Overall design
Using available multisource data (including published literature, open databases and local studies), a prevalence-based model was built [5]. Then, Joinpoint regression was applied to predict the prevalence and mortality rates of breast cancer and associated DALYs from 2020 to 2030. Finally, sensitivity analyses were performed to evaluate the impacts of demographics, DWs and screening parameters. The analysis indicators were DALYs and the age-standardized DALY rate (ASDR) considering Segis’ world standardized population. The overall methodology was shown in Fig. 1.
Data sources
Data on the incidence, mortality, and DALYs rates of breast cancer in China from 2006 to 2015 were extracted from the Chinese cancer registry annual reports (2009–2018) [20,21,22,23,24,25,26,27,28,29]. A national cancer surveillance network exists in China, and cancer registration is carried out in all provinces. In 2015, the national cancer surveillance network was expanded to include 501 cancer registries and covered 387,872,825 people, including 197,211,672 males and 190,661,153 females, accounting for 28.22% of the national population. All patients included in the registry had invasive cases, and all cancer cases were coded according to the International Classification of Diseases for Oncology, 3rd revision (ICD-O-3), and the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10). Provincial cancer registry centers are responsible for collecting, evaluating, and publishing provincial cancer statistics (including incidence, mortality and survival). All hospitals, medical and health institutions in administrative regions are required to submit cancer records to local cancer registries; this data is then submitted to the national cancer registry [29].
Demographic data were obtained from officially published reports, including China Population Statistics Yearbook [30] and World Population Prospects 2019 [31], which contain age-specific demographic information. Female standardized life expectancies were obtained from the GBD 2015 [32] and World Population Prospects 2019 [31]. Epidemiological parameters of breast cancer (incidence rate, mortality rate and survival rate) were obtained from the Chinese Cancer Register annual report [20,21,22,23,24,25,26,27,28,29] and the previous study [9]. DWs were obtained by converting the health utility values (calculated by “health utility scorenormal-health utility scorebreast cancer”) [18, 19], with the data of utilities obtained from two Chinese studies [15, 18, 19, 33]. This method was the most commonly used method summarized by a previous systematic review (five calculation measures in total) [17]. The proportions of clinical stages were obtained from national research in China [34, 35]. The detailed parameter sources and methods were shown in Additional files 1 and 2.
Parameters of breast cancer screening
Previous studies have shown that breast cancer screening plays a vital role in disease progression and the distribution of disease factors, such as the distributions of clinical stage and mortality. (1) The distributions of clinical stages were calculated from base-case data obtained from a multicenter hospital-based clinical epidemiologic survey among females who did not undergo screening in 2015; the population included 4211 breast cancer patients from 7 provinces in China (stage I: 19.2%) [35]. The data of Chinese females who underwent screening in 2015 were obtained from the “Central Financial Transfer Payment Project-Chinese Women Breast Cancer Screening Study”, which included 0.4 million women from 30 provinces in China (stage I: 35.9%) [34]; (2) The ten-year reduction in breast cancer mortality after screening was calculated from base-case data from a recent global meta-analysis (including 27 studies), which reported that the integrated mortality reduction rate was 22% (95% confidence interval [CI]: 18–25%)[12]. A linear hypothesis that the breast cancer mortality after screening five years and 15 years declined by 11% and 33%, respectively, was made. According to the results of the global meta-analysis and hypothesized data, “42%” was the maximum value of the included studies in the meta-analysis; since there were differences among the included studies, the sensitivity analyses tried the values of 25% (upper 95% CI of the meta-analysis results), 15%, and 5% in turn. (3) The coverage of breast cancer screening: According to the population-level screening coverage rates reported in the previous studies in China, breast cancer screening rates were 25.7% in women aged 35–64 years in 2015, and increased approximately 1% per year from 2013 to 2015 [36, 37]. Accordingly, the coverage rates (at the individual level) of breast cancer screening in Chinese females were estimated to be 30.7%, 35.7%, and 40.7% for 2020, 2025, and 2030, respectively.
YLDs estimation
The incidence rates of breast cancer from 2006 to 2015 in China were derived from the annual report of Chinese cancer register[20,21,22,23,24,25,26,27,28,29] and female population statistics[30], using numbers of females and new breast cancer cases, and local survival rates, to obtain the overall numbers of prevalence cases in 2015 according to the prevalence calculation formula[38] (see Fig. 1 Formula 1). Then using local data(detailed year- and age-specific data were shown in Additional file 3) on clinical stage-specific(I-IV) proportions for breast cancer and DWs to calculate overall YLDs. Survival rates were based on the 5-year survival rate of breast cancer in 2012–2015 in China (the best available local data at present) [9] and the survival rate ratios between age groups calculated from Surveillance, Epidemiology, and End Results (SEER) 5-year relative survival rates [39]. We extrapolated the age-specific 1- to 9-year survival parameters of breast cancer in China (detailed year- and age-specific data were shown in Additional file 3). The base-case DWs for different clinical stages (I–IV) were calculated as “health utility scorenormal-health utility scorebreast cancer”, which was the most commonly used method in previous systematic review [17,18,19]. The health utility scores were extracted from two previous Chinese articles [15, 34].
YLLs estimation
Total YLLs were estimated using mortality rates of breast cancer, female numbers and standard life expectancies [29, 30, 32]. The standard life expectancy (84.2 years) estimated by the GBD 2015 was used as the base-case parameter for external comparison [32]. Detailed year- and age-specific life expectancy data were shown in Additional file 4. According to the standard life expectancy in 2015 reported by the GBD 2015 and in 2050 reported by the World Health Organization (WHO) [40], the standard life expectancy in 2030 was estimated linearly, detailed age-specific data were shown in Additional file 4.
Predicting DALYs for 2020–2030
The incidence (strong linear correlation with the annual change rate of prevalence, correlation coefficient = 0.9, p < 0.001) and mortality rates of breast cancer were derived from the annual reports of Chinese cancer register covering the past several years [20,21,22,23,24,25,26,27,28,29]. A Joinpoint regression model (Joinpoint Regression Program 4.7.0.0) was used to calculate annual growth rates of prevalence and mortality rates (1.5% and 1.9%, respectively) and to predict the corresponding rates in 2020, 2025, and 2030 (assumption: there would be linear increases in the age-specific prevalence and mortality rates from 2015 to 2030). Standard life expectancies (84.2 years) from the GBD 2015 [32] and WHO 2050 (89.4 years) [40] were applied to the model. The standard life expectancies in 2020, 2025, and 2030 (84.9 years, 85.7 years, and 86.4 years, respectively) were estimated linearly. Changes in YLLs, YLDs, and DALYs were predicted for 2020, 2025, and 2030 using a similar approach. The formula was shown in Fig. 1.
Sensitivity analysis
The sensitivity analysis was based on the single factor analysis with replacing parameters. Limited by the availability of detailed data in China, some parameters were derived from assumptions and foreign studies. A series of sensitivity analyses were carried out for DWs, demographic, and screening parameters associated with breast cancer. Detailed sensitivity analysis factors were shown in Additional files 1 and 2.