Burden, Risk Factors, and Temporal Trends of Ovarian Cancer in Postmenopausal Women Aged 55 Years and Older: A Comparative Analysis in India, China, and the United States Based on the Global Burden of Disease Study 2023

Abstract

Background

Evidence remains limited on the ovarian cancer (OC) burden among women aged 55 years and older, used here as an operational proxy for the postmenopausal population, across India, China, and the United States.

Methods

We extracted data on OC-related incidence, mortality, disability-adjusted life years (DALYs), and relevant risk factors from the Global Burden of Disease (GBD) 2023 for the period 1990–2023. We assessed age-specific patterns; temporal trends using the estimated annual percentage change and Joinpoint regression; age–period–cohort (APC) patterns and estimable functions; demographic decomposition; and DALY burdens attributable to high body-mass index (BMI) and occupational asbestos exposure. The Autoregressive Integrated Moving Average (ARIMA) model and Bayesian age-period-cohort (BAPC) model were applied to predict age-standardized DALY rates of OC through 2040.

Results

Absolute incident cases, deaths, and DALYs increased in all three countries, whereas age-standardized rates increased in India and generally declined in China and the United States. India had a comparatively greater burden in younger age groups within the study population, whereas the United States had the highest burden at older ages. In China, despite an overall negative net drift, local-drift point estimates were positive among women aged 85 years and older (0.03%–0.82% per year), although the 95% confidence intervals included zero. The proportion of DALYs attributable to high BMI increased in all three countries, and population growth was the principal positive driver of the rising absolute burden. Projections suggested a continued increase in India’s age-standardized DALY rate and further declines in China and the United States through 2040.

Conclusion

Age-standardized OC burden increased in India but declined in China and the United States, whereas the absolute burden rose in all three countries; the non-significant positive local-drift point estimates among the oldest-old in China warrant monitoring. These ecological findings do not establish causality and support country-specific hereditary risk management and strengthened pathways for timely diagnosis and specialist referral.

Introduction

Ovarian cancer (OC) is the most lethal malignancy of the female reproductive system worldwide and a major threat to women’s health.¹ Given its asymptomatic early presentation and frequent late-stage diagnosis, OC is associated with poor long-term survival.² In terms of disease burden, OC accounts for approximately 23% of all gynecologic malignancies but contributes nearly 47% of all reproductive tract cancer-related deaths, underscoring its highly lethal nature.³ In high-income countries, OC ranks as the fifth leading cause of cancer death among women, with a lifetime risk of approximately 1.3% and a corresponding mortality risk of nearly 0.9%.^4,5 Despite lower overall case counts relative to breast and cervical cancer, OC remains a top cause of gynecologic cancer mortality and a persistent global public health challenge.⁶

Age constitutes a key OC risk factor, with disease risk climbing markedly in older females.⁴ Global population aging is an irreversible demographic shift, especially in middle- and high Socio-demographic Index (SDI) regions with rapidly expanding elderly female populations.⁷ Such demographic changes are projected to drive rising OC caseloads and strain local healthcare systems over upcoming decades. Because countries differ in demographic structure, risk exposures, diagnostic pathways, specialist-care capacity, and treatment access, OC burden among women aged 55 years and older may present distinct evolving patterns, necessitating targeted cross-country comparative studies.

Women aged 55 years and older constitute an important population for evaluating OC burden.⁴ In this study, women aged 55 years and older were used as an operational proxy for the postmenopausal population. This threshold aligns with the five-year age structure of the Global Burden of Disease Study 2023 (GBD 2023) and reduces the likelihood of including women who have not undergone natural menopause. However, GBD does not record individual menopausal status; therefore, some premenopausal women may remain in the study population, and individual variation in the timing of menopause cannot be captured. Previous studies have shown that OC patterns vary across countries and birth cohorts.^8,9 Hospital-based evidence from India has reported a median age at diagnosis of around 50 years,¹⁰ while regional inequalities in oncology resources and gynecological care capacity persist in China.^11,12 In the comparative framework of this study, India, China, and the United States represent developing, emerging, and developed economies, respectively. Their differences in demographic structure, socioeconomic conditions, health-care systems, diagnostic and specialist-care capacity, and population risk profiles create an ideal natural cohort for cross-country epidemiologic research. Nevertheless, direct comparisons of OC burden, attributable risk factors, demographic drivers, and projected trends among women aged 55 years and older in these three countries remain limited, creating an evident research gap.

GBD 2023 provides age- and sex-specific estimates of disease burden from 1990 to 2023.¹³ Using these data, this study assessed OC burden, attributable risk factors, demographic drivers, age–period–cohort (APC) patterns, and projected trends through 2040 among women aged 55 years and older in India, China, and the United States. Unlike previous global or single-country analyses based on earlier GBD iterations, this study provides a direct three-country comparison using GBD 2023 and combines Joinpoint regression, APC modeling, decomposition analysis, and Bayesian age–period–cohort (BAPC) and autoregressive integrated moving average (ARIMA) forecasting.

