1214_yang
6/29/07
10:01 AM
Page 1214
Clinical Therapeutics/Volume 29, Number 6, 2007
Factors Related to Antipsychotic Oversupply Among Central Texas Veterans Min Yang, PhD, MD1*; Jamie C. Barner, PhD1; and Jason Worchel, MD2 1College
of Pharmacy, The University of Texas at Austin, Austin, Texas; and 2US Department of Veterans Affairs, Central Texas Veterans Health Care System, San Antonio, Texas ABSTRACT Background: There have been many studies of underadherence to antipsychotics, but antipsychotic overadherence, or medication oversupply, in which patients receive more prescription medications than are needed, has been overlooked. Both underadherence and oversupply can have an important impact on clinical outcomes. Objectives: This study examined adherence (based on the medication possession ratio [MPR]) among patients treated with antipsychotics in the Central Texas Veterans Health Care System (CTVHCS) and investigated factors associated with their adherence status. Methods: Data from September 1995 to October 2002 were extracted from the computerized patient record system of the CTVHCS for continuously enrolled adult outpatients receiving antipsychotic monotherapy and filling at least 2 prescriptions within a year of the index date. Patients’ prescription records were tracked for up to 12 months. Underadherence was defined as an MPR <0.8, good adherence as an MPR from 0.8 to 1.2, and oversupply as an MPR >1.2. Results: Of 3268 eligible patients, 49.9% had good adherence, 42.6% were underadherent, and 7.6% had medication oversupply. The overall mean (SD) MPR was 0.83 (0.33). Multinomial logistic regression analysis revealed that compared with patients with good adherence, underadherent patients were significantly more likely to be nonwhite (P < 0.001), younger (P < 0.01), and receiving chlorpromazine therapy (P < 0.05), and were less likely to be receiving fluphenazine (P < 0.01), olanzapine (P < 0.05), or risperidone (P < 0.05). Patients with medication oversupply were significantly more likely to be receiving olanzapine (P < 0.001), quetiapine (P < 0.01), or risperidone (P < 0.05) than those with good adherence. Conclusions: Although half of adult outpatients receiving antipsychotic monotherapy in the CTVHCS were adherent to their treatment regimens, a large pro1214
portion were underadherent, and a small proportion had medication oversupply. Patients receiving secondgeneration antipsychotics were more likely to be adherent and were more likely to have medication oversupply than patients receiving first-generation antipsychotics. (Clin Ther. 2007;29:1214–1225) Copyright © 2007 Excerpta Medica, Inc. Key words: adherence, oversupply, medication possession ratio, antipsychotic, medication utilization.
INTRODUCTION Because of the substantial personal, social, and economic consequences of patients’ underadherence to medication regimens, the prevalence and impact of underadherence to antipsychotic regimens have been studied extensively.1–5 Lacro et al4 conducted a metaanalysis of 39 published articles examining the prevalence of and risk factors for medication underadherence in patients with schizophrenia. They found a number of factors that were consistent predictors of underadherence (eg, poor insight, negative attitude toward medication) and a number that were inconsistent predictors (eg, side effects, receipt of firstgeneration antipsychotics [FGAs] or second-generation antipsychotics [SGAs]). Although previous studies have provided valuable insights into the matter of underadherence in patients with schizophrenia, an important aspect of adherence behavior has been overlooked—antipsychotic overadherence, or medication oversupply, in which patients *Current affiliation: QualityMetric Incorporated, Lincoln, Rhode Island. Accepted for publication April 16, 2007. doi:10.1016/j.clinthera.2007.06.013 0149-2918/$32.00 Printed in the USA. Reproduction in whole or part is not permitted. Copyright © 2007 Excerpta Medica, Inc.
Volume 29 Number 6
1214_yang
6/29/07
10:01 AM
Page 1215
M. Yang et al. receive more medications than are needed. Both underadherence and oversupply can have a substantial impact on clinical outcomes. For example, Valenstein et al6 used the medication possession ratio (MPR) to evaluate adherence status among patients with schizophrenia and found that patients with both poor adherence (MPR <0.8) and excess medication fills (MPR >1.1) had higher psychiatric admission rates than patients with good adherence (MPR 0.8–1.1). Although public awareness of medication oversupply may not be as great as that of medication underadherence, oversupply is likely to be more common than expected. A number of studies that used pharmacy records to assess adherence rates found that many patients acquired more medications than were needed.7–13 In these studies, the prevalence of an adherence rate ≥1.1, which was termed drug stockpiling, ranged from 4.8% to 35.1%. A study that used pill counts during home visits to examine adherence rates found that 18.4% of patients were overadherent (adherence rate >1.2) to at least 1 medication.14 A number of studies have examined the magnitude of wasted and/or unused medications associated with either underadherence or oversupply.15–20 Several of these studies were based on community pharmacy disposal programs.15–18 In a pharmacy-based wastedmedication collection program in Alberta, Canada, >204 tons of unused medicinal products were accumulated over a 7-year period.17 One cross-sectional study found that drug wastage among people aged >65 years accounted for 2.3% of all drug costs, representing over $1 billion of drug wastage in the elderly US population.19 A preliminary report on unused medications among 68 patients newly discharged from the hospital noted that between 2 and 22 bottles of antipsychotic medication had been found in patients’ medicine cabinets, although the adherence rate was not reported.20 Valenstein et al6 noted that although patients’ adherence status is customarily characterized as either nonadherent or adherent, adherence is actually a matter of degree; their later study in a Veterans Affairs population supported this statement.12 It would be logical to consider adherence as a continuum composed of 3 main areas: underadherence, good adherence, and overadherence (or oversupply). Because the term overadherence may imply that a patient actually consumes more medications than are needed, the term oversupply may be more descriptively accurate when June 2007
using an indirect measurement, such as a retrospective review of pharmacy records. While usually used to characterize underadherence, nonadherence should be examined using a 2-tailed test to incorporate the concept of medication oversupply. Because both underadherence and oversupply are inappropriate adherence behaviors, the term nonadherence contains both of these behaviors. Underadherence can occur when patients delay their medication refills, resulting in suboptimal medication availability. Underadherence can also occur when patients refill prescriptions on time but intentionally or unintentionally reduce the dosage or frequency of dosing. As a result, the remaining medications may accumulate over time. In contrast, medication oversupply may occur when patients repeatedly refill prescriptions before the previous prescription is exhausted. If patients consume the extra medications, serious medical problems may occur, with higher psychiatric admissions.6 If patients do not consume the extra medications, unused medications can accumulate. The medical costs of both underadherence and oversupply may be substantial. It has been estimated that 40% of the annual costs of rehospitalization in patients with schizophrenia are attributable to nonadherence.21 Given the high acquisition costs of the SGAs and their more frequent use in recent years, the costs associated with inappropriate medicationadherence behaviors are likely to increase. This retrospective cohort study used a prescription claims–based adherence measure, the MPR, to assess adherence rates. The concept of oversupply (based on MPR) was added to the existing concepts of underadherence and good adherence. A novel statistical approach was employed to support the feasibility of using the concept of oversupply and to examine factors related to adherence status. To explore the feasibility of this concept, a relatively conservative approach was taken to estimating the MPR. The objectives of the study were to assess MPR among patients taking antipsychotics; to examine adherence status (underadherence, good adherence, or oversupply) among patients taking antipsychotics of different classes; and to investigate factors related to adherence status.
