Temporal Trends in Mortality of Female Cancers Among Women and the Impact of COVID-19 in Low- and Middle-Income Countries, 1990–2023: An Age-Period-Cohort Analysis from the Global Burden of Disease Study 2023

Introduction

Female cancers among women, including breast cancer (BC), cervical cancer (CC), ovarian cancer (OC) and uterine cancer (UC), pose a major global health challenge, collectively accounting for a substantial share of cancer-related morbidity and mortality among women. In 2022 alone, BC, which is now the most frequently diagnosed cancer worldwide, resulted in about 665,684 deaths, representing nearly one-sixth of cancer deaths among women.¹ CC was responsible for an estimated 348,189 deaths in 2022, ranking as the fourth leading cause of cancer mortality among women.¹ OC, the eighth most common malignancy in women, accounted for roughly 206,839 deaths globally in 2022, while UC caused approximately 97,704 deaths in the same year.¹ This persistent burden of female cancers among women underscores their profound impact on women’s health worldwide, and it has far-reaching implications for public health and socioeconomic development.

Despite overall advances in cancer control, the burden of cancers is increasingly concentrated in low-resource regions. For female cancers among women, the inequity is especially pronounced. Approximately 90% of CC deaths occurred in low- and middle-income countries (LMICs),^2,3 and transitioning countries continue to carry a disproportionate share of BC fatalities relative to transitioned countries.⁴ These patterns are driven by multiple factors, including limited access to prevention, early detection, and treatment in poorer countries. Women in LMICs are often diagnosed at later stages due to inadequate screening and awareness. Only around 19% of eligible women undergo CC screening in developing countries on average, compared to about 63% in developed countries.⁵ Similarly, late-stage BC diagnosis is more common in LMICs.⁶ Previous studies have shown that nearly 70% of BC cases in high-income countries are diagnosed at stages I–II, whereas in LMICs, the rate of early-stage diagnoses is persistently below 50%.^6,7 These disparities underline an urgent need to focus global cancer control efforts on lower-income countries, where improvements in early detection and treatment could yield significant reductions in female cancer mortality.

Previous research has extensively documented the global and national burden of female cancers among women. For example, studies using Global Burden of Disease (GBD) 2021 data have analyzed incidence, mortality, and disability-adjusted life-years (DALYs) for breast,^8–10 cervical,¹¹ ovarian,¹² and uterine cancers¹³ at global, regional, and national levels, revealing the increasing burden of female cancers among women and their geographic disparities.^10,14 National-level analyses, such as those in China, have demonstrated an increase in deaths from breast,¹⁵ cervical,¹⁶ and ovarian cancers¹⁷ from 1990 to 2019. Similarly, global trends in adolescent and young adult female cancers highlight regional disparities, with Central Latin America exhibiting the highest age-standardised incidence rates and South sub-Saharan Africa the highest age-standardised DALY rates.¹⁸ Moreover, our previous work using GBD 2021 data quantified global burden, trends, and inequalities in total and site-specific cancers attributable to high body mass index (BMI) among older adults (aged ≥60 years), using DALYs as the primary summary measure.¹⁹ That analysis, however, focused on high BMI–attributable cancers in older adults at the global level and did not specifically examine mortality from female cancers among women or the impact of the COVID-19 pandemic in LMICs. To date, few studies have examined how mortality trends for four major female cancers among women vary across World Bank income strata within LMICs. Furthermore, existing studies have not utilized the most recent GBD Study 2023 data to apply age-period-cohort models. By separating age-related risk from period-driven influences (eg, changes in screening coverage, human papillomavirus [HPV] vaccination uptake, treatment availability, or health-system disruptions) and cohort-related influences (eg, generational shifts in reproductive patterns, obesity, and other exposures), age-period-cohort analysis offers a policy-relevant framework to clarify whether observed mortality changes are more consistent with short-term system-level factors or longer-term transitions in population risk.^20–22

To address these gaps, we conducted a second analysis using the latest data from the GBD Study 2023. Our study examined temporal trends in age-standardized mortality rates (ASMRs) from 1990 to 2023 for four major female cancers among women combined, as well as for each of the four major cancer types (breast, cervical, ovarian, and uterine) individually, across 130 countries classified as low-, lower-middle or upper-middle-income according to the World Bank’s 2023 criteria. By highlighting the unique challenges in LMICs, this study aims to support equitable resource allocation and the development of context-specific strategies to reduce the disproportionate burden of female cancers among women in LMICs.

Material and Methods

Data Sources

This study constitutes a secondary analysis of publicly accessible estimates from the GBD Study 2023, produced by the Institute for Health Metrics and Evaluation. The GBD Study 2023 systematically quantified cause-specific mortality, incidence, and prevalence for 375 diseases and 88 risk factors across 204 countries and territories from 1990 to 2023.^23–25 The analytical framework, modeling strategies, and data sources adopted by the GBD study have been comprehensively described in the GBD publications.^23–25 We extracted cause-specific mortality data for four female cancers among women (breast, cervical, ovarian, and uterine) for 130 LMICs. These countries were stratified into three income categories based on the World Bank 2023 income classification (Table S1), using Gross National Income (GNI) per capita—a composite measure capturing both domestic output (GDP) and net income from abroad. Specifically, countries were classified as low-income countries (LCs, n = 26; GNI per capita ≤ US$1,145), lower-middle-income countries (LMCs, n = 51; US$1,146 to 4,515), and upper-middle-income countries (UMCs, n = 53; US$4,516 to 14,005).²⁶ Cancer definitions in this study were based on the International Classification of Diseases, 10th Revision (Table S2).^23,27 Mortality estimates were disaggregated by 5-year age groups (15–19 to 95+ years; for UC, from 20–24 years), calendar year, and location. All data were accessed through the GBD Results Tool (https://vizhub.healthdata.org/gbd-results/).

