The Global Burden of Disease (GBD) database is a major international resource for health data, providing comprehensive data on a wide range of diseases, injuries, and risk factors that affect global health (https://vizhub.healthdata.org/gbd-results). The GBD study is conducted by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington and systematically quantifies health loss across different populations and regions worldwide. This dataset tracks temporal trends in health outcomes and disease burden, offering valuable insights for policymakers, researchers, and healthcare providers^[11,12].

In the present study, data on AF/AFL in China (excluding Taiwan) from 1990 to 2023 were extracted from the GBD database. AF/AFL cases in the GBD study were identified based on standard clinical diagnostic criteria, primarily relying on electrocardiographic evidence^[13]. The analyses encompassed four key measures: incidence, prevalence, mortality, and DALYs. Sex-specific analyses were conducted, and the study population included individuals aged ≥ 30 years. As the GBD database contains aggregated data without individual-level identifiers, ethical approval was not required for this study.

Joinpoint regression analysis was applied to identify temporal trends and potential changes in trends over time. This method employs segmented linear regression models to divide the time series into multiple contiguous intervals, allowing distinct linear trends to be estimated for different time periods^[14]. In this study, joinpoint analyses were conducted using the Joinpoint Regression Program (version 5.1.0). The analysis was performed on a logarithmic scale. Standard errors were calculated based on 95% uncertainty intervals (UI). Model selection was performed using the permutation test approach, with an overall significance level set at 0.05. A total of 4,499 permutations were used, and the model permitted up to six joinpoints.

The age-period-cohort model is a widely used statistical framework to analyze and interpret demographic data, particularly in epidemiology, sociology, and public health research. It facilitates the separation and evaluation of the distinct contributions attributable to age, temporal period, and birth cohort with respect to specific health outcomes or behavioral patterns. In the present study, a web-based analytical platform was utilized to examine how these three dimensions influence the aforementioned measures of disease burden; further methodological details are available at: https://analysistools.cancer.gov/apc^[15]. The analysis incorporated data collected over three decades (1994-2023), with the overall timeframe segmented into successive five-year intervals to support age-period-cohort modeling.

Decomposition analysis is an important statistical approach for identifying the underlying drivers of changes in specific outcomes over time. This method has been widely applied in demographic studies, econometrics, and public health research, as it enables the attribution of changes in aggregate indicators, such as mortality rates, disease incidence, or economic ratios- to distinct contributing factors. In this study, we applied the Das Gupta method to decompose changes in the burden of AF/AFL over the past 34 years into three components: population aging, population growth, and epidemiological change. This approach allowed us to quantify the relative contributions of demographic structure and disease risk to the observed temporal trends in AF/AFL burden.

A nonparametric data envelopment analysis (DEA) framework was applied to conduct frontier analysis. To assess the robustness and stability of the estimated frontier, a resampling procedure was implemented. The original dataset was resampled with replacement for 100 iterations. In each iteration, a bootstrap sample of the same size as the original dataset was generated through random sampling with replacement^[16]. In this analysis, the Socio-demographic Index (SDI) was defined as the input variable and DALYs as the output variable. The frontier was then estimated to represent the minimum achievable DALYs at different levels of SDI. Finally, locally weighted regression (LOESS) was used for data visualization to illustrate temporal trends and relative performance. This approach was applied to depict changes in DALYs attributable to AF/AFL in China from 1990 to 2023, in comparison with 204 countries.

The ARIMA model was employed as a time-series forecasting approach to project future disease burden trends. The ARIMA framework consists of three components: the autoregressive (AR), the differencing (I), and the moving average (MA). Annual data were used, with 1990 defined as the starting year and each calendar year treated as a single observation. If the time series was non-stationary, first-order differencing was applied to achieve stationarity. Model parameters - including the autoregressive order (p), degree of differencing (d), and moving average order (q) - were automatically selected using the auto.arima function based on optimal model fit criteria. Finally, the Ljung-Box test was performed to ensure model adequacy by verifying that the residuals were white noise [Supplementary Table 1]. The finalized ARIMA model was then applied to forecast disease burden indicators for the 15-year period following 2023 to estimate future trends in atrial fibrillation and flutter - related burden.

Statistical analyses were performed in R (version 4.3.2), with statistical significance defined as P value < 0.05.

From 1990 to 2023, the numbers of incident cases, prevalent cases, deaths, and DALYs attributable to AF/AFL in China exhibited an overall increasing trend in both males and females. In contrast, the age-standardized incidence rate (ASIR) and age-standardized prevalence rate (ASPR) demonstrated a non-linear temporal pattern, characterized by an initial decline followed by a subsequent increase. Notably, the upward trend was most pronounced during the period from 2005 to 2010 [Figure 1A-D]. Overall, these findings indicate that although the absolute burden of AF/AFL increased substantially, the age-standardized rates followed a non-linear pattern with a marked rise during 2005-2010.

