Cost Effectiveness of Lung Cancer Screening With Low-Dose CT in Heavy Smokers in China

Lung Cancer has the second-highest incidence rate and the highest mortality rate globally (1). New cases and deaths related to lung cancer in China account for 37.3% and 39.4%, respectively, in the whole world while the 5-year survival rate is only 19.7% (2). Early diagnosis and treatment is an important way to alleviate such burdens. In 2012, Chinese government launched a key national plan for public health, Cancer Screening Program in Urban China (CANSPUC), in 18 provincial administrative regions (3). In this program, participants at high risk aged 40 to 74 years were provided free low-dose CT (LDCT) test for lung cancer screening, which increased the rate of early diagnosis of lung cancer. However, lung cancer screening may lead to certain consequences like false positive, negative emotion, overdiagnosis, overtreatment, and treatment-caused complications (4). In addition, LDCT will raise the concern for radiation exposure, especially in younger participants (5, 6). Therefore, an appropriate screening program should be developed to balance the advantages and disadvantages. In fact, the optimal age to start screening is still unclear in China, and different guidelines are seen in different recommendations, which have been modified several times in recent years (7–9).

A model-based cost-effectiveness analysis of different screening situations with LDCT can provide reference for decision makers to select the optimal screening strategy (6). Previous studies have shown that lung cancer screening is cost effective in Western countries (10–12). However, the epidemiologic characteristics and influencing factors of lung cancer will vary with region. For instance, in all age groups over 40 years old, the incidence of lung cancer in China exceeds that in America and shows an upward trend, which may directly lead to different model parameter settings (13, 14). Thus, it is necessary to conduct modeling analysis based on the epidemiologic characteristics of lung cancer in China.

In this study, based on CANSPUC results and epidemiologic characteristics of lung cancer in China, a Markov model was used to simulate the whole process of lung cancer screening with LDCT in different strategies. Through evaluation of their cost and effect from the perspective of healthcare system, we try to answer the following 2 questions: Is LDCT screening cost effective in China? What is the most appropriate age to start screening?

Target population was set to be 100,000 40-year-old Chinese heavy smokers in 2020. Samples are people with a 30-pack-year smoking history, including, as recommended by the latest Chinese lung cancer screening guideline, current smokers and those who have quit in the last 15 years (15).

The model in this study was modified based on the model of Jaine (12) and ultimately contained nine states (Fig. 1). It was assumed that lung cancer can be detected at an earlier stage if screening is adopted (6). The starting age of screening was set as 40, 45, 50, 55, 60, 65, and 70 years old, and the age to stop was set as 74 years old, with a frequency of screening once a year, as recommended in latest guidelines (7, 8). One cycle of Markov model was set to be 1 year and the model had 50 cycles in total and stopped when the patient reached 90 years or died.

In this study, the age-specific incidence of lung cancer in heavy smokers was determined by that of overall population in China (13) and the relative risk (RR) of heavy smokers to overall population. The annual mortality rate of different lung cancer stages were calculated by published corresponding 5-year survival rate (16), and annual mortality probability was converted from the mortality rate with other causes recorded in the statistical yearbook via a similar method (17). The distribution of lung cancer stages at screening rounds was determined by CANSPUC (18–23), the National Lung Screening Trial (NLST; ref. 24) and the Dutch–Belgian Randomised Lung Cancer Screening Trial(Nelson; ref. 25). The sensitivity and specificity of LDCT were obtained through NLST (24) results while the compliance of smokers came from CANSPUC (26). The increased risk of lung cancer caused by ionizing radiation from LDCT examination was detected via multiplying the radiation exposure dose of a single examination (5) by the increased risk brought by per unit exposure dose (6). Overdiagnosis was also considered in this study, whose probability came from NLST (27). Detailed parameter information can be found in the Supplementary Method and Table 1.

This study evaluated the screening effect of heavy smokers from healthcare system perspective. The cost of the model included the expenditure in LDCT test, human resources, publicity and participant management in screening, lung cancer diagnosis (including possible complications associated), and lung cancer treatment (stage-specific). All costs were discounted in the year of 2020 and then converted into US dollars according to the average exchange rate (ref. 28; Chinese Yuan (CNY) 6.8974 to US $1) in the whole year of 2020 (Table 1).

