1. health and fitness
2. therapy

Original Contributions
Lurasidone for the Acute Treatment of Adults
with Schizophrenia: What is the Number
Needed to Treat, Number Needed to Harm,
and Likelihood to be Helped or Harmed?
Leslie Citrome 1
Abstract
Objective: To describe the efficacy, safety and tolerability of lurasidone for the acute treatment of schizophrenia using
the metrics number needed to treat (NNT) and number needed to harm (NNH). Methods: Study data were pooled
from six Phase II and III, 6-week, randomized, placebo-controlled trials that were conducted to test the efficacy and
safety of lurasidone for the acute treatment of schizophrenia. Included were the following interventions: fixed doses of
lurasidone 20, 40, 80, 120 and 160 mg/d; haloperidol 10 mg/d; olanzapine 15 mg/d; quetiapine extended-release 600
mg/d; placebo. The following outcomes were assessed: responder rates as defined by a reduction of ≥20, 30, 40 or 50%
from baseline on the Positive and Negative Syndrome Scale (PANSS) total score; study completion; discontinuation
due to an adverse event (AE); weight gain ≥7% from baseline; incidence of spontaneously reported AEs; incidence of
total cholesterol ≥240 mg/dL, low-density lipoprotein cholesterol ≥160 mg/dL, fasting triglycerides ≥200 mg/dL and
glucose ≥126 mg/dL at endpoint. NNT for the efficacy outcomes were calculated after excluding one failed study. NNH
for the safety/tolerability outcomes were calculated using all six studies. Likelihood of being helped or harmed (LHH)
was also calculated to illustrate trade-offs between outcomes of improvement ≥30% on the PANSS vs. incidence of
akathisia, nausea, sedation, somnolence and parkinsonism. Results: NNT vs. placebo for PANSS reductions ≥30%
were 6, 6, 7 and 4 for lurasidone doses of 40, 80, 120 and 160 mg/d, respectively, and 4 and 3 for olanzapine 15 mg/d
and quetiapine extended-release 600 mg/d, respectively. Lurasidone was not associated with any statistically significant
disadvantages over placebo for weight gain or metabolic abnormalities; NNH vs. placebo for weight gain ≥7% from
baseline was 4 for olanzapine and 9 for quetiapine extended-release in contrast to a NNH for this outcome ranging
from 43 to 150 for lurasidone 40–160 mg/d. The 5 most consistently encountered adverse events attributable to lurasidone were akathisia, nausea, sedation, somnolence and parkinsonism, with NNH vs. placebo for lurasidone 40–120
mg/d ranging from 6 (akathisia with 120 mg/d) to 30 (parkinsonism with 80 mg/d). Lurasidone 160 mg/d appeared
better tolerated than doses of 40, 80 or 120 mg/d for akathisia, nausea, sedation or somnolence, with no NNH values
for these adverse events for 160 mg/d vs. placebo being statistically significant. LHH was favorable for lurasidone when
contrasting PANSS reductions vs. adverse events. Conclusions: NNT and NNH can help quantify efficacy, safety and
tolerability outcomes and place lurasidone into clinical perspective. Advantages for lurasidone include a low propensity
for weight gain and metabolic abnormalities. More commonly encountered adverse events include akathisia, nausea,
sedation, somnolence and parkinsonism, but NNH values are generally in the double digits, reflecting an overall tolerable profile. Individual patient characteristics, values and preferences will need to be considered when selecting lurasidone over other antipsychotics.
Key Words: Efficacy, Lurasidone, Number Needed to Harm, Number Needed to Treat,
Safety, Schizophrenia, Tolerability
1
New York Medical College, Valhalla, NY
Address for correspondence: Leslie Citrome, MD, MPH,
11 Medical Park Drive, Suite 106, Pomona, NY 10970
Phone: 845-362-2081; Fax: 845-362-8745; E-mail: [email protected]
Submitted: October 17, 2011; Revised: January 3, 2012;
Accepted: February 9, 2012
Introduction
Lurasidone is a second-generation antipsychotic agent
that was approved by the U.S. Food and Drug Administration (FDA) in 2010 (1). Although it is similar to other atypical antipsychotics in terms of antagonism at dopamine D2
and serotonin 5-HT2A receptors, lurasidone also has high
Clinical Schizophrenia & Related Psychoses April 2012
• 1
Lurasidone NNT NNH
Clinical Implications
Lurasidone is FDA-approved for the treatment of schizophrenia based on a subset of the clinical trial data now available.
NNT and NNH can help quantify efficacy, safety and tolerability outcomes and place lurasidone into clinical perspective. Advantages for lurasidone include a low propensity for weight gain and metabolic abnormalities. The five most
consistently encountered adverse events attributable to lurasidone are akathisia, nausea, sedation, somnolence and parkinsonism, but NNH values are generally in the double digits, reflecting an overall tolerable profile. Individual patient
characteristics, values and preferences will need to be considered when selecting lurasidone over other antipsychotics.
binding affinity (antagonist) for the serotonin 5-HT7 receptor. Lurasidone has moderate affinity for serotonin 5-HT1A
(partial agonist) and noradrenaline α2c (antagonist) receptors. Lurasidone has little to no appreciable affinity for histamine H1 and acetylcholine M1 receptors.
The efficacy and short-term tolerability of lurasidone
for the treatment of acute schizophrenia has been tested
in several 6-week, randomized, placebo-controlled trials at
fixed doses ranging from 20 to 160 mg, administered once
daily (2-6). As per the U.S. product label (1), and based on
the data available at the time of approval, the recommended
starting dose is 40 mg administered once daily. The maximum recommended dose is 80 mg/d. When examining
data from the 4 positive trials available at the time of FDA
approval, fixed doses of 120 mg/d did not produce greater
improvement than seen at lower doses, but were associated
with a higher propensity for somnolence and akathisia. As
per the product label, lurasidone is associated with minimal
weight gain and no clinically meaningful alterations in glucose, lipids, prolactin, or the electrocardiogram QT interval.
