3.2 Sensitivity Analyses
We performed a series of one-way sensitivity analyses to discover the effect of the different parameter’s uncertainty on the ICER. The Tornado diagram shows the effect of those parameters that led to a change in the ICER from big to small (Fig.2). In our model, the ICER was most sensitive to variations in OS of NAC and AC and, when the medial OS of NAC is 0.12 years more than AC, the NAC will be cost-effective at the 35,446 RMB threshold (Fig.3). The sensitivity analyses suggest that our model was robust enough to handle to the uncertainty surrounding the parameters of cost, base case probabilities, treatment time and utility weight.
In PSA, AC was the dominant strategy in 64.3% of 10,000 Monte Carlo simulations and, at the threshold of 35,446 RMB per QALY, there is a 54.1% probability that AC strategy is cost-effective. Fig.4 shows the corresponding cost effectiveness acceptability curve for NAC and AC. Therefore, the statement about basic simulation regarding incremental costs, effectiveness and the ICER, is stable.