February 2009  << Back  

  Can't open the PDF? Click here for help.



Association between antipsychotic combination therapy and treatment adherence among individuals with bipolar disorder

Frank D. Gianfrancesco, PhD

HECON Associates, Inc., Montgomery Village, MD, USA

Martha Sajatovic, MD

Case Western Reserve University, School of Medicine, Department of Psychiatry and University, Hospitals of Cleveland, Cleveland, OH, USA

Eskinder Tafesse, PhD

AstraZeneca Pharmaceuticals LP, Wilmington, DE, USA

Ruey-Hua Wang, MS

HECON Associates, Inc., Montgomery Village, MD, USA

BACKGROUND: This study investigated the effect on antipsychotic treatment adherence of combining quetiapine or risperidone with lithium, anticonvulsants, and/or antidepressants among bipolar individuals with predominantly manic/mixed or depressive symptoms.

METHODS: Treatment episodes with quetiapine or risperidone were identified from individuals with medical claims for bipolar/manic disorder. Multiple regression analysis was used to evaluate the impact of antipsychotic combinations on treatment adherence, as measured by intensity (medication possession ratio [MPR]) and treatment duration.

RESULTS: Among mixed/manic individuals, combination therapies were associated with lower antipsychotic MPRs than monotherapy (P < .05), with MPR decreasing with number of medications. Quetiapine showed a similar pattern among depressed individuals, whereas risperidone showed a weaker association. For both subgroups, antipsychotic combinations with anticonvulsants were associated with lower MPRs than combinations with lithium. For manic/mixed individuals, combining quetiapine with an anticonvulsant and lithium was associated with shorter treatment durations than quetiapine alone (P ≤ .05). For manic/mixed individuals receiving risperidone, treatment duration was not affected by type of combination therapy. Among depressed subgroups, treatment duration was unrelated to combination therapy.

CONCLUSIONS: Adherence to quetiapine and risperidone may be affected by whether they are prescribed alone or as combination therapy, the combination strategy, and the predominant symptomatology of the individual.

KEYWORDS: combination, bipolar disorder, adherence, quetiapine, risperidone



Polypharmacy, the use of 2 or more medications to treat the same medical condition, is an increasingly common treatment for bipolar disorder.1,2 Prescribing physicians may use a combination of therapies to improve functionality and protect against symptomatic relapse in individuals who do not respond to monotherapy, often as a consequence of treatment resistance or a particularly severe manifestation of the disease.1,3,4 Frequently used treatment strategies combine atypical antipsychotics with traditional mood stabilizers, such as lithium and valproate, other anticonvulsants, or antidepressants.3,5 A retrospective review of 178 individuals hospitalized with refractory bipolar disorder or unipolar depression revealed that the proportion of individuals receiving simultaneous treatment with 3 or more psychotropic medications increased steadily from 3.3% in 1974-1979 to 44% in 1990-1995.6 In a recent European survey of psychiatric inpatients, individuals with euphoric or mixed-state mania were receiving an average of 2.9 and 3.3 psychotropic agents, respectively.7

The usual strategy for treating patients with bipolar disorder is to commence treatment with a single, traditional mood stabilizer or antipsychotic and to supplement with additional medication if the patient does not respond or experiences recurrence of symptoms.8,9 Treatment guidelines reflect clinical practice, and the American Psychiatric Association, the British Association for Psychopharmacology, and the Canadian Network for Mood and Anxiety Treatments acknowledge that giving acutely ill individuals with manic/mixed symptoms an antipsychotic in combination with either lithium or valproate may enhance effectiveness beyond that provided by any of these agents given alone.10-12 For individuals with predominantly depressive symptoms who fail to respond to monotherapy, guidelines collectively endorse combining an antipsychotic or traditional mood stabilizer with an antidepressant in order to achieve symptom control.11-12

In clinical practice, however, there are problems in evaluating the incremental benefits of combination therapy; subsequent effectiveness may be derived from the supplementary medication(s) rather than the medication combination per se, and the addition of medication may be confounded with the natural progression of the illness (worsening or spontaneous subsidence of symptoms).4 Randomized clinical trials avoid these potentially confounding elements and provide convincing evidence in support of combination therapy. For example, studies have shown that approximately 20% more individuals with acute mania respond to combination therapy with lithium or valproate and an atypical antipsychotic than to monotherapy with an antimanic agent.13 However, most clinical studies involving atypical antipsychotics have either been placebo-controlled or have compared antipsychotic–mood stabilizer combinations with mood stabilizer monotherapy. Very few have compared atypical antipsychotics in combination with mood stabilizers and/or antidepressants with atypical monotherapy.1,14,15 In addition, trials have focused on drug efficacy rather than outcomes such as adherence.

