Annual probability of relapse   Distribution based on 10,000
MTC iterations 95% credible
intervals  
Olanzapine  0.1996  0.0146 to 0.7222  MTC competing risks
model  analysis of data included in the guideline
systematic review; results for 52 weeks assumed to reflect
annual probability; results for placebo assumed to apply to
no treatment in all years except the 1^{st} year
following the move to no treatment 
Amisulpride  0.2988  0.0197 to 0.9042 
Zotepine  0.1067  0.0023 to 0.5601 
Aripiprazole  0.2742  0.0130 to 0.8531 
Paliperidone  0.1625  0.0025 to 0.7008 
Risperidone  0.2761  0.0182 to 0.8785 
Haloperidol  0.3317  0.0262 to 0.9028 
No treatment –
following years  0.4361  0.0913 to 0.8613 
Flupentixol decanoate  0.2977  Beta distribution
(α=39, β=92
according to data reported in David and colleagues, 1999  David et al.,
1999. Metaanalysis of trials comparing
flupentixol decanoate versus other depot antipsyhcotics;
data on relapse 
No treatment – first
year following discontinuation of treatment  0.6062  Distribution based on 10,000
MTC iterations– results for placebo,
adding the effect of abrupt discontinuation on the risk for
relapse (Viguera
et al., 1997)  MTC competing risks model  a
higher probability of relapse over the first 7 months
(50%) was taken into account (Viguera et
al., 1997) 
Probability of discontinuation due
to intorelable side effects  1^{st} year of
initiation of a particular antipsychotic   Distribution based on 10,000
MTC iterations
95%
credible intervals  
Olanzapine  0.0783  0.0021 to 0.4784  MTC competing risks
model  analysis of data included in the guideline
systematic review; results for 52 weeks assumed to apply to
the 1^{st} year within initiation of a particular
antipsychotic only 
Amisulpride  0.0554  0.0006 to 0.3721 
Zotepine  0.3821  0.0120 to 0.9750 
Aripiprazole  0.1582  0.0026 to 0.7847 
Paliperidone  0.3287  0.0039 to 0.9770 
Risperidone  0.0994  0.0020 to 0.6471 
Haloperidol  0.0922  0.0017 to 0.5386 
Annual probability of
discontinuation due to other reasons   Distribution based on 10,000
MTC iterations 95% credible
intervals  
Olanzapine  0.2730  0.0207 to 0.8596  MTC competing risks
model  analysis of data included in the guideline
systematic review; results for 52 weeks assumed to reflect
annual probability 
Amisulpride  0.2435  0.0139 to 0.8324 
Zotepine  0.2253  0.0074 to 0.8189 
Aripiprazole  0.3520  0.0202 to 0.9218 
Paliperidone  0.3848  0.0090 to 0.9479 
Risperidone  0.1761  0.0086 to 0.7141 
Haloperidol  0.2516  0.0151 to 0.8290 
Weight gain  1^{st} year
of initiation of a particular antipsychotic    
 
Distribution based on 10,000 MTC iterations
 
Odds Ratios versus haloperidol
  95% credible
intervals  MTC simple random
effects model  analysis of data from guideline meta
analysis of side effects; only data reported as
“increase in weight gain of
≥7% from baseline” were
considered. 
Olanzapine  2.8631  1.7050 to 4.5090 
Amisulpride  1.8604  0.7345 to 4.0360 
Aripiprazole  0.7373  0.3498 to 1.3990 
Paliperidone  1.0779  0.4405 to 2.1640 
Risperidone  1.0895  0.5214 to 2.0850 
Zotepine  1.0895  As for risperidone  Odds ratio of zotepine versus
haloperidol assumed to be equal of that of risperidone
versus haloperidol 
Probability of weight gain
   
Haloperidol  0.2000  Beta distribution
(α=31, β=124
according to data reported in studies with time horizon up
to 12 weeks included in the guideline meta analysis of side
effects)  Extrapolation of data reported in
studies with time horizon up to 12 weeks included in the
guideline metaanalysis of side effects; only data reported
as “increase in weight gain of
≥7% from baseline” were
considered. 
Flupentixol decanoate  0.2000  As for haloperidol  Assumed to equal that for
haloperidol 
Acute EPS    
1^{st} year of initiation
of a particular antipsychotic    
 
