Content uploaded by Ibrahim Saud Al-Zakwani
Author content
All content in this area was uploaded by Ibrahim Saud Al-Zakwani
Content may be subject to copyright.
Objective: To examine the effects of typical and
atypical antipsychotics on medication adherence
and healthcare resource utilization.
Research design and methods: This was a
retrospective observational cohort analysis of
pharmacy and medical health insurance
reimbursement data of patients from a
southeastern United States health plan. Pharmacy
data of subjects between 6 and 65 years of age
were identified. Inclusion criteria included initiation
of a single antipsychotic agent between July 1,
1999 and September 30, 2000; no antipsychotic
medication usage 6 months prior to the index
prescription date; and continuous health plan
enrollment for the 18-month study period.
Multivariable methods were utilized to analyze
healthcare resource utilizations between groups.
Outcome measures: Primary outcome measures
included: (1) adherence and persistence with
antipsychotic therapy; (2) healthcare utilization for
outpatient office and hospital visits, inpatient
hospital visits, and emergency room visits; and (3)
therapy modifications and concomitant
medications.
Results: A total of 469 patients met initial study
criteria. Atypical and typical antipsychotics were
prescribed to 384 and 85 patients, respectively.
Length of therapy (days) for the atypical cohort was
significantly longer (136 vs 80; p< 0.001). As
defined using medication possession ratio (MPR),
the atypical cohort was significantly more adherent
to therapy than the typical cohort (mean MPR, 0.53
vs 0.24; p< 0.001). After adjusting for differences
in demographics, baseline utilization, MPR, and
length of therapy (n= 377), the atypical cohort
experienced significantly fewer office visits (2,635
vs 4,249 per 1000 patients per month [P1000PPM];
p= 0.005), fewer inpatient admissions (197 vs 511
P1000PPM; p= 0.032), and fewer emergency room
visits (125 vs 354 P1000PPM; p= 0.002).
Conclusions: Atypical antipsychotic users were
significantly more adherent to therapy, and had
lower rates of office, hospital and emergency room
utilization. Within the context of inherent limitations
associated with health insurance claims databases,
this study suggests that a relationship exists across
cohorts between medication adherence and use of
healthcare resources.
SUMMARY
ORIGINAL ARTICLE
Analysis of healthcare utilization
patterns and adherence in
patients receiving typical and
atypical antipsychotic
medications*
Ibrahim S. Al-Zakwani1, John J. Barron1, Michael F.
Bullano1, Steve Arcona2, Christopher J. Drury1and
Tara R. Cockerham3
1Health Core, Inc., Newark, DE, USA
2Novartis Pharmaceuticals Corp., East Hanover, NJ, USA
3Wellpoint Pharmacy Management, West Hills, CA, USA
Address for correspondence: Dr John J. Barron, Health Core, Inc., 4735 Ogletown-Stanton Road, Suite
3201, Newark, DE 19713, USA. Tel. 302-623-0513; Fax 302-623-0505; email jbarron@healthcore.com
Key words: Adherence – Antipsychotics, atypical, typical – Medication possession ratio – Persistence –
Resource utilization
Paper 2395 619
CURRENT MEDICAL RESEARCH AND OPINION®
VOL. 19, NO. 7, 2003, 619–626
© 2003 LIBRAPHARM LIMITED
0300-7995
doi:10.1185/030079903125002270
*This project was presented as a poster presentation at the International Society for Pharmacoeconomics and Outcomes Research
(ISPOR) international meeting in Arlington, VA, USA, 19–21 May 2003
Introduction
Each year, 44 million Americans are affected by
diagnosable mental disorders, and more than 5% of
American adults are diagnosed with ‘serious’ mental
illnesses, defined as disorders that interfere with some
area of social functioning1. The 1-year prevalence rates
of mental disorders for American adults have been
estimated at 22–23%2–4, and approximately 2.6% of
adults have ‘severe and persistent’ mental illness, a
category that includes schizophrenia, bipolar disorder,
panic disorder, obsessive-compulsive disorder, and
severe forms of depression5.
In 1996, the direct costs of mental health services in
the United States totaled $69.0 billion, a figure
representing 7.3% of total health spending. In addition,
lost productivity at the workplace, school, and home
due to premature death or disability was estimated at
$78.6 billion in 1990, making mental disorders one of
the most costly groups of illnesses in the country1,5–7.
