Analysis and prediction of the incidence and mortality trends of cervical cancer in Chinese women from 2003 to 2018
-
摘要:
目的 研究2003—2018年中国20~79岁女性宫颈癌发病率和死亡率的变化趋势,对未来五年宫颈癌发病及死亡率的趋势进行预测。 方法 收集我国2003—2018年20~79岁女性宫颈癌的发病和死亡数据,利用联结点回归模型分析趋势变化规律,进一步利用年龄-时期-队列模型探讨年龄、时期和队列因素对宫颈癌发病和死亡率的影响。分别建立自回归滑动平均混合模型(autoregressive integrated moving average model, ARIMA)、灰色模型(grey model, GM)(1,1)和误逆差传播(back propagation, BP)神经网络模型对发病率和死亡率进行拟合,选取预测精度高的模型预测未来五年宫颈癌的发病率和死亡率。 结果 2003—2018年间女性宫颈癌的发病率具有2个转折点,发病趋势先快速上升随后下降;死亡率具有1个转折点,趋势是先下降再上升。总体上看,宫颈癌的发生风险随着年龄的增长而增大,在55~<60岁达到峰值后缓慢下降。死亡风险从年龄上看不断上升,时期效应随着时期的推进而增大,队列效应则不断减弱。通过对比发现BP神经网络模型拟合的效果较好。 结论 2003—2018年间中国女性宫颈癌的发病率和死亡率整体上呈现下降的趋势,受年龄影响较大而受时期和队列的影响较小,未来五年发病率和死亡率将呈下降趋势。因此,应加强女性宫颈癌筛查和HPV疫苗接种工作,做好防控措施。 Abstract:Objective To study the change tendency of the incidence and mortality of the females with cervical cancer in China from 2003 to 2018, and to predict their trends in the next five years. Methods The data between 2003 and 2018 on the incidence and mortality of female cervical cancer cases aged 20-79 years old in China was collected. Then the Joinpoint regression model was used to analyze the regularity of the incidence and mortality on the base of the data, and the age-period-cohort (APC) model was further used to explore the influences of age, period and cohort on the numbers of the incidence and the mortality of females with cervical cancer. The autoregressive integrated moving average model (ARIMA), grey model (GM) (1, 1) and back propagation (BP) neural network model were developed to fit the incidence and mortality, and the model with the high-precision prediction was selected to foresee the incidence and mortality in the next five years. Results From 2003 to 2018, the incidence of female cervical cancer shows two turning points, with a rapid increase and then a decline; The mortality has a turning point, and decreases first and then increases. In general, the risk of cervical cancer cases increases with age, and slowly decreases after reaching a peak in their 55- < 60 years old. The risk of mortality keeps rising constantly with respect to age, the period effect increases with period evolving and the cohort effect decreases constantly. The fitting results of different models illustrate that the BP neural network model has better effect. Conclusions From 2003 to 2018, the incidence rate and the mortality rate of female cervical cancer cases decrease as a whole, and are more affected by age, but less affected by period and cohort. It is predicted that they will decline in the next five years. Therefore, it is necessity for women to strengthen the screening of cervical cancer and take HPV vaccination. -
Key words:
- Cervical cancer /
- Trend /
- APC model /
- Joinpoint regression model /
- Prediction
-
表 1 中国女性宫颈癌发病率、死亡率APC模型分析结果
Table 1. APC model results of the incidence and mortality of female cervical cancer in China