Distribution based on 10,000 MTC iterations
 
Odds Ratios versus haloperidol
  95% credible
intervals  
Olanzapine  0.2631  0.1832 to 0.3641  MTC full random
effects model  analysis of data from guideline
metaanalysis of side effects; only data on
“need for anticholinergic
medication” were considered 
Amisulpride  0.3993  0.2587 to 0.5836 
Zotepine  0.1476  0.0517 to 0.3132 
Aripiprazole  0.2517  0.1505 to 0.4002 
Paliperidone  0.2983  0.1179 to 0.6214 
Risperidone  0.4743  0.3680 to 0.5994 
Probability of acute EPS
   
Haloperidol  0.5367  Beta distribution
(α=928,
β=801 according to data reported in
RCTs with time horizon up to8 weeks included in the
guideline meta analysis of side effects)  Extrapolation of data reported in
studies with time horizon up to 8 weeks included in the
guideline metaanalysis of side effects; only data on
“need for anticholinergic
medication” were considered 
Flupentixol decanoate  0.4891  Beta distribution
(α=45, β=47
according to data reported in David and colleagues, 1999)  David et al.,
1999. Metaanalysis of trials comparing
flupentixol decanoate versus other depot antipsyhcotics;
data on need for anticholinergic medication 
Following years    
Probability of acute
EPS  all antipsychotics  10% of 1^{st}
year estimate  N/A (no distribution
assigned)  GDG expert opinion 
Probability of diabetes 
1^{st} year of initiation of a particular
antipsychotic  
Distribution based on 10,000 MTC iterations of data on
weight gain
 Probability of
haloperidol estimated from data reported in van Winkel
et al., 2006 &2008 and
considering the increased relative risk (RR) for diabetes of
SGAs versus FQAs; rest probabilities calculated by
multiplying respective RRs for weight gain of each SGA
versus haloperidol by the probability of diabetes for
haloperidol 
Olanzapine  0.0417  Relative risk of each
SGA versus haloperidol for diabetes was assumed to equal
their in between relative risk for weight gain; the latter
was determined by the posterior distribution of ORs of
weight gain for each SGA and haloperidol, respectively 
Amisulpride  0.0317 
Zotepine  0.0214 
Aripiprazole  0.0156 
Paliperidone  0.0212 
Risperidone  0.0214 
Haloperidol  0.0200  Beta distribution
(α=2, β=98
based on assumption) 
Flupentixol decanoate  0.0200  As for haloperidol 
Probability of glucose intolerance
 1^{st} year of initiation of a particular
antipsychotic  
Distribution based on 10,000 MTC iterations of data on
weight gain
 Probability of
haloperidol estimated from data identified in the guideline
systematic review; rest probabilities calculated by
multiplying respective RRs for weight gain of each SGA
versus haloperidol by the probability of glucose intolerance
for haloperidol 
Olanzapine  0.3129  Relative risk of each
SGA versus haloperidol for glucose intolernace was assumed
to equal their inbetween relative risk for weight gain; the
latter was determined by the posterior distribution of ORs
of weight gain for each SGA and haloperidol,
respectively 
Amisulpride  0.2381 
Zotepine  0.1606 
Aripiprazole  0.1167 
Paliperidone  0.1592 
Risperidone  0.1606 
Haloperidol  0.1500  Beta distribution
(α=15, β=85
based on assumption) 
Flupentixol decanoate  0.1500  As for haloperidol 
Annual transition probability of
impaired glucose tolerance to diabetes  0.0196  Beta distribution
Standard error 0.0025 (Gillies et al.,
2008)  Gillies et al.,
2008 
Annual probability of diabetes
complications   Beta
distribution
Determined from the numbers of people experiencing
each of the complications at each level of Haemoglobin
A_{1C} concentration in the UKPDS (Stratton
et al., 2000)  Based on UKPDS data
(Stratton
et al., 2000), assuming that
20% of people with schizophrenia and diabetes in
the model had Haemoglobin A_{1C} concentration
7% to<8%, 30% of
people had 8% to <9%,
30% of people had 9% to
<10% and 20% of people had
≥ 10% 
Fatal myocardial infarction  0.0042 
Nonfatal myocardial
infarction  0.0130 
Nonfatal stroke  0.0039 
Amputation  0.0023 
Macrovascular eventsheart
failure  0.0040 
Microvascular eventsischaemic
heart disease  0.0157 
Standardised Mortality Ratio  all
cause mortality  2.6  N/A (no distribution
assigned) 
McGrath
et al., 2008

