Projection of premature mortality from noncommunicable diseases for 2025: a model based study from Hunan Province, China, 1990–2016

Data source

We conducted the projection in Hunan Province, Central China, where both the burden of years of lost (YLLs) and the ratio of observed to expected disability-adjusted life-years (DALYs) are significantly higher than the national average (Zhou et al., 2019). Based on the Global Burden of Disease Study (GBD) 2016 for China, we extracted death data from 1990 to 2016 for the above four main NCDs for Hunan by age, sex, year, and underlying causes. The GBD study was a collaboration between the Institute for Health Metrics and Evaluation, University of Washington, and the Chinese Center for Disease Control and Prevention (CDC). With a highly standardized method, the mortality was estimated for China based on multisource death surveillance systems or surveys conducted in the country, which mainly consisted of national disease surveillance points system, population death information registry and management system, national maternal and child health surveillance system, local cancer registry and some other mortality reports (Zeng et al., 2017).

According to the International Classification of Diseases tenth revision (ICD-10), we distributed the ICD-10 codes for the four NCDs as follows: cancer: C00-C97; CVD: I00-I99; chronic respiratory diseases: J30-J98; and diabetes: E10-E14.

Statistical analysis

We set premature mortality from NCDs (also written as premature NCD mortality) as the key indicator for the analysis, with age-standardized rates (ASRs) standardized to the 2010 China census population as a minor indicator. Premature mortality is defined by the WHO as the probability of dying between the age of 30 and 70 from NCDs and is calculated by age-specific death rates with a life table method in the following manner: (World Health Organization, 2020a).

(1) Age-specific death rates in 5-year age groups (e.g., 30–34...65–69) were calculated by ${}_5^{\ast }{M}_x=\frac{TotaldeathsfromfourNCDsaged\left[x,x+5\right)}{Mid-yearpopulationaged\left[x,x+5\right)}$.

(2) The ${}_5^{\ast }{M}_x$ was translated into age-specific probability of death:${}_5^{\ast }{q}_x=\frac{{}_5^{\ast }{M}_x\ast 5}{1+{}_5^{\ast }{M}_x\ast 2.5}$.

(3) The probability of death for persons aged 30–70 was calculated last: ${}_{40}^{\ast }{q}_{30}=1-\prod \left(1-{}_5^{\ast }{q}_x\right)$

We performed a joinpoint regression to examine trends in premature NCD mortality, with a maximum of three joinpoints set for the analysis. The joinpoint regression describes continuous changes by connecting several line segments on a log scale and identifies statistically significant changes with a Monte Carlo permutation test (Jones et al., 2016; Kim et al., 2000). The three following indicators, annual percentage change (APC), average APC (AAPC) and overall percent change, reflected different changes in temporal trends. The former two indicators were calculated by APC_i= {exp(b_i) - 1}×100, and $AAPC=\left\{exp\left(\frac{\sum {w_{i}b}_i}{\sum w_i}\right)-1\right\}\times 100$,

where b_i represents the slope coefficients for each segment in the years studied, and w_i represents the length of each segment in the intervals. The final indicator was calculated with an AAPC-based exponentiation function by first converting the AAPC to the predicted single-year change and then exponentiating to the number of study years minus one to produce the overall change and its magnitude, which was finally converted to a percent change (Jones et al., 2016).

We projected premature mortality from NCDs (as well as ASRs) with 95% confidence intervals (CIs) for 2025 by fitting a linear regression over the most recent trend identified by the joinpoint model. To estimate the ability to meet the UN target in Hunan, we compared the projected premature mortality from NCDs in the province in 2025 with the level in 2010 to find the relative reduction with the following formula:

$relativereduction\left(\text{\%}\right)=\frac{\left(projectedprematuremortality2025-observedprematuremortality2010\right)}{observedprematuremortality2010}\ast 100$.

This allowed us to determine if the relative reduction would be greater than 25%.

To identify excess premature NCD mortality due to unfavorable changes (slowed, stalled or reversed) (Yang et al., 2017), we performed a three-step estimation: first, each most significant APC for the four NCDs was selected as a projection point to find the expected premature mortality, assuming it would continue to decline to 2025 at the same level as the selected APCs. Second, we compared the total differences among observed (1990–2016)-projected (2017–2025) premature mortality with the expected ones to obtain the absolute excess premature mortality. Third, the differences were divided by expected premature mortality to obtain the relative change in the excess premature mortality.