Current treatment of psychotic disorders includes
typical and atypical antipsychotic medications. Most
atypical agents are now generally considered the first
line of pharmacotherapy for these disorders8. Atypical
drugs have been associated with improved clinical
outcomes and quality of life with fewer relapses9,10, as
well as reductions in length of hospital stays, number of
physician visits, and overall managed care costs11–15. It
has been suggested that these reductions occur
secondary to improved adherence and fewer extra-
pyramidal-symptoms (EPS)16–18. A number of studies
have evaluated the adherence patterns of typical and
atypical antipsychotics19–21, but to our knowledge, no
research has focused on the association between
medication adherence and resource utilization.
Following this logic, the purpose of our study was to
examine the effects of typical and atypical
antipsychotics on medication adherence and healthcare
resource utilization in a managed care organization.
Methods
Patients
Patients for this study were identified using the
insurance reimbursement data of a private southeastern
United States health plan containing the information of
approximately 500 000 beneficiaries. Through use of
pharmacy service data, patients between 6 and 65 years
of age who were initiated on a single antipsychotic agent
between July 1, 1999 and September 30, 2000 (the
‘intake period’) were enrolled in the study. For each
patient, the date of the first antipsychotic prescription
filled (dispensed) during the intake period was
designated the ‘index prescription date’ (IPD), and the
medication initiated on the IPD was designated the
‘index drug.’ Upon identification, all computerized
records for medical and pharmacy services were
collected for 6 months before, and 12 months after the
IPD for this population (total study period spanned
from January 1, 1999 to September 30, 2001). Figure 1
presents patient selection criteria.
All patients who received an antipsychotic medication
during the 6-month period prior to their IPD were
excluded. Patients were then assigned to either an
atypical drug class cohort or a typical cohort, based on
the index drug prescribed. All patients who did not have
continuous plan enrollment for the 18-month study
period (6 months pre-IPD and 12 months post-IPD)
were excluded, as were those whose data did not
include reimbursement for mental health (behavioral
benefit carve-out), and those who had multiple index
medications. The remaining patients were included in
the analysis, and were used to determine the differences
in antipsychotic and concomitant medication use
patterns between the two groups.
620Healthcare Utilization and Patient Adherence to Antipsychotic Medications © 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7)
Figure 1. Summary of patient selection
Population = 497 000
Members who had a prescription for at least one
antipsychotic medication between 7/1/99 and 9/30/00
(total prescriptions amount to 14 277)
n = 2710
1081 members were excluded as they had at least one
prescription in the preceding 6 months
(1/1/99 to 6/30/99)
n = 1629
295 members were excluded as age was
< 6 or > 65 years
n = 1334
32 members were excluded because they had a
behavioral health benefit carve-out
n = 1302
833 members were excluded because they were not
continuously enrolled for 18 months (6 months before
and 12 months after the IPD) (n = 828), or because they
had multiple index drugs (n = 5)
n =
469
Data
Health insurance pharmacy data were used to examine
treatment patterns of antipsychotic agents as well as the
concomitant usage of mood stabilizers, antidepressants,
anxiolytics, sedative-hypnotics, and medications to treat
extrapyramidal symptoms, Alzheimer’s disease,
Parkinson’s disease, and attention deficit/hyperactivity
disorder.
Post-IPD antipsychotic use was examined for dose
adjustment, augmentation, switching, and mixed
therapy patterns. Dose adjustment was defined as an
increase or decrease in average daily dose of the index
drug of more than 30%. Therapy augmentation was
defined as the addition of a second antipsychotic
medication to the index drug if the second agent was
added at least 30 days after the IPD, and both
medications were filled at least one additional time after
the date the second antipsychotic agent was added.
Therapy switching was defined as the discontinuation of
the index drug and the fill of another antipsychotic
agent. Mixed therapy included cases where multiple
modifications to therapy occurred.
Length of antipsychotic pharmacotherapy was
defined as the difference in days between first and last
fill dates, plus last days supplied. The total days supplied
equaled the length of therapy for those with only one
fill. The maximum length of therapy for this study was
365 days; patients who received more than 365 days of
therapy had their records modified to reflect this.
Adherence to therapy for this study was quantified as a
medication possession ratio (MPR). Each patient’s MPR
was calculated by dividing the total days supplied (the
numerator) by a follow-up period of 365 days (the
denominator)22. Any subject who had an MPR greater
than one had their records modified to reflect a
maximum value of one.
