Cost-Effectiveness of Atypical
Antipsychotics: Implications for
Practice
Robert Rosenheck MD
New England MIRECC
Yale Medical School
April 21, 2011
VA Cooperative Study #451:
Olanzapine vs. haloperidol




Prospective 12 month double-blind trial
N=309 veterans
17 VA medical centers
Assesed multiple outcomes





Symptoms
Quality of Life
Cognitive functioning
Health service use
VA and Societal Costs
Figure 2: Compliance with Double Blind Study Drug
(log rank test p=.25)
Cummulative % Remaining in Study
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
50
100
150
Days
200
250
300
Figure 3. PANSS Total Symptom Score:
Intention-to-treat
PANSS Total
Score
100
75
50
25
0
0
1.5
3
Mixed model
analysis: ns
6
9
12
Months
Olanzapine
Haloperidol
Olz. N:
159
121
108
93
90
92
Hal. N:
150
115
105
94
83
85
VACS #451 Results(1)

Patients assigned to olanzapine had:




slightly less akathisia and
superior cognitive functioning (small effect <0.2 -not enough to affect quality of life);
Olanzapine showed no advantage in:
 Symptoms
 Quality of life
 Parkinsonian (EPS) side effects
Olanzapine incurred greater weight gain and
greater VA costs ($3,000 - $9,000/pt/year).
Cost Utility of the Latest
Antipsychotic Drugs in Schizophrenia
Study (CUtLASS)(S Lewis et al. 2005)







RCT conducted in the UK for the NHS Health
Tecnhnology Assessment Program
Recruited: Aug 1999 – April 2002
Randomly assigned patients to “doctor’s
choice” of any SGA other than clozpaine or
FGA (n=227)
Blinded assessments.
59% stayed on original drug for 1 year
81% were followed-up at 1 year
Primary outcome was Quality of Life Scale
(Heinrichs/Carpenter)
CUtLASS Results:No
significant differences on…










% compl. on original study drug (FGA<SGA)
Quality of Life (QOLS) (FGA>SGA)
Symptoms (PANSS) (FGA<SGA)
Global Assessment (GAF) (SGA>FGA)
Depression (Calgary) (FGA < SGA)
Compliance (FGA=SGA)
EPS (Simpson Angus) (FGA>SGA)
Akathisia (Barnes) (FGA>SGA)
TD (AIMS) (FGA>SGA)
Health costs (even without drugs)(FGA<SGA)
Estimate of difference
in QLS after imputation
of missing data, with
95% c.i. of difference
Hypothesis of 5
point advantage
for SGA excluded
Observed
-6
-5
-4
-3
Favors FGA
-2
Expected
-1
0
1
2
3
Equivalence
4
5
6
Favors SGA
Difference in QLS scores
Data from Table 3b
CATIE Schizophrenia Trial Design
Phase 2
Phase 1*
Double-blind, random
treatment assignment.
Participants who discontinue
Phase 1 choose either the
clozapine or the ziprasidone
randomization pathways
OLANZAPINE
CLOZAPINE
(open-label)
QUETIAPINE
1460 patients
with SCZ
Comorbidity
Other meds
R
R
OLANZAPINE,
QUETIAPINE or
RISPERIDONE
RISPERIDONE
Phase 3
Participants who discontinue
Phase 2 choose one of the
following open-label
treatments
•ARIPIPRAZOLE
•CLOZAPINE
•FLUPHENAZINE
DECANOATE
•OLANZAPINE
•PERPHENAZINE
ZIPRASIDONE
•QUETIAPINE
ZIPRASIDONE
R
OLANZAPINE,
QUETIAPINE or
RISPERIDONE
•RISPERIDONE
•ZIPRASIDONE
PERPHENAZINE
No one assigned to same
drug as in Phase 1
•2 of the antipsychotics
above
*Phase 1A: participants with TD do not get randomized to perphenazine; phase 1B: participants who fail perphenazine
will be randomized to an atypical (olanzapine, quetiapine, or risperidone) before they are eligible for phase 2.
Stroup TS et al. Schizophr Bull. 2003;29:15-31.
Proportion of Patients without Event
Time to Discontinuation for Any Reason
1
Overall p-value = 0.004*
0.8
0.6
0.4
0.2
0
P<0.001 for olanzapine vs quetiapine
P=0.002 for olanzapine vs risperidone
0
3
6
9
12
15
Time to Discontinuation for Any Cause (mo)
Olanzapine
Perphenazine
Quetiapine
Risperidone
18
Ziprasidone
OLZ
(n=330)
QUET
(n=329)
RISP
(n=333)
PER
(n=257)
ZPR
(n=183)
Discontinued
210 (64%)
269 (82%)
245 (74%)
192 (75%)
145 (79%)
Kaplan-Meier Median (mos)
[95%CI]
9.2
[6.9, 12.1]
4.6
[3.9, 5.5]
4.8
[4.0, 6.1]
5.6
[4.5, 6.3]
3.5
[3.1, 5.4]
Hazard ratios for Olanzapine
---
0.63
< 0.001*
0.78
0.021
0.76
0.028
0.75
0.002*
No advantage of any SGA over
perphenazine on…