METHODS Data Source The Central Texas Veterans Health Care System (CTVHCS) is one of the largest integrated health care 1215
1214_yang
6/29/07
10:01 AM
Page 1216
Clinical Therapeutics systems in the Department of Veterans Affairs (VA) and contains one of the largest inpatient psychiatric facilities in the country.22 Data were extracted from its computerized patient-record system (CPRS), which electronically captures patients’ medical information, including inpatient and outpatient records. Use of the CPRS medical records does not require patients’ informed consent. The present study employed data for adult outpatients receiving antipsychotic (FGA or SGA) monotherapy and filling at least 2 prescriptions for the same medication within 1 year from the date of the index prescription. The index prescription was the first antipsychotic prescription appearing in the database between September 30, 1995, and October 31, 2002. Eligible patients had to have been under CTVHCS care for ≥1 year from the index date. Patients who were switched from one antipsychotic to another or who were prescribed a combination of antipsychotics were excluded from the analysis, because the process of dose adjustment had the potential to result in higher numbers of unused medications. Patients with a record of hospital admission were also excluded, as these patients tend to have more severe illness and their inclusion might have biased the results; furthermore, inpatient medication utilization is usually under the supervision of nurses, whose medicationuse behavior may be systematically different from that in the outpatient setting. This study was approved by the institutional review boards of the CTVHCS and the University of Texas at Austin.
Study Variables Demographic and relevant clinical information were collected, including age, sex, race, psychiatric diagnoses, and antipsychotic utilization. Age was classified into 5 groups: 18–39, 40–49, 50–59, 60–69, and ≥70 years. Race was categorized as white or nonwhite. Psychiatric diagnoses were identified by International Classification of Diseases, Ninth Revision (ICD-9) codes. Four diagnoses (schizophrenia, bipolar disorder, depression, and substance abuse) were included in the study (yes/no). Schizophrenia and bipolar disorder were included because both are indications for antipsychotic use; depression and substance abuse were included because both have been associated with medication-adherence problems.23–27 A patient could have multiple mental health diag1216
noses. The FGAs included chlorpromazine, fluphenazine, haloperidol, loxapine, mesoridazine, perphenazine, thioridazine, thiothixene, and trifluoperazine. The SGAs included clozapine, olanzapine, quetiapine, risperidone, and ziprasidone. Haloperidol was used as the reference drug. The days’ supply of all prescriptions ranged from 1 to 90. The MPR was determined by dividing the sum of the days’ supply of all prescriptions dispensed within 1 year of the index date by the number of calendar days between the first and last fills plus the days’ supply of the last fill. The assumption underlying this calculation is that a patient receiving antipsychotic therapy should be treated continuously unless the patient chooses to discontinue or a physician discontinues the prescription; thus, an MPR of 1.0 would indicate that a patient received all medications as needed, and an MPR of 1.2 would indicate that a patient received 20% more medication than needed. For the purposes of this study, underadherence was defined as an MPR <0.8; good adherence was defined as an MPR between 0.8 and 1.2; and oversupply was defined as an MPR >1.2.
Statistical Analysis Descriptive analyses were used for overall patient characteristics and the characteristics of patients in each of the adherence categories. Unadjusted χ2 analyses were used to examine the bivariate association between underadherence, good adherence, and oversupply and particular demographic and clinical characteristics, as well as to examine whether there were differences between antipsychotic drug classes across adherence categories. A categorical multinomial logistic regression procedure was employed to evaluate the factors associated with being in a specific adherence category. Two models were analyzed. The dependent variable in both models was patients’ adherence status. The primary independent variable was the class of antipsychotics in model 1 (SGA vs FGA) and individual antipsychotic agents compared with haloperidol in model 2. All other covariates in the models were the same. SAS version 9.1.3 (SAS Institute Inc., Cary, North Carolina) was used for all data analyses.
