Prevalence, trends, and factors associated with antipsychotic polypharmacy among medicaid-eligible schizophrenia patients, 1998–2000

Antipsychotic Polypharmacy Among Medicaid Patients Prevalence, Trends, and Factors
Associated With Antipsychotic Polypharmacy Among
Medicaid-Eligible Schizophrenia Patients, 1998–2000
Rahul Ganguly, Ph.D.; Jeffrey A. Kotzan, Ph.D.; L. Stephen Miller, Ph.D.;
Klugh Kennedy, Pharm.D.; and Bradley C. Martin, Ph.D.
Received Aug. 15, 2003; accepted Feb. 17, 2004. From the Department of Pharmacy Administration, School of Pharmacy,University of Mississippi, Oxford (Dr. Ganguly); the Department Objective: To determine the prevalence, trends,
of Clinical and Administrative Pharmacy, College of Pharmacy (Drs. and factors associated with antipsychotic polyphar- Kotzan and Kennedy) and the Department of Psychology (Dr. Miller), macy categorized according to type of antipsy- University of Georgia, Athens; and the College of Pharmacy, chotic and duration of use and to contrast usage University of Arkansas for Medical Sciences, Little Rock (Dr. Martin). At the time this study was completed, Dr. Ganguly was enrolled as patterns with published treatment guidelines.
a Ph.D. candidate in the Pharmacy Care Administration Graduate Method: A retrospective cohort study was de-
Program at the University of Georgia, and Dr. Martin was affiliated signed, and Medicaid recipients ≥ 16 years of age with the Department of Clinical and Administrative Pharmacy, with a schizophrenia diagnosis (ICD-9-CM = College of Pharmacy, University of Georgia. 295.xx) between 1998 and 2000 were identified Supported, in part, by grant award RO3HS10815-01 from the from the California (20% random sample) and Agency for Healthcare Research and Quality (AHRQ), Rockville, Md. Presented, in part, at the 8th annual international meeting of the Georgia Medicaid claims databases. Use of anti- International Society for Pharmacoeconomic Research, Arlington, Va., psychotic polypharmacy was categorized based on duration (long-term polypharmacy was defined as Corresponding author and reprints: Bradley C. Martin, Ph.D., lasting > 2 months), and long-term use was further College of Pharmacy, University of Arkansas for Medical Sciences, categorized based on type of antipsychotic combi- 4301 W. Markham St., Slot 522, Little Rock, AR 72205-7122(e-mail: bmartin@uams.edu). nations (clozapine, conventional, and atypical).
The prevalence, mean duration, and frequency ofand yearwise trends in antipsychotic polypharmacywere estimated. A stepwise logistic variable selec- t is estimated that antipsychotic polypharmacy, de- tion procedure was used to identify factors associ- Ifined as concomitant use of multiple antipsychotics,
ated with long-term antipsychotic polypharmacy.
is prescribed for up to 40% of schizophrenia patients.1 Results: Of a total of 31,435 persons with
schizophrenia, the 1998–2000 prevalence of anti- However, there are no randomized controlled trials of psychotic polypharmacy was 40% (N = 12,549; combination therapy except for 1 with sulpiride and clo- mean age = 43 years; white, 47%; female, 48%; zapine, which provides little guidance in the United States mean duration of polypharmacy = 149 days), and since sulpiride is not available in the United States.2 Apart long-term antipsychotic polypharmacy prevalence from that study, there are case reports3–11 and open, uncon- was 23% (N = 7222, mean duration = 236 days).
The prevalence of atypical antipsychotic poly- trolled, nonrandomized trials12–17 that report the effects of antipsychotic polypharmacy. Many of these studies re- (p < .0001). Use of newer atypicals such as quetia- port improvements in symptom control3–5,7–9,11,12,15 and in- pine (OR = 18.32) and older conventionals such as stances of nonserious side effects such as drooling11 and chlorpromazine (OR = 28.87) was strongly associ- compulsive behavior.12 Other studies report an increased ated with long-term antipsychotic polypharmacy.
Conclusion: Antipsychotic polypharmacy
incidence of serious adverse events such as prolactin is widely prevalent, is prescribed for long dura- elevation, akathisia, and hypersalivation6,10,13,17 and even tions, and is an increasing phenomenon among an increased risk of mortality (study N = 88, relative Medicaid-eligible schizophrenia patients, indicat- risk = 2.46).16 However, it should be noted that besides ing a significant discrepancy with treatment guide- the obvious design limitations of such uncontrolled trials, lines (which do not advocate the use of any poly-pharmacy except for short-term periods when these studies were limited by small sample sizes (most of transitioning patients to new antipsychotics).
them are 1- or 2-patient case reports), short follow-up pe- Further research evaluating the effects of antipsy- riods, and incomplete reporting of adverse effects. The re- chotic polypharmacy in schizophrenia patients cent introduction of 4 antipsychotics (olanzapine in 1996, may assist in defining the scope and potential quetiapine in 1997, ziprasidone in 2001, and aripiprazole (J Clin Psychiatry 2004;65:1377–1388) in 2003) with differing receptor profiles has further in-creased the possibilities of combining these agents. Theobjective of our study was to estimate the prevalence and COPYRIGHT 2004 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Table 1. Antipsychotics Included in an Analysis
Figure 1. Hypothetical Antipsychotic Polypharmacy Episode
of Antipsychotic Polypharmacy
Involving Risperidone and Haloperidol
ThioridazineThiothixeneTrifluoperazineChlorprothixene trends of antipsychotic polypharmacy, categorize antipsy- chotic polypharmacy according to type of antipsychoticand duration of use, and contrast antipsychotic polyphar-macy usage patterns with published treatment guidelines.