Symptoms (Rosenheck et al., 2006)
Neurocognition (Keefe et al., 2007)
Violent behvaior (Swanson et al., 2008)
Employment (Resnick et al., 2008)
Community functioning (Swartz et al.,
2006)
Family Burden (Perlick et al, in press)
EPS/TD/Akithesia (Miller et al., 2008)
CATIE Cost-Effectiveness
Results (1)
COST
Service Use and Cost Measures

Service Use (Service Use and
Resources Form [SURF])

Outpatient



Inpatient







Mental health
Medical
Mental Health
Substance Abuse
Medical

Cost






Nursing home
Residential
Medication records
Criminal justice, public
support, productivity
Outpatient
All residential
Inpatient
All health care
Experimental medications



2003 Medicaid Discount
rates and mandated
company rebates
VA discount (40%)
Sensitivity analysis of price
discounts (as funded vs
Medicaid vs. VA).
Ancillary medication
(discounted cost to privately
insured Market Scan ®
patients).
Monthly Costs: All medication costs (experimental drugs
and concomitant medications)(actual prices, with
discounts)(mean=$499/month)(ITT)
$700
$500
$200/mo $2,400/yr
$400
$300
$200
$100
OLANZAPINE
QUETIAPINE
ZIPRASIDONE
PERPHENAZINE
RISPERIDONE
m
o.
18
m
o.
17
m
o.
16
m
o.
15
m
o.
14
m
o.
13
m
o.
12
m
o.
11
m
o.
10
m
o.
9
m
o.
8
m
o.
7
m
o.
6
m
o.
5
m
o.
4
m
o.
3
m
o.
2
m
o.
1
se
lin
e
$0
Ba
Monthly expenditures
$600
Monthly service use: All inpatient days (mental health
and medical/surgical)(mean=0.70/month)(ITT)(p=ns)
Monthly inpatient days
3
2
1
Olanzapine
Quetiapine
Ziprasidone
Perphenazine
Risperidone
m
o.
18
m
o.
17
m
o.
16
m
o.
15
m
o.
14
m
o.
13
m
o.
12
m
o.
11
m
o.
10
m
o.
9
m
o.
8
m
o.
7
m
o.
6
m
o.
5
m
o.
4
m
o.
3
m
o.
2
m
o.
1
Ba
se
lin
e
0
Monthly Costs: All inpatient (mental health and
medical/surgical)costs (mean=$363/month)(ITT)(p=ns)
$1,500
$1,000
$500
Olanzapine
Quetiapine
Ziprasidone
Perphenazine
Risperidone
m
o.
18
m
o.
17
m
o.
16
m
o.
15
m
o.
14
m
o.
13
m
o.
12
m
o.
11
m
o.
10
m
o.
9
m
o.
8
m
o.
7
m
o.
6
m
o.
5
m
o.
4
m
o.
3
m
o.
2
m
o.
1
se
lin
e
$0
Ba
Monthly expenditures
$2,000
Monthly Costs: All outpatient (mental health and
medical/surgical)costs (mean=$372/month)(ITT)(p=ns)
$700
$500
$400
$300
$200
$100
Olanzapine
Quetiapine
Ziprasidone
Perphenazine
Risperidone
m
o.
18
m
o.
17
m
o.
16
m
o.
15
m
o.
14
m
o.
13
m
o.
12
m
o.
11
m
o.
10
m
o.