RESULTS Within the study period, 6733 adult patients filled ≥1 antipsychotic prescription. Nine hundred thirtyVolume 29 Number 6
1214_yang
6/29/07
10:01 AM
Page 1217
M. Yang et al. two (13.8%) patients were excluded because of a hospital admission. A total of 1292 (19.2%) patients filled 1 prescription with no refills, and 838 (12.4%) received >1 type of antipsychotic in a year. An additional 403 (6.0%) patients filled multiple prescriptions for the same antipsychotic (either at the same or a different strength) on a single day. Patients filling prescriptions for different strengths of the same medication on a single day were excluded because of the difficulty of determining actual use of the prescription (eg, combinations of different strengths of an antipsychotic to achieve the desired dose). Patients filling multiple prescriptions for the same strength of the same medication on a single day were also excluded because of the possibility of a prescription reversal order. Thus, 3268 (48.5%) patients were included in the final analyses. Among these 3268 patients, 728 (22.3%) received prescriptions for at least 2 strengths of the same medication, but not on the same day. However, all patients with a recorded dose change filled subsequent prescriptions (for a different strength of that medication) after an elapsed period sufficient for them to have used half of the days’ supply of the previous prescription. Such prescriptions were considered to be filled within a reasonable period and thus were retained in the final analysis. Comparisons between demographic characteristics of the included (n = 3268) and excluded (n = 3465) patients showed no differences in mean age, sex, or racial distribution.
Patient Characteristics The demographic and clinical characteristics of the included patients are shown in Table I. The majority were male (94.5%) and white (67.7%). The mean (SD) age was 55.2 (14.4) years, and the majority (56.2%) of patients were aged ≥50 years. Based on ICD-9 codes, 9 (90.0%) of 10 patients had at least 1 mental health disorder. Schizophrenia (48.6%), depression (48.4%), bipolar disorder (36.0%), and substance abuse (39.7%) were the 4 most common mental health diagnoses.
Antipsychotic Supply Table II summarizes data on the number of antipsychotic prescriptions per patient over 1 year from the index data and days’ supply per prescription. The 3268 patients filled and/or refilled a total of 21,036 antipsychotic prescriptions. More than half (57.9%) June 2007
of patients had ≤6 prescriptions in a year, and the majority (91.5%) had ≤12 prescriptions (mean [SD], 6.4 [4.3] prescriptions/patient per year; median, 5 prescriptions/patient per year). The majority (85.4%) of prescriptions were for a 30-day supply (mean days’ supply, 32.0 [15.2] days; median, 30 days).
Medication Possession Ratio The figure shows the distribution of MPR values in the study population. Although the proportion of patients with an MPR >1.2 was relatively small, MPR values were distributed across the 3 adherence categories. With the exception of 5 patients, all MPR values were between 0 and 2.0. The frequency distribution peaked in the middle of the adherence continuum; 485 patients each had an MPR from 0.9 to <1.0 and from 1.0 to <1.1. Table III presents overall MPRs and the MPRs for each category of adherence, by type of antipsychotic. The number of patients receiving FGAs and SGAs was almost evenly divided (1668 and 1600, respectively). Half (49.9%) of patients were adherent to their regimens, 42.6% were underadherent, and 7.6% had medication oversupply. Among those with medication oversupply, 65.6% were receiving SGA therapy. Olanzapine was the most frequently prescribed antipsychotic (24.7%), followed by risperidone (19.2%) and haloperidol (16.6%). Olanzapine (33.6%), risperidone (24.3%), and haloperidol (13.4%) were also the most frequently prescribed antipsychotics in patients with an MPR >1.2. The difference in the likelihood of being in a particular adherence category differed significantly between patients receiving SGAs and those receiving FGAs (χ22 = 47.571; P < 0.001). Patients receiving SGAs were more likely to have medication oversupply than those receiving FGAs (χ21 = 29.561; P < 0.001) and were less likely to be underadherent (χ21 = 29.719; P < 0.001). Among recipients of FGAs, the likelihood of a patient being in a specific adherence category varied significantly across agents (χ24 = 27.578; P = 0.036), whereas no significant difference was found among SGA users (χ2 = 1.468). Due to small sample sizes, clozapine (n = 5) and ziprasidone (n = 4) were excluded from the bivariate analysis and, later, the multinomial logistic regression. The overall mean (SD) MPR was 0.83 (0.33). Recipients of SGAs had a higher mean MPR (0.87 [0.32]) than recipients of FGAs (0.79 [0.34]). Among underadherent patients, the lowest MPRs were ob1217
1214_yang
6/29/07
10:01 AM
Page 1218
Clinical Therapeutics
Table I. Demographic and clinical characteristics overall and by adherence status, based on the medication possession ratio (MPR). Unless otherwise specified, all values are expressed as no. (%). MPR
Variable
Overall (N = 3268)
<0.8 (n = 1391)
0.8–1.2 (n = 1630)
>1.2 (n = 247)
Age, mean (SD), y
55.2 (14.4)
53.6 (14.1)
56.4 (14.4)
55.7 (14.9)
Age group, y 18–39 40–49 50–59 60–69 ≥70
418 (12.8) 1014 (31.0) 723 (22.1) 456 (14.0) 657 (20.1)
190 (13.7) 491 (35.3) 295 (21.2) 167 (12.0) 248 (17.8)
198 (12.1) 444 (27.2) 376 (23.1) 260 (16.0) 352 (21.6)
30 (12.1) 79 (32.0) 52 (21.1) 29 (11.7) 57 (23.1)
Sex Male Female
3089 (94.5) 179 (5.5)
1317 (94.7) 74 (5.3)
1544 (94.7) 86 (5.3)
228 (92.3) 19 (7.7)
Race* White Nonwhite
1903 (67.7) 910 (32.3)
740 (60.3) 488 (39.7)
1008 (72.9) 374 (27.1)
155 (76.4) 48 (23.6)
Diagnosis Schizophrenia Yes No
1589 (48.6) 1679 (51.4)
683 (49.1) 708 (50.9)
851 (52.2) 779 (47.8)
119 (48.2) 128 (51.8)
Depression Yes No
1581 (48.4) 1687 (51.6)
683 (49.1) 708 (50.9)
851 (52.2) 779 (47.8)
119 (48.2) 128 (51.8)
Bipolar disorder Yes No
1178 (36.0) 2090 (64.0)
503 (36.2) 888 (63.8)
588 (36.1) 1042 (63.9)
160 (64.8) 87 (35.2)
Substance abuse Yes No
1298 (39.7) 1970 (60.3)
671 (48.2) 720 (51.8)
798 (49.0) 832 (51.0)
120 (48.6) 127 (51.4)
*n = 2813.