In this study, the health utility value and its fluctuation range suitable for different stages in Chinese patients with lung cancer were obtained through relevant literatures (29–31). An annual discount rate of 3% was adopted to discount the cost and utility value into 2020 equivalents. The fluctuation range of the discount rate in sensitivity analysis was approximately 0% to 8%.

R (Version 4.0.2) was selected to construct the Markov model. Primary study manifested incremental cost-effectiveness ratio (ICER), which was calculated as incremental cost divided by incremental effectiveness, compared with no screening. In 2020, China's GDP per capita was 10,504 US dollars (28). According to the World Health Organization's (WHO) recommendation, the program is cost effective if the ICER is less than three times per of GDP capita (31,512 US dollars in this study). Besides, the Net Health Benefit (NHB) was calculated by dividing the cost of different screening programs by willingness to pay (WTP) values, and subtracting the result from the quality-adjusted life-years (QALY) produced. A wide range of fluctuations and appropriate distribution types were set for the parameters (Table 1), and univariate sensitivity analysis and probabilistic sensitivity analysis were conducted.

The model was validated by comparing the simulated number of screen-detected and interval lung cancers per 1,000 individuals (in the first and second screening rounds) with the observed data from CANSPUC in Sichuan province (22), as shown in Supplementary Table S1.

In this study, we simulated 100,000 heavy smokers aged 40 years old who received screening at different starting ages. The baseline results showed that ICERs in all situations were lower than three times of GDP per capita, in which the minimum one appeared at the age of 65 years (Fig. 2). The younger the people initiated screening, the greater QALYs would be obtained. That means the maximum QALY was obtained at the age of 40. We also calculated the NHB obtained through different screening strategies, which showed that when the WTP was less than three times of GDP per capita, patients who started screening at the age of 50 obtained the maximal NHB (Fig. 3).

Through univariate sensitivity analysis, it was noticed that in the 40-year-old screening strategy, the ICER was unstable and would excess of three times of GDP per capita as parameters changed. But in other strategies ICERs all varied within acceptable threshold. The most sensitive factors affecting ICER were discount rate, cost of lung cancer diagnosis, and the health utility value of patients with lung cancer at stage I (Supplementary Table S2). Probabilistic sensitivity analysis showed that cost and effect scatter of all strategies (except for the 40-year-old strategy) fell within three times of GDP per capita on the whole (Supplementary Fig. S1). The cost-effectiveness acceptability curve showed that, when the WTP was less than three times of GDP per capita, the 65- and 50-year-old strategies had the highest cost-effectiveness probability, and latter one was higher than former (Supplementary Fig. S2).

For Chinese population, the cost effectiveness of LDCT screening for lung cancer at high-risk remains unclear, and the optimal age to start screening has not been clarified. We adopted Markov Model firstly to analyze the cost effectiveness of lung cancer screening with LDCT at different starting ages in heavy smokers in China. Taking the results of ICER, NHB, and univariate and probabilistic sensitivity analysis into consideration, we concluded that the optimal age to start screening in Chinese heavy smokers should be 50 years old, yielding a cost of 12,547 US dollars per QALY gained from screening and the highest NHB value among all scenarios.

Based on sensitive analysis, our model showed that all situations, except for the 40-year-old strategy, were cost effective and had significantly higher NHB values than those without screening. The incidence rate of lung cancer in China is still on rise while the mortality rate does not decreased significantly (14). Lung cancer screening with LDCT will greatly promote its early diagnosis and treatment, thus improving the survival rate. The CANSPUC program has proved this point: patients who are diagnosed as stage I lung cancer reached over 50% compared with 19% without screening (32).