9
m
o.
8
m
o.
7
m
o.
6
m
o.
5
m
o.
4
m
o.
3
m
o.
2
m
o.
1
se
lin
e
$0
Ba
Monthly expenditures
$600
Monthly health costs (IP and OP) excluding
medications (mean=$1,047/month)(ITT)(p=ns)
$2,100
$1,400
$700
Olanzapine
Quetiapine
Ziprasidone
Perphenazine
Risperidone
m
o.
18
m
o.
17
m
o.
16
m
o.
15
m
o.
14
m
o.
13
m
o.
12
m
o.
11
m
o.
10
m
o.
9
m
o.
8
m
o.
7
m
o.
6
m
o.
5
m
o.
4
m
o.
3
m
o.
2
m
o.
1
se
lin
e
$0
Ba
Monthly expenditures
$2,800
Monthly Costs: All healthcare costs including
medications (mean=$1,544/month)(actual drug
prices with discounts)(ITT)(p=P<O,Q,R,P)(APA).
Monthly expenditures
$2,800
$2,100
$1,400
`
$700
Olanzapine
Quetiapine
Ziprasidone
Perphenazine
Risperidone
m
o.
18
m
o.
17
m
o.
16
m
o.
15
m
o.
14
m
o.
13
m
o.
12
m
o.
11
m
o.
10
m
o.
9
m
o.
8
m
o.
7
m
o.
6
m
o.
5
m
o.
4
m
o.
3
m
o.
2
m
o.
1
Ba
se
lin
e
$0
Cost Summary





In CATIE average total health care costs
were 20%-30% lower for perphenazine than
for second generation antipsychotics,
because of lower drug cost.
$300-$500/month
$3,600-$6,000/year
Cost-effectiveness question: Are benefits
(quality of life) or avoidance of risks (TD)
worth this public expenditure?
Magnitude of cost should = magnitude of
benefit
CATIE Cost-Effectiveness (2)
EFFECTIVENESS
PANSS TOTAL SCORE (LS Means from Mixed Models
(Adj for site, baseline, exacerbation)
80
PANSS
70
60
50
40
30
Baseline
1 mo.
3 mo.
6 mo.
Olanzapine
Quetiapine
Ziprasidone
9 mo.
12 mo.
15 mo.
Perphenazine
Risperidone
Olanzapine<risperidone, quetiapine (with Hochberg adjustment for multiple
comparisons)
18 mo.
Quality Adjusted Life Years (LS Means from Mixed
Models (Adj for site, baseline, exacerbation)
(all group p<.0001; time p<.0001)
0.80
QALYs
0.60
0.40
0.20
0.00
Baseline
1 mo.
3 mo.
6 mo.
Olanzapine
Quetiapine
Ziprasidone
9 mo.
12 mo.
15 mo.
Perphenazine
Risperidone
Perphenazine>risperidone (with Hochberg adjustment for multiple comparisons)
18 mo.
Visual Analogue Scale: 0-100
[death - perfect health] (LS Means from Mixed
Models (Adj for site, baseline, exacerbation)(all
group p= ns, time p<.0002)
Monthly expenditures
75
60
45
30
15
0
Baseline
1 mo.
3 mo.
6 mo.
Olanzapine
Quetiapine
9 mo.
12 mo.
15 mo.
Perphenazine
Risperidone
18 mo.
Aripiprazole vs. Perphenazine in
patients. Refractory olz and risp:
Kane Meltzer, 2007