General clinical evidence suggests that the use of multiple medications may increase the risk of patient nonadherence.16 If the patient perceives that he or she has been overmedicated, that patient may be more likely to taper the medications or stop taking them altogether.17 Complicated treatment regimens may also promote nonadherence.18 In a recent study of patients with schizophrenia for whom treatment was characterized by significant polypharmacy, 37% were experiencing problems with treatment adherence.19 A therapeutic regimen that is difficult to tolerate may also decrease treatment adherence among individuals with bipolar disorder,15,20 and exposure to multiple medications may increase the risk of adverse events. In a large clinical trial of olanzapine and traditional mood stabilizers, patients on combination therapy vs mood stabilizer monotherapy were more likely to discontinue medication as a result of treatment-emergent adverse events.13 Considering the high rate of partial or complete nonadherence among individuals with bipolar disorder who are prescribed antipsychotics (33% to 48%),21-23 and the frequency of combination therapy in clinical practice, it is reasonable to suspect that polypharmacy may be at least partly accountable.

The purpose of this retrospective claims-based study was to assess and compare treatment adherence to quetiapine or risperidone as monotherapy or in combination with traditional mood stabilizers and antidepressants among commercially insured individuals with bipolar disorder. Quetiapine and risperidone were selected on the basis of their similar effects on body weight in the short term and long term.24

Treatment effects on the regularity of prescription refills were assessed using medication possession ratios (MPRs), and the effects of combination treatment on the duration (or persistence) of treatment with each antipsychotic were also examined. Since treatment adherence may also be differentially affected by bipolar subtype, individuals with predominantly manic or mixed symptoms and those with predominantly depressive symptoms were studied as separate groups.25,26 A retrospective study design was deemed appropriate for investigating the effects of combination therapy on treatment adherence for 2 reasons. First, cost constraints make it unlikely that a large clinical trial comparing a wide range of treatment combinations for bipolar disorder will take place.4 Second, a retrospective study may be more appropriate than a prospective trial for assessing adherence, since the behavior of both the patient and the physician may be influenced by participants’ awareness of being in a trial.27


Study population and data collection

A data extract for individuals with bipolar or manic disorder from the PharMetrics (Watertown, MA) database, covering the period January 1999 through May 2005, was used for this study. Individuals with bipolar or manic disorder were identified by the corresponding International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM)28 codes reported on medical claims (296.4 to 296.8 for bipolar disorder, 296.0 and 296.1 for manic disorder). Individuals were excluded if they had ICD-9-CM codes for schizophrenia (295.x) or major depression (296.2 and 296.3) that were more recent than the codes for bipolar or manic disorder. Owing to the possibility of misdiagnoses between bipolar disorder and these other conditions, the most recent diagnosis was considered the most accurate.12,29

The individuals selected were divided into 3 clinical groups (a grouping based on up to the 10 most recent bipolar disorder claims): (1) those with predominantly manic or mixed symptoms (ICD-9-CM codes 296.0, 296.1, 296.4, and 296.6); (2) those with predominantly depressive symptoms (ICD-9-CM code 296.5); and (3) those with unspecified bipolar disorder (ICD-9-CM codes 296.7 and 296.8). Individuals were classified as having unspecified bipolar disorder only if all of their claims included these codes, and data pertaining to this group are not discussed in this article. For those individuals who had a combination of manic or mixed symptoms and depressive symptoms, classification was based on the largest proportion of episode subtype, defaulting to manic or mixed if proportions were even. Individuals with bipolar disorder classified as such were retained for study only if they had at least 1 treatment episode of antipsychotic monotherapy with quetiapine or risperidone. Treatment episodes with quetiapine or risperidone that involved concurrent use of any other antipsychotic were excluded from the analysis. Treatment episodes were also excluded if they were <150 days (5 months) in duration (see explanation below).

Therapeutic categories

Eight therapeutic categories were defined for the purposes of this study: (1) (atypical) antipsychotic alone; (2) antipsychotic with lithium; (3) antipsychotic with an anticonvulsant; (4) antipsychotic with an antidepressant; (5) antipsychotic with lithium and an antidepressant; (6) antipsychotic with an anticonvulsant and an antidepressant; (7) antipsychotic with lithium and an anticonvulsant; and (8) antipsychotic with lithium, an anticonvulsant, and an antidepressant. Treatment with quetiapine or risperidone, alone and in various lithium/anticonvulsant/antidepressant combinations, was assessed over the 150-day period following treatment initiation. (Specific anticonvulsants and antidepressants included in combination categories appear in the appendix.)

Definition of treatment episode

A treatment-episode approach was taken to assess associations between antipsychotic dose and treatment adherence. A treatment episode consisted of at least 2 sequential monotherapy prescriptions for quetiapine or risperidone. A significant proportion of individuals in the study had >1 antipsychotic treatment episode because they were treated on separate occasions with either the same or a different agent.

Treatment episodes were measured from the date of the first prescription to the final date of treatment, which was calculated from the date of the last prescription plus the number of days for which it was supplied (unless preceded by patient disenrollment from the health plan or the end of the data).30 A prescription was assumed to be the first in a treatment episode if no other prescriptions for quetiapine or risperidone were filled within the preceding 120 days. Similarly, a prescription was assumed to be the last in a treatment episode if no other prescriptions for quetiapine or risperidone were filled within the following 120 days. Gaps of ≤120 days between prescription fill dates were permitted within treatment episodes.

Assessment of treatment adherence

Two measures of treatment adherence were investigated: adherence intensity and treatment duration.