Methods

Data Source

This study utilized data from the GBD Study 2023, which is publicly accessible via the GBD Results Tool (https://ghdx.healthdata.org/gbd-results-tool). The core indicators adopted to evaluate OC burden included incident cases, deaths, and disability-adjusted life years (DALYs). To facilitate comparisons across countries and years, age-standardized rates (ASRs) were used for the principal epidemiological outcomes. The standardized metrics included the age-standardized incidence rate (ASIR), age-standardized mortality rate (ASMR), and age-standardized DALY rate (ASDR). DALYs were calculated as the sum of years of life lost (YLLs) due to premature mortality and years lived with disability (YLDs) attributable to OC.¹⁴ All burden estimates were generated using the Bayesian meta-regression tool DisMod-MR 2.1, the standardized analytical framework applied in the GBD study to synthesize heterogeneous population data and produce consistent disease burden estimates. Uncertainty intervals (UIs) for each indicator were derived from 1000 posterior draws, with 95% UIs defined as the 2.5^th and 97.5^th percentiles of the estimated distribution.

Study Population

Our analysis focused on OC data from the three countries with the largest populations in the world: India, China, and the United States. Estimates were limited to women aged 55 years and older; this age-defined group served as an operational proxy for the postmenopausal population rather than representing biologically confirmed postmenopausal individuals. This age threshold was consistent with prior GBD-based studies investigating postmenopausal populations. It ensured near-complete coverage of naturally postmenopausal women while accommodating the age-stratified structure of the GBD database, which lacks individual-level data on menopausal status.

Descriptive Analysis

In line with established practices, we used open-access estimates from GBD 2023 to derive the number and rates of incident cases, mortality and DALYs attributable to OC. All rates were age-standardized using the global population standard (per 100,000 population) and are reported as ASRs.¹⁵ ASRs facilitate comparisons of disease burden across populations and over time and may help identify priority populations for prevention; however, they do not directly measure or identify underlying risk-factor patterns. This study employed the Estimated Annual Percentage Change (EAPC) to evaluate trends in ASIR, ASMR, and ASDR. EAPCs were derived from a regression model (Y=α+βX+e) applied to ASR over time. The EAPC formula is expressed as 100×[exp(β)–1]. A linear regression model was used to compute the 95% confidence interval (CI) for EAPC.¹⁶ If both the EAPC and its 95% CI lower limit are positive, ASR exhibits an upward trend; if both the EAPC and its 95% CI upper limit are negative, ASR shows a downward trend; otherwise, ASR remains stable.¹⁷

Joinpoint Analysis

In this study, we employed the Joinpoint regression model to assess the trends in the incidence and DALY rates of OC over time.¹⁸ This model is commonly utilized in epidemiological studies and serves as a crucial tool for calculating annual percentage changes and their 95% CI.¹⁹ These calculations enable the analysis of trends in mortality, incidence, and DALYs over time. Conversely, the Average Annual Percentage Change (AAPC) was used to evaluate the long-term trend in annual percentage changes. We employed Joinpoint software (Version 5.4.0.0; National Cancer Institute) to detect significant joinpoints and divide the time series into distinct segments. For each segment, we fitted log-linear regression models and computed the annual percentage change to characterize trend magnitude and direction. The maximum number of joinpoints was set to 5, and model selection was based on permutation tests. If the annual percentage change or AAPC, along with its 95% CI lower limit, is greater than zero, it indicates an upward trend over the specified time frame. Conversely, a downward trend is defined when the annual percentage change or AAPC and the upper bound of its 95% CI are both below zero; otherwise, the trend is considered stable.

Risk Factor Analysis

We extracted relevant risk factor data from GBD 2023 and calculated OC burden attributable to high body-mass index (BMI) and occupational asbestos exposure via the GBD Comparative Risk Assessment (CRA) framework.²⁰ The calculation methodology has been documented in prior publications.²¹ We quantified the DALY-attributable fractions of the two risk factors in India, China, and the United States in 1990 and 2023.

Age–Period–Cohort Model

The age–period–cohort (APC) model decomposes mortality trends into age, period, and birth-cohort effects and enables detection of both overall and age-group-specific temporal changes. In this study, the APC model was employed to characterize OC mortality trends stratified by age group, calendar period, and birth cohort. Death counts were modeled by log-linear Poisson regression with log person-years as the offset, using the web-based APC tool developed by the National Cancer Institute.²² Age was grouped into 5-year categories from 55–59 to 90–94 years, with an open-ended category aged ≥95 years; period was grouped into consecutive 5-year intervals spanning the study period, and corresponding 5-year birth-cohort groups were derived. Because age, period, and cohort are exactly linearly dependent (period = age + cohort), the model is intrinsically non-identifiable; rather than imposing an arbitrary external constraint, the tool reports uniquely defined estimable functions obtained via the Moore–Penrose generalized inverse.²² We report the net drift (overall annual percentage change in mortality), local drifts (age-group-specific annual percentage changes) with Wald chi-square tests, longitudinal and cross-sectional age-specific mortality curves, and period- and cohort-specific rate ratios (RRs) relative to their weighted mean.²³ This approach delineates long-term mortality patterns and demographic differences in OC, thereby supporting the formulation of targeted prevention and treatment strategies.