served for ziprasidone (0.43 [0.10]), chlorpromazine (0.49 [0.22]), and loxapine (0.49 [0.22]); however, the value for ziprasidone may not have been representative due to the small sample size. For descriptive purposes, the means and frequencies were also examined after categorizing patients based on an MPR >1 and ≤1. Nearly one third (29.6%) of patients had an MPR >1 (mean [SD], 1.18 [0.25]). Among patients with an MPR ≤1, the mean MPR value was 0.68 (0.24). 1218
In the bivariate analysis, no significant differences in the likelihood of a patient being in a specific adherence category were found by sex or mental health illness status. There were significant differences across adherence categories by race and age group (P < 0.001).
Multinomial Logistic Regression Analyses The findings of the 2 categorical multinomial logistic regression models were similar in many respects. In Volume 29 Number 6
1214_yang
6/29/07
10:01 AM
Page 1219
M. Yang et al.
Table II. Frequency of antipsychotic prescriptions per patient and days’ supply per prescription over 1 year from the index date.
Variable No. of antipsychotic prescriptions/patient Mean (SD) Median
Frequency, Cumulative, % %
6.4 (4.3) 5
– –
No. (%) of patients, by no. of prescriptions (N = 3268) 2–3 4–6 7–9 10–12 13–55
1027 (31.4) 865 (26.5) 600 (18.4) 499 (15.3) 277 (8.5)
31.4 57.9 76.3 91.5 100
Days’ supply/prescription Mean (SD) Median
32.0 (15.2) 30
– –
No. (%) of prescriptions, by days’ supply/ prescription (N = 21,036) 1–29 1534 (7.3) 30 17,961 (85.4) 31–60 507 (2.4) 61–90 1034 (4.9)
7.3 92.7 95.1 100
model 1, compared with patients with good adherence, patients receiving SGAs were more likely to have an oversupply and less likely to be underadherent than patients receiving FGAs. In model 2, the likelihood of a patient being in a specific adherence category varied significantly for particular antipsychotics compared with haloperidol (Table IV). Chlorpromazine was associated with a significantly greater likelihood of underadherence compared with haloperidol (P < 0.05), while fluphenazine (P < 0.01), olanzapine (P < 0.05), and risperidone (P < 0.05) were associated with a significantly lower likelihood of underadherence compared with haloperidol. None of the FGAs were associated with an increased likelihood of oversupply compared with haloperidol, whereas all 3 SGAs were associated with a significantly greater likelihood of oversupply (olanzapine, P < 0.001; quetiapine, P < 0.01; risperidone, P < 0.05). The effects of the covariates were similar in the 2 models. In the comparison between underadherence June 2007
and good adherence, older patients were less likely than younger patients to be underadherent (P < 0.01), and nonwhite patients were more likely than white patients to be underadherent (P < 0.001). The difference between the presence and absence of a mental health diagnosis was not statistically significant. In the comparison between oversupply and good adherence, none of the covariates were statistically significant.
DISCUSSION This study extended the concept of medication adherence beyond the dichotomous categories of poor adherence and good adherence. The rationale was that it may be more useful to view adherence as a continuum— from underadherence to good adherence to oversupply— along which the relationship between prescribed medications and patients’ medication-taking behavior can be quantified. Although these terms are similar to those used in the 2 studies by Valenstein et al,6,12 the present study differed in examining the full spectrum of adherence and directly focusing on factors associated with being in a specific adherence category. This study found that the mean (SD) adherence rate in these adult outpatients receiving antipsychotic monotherapy was 0.83 (0.33), comparable to other values reported in the literature (0.74–0.83).6,12,28–30 The publications by Valenstein et al12 and Gray et al14 included graphs of the adherence distribution in their populations that can be compared with the distribution in the present study (Figure). Although the proportion of patients in the oversupply category was higher in the study by Gray et al than in the present study, resulting in a higher mean MPR (0.95 [0.15]), the overall distribution between adherence categories was similar. Regardless of differences in the inclusion/ exclusion criteria and sample sizes, the MPR distribution in this study was comparable to that in the study by Valenstein et al. Underadherence was far more common than oversupply in the present study. In the studies by Valenstein et al6,12 and Gray et al,14 rates of underadherence were 40.3%, 39.6%, and 30.6%, respectively. The meta-analysis by Lacro et al4 found a mean nonadherence rate of 41.2%. Thus, the rate of underadherence in the present study (42.6%) was consistent with reports in the literature. In the studies by Valenstein et al6 and Gray et al,14 the rates of oversupply (defined as an MPR >1.2) were 2.5% and 18.4%, respectively. In the earlier litera1219
1214_yang
6/29/07
10:01 AM
Page 1220
Clinical Therapeutics
485 485
500 450 390
400
No. of Patients
350 301
300 250
319
227
200
169
215
186 140
150 105
100
64
60
50
31
24
17
11
3
5
5