Adherence intensity, a measure of the shortfall in prescription refills during a treatment episode or segment within a treatment episode with quetiapine or risperidone, was assessed over the 150-day period following the initiation of the antipsychotic. Expert Consensus Guidelines recommend that nonmaintenance treatment with atypical antipsychotics extend for 20 weeks after response.31 Although antipsychotic treatment episodes for bipolar disorder can extend well beyond the 150-day window, we considered it useful to enforce a uniform time frame for assessing adherence intensity. Adherence intensity can vary, depending on whether the patient is in an acute or maintenance phase. Asymptomatic patients may consider it less necessary to take their medications.32

Adherence intensity was quantified using the MPR.23,33 The MPR was calculated by dividing the total number of days for which the antipsychotic was supplied (number of days’ supply of medication that an individual actually received, as indicated on prescription claims) by the total number of days encompassed by the treatment episode (number of days’ supply that an individual should have received had he or she obtained the medication as prescribed). This methodology has been described previously.30 The higher the MPR, the greater (or higher) the adherence intensity. An MPR of 1 indicates that the individual has received all the medication required to take as prescribed, whereas an MPR of 0.5 indicates that the individual has received enough medication to take only half of the prescribed amount. MPRs >1 were possible in cases in which prescriptions may not have been completely used due to interim changes in dosage. For individuals who were hospitalized for any reason during a treatment episode, the total days encompassed by a treatment episode were reduced by the number of inpatient days before calculation of MPR in order to avoid understating the MPR (hospitalized individuals were assumed to have received all required medication from the hospital pharmacy).

Since adherence intensity was measured over the 150-day period following the start of antipsychotic treatment, the calculation of MPRs over this period required some modification. The number of days supplied associated with the last prescription within the 150-day period was compared with the number of remaining days in the period, and any excess was subtracted. For example, if the last prescription was filled on the 133rd day of the period and the prescription supplied the medication for 30 days, an excess of 13 days would be subtracted.

Treatment duration, the second measure of treatment adherence, was simply the length of the treatment episode beyond the 150-day period. Analysis of treatment duration was adjusted for treatment episodes that were censored (either by patient disenrollment from a health plan or by the end date of the data). Since the probability of being censored increased with treatment episodes starting nearer to the end point of data collection (May 2005), adjustments were made using information on the start date as well as censoring.

Statistical analyses

Linear regression models were estimated to assess the relationship between the various treatment options and treatment adherence (MPR during or treatment duration >150 days). In the main analysis, binary variables were specified to indicate therapies combining the antipsychotic with 1 or more of these agents: lithium, an anticonvulsant, and/or an antidepressant. The excluded therapeutic category in each instance was quetiapine or risperidone monotherapy. Separate sets of models for MPR and treatment duration were estimated for bipolar manic/mixed and bipolar depressed subgroups.

Models controlled for the following: days supplied of lithium, an anticonvulsant, an antidepressant, and other nonantipsychotic medications; use of gabapentin or oxcarbazepine (no longer recommended in guidelines because of multiple negative randomized controlled trials); age22,34-38; gender35,38-40; predominantly mixed vs manic episodes (manic/mixed models only)41,42; prior mental health expenditure (marker for illness severity)34,37,38,43,44; prior treatment with an antipsychotic,37 traditional mood stabilizers, antidepressants, or other psychotropics; comorbid schizophrenia, other psychoses, or nonpsychotic mental disorders,38,39,44,45 and substance dependence/abuse (markers for illness severity)34,36,38,46,47; preexisting antipsychotic-related side effects47-52; changes in antipsychotic pill strength (MPR models only)33; censoring and start date of treatment episode (treatment duration models only); type of health coverage53; and geographic location. Select variables are quantified in TABLES 1A and 1B.

In a secondary analysis, the antipsychotic combined with lithium (with or without an antidepressant) was compared with an antipsychotic combined with an anticonvulsant (with or without an antidepressant). MPR and treatment duration regressions were reestimated based on these observations only.

All regression analyses were tested for heteroscedasticity using White’s General Test. Where significant heteroscedasticity was found, models were reestimated with weighted least squares (WLS). Data were also tested for nonnormality. Because all patient samples were sufficiently large and deviations from normality were moderate, no adjustments were made for nonnormal data distributions unless the deviation from normality was extreme. The traditional P < .05 threshold was used to establish statistical significance. However, results that approached statistical significance at P < .10 were also noted. Results not meeting these thresholds are not reported in this article.

CLICK HERE for PDF Version of Table 1A

CLICK HERE for PDF Version of Table 1B


A total of 1767 treatment episodes with quetiapine and 2051 with risperidone met study inclusion criteria among 3626 patients with predominantly manic or mixed symptoms. Among patients with predominantly depressive symptoms, 899 episodes with quetiapine and 814 with risperidone met inclusion criteria among 1618 patients. Approximately 5% of patients in the manic/mixed group and 6% in the depressed group had multiple episodes, with the same or a different antipsychotic, occurring at different times. Antipsychotic treatment episodes falling to each of the 2 bipolar subgroups were subdivided into the 8 bipolar therapies identified previously (TABLES 1A and 1B).