Decomposition Analysis

The Das Gupta decomposition method is an epidemiological way of pinpointing factors causing temporal variations in disease burden.²⁴ It decomposes temporal fluctuations in incident cases and DALYs into three main components, namely population aging, population expansion and epidemiological change.²⁵ Population expansion means changes in total population size that exert an influence on disease burden. For example, the accelerated growth of the population is likely to increase the overall burden of disease, though incidence and mortality rates are stable. Population aging depicts the trend whereby an increasing number of old people could elevate the burden of chronic and non-infectious diseases. Epidemiological change represents the residual change in age-specific rates after accounting for population size and age structure; it should not be interpreted as the effect of a single intervention.

BAPC Model Analysis

To forecast OC burden among women aged 55 years and older in India, China, and the United States from 2024 to 2040, we implemented a Bayesian age-period-cohort (BAPC) model in R, with baseline estimates derived from the Global Burden of Disease 2023 dataset. The reported outcome was the age-standardized DALY rate (ASDR). Smoothing of age, period, and cohort effects was achieved by assigning second-order random walk (RW2) priors, while weakly informative priors were adopted for precision parameters and the intercept, in line with established recommendations. Model estimation was carried out using Integrated Nested Laplace Approximation (INLA). This framework supports computationally efficient and methodologically rigorous projections of core burden indicators, allowing for a structured comparison of projected OC burden trajectories across the three included countries.^26,27 Projected ASDRs are reported with nested posterior predictive credible intervals (50%, 60%, 70%, 80%, and 95%; shown as shaded bands in the figures), which widen from 2024 to 2040 as forecast uncertainty accumulates.

ARIMA Model Analysis

The Autoregressive Integrated Moving Average (ARIMA) model is a common tool for time series forecasting. It integrates autoregressive, differencing and moving average components to capture temporal trends and cyclical patterns. In the ARIMA (p, d, q) structure, p, d and q denote the orders of the autoregressive term, differencing and moving average term, respectively.²⁸ We first applied differencing to stabilize the time series and used the Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test for stationarity assessment. Quantile–quantile (Q–Q) plots were adopted to examine the normality of residuals. The auto.arima function was used to select the optimal model according to the minimum Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC).²⁹ The Ljung–Box test was further conducted to verify residual whiteness. A model with white noise residuals was regarded as reliable for short-term prediction.

All statistical analyses and data visualizations were performed using R (version 4.4.2) and JD_GBDR (V2.37, Jingding Medical Technology Co., Ltd.). A two-tailed p-value below 0.05 was considered to show statistical significance. Data were visualized using “ggplot2”.

Results

Change in Disease Burden

Table 1 summarizes the absolute number of cases, ASIR, ASDR, and ASMR of OC among women aged 55 years and older in India, China, and the United States from 1990 to 2023. Figure 1 illustrates the temporal trends of related disease burden indicators for the same population from 1990 to 2023. The absolute numbers of incident cases, deaths, and DALYs increased across the three countries during the study period, whereas ASIR, ASDR, and ASMR followed divergent trends. Overall, these three indicators rose significantly in India, declined markedly in the United States, and showed a fluctuating pattern with a slight decrease in China (Figure 1, and Supplementary Data 1).

Table 1.

Case Numbers and Age-Standardized Rates (ASRs) of Ovarian Cancer Incidence, DALYs, and Mortality Among Women Aged 55 years and Older in India, China, and the United States, 1990–2023

Location	Measure	1990 Number (95% UI)	1990 Age-Standardized Rates per 100,000 People (95% UI)	2023 Number (95% UI)	2023 Age-Standardized Rates per 100,000 People (95% UI)	1990–2023 EAPC (95% CI)
India	Incidence	5184 (3472, 7446)	14.03 (9.40, 20.15)	27,443 (19229, 37,007)	24.88 (17.44, 33.56)	1.62 (1.42, 1.83)
	DALYs	119,760 (80535, 172,399)	324.09 (217.94, 466.54)	577,589 (400463, 780,370)	523.74 (363.13, 707.62)	1.33 (1.12, 1.54)
	Mortality	4604 (3082, 6604)	12.46 (8.34, 17.87)	23,586 (16544, 32,042)	21.39 (15.00, 29.06)	1.54 (1.35, 1.73)
China	Incidence	9871 (7498, 13,205)	13.40 (10.18, 17.93)	23,707 (18276, 29,789)	11.10 (8.56, 13.95)	−1.10 (−1.35, −0.85)
	DALYs	212,642 (158890, 282,950)	288.73 (215.75, 384.20)	428,862 (335280, 531,853)	200.83 (157.01, 249.06)	−1.65 (−1.90, −1.40)
	Mortality	8205 (6183, 11,046)	11.14 (8.40, 15.00)	17,901 (13887, 21,849)	8.38 (6.50, 10.23)	−1.46 (−1.72, −1.20)
United States	Incidence	15,896 (14236, 17,502)	53.37 (47.80, 58.77)	19,067 (16829, 21,411)	35.46 (31.30, 39.82)	−1.46 (−1.65, −1.27)
	DALYs	258,787 (234145, 280,767)	868.93 (786.19, 942.74)	300,346 (268237, 333,025)	558.62 (498.90, 619.40)	−1.47 (−1.63, −1.32)
	Mortality	12,244 (10924, 13,409)	41.11 (36.68, 45.02)	14,800 (13033, 16,485)	27.53 (24.24, 30.66)	−1.46 (−1.65, −1.26)

Note: Rates are per 100,000 women aged 55 years and older.