Number needed to treat (NNT) and number needed to
harm (NNH) are measures of effect size and indicate how
many patients would need to be treated with one agent instead of the comparator in order to encounter one additional
outcome of interest (7, 8). NNT is traditionally used to denote potential desirable outcomes (benefits) that a clinician
and patient would want to encounter and NNH is ordinarily used to denote potential undesirable outcomes (harms)
that a clinician and patient would like to avoid. Both NNT
and NNH are calculated using the same formula. Some advocate for using the terms number needed to treat for an additional beneficial outcome (NNTB) and number needed to
treat for an additional harmful outcome (NNTH), instead of
NNT and NNH, respectively (8). Both sets of terms are commonly found in the medical literature. Nevertheless, NNT
and NNH offer the clinician a means of quantifying clinical
relevance or “clinical significance.” Lower NNTs (or NNHs)
are evidenced when there are large differences between the
interventions in question. For example, a NNT of 2 would
be a very large effect size, as a difference is encountered after
treating 2 patients with one of the interventions versus the
other. A NNT of 50 would mean little difference between
the two interventions, as it would take treating 50 patients
2 •
Clinical Schizophrenia & Related Psychoses April 2012
to encounter a difference in outcome. A useful medication
is one with a low NNT and a high NNH when comparing
it with another intervention; a low NNT and a high NNH
would mean one is more likely to encounter a benefit than
a harm. A rule of thumb is that a single digit NNT vs. placebo for an efficacy measure suggests that the intervention
has potentially useful advantages, and that double digit or
higher NNHs vs. placebo for adverse outcomes indicate that
the intervention is potentially tolerable.
Likelihood of being helped or harmed (LHH) can be a
useful way of synthesizing data regarding benefits and risks
(10-12). LHH is calculated using the following formula:
LHH=(1/NNT)/(1/NNH)=NNH/NNT. In general, a LHH
greater than 1 would mean the likelihood to be helped is
greater than the likelihood to be harmed. For a LHH less
than 1, the reverse is true. Choosing which NNH and NNT
to use in calculating LHH requires careful consideration so
that the outcomes being assessed are well matched and consistent with a patient’s values and preferences (11, 12).
The aim of this report is to apply NNT, NNH and LHH
in order to appraise the lurasidone acute schizophrenia clinical trial data.
Methods
Study data were pooled from six Phase II and III, 6-week,
randomized, placebo-controlled trials that were conducted
to test the efficacy, safety and tolerability of lurasidone for
the acute treatment of schizophrenia and that were conducted either exclusively in the U.S. or that included study sites
within the U.S. (2, 3, 13-15). One additional 6-week lurasidone placebo-controlled trial (conducted entirely in Japan,
Korea and Taiwan) has been recently completed but was not
included in this analysis (16). Table 1 provides an outline of
the number of subjects in each arm for each of the six clinical trials included in this report. More detailed descriptions
of the individual studies can be found elsewhere in peerreviewed journals (2, 3), posters presented at scientific meetings (13, 14), or publicly available regulatory documents
(15). The subjects themselves were typical of those enrolled
in registration studies for the acute treatment of schizophrenia (4, 5, 13, 15). Mean age ranged from approximately
36–41 years, depending on the treatment arm, with 64–79%
being men (13, 14). Mean baseline PANSS total score ranged
Leslie Citrome
Table 1
Phase II and III, 6-Week, Randomized, Controlled Trials for Lurasidone
in Patients with Schizophrenia: Treatment Arms
Lurasidone—Safety Population* N; ITT† N
Phase
Study (Ref #)
80 mg/d
120 mg/d
160 mg/d
Antipsychotic
Safety
Population* N;
ITT† N
D1050006 (14,15)
50; 48
NA
50; 49
NA
49; 47
NA
NA
D1050049 (15)
72; 70
71; 71
67; 65
71; 69
NA
NA
Haloperidol
10 mg/d
D1050196 (2)
90; 90
NA
NA
90; 90
NA
NA
NA
D1050229 (14,15)
127; 124
NA
124; 121
121; 118
124; 123
NA
NA
D1050231 (3)
116; 114
NA
119; 118
NA
118; 118
NA
Olanzapine
15 mg/d
122; 121
D1050233§ (13)
121; 120
NA
NA
125; 125
NA
121; 121
Quetiapine
XR 600 mg/d
119; 116
‡
II
Placebo—
Safety Population* N;
20 mg/d 40 mg/d
ITT† N
Active Control
III
72; 72
ITT=Intent-to-Treat; NA=not applicable; XR=extended release
*Safety population consisted of all randomized subjects who received at least one dose of study medication.
†Intent-to-Treat population consisted of all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline Positive and Negative Syndrome Scale measure.
‡This was a failed study. Neither lurasidone nor the active control statistically separated from placebo on the primary psychopathological
outcome measure. §This study was completed after the manufacturer submitted their New Drug Application to the FDA.
from approximately 93 to 98, depending on the treatment
arm (13, 14).
Pooling across the available studies, the following proportions were calculated for subjects who were randomized
to placebo, lurasidone 20 mg/d, 40 mg/d, 80 mg/d, 120 mg/d
or 160 mg/d, haloperidol 10 mg/d, olanzapine 15 mg/d and
quetiapine extended-release (XR) 600 mg/d:
1. the proportion (n/N) of subjects for each treatment
arm meeting the criterion of a ≥20%, 30%, 40% and
50% reduction from baseline to last observation carried
forward (LOCF) endpoint on the Positive and Negative Syndrome Scale (PANSS) total score (reduction
calculated by [PANSS total score at baseline-PANSS
total score at LOCF endpoint]/[PANSS total score at
baseline-30]). N, the denominator, is the Intent-toTreat population (all subjects who were randomized,
received at least one dose of study medication, and
had a baseline and at least one post-baseline PANSS
assessment; excluded was the failed study where neither lurasidone nor the active control separated from
placebo and, hence, that study was not informative
about efficacy).
2. the proportion (n/N) of subjects for each treatment arm
who completed the study and the proportion of subjects
for each treatment arm who discontinued specifically
due to an adverse event (AE). N is the safety population (all randomized subjects who received at least one
dose of study medication; all six studies were included).
3. the proportion (n/N) of subjects for each treatment
arm who experienced weight gain of ≥7% from baseline at LOCF endpoint. N is the safety population.
4. the proportion (n/N) of subjects for each treatment
arm who spontaneously reported AEs by event type.
N is the safety population. AEs were obtained from the
clinical trial reports. Selected were those AEs where
the pooled incidence rate for all lurasidone doses was
at least 5% (rounded upwards in case of incidence
>4.5%): headache, akathisia, nausea, sedation, somnolence, insomnia, vomiting, dyspepsia, parkinsonism
(or extrapyramidal disorder), anxiety, agitation and
dizziness.
