Modeling structure
In this study, a decision tree model was built to compare the cost-effectiveness of PES vs MES in VATS lobectomies from the perspective of China hospital administrators. Our decision tree structure was shown in Figure 1. The modeled population was composed of adult Chinese lung cancer patients requiring VATS lobectomies in tertiary hospitals in China. The two interventions compared were VATS lobectomies using two different types of endoscopic staplers: PES and MES. The model assumes the surgery team had the same levels of skill and experience regardless of the choice of staplers. Prolonged air leaks are the most prevalent and costly surgery complications for VATS lobectomies. Our model explicitly tracked the incidence of PAL events. In addition to calculating total costs for the economic outcome, the 6-month re-operation rates post-discharge were measured for the clinical effectiveness outcome.
The entire evaluation period of the cost-effectiveness analysis (CEA) was less than 7 months, including both the initial surgery period and the post-discharge follow-up period. Discounting would not create any material impact for this analysis. No discounts for cost or outcome were explicitly considered. The final CEA analysis decision was based on the incremental cost-effectiveness ratio (ICER), expressed as cost per reduction of re-operation. The willingness to pay (WTP) threshold for ICER values in this analysis is set to be the average total costs of a single reoperation event.
The decision tree was implemented using TreeAge Pro (version 2014)(7). The deterministic analysis was performed for the base case scenario. One-way sensitivity analysis (OWSA) and probabilistic sensitivity analysis (PSA) were also used to evaluate the stability of model results against input variable uncertainty.
Clinical Inputs
Two types of clinical inputs were used for the CEA model to capture the peri-operative and short-term efficacies. The peri-operative efficacy of staplers in VATS lobectomies was measured by PAL rates reported in a published clinical study. A head-to-head comparison of PAL rates among one type of MES and two types of PES was reported by Xiao et al in 2019(3). The outcomes reported by Xiao et al were based on retrospective study of VATS lobectomy surgeries performed by a single surgeon in a large tertiary hospital in China. We combined the two PES data sets as the efficacy input for PES in our CEA model. The final PAL rates were 25.3% and 6.4% for MES and PES, respectively. Table 1 details the sample size and PAL cases. Beta distributions were used to model the probabilities of peri-operative PAL for the two intervention strategies in PSA.
The short-term efficacies of the staplers were measured using 6-month post-discharge re-operation rates. Electronic medical records (EMR) retrieved from 20 tertiary hospitals in China were used to estimate the six-month reoperation rates. The participating hospitals were selected based on their geographic location: five in the north (Beijing, Jinan, Shenyang), 3 in the south (Guangzhou, Fuzhou, Kunming), 4 in the east (Shanghai, Nanjing), 4 in the west (Chongqing, Lanzhou, Wulumuqi), and 4 in the middle (Wuhan, Xi’an). Patient inclusion criteria are: 1) 18 years or older, 2) diagnosed with lung cancer, 3) received a VATS lobectomy, and 4) surgery times were between 1/1/2018-6/30/2018. Robotic surgeries were excluded from the analysis. Patients with incomplete medical records or received more than one surgery in the same episode of care were also excluded. Eligible patients were divided into two groups based on the occurrence of PAL in the peri-operative period. Post-discharge re-operation rates were calculated separately for the PAL and no PAL groups.
Cost Inputs
Direct costs were derived from the same real-world EMR data set described above and followed the same inclusion/exclusion criteria. Costs were reported using Chinese Yuan (CNY). Except for the stapler and cartridge costs, all other cost items in PSA were assumed to have Gamma distributions defined by the observed means and standard deviations.
The peri-operative costs included all inpatient costs such as pharmacy costs, anesthesia costs, lab costs, medical service costs, etc. Both the means and standard deviations for total costs and categorical costs are reported in Table 2. The stapler and cartridge costs were calculated separately based on the prices listed on the government-sponsored hospital procurement platform. The model assumed each VATS lobectomy used one PES with 6.5 cartridges or one MES with 6.94 cartridges based on the average stapler to cartridge ratio per VATS lobectomy from a published real-world retrospective study reported by Xiao et al. The detailed product information and the costs of staplers and cartridges are reported in Table 3.
Readmission for re-operation is rare. We only observed 10 cases of re-operation in the EMR dataset. We calculated the mean and standard deviation of the total readmission costs based on all 10 cases without further stratifying them by peri-operative PAL status.