Abbreviations: UI, uncertainty interval; CI, confidence interval; EAPC, estimated annual percentage change; ASIR, age-standardized incidence rate; ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; DALYs, disability-adjusted life years.

Figure 1.

National trends in age-standardized incidence, DALY, and mortality rates of ovarian cancer among women aged 55 years and older in India, China, and the United States, 1990–2023. (A) Incidence; (B) DALYs; (C) mortality.

In India, ASIR, ASDR and ASMR all increased significantly. The ASIR rose from 14.03 per 100,000 in 1990 to 24.88 per 100,000 in 2023 (EAPC = 1.62, 95% CI: 1.42–1.83). The ASDR increased from 324.09 per 100,000 to 523.74 per 100,000 (EAPC = 1.33, 95% CI: 1.12–1.54), and the ASMR rose from 12.46 per 100,000 to 21.39 per 100,000 (EAPC = 1.54, 95% CI: 1.35–1.73).

In China, absolute case counts increased, whereas ASIR, ASMR and ASDR showed an overall declining trend. The ASIR declined from 13.40 per 100,000 to 11.10 per 100,000 (EAPC = −1.10, 95% CI: −1.35 to −0.85). The ASDR decreased from 288.73 per 100,000 to 200.83 per 100,000 (EAPC = −1.65, 95% CI: −1.90 to −1.40). The ASMR fell from 11.14 per 100,000 to 8.38 per 100,000 (EAPC = −1.46, 95% CI: −1.72 to −1.20).

In the United States, despite the growth in absolute case numbers, ASIR, ASDR and ASMR all decreased significantly. The ASIR dropped from 53.37 per 100,000 to 35.46 per 100,000 (EAPC = −1.46, 95% CI: −1.65 to −1.27). The ASDR decreased from 868.93 per 100,000 to 558.62 per 100,000 (EAPC = −1.47, 95% CI: −1.63 to −1.32). The ASMR fell from 41.11 per 100,000 to 27.53 per 100,000 (EAPC = −1.46, 95% CI: −1.65 to −1.26).

Age-Specific Pattern Analysis

The 2023 age-stratified analysis of OC incidence among women aged 55 years and older revealed differences in age-related patterns and peak burden across India, China, and the United States (Figure 2A and Supplementary Data 2). Although incidence rates varied by age in all three countries, differences were observed in peak age and magnitude. The incidence rate peaked at 28.13 per 100,000 among women aged 70–74 years in India and at 12.83 per 100,000 among women aged 65–69 years in China. In the United States, the incidence rate increased steadily with age, reaching a peak of 59.34 per 100,000 among women aged 85–89 years, and then declined slightly among women aged ≥90 years. Overall, the United States had higher age-specific incidence rates in most age groups, particularly among older women.

Figure 2.

Age-specific trends in incidence and DALYs of ovarian cancer among women aged 55 years and older in India, China, and the United States in 2023. (A) Age-specific incidence rate per 100,000 population; (B) Age-specific DALY rate per 100,000 population.

The 2023 age-specific DALY rates of OC among women aged 55 years and older also varied across India, China, and the United States (Figure 2B and Supplementary Data 2). In India, the DALY rate peaked at 573.11 per 100,000 among women aged 65–69 years and was higher than in the other two countries among women aged 55–59 and 60–64 years. In China, the DALY rate peaked at 234.15 per 100,000 among women aged 65–69 years and remained relatively stable across age groups from 55–59 to 70–74 years, with a markedly lower peak burden than that in the United States. In the United States, the DALY rate increased progressively with age, peaking at 654.54 per 100,000 among women aged 75–79 years, and then gradually declined among women aged ≥80 years. Overall, OC-related DALY burden was concentrated among women aged 55–69 years in India, whereas the peak burden occurred among women aged 75–79 years in the United States.

Joinpoint Trend Analysis

Joinpoint regression analysis was performed to evaluate the temporal trends in OC ASIR and ASDR among women aged 55 years and older in India, China, and the United States from 1990 to 2023 (Figure 3 and Supplementary Data 3). For ASIR, the overall AAPC was 1.71 (95% CI: 1.60–1.81) in India, −0.51 (95% CI: −0.62 – −0.40) in China, and −1.28 (95% CI: −1.34 – −1.23) in the United States. For ASDR, the corresponding AAPCs were 1.44 (95% CI: 1.32–1.55) in India, −1.04 (95% CI: −1.15 – −0.93) in China, and −1.43 (95% CI: −1.48 – −1.38) in the United States. Overall, both ASIR and ASDR exhibited sustained increasing trends in India, slight overall decreases in China, and long-term declining trends in the United States throughout the study period. Notably, all three countries displayed distinct inflection points during the study period; the timing of the joinpoints varied across countries and metrics.

