The present study utilizes data sourced from the Global Burden of Disease Study 2021 (GBD 2021; https://vizhub.healthdata.org/gbd-results/). GBD 2021 provides systematic health estimates for 369 diseases and injuries and 87 risk factors across 204 countries and regions. These estimates are primarily based on methods such as the Cause of Death Ensemble Modeling (CODEm), Spatiotemporal Gaussian Process Regression (ST-GPR), and the Bayesian meta-regression tool DisMod-MR, which estimate age-specific incidence rates, mortality rates, disability-adjusted life years (DALYs), and demographic data for each country, stratified by cause, age, sex, year, and location [8]. Compared to GBD 2019, GBD 2021 incorporates an additional 147 surveys, 21 censuses, and 634 country-year life and sample registration data, bringing the total to 1455 surveys and censuses and 8709 country-year life and sample registration datasets, along with 150 other sources. Methodological enhancements feature temporal weighting optimization in ST-GPR via beta density functions, improving trend estimation accuracy through enhanced data representativeness [9]. AF/AFL is categorized as a Level 3 cause within the broader Level 2 cause category of Cardiovascular Disease, which is a subclassification of Level 1 non-communicable diseases [10]. According to the 2021 Disease Classification, non-fatal causes and injuries from AF/AFL diagnosed by electrocardiogram are mapped to the corresponding International Classification of Diseases (ICD) codes in GBD 2021, which are 472.3–427.32 (ICD-9) and 148–148.92 (ICD-10) [11]. AF/AFL in GBD 2021 is defined as ECG-confirmed arrhythmia meeting all three criteria:Irregular RR intervals; Absence of distinct P waves; Atrial cycle length 200 ms when visible. Cases represent unique individuals with incident or prevalent AF/AFL within a calendar year, excluding duplicate records from multiple healthcare encounters [12]. GBD evaluates the availability and completeness of Vital registration and verbal autopsy data by reporting the percentage of total deaths from specific causes in a given location and year, excluding sources with completeness below 50%. The quality of mortality data is also assessed based on the availability and completeness of life registration and verbal autopsy data, with ratings ranging from 0 to 5 stars. A higher star rating indicates better completeness, availability, and fewer “garbage codes” (i.e., deaths assigned to ambiguous diagnoses or conditions that cannot be potential causes of death according to ICD-10 classification) [13]. This study assesses the quality of available mortality data for 19 EU15+ countries during the period 1990–2020, with 15 countries receiving a 5-star rating, indicating that 85%–100% of their mortality data are well-validated. The remaining four countries (Belgium, France, Greece, and Luxembourg) received a 4-star rating, signifying that 65%–84% of their data are well-validated [13]. These 19 countries have been used in previous observational reports [14, 15] because of their similarities in health expenditure and comparability.

The data on ASIR and ASMR per 100,000 population for EU15+ countries from 1990 to 2021 were extracted from the GBD Results Tool. These age-standardized rates were derived using a standard population based on the unweighted average of populations across all countries for each 5-year age group, as defined in the United Nations World Population Prospects (2012 Revision) [16] for 2010–2035. The differences in ASIR and ASMR at the beginning and end of the observation period were calculated for each gender in all 19 countries to facilitate comparisons of absolute and relative changes over time. To quantify the proportion of ASMR to ASIR per 100,000 population among males and females in EU15+ countries from 1990 to 2021, MII was calculated by dividing mortality rates by incidence rates and multiplying by 1000. MII provides a standardized measure to assess disease burden and is commonly used to identify disparities in cancer screening and treatment internationally [17]. By analyzing temporal changes in MII, long-term trends and progress in disease management can be effectively tracked.

