Stronger policy required to substantially reduce deaths from PM2.5 pollution in China

The study addresses how China’s APPCAP (2013–2017) affected mortality attributable to ambient PM2.5 (DAPP), a key health risk and indicator for UN SDG 3.9. PM2.5 exposure contributes to respiratory and cardiovascular mortality; DAPP depends on PM2.5 concentration, demographics, and disease-specific death rates. China’s rapid industrialization led to high PM2.5 levels; APPCAP aimed to cut urban PM2.5 by 10–25% by 2017 across sectors at substantial cost. Prior estimates of DAPP trends and APPCAP’s health benefits were inconsistent due to differing data, disease selections, and exposure–response functions. This study quantifies national, regional, and disease-specific DAPP dynamics (2000–2017), decomposes drivers (population, PM2.5, age structure, death rates), and projects future DAPP to 2030 under policy scenarios to inform progress toward SDG 3.9.

Previous analyses of APPCAP’s health benefits reported mixed results. Some directly equated declines in DAPP with policy benefits, while others isolated PM2.5 effects assuming other factors fixed. Trends post-2013 diverged, with some studies indicating decreases in DAPP whereas GBD 2017 reported increases. Discrepancies stem from different input datasets (PM2.5, demographics), disease sets, and exposure–response functions (e.g., IER vs. alternatives). A comprehensive decomposition that separates PM2.5 effects from demographic and mortality-rate changes has been lacking, motivating this study’s approach using updated GBD 2017 data and methods.

Data: Annual gridded PM2.5 concentrations (EFPM V4CH02) at 0.01° for 2000–2016 from the Atmospheric Composition Analysis Group; 2017 gridded PM2.5 extrapolated using 2016 grids scaled by the ratio of monitored PM2.5 (2016 vs. 2017) from China’s National Urban Air Quality Real Time Publishing Platform. Demographics: national age structure and age- and disease-specific death rates from GBD 2017; spatial population distribution from HYDE 3.2 (0.083°) reconciled with city statistical yearbooks (284 prefectural-level cities, 2000–2017); future national population and age structure from UN World Population Prospects. Administrative and regional boundaries from national datasets and Liu et al. (2005). Estimation of DAPP: Following GBD 2017 comparative risk assessment, DAPP_t = Σ(POP_t × AgeP_a × Rate_ad × PAF_ad), across 15 age groups (25–30 … 90–95, 95+; and <5 for lower respiratory infection) and six PM2.5-related diseases: ischemic heart disease (IHD), stroke, chronic obstructive pulmonary disease (COPD), lung cancer (LC), lower respiratory infection (LRI), and diabetes mellitus type 2 (DM2). PAFs derived from PM2.5 and Integrated Exposure–Response (IER) functions (age-specific for IHD and stroke; uniform for others) from GBD 2017. DAPP computed at ~10 km grid resolution; national sums and age-standardized DAPP rates (per 100,000) calculated using GBD standard population weights to remove demographic effects. Trend detection: Sequential Mann–Kendall test applied to annual change rates of DAPP and age-standardized rates to detect significant breakpoints (notably 2013). Decomposition of drivers: Changes in DAPP decomposed into contributions from four factors—PM2.5 concentration, population size, age structure, and disease death rates—by sequentially introducing factors into the DAPP equation across all 24 permutations and averaging contributions. Effects of death rate changes are defined independent of PM2.5 by removing PM2.5 influences on rates. Scenario projections to 2030: Two PM2.5 policy scenarios—Trend (population-weighted PM2.5 to 35 µg m−3 by 2030) and Ambitious (to 10 µg m−3 by 2030, WHO guideline). Common assumptions across scenarios: UN business-as-usual population and age structure; improved health care leading to 30% lower death rates for PM2.5-related diseases in 2030 relative to 2015 (Healthy China 2030). Factor decomposition applied to 2017–2030 changes under each scenario. Uncertainty reported as 90% confidence intervals.