Statistical Analysis

Joinpoint Regression Analysis

We applied joinpoint regression analysis to quantify temporal trends in ASMRs for each female cancers among women across different income groups from 1990 to 2023.²⁸ This method identifies significant time points (“joinpoints”) at which trends in mortality rates change direction or magnitude, allowing the entire time series to be segmented into distinct linear intervals. The analysis was conducted using the Joinpoint Regression Program (version 5.3.0) developed by the US National Cancer Institute.²⁹ Following National Cancer Institute default specifications for 34-year time series,³⁰ log-linear models were fitted with a maximum of six joinpoints permitted, and statistical significance of each segment was assessed using Monte Carlo permutation tests. For each linear segment, we calculated the annual percentage change (APC) and its 95% confidence interval (CI), while the overall trend across the entire period was summarized using the average annual percentage change (AAPC). A statistically significant increasing or decreasing trend was defined as an APC or AAPC with a 95% CI that did not cross zero. When the 95% CI included zero, the trend was considered statistically non-significant, indicating a stable mortality pattern. To investigate the potential impact of the COVID-19 pandemic on mortality trends, we further estimated APCs for the pre-pandemic period (1990–2019) and the pandemic period (2020–2023).

Age-Period-Cohort Analysis

To further disentangle the temporal dynamics of cancer mortality, we performed age-period-cohort analysis to decompose observed mortality trends into age, period, and cohort effects between 1994 and 2023.^31,32 Mortality data were aggregated into 5-year age intervals ranging from 15–19 to 95 years and older (20–24 to 95+ for UC) and aligned with six consecutive 6-year calendar periods from 1994–1998 to 2019–2023. The median period (2009–2013) and the central birth cohort (1959–1968) were selected as reference categories. We estimated four interpretable functions: (1) local drift, representing age-specific annual changes in mortality; (2) the longitudinal age curve, capturing age-related mortality patterns independent of period and cohort effects; (3) period rate ratios (RRs), comparing each period to the 2009–2013 reference; and (4) cohort RRs, reflecting generational variations in mortality relative to the 1959–1968 cohort. Wald chi-squared tests were used to assess the statistical significance of each parameter. Analyses were conducted using the Age-Period-Cohort Web Tool provided by the US National Cancer Institute (https://analysistools.cancer.gov/apc/).³³

All data processing, analysis, and visualization were executed using R (version 4.4.1; The R Foundation for Statistical Computing, Vienna, Austria).

Results

Overall Mortality Trends in Female Cancers Among Women Across LMICs

In 2023, LMICs reported a total of 1,053,147 deaths from female cancers among women, representing a marked increase from 410,667 deaths in 1990 (Table 1). From 1990 to 2023, the number of deaths from female cancers among women increased substantially across all three income groups (Table 1). In 1990, the highest number of deaths occurred in UMCs, with 231,409 deaths, followed by LMCs (148,400), and LCs (30,858) (Table 1 and Figure 1). By 2023, this pattern had shifted: LMCs became the largest contributor, reporting 499,504 deaths, surpassing UMCs (432,044), while LCs rose to 121,599 deaths. Despite differing mortality counts, LMs consistently exhibited the highest ASMRs across both time points (Table 1). In contrast, UMCs showed a decline in ASMR (AAPC: –0.77%), while LMCs experienced an increase (AAPC: 0.71%), and LCs reported the most rapid growth (AAPC: 1.40%). Notably, between 2020 and 2023, ASMRs rose in all income settings. The steepest recent increase was seen in LCs (APC: 7.83%), followed by LMCs (APC: 3.73%) and UMCs (APC: 1.70%) (Table 1). Across 130 LMICs, India registered the highest number of deaths, followed by China and Indonesia (Table S3). The highest ASMRs were observed in Zambia, Equatorial Guinea, and Eswatini (Table S3 and Figure S1). From 1990 to 2023, the sharpest rises in ASMRs were seen in Equatorial Guinea, Lao People’s Democratic Republic, and Egypt (Table S3 and Figure S2). Conversely, notable declines were reported in Kazakhstan, China, and Armenia (Table S3 and Figure S2). In summary, while total cancer deaths have more than doubled from 1990 to 2023 in LMICs, the mortality rates have diverged.

Table 1 The Mortality of Female Cancers Among Women in Low- and Middle-Income Countries

Figure 1 Numbers of deaths for female cancers among women in low- and middle-income countries in 1990 and 2023.

Figure 2 Joinpoint regression analysis of ASMR for four female cancers among women in low- and middle-income countries from 1990 to 2023. (A) Joinpoint regression analysis of ASMR for breast cancer among women across UMCs (Ai), LMCs (Aii), and LCs (Aiii) from 1990 to 2023; (B) Joinpoint regression analysis of ASMR for cervical cancer across UMCs (Bi), LMCs (Bii), and LCs (Biii) from 1990 to 2023; (C) Joinpoint regression analysis of ASMR for ovarian cancer across UMCs (Ci), LMCs (Cii), and LCs (Ciii) from 1990 to 2023; (D) Joinpoint regression analysis of ASMR for uterine cancer across UMCs (Di), LMCs (Dii), and LCs (Diii) from 1990 to 2023. *Indicates that the Annual Percent Change (APC) is significantly different from zero at the alpha = 0.05 level.