Specifically, the number of incident AF/AFL cases in China increased from 312,558 (95% UI: 233,376-399,993) in 1990 to 1,057,528 (95% UI: 809,494-1,352,720) in 2023. Over the same period, the number of prevalent cases rose from 3,307,425 (95% UI: 2,484,120-4,223,896) to 12,394,235 (95% UI: 9,435,370-15,598,110). For both incident and prevalent cases, the absolute numbers were consistently higher in males than in females [Table 1]. In contrast, the number of deaths attributable to AF/AFL increased from 19,947 (95% UI: 15,522-24,005) in 1990 to 65,008 (95% UI: 51,849-75,275) in 2023, while DALYs rose from 562,194 (95% UI: 452,649-720,720) to 1,745,719 (95% UI: 1,398,854-2,293,705). Notably, females accounted for a higher number of deaths and DALYs than males throughout the study period. In summary, while males consistently exhibited higher incidence and prevalence, females had higher burden in terms of mortality and DALYs.

With regard to age-standardized rates, the ASIR increased from 41.33 per 100,000 population (95% UI: 32.43-51.93) in 1990 to 46.36 per 100,000 population (95% UI: 36.04-58.14) in 2023, while the ASPR rose from 462.92 per 100,000 population (95% UI: 348.45-577.62) to 541.74 per 100,000 population (95% UI: 410.72-679.94). Across the study period, males exhibited higher ASIR and ASPR than females in most years. By contrast, the age-standardized mortality rate (ASMR) demonstrated an overall declining trend in China, decreasing from 5.12 per 100,000 population (95% UI: 3.89-6.35) in 1990 to 3.23 per 100,000 population (95% UI: 2.56-3.76) in 2023, with females exhibiting higher ASMR than males in all years. Similarly, the age-standardized disability-adjusted life years rate (ASDR) declined from 96.99 per 100,000 population (95% UI: 78.52-119.37) to 79.29 per 100,000 population (95% UI: 64.23-103.01). Females consistently had higher ASDR than males before 2007; however, after 2007, the ASDR was broadly comparable between females and males. Overall, these results suggest that although incidence and prevalence rates slightly increased, mortality and overall disease burden rates declined, with females generally experiencing higher rates than males.

With respect to age distribution, in 2023, incident and prevalent AF/AFL cases in China reached their highest levels among individuals aged 70-74 years in both sexes. Prior to this age group, the numbers were higher in males than in females, whereas beyond the 70-74-year age group, females outnumbered males. For mortality and DALYs, older females accounted for substantially higher numbers than males, particularly among individuals aged 75 years and older [Figure 1E-H]. In summary, the burden of AF/AFL was concentrated in older populations, peaking at ages 70-74, with females experiencing disproportionately higher mortality and DALYs at advanced ages.

To further characterize temporal trends in AF/AFL in China from 1990 to 2023 by sex, joinpoint regression analyses were conducted [Figure 2A-D]. The results indicated that the ASIR declined most rapidly during 1990-1994 in both males and females, with annual percent changes (APC) of -3.86 (P < 0.05) and -2.47 (P < 0.05), respectively. The most pronounced increases in ASIR were observed during 2000-2004 in males (APC = 6.37, P < 0.05) and during 2000-2005 in females (APC = 3.95, P < 0.05). After 2010, ASIRs in both sexes generally exhibited a slow upward trend.

Similar temporal patterns were observed for the ASPR. The steepest declines occurred during 1990-1994, with APCs of -3.93 (P < 0.05) in males and -2.08 (P < 0.05) in females. The fastest increases in ASPR were identified during 2000-2004 among males (APC = 7.45, P < 0.05) and during 2000-2005 among females (APC = 3.98, P < 0.05).

The ASMR demonstrated an overall declining trend in both males and females. The most rapid decreases were observed during 1998-2004 in males (APC = -5.24, P < 0.05) and during 1999-2004 in females (APC = -6.01, P < 0.05). This trend reversed during 2020-2023, with ASMR increasing in both sexes, with APCs of 4.47 (P < 0.05) in males and 3.80 (P < 0.05) in females.

For ASDR, females experienced a prolonged period of decline, with the steepest decrease occurring during 1996-2004 (APC = -2.15, P < 0.05). Similar to the pattern observed for ASMR, both males and females exhibited increasing ASDR during 2020-2023, with APCs of 0.78 (P < 0.05) and 2.10 (P < 0.05), respectively.

To further clarify the independent contributions of age, period, and cohort to AF/AFL patterns in China by sex, age-period-cohort model analyses were performed. Regarding incidence, rates in females surpassed those in males beyond the 75-79-year age group. Both period and cohort patterns suggested a progressive rise in incidence over successive calendar periods and cohorts [Figure 3A-C]. Among individuals born after 1944, males showed higher incidence rates compared with females [Supplementary Table 2].

The patterns observed for prevalence were generally consistent with those for incidence in both period and cohort effects, with higher prevalence rates among males born after 1944 [Supplementary Table 3], which may partly reflect the increased incidence in this birth cohort. In terms of age effects, prevalence rates were consistently higher in males than in females across all age groups [Figure 3D-F].