We also estimated the factors associated with antipsy- Medicaid reimburses 4 state psychiatric hospitals for in- patient services rendered to Medicaid-eligible patients, sothere was no need to link state psychiatric hospitals with The Georgia18–21 and California22–24 Medicaid data have been used in the past for epidemiologic studies and have A retrospective observational study design was em- been found to be valid. The accuracy of schizophrenia di- ployed using a combined 2-state Medicaid claims data- agnoses in Medicaid claims data has also been validated base. The study was approved by the University of Geor- gia Institutional Review Board. Claims data from January1998 through December 2000 for Medicaid recipients from the states of Georgia and California were combined Persons with schizophrenia were identified using the to build this 2-state database. Three sources of data were following inclusion criteria: primary diagnosis of schizo- used: the Georgia Medicaid files maintained by the Geor- phrenia (ICD-9-CM = 295.xx) recorded on at least 1 paid gia Department of Medical Assistance (GDMA), Georgia claim during the period January 1998 through December state-based institutional data files maintained by the De- 2000 and age of at least 16 years as of Jan. 1, 1998.
partment of Human Resources (DHR), and a CaliforniaMedicaid 20% random sample (Medi-Cal files).
The Medicaid files contain eligibility details, demo- After the schizophrenia patients were identified, anti- graphics, and claims history for various health care ser- psychotic polypharmacy episodes for each person were vices, including Medicaid paid amount, outpatient pre- identified. For the purpose of this study, antipsychotic scription drugs, inpatient stays, and disease diagnosis. A polypharmacy was defined as 2 or more chemically dis- common resource available to Georgia Medicaid patients tinct antipsychotics prescribed concurrently with an over- is the 8 psychiatric hospitals managed by the DHR that do lap of at least 14 or more days of therapy.26 The list of not bill Medicaid for services rendered to persons aged 21 antipsychotics is provided in Table 1. Antipsychotic poly- to 64 years. Records from these 8 hospitals were com- pharmacy episodes were constructed using a “date of ser- bined to form the DHR file that contains a system-wide vice” variable that recorded the prescription fill date and a record of each visit a patient received at any of the 8 “days supply” variable that recorded the intended dura- system inpatient institutions in operation. To capture tion of each antipsychotic prescription filled for each per- psychiatric episodes of care, we linked the DHR files son. Figure 1 shows a hypothetical antipsychotic poly- (state-based institutional data) by patient identifiers to pharmacy episode involving risperidone and haloperidol.
the GDMA files (Georgia Medicaid claims data). These The first day of the polypharmacy episode was considered patient-linked or merged data provide a complete picture of the medical resources consumed for each Medicaid- An episode of antipsychotic polypharmacy was de- eligible patient with schizophrenia in Georgia. California fined as a period of continuous antipsychotic polyphar- COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Antipsychotic Polypharmacy Among Medicaid Patients Figure 2. Categories of Antipsychotic Polypharmacy and Monotherapy
macy without a break period of 31 or more days.27 A break Identification of Factors Associated With period was defined as a period when the patient had no supply of drugs. Hospital stays that occurred within 31 To identify factors associated with long-term antipsy- days of an antipsychotic use period were considered as chotic polypharmacy, we included recipients who had a continuation of the preceding episode and not a part been treated with antipsychotic polypharmacy or mono- of the break period if the therapy remained the same after therapy for at least 61 days between 1998 and 2000. Anti- psychotic use episodes in which a single antipsychotic Antipsychotic polypharmacy was classified in a hierar- was prescribed for at least 61 days without a break period chical fashion, narrowing the definition of antipsychotic of 31 or more days were referred to as monotherapy polypharmacy with each consecutive step in accordance with published treatment guidelines as displayed in Figure Among these recipients, those who did not have at least 2. The Journal of Clinical Psychiatry treatment guideline28 6 months of continuous Medicaid eligibility prior to their is the only guideline that offers guidance on the duration of single longest polypharmacy or monotherapy episode antipsychotic polypharmacy; it recommends polypharmacy were excluded. In addition, recipients who did not have at only when switching from one antipsychotic to another least 1 paid claim every 90 days prior to their single long- (cross-titration or overlap and taper) and for not more than est polypharmacy or monotherapy episode were excluded.
8 weeks or 2 months. Other guidelines do not advocate The latter criterion was applied to ensure that persons who any antipsychotic polypharmacy. Based on the Journal of were eligible for Medicaid benefits had not withdrawn Clinical Psychiatry treatment guideline, a subject with an from the system (e.g., were in prison).22 The single longest episode of antipsychotic polypharmacy longer than 2 episode of antipsychotic polypharmacy or monotherapy, months (at least 61 days) was categorized into the long- i.e., period of maximum exposure to treatment, between term antipsychotic polypharmacy cohort(s) (Figure 2).
1998 and 2000 was identified for each recipient. Depend- Since clozapine is generally reserved for treatment- ing on the type of antipsychotic polypharmacy or mono- resistant and comparatively more ill patients, the long-term therapy prescribed during the longest episode, e.g., long- antipsychotic polypharmacy cohort was separated into clo- term, clozapine, nonclozapine, clozapine + atypical, the zapine and nonclozapine cohorts for analyses. The cloza- recipient was grouped into one of the several long-term pine and nonclozapine groups were further categorized into polypharmacy or monotherapy groups (Figure 2).
clozapine subgroups (clozapine + atypical, clozapine + A comprehensive list of possible factors associated conventional) and nonclozapine subgroups (atypical + with antipsychotic polypharmacy was identified by a sur- atypical, atypical + conventional, conventional + conven- vey of published literature and expert opinion (Table 2).