Site-Specific Mortality Patterns Across LMICs

Breast Cancer

From 1990 to 2023, deaths rose from 103,067 to 221,257 in UMCs and 54,547 to 250,802 in LMCs. In LCs, deaths increased from 9,649 to 49,272 (Table 1 and Figure 1). Despite a mild decline in ASMR in UMCs (AAPC: –0.38%) (Figure 2Ai), increasing ASMR trends were observed in LMCs (AAPC: 1.66%) and LCs (AAPC: 2.16%) (Table 1 and Figures 2Aii and iii). Between 2020 and 2023, ASMR for BC increased in all groups, with the sharpest post-2020 increase occurred in LCs (APC: 8.04%) (Table 1 and Figure 2Aiii). In 2023, the highest number of BC deaths occurred in India, followed by China and Indonesia (Table S4). The highest ASMRs were observed in Marshall Islands, Equatorial Guinea, and Coted’Ivoire (Table S4 and Figure S3). From 1990 to 2023, the greatest increases in ASMR were seen in Equatorial Guinea, Ethiopia, and Lao People’s Democratic Republic (Table S4 and Figure S4). In contrast, the largest declines were seen in Kazakhstan, Armenia, and Kyrgyzstan (Table S4 and Figure S4). In summary, BC remained the dominant cause of female cancer deaths in UMCs and LMCs in both 1990 and 2023 (Table 1 and Figure 1), with ASMR declining only in UMCs but increasing in LMCs and LCs.

Cervical Cancer

CC was the leading cause of female cancer deaths in LCs in 2023, accounting for nearly 47.0% of total female cancer deaths in this group (Table 1 and Figure 1). From 1990 to 2023, CC deaths increased from 78,915 to 114,485 in UMCs, 72,958 to 156,467 in LMCs, and 17,934 to 57,114 in LCs. Furthermore, ASMR declined substantially in UMCs (AAPC: –1.44%) and LMCs (AAPC: –0.67%) (Table 1 and Figures 2Bi and 2Bii). However, in LCs, the ASMR increased (AAPC: 0.74%) (Table 1, Figure 2Biii). Notably, the post-2020 period saw a shift in trajectory. All income groups recorded positive APCs, particularly LCs (APC: 6.72%). Across 130 LMICs, the highest number of CC deaths occurred in India, China, and Ethiopia in 2023 (Table S5). The top three countries with the highest ASMR were Eswatini, Malawi, and Zambia (Table S5 and Figure S5). From 1990 to 2023, South Africa, Lesotho, and Botswana showed the greatest increase in ASMR, while Mexico, Armenia, and Ukraine had the largest declines (Table S5 and Figure S6). Overall, CC deaths increased across all income groups, whereas ASMR exhibited contrasting patterns.

Ovarian Cancer

Of note, OC ASMR in UMCs exhibited a fluctuating pattern: a downward trend during 1990–2019 (APC: –0.32%), followed by an upward trend from 2020–2023 (APC: 1.95%), resulting in an overall flat long-term trend (AAPC: –0.12%) (Table 1 and Figure 2Ci). In contrast, LMCs and LCs exhibited larger increases in ASMRs, reaching 5.59 (AAPC: 1.81%) and 5.52 (AAPC: 2.32%), respectively (Table 1, Figures 2Cii and 2Ciii). From 2020 to 2023, OC mortality accelerated across all settings, particularly in LCs. At national levels, India, China, and Pakistan reported the highest absolute number of OC deaths (Table S6). The greatest ASMRs were observed in Pakistan, Ethiopia, and Equatorial Guinea (Table S6 and Figure S7). Over the 1990–2023 period, the most pronounced increases in ASMR were recorded in Ecuador, Cabo Verde, and Lao People’s Democratic Republic (Table S6 and Figure S8). Conversely, notable reductions were recorded in the Republic of Moldova, China, and Kazakhstan (Table S6 and Figure S8). Collectively, OC deaths more than doubled across all income groups, with ASMR remaining overall stable in UMCs but increasing substantially in LMCs and LCs.

Uterine Cancer

UC had the lowest mortality burden among the four female cancers studied but showed a notable upward shift in LMCs and LCs (Table 1). ASMR declined significantly in UMCs (AAPC: –1.54%) (Figure 2Di), whereas it increased in LCs (AAPC: 1.06%) and LMCs (AAPC: 0.44%) (Figures 2Dii and iii). Between 2020 and 2023, ASMR increased across all groups, with the largest APC observed in LCs at 5.32% (Figure 2Diii). In 2023, UC deaths were highest in China, India, and Brazil (Table S7). The greatest ASMRs were seen in Grenada, Marshall Islands, and Jamaica (Table S7 and Figure S9). Between 1990 and 2023, ASMR rose most sharply in Ethiopia, Jamaica, and Equatorial Guinea (Table S7 and Figure S10). In contrast, the largest declines occurred in China, Ecuador, and Kazakhstan (Table S7 and Figure S10). Overall, UC exhibited the lowest mortality burden among the four cancers. However, its ASMR increased in LCs and LMCs whilst declining markedly in UMCs.