Characteristics of the study population

Select treatment and patient characteristics for the 8 categories of combination therapy with quetiapine or risperidone are reported in TABLES 1A and 1B, respectively. Of the total sample, the highest proportion of treatment episodes occurred with use of the antipsychotic in combination with anticonvulsants and antidepressants, and the smallest proportion occurred with use of the antipsychotic combined with lithium and anticonvulsants. Antipsychotic monotherapy treatment episodes accounted for 7% to 10% and 5% to 6% of the total sample for predominantly manic/mixed and depressed groups, respectively. In addition to lithium, anticonvulsants, and antidepressants, other psychotropic medications (such as anxiolytics and hypnotics) were also given in combination with the antipsychotic. Individuals treated with antipsychotic monotherapy were generally younger than those treated with combination therapy. Antipsychotic monotherapy candidates also tended to have the lowest baseline severity as gauged by mental health care expenditure during the 90 days prior to the start of antipsychotic treatment. Comorbid schizophrenia, other psychoses, and nonpsychotic mental disorders showed no clear association with the type of combination therapy selected. Similarly, preexisting conditions possibly associated with antipsychotic treatment (eg, diabetes, extrapyramidal symptoms, weight gain, and sexual dysfunction) showed little connection to the type of combination therapy used.

TABLES 2A AND 2B report antipsychotic treatment intensity (in terms of MPR) and treatment duration for each of the 8 categories of combination therapy with quetiapine or risperidone. Unadjusted MPRs and treatment durations (for both censored and uncensored treatment episodes) showed little association with the type of combination therapy used. The need for adjustment is reflected in the clear association between MPR and number of pill-strength changes. Multiple regressions adjusted for this and other factors. Regression estimates of the effects of various combination therapies on antipsychotic adherence intensity (gauged by the MPR) and treatment duration are presented in TABLES 3 AND 4.

CLICK HERE for PDF Version of Table 2A

CLICK HERE for PDF Version of Table 2B

Adherence intensity

Individuals with predominantly manic/mixed symptoms. For individuals with predominantly manic/mixed symptoms, any combination of lithium, an anticonvulsant, and an antidepressant was associated with a reduction in antipsychotic MPR (TABLE 3) that tended to worsen with the number of medications included within the combination strategy. The largest reductions occurred when quetiapine or risperidone was used concurrently with all 3 of the other psychotropic medication types (–0.329; P < .001 for quetiapine; and –0.370; P < .001 for risperidone). Generally, combinations of 2 or more of these medications with the antipsychotic were associated with lower antipsychotic MPRs than when only 1 of the medications was combined. For quetiapine, the smallest reduction in MPR occurred when it was used with an antidepressant alone (–0.087; P=.018); for risperidone, the smallest reduction occurred when it was used with an anticonvulsant alone (–0.102; P=.043). For both risperidone and quetiapine, combinations with anticonvulsants were associated with lower antipsychotic MPR than combinations with lithium (–0.190; P < .001 for quetiapine; and –0.168; P=.002 for risperidone).

Among other significant (P < .05) factors, individuals who consistently filled prescriptions for lithium, anticonvulsants, and antidepressants (as reflected in days’ supply for these medications) were also more consistent in filling their antipsychotic prescriptions, as demonstrated by a significant (P < .05) positive association with MPR. The number of changes in pill strength during the course of treatment also showed a positive association with antipsychotic MPR, while in contrast, older age was associated with reduced MPR (P < .05). For risperidone only, prior use of antidepressants was also associated with reduced MPR.

Individuals with predominantly depressive symptoms. For quetiapine, individuals with predominantly depressive symptoms displayed the same negative association between combination therapy and antipsychotic MPR as individuals with manic/mixed symptoms (TABLE 3). Again, treatment approaches combining any of the other psychotropics with quetiapine were associated with lower MPRs than quetiapine monotherapy. For risperidone, only strategies combining 2 or more other psychotropic medications were associated with significantly lower antipsychotic MPRs than risperidone monotherapy (TABLE 3). The largest reductions in MPR occurred when the antipsychotics were used concurrently with lithium and anticonvulsant therapies (–0.349; P < .001 for quetiapine; and –0.197; P=.023 for risperidone). The smallest reduction for quetiapine occurred when it was used with an antidepressant alone (–0.116; P=.039). Antipsychotic-anticonvulsant combinations were associated with significantly lower MPRs than antipsychotic-lithium combinations (–0.101; P=.010, and –0.189; P < .001 for quetiapine and risperidone, respectively) in individuals with predominantly depressive symptoms.

Among other significant factors (P < .05), as for the manic/mixed subgroup, days’ supply of lithium, anticonvulsants, and antidepressants, and the number of changes in pill strength during the course of treatment were positively associated with antipsychotic MPR (P < .05). Older age, the use of oxcarbazepine, and the prior use of traditional mood stabilizers negatively impacted antipsychotic MPR (P < .05). Illness severity, as reflected in the prior level of mental health care expenditure (both quetiapine and risperidone) and comorbid schizophrenia (risperidone only), was associated with higher antipsychotic MPR. Comorbid nonpsychotic mental disorders (quetiapine only) and substance abuse (risperidone only) were associated with lower MPR.