5. the proportion of subjects (n/N) with a fasting
total cholesterol value ≥240 mg/dL, fasting low-density
lipoprotein cholesterol ≥160 mg/dL, fasting triglyceride ≥200 mg/dL and fasting glucose ≥126 mg/dL at
LOCF endpoint. N is the safety population and where
a baseline and at least one post-baseline assessment for
each metabolic variable were available.
The studies included in this analysis were all designed
to compare lurasidone with placebo and, thus, powered as
such. When an active control was used, it was for the purposes of assay sensitivity. Thus, all calculations for NNT or
NNH are for antipsychotic vs. placebo. The NNT vs. placebo
for the PANSS responder threshold outcomes and completer
rates and NNH vs. placebo for all other outcomes were calClinical Schizophrenia & Related Psychoses April 2012
• 3
Lurasidone NNT NNH
Table 2
Responder Rates and NNT vs. Placebo-Pooled Data (Intent-to-Treat Population),
Excluding Data from the Failed Study
PANSS Reduction ≥20% PANSS Reduction ≥30%
PANSS Reduction ≥40%
PANSS Reduction ≥50%
N
n (%)
NNT
(95% CI)
n (%)
NNT
(95% CI)
n (%)
NNT
(95% CI)
n (%)
NNT
(95% CI)
Placebo
496
223 (45.0%)
NA
158 (31.9%)
NA
117 (23.6%)
NA
76 (15.3%)
NA
Lurasidone 40 mg/d
288
168 (58.3%)
8 (5–17)
141 (49.0%)
6 (5–10)
109 (37.8%)
8 (5–14)
76 (26.4%)
10 (6–20)
Lurasidone 80 mg/d
333
214 (64.3%)
6 (4–8)
163 (48.9%)
6 (5–10)
120 (36.0%)
9 (6–17)
72 (21.6%)
16 (9–117)
Lurasidone 120 mg/d
288
175 (60.8%)
7 (5–12)
137 (47.6%)
7 (5–12)
93 (32.3%)
12 (7–47)
60 (20.8%)
19 (ns)
Lurasidone 160 mg/d
121
95 (78.5%)
3 (3–4)
76 (62.8%)
4 (3–5)
56 (46.3%)
5 (4–8)
35 (28.9%)
8 (5–21)
Olanzapine 15 mg/d
121
89 (73.6%)
4 (3–6)
78 (64.5%)
4 (3–5)
59 (48.8%)
4 (3–7)
41 (33.9%)
6 (4–11)
Quetiapine XR 600 mg/d
116
92 (79.3%)
3 (3–4)
82 (70.7%)
3 (3–4)
65 (56.0%)
4 (3–5)
33 (28.4%)
8 (5–24)
Intervention
Shaded entries represent NNT values where the 95% CI does not include infinity and, hence, considered statistically significant.
CI=confidence interval; NA=not applicable; NNT=number needed to treat; ns=not significant (the 95% CI contains “infinity”);
PANSS=Positive and Negative Syndrome Scale; XR=extended release
Table 3 Rates and NNT vs. Placebo for Study Completion and NNH
vs. Placebo for Discontinuation Due to an Adverse EventPooled Data (Safety Population)
Study Completion
Discontinuation Due to AE
N
n (%)
NNT
(95% CI)
n (%)
NNH
(95% CI)
Placebo
576
316 (54.9%)
NA
40 (6.9%)
NA
Lurasidone 20 mg/d
71
27 (38.0%)
-6 (-4 to -21)*
1 (1.4%)
-19 (-12 to -48)*
Lurasidone 40 mg/d
360
205 (56.9%)
48 (ns)
48 (13.3%)
16 (10–44)
Lurasidone 80 mg/d
407
258 (63.4%)
12 (7–43)
35 (8.6%)
61 (ns)
Lurasidone 120 mg/d
291
171 (58.8%)
26 (ns)
43 (14.8%)
13 (9–31)
Lurasidone 160 mg/d
121
93 (76.9%)
5 (4–8)
4 (3.3%)
-28 (ns)*
Haloperidol 10 mg/d
72
29 (40.3%)
-7 (-4 to -40)*
11 (15.3%)
12 (ns)
Olanzapine 15 mg/d
122
84 (68.9%)
8 (5–21)
16 (13.1%)
17 (ns)
Quetiapine XR 600 mg/d
119
97 (81.5%)
4 (3–6)
4 (3.4%)
-28 (ns)*
Intervention
Shaded entries represent NNH values where the 95% CI does not include infinity and, hence,
considered statistically significant. AE=adverse event; CI=confidence interval; NA=not
applicable; NNT=number needed to treat; NNH=number needed to harm; ns=not significant
(the 95% CI contains “infinity”); XR=extended release. *A negative NNT denotes an advantage for placebo regarding the potential benefit; a negative NNH denotes an advantage for
study medication regarding the potential harm.
culated from the pooled proportions. NNT or NNH can be
calculated by first computing the difference in proportions
between the experimental intervention and the control intervention for the outcome of interest (risk difference), and
then computing the reciprocal of the risk difference and
rounding up to the next highest whole number (9). Calculating the 95% confidence interval (CI) for the NNT or NNH
involves first calculating the 95% CI for the risk difference;
4 •
Clinical Schizophrenia & Related Psychoses April 2012
the formula for this can be found elsewhere (9). When the
95% CI for the NNT or NNH estimate contains infinity, the
NNT or NNH estimate is not statistically significant at the P
threshold of <0.05 and under those circumstances, instead
of providing the 95% CI, the notation “ns” for not significant
is made. All calculations for NNT and NNH were done using the active medication as the experimental intervention
and placebo as the control intervention, with the assumption
Leslie Citrome
that the study medication would be superior to placebo on
efficacy outcomes but inferior to placebo on safety or tolerability outcomes. Thus, a negative NNT denotes an advantage for placebo over study medication regarding a potential
benefit and a negative NNH denotes an advantage for study
medication over placebo regarding a potential harm.