Experimental verification was carried out to evaluate the prediction accuracy of the Joinpoint model. Death data of all NCDs combined during 1990–2011 from the GBD was selected as a sample, to project premature mortality rate for 2012–2016. The results were then compared with the real data with three metrics: mean Square Error (MSE), Percentage Error (PE) and Mean Absolute Percentage Error (MAPE ) (Zeng, 2019). Among them, $MSE=\frac{1}{n}{\sum }_{i=1}^n{\left(\hat{y}-y_i\right)}^2,PE=\frac{\left|\hat{y}-y_i\right|}{y_i}\times 100\text{\%}$, $MAPE=\frac{1}{n}{\sum }_{i=1}^n\frac{\left|\hat{y}-y_i\right|}{y_i}\times 100\text{\%}$, where $\hat{y}$ represents projected value, y_i represents observed value. MAPE less than 10% were considered good accurate (Cesnaite et al., 2020).

Analyses for temporal trends and projection through 2025 were performed using the Joinpoint program version 4.7.0.0 (Statistical Research and Applications Branch, National Cancer Institute, USA). Line art was produced by R version 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria). A p-value <0.05 was considered statistically significant.

Modified high blood pressure control

Every 10 mm Hg reduction in systolic blood pressure significantly lowered the risk of major CVD events (relative risk 0.80, 95% CI [0.77–0.83]), resulting in a 13% reduction in all-cause mortality (Ettehad et al., 2016). However, high blood pressure management, from awareness to treatment or control, is poor in China (Li et al., 2017). A fact is that only 45% of Chinese adults with hypertension were aware of their condition, only 30% were taking anti-hypertensive drugs, and just 7% had achieved normal blood pressure levels (Lu et al., 2017). A comprehensive, multistage strategy is needed that involves a diet low in salt and rich in polyunsaturated fatty acids, adequate physical activity (no less than a metabolic equivalent of 600 min per week), and an improved primary health-care system. Health authorities and professional institutions need to work together to promote an integrated prevention-control-treatment model at the community level leveraging the internet and health information technology. Within such a model, a hypertension outpatient service in community medical institutions and a family doctor contracting service are required to provide individuals with regular hypertension management with respect to screening, essential anti-hypertensive medications, health counseling, follow-up services, etc. It is particularly crucial for implementing community-based hypertension screening, as it could have a significant long term impact on systolic blood pressure at the population level (Chen et al., 2019b).

More ambitious tobacco control measures

Although the smoking rate has fallen in many high-income countries, it is rising rapidly in China, with a prevalence of 50.5% in male adults (Chinese Center for Disease Control and Prevention, 2019). It would impose a high macroeconomic burden of tobacco-attributable NCDs for China: Tobacco-attributable NCDs would cost China 16.7 trillion yuan (US$2.3 trillion in constant 2018 prices) from 2015 to 2030, equivalent to a 0.9% annual tax on aggregate income. Secondhand smoke exposure would contribute to 14% of the burden (Chen et al., 2019a). As the hometown for of the two best-selling brands of cigarettes in China, Hunan has been slow to take measures for tobacco control. Many successful policies for tobacco control in other countries have shown that a 50% reduction in smoking is feasible (Jha & Peto, 2014; Ng et al., 2014). Such feasibility could be achieved by the following actions: first, raise cigarette taxes. Although China implemented a tax linkage in 2015, raising the wholesale and price taxes on cigarettes from 5% to 11% (National Health Commission of the People’s Republic of China (2015), 2015), cigarettes are much more affordable here than in other countries, with most costing 10 CNY (1.4 USD) a package. It is estimated that a 50% price increase in cigarettes due to taxes in China would yield an additional 231 million years of life (Verguet et al., 2015). Therefore, a higher cigarette tax rate should be the first action. Second, regulate smoking with reference to practices in developed cities such as Beijing, Shanghai, and Shenzhen by comprehensively enacting smoking bans in indoor workplaces, indoor public places, and public transportation. Third, enforce strict bans on tobacco advertising, sponsorship, or any other activity that may weaken smoking control. Fourth, reform the design of cigarette packaging. Cigarette packages in China are all beautifully designed due to a deeply rooted smoking culture, and the warning occupies only a small part of the design space. Packaging should be redesigned to feature the warning in text and graphics. Fifth, conduct targeted health education. The public’s awareness is often confined to ”smoking is harmful to health”, and many people do not know exactly how smoking is harmful. Health education could be conducted by combining traditional media with new media (such as the Internet or WeChat) or through health campaigns hosted by professional doctors.

