26.7 ± 8.4
Current <4 (n = 43)	28.4 ± 7.3	24.5 ± 8.1	31.3 ± 6.6	28.4 ± 7.5	25.9 ± 9.4
Lifetime ⩾11 (n = 56)	28.4 ± 5.3	24.5 ± 8.3	30.3 ± 6.6	29.5 ± 7.2	25.8 ± 8.1
Lifetime <11 (n = 33)	28.0 ± 6.1	24.4 ± 7.8	31.5 ± 7.4	29.0 ± 6.6	27.1 ± 9.3
Lifetime ⩾18 (n = 24)	29.0 ± 4.9	21.8 ± 8.9	29.6 ± 6.4	30.2 ± 7.5	21.9 ± 8.2
Lifetime <18 (n = 65)	28.0 ± 5.8	25.4 ± 7.6*	31.2 ± 7.0	29.04 ± 6.6	27.9 ± 8.2**

*

p < 0.05

**

p < 0.01

Discussion

Use of multiple psychotropic medications can be helpful as well as problematic for patients with bipolar disorder. There are two indicators of when polypharmacy may be helpful. The first is when clinicians use combinations of treatments consistent with evidence-based guidelines. Polypharmacy may also be acceptable for bipolar patients with inadequate response to proven combination treatments but who apparently benefit from effective complex care regimens. Beyond these subgroups, polypharmacy in the treatment of bipolar disorder often leads to ICCC (inadequate response to five or more psychotropic medications over six months).

ICCC is a common problem for patients with bipolar disorder and a challenge for their clinicians. The risk and cost of multiple ineffective medications cannot be justified. Thus, treatment guidelines recommending combination treatment for patients with unsatisfactory response to monotherapy should be applied judiciously. These recommendations are drawn from a modest number of double-blind clinical trials showing better efficacy for combinations of lithium or valproate with atypical agents. Moreover, these studies generally recruit and randomize subjects who have not responded to monotherapy. Such designs do not permit distinguishing whether combination treatment is more effective than simply increasing the dose of a previously ineffective monotherapy, nor whether the apparent response to combination treatment could represent a simple response to the newly-added medication. The current literature does not address the efficacy or effectiveness of polypharmacy for bipolar disorder involving more than two psychotropic agents, and does not offer guidance for prevention or management of ICCC. Prior reports are compatible with the suggestion that progression to ICCC may be associated with clinical severity and the availability of a large number of well-tolerated psychotropic medications with relatively selective mechanisms of action.

We hypothesized that personality dimensions such as high neuroticism and low conscientiousness would increase the risk that a patient will be exposed to ICCC. Specifically, we expected that psychotropic medication use would be associated with high neuroticism and low conscientiousness. Although we found high lifetime medication use associated with lower scores for extroversion and conscientiousness, we did not find the expected relationship between neuroticism and psychotropic medication use. This data may also be an indication that the number of current medications is not a sensitive measure of a patient's propensity for ICCC. The finding of higher current psychotropic medication utilization by patients with lower scores for openness was unexpected. It may be that our sample of patients coming for consultation is biased for patients who are more open to medication taking then they are to other management strategies.

This study is exploratory and has several important limitations, such as having a small sample size and being highly selective (i.e. individuals seeking a consultation in a speciality bipolar clinic). While our sample was likely biased to treatment-seeking individuals with refractory illness, the results are at least pertinent to such patients.

Our data does yield some interesting avenues of future research. For example, we found that even when patients report a medication as ineffective, stopping a longstanding treatment is not necessarily an attractive intervention for them. Future research should examine if the duration of exposure to an ineffective medication is correlated with reluctance to discontinue the ineffective agent. Other areas of future research include investigating the use of treatment contracts that clearly define treatment response based on standardized assessments, and the expectation of early discontinuation of ineffective interventions to reduce ICCC.

Further investigation is also warranted for the assessment of personality traits and ICCC as personality factors may be potential moderators of treatment success. Such work could allow treatment plans to be personalized for each patient, and even incorporate psychosocial strategies surrounding the utilization of psychotropic drugs. For example, self-management strategies may be more appealing than group therapy to individuals with low extraversion. Additionally, external supports could be encouraged for patients with low conscientiousness, and motivational interviewing may help manage low agreeableness.

Polypharmacy in the management of bipolar disorder may be expectable given aspects of the disorder (e.g. chronicity, comorbidity, uncertainty associated with shifting polarity of illness), characteristics of the modern medications used to treat the disorder (e.g. time delay in action, tolerability, selectivity) and the high frequency of inadequate response to proven therapies. Use of multiple medications may help some patients when carefully managed by skillful clinicians. How can clinicians responsibly manage polypharmacy strategies given the potentially harmful effects of polypharmacy and the risk of ICCC? The absence of evidence-based guidance for the management of bipolar patients with inadequate response to two psychotropic medications often leaves clinicians and their bipolar patients to face a confusing dilemma: when and how do we add another medication to ameliorate symptoms or side effects? Or, when and how do we taper off an apparently ineffective medication and try a new medication? Thus, it is vital that we begin to understand the factors that will enable us to parse the polypharmacy of effective complex care from ICCC. Our data indicate that individual patient characteristics such as personality factors, could be promising moderators, and warrant future research.

Acknowledgments

None.

Statement of Interest

Dr Sachs: grants/research support, consulting fees and honoraria within the last three years from Astra Zeneca, Abbott Labs, BMS, Cephalon, Eli Lilly, Forest, Merck, Organon, Otsuka, Pfizer, Sanofi-Aventis, Sepracor, Sunovion, Takeda, Wyeth. Employee: Massachusetts General Hospital, Bracket. Patent holder: Interactive computer interviews for rating scales in psychiatric research. Shares in Concordant Rater Systems, Medco Health Systems.

Dr Grunze: grants/research support, consulting fees and honoraria within the last three years from Astra Zeneca, Bial, BMS, Cephalon, Eli Lilly, Gedeon Richter, Janssen-Cilag, Merck, Organon, Pfizer Inc, Sanofi-Aventis, Sepracor, Servier and UBC. Neither he nor any member of his family have shares in any pharmaceutical company or could benefit financially from increases or decreases in the sales of any psychotropic medication.

Dr Sylvia: grants/research support from NIMH. Consulting fees from Bracket and Clintara.

Ms Peters: no disclosures to report.