Table 4 Rates and NNH vs. Placebo for Weight
Increase ≥7% from Baseline—Pooled
Data (Safety Population)
At Study Endpoint (LOCF)
Intervention
N
n (%)
NNH (95% CI)
Placebo
565
21 (3.7%)
NA
Lurasidone 20 mg/d
71
1 (1.4%)
-44 (ns)*
Lurasidone 40 mg/d
358
21 (5.9%)
47 (ns)
Lurasidone 80 mg/d
395
24 (6.1%)
43 (ns)
Lurasidone 120 mg/d
291
15 (5.2%)
70 (ns)
Lurasidone 160 mg/d
114
5 (4.4%)
150 (ns)
Haloperidol 10 mg/d
71
3 (4.2%)
197 (ns)
Olanzapine 15 mg/d
122
42 (34.4%)
4 (3–5)
Quetiapine XR 600 mg/d
111
17 (15.3%)
9 (6–22)
Shaded entries represent NNH values where the 95% CI does not
include infinity and, hence, considered statistically significant.
CI=confidence interval; NA=not applicable; NNH=number needed
to harm; ns=not significant (the 95% CI contains “infinity”);
XR=extended release. *A negative NNH denotes an advantage
for study medication regarding the potential harm.
Results
Table 2 provides the responder rates and the NNT vs.
placebo for the different treatment arms using pooled data
for each dose of lurasidone, excluding data from a single
failed study where neither lurasidone nor haloperidol statistically separated from placebo (15). Responder rates decreased as the threshold for response increased from 20 to
50%. All doses of lurasidone demonstrated a statistically
significant NNT vs. placebo at all thresholds for response,
with the exception of lurasidone 120 mg/d for the outcome
of ≥50% reduction in PANSS total score. The most robust
NNT values for lurasidone were observed for 160 mg/d;
the data for this dose are limited to one study (13). Figure 1
graphically displays the NNT vs. placebo for response for the
different doses of lurasidone.
In order to determine the impact of the failed study on
the overall pattern of results, these PANSS analyses were repeated by including the failed study in the pooling process.
Results were similar and are presented in Table S1 in the
supplementary appendix.
Table 3 provides the NNT for study completion and the
NNH for discontinuation due to an AE. Completion rates
vs. placebo demonstrated a statistically significant advantage for lurasidone 80 and 160 mg/d, as well as for the active
controls olanzapine and quetiapine extended-release. Discontinuation due to an AE demonstrated a disadvantage for
lurasidone 40 mg/d vs. placebo with a NNH of 16 (95% CI
10–44) and for lurasidone 120 mg/d vs. placebo with a NNH
of 13 (95% CI 9–31).
Table 4 provides the rates and NNH for clinically relevant weight gain, as defined by an increase in body weight
from baseline ≥7%. No statistically significant disadvantage
from placebo on this outcome was observed for any dose of
lurasidone or for haloperidol. However, NNH vs. placebo for
clinically relevant weight gain at study endpoint was 4 (95%
CI 3–5) for olanzapine and 9 (95% CI 6–22) for quetiapine
extended-release, in contrast to a NNH for this outcome
ranging from 43 (ns) to 150 (ns) for doses of lurasidone
40–160 mg/d. This is consistent with the outlier analysis for
abnormal metabolic variables as presented in Table 5, where
the only statistically significant differences from placebo
were observed for the outcome of triglycerides ≥200 mg/dL
at LOCF endpoint demonstrating an advantage for lurasidone 160 mg/d vs. placebo with a NNH of -11 (95% CI -7
to -34) and a disadvantage of olanzapine vs. placebo with a
NNH of 9 (95% CI 5–40).
Table 6A outlines the incidence of the 12 most commonly encountered spontaneously reported AEs associated
with lurasidone (all doses; incidence at least 5% after rounding). However, for many of these AEs, the difference from
placebo was small. Only 5 AEs had consistently different
rates when comparing most of the doses of lurasidone vs.
placebo. These are noted in Table 6B listing the NNH values
and are: akathisia, nausea, sedation, somnolence and parkinsonism. These are also graphically presented in Figure 2.
Lurasidone 160 mg/d appears better tolerated than lurasidone 40, 80 and 120 mg/d regarding akathisia, nausea, sedation and somnolence. As noted, the data for lurasidone 160
mg/d came from a single study. The design and conduct of
that study was similar to the others with the exception that
lurasidone was administered in the evening, in contrast to
the five other studies where study medication was administered in the morning. The active controls demonstrated clinically relevant NNH values vs. placebo as would be expected
from their individual clinical profiles: haloperidol was associated with low (i.e., more robust) NNH values for akathisia,
parkinsonism and sedation, olanzapine with sedation, and
quetiapine extended-release with dizziness.
Figures 1 and 2 can be used to contrast trade-offs between efficacy and tolerability outcomes. Although in most
instances the NNT would be lower than the NNH (and,
Clinical Schizophrenia & Related Psychoses April 2012
• 5
Lurasidone NNT NNH
Table 5 Rates and NNH vs. Placebo for Subjects with Abnormal Metabolic Variables
(LOCF)-Pooled Data (Safety Population)
Total Cholesterol ≥240 mg/dL LDL Cholesterol ≥160 mg/dL
Intervention
N
n (%)
Triglycerides ≥200 mg/dL
NNH
(95% CI)
N
n (%)
NNH
(95% CI)
NA
409
32 (7.8%)
NA
530 100 (18.9%)
No data
No data
NA
71
N
n (%)
Glucose ≥126 mg/dL
NNH
(95% CI)
N
n (%)
NNH
(95% CI)
Placebo
530 60 (11.3%)
Lurasidone 20 mg/d
71
Lurasidone 40 mg/d
341 35 (10.3%)
-95 (ns)*
228
15 (6.6%) -81 (ns)* 341 71 (20.8%)
52 (ns)
352 29 (8.2%) 130 (ns)
Lurasidone 80 mg/d
375 43 (11.5%)
686 (ns)
303
25 (8.3%) 235 (ns) 375 59 (15.7%)
-32 (ns)*
382 29 (7.6%) 810 (ns)
Lurasidone 120 mg/d
268
24 (9.0%)
-43 (ns)*
221
12 (5.4%) -42 (ns)* 268 52 (19.4%)
187 (ns)
283 30 (10.6%) 32 (ns)
Lurasidone 160 mg/d
115 12 (10.4%)
-113 (ns)*
115
15 (13.0%) 20 (ns)
Haloperidol 10 mg/d
70
-76 (ns)*
No data
No data
NA
Olanzapine 15 mg/d
115 21 (18.3%)
15 (ns)
115
11 (9.6%)
Quetiapine XR 600 mg/d 107 17 (15.9%)
22 (ns)
107
8 (11.3%) -1,882 (ns)*
7 (10.0%)
13 (18.3%)
NA
549 41 (7.5%)
NA
-180 (ns)*
71 9 (12.7%)
20 (ns)
115
11 (9.6%)
-11 (-7 to -34)* 113 5 (4.4%) -33 (ns)*
70
14 (20.0%)
89 (ns)
70
58 (ns)
115 35 (30.4%)
9 (5–40)
121 12 (9.9%)
41 (ns)
14 (13.1%) 19 (ns)
106 27 (25.5%)
16 (ns)
107 10 (9.3%)
54 (ns)
5 (7.1%) -308 (ns)*
Shaded entries represent NNH values where the 95% CI does not include infinity and, hence, considered statistically significant.