tional). The prevalence of each type of antipsychotic This list included demographics, diagnosis-related co- polypharmacy was calculated over the 3 years and sepa- morbidities, drug classes, antipsychotic agents, and prior rately for each year. The Cochran-Armitage trend test was health care utilization variables. The diagnosis-related co- performed to estimate temporal changes in prevalence of morbidities and drug classes were obtained from a cost antipsychotic polypharmacy, and t tests were performed to prediction model for schizophrenia patients. This model estimate differences in prevalence between various catego- has been developed and validated on the Georgia Med- icaid database as a part of an Agency for Healthcare Re- COPYRIGHT 2004 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Table 2. Initial List of Candidate Factors Associated With Antipsychotic Polypharmacy
Demographics
(antiarrhythmic + inotropic + vasopressor, ACE inhibitors, (adrenergic bronchodilators + asthma vasopressors + combinations, methylxanthines, inhalants + leukotrienes + (hydantoin + succinimide + oxazolidinedione, barbiturates + certain benzodiazepines, miscellaneous anticonvulsants [valproic acid and derivatives, carbamazepine and derivatives, gabapentin, lamotrigine, tiagabine, topiramate, Hyperlipidemia, hypercholesterolemia drugs Menopause drugs (hormone replacement therapy) Pain (terminal) drugs, narcotic, analgesic Rheumatoid arthritis/collagen vascular disease Rheumatologic drugs/Crohn’s disease drugs/ulcerative colitis drugs Atypicals: olanzapine, risperidone, quetiapine, clozapine Conventionals: haloperidol oral and injectable, fluphenazine oral and injectable, thioridazine, chlorpromazine, thiothixene Depression or schizoaffective disorder† Psychiatric inpatient episode, latest inpatient days, Regular antipsychotic use (antipsychotic prescription filled every aThe prior period was the 6 months preceding the polypharmacy or monotherapy episode.
*Indicates that a hierarchy exists in relation to the cost categories below, which are denoted by a dagger (†). If both comorbidities were present, only the higher cost category was counted.
Abbreviations: ACE = angiotensin-converting enzyme, AIDS = acquired immunodeficiency syndrome, ESRD = end-stage renal disease.
search and Quality project to develop risk adjustment administration to differentiate between the injectable and indices for persons suffering from schizophrenia. The oral dosage forms; this was done because the injectable month and year of the episode start date were also in- forms are generally prescribed to a less compliant group cluded to identify any yearwise or seasonal trend in use.
of patients28 and compliance in turn may be an important The list of antipsychotic agents consisted of the 10 most factor associated with choice of therapy.
prevalent drugs identified from a frequency analysis of A stepwise logistic variable selection procedure was the prescription records from the prior period (the 6 used to test the hypothesis that demographics, diagnosis- months preceding polypharmacy or monotherapy). Halo- related comorbidities, drug classes, antipsychotic agents, peridol and fluphenazine were categorized by mode of and prior health care utilization variables (identified in COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Antipsychotic Polypharmacy Among Medicaid Patients Table 3. Demographic Characteristics of a Sample
Table 4. Prevalence of Long-Term Antipsychotic
of Medicaid-Eligible Schizophrenia Patients at Least
Polypharmacy in a Sample of Medicaid-Eligible
16 Years of Age
Schizophrenia Patients at Least 16 Years of Age, 1998–2000
Table 2) are independently associated with long-term antipsychotic polypharmacy. The binary treatment indi- cator (1 = long-term antipsychotic polypharmacy, 0 = monotherapy) was modeled, and main effects of the ini- tial list of factors were entered into the model if they met the significance level (p value) of .2 and removed if they did not meet the significance level of .1. To guard against model specification errors using stepwise procedures, we developed the initial model with a 70% random sample and utilized the remaining 30% of the subjects to validate the final model. The primary analysis was performed to identify factors associated with long-term antipsychotic polypharmacy, since long-term usage is not a recom- mended practice and is of greater policy relevance thanall usage that combines both short- and long-term use.
Subanalyses were performed to identify factors associ- (Table 4). California had a significantly higher prev- ated with usage by type of antipsychotic, e.g., clozapine, alence of antipsychotic polypharmacy across all the atypical + atypical, atypical + conventional, and conven- antipsychotic polypharmacy categories (p < .0001). The tional + conventional polypharmacy, and any differences prevalence of long-term polypharmacy was highest from the primary analysis were reported. The database among subjects exposed to clozapine (36.2%).
was managed using SAS software Version 8.02,29 and sta- Among the 7222 persons with 1 or more long-term tistical analysis was performed using SAS and STATA episodes, there were over 2.0 episodes (95% CI = 2.0 to 2.1) per recipient between 1998 and 2000, with a meanduration of 236 days per episode (95% CI = 230.3 to 241.6) (Table 5). On average, a long-term polypharmacypatient was exposed to polypharmacy for a total of 1 year (367 days) over the 3-year period, which after ad- We found that 32,280 persons (Georgia, 18,373; Cali- justment for differences in total months of Medicaid eli- fornia, 13,907) had received at least 1 primary diagnosis gibility translated to 388 days of polypharmacy for each of schizophrenia between 1998 and 2000, of whom recipient who was continuously eligible for 3 years. A 31,435 were at least 16 years of age as of Jan. 1, 1998, trend was observed toward higher polypharmacy episode and were retained in the cohort. The mean age of the duration for California Medicaid recipients compared 31,435 persons was 43 years (SD = 14 years) (both Geor- with Georgia recipients across all categories. Long-term gia and California patients had mean ages of 43 years), clozapine polypharmacy episodes had a longer mean du- 50% were female (Georgia, 56%; California, 43%), and ration of 301 days than long-term nonclozapine poly- 46% were white (Georgia, 38%; California, 57%) (Table pharmacy episodes, which lasted for a mean of 225 days 3). Of 31,435 persons, 88% (N = 27,757) had received at least 1 prescription for an antipsychotic drug. Of those Among the long-term antipsychotic polypharmacy who received any polypharmacy, 48% were female and groups, clozapine polypharmacy accounted for 11% of 47% were white. The overall prevalence of any antipsy- all long-term polypharmacy, and atypical + conventional chotic polypharmacy was 40% (N = 12,549) over the pe- polypharmacy accounted for 68% of all long-term poly- riod 1998–2000 and was 46% in California compared with 35% in Georgia (p < .0001). The mean duration ofpolypharmacy was 149 days.