变量 发病率 死亡率 β sx Z值 P值 β sx Z值 P值 年龄(岁) 20~<25 -1.86 0.43 -4.34 <0.001 -2.55 1.22 -2.10 0.036 25~<30 -0.71 0.22 -3.22 0.001 -1.51 0.64 -2.36 0.018 30~<35 -0.07 0.17 -0.40 0.689 -0.75 0.48 -1.59 0.113 35~<40 0.15 0.15 0.95 0.344 -0.27 0.40 -0.68 0.496 40~<45 0.33 0.15 2.25 0.025 0.16 0.34 0.47 0.638 45~<50 0.39 0.14 2.83 0.005 0.44 0.29 1.53 0.126 50~<55 0.42 0.13 3.24 0.001 0.69 0.24 2.93 0.003 55~<60 0.44 0.12 3.64 <0.001 0.70 0.20 3.56 <0.001 60~<65 0.32 0.12 2.77 0.006 0.69 0.17 4.16 <0.001 65~<70 0.26 0.11 2.33 0.020 0.78 0.16 5.00 <0.001 70~<75 0.18 0.12 1.48 0.139 0.80 0.18 4.47 <0.001 75~79 0.17 0.14 1.22 0.224 0.82 0.23 3.62 <0.001 时期(年) 2003 -0.16 0.07 -2.42 0.015 -0.25 0.13 -1.99 0.047 2008 -0.13 0.06 -2.11 0.035 -0.18 0.09 -2.04 0.042 2013 0.08 0.06 1.40 0.162 0.12 0.09 1.32 0.186 2018 0.21 0.06 3.48 <0.001 0.32 0.11 2.75 0.006 队列(年) 1928—1932 0.42 0.23 1.79 0.073 0.92 0.34 2.68 0.007 1933—1937 0.35 0.17 2.04 0.041 0.81 0.26 3.05 0.002 1938—1942 0.27 0.14 1.95 0.051 0.66 0.22 3.07 0.002 1943—1947 0.18 0.12 1.44 0.149 0.50 0.19 2.67 0.008 1948—1952 0.23 0.13 1.84 0.065 0.47 0.20 2.36 0.018 1953—1957 0.21 0.13 1.53 0.125 0.34 0.23 1.50 0.135 1958—1962 0.10 0.15 0.66 0.511 0.13 0.28 0.47 0.639 1963—1967 0.10 0.16 0.62 0.534 0.09 0.32 0.29 0.769 1968—1972 0.16 0.16 1.02 0.309 -0.01 0.37 -0.02 0.980 1973—1977 -0.01 0.17 -0.04 0.966 -0.26 0.42 -0.60 0.547 1978—1982 -0.08 0.18 -0.41 0.681 -0.40 0.48 -0.83 0.405 1983—1987 -0.15 0.20 -0.74 0.461 -0.51 0.53 -0.95 0.345 1988—1992 -0.38 0.25 -1.52 0.127 -0.72 0.68 -1.05 0.295 1993—1997 -0.80 0.41 -1.98 0.048 -1.07 1.11 -0.97 0.334 1998—2002 -0.60 0.87 -0.69 0.490 -0.98 2.38 -0.41 0.681 b 2.64 0.07 36.28 <0.001 1.61 0.18 8.75 <0.001 偏差 1.90 1.05 自由度 20 20 AIC 5.74 4.77 BIC -75.52 -75.57 表 2 标化发病率及标化死亡率模型拟合值
Table 2. Fitting values of the age-adjusted incidence and mortality
年份(年) 标化发病率
(/10万)ARIMA(0, 1, 0) GM(1, 1) BP神经网络 标化死亡率
(/10万)ARIMA(0, 1, 0) GM(1, 1) BP神经网络 2003 11.26 — — — 4.88 — — — 2004 11.45 11.42 10.91 — 4.81 4.76 4.48 — 2005 11.37 11.61 11.10 — 4.91 4.71 4.54 — 2006 11.23 11.52 11.30 — 4.49 4.84 4.60 — 2007 11.25 11.39 11.51 — 4.56 4.45 4.66 — 2008 11.28 11.41 11.72 11.46 4.65 4.54 4.72 4.38 2009 11.55 11.44 11.93 11.49 4.66 4.66 4.78 4.61 2010 11.85 11.71 12.14 11.92 4.53 4.70 4.84 4.33 2011 12.16 12.01 12.36 12.26 4.49 4.60 4.90 4.97 2012 12.51 12.32 12.58 12.57 4.81 4.59 4.97 4.98 2013 12.98 12.67 12.81 12.97 5.30 4.94 5.03 5.02 2014 13.55 13.14 13.04 13.49 4.95 5.45 5.10 5.08 2015 13.88 13.70 13.28 13.90 5.16 5.13 5.17 4.83 2016 13.87 14.03 13.52 14.20 5.28 5.37 5.23 5.15 2017 13.31 14.03 13.76 13.45 5.00 5.52 5.30 4.96 2018 13.73 13.47 14.01 13.65 5.89 5.27 5.37 5.91 MAE — 0.22 0.31 0.10 — 0.08 0.07 0.04 MSE — 0.08 0.13 0.02 — 0.21 0.20 0.19 MAPE(%) — 2.45 1.71 0.79 — 4.27 4.12 3.96 -