Age-Specific Mortality Patterns Across LMICs

Across 17 age groups, the proportion of female cancer deaths peaked at 13.06% in the 55–59 age group for UMCs in 2023, 12.20% in the same age group for LMCs, and 14.94% in the 40–44 age group for LCs (Table S8 and Figure S11). Deaths of CC peaked in the 40–44 age group in LCs (16.55%), 50–54 in LMCs (12.33%), and 55–59 in UMCs (12.98%) (Table S8 and Figure S12). OC showed the highest proportions at 65–69 years (13.88%) in UMCs, 60–64 years (14.06%) in LMCs, and 50–54 years (12.02%) in LCs (Table S8 and Figure S13). Deaths of BC peaked at 55–59 years in UMCs (13.24%), 50–54 in LMCs (12.23%), and 40–44 in LCs (14.52%) (Table S8 and Figure S14). UC showed later-age peaks at 65–69 in UMCs (15.99%), 60–64 in LMCs (16.6%), and 60–64 in LCs (13.02%) (Table S8 and Figure S15). Furthermore, the age distribution of deaths across the four female cancers showed marked income-stratified contrasts in 2023 (Table S9 and Figure 3). In UMCs, OC had the largest share at ages 15–19 (49.73%) and 20–24 (35.44%). From 25–29 onward, BC became predominant (39.96% at 25–29, 50.28% at 30–34), rising to 62.45% at ≥95. CC peaked at 25–29 (38.61%) and declined thereafter to 21.44% at ≥95. UC was minimal before 30 (6.12% at 20–24) and increased modestly in older ages, peaking at 10.92% at 70–74. In LMCs, CC led at 15–19 (45.81%) and 20–24 (40.50%), with a shift to BC at 25–29 (45.78%) and 30–34 (53.20%), reaching 73.19% at ≥95. OC formed a mid-life shoulder (15–18% at 55–84, peaking at 18.10% at 65–69), while UC rose with age to 7.24% at 65–69 and tapered thereafter. In LCs, CC remained the largest component from 15–19 (65.61%) through 60–64 (41.95%); the crossover to BC occurred at 65–69 (41.54% BC vs 38.92% CC), after which BC dominated (up to 78.29% at ≥95). OC decreased from 15–19 (19.89%) to 35–39 (5.89%) and showed a secondary rise around 60–79 (12–13%). Across all strata, UC contributed ≤6% before 60, increasing gradually in older ages.

Figure 3 Proportions of deaths for four female cancers among women by age groups in low- and middle-income countries in 2023.

Age–Period–Cohort Effects on the Mortality Across LMICs

Breast Cancer

The age–period–cohort analysis demonstrated pronounced disparities in BC mortality across income groups (Figure 4). Age effects increased steadily with age in all regions, with the highest mortality risk observed in individuals aged ≥95 years (Table S10 and Figure 4A). Local drift showed declining trends across all age groups in UMCs, with the largest annual decrease in those aged 50–54 years (−1.09%) (Table S11 and Figure 4B). In contrast, LMCs and LCs experienced rising trends, with the fastest annual increases seen in individuals aged ≥95 years (2.45%) and 30–34 years (2.90%). Period effects declined in UMCs, with RRs decreasing from 1.15 in 1994–1998 to 0.97 in 2019–2023. Conversely, period risks rose substantially in LMCs and LCs. From the reference period (2009–2013), period RRs increased to 1.34 in LMCs and 1.44 in LCs by 2019–2023. (Table S12 and Figure 4C). Regarding cohort effects, relative mortality risk decreased steadily in UMCs, from 1.43 in the 1894–1903 birth cohort to 0.79 in those born in 1999–2008. In contrast, cohort effects in LMCs and LCs showed consistent and accelerating increases across successive generations. Among those born in 1999–2008, cohort RRs reached 1.85 and 1.77 in LMCs and LCs, respectively (Table S13 and Figure 4D). In summary, mortality of BC showed uniformly rising age-related risk but divergent temporal patterns by income level, with declining local drift, period, and cohort risks in UMCs versus increasing local drift and rising period/cohort relative risks in LMCs and LCs.

Figure 4 Age-period-cohort model analysis on mortality for breast cancer in low- and middle-income countries. (A) Local drift of mortality from 1990 to 2023. The dots and shaded areas indicate the annual percentage change of mortality rate (% per year) and the corresponding 95% CIs. (B) Age effects are shown by the fitted longitudinal age curves of mortality rate (per100000person-years) adjusted for period deviations. (C) Period effects are shown through the relative risk of mortality during different periods, calculated as the ratio of the age-specific rates from the period from 1994–1998 to 2019–2023, with the baseline period set as 2009–2013. (D) Birth cohort effects are demonstrated by the cohort relative risk of mortality and calculated as the ratio of age-specific rates from 1894–1903 cohort to 1999–2008 cohort, with the reference cohort set at 1959–1968. The dots and shaded areas denote the mortality rates or rate ratios and their corresponding 95% CIs.

Cervical Cancer

Age–period–cohort effects on CC mortality are presented in Figure 5 and Tables S10–S13. Age effects also showed progressive increases in mortality risk with advancing age in all regions (Table S10 and Figure 5A). Local drift values indicated consistent declines across all age groups in UMCs, especially among those aged 65–69 years (−1.97%) (Table S11 and Figure 5B). In LMCs, annual declines were less pronounced but evident across most age groups. In contrast, LCs demonstrated increasing trends in younger and middle-aged groups, with the highest rise at ages 55–59 (0.87%). Period effects in UMCs declined steadily, with RRs decreasing from 1.30 in 1994–1998 to 0.89 in 2019–2023 (Table S12 and Figure 5C). In contrast, period RRs in LMCs fluctuated slightly and showed a minor uptick in the most recent period (1.03 in 2019–2023). Notably, LCs exhibited a reverse trend, with a substantial increase in mortality risk in 2019–2023 (1.19), after a relatively stable risk in previous periods. Among women born before 1920 in UMCs, the relative mortality risk exceeded 2.0 and declined steadily across successive cohorts, dropping to 0.63 for the 1999–2008 birth cohort (Table S13 and Figure 5D). A similar declining pattern was observed in LMCs, though at a slower rate. In contrast, LCs showed rising cohort effects across successive generations, with RR increasing from 0.77 in the 1894–1903 cohort to 1.19 in the 1979–1988 cohort, before plateauing. Overall, mortality of CC revealed increasing age-related risks across all settings. UMCs demonstrated consistent declines in local drift, period, and cohort effects—contrasted by modest reductions in LMCs. Conversely, LCs exhibited increasing local drift among young and middle-aged groups coupled with growing recent period risks.