A total of 23% (N = 7222) of recipients had 1 or The 3-year trend of long-term antipsychotic polyphar- more long-term antipsychotic polypharmacy episodes macy is presented in Figure 3. The overall prevalence COPYRIGHT 2004 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Table 5. Number and Length of Long-Term Antipsychotic Polypharmacy Episodes by Type of Polypharmacy, 1998–2000
All long-term polypharmacyNo. of episodes Long-term clozapine polypharmacyaNo. of episodes Long-term nonclozapine polypharmacyNo. of episodes aDefined as concurrent use of clozapine and a nonclozapine antipsychotic. At this level, “nonclozapine” is not differentiated into atypical or conventional. Therefore, a clozapine + atypical episode and clozapine + conventional episode occurring within 31 days of each other would beconsidered 1 episode and not 2 separate episodes.
Figure 3. Prevalence of Long-Term Antipsychotic Polypharmacy by Type, 1998–2000
of all antipsychotic polypharmacy increased significantly ciated with antipsychotic polypharmacy. A total of 8757 from 32% in 1998 to 41% in 2000 (Cochran-Armitage patients received long-term monotherapy and met the test: p < .0001); the increase in Georgia was from 24% to inclusion criteria. A 70% random sample, 4422 antipsy- 30% and in California from 43% to 62%. Except for cloza- chotic polypharmacy subjects and 6162 monotherapy pine + conventional polypharmacy (no change) and con- subjects, was retained for the primary analysis, and the re- ventional + conventional polypharmacy (decreased), all maining subjects were analyzed separately to estimate the antipsychotic polypharmacy prevalences increased from validity of the final model specification. Table 6 provides 1998 through 2000 (Cochran-Armitage test: p < .0001).
the adjusted odds ratios, 95% confidence intervals, anddistribution of the factors identified from the stepwise lo- Factors Associated With Antipsychotic Polypharmacy gistic regression analysis for the long-term antipsychotic Of the 7222 schizophrenia patients who received long- polypharmacy outcome in the primary sample. Forty vari- term antipsychotic polypharmacy between 1998 and 2000, ables were retained in the final model, and all variables 6438 were continuously eligible for Medicaid and had at were associated with long-term antipsychotic polyphar- least 1 claim every 90 days in the 6-month period preced- macy at a significance level of < .05 except for use of an- ing the episode and were retained to study the factors asso- tihypertensive drugs (p = .0579), insulins (p = .0649), and COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Antipsychotic Polypharmacy Among Medicaid Patients Table 6. Independent Factors Associated With Long-Term Antipsychotic Polypharmacy Identified From the Stepwise Logistic
Variable Selection Procedurea
Eligible for Medicaid in Georgia (vs California) First- and second-line antihypertensive drugs Exposure to 3 of 3 respiratory drug classesc Exposure to 1 of 2 antiepileptic drug classesd Hyperlipidemia, hypercholesterolemia drugs Health care utilization in 6 months preceding therapy episode Mental health cost in 6-month period, mean (SD), $ No. of psychiatric outpatient physician visits, mean (SD) (antipsychotic prescription filled every 2 mo) 3rd quarter start date (July, August, September)e 4th quarter start date (October, November, December)e aValues shown as N (%) unless otherwise noted. Association of predicted probabilities and observed responses: c-statistic for the final model = 0.9143, c-statistic for validation sample = 0.9174.
bCardiac drug classes are (1) antiarrhythmic + inotropic + vasopressor, (2) angiotensin-converting enzyme (ACE) inhibitors, (3) antianginal agents, cRespiratory drug classes are (1) adrenergic bronchodilators + asthma vasopressors + combinations, (2) methylxanthines, and (3) inhalants + leukotrienes + combinations.
dAntiepileptic drug classes are barbiturates + certain benzodiazepines and miscellaneous anticonvulsants (valproic acid and derivatives, carbamazepine and derivatives, gabapentin, lamotrigine, tiagabine, topiramate, levetiracetam).
fYear 1998 odds ratio = 1.
Abbreviation: AIDS = acquired immunodeficiency syndrome.
COPYRIGHT 2004 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
gout drugs (p = .0906). The c-statistic for the model was personality disorders (OR = 0.71) and drug use for car- 0.914 (0.5 for model with no predictive power; 1 for per- diac conditions (OR = 0.81) were negatively associated fect model), which shows that the model could discrim- with long-term antipsychotic polypharmacy, although the inate well between those who used long-term antipsy- strength of association was low (p ~ .05) for AIDS and chotic polypharmacy and those who used monotherapy.