Figure 3.

Joinpoint analysis of ovarian cancer trends among women aged 55 years and older in India, China, and the United States, 1990–2023. (A) Age-standardized incidence rate; (B) age-standardized DALY rate. The asterisk (*) indicates that the annual percentage change for the corresponding joinpoint segment was statistically significant (two-tailed P < 0.05).

Attributable Burden

Figure 4 shows the proportions of attributable risk factors for OC DALYs among women aged 55 years and older in India, China, and the United States in 1990 and 2023. Across all three countries, the proportions of DALYs attributable to high body-mass index (BMI) and occupational asbestos exposure varied substantially, with particularly pronounced disparities in high BMI-related burdens.

Figure 4.

Proportions of ovarian cancer DALYs attributable to high body-mass index and occupational asbestos exposure in India, China, and the United States in 1990 and 2023.

For high BMI, the United States had the highest attributable proportion (10.0% in 1990, 14.5% in 2023), followed by China (2.0% in 1990, 5.3% in 2023) and India (0.3% in 1990, 4.0% in 2023). All three nations saw rising attributable proportions over the study period, with the largest absolute increment of 4.5% points recorded in the United States. For occupational asbestos exposure, the United States again presented the highest attributable proportion (4.3% in 1990, 4.2% in 2023), followed by China (1.3% in 1990, 1.6% in 2023) and India (0.3% in 1990, 0.4% in 2023) (Figure 4 and Supplementary Data 4).

Age-Period-Cohort Analysis

We performed an APC analysis of OC mortality among women aged 55 years and older in India (Figure 5A and Supplementary Data 5). Fitted age-specific mortality increased from 8.35 per 100,000 person-years at 55–59 years to 24.91 at 85–89 years, then declined to 23.01 at 90–94 years and 18.60 at 95 years and older. Relative to the 1927–1931 reference cohort (RR = 1.00), cohort RRs were below 1.00 in earlier cohorts and increased progressively in cohorts born after the reference, reaching 1.83 (95% CI: 1.69–1.99) in 1962–1966. Relative to the 2004–2008 reference period, period RRs were lower in 1989–1993, 1994–1998, and 1999–2003; were not significantly different from 1.00 in 2009–2013; and were higher in 2014–2018 and 2019–2023. Net drift was 1.67% per year (95% CI: 1.10%–2.24%). Local-drift estimates were positive across age groups, although the 95% CI included zero for the group aged 95 years and older.

Figure 5.

Age-period-cohort analysis of ovarian cancer mortality among women aged 55 years and older. (A) India; (B) China; (C) the United States. The horizontal dashed pink line indicates the net drift, and the horizontal dashed black line indicates a rate ratio of 1.00.

An APC analysis was performed to evaluate OC mortality among women aged 55 years and older in China (Figure 5B and Supplementary Data 5). Fitted age-specific mortality peaked at 12.21 per 100,000 person-years at 70–74 years and then generally declined, with estimates of 9.56 at 85–89 years, 9.89 at 90–94 years, and 8.43 at 95 years and older. Relative to the 2004–2008 reference period, period RRs were higher in 1989–1993, 1994–1998, and 1999–2003; were not significantly different from 1.00 in 2009–2013 and 2014–2018; and were higher again in 2019–2023. The cohort RR peaked at 1.06 (95% CI: 1.01–1.11) in the 1917–1921 cohort and declined after the 1927–1931 reference cohort to 0.55 (95% CI: 0.52–0.59) in 1962–1966. Net drift was −0.78% per year (95% CI: −1.10% to −0.45%). Local-drift point estimates were negative through 80–84 years but became slightly positive at 85 years and older, ranging from 0.03% to 0.82% per year; all corresponding 95% CIs included zero.

An APC analysis was conducted on OC mortality burden among women aged 55 years and older in the United States (Figure 5C and Supplementary Data 5). Age-specific mortality increased monotonically with age and peaked at 67.59 per 100,000 person-years (95% CI: 65.24–70.03) in the 90–94 years group before marginally dropping to 57.78 per 100,000 person-years (95% CI: 54.37–61.41) for those aged ≥95 years. Setting 2004–2008 as the reference period (RR = 1.00), the period RR declined steadily from 0.92 (95% CI: 0.90–0.95) in 2009–2013 to 0.78 (95% CI: 0.76–0.80) in 2019–2023. The cohort effect peaked in the 1917–1921 birth cohort (RR = 1.10, 95% CI: 1.07–1.13) when the 1927–1931 cohort served as the reference (RR = 1.00), followed by a monotonic decline across successive cohorts to RR = 0.57 (95% CI: 0.54–0.61) in the 1962–1966 cohort. The overall net drift was −0.94% per year (95% CI: −1.02% to −0.86%), confirming a significant long-term reduction in mortality. Local-drift estimates were significantly negative from 55–59 through 85–89 years. The estimates for women aged 90–94 years (−0.02% per year; 95% CI: −0.25% to 0.20%) and 95 years and older (0.12% per year; 95% CI: −0.34% to 0.58%) were not statistically different from zero.