CLICK HERE for PDF Version of Table 3

Treatment duration

Individuals with predominantly manic/mixed symptoms. Combining quetiapine with lithium or with lithium and an anticonvulsant was associated with shorter treatment durations compared with quetiapine monotherapy (–3.4; P=.007 for quetiapine with lithium; and –2.7; P=.032 for quetiapine with lithium and an anticonvulsant; TABLE 4). For risperidone, treatment duration was not significantly affected by the type of combination therapy, but the reduced duration when combined with an antidepressant and lithium approached significance (–2.3; P=.083). Quetiapine, when combined with an anticonvulsant, was associated with longer treatment duration than when combined with lithium (1.4; P=.056), although this observation did not reach significance. The difference in treatment duration between risperidone with an anticonvulsant and risperidone with lithium was also nonsignificant.

Antipsychotic treatment duration was significantly (P < .05) affected by other factors. A higher level of lithium use (as measured by days’ supply) was associated with a longer duration of quetiapine treatment (P < .05), whereas for risperidone, treatment duration was positively associated with antidepressant use but negatively affected by the use of other psychotropics (P < .05). Males receiving treatment with risperidone had longer treatment durations than females (P < .05), whereas gender had no effect on the duration of quetiapine treatment. Type of health coverage and geographic location had varying effects across antipsychotics. For quetiapine and risperidone, uncensored treatment episodes had shorter durations, reflecting the greater likelihood that episodes of longer duration were censored by the end point of the database (May 2005).

Individuals with predominantly depressive symptoms. None of the combination therapies showed significant differences in treatment duration from that of antipsychotic monotherapy (TABLE 4). Antipsychotic combinations with lithium vs combinations with anticonvulsants also showed no significant differences in treatment duration. Aside from variables that controlled for treatment episode censoring and type of health coverage, no other factors significantly affected antipsychotic treatment duration.

CLICK HERE for PDF Version of Table 4


In recent years, atypical antipsychotics have become a mainstay in the treatment of bipolar disorder.10 There has also been an increasing trend toward polypharmacy, often a combination of antipsychotics with lithium, anticonvulsants, and/or antidepressants.1,2 Very few clinical studies have assessed the outcomes of antipsychotic combination therapies vs antipsychotic monotherapy in the treatment of bipolar mania and depression, and those that have done so have focused on efficacy.1,14,15 Since polypharmacy may increase the financial cost, the time/management burden, and the risk of adverse events, patient adherence to medication may be reduced.15,20 In addition to increased side effects, out-of-pocket costs associated with polypharmacy may also contribute to nonadherence.54 To the authors’ knowledge, this study is the first to examine the association between combination treatment with antipsychotics and antipsychotic adherence among individuals with bipolar disorder stratified according to predominant symptomatology.

Among individuals with predominantly manic or mixed symptoms, the most common treatment approach involving quetiapine or risperidone combined the antipsychotic with an anticonvulsant and an antidepressant. Pairing the antipsychotic with lithium alone was the least favored treatment approach. Individuals with depressive symptoms who were prescribed quetiapine or risperidone were most commonly treated with the antipsychotic in combination with an antidepressant. Treatment approaches that combined the antipsychotic with lithium or with lithium and anticonvulsants were least used.

Various antipsychotic combination therapies were compared with antipsychotic monotherapy with respect to adherence intensity (measured by the MPR) and treatment duration. Unadjusted antipsychotic MPRs and treatment durations showed no clear patterns across therapies. However, regression estimates, which adjusted for a range of other factors potentially affecting antipsychotic adherence, did reveal differences.

In manic or mixed individuals, all combination therapies were associated with lower antipsychotic MPRs than when the antipsychotic was used alone, particularly therapies combining 2 or more of the other psychotropics. The same trend was exhibited in depressed individuals treated with quetiapine, but for those receiving risperidone, only therapies combining 2 or more of the other psychotropics were associated with lower MPRs. For both bipolar subgroups and both antipsychotics, combinations with anticonvulsants were associated with lower antipsychotic MPRs than combinations with lithium.

For the manic/mixed subgroup, combining quetiapine with lithium or with lithium and an anticonvulsant was associated with shorter antipsychotic treatment durations compared with quetiapine alone. Combining quetiapine with lithium was also associated with shorter treatment durations than when combining quetiapine with an anticonvulsant. No differences were found for risperidone. In the depressed subgroup, antipsychotic treatment durations did not differ between monotherapy and combination therapy for either of the antipsychotics. However, quetiapine-anticonvulsant combinations were associated with shorter antipsychotic treatment durations than quetiapine-lithium combinations.

The relationship between polypharmacy and treatment adherence among individuals with bipolar disorder has been investigated previously.23,55,56 In these studies polypharmacy was associated with increased MPR, although these data were unadjusted and confined to individuals receiving 2 antipsychotics24 or 2 medications for mood stabilization.56 In contrast to the Veterans study, in which MPRs were generally in the order of 0.75 to 0.80,55 participants in the current study tended to be more adherent to antipsychotic therapy as evidenced by higher MPRs, ranging between 0.86 and 1.1.

Limitations of the study

Assessment of antipsychotic adherence intensity was limited to the 150-day period following treatment initiation, which corresponds to Expert Consensus Guidelines recommended time frame for nonmaintenance treatment with atypical antipsychotics.31 Effects of combination therapy beyond this period were not evaluated. Also, effects of combination therapy on treatment duration were assessed only for durations extending beyond the 150-day period, and effects on shorter durations were not investigated. These limitations may reduce the generalizability of the findings.