The medical data were used to measure the utilization
of office-based outpatient, hospital inpatient and
outpatient, emergency room, and mental health
(psychiatric) visits for each patient. Patient diagnoses
were identified as outlined in the International
Classification of Diseases (9th revision) Clinical
Modification (ICD-9-CM codes)23. All ICD-9-CM
codes from 290.xx to 319.xx (mental disorders) and
331.xx to 333.xx (Alzheimer’s and Parkinson’s diseases)
were identified from the health insurance data. The
numbers of visits for each resource utilization category
were recorded for the 6-month pre-IPD period as well
as during the 12 months following the IPD, and were
separated among the typical and atypical cohorts. For
visits occurring during the 12-month post-IPD period,
the data were presented in two forms: unadjusted,
which did not account for variables such as age, gender,
MPR, length of therapy, and treatment augments,
switches, or mixed therapy; and adjusted, which did
account for the above confounding variables by utilizing
multivariable regression techniques.
All healthcare utilization visits were calculated as
rates per 1000 patients per month (P1000PPM).
Unadjusted and adjusted incident rate ratios with their
respective 95% confidence intervals (CI) were
calculated for each utilization category. Psychiatric visits
included inpatient or residential treatment facility stays,
and visits to partial hospitalization programs or
community mental health centers.
Statistical Analysis
For each antipsychotic group, means ±standard deviation
(SD) (for continuous variables) and percentages (for
categorical variables) were reported for the set of potential
predictors of outcomes. To test for differences between
treatment groups, independent t tests were used for the
continuous variables, and χ2tests (or Fisher’s exact tests
for cells less than 5) were used for the categorical
variables. Service use counts (the outcome variables
defined below by regression models 1 through 4) were
expressed as numbers of visits P1000PPM. Crude incident
densities were compared by calculating incident rate ratios
and their associated 95% confidence interval (CI) using
the exact method.
An appropriate regression model for count data often
follows a Poisson distribution or one of its variants. One of
the rarely met assumptions of a Poisson model is that the
mean must equal the variance. When the conditional
variance is greater than the mean, overdispersion may
occur. An overdispersed Poisson model produces incorrect
variance estimates which are biased downwards. When
this occurs, a negative binomial model, which does not
constrain the conditional variance to equal the mean, is
preferred over a Poisson model24–26. Since the likelihood
ratio tests (for each of the four adjusted regression
models) indicated that our data were significantly
overdispersed ( p<0.001), negative binomial models
were utilized in a generalized linear model framework
using negative binomial as the family and logarithm as the
canonical link27. The main independent variable in all
negative binomial regression models was antipsychotic
drug class. Model fit statistics, assessed using deviance
residuals, revealed minor deviations of the independent,
identically distributed (iid) errors assumption. Hence,
robust standard errors, which provide efficient estimates
without relying on the strong and sometimes implausible
iid errors assumption, were calculated using the
Huber/White sandwich variance estimator28,29. Incident
rate ratios (IRR) were derived by exponentiation of
parameter estimates from the models.
Four separate models were estimated to predict the
following outcome variables: (1) office-based outpatient
© 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7) Healthcare Utilization and Patient Adherence to Antipsychotic Medications Al-Zakwani et al. 621
utilization; (2) hospital-based outpatient utilization; (3)
inpatient admission; and (4) emergency room
utilization. Model covariates included drug class at IPD,
age, gender, MPR, length of therapy, use of
antidepressants and mood stabilizers, and respective
medical care use at baseline (e.g. previous use of
emergency room for model 4). Baseline resource use
was based on utilization during the 6-month pre-IPD
period. For all models, an a priori two-tailed level of
significance was set at the 0.05 level.
Statistical analyses were performed using STATA
version 7.030. Patient identity was masked throughout in
a limited data set format, in line with US privacy and
confidentiality regulations.
Results
A total of 469 patients met the study criteria, and their
data were used to analyze the antipsychotic and
concomitant use patterns between the two groups.
Atypical and typical antipsychotics were prescribed to
384 (81.9%) and 85 (18.1%) patients, respectively.
Mean age at initiation of antipsychotic therapy was
significantly higher among patients receiving typical
therapy than atypical therapy (43 ±15 years vs 34 ±15
years, respectively; p< 0.001). There were no
significant gender differences between cohorts (59%
male vs 48% male for typical and atypical agents,
respectively; p= 0.069).
Table 1 presents the frequencies and percentage
distribution of diagnoses upon initiation of antipsychotic
therapy, as well as their associated ICD-9-CM codes. The
three most common diagnoses at index were psychosis
disorders (ICD-9-CM = 296.xx; n= 249; 53.1%);
neurotic disorders (ICD-9-CM = 300.xx; n= 173;
36.9%); and depressive disorders (ICD-9-CM =
311.xx; n=147; 31.3%). Please note that the diagnosis
breakdown is not mutually exclusive among patients as
some patients had multiple diagnoses.