Demographic Decomposition of Burden Changes

Decomposition analysis revealed that regarding changes in incident cases, population growth was the leading driver of rising case numbers across all included countries. In China, population growth contributed 114.04% to the total increase. In India, population growth accounted for 57.79% of the growth. Notably, epidemiological change contributed 35.04%, substantially higher than the figure in China. The United States showed a distinct pattern: population growth (597.89%) was the sole positive driver, while population aging (−254.32%) and epidemiological change (−243.57%) both made negative contributions (Figure 6A and Supplementary Data 6).

Figure 6.

Demographic decomposition of changes in ovarian cancer burden among women aged 55 years and older in India, China, and the United States, 1990–2023. (A) Incident cases; (B) DALYs. Black dots indicate the observed total net change between 1990 and 2023.

Decomposition analysis revealed that population growth was the primary positive driver of changes in total DALYs across the three countries. In China, population growth contributed 145.48% to the increase in total DALYs, while population aging accounted for 6.97%. Epidemiological change exerted a substantial negative effect, contributing −52.45%. In the United States, population growth was the sole major positive driver (733.72%), whereas population aging (−304.01%) and epidemiological change (−329.71%) both had pronounced negative effects. In India, total DALYs increased most markedly, with positive contributions from population growth (61.25%), population aging (6.00%), and epidemiological change (32.75%) (Figure 6B, Supplementary Data 6).

Future Trend of DALYs

According to the BAPC model, the ASDR of OC among women aged 55 years and older in India is projected to rise to 815.28 per 100,000 by 2040. In China, the ASDR is projected to decrease, reaching 168.65 per 100,000 by 2040. In the United States, the future ASDR is projected to decline continuously, falling to 405.21 per 100,000 in 2040 (Figure 7 and Supplementary Data 7). We also applied the ARIMA model to predict ASDR from 2024 to 2040 and compared its outputs with projections from the BAPC model. Both models yielded consistent trends, supporting the robustness of the projected trend directions, although the predicted magnitudes differed between models (Figure 8 and Supplementary Data 8).

Figure 7.

Age-standardized ovarian cancer DALY rates among women aged 55 years and older, 1990–2040, estimated with the Bayesian age-period-cohort model. Shaded bands show nested 50%, 60%, 70%, 80%, and 95% posterior predictive credible intervals. (A) India; (B) China; (C) the United States. The vertical dashed grey line marks the boundary between observations up to 2023 and projections.

Figure 8.

Age-standardized ovarian cancer DALY rates among women aged 55 years and older, 1990–2040, estimated with the autoregressive integrated moving average model. Shaded areas show 95% prediction intervals. (A) India; (B) China; (C) the United States. The vertical dashed grey line marks the boundary between observations up to 2023 and projections.

Discussion

Based on GBD 2023, this study compared OC burden, attributable risk factors, demographic drivers, APC patterns, and projected trends among women aged 55 years and older in India, China, and the United States. Our findings extend previous GBD 2021 analyses of OC in older women by providing a focused comparison of these three countries. Previous global analyses have reported a nonlinear association between OC burden and SDI, with burden increasing up to an SDI of approximately 0.8 and declining thereafter.⁹ However, this ecological association should not be interpreted as evidence that SDI or disease-control performance caused the observed national trends. The divergent trajectories in India, China, and the United States indicate that SDI alone does not capture heterogeneity in demographic structure, attributable risk factors, access to diagnosis, and cancer care; the country-specific explanations discussed below should therefore be regarded as hypothesis-generating rather than causal.

First, distinct long-term trends in OC burden were observed among women aged 55 years and older across India, China, and the United States. In India, the sustained rise in age-standardized mortality and DALY rates should be interpreted in the context of limitations in timely diagnosis and equitable access to cancer treatment,^30,31 together with marked regional differences in cancer burden and health-care resources.³² The present analysis cannot determine the reasons for the concurrent rise in incidence. Inequities in access to specialist care remain important barriers in resource-constrained OC settings.³³ The increase in the burden attributable to high BMI identified in this study indicates that the modeled contribution of high BMI increased over time; broader nutritional and metabolic transitions were not assessed directly. The rising cohort pattern in the APC analysis should not be interpreted as evidence of a specific causal mechanism. Individual-level studies are needed to test these explanations.

China’s declining age-standardized OC incidence may be considered in the context of possible cohort-related variation in reproductive and hormonal exposures associated with epithelial OC risk in Asian populations,^8,34 although these exposures were not measured in the present study. The declines in mortality and DALY rates may also be consistent with broader improvements in cancer care. Standardized OC management and quality-control pathways have been issued for county-level settings,³⁵ and population-based cancer survival has improved in China,³⁶ although the latter evidence is not specific to OC. Regional inequities in oncology resources and gynecological care capacity persist.^11,12 These regional disparities may limit access to timely diagnosis, specialist referral, and guideline-concordant management, particularly in rural and remote areas. Because direct OC-specific evidence remains limited, they should be interpreted as potential barriers rather than established explanations for the observed trends.

