Cost-Effectiveness of Surveillance Scanning Strategies after Curative Treatment of Non-Small-Cell Lung Cancer

Abstract

Background: After curative treatment of primary non-small-cell lung cancer (NSCLC), patients undergo intensive surveillance with the aim to detect recurrences from the primary tumor or metachronous second primary lung cancer as early as possible and improve overall survival. However, the benefit of surveillance is debated. Available evidence is of low quality and conflicting. Microsimulation modeling facilitates the exploration of the impact of different surveillance strategies and provides insight into the cost-effectiveness of surveillance.

Methods: A microsimulation model was used to simulate a range of computed tomography (CT)-based surveillance schedules, differing in the frequency and duration of CT surveillance. The impact on survival, quality-adjusted life-years, costs, and cost-effectiveness of each schedule was assessed.

Results: Ten of 108 strategies formed the cost-effectiveness frontier; that is, these were the strategies with the optimal cost-health benefit balance. Per person, the discounted QALYs of these strategies varied between 5.72 and 5.81 y, and discounted costs varied between €9892 and €19,259. Below a willingness-to-pay threshold of €50,000/QALY, no scanning is the preferred option. For a willingness-to-pay threshold of €80,000/QALY, surveillance scanning every 2 y starting 1 y after curative treatment becomes the best option, with €11,860 discounted costs and 5.76 discounted QALYs per person. The European Society for Medical Oncology guideline strategy was more expensive and less effective than several other strategies.

Conclusion: Model simulations suggest that limited CT surveillance scanning after the treatment of primary NSCLC is cost-effective, but the incremental health-benefit remains marginal. However, model simulations do suggest that the guideline strategy is not cost-effective.

Keywords: CT-scan; cost-effectiveness analysis; non-small cell lung-cancer; surveillance.

Figure 1

The disease model (ovals) interacts with the clinical pathway (rectangles). Death states are shown with triangles. This combination determines the timing of detection on a scan or symptomatic detection of metastases and second primary tumors. There are 4 parallel chains that operate simultaneously: 1) recurrences of the primary tumor, 2) the hazard of developing a second primary tumor (SPLC), 3) the hazard of becoming ineligible for surgery or chemotherapy, and 4) the hazard of death from other causes (DOC). As a result, patients can for instance simultaneously have recurrences, and a second primary tumor, and be untreatable. Patients can also die of cancer (DOD), or from the treatment (excess mortality). Death states are mutually exclusive. Example life histories are shown in Appendix 1.1.

Figure 2

Cost-effectiveness plane and frontier (using Dutch discounting rates of 4% for costs and 1.5% for effects). The cost-effectiveness frontier (line) connects all potentially cost-effective strategies that dominate all other strategies directly (squares). All other strategies are either dominated (circles), meaning that they are more expensive and less effective than a strategy on the frontier, or subject to extended dominance (triangles), meaning that a combination of strategies on the frontier can be found that leads to higher effectiveness at the same costs or lower costs at equal effectiveness. The European Society for Medical Oncology guideline strategy (cross) is also dominated. The numbers of the strategies on the frontier correspond to the numbering of the strategies shown in Table 2.

Figure 3

Graphical representations of the output from the probabilistic sensitivity analysis. (A) The cost-effectiveness acceptability curve (CEAC) shows for each strategy the proportion of simulations in which this strategy has the highest net monetary benefit (NMB), given the willingness to pay (WTP) in €/quality-adjusted life-year on the x-axis. (B) The expected loss curve shows for each strategy the expected difference between the NMB of that strategy and the maximum achieved NMB as a function of the WTP threshold. The frontier (dashed black line) follows the CEAC and expected loss curve of the strategy that has the highest expected NMB at each WTP and fully corresponds to the cost-effectiveness frontier in Figure 2. The vertical dashes on the x-axis correspond to the incremental cost-effectiveness ratios in Table 2 and depict the optimal strategy (numbers).