Because there is a clear strategy of adding drugs to a treatment regimen if monotherapy proves ineffective, this study’s findings may have been influenced by selection bias to the extent that it was not fully adjusted with the variables specified in the models, particularly those intended to capture illness severity and treatment resistance. Unmeasured patient characteristics may have been associated with both treatment adherence and the choice of therapy. Consequently, effects on antipsychotic treatment adherence attributed to combination therapies may in part have been caused by the unmeasured patient factors. For example, patients who received antipsychotic monotherapy had lower baseline severity than did patients treated with combination therapies, as reflected by markers such as prior mental health care expenditure and the proportion of individuals with substance dependence/abuse. These markers, however, are imperfect measures and do not fully adjust for the association between type of treatment and severity, and this failure may have significantly affected the results. Greater adherence intensity may be more easily attained among less severely ill patients. This limitation affects most claims-based studies because clinical details are lacking in such data.

Another limitation of this study was the use of prescription refills to gauge adherence intensity. Low adherence intensity associated with prescriptions that are filled but not used is not measured. This could create bias if the degree to which prescriptions are unused differs among the therapies included in this study. Unused prescriptions cannot be identified in claims data. Also, unlike other studies,57 in calculating MPRs a maximum ratio of 1 was not enforced, as this would bias comparisons against agents with a relatively high adherence intensity. Interpretation of MPRs >1 also requires caution in that they may reflect overprescribing34 as well as unused medication due to changes in pill strength adjusted for in the models.33 Overprescribing, however, cannot be determined solely from claims data.37


Treatment adherence to the atypical antipsychotics quetiapine and risperidone, prescribed alone and as part of combination therapy and as measured by adherence intensity and persistence of treatment, is dependent on the nature of the medication combination strategy and the predominant symptomatology of the individual receiving treatment. It may also depend on the specific antipsychotic medication. Although these findings may affect medication selection when treating bipolar illness, adherence is only 1 contributing element, albeit a very important 1, to a therapy’s effectiveness. Therapies combining multiple medications may have benefits that outweigh any negative impact on adherence.

ACKNOWLEDGEMENTS: This study was supported by AstraZeneca Pharmaceuticals LP. The authors would like to acknowledge the editorial assistance of Eleanor Bull, PhD (PAREXEL MMS), with manuscript review and revisions. Financial support for this assistance was provided by AstraZeneca Pharmaceuticals LP.

DISCLOSURES: When the article was written in 2007, Dr. Gianfrancesco’s company, HECON Associates, Inc., was under contract with AstraZeneca Pharmaceuticals. Dr. Tafesse is an employee of AstraZeneca Pharmaceuticals. Dr. Sajatovic receives grant/research support from AstraZeneca Pharmaceuticals and GlaxoSmithKline and is a consultant to GlaxoSmithKline. Mr. Wang reports no financial relationship with any company whose products are mentioned in this article or with manufacturers of competing products.

Appendix (Specific anticonvulsants and antidepressants included in combination categories)

Traditional mood stabilizers

Divalproex sodium (Depakote)
Valproic acid (Depakene)
Valproate sodium (Depacon)
Carbamazepine (Tegretol, Equetro, Carbatrol, Atretol, Convuline, Epitol, Macrepan)
Oxcarbazepine (Trileptal)
Lamotrigine (Lamictal)
Gabapentin (Neurontin)

Antidepressants Multiple reuptake inhibitors
Bupropion (Wellbutrin)
Venlafaxine (Effexor)

Selective serotonin reuptake inhibitors
Fluoxetine (Prozac)
Paroxetine (Paxil)
Citalopram oxalate (Celexa, Cipramil)
Escitalopram oxalate (Lexapro, Cipralex)
Sertraline (Zoloft, Lustral)
Fluvoxamine maleate (Luvox, Floxyfral, Faverin)
Dapoxetine (no known trade names)

Tricyclic antidepressants
Imipramine (Tofranil)
Amitriptyline (Elavil; Larozyl; Vanatrip)
Doxepin HCl (Sinequin)
Desipramine HCl (Norpramin)
Trimipramine (Surmontil)
Protriptyline HCl (Vivactil)

Monoamine oxidase inhibitors
Isocarboxazid (Marplan)
Tranylcypromine (Parnate)
Moclobemide (Manerix)