The findings in this study are close to the results of Taiwan province, which concluded that lung cancer screening with LDCT in high-risk groups was cost effective and the ICER was 10,947 US dollars to 29,349 US dollars per QALY gained (33). Our study showed better effect than study in Iran, but was not as good as those conducted in New Zealand, United States, and Canada. In the study of Iran, LDCT was implemented every 3 years and the ICER increased from screening versus no screening was about 3,283.83 US dollars per QALY (34). In studies of New Zealand, United States, and Canada, annual screening was recommended and the corresponding ICER was 24,934.84 US dollars (35), 49,200 to 96,700 US dollars (10), and 33,825 US dollars (11) per QALY, respectively.

Several reasons may be responsible for these different results. Firstly, the frequency of screening will greatly influence the results, which could explain why the ICER is lower in Iran's study. Screening every 3 years significantly reduced the cost compared with annual screening. However, previous studies have indicated that lung cancer in stage IA may be more difficult to detect when the screening cycle is longer than 1 year (36). Since the survival rate of stage IA is much higher than other stages, annual screening is more probable to reduce mortality, which has been proved by the analysis in an Canadian study (11). Secondly, Different countries have different economic development levels, and costs of screening tests, diagnostic tests, and cancer treatment vary in different health systems (12). In China, the cost of LDCT test, lung cancer diagnosis, and treatment are lower than that of United States and Canada, which can explain why ICER of our study are lower than in those countries.

Thirdly, the results may also be influenced by screening eligibility criteria, which is exactly the purpose of this study. The setting of age threshold should be built on lung cancer epidemiology characters, health resources, and economic development level of the screening area. Although strategies with older starting screening age basically produce lower ICER, they obtained shorter QALYs. It is inconsiderate to simply choose the strategy with the lowest ICER, for the aim of screening is to achieve the maximal health benefits within the appropriate expenditure range. In this study, we took NHB into consideration, which achieved the highest value within the WTP less than three times of GDP per capita when screening started at 50 years old. It could be explained by two aspects. On one hand, the lung cancer incidence among 50 to 55 years old Chinese in 2015 was 56.13 per 100,000 people, which is more than twice of people aged 45 to 49 years old (13). On the other hand, the life expectancy in China was 77.3 years in 2019 (37) and screening started at 50 years can gain substantial QALYs compared with an older age (10).

There are some limitations of this study. First, the basic assumption of the screening model was to shift diagnosed cancers to earlier stages, for which benefits of early detection and management of precursor glandular lesions (which with excellent survival outcomes with treatment) were not considered. Similarly, LDCT for lung cancer screening offers an opportunity for simultaneous cardiovascular disease risk estimation and reduces cardiovascular morbidity, which also not considered here. This means benefits of the results were underestimated. Second, with the high smoking rate and large smoker number in China, screening combined with smoking cessation intervention may bring better effect, which is not reflected in this study. Third, this model is inapplicable for nonsmokers. Due to the lack of data, other important high-risk factors for lung cancer in China, such as second-hand smoke exposure and air pollution, have not been considered. Finally, although all risk factors for lung cancer (age, smoking history, quit date) are interdependent variables, only age was considered because the starting age of screening has been the most controversial factor in China's lung cancer screening guidelines in recent years. Given that there is no ideal lung cancer risk prediction model in China so far, it may lead to greater bias if we take other variables into the model.

In conclusion, in heavy smokers, implementing lung cancer screening with LDCT is cost effective and can be conducive to early diagnosis and early treatment of patients. We recommend that the optimal age to start screening is 50 years old. High-risk population can be identified through inquiry into their smoking history and age. Through explanation on the screening significance, we may encourage patients to participate in if they meet the inclusion criteria. At the same time, it may also be beneficial to further explore the feasibility of post setting for nursing navigators in China to reduce screening barriers for participants.

No disclosures were reported.

J. Yuan: Conceptualization, data curation, writing–original draft. Y. Sun: Conceptualization, data curation. K. Wang: Investigation. Z. Wang: Investigation. D. Li: Investigation. M. Fan: Investigation. X. Bu: Investigation. M. Chen: Writing–review and editing. H. Ren: Writing–review and editing.

The authors appreciate the contribution of Chao Li (School of Public Health, Xi'an Jiaotong University) to parameters setting and sensitivity analysis.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.