Ninety per cent (n= 422) of patients received at
least one concomitant medication over the one-year
observation period. A majority (n= 355, 75.7%) of the
patients received concomitant prescriptions for
antidepressants (Table 2). Mood stabilizers were
prescribed concurrently for 42.2% (n= 198) of the
patients. In general, patients receiving atypical
antipsychotics used concomitant medications with
greater frequency than those patients prescribed typical
antipsychotics. However, the only concomitant drug
types to be significantly associated with the atypical
cohort were mood stabilizers (44.5% vs 31.8%;
p= 0.031) and antidepressants (81.0% vs 51.8%;
p= 0.001).
Observed post-IPD antipsychotic therapy
modification patterns, including dose adjustments,
treatment augmentations, switches, and mixed
therapies, are outlined in Table 3. Two patients were
removed from the dose adjustment analysis because it
was not possible to elicit whether these patients had
dose changes or whether they were taking two different
strengths concurrently. Approximately 30% of patients
(n= 142) had at least one dose adjustment. The
number of dose adjustments per patient ranged from
one to five with a mean of 1.7 ±1.0 . The atypical group
had a significantly higher percentage of dose
adjustments than the typical group (33.5% vs 16.5%;
p= 0.002).
Three patients in the atypical cohort (0.8%) and one
patient in the typical cohort (1.2%) had a treatment
augmentation. A total of 81 patients (17.3%) had
treatment switches. The majority of the switches
(n= 56, 11.9%) were from one atypical agent to
622 Healthcare Utilization and Patient Adherence to Antipsychotic Medications © 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7)
Table 1. Frequency and percentage distribution of diagnosis upon initiation of antipsychotic therapy
ICD-9-CM code Diagnosis Number of patients* Percentage of patients†
290.xx – 299.xx Psychosis disorders 372 79.3
300.xx – 302.xx Neurotic, personality and sexual
disorders 199 42.4
303.xx – 305.xx Drug and alcohol dependence
syndromes 105 22.4
306.xx – 310, 312.xx Psychological malfunction arising
from mental disorders 97 20.7
311.xx Depressive disorder 147 31.3
313.xx – 315.xx Childhood and adolescence
emotional disturbance and
developmental delays
74 15.8
317.xx – 319.xx, 331.xx – 333.xx Mental retardation, Alzheimer’s
and Parkinson’s diseases
14 3.0
*Numbers of patients will not add up to our total cohort (n = 469) as some patients had more than one diagnosis
†Percentages of patients will not add up to 100% because some patients had more than one diagnosis
another. Fourteen patients (3.0%) had their treatment
switched from an atypical antipsychotic agent to a
typical medication, while eleven patients (2.4%) had
their treatments switched from typical agents to
atypical antipsychotic medications. Seven patients had
mixed therapy; all but one of these patients were on
atypical agents. These patients had one fill of an
additional antipsychotic agent during treatment with
the index drug.
Length of therapy and MPR analyses were performed
on the cohort that did not have switches, augments or
mixed therapy (n= 377). Mean length of therapy
for the atypical cohort was significantly longer than
for those in the typical group (136 days; [95% CI
121–150] vs 80 days; [95% CI: 53 to 106]; p= 0.001).
The atypical cohort was also significantly more ad-
herent with therapy than the typical cohort (mean
MPR = 0.53; [95% CI: 0.48 to 0.57] vs 0.24; [95% CI
0.15–0.32]; p= 0.001), where a lower MPR represents
lower adherence.
Only those patients who did not have switches,
augments, or mixed therapy (n= 377) were included in
the examination of medical care use patterns (Table 4).
The numbers of office-based outpatient visits
P1000PPM were 2635 visits for the atypical cohort and
4249 visits for the typical group (unadjusted IRR =
0.62; [95% CI 0.57–0.67]). After controlling for age,
gender, baseline office visit use, length of therapy, MPR,
and whether the patient was prescribed antidepressants
or mood stabilizers, the atypical cohort experienced
45% (95% CI 17–64%) fewer office visits compared to
the typical cohort. Furthermore, results from the
regression model suggest that for every 10% increase
in MPR across cohorts, there was a 17.6% (95% CI
14.9–20.3%) decrease in outpatient visits during the 12-
month observation period.
