Data on incidence, prevalence and YLDs of fall-related SCI were retrieved from the 2023 Global Burden of Disease (GBD 2023) for China, India, Japan, the United States and the global average [15]. Covering 375 diseases and injuries across 204 countries/territories from 1990 to 2023, all estimates including 95% uncertainty intervals (UIs) were acquired via the Global Health Data Exchange tool (https://vizhub.healthdata.org/gbd-results/). We extracted data with settings: cause as spinal cord injury, risk factor as falls, and metrics as incidence, prevalence and YLDs. All rates were age-standardized per 100,000 population based on the GBD global standard population.

GBD 2023 integrated over 310,000 data sources via standardized models. It linked ICD codes for SCI (ICD-10: S14, S24, S34, T06.0; ICD-9: 806, 952) and falls (ICD-10: W00-W19; ICD-9: E880-E888) to confirm their causal association, and classified SCI into neck-level and below-neck-level subtypes differing in clinical severity. Estimates were produced using Bayesian models including DisMod-MR 2.1 and comparative risk assessment, stratified by age, sex, region and year with 95% UIs. Strict data cleaning, outlier screening, model imputation, cross-validation and sensitivity analyses were adopted to guarantee reliable results [16].

To quantify the overall linear trend from 2004 to 2023, we calculated the EAPC for each age-standardized rate (ASR). This was achieved by fitting a linear regression model to the natural logarithm of the annual ASR: ln ASR = α + β × year + ε

Where β represents the slope. The EAPC and its 95% confidence interval (CI) were derived as 100×(e^β-1). A trend was deemed statistically significant if the 95% CI of the EAPC did not include zero.

To detect significant inflection points and temporal phases of trends, we conducted joinpoint regression using the National Cancer Institute’s Joinpoint Regression Program (Version 5.0.2) [17]. This iterative algorithm determines the optimal number of valid joinpoints and divides time series into linear segments, with a maximum of 3 joinpoints allowed. The optimal model was selected based on the Bayesian Information Criterion (BIC). The model estimates segmental annual percentage change (APC) and overall average annual percentage change (AAPC), a weighted average of individual APCs. Results were regarded as statistically significant if the 95% CI excluded zero.

An age-period-cohort model was built to explore trend drivers. The intrinsic estimator method was adopted to address the inherent identifiability issue of the model [18, 19]. This method applies constraints orthogonal to the design matrix null vector to eliminate bias from reference category selection, and decomposes age, period and cohort effects into orthogonal components free of arbitrary parameterization [20]. Data were grouped into 5-year age bands (<5 to 95+ years) and 5-year calendar periods (2004–2008 to 2019–2023) to define corresponding birth cohorts.

The model assumes Poisson-distributed case counts and adopts a log-linear equation. Coefficients of period and cohort effects were exponentiated to calculate risk ratios relative to the overall mean, and age effect coefficients were exponentiated to obtain age-specific rate ratios. Analyses were conducted using the apc package in R (version 4.3.1).

Demographic decomposition analysis was conducted to quantify drivers of temporal changes in fall-related SCI burden. Total variations in incidence, prevalence and YLD case counts from 2004 to 2023 were decomposed into three factors: population growth, population aging, and shifts in age-specific epidemiological rates. A Kitagawa-based approach was applied for exact additive decomposition of case count differences between the 2 years. Each factor’s contribution was assessed via counterfactual analyses: one factor was adjusted to its 2023 level while the other two remained at 2004 values. The total case change ΔY was split as shown in the formula. Δ Y = Δ Y  population + Δ Y  aging + Δ Y  rates

Positive values indicate increased disease burden, while negative values denote reduced burden. This method enables accurate attribution of overall burden changes to demographic and epidemiological factors.

This study used aggregated, de-identified public data from GBD, so ethics approval and informed consent were waived. Raw data are accessible via the Global Health Data Exchange. Analytical code can be obtained from the corresponding author upon reasonable request.

Cross-country comparisons revealed substantial disparities in the burden of fall-related SCI in 2023 (Supplementary Material 1). The highest ASIRs were observed in the United States (13.78 per 100,000; 95% UI: 7.98–24.26) and China (12.61; 7.79–20.23), both markedly exceeding the global average (6.93; 4.32–11.32). China also recorded the highest ASPR (359.00; 308.05–410.74) and ASYR (99.32; 66.97–132.15), followed by the United States, whereas India and Japan consistently exhibited the lowest burden. In absolute terms, China accounted for approximately 185,462 incident cases, 6.44 million prevalent cases, and 1.75 million YLDs in 2023. From 2004 to 2023, India showed the greatest declines across all burden indicators, while Japan and the United States also experienced significant reductions. In contrast, China exhibited slight increases in ASIR (EAPC = 0.84%) and ASPR (EAPC = 0.50%), with ASYR remaining relatively stable (EAPC = −0.16%).