Figure 5 Age-period-cohort model analysis on mortality for cervical cancer in low- and middle-income countries. (A) Local drift of mortality from 1990 to 2023. The dots and shaded areas indicate the annual percentage change of mortality rate (% per year) and the corresponding 95% CIs. (B) Age effects are shown by the fitted longitudinal age curves of mortality rate (per100000person-years) adjusted for period deviations. (C) Period effects are shown through the relative risk of mortality during different periods, calculated as the ratio of the age-specific rates from the period from 1994–1998 to 2019–2023, with the baseline period set as 2009–2013. (D) Birth cohort effects are demonstrated by the cohort relative risk of mortality and calculated as the ratio of age-specific rates from 1894–1903 cohort to 1999–2008 cohort, with the reference cohort set at 1959–1968. The dots and shaded areas denote the mortality rates or rate ratios and their corresponding 95% CIs.

Ovarian Cancer

Figure 6 and Tables S10–S13 show the age–period–cohort effects in OC mortality across LMICs from 1994 to 2023. Age effects demonstrated a consistent and exponential increase in OC mortality with advancing age across all three World Bank income groups (Table S10 and Figure 6A). Notably, LMICs had the highest age-specific relative risks across most age groups. Local drift analysis indicated consistent annual increases in age-specific mortality rates across nearly all age groups among LMCs and LCs (Table S11 and Figure 6B). Conversely, UMCs experienced negative local drifts for most age groups, particularly between ages 25–59. After controlling for age and cohort effects, a marked increase in mortality risk was observed in the most recent period (2019–2023) in both LCs (1.54) and LMCs (1.34) (Table S12 and Figure 6C). In contrast, UMCs exhibited a distinct biphasic pattern: the period effect initially declined from 1994–1998 (1.13) to 2014–2018 (1.00), followed by a slight uptick in 2019–2023 (1.03). In LCs, a progressive increase in risk was evident from cohorts effects, peaking in the 1984–1993 cohort (1.86) (Table S13 and Figure 6D). Similarly, LMCs exhibited rising risks, reaching a maximum in the 1989–1998 cohort (1.70). In contrast, UMCs experienced a declining cohort trend, with relative risks falling steadily from earlier cohorts (eg, 1904–1913: 1.24) to a nadir around the 1984–1993 cohort (0.83). Collectively, mortality of OC showed exponentially increasing age-related risk across all income groups, with rising age-specific rates, elevated recent-period risk (2019–2023), and increasing cohort risks in LMCs and LCs, contrasted by generally declining local drift and cohort effects in UMCs.

Figure 6 Age-period-cohort model analysis on mortality for ovarian cancer in low- and middle-income countries. (A) Local drift of mortality from 1990 to 2023. The dots and shaded areas indicate the annual percentage change of mortality rate (% per year) and the corresponding 95% CIs. (B) Age effects are shown by the fitted longitudinal age curves of mortality rate (per100000person-years) adjusted for period deviations. (C) Period effects are shown through the relative risk of mortality during different periods, calculated as the ratio of the age-specific rates from the period from 1994–1998 to 2019–2023, with the baseline period set as 2009–2013. (D) Birth cohort effects are demonstrated by the cohort relative risk of mortality and calculated as the ratio of age-specific rates from 1894–1903 cohort to 1999–2008 cohort, with the reference cohort set at 1959–1968. The dots and shaded areas denote the mortality rates or rate ratios and their corresponding 95% CIs.

Uterine Cancer

For UC, age effects also demonstrated a monotonic increase in mortality risk with advancing age across all settings (Table S10 and Figure 7A). Local drift values indicated a consistent annual decline in mortality in UMCs, most prominent among younger adults aged 25–34 years (Table S11 and Figure 7B). Conversely, LMCs and LCs experienced rising mortality risks across most age groups, especially in LCs where increases exceeded 1.7% annually for those aged 30–44 years. In UMCs, the adjusted period RR declined progressively from 1.47 in 1994–1998 to 0.86 in 2019–2023 (Table S12 and Figure 7C). In contrast, period risks increased in LMCs and LCs, particularly in the latter where the RR rose to 1.22 in the most recent period. Cohort effects declined consistently in UMCs, with risk falling from 2.94 in the earliest cohort (1894–1903) to 0.32 in the most recent (1994–2003) (Table S13 and Figure 7D). In contrast, cohort effects in LMCs and LCs showed long-term upward trends, with peak risks observed among those born in 1984–1993 (LMCs: 1.29; LCs: 1.54), followed by modest declines in the most recent cohorts. In summary, mortality of UC showed steadily increasing age-related risk across all settings. UMCs exhibited declining local drift, period, and cohort effects, whereas LMC and LCs showed rising age-specific mortality, increasing period risk in recent years, and upward cohort risks.