Longitudinal trends in AF/AFL disease burden across EU15+ nations were analyzed using Joinpoint Regression Analysis v4.5.0.1 (National Cancer Institute Surveillance Research Program, Bethesda, MD, USA). This methodology identifies temporal inflection points in epidemiological trends by fitting the most parsimonious segmented linear model to log-transformed annual rates. The algorithm iteratively evaluates potential trend breaks through permutation testing with Bonferroni correction, permitting $\le$5 significant joinpoints (p 0.05, Monte Carlo-adjusted for multiple comparisons) [18]. Joinpoint regression has been extensively employed in previous cardiovascular disease burden studies to evaluate long-term trend changes [19]. This analysis calculates average percentage change, average annual percentage change (AAPC), and their 95% confidence intervals (CI) to assess temporal trends in AF/AFL morbidity and mortality from 1990 to 2021.

The age-period-cohort (APC) model, a widely used framework in contemporary epidemiology, is built upon the Poisson distribution and allows simultaneous assessment of age, period, and cohort effects on AF/AFL incidence trends. Beyond traditional epidemiological analysis, the APC model aids in uncovering societal changes, disease etiology, aging processes, and demographic dynamics. It has been extensively applied to descriptive epidemiological studies on cardiovascular diseases [20].

The longitudinal age curve represents the relative risk of age groups compared to a reference group, adjusted for period effects. Rate ratios (RR) for periods or cohorts indicate relative risks compared to reference cohorts or periods, accounting for nonlinear period or cohort effects.

We used the GBD 2021 database to estimate the morbidity of AF/AFL as well as demographic data from individual EU15+ countries as inputs to the APC model.To minimize overlap between adjacent cohorts, mortality and population data were divided into consecutive 5-year age groups (1992–2021) and corresponding 5-year birth cohorts. Midpoint years (e.g., 1992–1996 represented by 1994) were used to approximate specific time intervals, resulting in 14 age groups (30–34 to 95–99 years) and 19 birth cohorts (1897–1901 to 1987–1991). The analysis was conducted using the R-based web tool provided by the National Cancer Institute’s Division of Cancer Epidemiology and Genetics (http://analysistools.nci.nih.gov/apc/), which reduces potential confounding and bias. To address limitations of the APC model, stricter selection criteria were applied.Statistical analyses were performed using R software v4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). The significance of estimable functions was evaluated using the Wald χ² test, with two-sided p-values 0.05 indicating statistical significance.

Fig. 1 shows the ASIR for males and females in EU15+ countries for the years 1990 and 2021. In 1990, Canada had the highest ASIR for both males (125.34/100,000) and females (80.34/100,000). In 2021, Sweden had the highest ASIR for both males (136.71/100,000) and females (109.47/100,000). In 1990, the UK had the lowest ASIR for both males (64.93/100,000) and females (41.58/100,000). By 2021, Belgium had the lowest ASIR for males (63.04/100,000), the UK had the lowest ASIR for females (40.68/100,000), and Belgium’s female ASIR was also among the lowest (42.83/100,000). Throughout both 1990 and 2021, the ASIR for males was consistently higher than that for females in all 19 EU15+ countries.

Fig. 1.

Age-standardized incidence rates (ASIR) per 100,000 for atrial fibrillation and atrial flutter (AF/AFL) in 1990 and 2021 in European Union 15+ (EU15+) countries. (A) Female ASIR in 1990. (B) Male ASIR in 1990. (C) Female ASIR in 2021. (D) Male ASIR in 2021.

Fig. 2 presents the ASMR for males and females in EU15+ countries for 1990 and 2021. In 1990, Australia had the highest ASMR for males (7.38/100,000), while Finland had the highest ASMR for females (7.79/100,000). By 2021, Sweden had the highest ASMR for both males (10.29/100,000) and females (8.83/100,000). In 1990, Italy had the lowest ASMR for males (3.45/100,000), while the USA had the lowest ASMR for females (3.28/100,000). In 2021, Portugal had the lowest ASMR for both males (3.89/100,000) and females (3.18/100,000). In both 1990 and 2021, the ASMR for males was higher than that for females in all 19 EU15+ countries.

Fig. 2.

ASMR per 100,000 for AF/AFL in 1990 and 2021 in EU15+ countries. (A) Female ASMR in 1990. (B) Male ASMR in 1990. (C) Female ASMR in 2021. (D) Male ASMR in 2021.