0 to 0. <0 1 to .1 0. <0 2 . to 2 0. <0 3 . to 3 0. <0 4 . to 4 < 0. 5 0.5 to 0. <0 6 . to 6 0. <0 7 . to 7 < 0. 8 0.8 to 0. <0 9 . to 9 1. <1 . 0 to 0 < 1. 1 1.1 to 1. <1 2 . to 2 1. <1 3 . to 3 < 1. 4 1.4 to 1. <1 5 . to 5 1. <1 6 . to 6 < 1. 7 1.7 to 1. <1 8 . to 8 1. <1. 9 9 to <2 .0 ≥2 .0
0
26
MPR Range Figure. Medication possession ratio (MPR) distribution (N = 3268). ture7–11 and the more recent study by Valenstein et al,12 rates of oversupply (defined as an MPR ≥1.1) ranged from 4.8% to 35.1%. The rate of oversupply in the present study (7.6%) approached the lower end of these findings. When a post hoc analysis was conducted in which oversupply was defined as an MPR >1.1, the oversupply rate was 15.8%. When both underadherence and medication oversupply were included, 50.1% of patients in the present study were nonadherent. The study by Valenstein et al6 found that both types of nonadherent behavior could jeopardize the treatment outcome, leading to a higher risk of psychiatric admission and longer inpatient stays. Both factors also can lead to increases in health care costs. Although cost was not a focus of this study, it should be a concern in the long run because of the higher acquisition costs of the newer antipsychotics. The SGAs are the first-line medications for schizophrenia and bipolar disorder, and have been prescribed far more frequently than FGAs in recent years. In the present study, although the antipsychotic classes comprised similar numbers of patients, the majority of FGAs were prescribed before or in 1998, whereas the majority of the SGAs were prescribed after 1998. Oversupply of 1220
more expensive medications could have a prolonged impact on the allocation of health care resources. Multinomial logistic regression analysis was used to examine the relationship between adherence status and the type of antipsychotic prescribed. Reviews of the literature have reported that studies examining the relationship between antipsychotic class (ie, SGA vs FGA) and underadherence yielded inconsistent findings.4,5,31 The results of the present study indicate that patients receiving SGAs (olanzapine, risperidone, and/ or quetiapine) were more likely to be adherent and to have medication oversupply than patients receiving FGAs. A potential explanation for these findings is that physicians’ perceptions of medications may be involved in medication oversupply. Studies have indicated that although FGAs and SGAs may have similar efficacy, FGAs are associated with a high incidence of extrapyramidal symptoms.32,33 This adverse effect causes considerable discomfort and is an important cause of medication underadherence.32,33 Thus, physicians may be monitoring patients receiving FGAs more closely and may be more cautious in prescribing these agents for long durations or for multiple refills. On the other hand, physicians may be more confident in prescribing SGAs with longer intervals between ofVolume 29 Number 6
June 2007 1600 (49.0) 5 (0.2) 807 (24.7) 158 (4.8) 626 (19.2) 4 (0.1) 1668 (51.0) 150 (4.6) 76 (2.3) 542 (16.6) 50 (1.5) 30 (0.9) 190 (5.8) 284 (8.7) 268 (8.2) 78 (2.4)
Second generation* Clozapine Olanzapine Quetiapine Risperidone Ziprasidone
First generation† Chlorpromazine Fluphenazine Haloperidol Loxapine Mesoridazine Perphenazine Thioridazine Thiothixene Trifluoperazine 0.79 (0.34) 0.75 (0.34) 0.87 (0.24) 0.78 (0.36) 0.79 (0.30) 0.76 (0.29) 0.81 (0.46) 0.79 (0.29) 0.78 (0.29) 0.79 (0.25)
0.87 (0.32) 1.02 (0.03) 0.87 (0.32) 0.88 (0.33) 0.87 (0.31) 0.70 (0.56)
0.83 (0.33)
Mean (SD)
787 (56.6) 81 (5.8) 23 (1.7) 261 (18.8) 20 (1.4) 18 (1.3) 93 (6.7) 126 (9.1) 132 (9.5) 33 (2.4)
604 (43.4) – 311 (22.4) 61 (4.4) 229 (16.5) 3 (0.2)
1391 (42.6)
No. (%)
0.52 (0.20) 0.49 (0.22) 0.61 (0.19) 0.50 (0.20) 0.49 (0.22) 0.59 (0.20) 0.55 (0.19) 0.53 (0.18) 0.55 (0.19) 0.56 (0.19)
0.55 (0.17) – 0.55 (0.39) 0.55 (0.17) 0.55 (0.18) 0.43 (0.10)
0.53 (0.19)
Mean (SD)
796 (48.8) 59 (3.6) 52 (3.2) 248 (15.2) 27 (1.7) 10 (0.6) 89 (5.5) 145 (8.9) 123 (7.5) 43 (2.6)
834 (51.2) 5 (0.3) 413 (25.3) 79 (4.8) 337 (20.7) –
1630 (49.9)
No. (%)
0.97 (0.10) 1.00 (0.09) 0.96 (0.09) 0.97 (0.11) 0.95 (0.07) 0.94 (0.10) 0.98 (0.10) 0.97 (0.10) 0.96 (0.10) 0.95 (0.08)
1.00 (0.11) 1.02 (0.03) 1.01 (0.11) 1.01 (0.13) 1.01 (0.11) –
0.99 (0.11)
Mean (SD)
MPR 0.8–1.2
85 (34.4) 10 (4.0) 1 (0.4) 33 (13.4) 3 (1.2) 2 (0.8) 8 (3.2) 13 (5.3) 13 (5.3) 2 (0.8)
162 (65.6) – 83 (33.6) 18 (7.3) 60 (24.3) 1 (0.4)
247 (7.6)
No. (%)
1.51 (0.58) 1.34 (0.14) – 1.55 (0.51) 1.34 (0.06) 1.38 (0.03) 1.99 (1.49) 1.42 (0.22) 1.42 (0.17) 1.22 (0.02)
1.40 (0.20) – 1.40 (0.20) 1.44 (0.20) 1.39 (0.20) –
1.44 (0.38)
Mean (SD)
MPR >1.2
MPR <0.8 = underadherence; MPR 0.8–1.2 = good adherence; MPR >1.2 = oversupply. *χ2 analysis between atypical and typical antipsychotics: χ22 = 47.571; P < 0.001. χ2 analysis between individual atypical antipsychotics (clozapine and ziprasidone were excluded due to small sample sizes): χ24 = 1.468; P = 0.832. † χ2 analysis between atypical and typical antipsychotics: χ2 = 47.571; P < 0.001. χ2 analysis between individual typical antipsychotics: χ2 = 27.578; P = 0.036. 2 4
3268 (100)
All
No. (%)
MPR <0.8
10:01 AM
Overall MPR
6/29/07
Antipsychotic Category
Table III. Medication possession ratio (MPR), by antipsychotic category.