Figure 7 Age-period-cohort model analysis on mortality for uterine cancer in low- and middle-income countries. (A) Local drift of mortality from 1990 to 2023. The dots and shaded areas indicate the annual percentage change of mortality rate (% per year) and the corresponding 95% CIs. (B) Age effects are shown by the fitted longitudinal age curves of mortality rate (per100000person-years) adjusted for period deviations. (C) Period effects are shown through the relative risk of mortality during different periods, calculated as the ratio of the age-specific rates from the period from 1994–1998 to 2019–2023, with the baseline period set as 2009–2013. (D) Birth cohort effects are demonstrated by the cohort relative risk of mortality and calculated as the ratio of age-specific rates from 1894–1903 cohort to 1994–2003 cohort, with the reference cohort set at 1959–1968. The dots and shaded areas denote the mortality rates or rate ratios and their corresponding 95% CIs.

Discussion

Leveraging the most recent GBD 2023 data, this study provides the first comprehensive analysis of mortality for female cancers among women in 130 LMICs from 1990 to 2023 using age-period-cohort methodology. We found that total deaths from BC, CC, OC, and UC in LMICs more than doubled over the 34-year study period. Notably, LMCs surpassed UMCs in absolute number of female cancer deaths by 2023. Additionally, UMCs showed a downward trend in ASMR, whereas LMCs and LCs exhibited ongoing increases. Since 2020, all income groups have experienced a rise in ASMRs, with the steepest recent increase observed in LCs. Site‑specific analyses indicate that BC remained the dominant cause of female cancer deaths among women in UMCs and LMCs, whereas CC continued to predominate in LCs. Furthermore, our age-period-cohort analysis revealed that period RRs for BC, OC, and UC have been decreasing over time in UMCs, but conversely increasing in LMCs and LCs. For CC, U-shaped period effects were observed in LMCs and LCs, with rising trends noted in recent periods. Meanwhile, cohort effects highlighted a significant generational disparity: in UMCs, each successive birth cohort experiences a consistent decline in mortality risk for all four cancers. In contrast, in LMCs and LCs, women born in more recent decades face higher risks compared to their predecessors.

Female cancers among women remain a substantial public‑health challenge, with LMICs bearing a disproportionate burden. Data from the GBD 2023 indicate that approximately 1.46 million deaths globally were attributed to four female cancers.^27,34 Of these deaths, 1.05 million (72%) occurred in LMICs. The 2.5‑fold increase in deaths from 1990 to 2023 likely reflects population growth and demographic change in LMICs. Meantime, it may also partly reflect temporal improvements in diagnosis, death certification, and cancer registration that have increased case ascertainment compared with the early 1990s. Importantly, the divergent trends in ASMRs suggest that changes in burden are not solely driven by demographic shifts or improved ascertainment, underscoring the influence of socioeconomic development and health-system capacity. India, classified as an LMC, contributed the highest number of deaths among LMICs in 2023, which largely explains why the LMCs became the largest contributor. When adjusting for population size and age structure, stark differences emerge. UMCs saw declining ASMRs, likely thanks to improved access to interventions such as mammographic screening, Papanicolaou (Pap) testing and HPV vaccination, and timely oncology care.^35–37 In LMCs and LCs, however, these lifesaving interventions have not been implemented at scale or with sufficient quality, resulting in limited ASMR reductions. China, categorized as an UMC, exemplifies how socio-economic development and health infrastructure investment could translate into better cancer outcomes. Beginning with its first National Cancer Prevention and Control Plan (1986–2000) released in 1985, China progressively strengthened cancer control through the establishment of registries, public education, and expanded screening programs.³⁸ This was accompanied by health insurance reforms that reduced financial barriers to care. For instance, the launch of the New Rural Cooperative Medical Scheme in 2003 and subsequent insurance expansions significantly increased reimbursement for catastrophic illnesses, improving access to cancer treatment in rural areas.³⁹ Partially due to these factors, China has seen substantial declines in ASMR, thereby alleviating the overall burden within UMCs. Conversely, several sub‑Saharan African countries, such as Zambia and Malawi, record the highest ASMRs. These patterns likely reflect resource constraints, low HPV vaccination uptake, and a high prevalence of risk factors and comorbid conditions that can worsen prognosis and increase case fatality. In particular, HIV infection is both a major etiologic co-factor for CC and a clinically important comorbidity that may accelerate disease progression and complicate treatment through immunosuppression, higher susceptibility to infections, and reduced tolerance to chemoradiotherapy. The World Health Organization (WHO) reports that women living with HIV have a substantially higher risk of CC and that 5% of CC cases are attributable to HIV,^38,40 highlighting how comorbid epidemics amplify the burden in LCs. Beyond HIV, the burden of coexisting conditions common in resource-limited contexts, such as tuberculosis, chronic infections, anemia, and undernutrition, may contribute to delays in diagnosis, limitations in treatment eligibility or completion, and poorer survival, thereby increasing mortality even when incidence patterns are comparable. Together, these country-specific patterns illustrate a broader phenomenon: socio-economic development and targeted health investments are associated with improved cancer outcomes,^41,42 whereas resource-limited settings struggle to make similar gains.⁴³ National income level remains a critical determinant of progress in cancer control, highlighting the urgent need for international support and capacity-building in the lowest-income regions.⁴³