  1. Zarate CA, Quiroz JA. Combination treatment in bipolar disorder: a review of controlled trials. Bipolar Disord. 2003;5:217–225.
  2. Keck PE. Long-term management strategies to achieve optimal function in patients with bipolar disorder. J Clin Psychiatry. 2006;67(suppl 9):19–24.
  3. Bowden CL. Atypical antipsychotic augmentation of mood stabilizer therapy in bipolar disorder. J Clin Psychiatry. 2005;66(suppl 3):12–19.
  4. Grof P. Selecting effective long-term treatment for bipolar individuals: monotherapy and combinations. J Clin Psychiatry. 2003;64(suppl 5):53–61.
  5. Nemeroff CB. Use of atypical antipsychotics in refractory depression and anxiety. J Clin Psychiatry. 2005;66(suppl 8):13–21.
  6. Frye MA, Ketter TA, Leverich GS, et al. The increasing use of polypharmacotherapy for refractory mood disorders: 22 years of study. J Clin Psychiatry. 2000;61:9–15.
  7. Wolfsperger M, Greil W, Rössler W, et al. Pharmacological treatment of acute mania in psychiatric in-patients between 1994 and 2004. J Affect Disord. 2007;99:9–17.
  8. Keck PE Jr. The role of second-generation antipsychotic monotherapy in the rapid control of acute bipolar mania. J Clin Psychiatry. 2005;66(suppl 3):5–11.
  9. Suppes T, Dennehy EB, Hirschfeld RM, et al; Texas Consensus Conference Panel on Medication Treatment of Bipolar Disorder, The Texas implementation of medication algorithms: update to the algorithms for treatment of bipolar I disorder. J Clin Psychiatry. 2005;66:870–886.
  10. Hirschfeld RMA, Bowden L, Gitlin MJ, et al, for the Work Group on Bipolar Disorder. Practice Guideline for the Treatment of Patients With Bipolar Disorder, 2nd ed. Arlington, VA: American Psychiatric Association; 2002:1-82.
  11. Goodwin GM. Consensus Group of the British Association for Psychopharmacology, Evidence-based guidelines for treating bipolar disorder: recommendations from the British Association for Psychopharmacology. J Psychopharmacol. 2003;17:149–173.
  12. Yatham LN, Kennedy SH, O’Donovan C, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) guidelines for the management of patients with bipolar disorder: consensus and controversies. Bipolar Disord. 2005;7(suppl 3):5–69.
  13. Lin D, Mok H, Yatham LN. Polytherapy in bipolar disorder. CNS Drugs. 2006;20:29–42.
  14. Bowden CL, Ketter TA, Sachs GS, et al. Focus on bipolar disorder treatment. J Clin Psychiatry. 2005;66:1598–1609.
  15. Ketter TA, Nasrallah HA, Fagiolini A. Mood stabilizers and atypical antipsychotics: bimodal treatments for bipolar disorder. Psychopharmacol Bull. 2006;39:120–146.
  16. Kingsbury SJ, Yi D, Simpson GM. Psychopharmacology: rational and irrational polypharmacy. Psychiatr Serv. 2001;52:1033–1036.
  17. Fincke BG, Miller DR, Spiro A. The interaction of patient perception of overmedication with drug compliance and side effects. J Gen Intern Med. 1998;13:182–185.
  18. Burton SC. Strategies for improving adherence to second-generation antipsychotics in patients with schizophrenia by increasing ease of use. J Psychiatr Pract. 2005;1:369–378.
  19. West JC, Wilk JE, Olfson M, et al. Patterns and quality of treatment for patients with schizophrenia in routine psychiatric practice. Psychiatr Serv. 2005;56:283–291.
  20. Bowden CL. Making optimal use of combination pharmacotherapy in bipolar disorder. J Clin Psychiatry. 2004;65(suppl 15):21–24.
  21. Svarstad BL, Shireman TI, Sweeney JK. Using drug claims data to assess the relationship of medication adherence with hospitalization and costs. Psychiatr Serv. 2001;52:805–811.
  22. Dolder CR, Lacro JP, Dunn LB, et al. Antipsychotic medication adherence: is there a difference between typical and atypical agents? Am J Psychiatry. 2002;159:103–108.
  23. Sajatovic M, Valenstein M, Blow FC, et al. Treatment adherence with antipsychotic medications in bipolar disorder. Bipolar Disord. 2006;8:232–241.
  24.  American Diabetes Association, the American Psychiatric Association, the American Association of Clinical Endocrinologists, and the North American Association for the Study of Obesity. Consensus development conference on antipsychotic drugs and obesity and diabetes. Diabetes Care. 2004;27:596–601.
  25. Perlick DA, Rosenheck RA, Kaczynski R, et al. Medication non-adherence in bipolar disorder: a patient-centered review of research findings. Clinical Approaches in Bipolar Disorder. 2004;3:56–64.
  26. Singh V, Bowden CL. Concepts surrounding the diagnosis and treatment of mixed states in bipolar disorders. Clinical Approaches in Bipolar Disorder. 2005;4:35–43.
  27. Paykel ES. The evolution of life events research in psychiatry. J Affect Disord. 2001;62:141–149.
  28.  Centers for Disease Control and Prevention, Centers for Medicare & Medicaid Services. International Statistical Classification of Diseases, 9th revision, clinical modification (ICD-9-CM).; 2008.
  29. Das AK, Olfson M, Gameroff MJ, et al. Screening for bipolar disorder in a primary care practice. JAMA. 2005;293:956–963.