© 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7) Healthcare Utilization and Patient Adherence to Antipsychotic Medications Al-Zakwani et al. 623
Table 2. Concomitant medication use
Antipsychotic drug EPS Alzheimer’s
disease
Parkinson’s
disease
Mood
stabilizers
Anti-
depressants
Anxiolytics Sedatives/
hypnotics
ADHD
Typical antipsychotics (n)
chlorpromazine (33) 1 0 0 8 12 11 10 4
haloperidol (16) 4 0 0 4 8 9 2 4
thioridazine (13) 2 0 0 6 9 3 2 3
perphenazine (13) 4 0 0 5 10 8 5 0
other† (10) 1 0 0 4 5 2 0 0
Sub-total, n = 85 (%) 12 (14.1) 0 (0) 0 (0) 27 (31.8)* 44 (51.8)* 33 (38.8) 19 (22.4) 11 (12.9)
Atypical antipsychotics
risperidone (205) 17 3 5 89 170 78 36 53
olanzapine (135) 19 2 2 67 101 60 25 18
quetiapine (41) 2 0 0 15 39 23 8 5
clozapine (3) 1 0 0 0 1 0 1 1
Sub-total, n = 384 (%) 39 (10.2) 5 (1.3) 7 (1.8) 171 (44.5) 311 (81.0) 161 (41.9) 70 (18.2) 77 (20.1)
Total, n = 469 (%) 51 (10.9) 5 (1.1) 7 (1.5) 198 (42.2) 355 (75.7) 194 (41.4) 89 (19.0) 88 (18.8)
*Significant below 0.05 level as compared to atypical cohort
†Other category includes fluphenazine, loxapine, trifluoperazine, and thiothixene
EPS, extrapyramidal symptom; ADHD, attention deficit/hyperactive disorder
Table 3. Antipsychotic therapy modification summary
Index antipsychotic prescription Dose adjustments Treatment augmentation Treatment switch Mixed therapy
Typical antipsychotics
chlorpromazine 3 0 1 1
haloperidol 1 1 2 0
thioridazine 2 0 4 0
perphenazine 5 0 1 0
other† 3 0 3 0
Sub-total, n = 85 (%) 14 (16.5)* 1 (1.2) 11 (12.9) 1 (1.2)
Atypical antipsychotics
risperidone 74 1 33 2
olanzapine 44 1 31 3
quetiapine 10 1 6 1
clozapine 0 0 0 0
Sub-total, n = 384 (%) 128 (33.5) 3 (0.8) 70 (18.2) 6 (1.6)
Total, n = 469 (%) 142 (30.4) 4 (0.9) 81 (17.3) 7 (1.5)
*Significant below 0.05 level as compared to atypical cohort
†Other category includes fluphenazine, loxapine, trifluoperazine, and thiothixene
Atypical users had fewer inpatient hospital admissions
compared to the typical cohort (197 vs 511 P1000PPM;
IRR = 0.39; [95% CI 0.30–0.49]. After adjustment, the
atypical cohort was associated with 76% (95% CI
11–94%) fewer hospital admissions compared to the
typical cohort. Increased adherence also had a
statistically significant inverse effect on hospital
admissions; a 10% increase in MPR resulted in an 18.3%
(95% CI 12.1–24.1%) decrease in admissions. The
atypical cohort also had fewer hospital-based outpatient
visits compared to the typical group, and the difference
was also statistically significant ( p< 0.001). Across
cohorts, an increase of 10% in the MPR again led to an
18.7% (95% CI 13.5–23.7%) reduction in hospital-
based outpatient visits.
Members of the atypical cohort used the emergency
room less frequently than typical cohort patients (125
vs 354 P1000PPM; IRR = 0.35; [95% CI 0.26–0.48]).
After adjustment, the atypical cohort was associated
with 67% (95% CI 33%–84%) fewer emergency room
visits compared to the typical cohort. Again, a 10%
increase in MPR translated into a 20.4% (95% CI
15.6–25.0%) decrease in emergency room utilization.
Atypical medication users had fewer psychiatric visits
compared to the typical cohort (16 vs 28 P1000PPM;
IRR = 0.57; [95% CI 0.21–1.93]), but this association
was not statistically significant ( p=0.275). Further
multivariable adjustment analysis was not performed
due the small sample size (44 P1000PPM).