6 week run-in trial with olanzapine or
ripseridone with <20% PANSS response

334/416 completed open label phase



2% responded
Mean PANSS increased by 3.5% on olanz and 3.2% of
risperidone
225 (75%) completed random phase

6 weeks outcome



Completed
DC Adevrse event
DC lack of Effic
Aripiprazole
71%
14%
6%
Perphenazine
79%
8%
5%
Aripiprazole vs. Perphenazine in
patients. Refractory olz and risp:
Kane Meltzer, 2007
Aripip
PANSS (mean)
-9.8
PANSS Responder (25%)
27%
CGI
-0.3
QLS responders
35%
Serious Adv evnt.
21%
EPS
13.7%
Akithesia
6.3%
EPS meds
27.8%
Body weight (loss)
-1.5
Perph.
-10.5 ns
25% ns
-0.3 ns
26% ns
17% ns
19.4% ns
9.0% ns
17.6% ns
-2.2
ns
A Randomized Clinical Trial of LongActing Injectable Risperidone
(CONSTA ®) and Oral Antipsychotics
in Unstable Chronic Schizophrenia
RA Rosenheck, JH Krystal, R Lew,
PG Barnett, L Fiore, D Valley, SS
Thwin, JE Vertrees, MH Liang for
the CSP555 Investigators
VA Cooperative Studies Program
NEJM, March 4, 2011
Time on randomized drug (log
rank=1.7, df=1. p<0.19)
1
Survival Distribution Function
0.9
0.8
0.7
0.6
0.5
0.4
OralOral
0.3
LAI Risp
LAI-Risperidone
0.2
0.1
0
0
1
2
3
4
5 6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Month
Time to rehospitalization: p=.39:
LAI HR .87 (95% CI .63-1.2)
PANSS Follow-up score*
Total PANSS
85
80
75
70
Oral
65
LAI
60
55
50
Baseline 3 Mos
6 Mos
9 Mos
12 Mos
15 Mos
18 Mos
21 Mos
24 Mos
Oral
N= 178
140
122
117
112
99
84
68
50
LAI
N= 182
147
132
120
111
100
86
71
57
* (t=.47, df=1944, p=.65)
Tardive Dyskinesia (TD) Risk
SGA=vs. FGA Annual Incidence (0.8% vs
5.4%)* vs Health Outcomes
Six Dimensions of Sensitivity analysis

i) severity,

ii)duration,

iii)treatment with SGAs,

iv)QOL,

iv)QALYs

v) Annual cost
* Correll. Leucht and Kane, AJP 161(3): 596-601, 2004
ICERs for TD

If ,as per CATIE cost difference is $2,400- $3,600$6,000 cost/case of TD yields the following matrix:
D iff in
CATIE
CATIE Total
Effe ct
low/long-term risperidone
D iff. in Annua lize d H e a tlh Cost vs Pe rph.
$2,400
$6,216
Difference in Risk of TD cases
4.6%
$52,174
$135,130
Assume all cases are severe
0.14
$372,671
$965,217
Asssume 2/3 mild (QALY loss= 0.07)
0.093
$561,010 $1,453,015
Assume 15% of cases last < 3 months
0.85
$660,012 $1,709,430
Off Label Use

Most patients prescribed SGAs do not
have indicated treatments.