1214_yang Page 1221
M. Yang et al.
1221
1214_yang
6/29/07
10:01 AM
Page 1222
Clinical Therapeutics
Table IV. Multinomial logistic regression analysis of factors related to antipsychotic adherence status, by medication possession ratio (MPR) (N = 2806). Model likelihood ratio = 2009.09; P = 0.982 (indicating good fit of the model). MPR <0.8 vs MPR 0.8–1.2 Variable
MPR >1.2 vs MPR 0.8–1.2
OR
95% CI
OR
95% CI
1.577† 0.470‡ 0.867 1.946 1.120 1.014 1.188 0.790
1.105–2.250 0.288–0.767 0.494–1.520 0.937–4.041 0.824–1.523 0.780–1.318 0.914–1.543 0.500–1.249
1.392 0.189 1.162 1.898 0.908 0.884 0.878 0.226
0.632–3.066 0.030–1.188 0.371–3.640 0.457–7.883 0.435–1.896 0.470–1.654 0.460–1.677 0.036–1.422
0.820† 0.929 0.770†
0.676–0.996 0.629–1.371 0.625–0.948
1.989§ 2.582‡ 1.628†
1.341–2.948 1.380–4.829 1.063–2.493
Age group
0.900‡
0.845–0.959
0.910
0.807–1.026
Nonwhite (reference: white)
1.311§
1.204–1.426
0.883
0.740–1.052
Female (reference: male)
1.018
0.850–1.218
1.190
0.884–1.602
0.973 1.005
0.898–1.054 0.923–1.094
1.020 1.055
0.876–1.187 0.897–1.241
1.001
0.917–1.093
0.925
0.781–1.095
1.009
0.927–1.097
1.121
0.955–1.315
Antipsychotic (reference: haloperidol) First generation* Chlorpromazine Fluphenazine Loxapine Mesoridazine Perphenazine Thioridazine Thiothixene Trifluoperazine Second generation Olanzapine Quetiapine Risperidone
Comorbidity Schizophrenia (reference: no schizophrenia) Depression (reference: no depression) Bipolar disorder (reference: no bipolar disorder) Substance abuse (reference: no substance abuse)
MPR <0.8 = underadherence; MPR 0.8–1.2 = good adherence; MPR >1.2 = oversupply; OR = odds ratio. *Clozapine and ziprasidone were excluded from the analysis due to small sample sizes. † P < 0.05. ‡ P < 0.01. §P < 0.001.
fice visits and/or multiple refills. A post hoc analysis of data from the present study found that the mean (SD) days’ supply of FGAs was 28.80 (12.75), compared with 31.02 (15.04) for the SGAs (t = 4.557; P < 0.001). Among the covariates, nonwhite race was identified as a significant factor in the comparison between underadherence and good adherence (P < 0.001). Al1222
though some studies have found race to be a nonsignificant factor with respect to adherence status,34–40 many have reported that nonwhite patients had a lower adherence rate than white patients (including the study by Valenstein et al12: odds ratio = 2.38; 95% CI, 2.28–2.49).41–46 The result of the present study was consistent with a greater likelihood of underadherence in nonwhite patients, although the exact reaVolume 29 Number 6
1214_yang
6/29/07
10:01 AM
Page 1223
M. Yang et al. sons are not clear. It has been speculated that race may be an intermediate factor related to health beliefs, medication management, and access to care.12 The present study also found age to be a significant factor, with younger patients having a greater likelihood of underadherence than older patients (P < 0.01). This is consistent with the results of a number of other studies,12,43,47–50 although one study reported that older patients had a greater likelihood of underadherence.42 As noted by Schectman et al,43 higher adherence rates among older patients may be an effect of training and/or maturation. This study found no association between the presence of particular mental health diagnoses (ie, bipolar disorder, depression, schizophrenia, or substance abuse) and adherence status. Although other studies have reported that the presence of depression is an important factor in underadherence in other disease states (eg, coronary heart disease,23 type 2 diabetes,24 hypertension,25 asthma26), the present study found no evidence to support this suggestion, probably because many of the patients in these studies had multiple mental health comorbidities.
Study Limitations This study, which was based on a retrospective review of claims data, used the MPR to determine adherence rates. It was assumed that patients took medications as prescribed. Pharmacy records indicate receipt of a medication but not whether patients actually ingested the medication. Thus, nonuse of medication cannot be observed directly. Furthermore, if patients had refilled their prescriptions in a pharmacy outside the VA system, the MPR values could have been underestimated. However, given the extensive level of VA service–related medication coverage, few prescriptions are filled at locations other than VA pharmacies. There has been little research on medication oversupply. To the best of our knowledge, this is the first study to use multinomial logistic regression to incorporate all 3 aspects of adherence (underadherence, adherence, and oversupply) in a single model to evaluate factors related to specific categories of adherence. An adherence rate <0.8 has commonly been used to define underadherence. However, use of an MPR >1.2 to define oversupply may be controversial, as an adherence rate/MPR of 1.1 has been used to define oversupply in other studies.6,12–18 Because of the potential June 2007
policy implications, we feel it was appropriate to use a relatively conservative threshold to define oversupply. Furthermore, because this study used prescription information to estimate adherence, it was not possible to gain direct insights into the reasons for medication oversupply. Additional studies are needed that employ personal interviews with patients and/or surveys of physicians’ perceptions to obtain first-hand knowledge of this subject. This information could be of value in improving treatment outcomes, reducing medication nonuse, and maximizing the benefits of limited medical resources. In addition, because of the stringent inclusion and exclusion criteria employed in this study, >50% of patients were excluded from the total sample. This could limit the generalizability of the findings. However, the MPRs for the realized sample can represent only a conservative estimate. For example, patients switched between antipsychotics or receiving combination treatment have been reported to have higher MPRs than patients receiving monotherapy.12 Hospitalized patients usually have improved adherence because of nurse supervision. On the other hand, this study was not limited to patients with a diagnosis of schizophrenia, which extended the applicability of the findings. The similarity between the MPR distributions in our study and in the study by Valenstein et al,12 which was based on all patients with schizophrenia in the VA system, supports the methods used in the present study and suggests an overall commonality of inappropriate medication-use behavior. Finally, the results of this study may not be generalizable to other health care systems. In the VA health care system, mental health disorders are usually classified as service related, which means that there is usually no out-of-pocket cost for patients receiving antipsychotic therapy. Thus, it is necessary to examine whether medication oversupply exists in other health care systems, as well as in association with medications for other conditions.