The c-statistic for the final model in the 30% validation No additional factors were identified in the clozapine sample was 0.917, which shows that the model could dis- polypharmacy versus clozapine monotherapy analysis.
criminate equally well between antipsychotic polyphar- However, in the nonclozapine groups, arrhythmia (OR = macy and monotherapy in an external sample and sug- 2.0, 95% CI = 0.97 to 4.33), chronic obstructive pulmo- gests that the initial model was correctly specified.
nary disease (COPD) (OR = 1.7, 95% CI = 1.14 to 2.57), Being eligible for Georgia Medicaid was significantly asthma (OR = 2.3, 95% CI = 1.18 to 4.60), and compli- associated with a reduced likelihood of receiving long- cated diabetes (OR = 2.7, 95% CI = 1.10 to 6.90) were term antipsychotic polypharmacy as compared with positively associated with atypical + atypical polyphar- being eligible for California Medicaid (OR = 0.62, 95% macy; myocardial infarction (OR = 3.14, 95% CI = 0.90 CI = 0.54 to 0.70). Being of male gender and belonging to 10.93) was associated with atypical + conventional to the aid category (aged, blind, disabled) were associated polypharmacy; and coagulopathy (OR = 6.8, 95% CI = with an increased likelihood of receiving long-term anti- 1.14 to 40.94) was associated with conventional + con- psychotic polypharmacy. Among the diagnosis-related comorbidities, weight loss or malnutrition was stronglyassociated with long-term antipsychotic polypharmacy (OR = 4.50, 95% CI = 1.54 to 13.17), although the abso-lute numbers were small in both the antipsychotic poly- The Journal of Clinical Psychiatry treatment guide- pharmacy and monotherapy groups. Diagnosis of epi- line28 and some review articles on antipsychotic usage1,2,31 lepsy, other psychoses, and other mental disorders was recognize antipsychotic polypharmacy as a possible also positively associated with long-term antipsychotic option in 2 specific situations: short-term or p.r.n. use for “symptom control” and as a short-term tactic while Among the drug classes, exposure to drugs used to switching from one monotherapy to another. The Journal treat Parkinson’s disease, respiratory disorders, and can- of Clinical Psychiatry guideline is the only one that de- cer was associated with long-term antipsychotic poly- fines this short-term usage period, placing it at 2 months.
However, the authors also acknowledge the lack of pub- All antipsychotic drugs selected in the model were lished evidence and potential for adverse events with anti- strongly associated (ORs of 5 to 28, p < .0001) with a psychotic polypharmacy. Therefore, it is of concern to see higher likelihood of long-term antipsychotic polyphar- that 23% of patients received polypharmacy for more than macy. Among the atypical antipsychotics, quetiapine 2 months and the mean duration of use was almost 8 had the highest positive association with long-term anti- psychotic polypharmacy (OR = 18.32, 95% CI = 13.07 We found no previous studies that report prevalence to 25.68), followed by olanzapine (OR = 14.45) and ris- data for Medicaid-eligible schizophrenia patients or any peridone (OR = 9.18). Among the conventionals, chlor- that report prevalence of long-term polypharmacy for any promazine (OR = 28.87, 95% CI = 21.14 to 39.42), population. However, the overall combined prevalence followed by thioridazine (OR = 18.61) and thiothixene rate of long-term and short-term polypharmacy in our (OR = 8.44), had the highest positive associations. Cloza- study was similar to that reported in a 1985 survey of 768 pine (OR = 11.77) was also associated with long-term patients in 8 countries (overall prevalence = 40%, U.S.
prevalence = 36%)1 but was considerably higher than the Among the prior utilization variables, regular antipsy- prevalence reported in a more recent Veterans Affairs chotic use (1 antipsychotic prescription filled every 2 (VA) study looking at data over a short 4-month time pe- months) was associated (OR = 4.8) with long-term anti- riod (6.8%)32 and was also higher than those in a 1-year psychotic polypharmacy. Among the temporal variables, prevalence study using physician office–based data from index dates in the fourth quarter (October, November, 1997 (16.7%)33 and a Canadian study of hospital out- December) had a higher association with antipsychotic patients (27.5%).34 Some of the factors responsible for a polypharmacy (OR = 2) compared with those in the first higher prevalence estimate in our study could be a longer quarter. Also, the years 1999 (OR = 5.53) and 2000 study time frame (the VA study captured prescription (OR = 9.67) had a higher association with long-term anti- records during a 1-week period), a more recent year of psychotic polypharmacy compared with 1998.
study (the physician office–based study was performed in Diagnosis of acquired immunodeficiency syndrome 1997), Medicaid system-specific policies (California and (AIDS) (OR = 0.52), alcohol abuse (OR = 0.58), and Georgia Medicaid had removed a prior authorization rule COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Antipsychotic Polypharmacy Among Medicaid Patients restricting newer antipsychotic use in 1997, which may lence of long-term antipsychotic polypharmacy was 36% have increased antipsychotic switches, inflating our prev- in the clozapine-exposed group, which was higher than in alence estimates), and analysis of a Medicaid population, the atypical (29%) and conventional groups (31%), sug- which is considered a more ill group of patients than non– gesting refractory patients and less control with mono- Medicaid-eligible persons with schizophrenia.
The 11% difference in prevalence of long-term poly- The long-term nonclozapine atypical + conventional pharmacy between California and Georgia is interesting.
group had the highest prevalence, 16%. The therapeutic Additionally, the rate of increase in the prevalence of actions of conventional antipsychotic drugs are due to long-term antipsychotic polypharmacy from 1998 to blockade of dopamine (D ) receptors, whereas the atypi- 2000 was notably higher in California (19%) than in cal antipsychotics block both D and serotonin 5-HT Georgia (6%). There are several potential reasons for this receptors.35 Due to the differing receptor profiles, there occurrence. First, these findings could be a result of dif- may be a pharmacologic justification for combining ferences in reimbursement policies. We compared the atypicals with conventionals and using them for a long prescription and inpatient reimbursement policies be- duration, but there are no clinical studies that provide evi- tween the 2 states and found no substantial difference.
