Page 91 - Htain Manual

assess the minimum coverage of an immunization or a screening program necessary to

maintain the cost-effectiveness of the program.

One type of multi-variate analysis is scenario analysis. Under this, a range of scenarios

is constructed based on the set of certain circumstances across parameters and is compared
with the ‘base’ case analysis. For example, the worst and best case scenarios comprise of

those extreme circumstances where parameter values lead to the highest and lowest ICERs

respectively.

In probabilistic sensitivity analysis (PSA), all the uncertain parameters are varied as per

pre-defined distributions by specific statistical means – for example using Monte Carlo

Simulations. (2) Appropriate distributions are needed to be used for each of the uncertain

modelling parameters depending upon the type and nature of the variable. In comparison to
deterministic analysis, PSA allows us to quantify the level of confidence in the output of the

analysis, in relation to uncertainty in the model inputs. In PSA, the first step is to convert point

estimates of each of the parameter values into distributions. Once the distributions has been
assigned, a number of simulations are run (say 10,000), which allow sampling from the various

distributions. For each simulation, different values are picked from the distributions, and

many different ICER values are calculated. Finally, a mean or median ICER value is reported

th
th
along with 95% confidence interval or 2.5 and 97.5 percentile respectively. This gives a
better representation of the result as we get the conclusive range in which the ICER will fall
with statistical significance.

References
1. Hutubessy R, Chisholm D, Edejer TT. Generalized cost-effectiveness analysis for national-
level priority-setting in the health sector. Cost Eff Resour Alloc. 2003 Dec 19;1(1):8.
2. Doubilet P, Begg CB, Weinstein MC, Braun P, BJ. M. Probabilistic sensitivity analysis using
Monte Carlo simulation. A practical approach. Medical Decision Making. 1985;5:157-77.

79