CI=confidence interval; LDL=low density lipoprotein; NA=not applicable; NNH=number needed to harm; ns=not significant (the 95% CI
contains “infinity”); XR=extended release. *A negative NNH denotes an advantage for study medication regarding the potential harm.
Table 6A Number of Patients and Rate (%) for Common Adverse Events-Pooled Data
(Safety Population)
Adverse
Event
Placebo
(N=576)
Lurasidone Lurasidone Lurasidone Lurasidone Lurasidone Lurasidone Haloperidol Olanzapine Quetiapine XR
All Doses 20 mg/d
40 mg/d
80 mg/d 120 mg/d 160 mg/d
10 mg/d
15 mg/d
600 mg/d
(N=1,250)
(N=71)
(N=360)
(N=407)
(N=291)
(N=121)
(N=72)
(N=122)
(N=119)
Headache
96 (16.7%) 201 (16.1%) 15 (21.1%)
74 (20.6%)
55 (13.5%)
45 (15.5%)
12 (9.9%)
14 (19.4%)
18 (14.8%)
13 (10.9%)
Akathisia
16 (2.8%) 170 (13.6%)
4 (5.6%)
41 (11.4%)
52 (12.8%)
64 (22.0%)
9 (7.4%)
14 (19.4%)
9 (7.4%)
2 (1.7%)
Nausea
31 (5.4%) 138 (11.0%)
8 (11.3%)
39 (10.8%)
46 (11.3%)
37 (12.7%)
8 (6.6%)
4 (5.6%)
6 (4.9%)
4 (3.4%)
Sedation
25 (4.3%) 123 (9.8%)
8 (11.3%)
38 (10.6%)
39 (9.6%)
36 (12.4%)
2 (1.7%)
15 (20.8%)
19 (15.6%)
5 (4.2%)
Somnolence
22 (3.8%) 120 (9.6%)
3 (4.2%)
33 (9.2%)
34 (8.4%)
42 (14.4%)
8 (6.6%)
9 (12.5%)
11 (9.0%)
16 (13.4%)
Insomnia
41 (7.1%) 106 (8.5%)
6 (8.5%)
31 (8.6%)
36 (8.8%)
25 (8.6%)
8 (6.6%)
12 (16.7%)
13 (10.7%)
5 (4.2%)
Vomiting
33 (5.7%)
97 (7.8%)
5 (7.0%)
19 (5.3%)
37 (9.1%)
27 (9.3%)
9 (7.4%)
4 (5.6%)
3 (2.5%)
6 (5.0%)
Dyspepsia
31 (5.4%)
86 (6.9%)
8 (11.3%)
24 (6.7%)
25 (6.1%)
22 (7.6%)
7 (5.8%)
5 (6.9%)
7 (5.7%)
3 (2.5%)
Parkinsonism* 9 (1.6%)
84 (6.7%)
3 (4.2%)
25 (6.9%)
20 (4.9%)
28 (9.6%)
8 (6.6%)
13 (18.1%)
7 (5.7%)
4 (3.4%)
Anxiety
25 (4.3%)
76 (6.1%)
2 (2.8%)
29 (8.1%)
22 (5.4%)
19 (6.5%)
4 (3.3%)
10 (13.9%)
7 (5.7%)
1 (0.8%)
Agitation
24 (4.2%)
74 (5.9%)
7 (9.9%)
32 (8.9%)
12 (2.9%)
17 (5.8%)
6 (5.0%)
3 (4.2%)
8 (6.6%)
3 (2.5%)
Dizziness
14 (2.4%)
59 (4.7%)
4 (5.6%)
17 (4.7%)
17 (4.2%)
14 (4.8%)
7 (5.8%)
3 (4.2%)
3 (2.5%)
16 (13.4%)
*Alternative coding is extrapyramidal disorder (studies D1050006, D1050049, D1050196). XR=extended release
hence, the benefit would be encountered more often than
the harm), this may not always be the case as, for example,
lurasidone 120 mg/d and the NNH vs. placebo for akathisia.
LHH can be calculated to help quantify these trade-offs, as
illustrated in Table 7 where NNT for response (defined by
≥30% decrease in PANSS from baseline; data excluding the
failed study) is contrasted with NNH for akathisia, nausea,
6 •
Clinical Schizophrenia & Related Psychoses April 2012
sedation, somnolence and parkinsonism, for lurasidone 40,
80, 120 and 160 mg/d. Akathisia, nausea, sedation, somnolence and parkinsonism are considered commonly observed
AEs associated with lurasidone (see Tables 6A and 6B). The
LHH is greater than 1 for all scenarios, with the exception of
lurasidone 120 mg/d for response vs. akathisia. Advantages
appear strongest for lurasidone 160 mg/d over lower doses. Leslie Citrome
Number Needed to Treat vs. Placebo
Figure 1
PANSS Responders with Lurasidone: NNT and 95% Confidence Interval vs.
Placebo By Dose
100
16
12
10
8
7
6
6
7
6
10
9
8
19
n.s.
8
5
4
3
1
40
Dose
(mg/d)
80
120
160
40
80
≥20%
120
160
40
80
≥30%
120
160
40
≥40%
80
120
160
≥50%
PANSS Reduction from Baseline
NNT=number needed to treat; n.s.=not significant; PANSS=Positive and Negative Syndrome Scale
Number Needed to Harm vs. Placebo
Figure 2
Common Adverse Events Associated with Lurasidone: NNH And 95% Confidence
Interval vs. Placebo by Dose
100
82
n.s.