dence for such use. Individuals in the atypical + conven- Both states had rules requiring prior authorization for tional group in our study could have been patients receiv- atypical agents before 1998, reimbursed mental health ing treatment on a switchover or p.r.n. basis who were hospitalizations, had state psychiatric hospitals that bill started on a short-term combination therapy and then re- Medicaid, and had similar treatment authorization proce- mained on it for some reason. For example, the therapy dures. However, we cannot discount the fact that there may have been continued if a patient was stable and the could have been policy changes between 1998 and 2000 physician did not want to risk a relapse; this has been that we are not aware of that could have driven these quoted as a reason in a study of antipsychotic polyphar- prevalence rates. Second, the difference in prevalence macy.36 Physicians may also be observing that their pa- rate could be a reflection of dissimilar prescribing habits tients are stabilized on atypical + conventional combina- in the 2 states. However, among schizophrenia patients tion treatment and that there is real merit to such therapy.
who received at least 1 prescription for an antipsychotic, It is possible that diagnosis of epilepsy and drug use antipsychotic prescribing rates, which are a marker for for Parkinson’s disease may be predictors of long-term prescribing habits, were similar in the 2 states (atypicals: antipsychotic polypharmacy, as presence of these com- Georgia, 49%; California, 48%; conventionals: Georgia, mon conditions may induce a physician to switch to a 47%; California, 44%) except for clozapine (Georgia, better-tolerated treatment. However, since our study did 4%; California, 7%). Other process-related measures, for not have a medication-free washout period (49% of the example, the average number of physicians seen by Cali- antipsychotic polypharmacy patients had received poly- fornia subjects compared with Georgia subjects, could be pharmacy in some prior period), these could be side ef- responsible for the difference in long-term polypharmacy fects of prior treatment with antipsychotic polypharmacy rates. These are, however, speculations and cannot be or monotherapy. The association between diagnosis of cardiac arrhythmia and nonclozapine atypical + atypical The rising trend for all atypical polypharmacy com- polypharmacy (OR = 2.0) was as expected, given that binations (atypical + atypical, atypical + conventional, cardiovascular side effects such as tachycardia are com- atypical + clozapine) may be because of increased avail- mon with atypical monotherapy. Although a causal rela- ability of newer antipsychotics or changing prescribing tionship cannot be established from our analysis, this as- habits. As newer antipsychotics become available, there sociation may reflect an increased risk of cardiac side is a higher probability of receiving antipsychotic poly- effects with atypical polypharmacy. At the same time, pharmacy as patients on treatment with older antipsy- this association may be indicative of an effort to reduce chotics are switched to newer medications. The rise in cardiovascular side effects by titrating the patient to a atypical polypharmacy could also be due to changing pre- better-tolerated therapy. However, in the absence of any scribing habits among physicians as they find success and other information on this association, it might be prudent a growing confidence in treating patients with antipsy- to exercise caution, e.g., regular electrocardiogram moni- toring, especially while adding one atypical to another.
Long-term clozapine polypharmacy episodes had a Similarly, Parkinson’s disease was strongly associated mean duration of 301 days (95% CI = 281 to 320) and (p < .0001) with nonclozapine polypharmacy involving lasted longer than nonclozapine episodes (225 days, 95% conventional antipsychotics (OR > 3.5). The positive as- CI = 219 to 231); this may be explained by the fact that sociation between complicated diabetes and atypical + clozapine is reserved for treatment-refractory patients atypical polypharmacy may reflect recent concerns about who require longer durations of treatment than the rela- the occurrence of diabetes associated with the use of tively better controlled nonclozapine patients. The preva- olanzapine.37 The positive association between asthma/ COPYRIGHT 2004 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
COPD and atypical + atypical polypharmacy may be ex- pharmacy. The annual cost to the Medicaid systems of plained through evidence in the medical literature that adding a second antipsychotic in 2000 U.S. dollars was suggests noncausal mechanisms such as noncompliance to $494,877 for Georgia and $6,032,775 for California. From asthma therapy in this population and causal mechanisms a policy perspective, if Medicaid considered implement- such as depression of the central nervous system and im- ing a prior authorization rule for long-term polypharmacy paired respiratory drive during asthma attacks.38 (usage beyond 60 days), the Medicaid systems could save The odds ratios for all of the antipsychotic drugs $412,397 and $5,027,312 per year in antipsychotic pre- were high because we included only antipsychotics that scription costs in Georgia and California, respectively.
were highly prevalent in the long-term polypharmacy co- These estimates are based only on the incremental cost of hort. We did this because we were interested in under- the additional antipsychotic drugs when used concurrently standing the relative effects of being prescribed these 10 with another antipsychotic and do not include any admin- drugs, which are evident from the odds ratios. One of the istrative or indirect costs (savings) such as drug level interesting findings was that being placed on more re- monitoring costs or additional or reduced hospitalizations cently marketed atypical antipsychotics was associated that may be associated with polypharmacy.
with an increased likelihood of being exposed to long- The higher cost in California was driven by the fact that term polypharmacy; for example, for quetiapine, launched California has a larger Medicaid population, higher preva- in 1997, OR = 18.32; for olanzapine, launched in 1996, lence of clozapine + atypical and atypical + atypical poly- OR = 14.45; and for risperidone, launched in 1994, OR = pharmacy, and longer durations of polypharmacy. It is also 9.18. One explanation for this could be that more people important to note that in California there was a 9-fold on treatment with the most recent atypical compared with increase in the likelihood of atypical + atypical long-term other medications are in a switchover stage. This trend polypharmacy in 1999 over 1998 and that there was a was similar across all the groups. Also, use of low-potency 21-fold increase in 2000. This change was the highest antipsychotic medications such as quetiapine or chlorpro- among all other antipsychotic polypharmacy groups.
mazine was strongly associated with long-term antipsy- While evaluating these speculative cost estimates, it is im- portant to bear in mind that they are based on average Long-term antipsychotic polypharmacy also raises wholesale prices and recommended daily dosing, which some financial concerns, as the Medicaid systems have act as proxy measures for Medicaid-specific drug prices limited resources that are being allocated to expensive and actual average prescribed doses, respectively.
