ICE BOX PLOTTER
INCREMENTAL COST EFFECTIVENESS
(ICE)
NNT (Number Needed to Treat)
and CNT (Cost Needed to Treat)
Estimating Incremental
Cost Effectiveness Ratios (ICE Ratios)
with Consideration of Confidence
Intervals
Beta Version - for Testing
Only.
This calculator examines various clinical
and economic elements of treatment:
1) Number Needed to Treat (NNT) to attain
one additional good outcome.
2) Cost of treatment, upper and lower
limits.
3) Approximating Incremental cost-effectiveness
ratio (ICE Ratio) of treatment.
4) Confidence intervals considerations
on NNT and Cost Needed to Treat (CNT).
The relationship between treatment and
outcome is represented
in a 2 by 2 table, from which can be
calculated several
measures of association.
[You can change values in white boxes and hit "calculate"]
Clinical Data (enter
values in blue boxes)
Economic Data (enter values in green
boxes)
FAILURE
SUCCESS
Average Cost
Low Cost
High Cost
Per Patient
Per Patient
Per Patient
a+b
CONTROL GROUP
a
b
..
CONTROL
$
$
$
c+d
TREATMENT GROUP
c
d
..
TREATMENT
$
$
$
a+c
b+d
a+b+c+d
Incremental Cost Per Patient
RELATIVE RISK = RR =[a/(a+b)]/[c/(c+d)] >>>>
RR =
..
(Risk Ratio, Relative Risk)
ODDS RATIO >>>> OR = ad/bc
=
OR =
..
RELATIVE RISK REDUCTION
RRR = ( c/(c+d) - a/(a+b))/ (c/(c+d)) >>>>
RRR =
..
RATE DIFFERENCE = RD
RD or AR = c/(c+d)-a/(a+b)
(Absolute Risk Change)
Failure Rate
Success Rate
Outcome rates change
from >>>
To >>>>
RD (AR)
Upper Value
Lower Value
Due to treatment
group, for a change of >>>>>
= RD (or AR) =
Upper and Lower values of RD above are
based on 95% Confidence Interval = +/- =
..
= +/- Confidence Interval
=1.96*(SQRT(((a/a+b)*(1-(a/a+b))/(a+b))+((c/c+d)*(1-(c/c+d))/(c+d))))
NUMBER
NEEDED TO TREAT TO OBTAIN ONE EXTRA "SUCCESS".
NNT
Upper Value
Lower Value
NNT = 1/RD = 1/AR =
= 1/(c/(c+d)-a/(a+b)) =
= NNT =
..
..
..
COST NEEDED
TO TREAT TO OBTAIN ONE EXTRA "SUCCESS"
CNT
Upper Value
Lower Value
CNT = Incremental cost per patient *
NNT >>>>
= CNT =
$
$
$
Cost needed to treat to obtain one extra
"success" outcome,
which roughly estimates an Incremental
Cost Effectiveness (ICE Ratio).
========================
Notes: A negative NNT value can indicate
that there is a chance that some patients could be harmed by the treatment.
If the upper and lower value of RD (AR)
passes through zero, then the NNT passes through infinity
and the difference in the two clinical
groups would not be considered statistically significant.
The default values entered in the clinical
boxes above, are roughly based on the percentages and therapy costs for
treating myocardial infarction patients
in the classic TPA vs. Streptokinase paper by D. Mark, NEJM 1995.
For simplicity, this ICE Box example
uses only 200 patients, while the Mark paper sample size total, based
on
the GUSTO study was 41,000 patients for
survival analysis and about 17,000 cases for economic analysis.
When a total of 20,000 patients are entered
into the clinical boxes above with the same success and failure rates,
In the Mark paper, patients successfully
treated were expected to live at least 5 years,
and the resulting incremental
cost-effectiveness ratio was about $40,000 per life year saved.
This published result is similar to our
default CNT divided by 5 years.
Thus if a larger numbers
of patients are entered into the default clinical boxes, the confidence
intervals should be more narrow.
The CNT values for one additional "Success"
outcome are similar to the incremental cost effectiveness ratio (ICE Ratio)
since
the ICE numerator and denominator are
related to the incremental costs divided by the incremental effects.
In this example, it is assumed that the
lower NNT limits had the lower costs and high NNT limits had higher costs,
but this may not always be the case.
You can change the entered cost values to explore different results.
.
========================
-- Some of the NNT calculations are based
on materials developed by R. Hamm.
-- Confidence Interval calculations are
based on the method discussed
in the EBM Bandolier bulletin at www.jr2.ox.ac.uk/bandolier/band18/b18-9.html