[1] 魏长慧, 朱继存, 牛媛娜, 等. 2004-2016年中国女性生殖系统恶性肿瘤死亡趋势分析[J]. 中华疾病控制杂志, 2019, 23(5): 506-511. DOI: 10.16462/j.cnki.zhjbkz.2019.05.003.Wei CH, Zhu JC, Niu YN, et al. Analysis of the mortality trend of female reproductive system cancer in China from 2004 to 2016[J]. Chin J Dis Control Prev, 2019, 23(5): 506-511. DOI: 10.16462/j.cnki.zhjbkz.2019.05.003. [2] 唐大先. 年轻女性宫颈癌患者临床特征与早期宫颈癌预后因素分析[J]. 中外女性健康研究, 2019, (15): 53-54. https://www.cnki.com.cn/Article/CJFDTOTAL-ZWVJ201915033.htmTang DX. Analysis of clinical features and prognostic factors of early cervical cancer in young women[J]. Women's Health Research, 2019, (15): 53-54. https://www.cnki.com.cn/Article/CJFDTOTAL-ZWVJ201915033.htm [3] 胡文斌, 史建国, 张婷, 等. 江苏省昆山市2006-2013年宫颈癌发病趋势分析[J]. 中华肿瘤防治杂志, 2015, 22(21): 1658-1661. DOI: 10.16073/j.cnki.cjcpt.2015.21.004.Hu WB, Shi JG, Zhang T, et al. Temporal trend analysis of cervical cancer incidence during 2006-2013 in Kunshan, Jiangsu Province[J]. Chin J Cancer Prev Treat, 2015, 22(21): 1658-1661. DOI: 10.16073/j.cnki.cjcpt.2015.21.004. [4] Dhillon PK, Yeole BB, Dikshit R, et al. Trends in breast, ovarian and cervical cancer incidence in Mumbai, India over a 30-year period, 1976-2005: an age-period-cohort analysis. [J]. Br J Cancer, 2011, 105(5): 723-730. DOI: 10.1038/bjc.2011.301. [5] Gonzaga CMR, Freitas JR, Barbaresco AA, et al. Cervical cancer mortality trends in Brazil: 1980-2009[J]. Cad Saude Publica, 2013, 29(3): 599-608. DOI: 10.1590/S0102-311X2013000300017. [6] 原瑞霞. 基于GBD大数据分析与预测中国女性乳腺癌发病与死亡趋势的研究[D]. 武汉: 武汉大学, 2018.Yuan RX. Analysis and prediction of morbility and mortality trend of breast cancer in Chinese women based on GBD big data[D]. Wuhan: Wuhan University, 2018. [7] Yang Y, Fu WJ, Land KC. A methodological comparison of age-period-cohort models: the intrinsic estimator and conventional generalized linear models[J]. Sociol Methodol, 2004, 34(1): 75-110. DOI: 10.1111/j.0081-1750.2004.00148.x. [8] 陈磊, 徐杰茹, 王冕, 等. 1990-2019年中国肾癌死亡趋势及其年龄-时期-队列分析[J]. 中华疾病控制杂志, 2021, 25(9): 1026-1033. DOI: 10.16462/j.cnki.zhjbkz.2021.09.007.Chen L, Xu JR, Wang M, et al. Analysis for mortality trend and age-period-cohort of kidney cancer in China from 1990 to 2019[J]. Chin J Dis Control Prev, 2021, 25(9): 1026-1033. DOI: 10.16462/j.cnki.zhjbkz.2021.09.007. [9] 程志伟. 利用年龄-时期-队列模型探讨中国膀胱癌死亡率的影响因素[D]. 郑州: 郑州大学, 2018.Cheng ZW. Study on the influencing factors of bladder cancer mortality in China by age-period-cohort model[D]. Zhengzhou: Zhengzhou University, 2018. [10] 向芳, 李胜, 郝莲, 等. 2003-2017年中国女性宫颈癌死亡趋势分析-基于年龄-时期-队列模型[J]. 卫生经济研究, 2020, 37(5): 23-27. DOI:10.14055/j.cnki.33-1056/f.2020.05.024.Xiang F, Li S, Hao L, et al. Analysis on Chinese women's cervical cancer death trends from 2003 to 2017[J]. Health Economics Research, 2020, 37(5): 23-27. DOI: 10.14055/j.cnki.33-1-1056/f.2020.05.024. [11] 包娅薇, 邵明, 陈雨婷, 等. 自回归求和滑动平均(ARIMA)模型在全球新型冠状病毒肺炎发病人数预测中的应用[J]. 中华疾病控制杂志, 2020, 24(5): 543-548. DOI: 10.16462/j.cnki.zhjbkz.2020.05.010.Bao YW, Shao M, Chen YT, et al. Application of autoregressive integrated moving average (ARIMA) model in global prediction of COVID-19 incidence[J]. Chin J Dis Control Prev, 2020, 24(5): 543-548. DOI: 10.16462/j.cnki.zhjbkz.2020.05.010. [12] 王峰. 基于灰色神经网络模型的全国病毒性肝炎发病率预测[D]. 太原: 山西医科大学, 2012.Wang F. Predict the incidence of hepatitis B based on Grey and neural network model[D]. Taiyuan: Shanxi Medical University, 2012. [13] 颜康康. 多种模型在梅毒月发病率预测中的比较研究[D]. 长春: 吉林大学, 2015.Yan KK. The comparative study of many models on the forecasting incidence of syphilis[D]. Changchun: Jilin University, 2015.