antipsychotic therapy. The clozapine + atypical and atypi- Some of the limitations of the study are that inpatient cal + atypical combinations are especially expensive, and medication use is not recorded in this database and has not the atypical + atypical combinations have been mentioned been accounted for except where the patient was pre- as cost drivers and a concern in the California Medicaid scribed the same medication before and after hospitaliza- system.39 Using the estimates for mean length of episode, tion. In that case, the patient was assumed to be on that number of episodes, prevalence by type of polypharmacy medication during the inpatient stay. As mentioned earlier, and a per-day average cost of antipsychotic,40 we calcu- we singled out the longest episodes in our analysis to iden- lated a rough speculative estimate* for the excess antipsy- tify factors associated with polypharmacy. The identified chotic cost per year to Medicaid due to long-term poly- longest episode may not be the first episode of antipsy-chotic polypharmacy (49% of the polypharmacy patientshad received polypharmacy in the prior period) for the pa- *Drug cost per day was calculated as the unit cost (average wholesale tient in the study period. Therefore, treatment factors, for price) of each drug multiplied by the average maintenance phase daily example, comorbid conditions or medication use, cannot dose (mg/day) of the drug as derived from the Journal of Clinical Psy- be interpreted as predictors of antipsychotic polyphar- chiatry treatment guideline. The average maintenance phase dailydoses for commonly prescribed agents were olanzapine 10 mg, risperi- macy as they may be the result of prior polypharmacy done 4 mg, quetiapine 300 mg, haloperidol 5 mg, fluphenazine 5 mg, and thioridazine 300 mg. A weighted average daily cost of atypicals Unless specifically mentioned, prevalence data re- (~ $8) and conventionals (~ $0.23) was obtained using the product-specific drug cost per day and the utilization of the different atypical ported here have not been adjusted for the fact that total and conventional drugs observed in Georgia and California Medicaid.
months of Medicaid eligibility over the 3-year period var- The annual cost of adding a second antipsychotic was calculated sepa- ied between recipients. However, this may have affected rately for each polypharmacy group by multiplying the weighted aver-age daily cost for atypical or conventional, multiplied by the average the results marginally, because over 90% of polypharmacy duration of polypharmacy, multiplied by the number of episodes of that subjects were continuously eligible for more than 2 polypharmacy combination. It was assumed that when there were com- years and 79% were continuously eligible for all 3 years.
binations using conventional drugs, the conventional drug would be thedrug discontinued and represent the incremental cost (atypicals and clo- Additionally, the eligibility profile was very similar zapine were assumed to be the drugs that would be kept if the patient (California vs. Georgia, 78% vs. 80% continuously eli- had to be treated with only 1 antipsychotic), and atypical costs were gible for 3 years) for both states and would not have counted only for atypical + atypical or clozapine + atypical polyphar-macy episodes.
COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
Antipsychotic Polypharmacy Among Medicaid Patients Like other administrative claims databases, the Med- 10. Cooke C, de Leon J. Adding other antipsychotics to clozapine [letter].
icaid databases have coding biases, as coding is dependent on reimbursement incentives and may not be totally com- 11. Gupta S, Sonnenberg SJ, Frank B. Olanzapine augmentation plete. These administrative databases do not include many of clozapine. Ann Clin Psychiatry 1998;10:113–115 12. Taylor CG, Flynn SW, Altman S, et al. An open trial of risperidone direct disease measures such as Positive and Negative Syn- augmentation of partial response to clozapine. Schizophr Res 2001; drome Scale scores that may be important predictors of antipsychotic polypharmacy. Our results are specific to 13. Kapur S, Roy P, Daskalakis J, et al. Increased dopamine d(2) receptor occupancy and elevated prolactin level associated with addition of Medicaid-eligible schizophrenia patients and may not be haloperidol to clozapine. Am J Psychiatry 2001;158:311–314 generalizable to other patient populations.
14. de Groot IW, Heck AH, van Harten PN. Addition of risperidone to clozapine therapy in chronically psychotic inpatients [letter]. J Clin 15. Waring EW, Devin PG, Dewan V. Treatment of schizophrenia with antipsychotics in combination [letter]. Can J Psychiatry 1999;44: Long-term antipsychotic polypharmacy is widely preva- 16. Waddington JL, Youssef HA, Kinsella A. Mortality in schizophrenia: lent (23%), is prescribed for long durations (~ 8 months), antipsychotic polypharmacy and absence of adjunctive anticholinergics and is an increasing phenomenon among Medicaid-eligible over the course of a 10-year prospective study. Br J Psychiatry 1998; schizophrenia patients, indicating a significant discrepancy 17. Henderson DC, Goff DC, Connolly CE, et al. Risperidone added to with practice guidelines. Effectiveness research is needed clozapine: impact on serum prolactin levels. J Clin Psychiatry 2001; to define the full scope and potential for prescribing anti- psychotic polypharmacy. In the absence of this infor- 18. Martin BC, Miller LS. Expenditures for treating schizophrenia: a population-based study of Georgia Medicaid recipients. Schizophr mation, clinicians have little or no prescribing guidance to treat schizophrenia patients who have had a partial re- 19. Martin BC, Kotzan JA, Reeves J. The direct medical costs of sponse or nonresponse to adequate doses of single anti- schizophrenia: a population based study of Georgia Medicaidrecipients [abstract]. Value Health 1998;1:65 20. Kotzan JA, Martin BC, Reeves JH, et al. The impact of race and fractures on mortality in a postmenopausal Medicaid population.
