Donepezil, rivastigmine, galantamine and memantine for ...

120
Economic analysis
The model structure has been informed by the
systematic review of the literature on the costeffectiveness
of pharmaceuticals for AD and on the
available methods for modelling disease
progression in AD (see discussion above) and
discussion with physicians involved in the
treatment of AD. The model structure used is
based on the AHEAD model presented by Caro
and colleagues 131 (discussed above), which has
been used in the published literature to estimate
the cost-effectiveness of galantamine. Figure 19
describes the simple model structure. The
approach applies the predictive risk equation from
the AHEAD model to determine a monthly hazard
for progression of disease to a point where
patients require FTC. This hazard rate has been
used in a disease progression model, together with
mortality data for AD patients, to model over time
the experiences of a cohort of AD patients. The
cohort analysis predicts the proportion of patients
that will be located in the three possible health
states (pre-FTC, FTC and death) at each monthly
cycle (i.e. a time horizon of 60 months). Costs and
benefits have been allocated to each of these
health states, and the analysis compares disease
progression and subsequent costs and benefits
over time to estimate the cost-effectiveness of the
intervention.
The AHEAD model presented by Caro and
colleagues 131 has been described above. The
model is based on the progression of disease to a
Drug therapy
(improved ADAS-cog)
COHORT OF
AD PATIENTS
Costs outcomes
cost-effectiveness?
Usual care
(baseline ADAS-cog)
Pre-FTC Pre-FTC
Death FTC FTC Death
FIGURE 19 Diagrammatic representation of SHTAC model/approach
point where patients require FTC; where they have
a requirement for a significant amount (for the
greater part of the day) of paid care and
supervision each day, regardless of the location of
care (institution or community setting) or who
provides the care. Given the concerns in the
literature over the use of cognition alone to model
disease progression over time, and based on
discussions with physicians, it was felt that of the
methods available, the use (in our modelling
approach) of the AHEAD methodology and the
endpoint FTC was preferable to the use of transit
probabilities from trial data for health states
described solely by cognitive scores (i.e. MMSE
scores). The statistical techniques used by Stern
and colleagues 132 and thereafter Caro and
colleagues 131 to determine the predictive risk
equations used in the AHEAD model are subject
to a number of concerns, but the method could be
used to illustrate the potential progression of AD
over time to inform on the current consideration
of the cost-effectiveness of interventions.
The predictive risk equation for FTC (used in the
SHTAC model) has a two-stage process to
calculate a monthly hazard (risk) of patients
entering FTC from the starting health state of pre-
FTC. A Cox proportional hazards model is used to
calculate a risk index, and this has coefficients for
the presence of EPS, the presence of psychotic
symptoms, a young age at disease onset (i.e.