In the United States, the decline in age-standardized OC incidence and mortality is consistent with long-term national trends.³⁷ The incidence decline may be considered in the context of risk-based genetic assessment and counseling for high-risk women,³⁸ risk-reducing salpingo-oophorectomy among identified BRCA1/2 carriers,³⁹ and changes in menopausal hormone therapy use.⁴⁰ Current professional guidance in the United States recommends individualized menopausal hormone therapy (MHT) use for appropriate indications, with periodic reassessment of benefits and risks.⁴¹ Declines in mortality and DALY rates may also be consistent with access to gynecologic oncology specialists and guideline-concordant care.^42,43 These factors are plausible contextual contributors, but this ecological analysis did not test their effects directly. Routine population-based OC screening is not recommended for asymptomatic women at average risk.⁵ For women with elevated hereditary risk, prevention instead emphasizes risk assessment and, when indicated, risk-reducing management.

These findings also have country-specific implications for prevention and control. In India, the barriers described above indicate priorities for improving timely diagnosis, specialist referral, and treatment capacity, alongside gynecologic oncology workforce development⁴⁴ and wider access to genetic testing and counseling for hereditary breast and ovarian cancer (HBOC).⁴⁵ In China, county-integrated health-care and referral reforms may provide a broader health-system framework for improving access to standardized OC care.^46,47 In the United States, the US Preventive Services Task Force (USPSTF) recommends risk assessment for women with relevant personal or family histories or ancestry and, when indicated, genetic counseling and testing.⁴⁸ Published evidence indicates persistent gaps in referral for genetic counseling and completion of genetic testing,⁴⁹ while data from the United States indicate incomplete or delayed uptake of risk-reducing surgery.⁵⁰ These measures should be regarded as intervention priorities rather than as established explanations for the national trends observed in this ecological analysis.

Second, our GBD-based analysis examined the OC burden attributable to high BMI and occupational asbestos exposure. High BMI is positively associated with OC risk; a dose-response meta-analysis reported a 7% increase in risk for each 5-unit increase in BMI (RR, 1.07; 95% CI: 1.03–1.11).⁵¹ In our analysis, the proportion of OC DALYs attributable to high BMI increased in India, China, and the United States and was highest in the United States, consistent with previous GBD-based evidence.⁵² Obesity may influence carcinogenesis through chronic low-grade inflammation, insulin resistance, altered hormonal signaling, and adipokine dysregulation.⁵³ OC-specific evidence further suggests that adipokines such as resistin may promote tumor-cell growth.⁵⁴ Given the continuing global rise in obesity,⁵⁵ these findings support incorporating weight management into broader chronic-disease prevention for women aged 55 years and older, while recognizing that this ecological analysis cannot establish that BMI caused the observed national trends. Compared with high BMI, occupational asbestos exposure accounted for a smaller attributable burden in our analysis. Nevertheless, meta-analyses support an association between asbestos exposure and OC risk.^56,57 Because latency periods of several decades have been documented for other asbestos-related cancers, past occupational exposure may remain relevant to older populations,⁵⁸ although OC-specific latency data remain limited.

Third, demographic decomposition and APC analyses provided complementary evidence on the drivers of changing OC burden. In the decomposition analysis, population growth increased absolute case counts in India, China, and the United States. Favorable changes in age-specific rates partly offset the increase associated with population growth in China and the United States. Population aging contributed positively in China but negatively in the United States, whereas all three components contributed positively in India. This distinction is important because an increasing absolute burden does not necessarily indicate worsening age-specific risk.⁵⁹ The country-specific age patterns may inform the allocation of diagnostic services and specialist referral pathways, but they do not justify population-based screening of asymptomatic women at average risk.⁵ In the APC analysis, India showed rising cohort RRs and positive net drift, whereas cohort RRs generally declined in China and the United States. These estimates describe population-level rate patterns and cannot establish independent causal effects of reproductive behavior, lifestyle, health education, or health policy. Metabolic transitions and health-system constraints are plausible contextual considerations for India,^60,61 but their contributions were not tested directly. Of particular concern, despite the overall decline in China, local-drift point estimates were slightly positive among women aged 85 years and older, ranging from 0.03% to 0.82% per year. Potential explanations include delayed diagnosis, multimorbidity and frailty, and reduced eligibility for or receipt of standard combined treatment among very old women.^62–64 However, this clinical evidence was derived mainly from settings outside China, and these explanations should therefore be regarded as hypothesis-generating. Expansion of the oldest-old population may increase absolute case counts,⁵⁹ but it cannot by itself explain positive local drift, which reflects changes in age-specific rates. Chinese women aged 85 years and older should therefore be considered a priority group for symptom recognition, timely diagnostic evaluation, geriatric assessment, specialist referral, and individualized treatment rather than routine population screening.