Fig. 3 presents the AF/AFL incidence rate data from 1990 to 2021 for each EU15+ country, disaggregated by gender. Among women, a decrease in ASIR was observed in 14 countries from 1990 to 2021. Finland showed the largest decrease (–27.3%), followed by Canada (–25.5%), while Austria (+75.9%) and Sweden (+42.9%) experienced the greatest increases in ASIR. In men, during the same period, the ASIR decreased in 12 countries, with Finland (–25.7%) and Canada (–17.8%) showing the largest decreases. Austria (+82.1%) and Sweden (+48.2%) saw the largest increases in male ASIR.

Fig. 3.

Trends in age-standardized incidence rates per 100,000 for AF/AFL in EU15+ countries between 1990 and 2021. Blue squares indicate males and red squares indicate females.

Tables 1,2 summarize the joint point analysis results of male and female AF/AFL ASIR from 1990 to 2021, including average percentage change and AAPC for each trend interval. From the AAPC, it is evident that the incidence rates of AF/AFL have generally decreased for women in 15 countries between 1990 and 2021. Finland showed the most significant decrease in incidence (–1.04%, 95% CI: –1.08%, –0.99%), while Austria, Australia, Sweden, and the USA showed increases, with Austria experiencing the largest increase (+1.86%, 95% CI: 1.82%, 1.90%). In men, 12 countries saw a decline in incidence, with Finland showing the largest decrease (–0.95%, 95% CI: –1.02%, –0.88%). Austria, Denmark, Germany, Luxembourg, Spain, Sweden, and the USA experienced increases, with Austria again showing the largest increase (+1.97%, 95% CI: 1.93%, 2.01%). However, the rate of decline varied across countries. According to the average percentage change, Ireland exhibited the fastest decrease in AF/AFL incidence for women between 1996–2000 (–4.9%), while Australia saw the fastest decline in male incidence from 2019–2021 (–5.04%). The most recent trends show that, among the 19 EU15+ countries, 16 countries reported an increase in female AF/AFL incidence, and 15 countries observed an increase in male incidence. The highest increases in incidence from 2019–2021 were observed in Italy for men (+10.25%) and Sweden for women (+13.54%).

Fig. 4 shows the AF/AFL mortality rate data for each EU15+ country from 1990 to 2021, disaggregated by gender. Among women, a decrease in ASMR was seen in 12 countries, with Finland (–46.0%) and Portugal (–31.1%) showing the largest decreases, while Sweden (+85.0%) and the USA (+38.8%) had the largest increases. In men, during the same period, 11 countries experienced a decline in ASMR, with Finland (–27.2%) and Ireland (–18.9%) showing the largest decreases. However, Sweden (+107.3%) and Denmark (+42.0%) experienced the greatest increases in male ASMR.

Fig. 4.

Trends in age-standardized mortality rates per 100,000 for AF/AFL in EU15+ countries between 1990 and 2021. Blue squares indicate males and red squares indicate females.

Tables 3,4 summarize the joint point analysis results of male and female AF/AFL ASMR from 1990 to 2021. From the AAPC, it is observed that, for most countries, both male and female mortality rates showed a general decreasing trend. Finland showed the largest decrease in female (–1.93%, 95% CI: –2.39%, –1.47%) and male mortality rates (–1.06%, 95% CI: –1.43%, –0.69%). Conversely, Sweden experienced the largest increase in mortality rates for both sexes, with females showing a +1.97% increase (95% CI: 1.49%, 2.46%) and males a +2.38% increase (95% CI: 1.27%, 3.50%). According to the average percentage change, Finland exhibited the most significant decrease in female mortality rates during the period 2007–2010 (–8.05%), while Ireland showed the largest decline in male mortality rates from 2005–2008 (–7.92%). Additionally, Sweden showed the most rapid increase in mortality rates during 2003–2006, with female mortality increasing by 9.87% and male mortality by 13.48%.

