In search of the most cost-effective monitoring strategy for vestibular schwannoma: A decision analytical modelling study

Abstract

Objectives: To assess the cost-effectiveness of frequently used monitoring strategies for vestibular schwannoma (VS).

Design: A state transition model was developed to compare six monitoring strategies for patients with VS: lifelong annual monitoring; annual monitoring for the first 10 years after diagnosis; scanning at 1-5, 7, 9, 12, 15 years after diagnosis and subsequently every 5 years; a personalised monitoring strategy for small and large tumours; scanning at 1, 2 and 5 years after diagnosis and no monitoring. Input data were derived from literature and expert opinion. Quality-adjusted life years (QALYs) and healthcare costs of each strategy were modelled over lifetime. Net monetary benefits (NMBs) were calculated to determine which strategy provided most value for money. Sensitivity analyses were performed to address uncertainty.

Results: Omitting monitoring is least effective with 18.23 (95% CI 16.84-19.37) QALYs per patient, and lifelong annual monitoring is most effective with 18.66 (95% CI 17.42-19.65) QALYs. Corresponding costs were €6526 (95% CI 5923-7058) and €9429 (95% CI 9197-9643) per patient, respectively. Lifelong annual monitoring provided the best value with a NMB of €363 765 (339 040-383 697), but the overall probability of being most cost-effective compared to the other strategies was still only 23%. Sensitivity analysis shows that there is large uncertainty in the effectiveness of all strategies, with largely overlapping 95% confidence intervals for all strategies.

Conclusions: Due to the largely overlapping 95% confidence intervals of all monitoring strategies for VS, it is unclear which monitoring strategy provides most value for money at this moment.

Keywords: acoustic neuroma; cost-effectiveness analysis; magnetic resonance imaging; monitoring; vestibular schwannoma; wait and scan.

Figure 1

Influence diagram of the Markov model. Patients could enter the model via one of the Koos states in the monitoring strategy. Koos 1 corresponds to an intracanalicular VS, Koos 2 to an extracanalicular VS without brainstem contact, Koos 3 to VS with brainstem contact and Koos 4 corresponds to VS that compresses the brainstem. When tumour growth was present, patients entered the next Koos state. In case of Koos state 3 and 4, patients exited the monitoring strategy when tumour growth was detected on MRI. Leaving the monitoring strategy meant transition to one of the treatment options; stereotactic radiosurgery (SRS) or microsurgery. After treatment, patients were monitored for tumour growth. If tumour growth was detected after treatment, patients could receive additional treatment. The health state “dead” is not displayed, but could be entered from all health states

Figure 2

Outcomes of the probabilistic sensitivity analysis. This analysis quantifies the level of confidence of the model's conclusions. All six monitoring strategies are displayed. Every dot represents the outcome of one analysis

Figure 3

Cost‐effectiveness acceptability curve. This graph shows the probability that one of the strategies is most cost‐effective for different willingness to pay values. The willingness to pay represents an estimate of what we might be prepared to pay for the health benefit