Finally, the BAPC and ARIMA models produced concordant trend directions through 2040, although the projected magnitudes differed. This directional agreement should be regarded as a cross-model sensitivity check rather than independent validation. In India, the projected increase in the age-standardized DALY rate should be interpreted alongside documented limitations in cancer-care infrastructure, timely diagnosis, treatment access, and referral pathways.^30,65 Similar inequalities in access to OC care have been reported across low- and middle-income countries.³³ In China and the United States, projected declines in age-standardized DALY rates should be considered alongside the historical increases in absolute burden observed in this study. Declining age-standardized rates do not necessarily imply declining absolute burden, a distinction also emphasized in broader GBD cancer analyses.⁵⁹ The public-health implications should remain proportionate to the available evidence. Risk assessment should focus on women with personal or family histories suggestive of hereditary cancer, followed by genetic counseling, testing, and structured risk management when indicated.^48,66 Symptom recognition, timely diagnostic evaluation, and referral pathways should be strengthened for symptomatic women aged 55 years and older.⁶⁷ Multimodal or multi-marker screening strategies should remain research questions rather than population-health recommendations, because long-term trial evidence has not demonstrated a reduction in OC mortality and routine screening is not recommended for asymptomatic women at average risk.^5,68 These long-range projections cannot anticipate future structural changes in prevention, diagnosis, treatment, or data quality. Differences in projected magnitude and widening credible and prediction intervals therefore require cautious interpretation.^27,69

This study has some limitations. First, our research results rely on the data from the GBD 2023 study. These data have inherent uncertainties due to the quality of input data and the limitations of the model, especially in regions with missing data. Second, as an ecological analysis of aggregated data, it can identify population-level associations but cannot determine causal relationships or identify confounding factors at the individual level. Third, women aged ≥55 years were used as a proxy for the postmenopausal population; because GBD data are population-aggregated, individual menopausal status cannot be ascertained, and the ≥55 group may include a small proportion of premenopausal women, which may introduce non-differential misclassification. Our findings should therefore be interpreted as applying to women aged ≥55 years rather than to biologically defined postmenopausal women. Finally, the projection models cannot anticipate structural changes in prevention, diagnosis, treatment, or data quality; their widening uncertainty intervals should therefore be considered when interpreting long-range estimates. Despite these limitations, this analysis provides a comparative description of OC burden among women aged 55 years and older in the world’s three most populous countries.

Conclusion

From 1990 to 2023, age-standardized OC incidence, mortality, and DALY rates increased in India and generally declined in China and the United States, while absolute burden increased in all three countries. Population growth was the principal positive driver of the rising absolute burden. In China, local-drift point estimates became slightly positive among women aged 85 years and older, although the 95% confidence intervals included zero; this pattern warrants monitoring and age-appropriate diagnostic and care pathways. These population-level findings do not demonstrate causal mechanisms. Public-health responses should emphasize risk stratification and structured management of women with elevated hereditary risk, stronger diagnostic capacity and specialist referral pathways, and prospective multimodal or multi-marker screening pilots only in settings with adequate resources and formal evaluation.

Funding Statement

This research was funded by the Fujian Provincial Health Technology Project (2024TG010); the Joint Funds for the Innovation of Science and Technology, Fujian Province (2025Y9642 and 2024Y9553); the Startup Fund for Scientific Research, Fujian Medical University (2021QH1171); the National Key Clinical Specialty Construction Program of China (Gynecology); and the Fujian Provincial Natural Science Foundation of China (2022J011033).

Abbreviations

AAPC, average annual percentage change; AIC, Akaike information criterion; APC, age–period–cohort; ARIMA, autoregressive integrated moving average; ASDR, age-standardized DALY rate; ASIR, age-standardized incidence rate; ASMR, age-standardized mortality rate; ASR, age-standardized rate; BAPC, Bayesian age–period–cohort; BIC, Bayesian information criterion; BMI, body mass index; CI, confidence interval; CRA, comparative risk assessment; CrI, credible interval; DALY, disability-adjusted life year; EAPC, estimated annual percentage change; GBD, Global Burden of Disease; GHDx, Global Health Data Exchange; HBOC, hereditary breast and ovarian cancer; INLA, integrated nested Laplace approximation; KPSS, Kwiatkowski–Phillips–Schmidt–Shin test; MHT, menopausal hormone therapy; OC, ovarian cancer; PI, prediction interval; Q–Q, quantile–quantile; RR, rate ratio; RW2, second-order random walk; SDI, Socio-demographic Index; UI, uncertainty interval; USPSTF, US Preventive Services Task Force; YLD, year lived with disability; YLL, year of life lost.

Data Sharing Statement

Data are available in a public, open access repository. Data from the GBD 2023 can be accessed using the Global Health Data Exchange (GHDx) query tool (https://vizhub.healthdata.org/gbd-results/) which the Institute for Health Metrics and Evaluation maintains. The Supplementary File provides tabulated results associated with Figures 1–8.

Ethics Approval

This study used publicly available, de-identifed, summary-level data from the Global Burden of Disease Study 2023 database and did not involve direct contact with human participants or access to identifiable personal information. The study was reviewed by the Ethics Committee of the Fujian Maternity and Child Health Hospital and was determined to be exempt from ethical review. The requirement for informed consent was also waived because only publicly available, aggregated, and de-identifed data were used. This exemption is consistent with Article 32, Items (1) and (2), of the Measures for Ethical Review of Life Science and Medical Research Involving Human Subjects.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare no competing interests.