- PM2.5 and demographics: Population-weighted PM2.5 decreased from 52.5 µg m−3 (2013) to 42.2 µg m−3 (2017) after APPCAP. Total population grew 13.2% and the share aged ≥60 grew 66.0% from 2000 to 2017. - National DAPP trend: DAPP increased from 714 thousand (458–950) in 2000 to 971 thousand (635–1284) in 2017, a 36.1% (35.2–38.6) rise; age-standardized DAPP rate fell from 83 (54–109) to 65 (42–85) per 100,000. - Post-2013 slowdown: Average annual DAPP growth declined from 2.1% (2000–2013) to 1.0% (2013–2017); age-standardized rate decline accelerated from about −1.1%/yr (pre-2013) to −2.7%/yr (2013–2017). Significant trend break in 2013 detected nationally and in several regions. - Regional contributions: The North region accounted for about 30% of the national DAPP increase (2000–2017); other regions added 13–57 thousand. - Disease-specific changes (2000–2017): IHD +145 thousand (101–188) and LC +81 thousand (61–100) accounted for >80% of net increase; stroke and DM2 increased; LRI and COPD decreased. - Decomposition (2000–2017): Net change driven by PM2.5 (+63 thousand [54–67]), population growth (+110 thousand [71–146]), and aging (+424 thousand [273–563]); partially offset by declining death rates (−340 thousand [222–442]). - APPCAP effect (PM2.5 component): PM2.5 changes increased DAPP by +127 thousand (111–131) during 2000–2013, but decreased DAPP by −64 thousand (57–64) during 2013–2017. Per year: +10 thousand/yr (2000–2013) versus −16 thousand/yr (2013–2017). The −64 thousand corresponds to a 6.8% (5.3–9.1) reduction relative to 2013 DAPP (935 thousand [627–1,215]). - Other factors (2013–2017 per year): Aging +29 thousand (18–39); population +4 thousand (2–5); death rate changes −8 thousand (4–11), with smaller reductions due mainly to slower COPD mortality declines. - Projections to 2030: • Trend (PM2.5 to 35 µg m−3): DAPP 953 thousand (608–1279), net −18 thousand (5–26) vs 2017. PM2.5 alone −69 thousand (59–74; ~−7.1% [5.7–9.3]); death rates −299 thousand (194–397); population +17 thousand (11–22); aging +333 thousand (216–443). • Ambitious (PM2.5 to 10 µg m−3): DAPP 550 thousand (275–850), net −421 thousand (359–433) vs 2017. PM2.5 alone −511 thousand (415–556; ~−52.6% [43.4–65.4]); death rates −177 thousand (111–243); population +1 thousand (−3–4); aging +266 thousand (163–369).

APPCAP markedly reduced PM2.5 concentrations, shifting the PM2.5 contribution to DAPP from positive pre-2013 to negative post-2013, yielding 64 thousand fewer deaths in 2017 relative to 2013. Nonetheless, overall DAPP continued to rise due to strong opposing demographic forces—population growth and rapid aging—and a diminished mitigating effect from declining disease death rates post-2013. Thus, while APPCAP generated meaningful health benefits, it was insufficient to reverse overall mortality trends attributable to PM2.5. Projections indicate that maintaining current policy trajectories (Trend) marginally reduces DAPP by 2030, whereas adopting a more ambitious policy achieving WHO PM2.5 guidelines could halve DAPP relative to 2017. The findings support integrated policy strategies: deeper PM2.5 emission cuts (especially in the energy sector and northern heating), tighter industrial regulation and enforcement, economic instruments (pollution-related taxes), public health advisories to reduce exposure, and alignment with climate policy to capture co-benefits for SDGs 3 and 13.

DAPP in China rose 36% from 2000 to 2017, with growth slowing after APPCAP’s 2013 implementation as PM2.5 concentrations declined. APPCAP reduced DAPP by an estimated 64 thousand (6.8%) in 2017 versus 2013 through air quality improvements, but demographic pressures offset much of this benefit. To achieve substantial reductions by 2030 and meet SDG 3.9, China must adopt stronger air pollution control policies capable of lowering population-weighted PM2.5 to WHO’s 10 µg m−3 guideline. Future work should refine estimates with higher-resolution mortality and exposure data, incorporate meteorology and socioeconomic drivers, and use integrated scenario frameworks linking emissions, air quality, climate, and health.

Key limitations and uncertainties include: (1) Epidemiological and data uncertainties beyond the IER 90% confidence intervals, including input data sources, exposure assessment, and choice of exposure–response function (IER vs. GEMM). National-level age structure and death rates were used; limited province-level rates indicate potential bias (e.g., Beijing ~50% lower DAPP vs. national-rate estimates). (2) Exposure proxied by annual average outdoor PM2.5, potentially overestimating personal exposure by not accounting for behavior (inhalation rate, time spent outdoors). (3) APPCAP health benefit estimates assume a counterfactual of PM2.5 remaining at 2013 levels absent policy; real-world economic, trade, weather, and behavioral interactions could alter trends. (4) Meteorological influences on PM2.5 were not explicitly modeled. (5) Projection scenarios (Trend, Ambitious) are simplified and do not fully capture complex relationships among emissions abatement, climate change, and air quality; decomposition excludes other socioeconomic/behavioral factors. Future improvements could leverage higher-resolution datasets, province-level mortality, and integrated emissions–chemistry–transport modeling within advanced scenario frameworks.