36
n.s.
22
n.s.
12
10
Dose
(mg/d)
19
17
17
14
20
19
30
23
20
19
13
13
10
10
6
1
40
80 120 160
Akathisia
40
80 120 160
40
Nausea
80 120 160*
Sedation
40
80
120
Somnolence
160
40
80
120 160
Parkinsonism
Adverse Event
NNH=number needed to harm; n.s.=not significant.
*NNH for sedation for lurasidone 160 mg/d is -38 (n.s.).
Clinical Schizophrenia & Related Psychoses April 2012
• 7
Lurasidone NNT NNH
Table 6B
Adverse
Event
NNH (95% CI) vs. Placebo for Common Adverse Events-Pooled Data (Safety Population)
Lurasidone Lurasidone Lurasidone Lurasidone Lurasidone
All Doses
20 mg/d
40 mg/d
80 mg/d
120 mg/d
(N=1,250)
(N=71)
(N=360)
(N=407)
(N=291)
Lurasidone
160 mg/d
(N=121)
Haloperidol Olanzapine Quetiapine XR
10 mg/d
15 mg/d
600 mg/d
(N=72)
(N=122)
(N=119)
Headache
-171 (ns)*
23 (ns)
26 (ns)
-32 (ns)*
-84 (ns)*
-15 (-8 to -163)*
36 (ns)
-53 (ns)*
-18 (ns)*
Akathisia
10 (8–12)
35 (ns)
12 (9–20)
10 (8–16)
6 (5–7)
22 (ns)
6 (4–14)
22 (ns)
-92 (ns)*
Nausea
18 (13–32)
17 (ns)
19 (11–58)
17 (11–43)
14 (9–33)
82 (ns)
576 (ns)
-216 (ns)*
-50 (ns)*
Sedation
19 (13–32)
15 (ns)
17 (11–38)
20 (12–52)
13 (9–26)
-38 (ns)*
7 (4–15)
9 (6-22)
-722 (ns)*
Somnolence
18 (13–29)
247 (ns)
19 (12–51)
23 (14–71)
10 (7–16)
36 (ns)
12 (7–114)
20 (ns)
11 (7-31)
Insomnia
74 (ns)
75 (ns)
67 (ns)
58 (ns)
68 (ns)
-198 (ns)*
11 (6–146)
29 (ns)
-35 (ns)*
Vomiting
50 (ns)
77 (ns)
-222 (ns)*
30 (ns)
29 (ns)
59 (ns)
-576 (ns)*
-31 (ns)*
-146 (ns)*
Dyspepsia
67 (ns)
17 (ns)
78 (ns)
132 (ns)
46 (ns)
248 (ns)
64 (ns)
282 (ns)
-35 (ns)*
38 (ns)
19 (13–39)
30 (18–98)
13 (9–23)
20 (11–198)
7 (4–14)
24 (ns)
56 (ns)
Parkinsonism 20 (15–30)
Anxiety
58 (ns)
-66 (ns)*
27 (15–223)
94 (ns)
46 (ns)
-97 (ns)*
11 (6–72)
72 (ns)
-29 (-18 to -86)*
Agitation
57 (ns)
18 (ns)
22 (13–74)
-83 (ns)*
60 (ns)
127 (ns)
infinity
42 (ns)
-61 (ns)*
Dizziness
44 (25–177)
32 (ns)
44 (ns)
58 (ns)
42 (ns)
30 (ns)
58 (ns)
3,514 (ns)
10 (6–21)
Shaded entries represent NNH values where the 95% CI does not include infinity and, hence, considered statistically significant.
CI=confidence interval; NA=not applicable; NNH=number needed to harm; ns=not significant (the 95% CI contains “infinity”);
XR=extended release. *A negative NNH denotes an advantage for study medication regarding the potential harm.
Table 7
LHH: Response (≥30% Reduction from Baseline in PANSS Total Score)
vs. Akathisia, Nausea, Sedation, Somnolence and Parkinsonism by
Lurasidone Dose
Lurasidone 40 mg/d
Lurasidone 80 mg/d
NNT
NNH
LHH
NNT
NNH
LHH
NNT
NNH
LHH
NNT
NNH
LHH
Akathisia
6
12
2.0
6
10
1.7
7
6
0.9
4
22*
NA
Nausea
6
19
3.2
6
17
2.8
7
14
2.0
4
82*
NA
Sedation
6
17
2.8
6
20
3.3
7
13
1.9
4
-38*
NA
Somnolence
6
19
3.2
6
23
3.8
7
10
1.4
4
36*
NA
Parkinsonism
6
19
3.2
6
30
5.0
7
13
1.9
4
20
5.0
Adverse Event
Lurasidone 120 mg/d
Lurasidone 160 mg/d
*When the NNT and/or NNH are not statistically significant, LHH cannot be reliably calculated and is denoted
by NA. In the case where the NNH is a negative number, LHH is rendered meaningless. LHH=likelihood of being
helped or harmed calculated by NNH/NNT; NA=not applicable; NNH=number needed to harm; NNT=number
needed to treat; PANSS=Positive and Negative Syndrome Scale.
A LHH of 5.0 for lurasidone 160 mg/d for response vs.
parkinsonism can be interpreted that “lurasidone treatment
at 160 mg/d is 5 times more likely to help the patient (≥30%
decrease in PANSS) than harm the patient (parkinsonism).”
For olanzapine 15 mg/d, NNT vs. placebo for ≥30% decrease
in PANSS from baseline is 4 and NNH for weight increase
≥7% at study endpoint is 4, yielding a LHH of 1.0, which can
be interpreted as “olanzapine treatment at 15 mg/d is equally
likely to help the patient (≥30% decrease in PANSS) as harm
the patient (weight gain ≥7%).” Note that when calculating
and applying the LHH the benefit and the harm must be
clinically relevant with regard to the individual being treat-
8 •
Clinical Schizophrenia & Related Psychoses April 2012
ed. Benefits and harms can also have different time courses
and/or enduring consequences that also require careful consideration (11).