CONCLUSIONS This study added the concept of oversupply (based on MPR) to the retrospective examination of adherence in adult outpatients treated with antipsychotic monotherapy in the CTVHCS and employed a novel statistical approach to investigate this concept’s feasibility. Descriptive and multinomial logistic regression analyses were used to assess the extent of underadherence, 1223
1214_yang
6/29/07
10:01 AM
Page 1224
Clinical Therapeutics good adherence, and oversupply, and to identify factors related to adherence status. Half (49.9%) of patients were adherent to their treatment regimen, 42.6% were underadherent, and 7.6% had medication oversupply. Patients receiving SGAs were more likely to be adherent and more likely to have medication oversupply than patients receiving FGAs. Nonwhite patients were more likely to be underadherent than white patients, and older patients had better adherence than younger patients. More investigation is needed to help understand the concept of medication oversupply and its relationship to underadherence and good adherence. Further research on oversupply is warranted in other health care systems and involving other drug classes.
REFERENCES 1. Babiker IE. Noncompliance in schizophrenia. Psychiatr Dev. 1986;4:329–337. 2. Fenton WS, Blyler CR, Heinssen RK. Determinants of medication compliance in schizophrenia: Empirical and clinical findings. Schizophr Bull. 1997;23:637–651. 3. Hughes I, Hill B, Budd R. Compliance with anti-psychotic medications: From theory to practice. J Ment Health. 1997; 6:473–489. 4. Lacro JP, Dunn LB, Dolder CR, et al. Prevalence of and risk factors for medication nonadherence in patients with schizophrenia: A comprehensive review of recent literature. J Clin Psychiatry. 2002;63:892–909. 5. Pinikahana J, Happell B, Taylor M, Keks NA. Exploring the complexity of compliance in schizophrenia. Issues Ment Health Nurs. 2002;23:513–528. 6. Valenstein M, Copeland LA, Blow FC, et al. Pharmacy data identify poorly adherent patients with schizophrenia at increased risk for admission. Med Care. 2002;40:630– 639. 7. Steiner JF, Koepsell TD, Fihn SD, Inui TS. A general method of compliance assessment using centralized pharmacy records. Description and validation. Med Care. 1988;26: 814–823. 8. Meyer TJ, Van Kooten D, Marsh S, Prochazka AV. Reduction of polypharmacy by feedback to clinicians. J Gen Intern Med. 1991;6:133–136. 9. Steiner JF, Fihn SD, Blair B, Inut TS. Appropriate reductions in compliance among well-controlled hypertensive patients. J Clin Epidemiol. 1991;44:1361–1371. 10. Steiner JF, Robbins LJ, Roth SC, Hammond WS. The effect of prescription size on acquisition of maintenance medications. J Gen Intern Med. 1993;8:306–310. 11. Porter AM. Drug defaulting in a general practice. Br Med J. 1969;1:218–222.
1224
12. Valenstein M, Blow FC, Copeland LA, et al. Poor antipsychotic adherence among patients with schizophrenia: Medication and patient factors. Schizophr Bull. 2004;30: 255–264. 13. Wandless I, Mucklow JC, Smith A, Prudham D. Compliance with prescribed medicines: A study of elderly patients in the community. J R Coll Gen Pract. 1979;29:391–396. 14. Gray SL, Mahoney JE, Blough DK. Medication adherence in elderly patients receiving home health services following hospital discharge. Ann Pharmacother. 2001;35:539–545. 15. Boivin M. The cost of medication waste. Can Pharmaceutical J. 1997;130:32–39. 16. Cameron S. Study by Alberta pharmacists indicates drug wastage a “mammoth” problem. CMAJ. 1996;155:1596– 1598. 17. Carter BA, Holland CL. Drug non-utilization review: EnvirRx Research Project on Drug Waste. Drug Use Elderly Q. 1996;12:1–4. 18. Garey KW, Johle ML, Behrman K, Neuhauser MM. Economic consequences of unused medications in Houston, Texas. Ann Pharmacother. 2004;38:1165–1168. 19. Morgan TM. The economic impact of wasted prescription medication in an outpatient population of older adults. J Fam Pract. 2001;50:779–781. 20. Velligan DI, Lam F, Ereshefsky L, Miller AL. Psychopharmacology: Perspectives on medication adherence and atypical antipsychotic medications. Psychiatr Serv. 2003;54: 665–667. 21. Weiden PJ, Olfson M. Cost of relapse in schizophrenia. Schizophr Bull. 1995;21:419–429. 22. US Department of Veterans Affairs. VISN 17: VA Heart of Texas Health Care Network. Available at: http://www1.va. gov/directory/guide/facility.asp?ID=1017. Accessed October 4, 2006. 23. Gehi A, Haas D, Pipkin S, Whooley MA. Depression and medication adherence in outpatients with coronary heart disease: Findings from the Heart and Soul Study. Arch Intern Med. 2005;165:2508–2513. 24. Kalsekar ID, Madhavan SS, Amonkar MM, et al. Depression in patients with type 2 diabetes: Impact on adherence to oral hypoglycemic agents. Ann Pharmacother. 2006;40:605–611. 25. Morris AB, Li J, Kroenke K, et al. Factors associated with drug adherence and blood pressure control in patients with hypertension. Pharmacotherapy. 2006;26:483–492. 26. Smith A, Krishnan JA, Bilderback A, et al. Depressive symptoms and adherence to asthma therapy after hospital discharge. Chest. 2006;130:1034–1038. 27. Wilk J, Marcus SC, West J, et al. Substance abuse and the management of medication nonadherence in schizophrenia. J Nerv Ment Dis. 2006;194:454–457. 28. Kruse W, Koch-Gwinner P, Nikolaus T, et al. Measurement of drug compliance by continuous electronic monitoring:
Volume 29 Number 6
1214_yang
6/29/07
10:01 AM
Page 1225
M. Yang et al.
29.
30.
31.
32.
33.
34.
35.
36.
37.
A pilot study in elderly patients discharged from hospital. J Am Geriatr Soc. 1992;40:1151–1155. Lowe CJ, Raynor DK, Courtney EA, et al. Effects of self medication programme on knowledge of drugs and compliance with treatment in elderly patients. BMJ. 1995;310:1229–1231. Rich MW, Gray DB, Beckham V, et al. Effect of a multidisciplinary intervention on medication compliance in elderly patients with congestive heart failure. Am J Med. 1996;101:270–276. Keith SJ, Kane JM. Partial compliance and patient consequences in schizophrenia: Our patients can do better. J Clin Psychiatry. 2003;64:1308–1315. Casey DE. Neuroleptic-induced acute extrapyramidal syndromes and tardive dyskinesia. In: Hirsch SR, Weinberger DR, eds. Schizophrenia. London, UK: Blackwell Science; 1995:546–565. Kapur S, Remington G. Atypical antipsychotics: New directions and new challenges in the treatment of schizophrenia. Annu Rev Med. 2001; 52:503–517. Dolder CR, Lacro JP, Dunn LB, Jeste DV. Antipsychotic medication adherence: Is there a difference between typical and atypical agents [published correction appears in Am J Psychiatry. 2002;159:514]? Am J Psychiatry. 2002;159:103–108. Grunebaum MF, Weiden PJ, Olfson M. Medication supervision and adherence of persons with psychotic disorders in residential treatment settings: A pilot study [published correction appears in J Clin Psychiatry. 2001;62:480]. J Clin Psychiatry. 2001; 62:394–399. Nageotte C, Sullivan G, Duan N, Camp PL. Medication compliance among the seriously mentally ill in a public mental health system. Soc Psychiatry Psychiatr Epidemiol. 1997; 32:49–56. Dixon L, Weiden P, Torres M, Lehman A. Assertive community treatment and medication compliance in the homeless mentally ill. Am J Psychiatry. 1997;154:1302–1304.
June 2007
38. Owen RR, Fischer EP, Booth BM, Cuffel BJ. Medication noncompliance and substance abuse among patients with schizophrenia. Psychiatr Serv. 1996;47:853–858. 39. Buchanan A. A two-year prospective study of treatment compliance in patients with schizophrenia. Psychol Med. 1992;22:787–797. 40. Heyscue BE, Levin GM, Merrick JP. Compliance with depot antipsychotic medication by patients attending outpatient clinics. Psychiatr Serv. 1998;49:1232–1234. 41. Fleck DE, Hendricks WL, DelBello MP, Strakowski SM. Differential prescription of maintenance antipsychotics to African American and white patients with new-onset bipolar disorder. J Clin Psychiatry. 2002;63: 658–664. 42. Rosenheck R, Chang S, Choe Y, et al. Medication continuation and compliance: A comparison of patients treated with clozapine and haloperidol. J Clin Psychiatry. 2000;61: 382–386. 43. Schectman JM, Bovbjerg VE, Voss JD. Predictors of medication-refill adherence in an indigent rural population. Med Care. 2002;40:1294–1300. 44. Sellwood W, Tarrier N. Demographic factors associated with extreme non-compliance in schizophrenia. Soc
45.
46.
47.
48.
49.
50.
Psychiatry Psychiatr Epidemiol. 1994;29: 172–177. Apter AJ, Boston RC, George M, et al. Modifiable barriers to adherence to inhaled steroids among adults with asthma: It’s not just black and white. J Allergy Clin Immunol. 2003; 111:1219–1226. Mark TL, Palmer LA, Russo PA, Vasey J. Examination of treatment pattern differences by race. Ment Health Serv Res. 2003;5:241–250. Monane M, Bohn RL, Gurwitz JH, et al. Noncompliance with congestive heart failure therapy in the elderly. Arch Intern Med. 1994;154:433–437. Schectman JM, Nadkarni MM, Voss JD. The association between diabetes metabolic control and drug adherence in an indigent population. Diabetes Care. 2002;25:1015– 1021. Agarwal MR, Sharma VK, Kishore Kumar KV, Lowe D. Non-compliance with treatment in patients suffering from schizophrenia: A study to evaluate possible contributing factors. Int J Soc Psychiatry. 1998;44:92–106. Duncan JC, Rogers R. Medication compliance in patients with chronic schizophrenia: Implications for the community management of mentally disordered offenders. J Forensic Sci. 1998;43:1133–1137.
Address correspondence to: Min Yang, PhD, MD, QualityMetric Incorporated, 640 George Washington Highway, Suite 201, Lincoln, RI 02865. E-mail:
[email protected] 1225