A notable finding of our study is the inflection in ASMR trends after 2020. All income groups showed an increase in ASMRs during 2020–2023, with the surge most pronounced in LCs. It is plausible that the COVID-19 pandemic played a major role in this short-term reversal of progress. Many LMICs experienced significant healthcare disruptions during the pandemic, including the suspension of routine cancer screening programs, delays in diagnostic workups, interruptions in treatment services, and diversion of hospital resources and personnel toward COVID-19 care.^44–46 Even UMCs, which previously had steady downward ASMR trajectories, saw a loss of momentum or setbacks in 2020–2023, suggesting a broad and detrimental impact of the pandemic on cancer control efforts. Emerging studies have attempted to quantify the pandemic’s toll on cancer outcomes. For example, modeling analyses have predicted that a 6-month halt in breast and cervical cancer screening without adequate catch-up would lead to measurable increases in cancer deaths, attributable to missed early detection opportunities.⁴⁷ Beyond the direct risk posed by SARS-CoV-2 infection to cancer patients, the indirect effects of pandemic-related health system disruptions are expected to have a far greater influence on population level cancer mortality.^48–50 Real-world data from several countries indeed showed sharp drops in new cancer diagnoses in 2020 as screenings and clinic visits plummeted, a pattern that experts warn will translate into more patients being diagnosed at advanced stages and hence higher mortality in subsequent years.^46,47,51 Our observation of rising ASMR in 2020–2023 across all LMIC income groups is consistent with these concerns. These findings call for urgent action to mitigate the pandemic’s impact on cancer outcomes.⁵² Health systems in LMICs should implement recovery strategies such as intensive catch-up screening campaigns, proactive patient outreach to re-engage individuals who missed routine check-ups or dropped out of care during the pandemic, and the use of telemedicine or community health programs to maintain continuity of oncology care even during public health crises. Ensuring that cancer control programs are resilient to system-wide shocks will be key. For instance, developing contingency plans to sustain essential services like chemotherapy, radiotherapy, and surgery during future emergencies. Notably, the observed mortality surge across all income groups after 2020 likely reflects short-term data fluctuations attributable to COVID-19 disruptions rather than actual cancer risk increases. Continued monitoring of the pandemic’s long-term effects on cancer outcomes, including shifts in diagnostic stage and overall survival, is essential in future years.

Beyond these overarching trends, our results highlight distinct patterns for each cancer type and their potential causes. BC remained the dominant cause of female cancer mortality in both UMCs and LMCs throughout 1990–2023. This aligns with global cancer statistics indicating that BC is now the leading cause of cancer death among women worldwide, and the foremost cause of female cancer mortality in at least 112 countries.¹ The divergent trajectories of BC and CC across income levels likely reflect the unequal distribution of medical advances. Widespread mammography screening programs, and the availability of multi-modal treatments, including improved surgery, radiotherapy, hormonal therapy, and HER2-targeted agents, have all contributed to substantial reductions in BC mortality in UMCs.⁵³ This is consistent with historical data from high-income countries, which have documented around a 40% decline in BC mortality since the 1980s due to the synergy of early detection and more effective adjuvant therapies, corresponding to annual mortality reductions of approximately 2–4% in many populations.⁵⁴ Conversely, the rising BC ASMR in LMCs and LCs likely result from weaker health infrastructure and gaps in care: limited screening coverage leading to late-stage presentation, shortages of pathology and imaging, and inequitable access to standard therapies, all of which constrain survival gains. CC patterns further illustrate both progress and persistent inequities. UMCs in our analysis showed steep declines in CC ASMR, and even many LMCs had modest downward trends prior to 2020, reflecting the benefits of widespread Pap smear screening, introduction of HPV DNA testing, early treatment of pre-cancers, and, more recently, HPV vaccination of adolescents. By contrast, CC remains the leading female cancer killer in LCs, with increasing ASMRs. Notably, our age-stratified findings suggest that the excess CC mortality in LCs is concentrated in younger and middle-aged women, a pattern that is plausibly driven by a combination of earlier and sustained exposure to high-risk HPV, high prevalence of HIV co-infection, and reproductive factors such as early childbirth. These factors may increase the risk of persistent HPV infection and accelerate progression from infection to invasive disease. Even when CC is diagnosed, access to timely treatment is a major barrier. Radiation therapy is an integral component of curative treatment for locally advanced CC and an important palliative tool for advanced gynecologic malignancies, yet more than half of patients who need radiotherapy in LMICs do not have access to it.⁵⁵ This dire reality lends urgency to the WHO’s Cervical Cancer Elimination Initiative.⁵⁶ We also observe disproportionately higher OC deaths in UMCs versus LCs among younger women. This pattern likely arises from a combination of improved diagnostic ascertainment in UMCs and underestimation of OC-related deaths in LCs. Furthermore, epidemiological transitions in UMCs may modify age-specific risk through changes in reproductive patterns—such as delayed parity and fewer births—and increased prevalence of metabolic risk factors, including smoking and high-fat dietary intake, thereby increasing susceptibility to OC at younger ages.