Supplementary Fig. 1 (Supplementary Material 1) presents the MII (Mortality-to-Incidence Ratio) data for both men and women in EU15+ countries in 1990 and 2021. In 1990, the median MII for women was 0.095, and for men, it was 0.067. The highest MII values for women were observed in the UK (0.127), Austria (0.118), and Australia (0.117), while Canada (0.054), the USA (0.058), and Italy (0.060) had the lowest values. For men in 1990, the highest MII values were in the Netherlands (0.089), the UK (0.088), and Luxembourg (0.087), while Italy (0.045), Canada (0.046), and the USA (0.051) had the lowest values. In 2021, the median MII for women was 0.092, and for men, it was 0.072. The highest MII values for women in 2021 were seen in the UK (0.138), Germany (0.135), and Luxembourg (0.122), while Portugal (0.057), the USA (0.062), and Canada (0.065) had the lowest values. For men in 2021, the highest MII values were in the Netherlands (0.092), the UK (0.090), and Luxembourg (0.083), while Canada (0.047), Austria (0.054), and Spain (0.056) had the lowest values. In 1990, the MII for women was higher than for men in all EU15+ countries. In 2021, except for Portugal, where the male MII was higher, the MII for women remained higher than for men.

The MII trends for each country, categorized by gender, are shown in Supplementary Fig. 2 (Supplementary Material 1). It is observed that the trend changes for MII in both men and women were not uniform across countries. Among women, 12 countries experienced an upward trend in MII, with the largest increase in Germany (+35.62%). Conversely, 7 countries showed a downward trend, with the largest decrease in Austria (–32.58%). Among men, 10 countries saw an increase in MII, with the largest increase in Sweden (+39.89%), while the remaining 9 countries showed a decrease, with Austria again showing the largest decrease (–30.66%).

Supplementary Tables 1,2 (Supplementary Material 1) present the joint point analysis results for MII data of men and women in EU15+ countries during the observation period. The trend changes for each country were not uniform across the different trend intervals, and the average percentage change values indicate the following key observations: From 1996 to 1999, the MII for women in Ireland increased the fastest (+8.83%), while from 2019 to 2021, it decreased the slowest (–16.07%). Among men, from 2003 to 2006, Sweden saw the fastest increase in MII (+15.00%), and from 2019 to 2021, Italy experienced the largest decrease in MII (–11.15%). These inflection points were all statistically significant. According to the AAPC values, most EU15+ countries exhibited an overall upward trend in MII for both men and women. The fastest increase in MII was observed in Germany for women (+1.00%, 95% CI: 0.58%, 1.41%) and Italy for men (+1.15%, 95% CI: 0.60%, 1.71%). The largest decrease in MII was seen in Austria, with a –1.06% decrease (95% CI: –1.66%, –0.45%) in men and –1.15% (95% CI: –1.61%, –0.70%) in women.

Net Drift represents the overall annual percentage change in incidence rates from 1992 to 2021. Local Drifts reflect the annual percentage change in incidence rates for specific age groups relative to the net drift. Table 5 and Supplementary Material 3 summarizes the Net Drift and Local Drifts for each age group across the EU15+ countries.During the observation period, Luxembourg exhibited a constant Net Drift with no gender differences. Specifically, the Net Drift for females was –0.46% (95% CI: –1.62%, 0.70%, p = 0.43), for males it was 0.00% (95% CI: –1.24%, 1.26%, p = 1.00), and for both genders combined, it was –0.19% (95% CI: –0.85%, 0.47%, p = 0.57). Additionally, Denmark showed a stable Net Drift for women at –0.19% (95% CI: –0.47%, 0.09%, p = 0.18), while Australia showed a stable Net Drift for both genders combined at –0.06% (95% CI: –0.14%, 0.02%, p = 0.12). Aside from these three countries, all other countries had statistically significant Net Drift (p 0.05). Among them, 13 countries showed a decreasing trend with no gender differences, while 3 countries (Austria, Sweden, and the USA) showed an increasing trend. Regarding gender-specific annual percentage changes, the range of Net Drift was as follows: For females: Ireland showed the most significant decrease at –1.32% (95% CI: –1.73%, –0.90%), while Austria showed the highest increase at 1.63% (95% CI: 1.37%, 1.89%). For males: Finland showed the most significant decrease at –1.17% (95% CI: –1.47%, –0.86%), while Austria showed the highest increase at 1.85% (95% CI: 1.60%, 2.09%). For both genders: Finland showed the most significant decrease at –1.28% (95% CI: –1.44%, –1.12%), while Austria showed the highest increase at 1.82% (95% CI: 1.69%, 1.96%). Local Drifts also reflect the differences in incidence trends for specific age groups. The trend changes for each country were not uniform, with most countries showing local drift values mainly below 0, indicating improvements in AF/AFL incidence. Specifically, excluding countries where time trends were the same for each age group (Finland, Ireland, Italy, Luxembourg, and Norway, with p $>$ 0.05), 12 countries showed a single downward trend for the elderly group (aged 75 and above). The most significant decrease in this group was observed in UK males at –3.55% (95% CI: –5.08%, –2.00%). For the younger age groups, countries like Australia, Austria, Belgium, Canada, France, Sweden, and the USA exhibited single upward trends in local drift. Among these, the most significant increase was seen in Belgian males, with an increase of 1.86% (95% CI: 1.19%, 2.54%).