Discussion
High acquisition costs of atypical antipsychotics have
continued to raise concerns in the healthcare industry
despite their beneficial clinical effects, lower EPS side-
effect profile, and improved humanistic outcomes. The
average wholesale price (AWP) of a 30-day supply of the
typical agent haloperidol 5 mg is under $5, while the
AWP for the same 30-day supply of the atypical agent
olanzapine 5 mg ranges from $150 to $20031. In spite of
this, the clinical and adherence-related benefits that
atypical agents offer may potentially offset acquisition
costs, making them efficient, cost-effective options
when compared to typical therapy14,32. This may be
observed if relapse with rehospitalization or other
healthcare utilization occurs at differential rates33,34.
In this study, patients using atypical antipsychotic
agents were shown to have a lower hospital admission
rate over a 1-year observation period compared to
typical antipsychotic medication patients. Furthermore,
we observed lower rates of office-based outpatient and
emergency room visits in the atypical cohort. These
findings are consistent with a number of other
investigations32,34–41. Given our findings of higher MPRs
among atypical antipsychotic users compared to typical
antipsychotic users, our data showed an association
between the reduced rates of medical care use and
higher MPR in the atypical cohort. To our knowledge,
this relationship has not been previously addressed.
Our study also showed that 90% of patients received
at least one concomitant medication over the 1-year
observation period, with patients receiving atypical
antipsychotics generally using more concomitant
medications than patients using typical antipsychotics.
The usage of mood stabilizers and antidepressants as
concomitant medications was significantly associated
with the atypical cohort. This association could have
contributed to the lower utilization of healthcare
resources of atypical agents; however, as this was not a
primary objective of our study, further research is
needed to evaluate this relationship.
This study has a number of limitations. First, this
analysis included therapy augments, switches, and
concomitant therapy, but the physicians’ justification for
altering a patient’s therapy was not available due to the
study’s retrospective nature. Due to the nature of health
insurance data, selection bias may exist. Although these
results suggest greater effectiveness of atypical over
typical agents, our study included a limited number of
control variables. While the strength of health insurance
624 Healthcare Utilization and Patient Adherence to Antipsychotic Medications © 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7)
Table 4. Healthcare resource use pattern of the two antipsychotic cohorts of patients who did not have treatment augments,
switches or mixed therapy (based on 1-year data post index drug) (n = 377)
Number of visits P1000PPM Incident rate ratios (95% CI)†
Service, (number of visits)
Atypical Typical Unadjusted Adjusted‡
Office visits (4370) 2635 4249 0.62 (0.57–0.67)* 0.55 (0.36–0.83)*
Hospital outpatient visits (534) 307 634 0.48 (0.39–0.60)* 0.59 (0.29–1.18)
Hospital visits (361) 197 511 0.39 (0.30–0.49)* 0.24 (0.06–0.89)*
Emergency room visits (235) 125 354 0.35 (0.26–0.48)* 0.33 (0.16–0.67)*
Psychiatric visits** (27) 16 28 0.57 (0.21–1.93) —
*Significant below 0.05 level
**The adjusted incident rate ratio for psychiatric visits was not performed due to the small sample size
†95% confidence interval
‡This analysis adjusted for age, gender, MPR, length of therapy, use of antidepressant and mood stabilizers, and respective medical care use
at baseline
data lies in its ‘real-world’ nature, the exclusion of clinical
confounders, especially measures of disease severity
associated with both antipsychotic use and subsequent
resource use, can bias estimates of effectiveness. Another
limitation was that medical costs could not be analyzed
since mental health-related costs were capitated. It is also
important to acknowledge that MPR was derived from
prescription refills, which is not without methodological
flaws42. Furthermore, the results can only be generalized
to the sample population from which they were drawn.
Conclusion
Atypical antipsychotic users were significantly more
adherent to therapy and were shown to have lower rates
of office, hospital and emergency room utilization.
Within the context of inherent limitations associated
with health insurance data, this study suggests that there
is a relationship across cohorts between medication
adherence and use of healthcare resources. Additional
research is necessary to estimate costs specifically
attributable to atypical antipsychotic drug adherence.
Economic evaluations that include explicit (and possibly
standardized) variables representing adherence between
drug classes, subsequently linked to a cost-effectiveness
analysis, would be valuable additions to the existing body
of research on antipsychotic drug adherence and its
effect on rates of resource use and relapse.
Acknowledgements
The authors would like to thank Joshua J. Spooner,
PharmD, of Health Core, for his assistance with this
manuscript, and Matthew C. Wood, of Health Core, for
his assistance with data management. This project was
presented as a poster presentation at the International
Society for Pharmacoeconomics and Outcomes
Reasearch (ISPOR) international meeting in Arlington,
VA: 19-21 May 2003. This project was funded in its
entirety by Novartis Pharmaceuticals.