Drug names: aripiprazole (Abilify), carbamazepine (Carbatrol, Tegretol, and others), chlorpromazine (Thorazine, Sonazine, and 21. Martin BC, Miller LS, Kotzan JA. Antipsychotic prescription use and others), clozapine (Fazaclo, Clozaril, and others), fluphenazine costs for persons with schizophrenia in the 1990s: current trends and (Prolixin, Permitil, and others), gabapentin (Neurontin and others), five year time series forecasts. Schizophr Res 2001;47:281–292 haloperidol (Haldol and others), lamotrigine (Lamictal), levetiracetam 22. McCombs JS, Nichol MB, Stimmel GL, et al. Use patterns for antipsy- (Keppra), lithium (Eskalith, Lithobid, and others), loxapine (Loxitane chotic medications in Medicaid patients with schizophrenia. J Clin and others), mesoridazine (Serentil), molindone (Moban), olanzapine (Zyprexa), perphenazine (Trilafon and others), pimozide (Orap), 23. Ganguly R, Barghout V, Feride F, et al. The economic burden of irritable prochlorperazine (Compazine, Compro, and others), quetiapine bowel syndrome to Medicaid [poster]. Presented at the 66th annual (Seroquel), risperidone (Risperdal), thiothixene (Navane and others), scientific meeting of the American College of Gastroenterology (ACG); tiagabine (Gabitril), topiramate (Topamax), trifluoperazine (Stelazine and others), valproic acid (Depakene and others), ziprasidone 24. Malkin JD, Ackerman SJ, Schein J, et al. Cost and utilization patterns of fentanyl transdermal system and oxycodone hydrochloride controlled-release in a California Medicaid population. J Manag Care Pharm 2002;8:132–140 25. Lurie N, Popkin M, Dysken M, et al. Accuracy of diagnoses of schizo- phrenia in Medicaid claims. Hosp Community Psychiatry 1992;43: 1. Canales PL, Olsen J, Miller AL, et al. Role of antipsychotic poly- pharmacotherapy in the treatment of schizophrenia. CNS Drugs 26. Kotzan JA, Maclean R, Wade W, et al. Prevalence and patterns of concomitant use of selective serotonin reuptake inhibitors and other 2. Yuzda MSK. Combination antipsychotics: what is the evidence? antidepressants in a high-cost polypharmacy cohort. Clin Ther 3. Stubbs JH, Haw CM, Staley CJ, et al. Augmentation with sulpiride for 27. Svarstad BL, Shireman TI, Sweeney JK. Using drug claims data to a schizophrenic patient partially responsive to clozapine. Acta Psychiatr assess the relationship of medication adherence with hospitalization and costs. Psychiatr Serv 2001;52:805–811 4. Lerner V, Chudakova B, Kravets S, et al. Combined use of risperidone 28. Expert Consensus Guideline Series: Treatment of Schizophrenia 1999.
and olanzapine in the treatment of patients with resistant schizophrenia: J Clin Psychiatry 1999;60(suppl 11):1–80 a preliminary case series report. Clin Neuropharmacol 2000;23:284–286 29. SAS 8.02 [computer program]. [8.02]. Cary, NC: SAS Institute Inc.; 5. Chue P, Welch R, Snaterse M. Combination risperidone and quetiapine therapy in refractory schizophrenia [letter]. Can J Psychiatry 2001;46: 30. STATA 6.0 [computer program]. [6.0]. College Station, Tex: STATA 6. Mujica R, Weiden P. Neuroleptic malignant syndrome after addition of 31. Stahl SM. Antipsychotic polypharmacy, pt 1: therapeutic option or haloperidol to atypical antipsychotic [letter]. Am J Psychiatry 2001;158: dirty little secret? [BRAINSTORMS] J Clin Psychiatry 1999;60:425–426 32. Leslie DL, Rosenheck RA. Use of pharmacy data to assess quality of 7. Rhoads E. Polypharmacy of 2 atypical antipsychotics [letter].
pharmacotherapy for schizophrenia in a national health care system: individual and facility predictors. Med Care 2001;39:923–933 8. Raskin S, Katz G, Zislin Z, et al. Clozapine and risperidone: 33. Wang PS, West JC, Tanielian T, et al. Recent patterns and predictors of combination/augmentation treatment of refractory schizophrenia: antipsychotic medication regimens used to treat schizophrenia and other a preliminary observation. Acta Psychiatr Scand 2000;101:334–336 psychotic disorders. Schizophr Bull 2000;26:451–457 9. Morera AL, Barreiro P, Cano-Munoz JL. Risperidone and clozapine com- 34. Procyshyn RM, Kennedy NB, Tse G, et al. Antipsychotic polyphar- bination for the treatment of refractory schizophrenia. Acta Psychiatr macy: a survey of discharge prescriptions from a tertiary care COPYRIGHT 2004 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.
psychiatric institution. Can J Psychiatry 2001;46:334–339 use and diabetes. Psychother Psychosom 2002;71:244–254 35. Stahl SM. Essential Psychopharmacology: Neuroscientific Basis 38. Joseph KS, Asthma mortality and antipsychotic or sedative use: and Practical Applications. 2nd ed. Cambridge, England: Cambridge what is the link? Drug Saf 1997;16:351–354 39. Stahl S. Antipsychotic polypharmacy: evidence based or eminence 36. Tapp A, Wood AE, Secrest L, et al. Combination antipsychotic therapy based? [editorial] Acta Psychiatr Scand 2002;106:321–322 in clinical practice. Psychiatric Serv 2003;54:55–59 40. Red Book 2003 Drug Topics. Montvale, NJ: Thomson Medical 37. Ananth J, Venkatesh R, Burgoyne K, et al. Atypical antipsychotic drug COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2004 PHYSICIANS POSTGRADUATE PRESS, INC.

Source: http://www.uga.edu/psychology/millerlab/publications/pubs_pdf/RGangulyJClinPsycArticle.pdf