Mortality rates per 1000 people in
general population by age  25–34 yrs:
0.69 35–44 yrs:
1.29 45–54 yrs:
3.10 55–64 yrs:
7.53 65–74 yrs:
20.48 75–84 yrs:
59.36 ≥85 yrs: 164.02  N/A (no distribution
assigned)  Office for National Statistics,
2008; mortality rates for England and Wales,
2005, estimated based on a male to female ratio 1.4 to 1,
characterising people with schizophrenia (McGrath,
2006). 
Utility scores  
Beta distribution
 
Model health states    
Remission  0.799  Determined using the
reported numbers of people valuing each PANSSgenerated
health state as in Lenert and colleagues (2004)  Lenert et
al., 2004; linking between model
states and states described in the study based on GDG
estimates – see text for details. Duration of
decrement in HRQoL caused by relapse: 6 months 
Relapse  0.670 
Death  0.000 
Side effects    
Acute EPS  −0.888%  Estimated from the
number of people valuing the presence of each side effect,
as reported in Lenert and colleagues (2004)  Lenert et
al., 2004; acute EPS causes HRQoL
reduction corresponding to that of pseudo Parsokinism,
lasting 3 months; weight gain causes permanent reduction in
HRQoL 
Weight gain  −0.959% 
Diabetes complications    
Myocardial infarction  −0.055  95% CIs:
−0.067 to −0.042  Clarke et
al., 2002; utility scores based on
patientreported EQ5D scores, valued using EQ5D UK tariff
values 
Stroke  −0.164  95% CIs:
−0.222 to −0.105 
Amputation  −0.280  95% CIs:
−0.389 to −0.170 
Macrovascular events  heart
failure  −0.108  95% CIs:
−0.169 to −0.048 
Microvascular events  ischaemic
heart disease  −0.090  95% CIs:
−0.126 to −0.054 
Annual drug acquisition costs
(remission state)   N/A (no distribution
assigned)  BNF 56 (British Medical
Association & the Royal Pharmaceutical Society
of Great Britain, 2008), except risperidone cost,
which was taken from the Electronic Drug Tariff (NHS, Business
Services Authority, 2008). Average daily dosage
taken from respective NHS data (NHS, The Information Centre,
2008C) and BNF guidance when no other data were
available. 
Olanzapine  £1,036 
Amisulpride  £696 
Zotepine  £767 
Aripiprazole  £1,325 
Paliperidone  £1,902 
Risperidone  £821 
Haloperidol  £175 
Flupentixol decanoate  £81 
Annual costs of remission   Gamma
distribution Standard error of all costs:
70% of mean value (assumption)  Details on outpatient,
primary and community care cost reported in ; details
on costs of residential and longterm hospital care reported
in ;
2007 prices 
Outpatient, primary &
community care  £5,401 
Residential and longterm hospital
care  £7,325 
Total
(cost of antipsychotic medication for relapse
prevention excluded)  £12,726 
Annual costs of relapse   Gamma
distribution Standard error of all costs:
70% of mean value (assumption)  Details on outpatient,
primary and community care cost reported in ; details
on costs of treating acute episode reported in ; details
on costs of residential and longterm hospital care reported
in ;
2007 prices 
Outpatient, primary &
community care  £4,323 
Residential and longterm hospital
care  £5,421 
Acute treatment (including
olanzanpine)  £23,274 
Total
(cost of antipsychotic medication for relapse
prevention excluded)  £33,018 
Cost of switching between
antipsychotics  £435  Gamma
distribution Standard error: 70%
of mean value (assumption)  3 visits to consultant
psychiatrist, lasting 20 min each; unit cost from Curtis,
2007 (2007 price) 
Cost of treating side effects   Gamma
distribution Standard error of all costs:
70% of the respective mean value
(assumption)  Details on resource
use and unit costs associated with acute EPS and weight gain
reported in ; 2007 prices
UKPDS (Curtis
et al., 2005); 2007 prices 
Acute EPS  £177 
Weight gain  £117 
Diabetes (without complications)
– annual  £199 
Fatal myocardial infarction  £1,531 
Nonfatal myocardial
infarction  
first year/following
years  £5,407/£616 
Nonfatal stroke  
first year/following
years  £3,144/£331 
Amputation  
first year/following
years  £11,238/£401 
Macrovascular eventsheart
failure  
first year/following
years  £418/£343 
Microvascular eventsischaemic
heart disease  
first year/following
years  £363/£271 
Discount rate (for
both costs and outcomes)  0.035  N/A (no distribution
assigned)  Recommended by NICE (NICE,
2008A) 