Figs. 5,6,7 provides the estimates of the effects of age, period, and cohort on the incidence of AF/AFL. According to the Wald chi-squared test results, 10 countries show statistically significant results. These countries (Austria, Belgium, Canada, Denmark, France, Greece, Spain, Sweden, UK, and USA) will be analyzed further.

Fig. 5.

The age effects on AF/AFL for EU15+ in female and male both. Red lines indicate both, green lines indicate females, and Blue lines indicate males.

Fig. 6.

The period effects on AF/AFL for EU15+ in female and male both. Red lines indicate both, green lines indicate females, and Blue lines indicate males.

Fig. 7.

The cohort effects on AF/AFL for EU15+ in female and male both. Red lines indicate both, green lines indicate females, and Blue lines indicate males.

From Fig. 5, we can observe the age-specific longitudinal curves of AF/AFL incidence by gender. In all countries, the incidence of AF/AFL generally increases with age. However, the highest risk of AF/AFL occurs in the age groups of 70–74 years and 75–79 years in all countries, followed by a decrease in risk with further age. A comparison of AF/AFL incidence rates across EU15+ countries for each age group shows notable age differences. Except for Canada and USA, where the male incidence rates are higher than those of females across all age groups, in the other 8 countries, the incidence of AF/AFL in the elderly population (aged 75 and above) shows a shift in gender differences. Specifically, in Denmark, Greece, Sweden, and UK, the shift occurs in the 75–79 years age group, while in France, Spain, and Belgium, the shift occurs in those aged below 80 years, and in Austria, the shift is observed in the 80–85 years age group. After these age groups, female incidence rates exceed male incidence rates, marking a reversal from previously being lower.

From Fig. 6, The trends in the period effects of AF/AFL incidence across all countries from 1990 to 2021 are not uniform. Seven countries show a general decline in period effects, whereas Austria, Sweden, and USA show an increasing trend. Specifically, France exhibits the most significant overall decline, while Austria shows the most noticeable increase. In more recent years, gender differences in the relative risk of AF/AFL incidence appear. In six countries, the relative risk for women is lower than that for men, while in Denmark and Sweden, women have a higher relative risk than men. In Belgium and UK, gender differences exhibit a fluctuating trend, with women in these countries showing an increasing risk of AF/AFL in recent years.

From Fig. 7, the cohort effects on AF/AFL incidence show that in seven countries, individuals born earlier (e.g., in the 1900s) have a higher incidence compared to those born later (e.g., in the 1980s). However, the overall cohort effect shows a decreasing trend in incidence. Conversely, Austria and Sweden exhibit an opposite trend, with incidence rates increasing in more recent birth cohorts. We also observe that the trends in cohort-related incidence rates are generally consistent for both males and females. The only exception is USA, where the cohort incidence for women born after 1967 shows a significant upward trend in contrast to males.