References
1. Kessler RC, Berglund PA, Zhao S, et al. The 12-month preva-
lence and correlates of serious mental illness. In Manderscheid
RW, Sonnenschein MA, eds. Mental Health, United State.
Washington DC: US Government Printing Office, 1996:59-70
2. United States Department of Health and Human Services.
National Institute of Mental Health. The epidemiology of mental
disorders: results from the Epidemiologic Catchment Area Study
(ECA). Psychopharmacol Bull 1982;18:222-5
3. Regier DA, Narrow WE, Rae DS, et al. The de facto mental and
addictive disorders service system. Epidemiologic Catchment
Area prospective 1-year prevalence rates of disorders and servic-
es. Arch Gen Psychiatry 1993;50:85-94
4. Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and 12-
month prevalence of DSM-III-R psychiatric disorders in the
United States. Results from the National Comorbidity Survey.
Arch Gen Psychiatry 1994;51:8-19
5. Satcher D. Mental health: a Report of the Surgeon General.
Rockville, MD: National Institute of Mental Health, 1999
6. Rice DP, Miller LS. The economic burden of schizophrenia: con-
ceptual and methodological issues, and cost estimates. In
Moscarelli M, Rupp A, Santorius N, eds. Schizophrenia.
Handbook of Mental Health Economics and Health Policy, Vol.
1. London: John Wiley & Sons, 1996:321-34
7. Mark T, McKusick D, King E, et al. National expenditures for
mental health, alcohol, and other drug abuse treatment, 1996.
1998. Rockville, MD: Substance Abuse and Mental Health
Services Administration, 1998
8. Worrel JA, Marken PA, Beckman SE, et al. Atypical antipsychotic
agents: a critical review. Am J Health Syst Pharm 2000;57:238-55
9. Revicki DA. Cost effectiveness of the newer atypical antipsy-
chotics: a review of the pharmacoeconomic research evidence
Curr Opin Invest Drugs 2001;2:110-17
10. Revicki DA. The new atypical antipsychotics: a review of pharma-
coeconomic studies Expert Opin Pharmacother 2000;1:249-60
11. Chouinard G. Effects of risperidone in tardive dyskinesia: an
analysis of the Canadian multicenter risperidone study. J Clin
Psychopharmacol 1995;15:36-44S
12. Albright PS, Livingstone S, Keegan DL, et al. Reduction of
healthcare resource utilization and costs following the use of
risperidone for patients with schizophrenia previously treated
with standard antipsychotic therapy: a retrospective analysis
using Saskatchewan health linkable database. Clin Drug Invest
1996;11:289-99
13. Addington DE, Jones B, Bloom D, et al. Reduction of hospital
days in chronic schizophrenic patients treated with risperidone: a
retrospective study. Clin Ther 1993;15:917-26
14. Menzin J, Boulanger L, Friedman M, et al. Treatment patterns
and medication adherence between users of conventional
and atypical antipsychotics in clinical practice: the Florida
Medicaid experience. National Institute of Mental Health:
NCDEU Poster Abstracts. Billerica, MA: Boston Health
Economics, 2000
15. Delgado L, McKoy Y, Rey JA, et al. A comparison of clinical out-
comes in patients receiving depot vs. atypical antipsychotics.
National Institute of Mental Health: NCDEU Poster Abstracts.
Ft. Lauderdale, FL: Nova Southeastern University - College of
Pharmacy, 2000
16. Csernansky JG, Schuchart EK. Relapse and rehospitalization
rates in patients with schizophrenia. Effects of second generation
antipsychotics. CNS Drugs 2002;16:473-84
17. Masukopf J, Muroff M, Gibson PJ, et al. Estimating the costs and
benefits of new drug therapies: atypical antipsychotic drugs for
schizophrenics. Schizophr Bull 2002;28:619-35
18. Richie CW, Chiu E, Harrigan S, et al. The impact upon extra-
pyramidal side effects, clinical symptoms and quality of life of a
switch from conventional to atypical antipsychotics (risperidone
or olanzapine) in elderly patients with schizophrenia. Int J
Geriatr Psychiatry 2003;18:432-40
19. Dolder CR, Lacro JP, Dunn LB, et al. Antipsychotic medication
adherence: is there a difference between typical and atypical
agents? Am J Psychiatry 2002;159:103-8
20. Olfson M, Mechanic D, Hansell S, et al. Predicting medication
noncompliance after hospital discharge among patients with
schizophrenia. Psychiatr Serv 2000;51:216-22
21. Cabeza IG, Amador MS, Lopez CA, et al. Subjective response to
antipsychotics in schizophrenia patients: clinical implications and
related factors. Schizophr Res 2000;41:349-55
22. Schectman JM, Bovbjerg VE, Voss J.D Predictors of medication-
refill adherence in an indigent rural population. Med Care
2002;40:1294-1300
© 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7) Healthcare Utilization and Patient Adherence to Antipsychotic Medications Al-Zakwani et al. 625