Our age-period-cohort analysis provides additional insight into the drivers behind these observed trends. As expected, mortality risk increased with advancing age for all four cancers, but the age at which mortality peaks differed by country income level. Cervical and breast cancer deaths in LCs peaked in women in their 40s, whereas in UMCs the peak contribution shifted to later ages (often in the 50s or 60s). In other words, women in poorer countries are dying at younger ages from these cancers compared to those in more developed settings. Several factors likely contribute to these age-pattern differences. LCs tend to have younger population age structures and a high prevalence of risk factors, including HIV co-infection, drive earlier CC onset and contribute to earlier mortality peaks.⁵⁷ In UMCs, longer life expectancy and better management of other health conditions mean that more women survive into older age. Furthermore, effective screening and early intervention could delay the occurrence of cancer deaths to older ages. UMCs exhibited steadily decreasing period RRs from 1990 to 2023, indicating a declining risk of cancer death across all age groups in a given calendar period. This trend likely reflects broad improvements, including strengthened health systems, increased access to cancer screening and treatment, and effective public health measures. Cohort effects in UMCs were also generally downward, with each successive generation experiencing a lower relative risk of cancer mortality. This trend likely results from long-term benefits of reduced risk factor exposure (eg, lower smoking rates, higher HPV vaccine uptake) and improvements in healthcare throughout the life course of more recent cohorts. By contrast, LMCs and LCs experienced concerning increases in period RRs, especially in the most recent periods. This suggests that any improvements from medical advances are being outpaced by other forces, such as increasing incidence or inadequate health system responses. Moreover, we observed that women born in more recent decades in LMCs/LCs have higher relative mortality risks for these cancers than those born a generation earlier. This alarming trend implies that younger women in these countries are facing greater inherent cancer risks than their mothers or grandmothers did at the same age. Possible explanations include epidemiologic transition and lifestyle changes. For instance, recent birth cohorts may have higher prevalence of risk factors like obesity, physical inactivity, later age at first childbirth, and fewer pregnancies, all of which are known to increase breast and ovarian cancer risk.^58–60 Similarly, rising adoption of “Westernized” diets and sedentary habits in LMCs could elevate cancer risks in younger generations. At the same time, few women in the youngest cohorts of many LCs have benefited from systematic prevention measures, as HPV vaccination and organized screening programs were either nonexistent or only introduced in the past decade or two. Hence, absent significant interventions, there is a danger that each new generation in low-resource settings might face a higher cancer burden than the one before. This worrisome scenario appears to be supported by our cohort findings.

The rising cancer mortality in LMCs and LCs carries significant implications for global health targets and underscores the need for intensified action. The United Nations Sustainable Development Goal (SDG) 3.4 calls for a one-third reduction in premature mortality from non-communicable diseases (including cancers) by 2030.⁶¹ Achieving more equitable reductions in female cancer deaths is not only a matter of meeting statistical targets but also one of global health equity and social justice. As the BC burden grows in many transitioning economies, there is an urgent call to implement the strategies outlined by the WHO’s Global Breast Cancer Initiative and related programs. These include raising public awareness about early signs and symptoms, improving access to diagnostic imaging and pathology, and ensuring that women diagnosed with BC could complete multidisciplinary treatment without experiencing financial catastrophe.⁶² Primary prevention still has a role to play. Current strategies to lower a woman’s risk of developing BC include health education to encourage avoidance of tobacco use and unnecessary hormone therapies, engaging in regular physical activity, breastfeeding when possible, consuming a balanced diet, and limiting alcohol intake.^60,63 Likewise, for CC, extensive educational and outreach efforts are critical to improve HPV vaccination and screening uptake. Experience has shown that multifaceted, culturally sensitive approaches could raise coverage: mass media campaigns that provide clear “calls to action,” community engagement and group education sessions, personal invitation letters for screening, and automated reminders have all been used to increase screening participation in some LMICs.⁶⁴ Governments and international partners should invest in scaling up such interventions, tailoring them to local context. In addition, substantial investment in healthcare infrastructure is imperative. Establishing regional cancer centers, training, and retaining oncology specialists are crucial steps to reduce the survival gap between high- and low-income regions. At the same time, improving treatment capacity, from surgery to radiation therapy, would ensure that once cancers are detected, patients have a fighting chance at cure. Furthermore, governments should incorporate cancer control into their universal health coverage plans, ensuring that preventive services and treatments are included in basic benefit packages so that cost is not a barrier to care.

This study has several limitations. First, it relies on GBD 2023 estimates derived from national vital registration, cancer registries and verbal autopsy data. Many LMICs lack complete and high-quality cancer mortality surveillance. Deaths may be misclassified or underreported due to weak vital registration systems and limited diagnostic capabilities. Although the GBD methodology employs sophisticated algorithms to adjust for biases and estimate causes of death where data are sparse, there remains uncertainty in the absolute death counts and ASMRs for countries with poor data. Second, trends observed during 2020–2023 should be interpreted cautiously because this is a relatively short post-pandemic window. The impact of the COVID-19 pandemic on cancer mortality warrants continued monitoring with future data updates. Third, because we used aggregated population-level mortality data, we could not account for individual-level factors such as tumor stage at diagnosis, molecular subtype, treatment modalities received, or behavioral risk factors. Consequently, the ability to identify specific determinants of mortality changes or to attribute variations to particular interventions is limited. Fourth, analyses were conducted at the national level and did not explore sub‑national or urban–rural disparities, which could be substantial in large countries such as India, China, and Brazil.

Conclusion

In summary, this study provides an up-to-date and comprehensive assessment of female cancer mortality in LMICs. Despite progress in UMCs, ASMRs continue to rise in LMCs and LCs from 1990 and 2023, widening health inequities and threatening achievement of the SDG. These disparities should not be attributed solely to income classification. Rather, income categories likely serve as broad proxies for multiple interacting determinants, including health-system capacity and accessibility, prevention and early-detection programs implementation, treatment availability and quality, and improvements in diagnostic and reporting practices. The COVID‑19 pandemic appears to have reversed recent gains, underscoring the vulnerability of cancer‑control programs in LMICs. Site‑specific and age-period-cohort analyses reveal distinct trajectories and drivers, demonstrating the necessity of tailored, context‑sensitive interventions. Achieving equitable reductions in burden of female cancer among women necessitates the rapid expansion of HPV vaccination, mammography, and other screening programs, coupled with investments in surgical oncology, radiotherapy, and systemic therapies. Robust health policies, integrated cancer control within universal health coverage, and community‑engaged approaches are critical to ensuring that every woman, regardless of where she lives, benefits from advances in cancer prevention and treatment.