For mortality, age effects showed no marked sex differences before the 80-84-year age group; however, beyond this age, mortality rates were higher in females than in males. Regarding period effects, males had lower mortality rates than females before 2004-2009, whereas males surpassed females thereafter [Supplementary Table 4]. Similarly, cohort effects revealed higher mortality rates among males born after 1944 compared with their female counterparts [Figure 3G-I].

To further quantify the contributions of aging, population growth, and epidemiological change to the burden of AF/AFL, a decomposition analysis was performed. Between 1990 and 2023, aging, population growth, and epidemiological change all contributed to the increases in incidence and prevalence, with aging as the predominant driver in both males and females [Figure 4A and B]. In the total population, aging accounted for 72.62% of the increase in incidence and 70.90% of the increase in prevalence [Supplementary Table 5].

In contrast, epidemiological change contributed negatively to mortality and DALYs [Figure 4C and D], indicating overall declines in AF/AFL-related mortality and DALYs in China from 1990 to 2023. This negative contribution was more pronounced among females, accounting for -86.62% of the change in mortality and -34.22% of the change in DALYs. Nevertheless, aging remained the dominant contributor to changes in mortality and DALYs, accounting for 147.95% and 103.17% of the total change in the overall population, respectively.

To examine changes in the ASDR in China in relation to increasing sociodemographic development, a frontier analysis was performed using data from 204 countries and territories spanning 1990-2023. The results showed that China’s ASDR declined steadily from 1990 (96.99 per 100,000 population) to 2013 (75.78 per 100,000 population) [Supplementary Table 6]. Thereafter, ASDR exhibited a fluctuating pattern and demonstrated an overall upward trend after 2020 [Figure 5A].

Furthermore, among countries and territories with an SDI < 0.5 in 2023, the five locations with the lowest ASDR were Somalia, Comoros, Mozambique, Djibouti, and Eritrea [Figure 5B], with Somalia showing the lowest burden (45.61 per 100,000 population). Similarly, among countries and territories with an SDI > 0.85, the five locations with the lowest ASDR were Singapore, Japan, the United Arab Emirates, Kuwait, and Taiwan (China), among which Singapore had the lowest ASDR (53.25 per 100,000 population).

ARIMA models were applied to estimate future patterns of AF/AFL burden in China for the upcoming 15-year period. The projections indicated that the ASIR among males is projected to decline from 49.22 per 100,000 population in 2023 to 41.14 per 100,000 population in 2028, followed by a plateau thereafter. In contrast, ASIR among females is projected to increase from 42.00 per 100,000 population in 2023 to 49.50 per 100,000 population in 2028 [Supplementary Table 7], followed by a gradual decline to 47.10 per 100,000 population by 2033 [Figure 6A-C].

For prevalence, the ASPR among males is projected to decrease from 614.89 per 100,000 population in 2023 to 410.89 per 100,000 population in 2027. A similar declining trend is projected among females, with ASPR decreasing from 472.17 per 100,000 population in 2023 to 439.25 per 100,000 population in 2028 [Figure 6D-F].

Regarding mortality and overall disease burden, both the ASMR and the ASDR in the total population are projected to decline steadily over the next 15 years, decreasing from 3.23 to 2.53 per 100,000 population and from 79.29 to 71.48 per 100,000 population, respectively, between 2023 and 2038 [Figure 6G-L].

This study has several limitations that warrant consideration. First, the analyses were based on nationally aggregated data, which may obscure substantial regional and subnational heterogeneity in the burden of AF/AFL. As a result, the findings may have limited applicability for informing individual-level research or locally tailored interventions. Second, reliance on secondary data sources may introduce potential biases, including inaccuracies in data collection, misclassification or misdiagnosis of disease, and reporting bias, particularly in low-SDI settings where limited healthcare access and weaker surveillance systems may lead to underdiagnosis and incomplete cause-of-death registration. Third, estimates from the GBD framework are model-based and depend on statistical assumptions and data imputation, especially in regions with sparse or low-quality data, which may introduce uncertainty and affect the robustness of the results. Although uncertainty intervals are provided, they may not fully capture all sources of systematic error. Fourth, temporal trends may be influenced by changes in diagnostic criteria, coding practices, and the accessibility of healthcare services over time, which can compromise the comparability of estimates across years and, consequently, reduce the accuracy of ARIMA models in forecasting future trends.

In addition, the GBD database categorizes multiple conditions into relatively broad disease groups, which may mask specific clinical subtypes or contextual health issues that warrant targeted prevention and management strategies. Future studies incorporating more granular, regionally representative, and individual-level data are needed to better characterize local patterns of AF/AFL burden and to support the development of more precise and effective public health interventions.

Despite improvements in age-standardized mortality and DALYs, the overall burden of AF/AFL in China continues to increase, primarily driven by population aging. Targeted prevention, early detection, and optimized management strategies - particularly among older adults and high-risk males - are urgently needed.