23. International Classification of Diseases, 9th revision. Clinical
Modification, 6th edn. 2001. Los Angeles, CA: Practice
Management Information Corporation, 2000
24. Moulton BR. Random group effects and the precision of regres-
sion estimates. J Econometrics 1986;32:385-97
25. Gardner W, Mulvey EP, Shaw EC. Regression analyses of counts
and rates: Poisson, overdispersed Poisson, and negative binomial
models. Psychol Bull 1995;118:392-404
26. Long JS. Regression Models for Categorical and Limited Dependent
Variables. Thousand Oaks, CA: Sage Publications, 1997
27. McCullagh P, Nedler JA. Generalized Linear Models, 2nd edn.
New York: Chapman and Hall, 1989
28. White H. A heteroskedasticity-consistent covariance matrix esti-
mator and a direct test for heteroskedasticity. Econometrica
1980;48:817-38
29. Huber PJ. The behavior of maximum likelihood estimates under
nonstandard conditions. In Proceedings of the Fifth Berkeley
Symposium on Mathematical Statistics and Probability, Vol. 1.
Berkeley, CA: University of California Press, 1967,221-3
30. STATA Corporation (Version 8.0). College Station, TX: STATA
Corporation
31. Reents S. Haloperidol/olanzapine: cost. Clinical Pharmacology
2000, Gold Standard Multimedia, Inc. Accessed August 8, 2002.
URL: http://cp.gsm.com
32. Rosenheck R, Cramer J, Xu W, et al. Multiple outcome assess-
ment in a study of the cost-effectiveness of clozapine in the
treatment of refractory schizophrenia. Health Serv Res
1998;33:1237-61
33. Svarstad BL, Shireman TI, Sweeney JK. Using drug claims data
to assess the relationship of medication adherence with hospital-
ization and costs. Psychiatr Serv 2001;52:805-11
34. Ghaemi SN, Kirkwood CK, Sambur MR, et al. Economic out-
comes of risperidone in comparison to typical neuroleptic agents
for treatment-resistant psychosis: a community-based study. J
Pharm Technol 2001;17:273-8
35. Reid WH, Mason M. Psychiatric hospital utilization in patients
treated with clozapine for up to 4.5 years in a state mental health
care system. J Clin Psychiatry 1998;59:189-94
36. Gianfrancesco F, Durkin MB, Mahmoud R, et al. Use of health-
care services by patients treated with risperidone versus conven-
tional antipsychotic agents. Pharmacoeconomics 2002;20:413-27
37. Galvin PM, Knezek LD, Rush AJ, et al. Clinical and economic
impact of newer versus older antipsychotic medications in a com-
munity mental health center. Clin Ther 1999;21:1105-16
38. Haywood TW, Kravitz HM, Grossman LS, et al. Predicting the
‘revolving door’ phenomenon among patients with schizophrenic,
schizoaffective, and affective disorders. Am J Psychiatry
1995;152:856-61
39. Weiden PJ, Olfson M. Cost of relapse in schizophrenia. Schizophr
Bull 1995;21:419-29
40. Sullivan G, Wells KB, Morgenstern H, et al. Identifying modifi-
able risk factors for rehospitalization: a case-control study of seri-
ously mentally ill persons in Mississippi. Am J Psychiatry
1995;152:1749-56
41. Glazer WM, Ereshefsky L. A pharmacoeconomic model of out-
patient antipsychotic therapy in ‘revolving door’ schizophrenic
patients. J Clin Psychiatry 1996;57:337-45
42. Steiner JF, Prochazka AV. The assessment of refill compliance
using pharmacy records: methods, validity, and applications. J
Clin Epidemiol 1997;50:105-16
626 Healthcare Utilization and Patient Adherence to Antipsychotic Medications © 2003 LIBRAPHARM LTD – Curr Med Res Opin 2003; 19(7)
CrossRef links are available in the online published version of this paper:
http://www.cmrojournal.com
Paper CMRO-2395, Accepted for publication: 25 July 2003
Published Online: 00 MMM 2003
doi:10.1185/030079903125002289
