Intended for healthcare professionals

Rapid response to:

Primary Care

Coronary heart disease prevention: insights from modelling incremental cost effectiveness

BMJ 2003; 327 doi: https://doi.org/10.1136/bmj.327.7426.1264 (Published 27 November 2003) Cite this as: BMJ 2003;327:1264

Rapid Response:

Tables of Marshall's article

The cost-effectiveness study published by Marshall
(BMJ 2003;327:
1264) is very difficult to
understand in its technical details not only for the average reader but also
for people with some expertise in pharmacoeconomics. In our view, the part of this
study that presents the data of the average cost-effectiveness (Table 1 of the
article) is useless because the practical implications of this study, along
with the assessment of the pharmacoeconomic profile of the various drugs, rely
exclusively on the findings of the incremental cost-effectiveness analysis
(i.e. the data presented in Table 2).

On the other hand, Table 2 has a drawback in that
it is not adequately explained by the author. For this reason, we propose herein
a more detailed printout of this table (see below) in which much more
information is presented and the labelling of each row is made much more
explicit.

 




 REARRANGEMENT OF TABLE 2

 

Treatments

being

compared

in the

cost

effectiveness

assessment

 

 

 

 

Undiscounted

event rate

with the

first

treatment

 

 

 

C1

Relative

risk (RR)

of the

second

treatment

vs placebo

 

 

C2

Undiscounted

absolute event

rate with the

second treatment

 

 

 

 

C3 =
C1*C2

Discounted

absolute event

rate with the

first treatment

 

 

 

 

C4#

Discounted

absolute event

rate with the

second treatment

 

 

 

 

C5#

 

Undiscounted

absolute

gain in the

event rate

 

 

 

 

C6=C1-C3

Discounted

absolute

gain in the

event rate

 

 

 

 

C7=C4-C5

 

Discounted

cost

per patient

with

the first

treatment

(pounds)

 

C8#

 

Discounted

cost

per patient

with

the second

treatment

(pounds)

 

C9#

Discounted

cost

effectiveness

ratio

(pounds per

event averted)

 

 

C10=
(C9-C8)/C7

 

P

vs A

0.1

0.75

0.075

0.09272165

0.069541

0.025

0.023180413

0

85

3666.88

A

vs A+DB

0.075

0.83

0.06225

0.069541238

0.057719227

0.01275

0.01182201

85

228

12096.08

A+DB

vs
A+DB+E

0.06225

0.81

0.0504225

0.057719227

0.046752574

0.0118275

0.010966653

228

602

34103.38

A+DB+E

vs
A+DB+E+S

0.0504225

0.69

0.034791525

0.046752574

0.032259276

0.015630975

0.014493298

602

2385

123022.37

#Discount rates are 6% per year for costs and 1.5%
per year for benefits. Abbreviations: P = placebo; A = aspirin; DB = diuretic
plus

beta-blocker; E = enalapril; S = statin; all
event rates (normalised to 1 patient) and all costs (normalised to 1 patient
as well) are at

5 years.

NOTE: Each value of RR refers to the adjunctive
drug (considered in each individual row of the table) in comparison with
placebo.

There seems to be a contradiction in this table
because the RRs (obtained from either clinical studies or meta-analyses)
compare the

drug under examination (adjunctive drug) with
placebo, whereas the cost-effectiveness assessment in the same row compares
the

combination with the absence of the adjunctive
drug vs the combination with the presence of the adjunctive drug (with no
comparison

vs placebo). Hence, the comparators used for the
RR assessment (i.e. adjunctive drug vs placebo) differ from those used for
the

cost-effectiveness comparison (i.e. absence vs
presence of the adjunctive drug). However, this contradiction is only
apparent

because RRs are applied in a chain of subsequent
multiplications throughout the whole table. For example, if the
cost-effectiveness

comparison of A vs A+DB appears together with the
RR of P vs DB (RR=0.83), there is actually no error in this strategy because

the relative risk of A vs P (RR=0.75) has already
been incorporated as a multiplicative factor in the previous row and its
presence

affects all the subsequent rows of the table
(including the one that compares A vs A+DB).

 

 

Competing interests:
None declared

Competing interests:  

07 December 2003
Andrea Messori
coordinator
Benedetta Santarlasci
Laboratorio SIFO di Farmacoeconomia, Careggi University Hospital, 50134 Firenze, Italy