A stepwise reduction in salt intake

High salt consumption is the leading cause of hypertension and is strongly tied to stroke in China (Mente et al., 2018). Chinese people have a daily average salt intake of 12–14 g, (Mente et al., 2018; Du et al., 2014), much greater than the WHO recommendation of <5 g/day. Due to the characteristics of Hunan cuisine, residents’ diets are particularly high in salt. We suggest a shift strategy for salt reduction involving both commercial foods and consumer behaviors. First, reduce salt in commercial or processed foods through an industry-wide shift. The key is gradual salt reduction in small steps. Following the UK’s practice (He, Brinsden & MacGregor, 2014), gradually lower salt targets (such as a 20% decrease) can be set in high-salt categories for the food industry. It is also encouraged that alternatives with the same or better taste be proposed, such as “less salt, more spices” or a “more potassium, less sodium” diet, which have been shown to be helpful in reducing blood pressure and CVD mortality (Peng et al., 2014). Second, shift consumer awareness to action. National salt campaigns can not only raise consumer awareness but also have a remarkable impact on salt intake in the population. In 2017, the Action on Salt China (ASC) program was established with four cluster randomized controlled trial packages (He et al., 2018). Hunan participated in the program at five county-level locations, but with 130 county-level areas in the province, actual participation was low. Thus, we suggest expanding salt-reducing interventions from the ASC to the whole province. The salt industry should also take responsibility, for instance, by developing saltshakers with smaller holes or convenient salt intake calculators to help consumers make essential behavior changes.

Reduce harmful alcohol intake

Health issues attributable to alcohol use, such as CVD and cancer, have been largely underemphasized in China. During the past 30 years, a striking increase has been seen in alcohol use among Chinese men, greater than that in most other countries (Jiang, Room & Hao, 2015), and this trend is forecasted to continue (Manthey et al., 2019). This increase is strongly associated with robust economic development and a deeply rooted alcohol culture. However, the government can play a substantial role in developing alcohol use policies to alter drinking levels. A good example would be Russia, where WHO’s recommended best buys interventions for alcohol use, including taxation, availability restrictions, and bans on marketing, were effectively carried out, leading to remarkable changes in both alcohol use and the burden of alcohol-related disease (Neufeld & Rehm, 2013). For Hunan, similar interventions are also needed and should be broadened through strict restrictions on alcohol advertising on television, legally binding regulations on alcohol sponsorship, and heavy punishments for drunk drivers.

Multisector cooperation to control PM pollution

Although the WHO has not set targets for environmental risk factors, PM pollution should be seriously addressed due to its striking position as the fourth-leading risk factor for death in China (Zhou et al., 2019). PM pollution causes 3⋅3 million premature deaths worldwide each year, with China being the largest contributor (Lelieveld et al., 2015). Meanwhile, PM pollution has a huge macroeconomic impact on NCD in China, where total losses from NCDs associated with air pollution were estimated to be $499 billion (constant 2010 USD) from 2015–2030 (Chen & Bloom, 2019). As in many other regions in China, PM2.5 reduction control in Hunan has just begun. The local government released in 2018 a 3-year action plan to reduce the annual PM2.5 concentration to less than 40 µg/m³ by 2020 (People’s Government of Hunan Province, 2018), but this plan still falls far short of the WHO guidelines (World Health Organization, 2020b). Thus, intersector collaboration in public–private partnerships should be encouraged by setting PM2.5 levels as an assessment indicator for local government development, supervising pollution from industrial enterprises, establishing reward-and-punishment mechanisms, prompting responses to heavy pollution weather, etc.

When discussing NCDs control, we must recognize that the COVID-19 is now capturing the world focus. As the COVID-19 pandemic is raging worldwide and spreading fast in many countries, China has largely controlled the epidemic. The Chinese government has spent substantial efforts in controlling the COVID-19 epidemic—such as lockdown, extension of the lunar new year holiday, and facility isolation of mild to moderate cases using Fangcang shelter hospitals (Chen et al., 2020b; Chen et al., 2020c). However, these measures impose costs, involving human resources, economic losses, public engagement, coordinated governance structures at the national and local levels, etc. Thus we need to think about long-term plans for epidemic control. In the post lock-down COVID-19 era in China, the management of NCDs becomes especially important as people with underlying chronic diseases are more likely to die from COVID-19 (Wu & McGoogan, 2020). Reducing the risk factors in NCDs such as smoking and air pollution can also lower the risk of death for COVID-19 (Chen et al., 2020a; Alqahtani et al., 2020).

Our study is subject to some limitations. First, the data were derived from the GBD 2016, and all the limitations in the GBD study are also applicable to this study. Second, only five risk factors were addressed in the recommended interventions because the effects of other factors can be partially replaced, but this may affect the most effective control of future premature NCD deaths. Third, we conducted the projection under the current trend, with no consideration for possible greater efforts to reduce NCD risk factors in the future. However, the benefits of lowering the risk factors are produced gradually, which should have a small impact on the results of the present study. Additionally, our projection did not consider the change of age structure of the population during the period studied. This may weaken to some extent, the extrapolation of the model presentation.