| ObjectlivesAs one of the most common cancers in the female population,cervical cancer has ranked as the second most incident gynecological cancer in recent years,trailing only breast cancer.In this study,according to the data of the cervical cancer mortality rates among urban and rural female populations in China during 1988-2014,we make a fully understanding of the cervical mortality trends among urban and rural female populations in China in the past 30 years.Furthermore,by setting up Poisson regression,negative binomial regression and age-period-cohort model,we analyze the epidemiological characteristics of cervical cancer mortality,meanwhile,to explore the application of ARIMA model of time series analysis methods and GM(1,1)model of gray system model on cervical cancer mortality forecasting,and to provide scientific reference and lay the foundation for future evaluation of the effects of cervical cancer prevention and control measures.MethodsIn this study,we take urban and rural women in China for research objects,and we collected mortality data for cervical cancer in urban and rural regions during the periods 1988-2014.This study mainly focused on studying the cervical cancer mortality in urban and rural female populations through the following five parts:(1)We studied the total mortality trends of cervical cancer in urban and rural regions during the periods 1988-2014 by descriptive research and describe the age variation trends,time variation trends and cohort variation trends of cervical cancer mortality.(2)We applied the joinpoint regression model to describe the mortality trends among urban and rural women in China during the periods 1988-2014.At the same time,the annual percentage change(APC),average annual percentage change(AAPC)and 95%confidence interval of age-specific mortality among different age groups were calculated.(3)We applied Poisson Regression and Negative Binomial Regression model to analyze the mortality data of cervical cancer for urban and rural women in China during the periods 1988-2014.Additionally,compare the fitting effects of the two models and analyze the risk factors for cervical cancer mortality.(4)We used the age-period-cohort model combined with the Intrinsic Estimator algorithm to describe the variation trends of cervical cancer mortality for urban and rural women in China during the periods 1988-2012 and analyze the independent effects of chronological age,time period and birth cohort.(5)We performed the ARIMA model of time series analysis and GM(1,1)model of gray system model based on the cervical cancer mortality data of urban and rural women in China during the periods 1987-2015 to predict the mortality of cervical cancer for urban and rural women in China during the periods 2016-2015 respectively,forecasting the mortality trends of cervical cancer in China for the next ten years.Results1.For the whole populations in China,the age-standardized mortality rate of cervical cancer showed a downward trend during the periods 1988-1998 and 2005-2012,while the mortality rate showed an upward trend during the periods 1998-2002.On the whole,the total mortality rate of cervical cancer experienced two significant declines and one increasing trend during the periods 1988-2014.The annual percent change rate gradually decreased at the speed of 2.57%?3.14%and 2.44%for age-standardized mortality rate of cervical cancer during the periods 1988-1993,1993-1998,and 2005-2012,respectively,while the annual percent change rate gradually increased at the speed of 2.28%during the periods 1998-2002.For female populations in urban China,the age-standardized mortality rate for cervical cancer has experienced two significant changes during the periods 1988-2014,the annual percent change rate gradually decreased at the speed of 6.72%for age-standardized mortality rate of cervical cancer during the periods 1988-2002,while the annual percent change rate gradually increased at the speed of 9.21%during the periods 2002-2014.For female populations in rural China,the age-standardized mortality rate for cervical cancer has also experienced two significant changes during the periods 1988-2014,the annual percent change rate gradually decreased at the speed of 14.45%during the periods 2000-2005,while the annual percent change rate gradually increased at the speed of 9.41%during the periods 2005-2014.In general,the mortality rate in rural areas was higher than that in urban areas.However,the mortality rate of urban females began to catch up with the mortality level in rural areas since 2005.1n addition,the mortality rate of cervical cancer increased with age and decreased with birth cohort in both urban and rural areas,and the mortality rate in age group of 75-79 was the highest in all age groups.2.For female populations in urban China,the annual percent change rate gradually decreased at the speed of 8.3%for age-specific mortality rate of cervical cancer for the age group including women aged 20-24 years during the periods 1988-2014.However,the age-specific mortality rates of cervical cancer for the four age groups