  30. Gianfrancesco FD, Rajagopalan K, Sajatovic M, et al. Treatment adherence among individuals with bipolar or manic disorder taking atypical and typical antipsychotics. J Clin Psychiatry. 2006;67:222–232.
  31. Keck PE. The Expert Consensus Guideline Series: Treatment of Bipolar Disorder. Postgrad Med. 2004;1–20.
  32. Colom F, Vieta E, Martínez-Arán A, et al. Clinical factors associated with treatment noncompliance in euthymic bipolar individuals. J Clin Psychiatry. 2000;61:549–555.
  33. Valenstein M, Copeland LA, Blow FC, et al. Pharmacy data identify poorly adherent individuals with schizophrenia at increased risk for admission. Med Care. 2002;40:630–639.
  34. Gilmer TP, Dolder CR, Lacro JP, et al. Adherence to treatment with antipsychotic medication and health care costs among Medicaid beneficiaries with schizophrenia. Am J Psychiatry. 2004;161:692–699.
  35. Linden M, Godemann F, Gaebel W, et al. A prospective study of factors influencing adherence to a continuous neuroleptic treatment program in schizophrenia individuals during 2 years. Schizophr Bull. 2001;27:585–596.
  36. Nose M, Barbui C, Tansella M. How often do individuals with psychosis fail to adhere to treatment programmes? A systematic review. Psychol Med. 2003;33:1149–1160.
  37. Rittmannsberger H, Pachinger T, Keppelmuller P, et al. Medication adherence among psychotic individuals before admission to inpatient treatment. Psychiatr Serv. 2004;55:174–179.
  38. Lacro JP, Dunn LB, Dolder CR, et al. Prevalence of and risk factors for medication nonadherence in individuals with schizophrenia: a comprehensive review of recent literature. J Clin Psychiatry. 2002;63:892–909.
  39. Leslie DL, Rosenheck RA. Use of pharmacy data to assess quality of pharmacotherapy for schizophrenia in a national health care system: individual and facility predictors. Med Care. 2001;39:923–933.
  40. Diaz E, Neuse E, Sullivan MC, et al. Adherence to typical and atypical antipsychotics after hospital discharge. J Clin Psychiatry. 2004;65:354–360.
  41. Bowden CL, Singh V. Valproate in bipolar disorder: 2000 onwards. Acta Psychiatr Scand. 2005;111:13–20.
  42. Dunner DL. Atypical antipsychotics: efficacy across bipolar disorder subpopulations. J Clin Psychiatry. 2005;66(suppl 3):20–27.
  43. Keck PE Jr, McElroy SL, Strakowski SM, et al. Factors associated with pharmacologic noncompliance in individuals with mania. J Clin Psychiatry. 1996;57:292–297.
  44. Corriss DJ, Smith TE, Hull JW, et al. Interactive risk factors for treatment adherence in a chronic psychotic disorders population. Psychiatry Res. 1999;89:269–274.
  45. Russo P, Smith MW, Dirani R, et al. Pharmacotherapy patterns in the treatment of bipolar disorder. Bipolar Disord. 2002;4:366–377.
  46. Owen RR, Fischer EP, Booth BM, et al. Medication noncompliance and substance abuse among individuals with schizophrenia. Psychiatr Serv. 1996;47:853–858.
  47. Olfson M, Mechanic D, Hansell S, et al. Predicting medication noncompliance after hospital discharge among individuals with schizophrenia. Psychiatr Serv. 2000;51:216–222.
  48. Lambert M, Conus P, Eide P, et al. Impact of present and past antipsychotic side effects on attitude toward typical antipsychotic treatment and adherence. Eur Psychiatry. 2004;19:415–422.
  49. Robinson DG, Woerner MG, Alvir JM, et al. Predictors of medication discontinuation by patients with first-episode schizophrenia and schizoaffective disorder. Schizophr Res. 2002;57:209–219.
  50. Buchanan A. A two-year prospective study of treatment compliance in individuals with schizophrenia. Psychol Med. 1992;22:787–797.
  51. Weiden PJ, Kozma C, Grogg A, et al. Partial compliance and risk of rehospitalization among California Medicaid individuals with schizophrenia. Psychiatr Serv. 2004;55:886–891.
  52. Zarate CA Jr. Antipsychotic drug side effect issues in bipolar manic individuals. J Clin Psychiatry. 2000;61(suppl 8):52–61.
  53. Gibson TB, Ozminkowski RJ, Goetzel RZ. The effects of prescription drug cost sharing: a review of the evidence. Am J Manag Care. 2005;11:730–740.
  54. Briesacher BA, Gurwitz JH, Soumerai SB. Patients at risk for cost-related medication nonadherence: a review of the literature. J Gen Int Med. 2007;22:864–871.
  55. Sajatovic M, Bauer MS, Kilbourne AM, et al. Self-reported medication treatment adherence among veterans with bipolar disorder. Psychiatr Serv. 2006;57:56–62.
  56. Sajatovic M, Valenstein M, Blow F, et al. Treatment adherence with lithium and anticonvulsant medications among patients with bipolar disorder. Psychiatr Serv. 2007;58:855–863.
  57. Al-Zakwani IS, Barron JJ, Bullano MF, et al. Analysis of healthcare utilization patterns and adherence in individuals receiving typical and atypical antipsychotic medications. Curr Med Res Opin. 2003;19:619–626.

CORRESPONDENCE: Frank Gianfrancesco, PhD, HECON Associates, Inc., 9833 Whetstone Drive, Montgomery Village, MD 20886 USA. E-MAIL: frank_gianfrancesco@heconassoc.com, heconassoc@comcast.net