Discussion
Kraemer and Kupfer (17) propose that NNTs of 3
(rounded up from 2.3), 4 (rounded up from 3.6), and 9
(rounded up from 8.9) correspond to a Cohen’s d of 0.8, 0.5,
and 0.2, respectively, representing effect sizes that are “large,”
“medium,” and “small.” In general, the therapeutic effects of
lurasidone vs. placebo resulted in single digit NNT values,
as did the active controls olanzapine 15 mg/d and quetiap-
Leslie Citrome
ine extended-release 600 mg/d. For lurasidone 160 mg/d
the NNT was as robust as 3 (medium-large effect size) for
a threshold of improvement of ≥20% on the PANSS, with
the 95% CI overlapping that for olanzapine and quetiapine. Effect size differences from placebo generally became
weaker as the threshold for response was increased from 20
to 50%. A caveat is that studies with high placebo response
rates will yield NNT values that are less robust and will make
cross-study comparisons difficult to interpret; this becomes
relevant in this report when examining the NNT vs. placebo
for the active controls.
Tolerability outcome differences for lurasidone vs. placebo resulted in NNH values that were in the double digits
(with only one exception). This produced LHH values that
were consistently favorable. Clinically, additional considerations include the time to onset of the adverse event versus
time to onset of a therapeutic response, as well as the severity and duration of the adverse event. The adverse event in
question may be easily manageable if it is non-serious and
short-lived. The numerically lower incidence of akathisia
with lurasidone 160 mg/d compared to 40, 80 or 120 mg/d
will need to be further explored in additional studies, but
may be reflective of the different time of administration of
the medication (evening vs. morning dosing). With respect
to metabolic outcomes, no adverse safety signal was evident
for lurasidone across all doses in the examination of clinically relevant weight gain or laboratory abnormalities, in
contrast to that observed with the active controls olanzapine
and quetiapine.
In another published paper, NNH for the outcomes
of weight gain ≥7% from baseline and AEs of somnolence,
akathisia, or discontinuation due to an AE have been reported for the oral, first-line, second-generation antipsychotics
vs. placebo, as calculated from their U.S. product labels (18).
In that analysis, NNH for weight gain was the most favorable for lurasidone (NNH 63) compared with a NNH of 6
for olanzapine or quetiapine immediate-release and NNH
values ranging from 12 to 35 for the other antipsychotics;
NNH for somnolence was 9 for lurasidone, compared with
7 for olanzapine or quetiapine XR and 10 to 42 for the other
antipsychotics; NNH for akathisia was 9 for lurasidone compared to 15 and higher for the other antipsychotics; NNH
for discontinuation due to an AE was 30 for lurasidone,
compared with 41 and higher for the other antipsychotics
(18). When examining the data for the active controls in the
included lurasidone studies, the NNH vs. placebo for discontinuation because of an AE was 12 for haloperidol, 17 for
olanzapine and -28 for quetiapine XR (all ns). From product
labeling (1), there are no AEs associated with discontinuation in subjects treated with lurasidone that were at least 2%
and at least twice the placebo rate. For example, from the
data available at the time of drug approval, akathisia resulted
in discontinuation in 1.7% of patients treated with lurasidone vs. 0% for placebo (15). Of interest are the rates of discontinuation because of an AE in the study whose results
were not available at the time of drug approval and where
medication was administered in the evening (13): 3.3% for
lurasidone 160 mg/d, 4.0% for lurasidone 80 mg/d, 3.4% for
quetiapine XR 600 mg/d and 4.1% for placebo.
A major limitation of NNT and NNH is that these
metrics are limited to dichotomous outcomes. Other effect
size measures are necessary when describing continuous
outcomes, such as mean changes in PANSS scores or mean
changes in fasting plasma glucose levels (17, 19).
A caveat regarding the crude metabolic outcomes presented here is that they do not take into account individual
baseline risk. It is important to distinguish shifts from normal to abnormal values vs. from borderline to abnormal
values. Moreover, high-density lipoprotein cholesterol was
not examined. Although the blood for metabolic testing was
to be obtained fasting, confirmation of this was not always
available.
Data were obtained from carefully structured studies
and, hence, the results may not be generalizable to patients
outside the confines of a clinical trial. This is always a concern for results of registration trials because of the strict inclusion/exclusion criteria that these studies require. In addition, the studies were of a fixed-dose design; although this
can reveal information about potential dose response for efficacy or tolerability outcomes, a flexible-dose methodology
would better mirror clinical practice where the prescriber
would adjust the dose based on the individual person’s relief
of symptoms and/or the emergence of adverse events.
Long-term efficacy, safety and tolerability, including
maintenance of response are also important considerations
in the selection of antipsychotics. Dichotomous outcomes
from long-term trials, such as relapse and weight gain exceeding a predefined threshold, will be of interest in further
appraising the clinical profile of lurasidone.
Conclusions
Lurasidone is FDA-approved for the treatment of
schizophrenia based on a subset of the clinical trial data now
available. NNT and NNH can help quantify efficacy, safety
and tolerability outcomes and place lurasidone into clinical
perspective. Advantages for lurasidone include a low propensity for weight gain and metabolic abnormalities. The
five most consistently encountered adverse events attributable to lurasidone are akathisia, nausea, sedation, somnolence and parkinsonism, but NNH values are generally in
the double digits, reflecting an overall tolerable profile. Individual patient characteristics, values and preferences will
need to be considered when selecting lurasidone over other
antipsychotics.
Clinical Schizophrenia & Related Psychoses April 2012
• 9
Lurasidone NNT NNH
Disclosures
This study was funded by Sunovion Pharmaceuticals
Inc., Fort Lee, NJ. The analyses were conducted independently by the author using data provided by Sunovion Pharmaceuticals Inc.
In the past twenty-four months, Leslie Citrome has
engaged in collaborative research with or received consulting or speaking fees from: Alexza, Alkermes, AstraZeneca,
Avanir, Bristol-Myers Squibb, Eli Lilly, Janssen, Lundbeck,
Merck, Novartis, Noven, Otsuka, Pfizer, Shire, Sunovion and
Valeant.
References
1. Sunovion. Latuda (lurasidone). Revised October 2010. Available from: http://
www.latuda.com/LatudaPrescribingInformation.pdf [Last accessed 2011 May
24].
clinician should know how to calculate it. J Clin Psychiatry 2011;72(3):412413.
9. Citrome L. Quantifying risk: the role of absolute and relative measures in interpreting risk of adverse reactions from product labels of antipsychotic medications. Curr Drug Saf 2009;4(3):229-237.