VA 1999: 33% off label
Market Scan® 2004: 56% off label
VA 2007: 60.2% off label
Recent review by AHRQ found weak
evidence of effectiveness of off-label
uses
CATIE AD Trial: Quality Adjusted Life Years from the
Health Utilities Index Mark III
0.4
Monthlyy QALY
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Baseline
3 mo.
6 mo.
Olanzapine (N=100, 91, 80, 74)
Quetiapine (N=94, 83, 74, 68)
Risperidone (N=85, 115, 74, 70)
Placebo (N=142, 123, 115, 109)
9 mo.
CATIE AD Trial: Monthly Costs: Experimental Medication
costs
Monthly expenditures
$200
$100
$0
1 mo.
2 mo.
3 mo.
4 mo.
5 mo.
6 mo.
7 mo.
Olanzapine (mean=$106)
Quetiapine (mean=$102)
Risperidone (mean = $92)
Placebo (mean = $45)
8 mo.
9 mo.
CATIE AD Trial: Average Monthly Costs: All healthcare
costs including medications*
Monthly expenditures
$1,600
$1,200
$800
$400
$0
Baseline
1
2
3
4
5
6
7
Olanzapine (mean=$1,118)
Quetiapine (mean=$1,215)
Risperidone (mean=$1,092)
Placebo (mean=$1,023)
8
9
How Could This Be? (1)


Leucht meta-analysis of 150 short-term
studies (2009) showed cloazpine (ES=.52), amisuplridide (ES=-.31),
olanzapine (ES=-.28) and risperidone
(ES=-.13) to be superior to FGAs.
“ Most previous studies addressed pure
efficacy and safety, whereas in the
CATIE and CUtLASS studies the
investigators focused on real-world
effectiveness. “
How Could This Be? (2)

“A strength of CATIE and CUtLASS was
the use of comparator drugs that are
less potent than haloperidol. A major
limitation of our metaanalysis is that
haloperidol was the comparator drug in
most of the studies, and the number of
studies of mid-potency first-generation
drugs was insufficient.” Leucht et al.
2009, p. 40.
Four groups of APS drugs


Risperidone or any FGA
Clozapine (2 or 3 failures)




Generic available
Weight gain risk is of concern and some
patients may not tolerate the required
blood monitoring
Aripiprazole, ziprasidone or quetiapine
Olanzapine: greatest weight gain
Implementation

Two principals (not a protcol: flexibility)





Do not create disincentive to take medication
Individualize treatment when needed
Apply preferential sequence when: a)
initiating therapy or b) change is required
Self-authortization of group 3 and 4 drug use
(aripiprazole, ziprasidone, quetiapine,
olanzapine)
Mandatory review form before Rx is filled
Review form







sociodemographic characteristics
diagnoses
risk factors for TD and metabolic disease
previous antipsychotic medication failures,
clinical reasons for selecting the proposed
agent.
Consent for FGAs (TD) clozpaine, quetiapine,
olanzapine (weight gain, DM risk)
All important criteria for selecting
medications.
Incentives to Change


Academic Detailing; Education/Information
Monitor prescribing decision making with a
structured form when NEW SGAs are
prescribed



Very weak disincentive (time burden for the form)
Feedback information in further academic
detailing
Caveat: Astra-Zeneca complained that we
were not treating all SGAs equally as per VA
agreement so we had to add risperidone to
level 2
VA CT: Monitoring Form Data On All new
Starts of Atypical Antipsychotics FY 20082009 (N=1,721): Diagnosis

Schizophrenia or BP




Schizophrenia
Bipolar disorder
Sleep
Other Diagnoses


Other affective
PTSD
44%
18.1%
26.6%
18%
38%
24.0%
23.4%
Monitoring Form Data (N=1,721):
Reason for new medication









Efficacy
Sleep
Patient preference
Less EPS
Less TD risk
Less akathisia
Less sedation
Treatment of TD
Other
39.3%
30.0%
27.0%
12.5%
8.9%
4.7%
5.8%
0.8%
20.7%
Monitoring Form Data (N=1,721):
Co-Morbidity
TD
 EPS
 Akathisia
 Diabetes
 Hyperlipidemia
 Obesity
 Hypertension
- ASCVD

4.3%
3.7%
3.0%
14.9%
29.0%
20.5%
34.0%
10.6%
(DK-17.0%)
(DK-11.3%)
(DK-10.2%)
(DK-24.5%)