Joinpoint regression revealed considerable non-linear variation in these trends (Figure 1; Supplementary Materials 2, 3). Most countries followed a three-phase pattern characterized by stability or increase before 2015, decline through 2020, and subsequent rebound. China showed the largest fluctuations, with all indicators peaking around 2015. ASIR increased rapidly during 2004–2010 (APC = 7.31%), declined during 2015–2020 (APC = −7.52%), and rose again thereafter (APC = 3.58%), resulting in a significant overall increase (AAPC = 1.62%). By contrast, the United States maintained relatively stable incidence while achieving marked reductions in prevalence (AAPC = −2.42%) and disability burden (AAPC = −2.58%). India and Japan demonstrated sustained improvements across all burden measures throughout the study period.

Age-stratified analyses showed that incidence, prevalence, and YLDs generally increased with age, although the magnitude varied across countries (Figure 2). China and the United States exhibited the steepest age-related increases, whereas India and Japan showed more gradual age gradients. China displayed a rise-peak-decline-rebound pattern across age groups, while the United States showed declining burdens below age 50 but increasing burdens among older adults. Males consistently experienced higher burdens than females, and cervical injuries contributed more to disability burden than injuries below the neck. Across sex and injury subgroups, China and the United States consistently exhibited the highest burdens, whereas India and Japan had the lowest (Supplementary Material 4–6).

Age effects revealed substantial heterogeneity in life-course risk profiles across countries (Figure 3). Prevalence and YLDs exhibited age patterns largely consistent with incidence within each country. China showed a pronounced age-progressive pattern, with incidence RR increasing from 0.43 (95% CI: 0.37–0.49) in the youngest age group to 2.78 (95% CI: 2.24–3.46) among those aged 95–99 years, indicating aging as the dominant risk factor. In contrast, India and the United States displayed age-regressive profiles, with incidence RR declining from 1.76 (1.65–1.88) to 0.60 (0.55–0.66) in India and from 2.03 (1.69–2.44) to 1.69 (1.45–1.97) in the United States. The United States also exhibited the steepest age gradient in prevalence, declining from 5.93 (5.12–6.88) to 0.53 (0.46–0.60). Japan demonstrated an intermediate pattern, with elevated risk persisting through mid-adulthood before gradually approaching baseline levels in older age groups.

Period effects demonstrated contrasting temporal dynamics among countries (Figure 4). China exhibited a pronounced inverted-V pattern across all burden metrics, with incidence RR increasing from 0.72 (0.70–0.75) in 2004–2008 to 1.10 (95% CI:1.06–1.14) in 2014–2018 before declining to 0.97 (0.93–1.01) in 2019–2023. Similar trajectories were observed for prevalence and YLDs. By contrast, India, Japan, and the United States experienced sustained reductions in period-related risk. The decline was most pronounced in the United States, where prevalence and YLD RRs decreased to 0.71, representing reductions exceeding 30%. India showed steady improvements across all metrics, whereas Japan experienced more gradual declines throughout the study period.

Cohort effects identified two distinct generational patterns (Figure 5). China and the United States exhibited unimodal trajectories characterized by an initial rise followed by a decline. In China, incidence RR increased from 0.19 (95% CI: 0.04–0.98) in the earliest birth cohort to a peak of 1.83 (1.61–2.09) around the 2004 cohort before falling to 0.89 (0.72–1.11) in the most recent cohort. The United States showed a similar pattern but peaked earlier and at a lower level, followed by a marked decline to 0.10 (0.07–0.14) in the youngest cohort. In contrast, India and Japan demonstrated sustained generational improvements after an early high-risk plateau. Incidence RR in India declined from approximately 1.38 among cohorts born between 1929 and 1939 to 0.38 (0.33–0.44), while Japan followed a comparable but less pronounced trajectory. Prevalence and YLD cohort effects closely mirrored the incidence patterns. Overall, cohort trends suggest a concentrated generational burden in China and the United States, but progressive attenuation of risk across successive generations in India and Japan.

Decomposition analysis revealed distinct drivers of changes in fall-related SCI burden between 2004 and 2023 (Figure 6, Supplementary Material 7). In China, epidemiological change was the primary driver of increased incident cases (+64,385; +53.18%), accounting for 68.37% of growth, whereas population aging contributed most to increases in prevalence (+61.07%) and YLDs (+39.10%), accounting for 39.53% and 44.55% of growth, respectively. In India, population growth was the dominant positive contributor across all burden measures, but its effect was largely offset by favorable epidemiological changes, particularly for incident cases (−78.45%). In Japan and the United States, demographic and epidemiological forces acted in opposite directions. Although population aging strongly increased incident cases (383.79% in Japan and 92.18% in the United States), improvements in age-standardized rates partially offset these effects. Consequently, Japan experienced declines in prevalent cases (−1.62%) and YLDs (−2.97%), whereas the United States showed modest increases in prevalence (+4.41%) and YLDs (+0.44%). Detailed results are presented in Supplementary Material 8.

This study examined the burden of fall-related SCI in China, the United States, Japan, and India from 2004 to 2023 using age-period-cohort and demographic decomposition analyses. The findings reveal substantial cross-country heterogeneity, largely reflecting differences in demographic transitions and health system responses. While aging and population growth remain key drivers of burden, their impacts vary according to national capacities for prevention, treatment, rehabilitation, and long-term care. Overall, the results underscore the importance of integrated, life-course approaches to injury prevention and management, and provide evidence to support context-specific strategies for reducing the burden of fall-related SCI in diverse demographic settings.