10. Citrome L. Adjunctive aripiprazole, olanzapine, or quetiapine for major
depressive disorder: an analysis of number needed to treat, number needed
to harm, and likelihood to be helped or harmed. Postgrad Med 2010;122(4):
39-48.
11. Citrome L, Kantrowitz J. Antipsychotics for the treatment of schizophrenia:
likelihood to be helped or harmed, understanding proximal and distal benefits
and risks. Expert Rev Neurother 2008;8(7):1079-1091.
12. Straus SE. Individualizing treatment decisions. The likelihood of being helped
or harmed. Eval Health Prof 2002;25(2):210-224.
13. Loebel A, Cucchiaro J, Sarma K, Xu J, Hsu J, Kalali AH, et al. Lurasidone in
the treatment of acute schizophrenia: results of the double-blind, placebo-controlled, 6-week, PEARL 3 trial. Poster presentation NR06-38. American Psychiatric Association 164th Annual Meeting. Honolulu, HI, 2011 May 14-18.
2. Nakamura M, Ogasa M, Guarino J, Phillips D, Severs J, Cucchiaro J, et al.
Lurasidone in the treatment of acute schizophrenia: a double-blind, placebocontrolled trial. J Clin Psychiatry 2009;70(6):829-836.
14. Loebel A, Cucchiaro J, Ogasa M, Silva R, Pikalov AA, Hsu J, et al. Efficacy of
lurasidone: summary of results from the clinical development program. Poster
presentation II-45. NCDEU 50th Anniversary Meeting. Boca Raton, FL, 2010
June 14-17.
3. Meltzer HY, Cucchiaro J, Silva R, Ogasa M, Phillips D, Xu J, et al. Lurasidone
in the treatment of schizophrenia: a randomized, double-blind, placebo- and
olanzapine-controlled study. Am J Psychiatry 2011;168(9):957-967.
15. FDA. Drug Approval Package. Latuda (lurasidone hydrochloride) Tablets.
3 December 2010. http://www.accessdata.fda.gov/drugsatfda_docs/nda/
2010/200603Orig1s000TOC.cfm [Last accessed 2010 December 16].
4. Citrome L. Lurasidone for schizophrenia: a review of the efficacy and safety
profile for this newly approved second-generation antipsychotic. Int J Clin
Pract 2011;65(2):189-210.
16. Dainippon Sumitomo Pharma. Dainippon Sumitomo Pharma announces
Pan-Asia study results for lurasidone, an atypical antipsychotic agent, in the
treatment of schizophrenia. Press Release 2011 May 11. http://www.ds-pharma.com/news/pdf/ene20110511_3.pdf [Last accessed 2012 January 03].
5. Citrome L. Lurasidone for schizophrenia: a brief review of a new secondgeneration antipsychotic. Clin Schizophr Relat Psychoses 2011;4(4):251-257.
6. Citrome L. Iloperidone, asenapine, and lurasidone: a brief overview of 3 new
second-generation antipsychotics. Postgrad Med 2011;123(2):153-162.
7. Citrome L. Compelling or irrelevant? Using number needed to treat can help
decide. Acta Psychiatr Scand 2008;117(6):412-419.
8. Citrome L. Number needed to treat: what it is and what it isn’t, and why every
17. Kraemer HC, Kupfer DJ. Size of treatment effects and their importance to
clinical research and practice. Biol Psychiatry 2006;59(11):990-996.
18. Citrome L, Nasrallah HA. On-label on the table: what the package insert informs us about the tolerability profile of oral atypical antipsychotics, and what
it does not. Expert Opin Pharmacother 2001 Oct 22. [Epub ahead of print).
19. Citrome L. Relative vs. absolute measures of benefit and risk: what’s the difference? Acta Psychiatr Scand 2010;121(2):94-102.
Supplementary Appendix
Table S1
Responder Rates and NNT vs. Placebo-Pooled Data (Intent-to-Treat Population)
PANSS Reduction ≥20% PANSS Reduction ≥30%
PANSS Reduction ≥40%
PANSS Reduction ≥50%
N
n (%)
NNT
(95% CI)
n (%)
NNT
(95% CI)
n (%)
NNT
(95% CI)
n (%)
NNT
(95% CI)
Placebo
566
255 (45.1%)
NA
184 (32.5%)
NA
137 (24.2%)
NA
90 (15.9%)
NA
Lurasidone 20 mg/d
71
27 (38.0%)
-15 (ns)*
20 (28.2%)
-24 (ns)*
16 (22.5%)
-60 (ns)*
11 (15.5%)
-246 (ns)
Lurasidone 40 mg/d
353
198 (56.1%)
10 (6–23)
162 (45.9%)
8 (6–15)
124 (35.1%)
10 (6–21)
88 (24.9%)
12 (7–28)
Lurasidone 80 mg/d
402
249 (61.9%)
6 (5–10)
194 (48.3%)
7 (5–11)
140 (34.8%)
10 (7–21)
84 (20.9%)
21 (11–13,763)
Lurasidone 120 mg/d
288
175 (60.8%)
7 (5–12)
137 (47.6%)
7 (5–13)
93 (32.3%)
13 (7–62)
60 (20.8%)
21 (ns)
Lurasidone 160 mg/d
121
95 (78.5%)
3 (3–4)
76 (62.8%)
4 (3–5)
56 (46.3%)
5 (4–8)
35 (28.9%)
8 (5–23)
Haloperidol 10 mg/d
72
36 (50.0%)
21 (ns)
29 (40.3%)
13 (ns)
20 (27.8%)
28 (ns)
13 (18.1%)
47 (ns)
Olanzapine 15 mg/d
121
89 (73.6%)
4 (3–6)
78 (64.5%)
4 (3–5)
59 (48.8%)
5 (3–7)
41 (33.9%)
6 (4–12)
Quetiapine XR 600 mg/d
116
92 (79.3%)
3 (3–4)
82 (70.7%)
3 (3–4)
65 (56.0%)
4 (3–5)
33 (28.4%)
8 (5–27)
Intervention
Shaded entries represent NNT values where the 95% CI does not include infinity and, hence, considered statistically significant. CI=confidence
interval; NA=not applicable; NNT=number needed to treat; ns=not significant (the 95% CI contains “infinity”); PANSS=Positive and Negative
Syndrome Scale; XR=extended release. *A negative NNT denotes an advantage for placebo regarding the potential benefit.
10 •
Clinical Schizophrenia & Related Psychoses April 2012