including women aged 30-49 years showed a general upward trend during the whole observation periods.Particularly,the annual percent change rates for the age groups including women aged 30-34 years,35-39 years,40-44 years and 45-49 years gradually increased at the speed of 5.0%,7.0%,6.0%and 4.6%during the whole observation periods,respectively.For female populations in rural China,the annual percent change rate gradually decreased at the speed of 14.6%,3.7%,3.6%and 4.8%for age-specific mortality rate of cervical cancer for the age groups including women aged 20-24 years,45-49 years,55-59 years and 60-64 years respectively during the periods 1988-2014.3.The results by Poisson regression and negative binomial regression model shows that all factors including region,age and periods are independent risk factors of cervical cancer;furthermore,the results of the comparison of the two model shows that negative binomial regression fitted better than Poisson regression,and the parameters estimated by negative binomial regression model are more accurate than Poisson regression model.The results by negative binomial regression model shows that the relative risk of mortality of cervical cancer in urban China was 73%of rural China in the same periods and same age groups(95%CI=0.67-0.79,P<0.01);moreover,the risk of mortality of cervical cancer increased with age,and with an average increase of 37%with the increase of every five age years;the risk of mortality of cervical cancer decreased with year and with an average decrease of 4%per year(OR=0.96,95%CI=0.96?0.97,P<0.01).4.In this study,we performed an age-period-cohort analysis using the intrinsic estimator method to estimate the independent effects of age,time period,and birth cohort based on the mortality data of cervical cancer in urban and rural regions during the periods 1988-2012.The results by APC model showed that risk of mortality for cervical cancer in urban and rural regions generally increased with age and decreased with birth cohort,and the variation trends in urban and rural regions were basically consistent,the risk of mortality for cervical cancer increased with age before 64 years old while the age effect of morality risk showed a general decreasing trend after 64 years old.Additionally,the period effects increased the risk of mortality of cervical cancer for urban women while it decreased the mortality risk for rural women when controlling age and cohort factors.At the same time,the cohort effects of mortality risk of cervical cancer for urban and rural women are almost the same,the cohort effects peaked in the cohort born in 1911-1915 in both rural and urban China,and cohort effects decreased the mortality risk of cervical cancer in both urban and rural China after controlling the other two factors.5.In this study,we performed the ARIMA model of time series analysis and GM(1,1)model of gray model based on the mortality data of cervical cancer in urban and rural regions during the periods 1987-2015,we finally established the ARIMA(0,1,1)model for cervical cancer mortality of urban female populations and ARIMA(1,1,0)model for cervical cancer mortality of rural female populations after repeated testing and diagnosis for the mortality data sequence,meanwhile,we established GM(1,1)model and gray residual GM(1,1)model.We finally concluded that the gray residual GM(1,1)model was the best model compared to another two models based on the fitting effects of three different models.According to the prediction results of gray residual GM(1,1)model,we concluded that the mortality level in urban regions may be higher in the next ten years,which may show an increasing trend at the same time,however,the cervical cancer mortality for rural women may show a gradual decreasing trend in the next ten years,which may contribute to the widening gap between urban and rural areas.Conclusions1.The mortality level of cervical cancer of rural women was higher than that of urban women,and the gap between urban and rural areas gradually narrowed during the periods 1988-2014.2.The fitting effect of negative binomial regression model was better than Poisson regression model,and the parameters estimated by negative binomial regression mode are more accurate.3.When controlling the periods and cohort factors,the age effect of mortality risk for cervical cancer increased with age before 64 years old while the age effect showed a general decreasing trend after 64 years old;while cohort effects decreased the mortality risk of cervical cancer in both urban and rural China after controlling the age and periods factors.Furthermore,in the 1860s,the cohort effect of mortality risk for cervical cancer in urban and rural regions showed a transient upward trend due to civil strife.4.The intrinsic estimator method solved the problem of "non-identification" of the APC model to a certain extent and it can obtain the unique solution of the parameters.5.The mortality level of cervical cancer of urban women may be higher in the next ten years and the gap between urban and rural areas may be widened. |
PDF Full Text Request