Global assessment of the quality of care index for Burkitt Lymphoma from 1990 to 2021

doi:10.21203/rs.3.rs-7944023/v1

Background

Advances in medical technology have significantly improved the prognosis for Burkitt lymphoma (BL), but the quality of care remains a concerning issue. This study utilizes a modified Quality Care Index (QCI) to assess the global status of Burkitt lymphoma care.

Methods

Based on GBD 2021 data, we analyzed the burden of BL and its changing trends. Integrate the four secondary indicators through principal component analysis to construct QCI. Utilize the machine learning interpretability tool SHAP (SHapley Additive exPlanations) to deeply analyze the key factors affecting QCI. Employ the Bayesian age-period-cohort model to predict the QCI trends from 2022 to 2035.

Results

From 1990 to 2021, the age-standardized incidence rate (ASIR) of BL showed a significant upward trend (EAPC = 2.179), while the age-standardized mortality rate (ASDR) increased slowly (EAPC = 0.623). High Socio-demographic Index (SDI) regions had the highest incidence but lower mortality, whereas low SDI regions showed the opposite pattern. QCI was highly correlated with SDI, with a median QCI of 78.50% in high SDI regions and only 21.60% in low SDI regions. SHAP analysis indicates that age is the most important factor affecting QCI, followed by year and gender. Gender differences have reversed in recent years, with the quality of care for female patients gradually surpassing that for male patients. Predictions show that by 2035, the global QCI will stabilize at over 90%, and the advantage in the quality of care for female patients will be further consolidated.

Conclusions

The global burden of BL is increasing, with significant disparities in care quality, closely associated with age, time, gender, and geographic regions. To comprehensively improve the quality of global BL care, targeted interventions must be strengthened for regions with low SDI.

Burkitt lymphoma

Quality of care

Global burden

Epidemiology

Burkitt Lymphoma (BL) is a highly aggressive malignant tumor originating from germinal center B cells, with its core molecular feature being MYC gene translocation leading to uncontrolled cell proliferation, short cell doubling time, and rapid disease progression(1). Based on clinical and epidemiological characteristics, the disease exhibits a unique bimodal age distribution and high geographic heterogeneity, which can be divided into three subtypes: endemic, sporadic, and immunodeficiency-type (2, 3)。Endemic BL primarily occurs in sub-Saharan Africa and is closely associated with Epstein-Barr virus and P. falciparum infections(4); while sporadic and HIV-related immunodeficiency-associated BL(5) occur worldwide. BL is sensitive to high-intensity chemotherapy, and the use of high-intensity, short-cycle combined chemotherapy regimens (such as R-CODOX-M/R-IVAC) can achieve cure rates exceeding 80% in children and over half in adult patients, but this treatment regimen is accompanied by significant toxicity, requiring strong supportive treatment and nursing interventions(6–8).

Nursing quality is defined as providing timely, appropriate, and effective medical services to patients to achieve ideal health outcomes. The disparity in nursing quality is a significant barrier in modern healthcare systems, and it is equally crucial in the management of hematologic malignant diseases (9, 10). The incidence and outcomes of BL exhibit significant geographic differences globally(11), suggesting substantial inequalities in disease burden and the potential nursing quality that can be accessed. However, there currently lacks a comprehensive indicator to globally measure and compare the nursing quality of BL. The Quality of Care Index (QCI) is an emerging comprehensive evaluation tool developed by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington, USA, based on GBD research data. By integrating epidemiological indicators and employing methods such as principal component analysis (PCA), it constructs a standardized score to quantify differences in nursing quality across various regions and disease areas(12) This index has been successfully applied in research on various malignant tumors, including leukemia(13) and multiple myeloma(14). Although it can effectively reveal nursing disparities between different regions and populations, traditional quality of care indices was criticized for their insufficient accuracy and the risk of oversimplification(14). The QCI constructed in this study demonstrates good validity, reflecting both the mortality dimension of the disease and the chronicity dimension, while avoiding the limitations of a single indicator(13, 15). Compared to previous studies that only used the first principal component, it offers greater comprehensiveness and interpretability(16), providing a multidimensional and more robust assessment of nursing quality for Burkitt lymphoma. It aims to provide a deeper understanding of the importance and direction of various influencing factors, offering a scientific basis for the optimal allocation of global health resources and the development of prevention and control strategies for BL.

2.1 Data Sources and Collection

This study utilized data from the Global Burden of Disease Study (GBD) 2021 to systematically extract global, regional, and national burden of disease data related to Burkitt lymphoma from 1990 to 2021, stratified by Socio-demographic Index (SDI). The GBD database integrates multiple sources worldwide, including vital registration systems, cancer registries, hospital records, and scientific literature(17). It employs standardized modeling approaches and Bayesian meta-regression tools (DisMod-MR 2.1) to generate consistent estimates of incidence, mortality, disability-adjusted life years (DALYs), and other key metrics, along with 95% uncertainty intervals (UIs), significantly enhancing international comparability and robustness of the data(18). The primary indicators used in this study encompassed incidence, mortality, prevalence, years of life lost (YLLs), years lived with disability (YLDs), and DALYs for Burkitt lymphoma. All data were obtained from the Global Health Data Exchange (GHDx, http://ghdx.healthdata.org/gbd-results-tool) and extracted by age, sex, country, region, and SDI category(17). The SDI—a composite measure of income per capita, average years of education, and total fertility rate—was used to classify countries and territories into five developmental levels: high, high-middle, middle, low-middle, and low, enabling comparative analysis across different development strata(19, 20). Furthermore, to maintain consistency with international classification standards, cases were identified and extracted using the corresponding Burkitt lymphoma codes from the International Classification of Diseases, Tenth Revision (ICD-10). All data processing steps adhered to the standardized protocols of the GBD study, including imputation of missing data, age standardization (using the WHO world standard population structure), and uncertainty evaluation, ensuring data reliability and comparability across time periods and geographic regions.

2.2 Time Trend Analysis

To accurately understand the complex temporal patterns of Burkitt lymphoma's disease burden, this study employed two complementary statistical methods: the Estimated Annual Percentage Change (EAPC) and Joinpoint regression analysis(21). The EAPC was derived by fitting a linear regression model to the natural logarithm of the Age-Standardized Rate (ASR) over time (calendar year), calculated using the formula: EAPC = [exp(β) − 1] × 100%(13, 22). The EAPC effectively quantifies the average annual magnitude of change in the rate over the entire period. The trend is interpreted as follows: an EAPC > 0 with a 95% CI not including 0 indicates a significant upward trend; an EAPC < 0 with a 95% CI not including 0 indicates a significant downward trend; if the 95% CI includes 0, the trend is considered statistically stable. Secondly, Joinpoint regression analysis was used to identify and characterize potential turning points and segmented trends within the study period. This method fits a piecewise linear model, automatically detects points in time where the trend in the age-standardized rate changes significantly (i.e., "joinpoints"), and partitions the entire time series into multiple segments of monotonic trend. For each segment, the model calculates the Annual Percent Change (APC) and its corresponding 95% Confidence Interval (CI). The interpretation of the APC follows principles similar to those of the EAPC.

2.3 Quality of care index

Incidence, prevalence, mortality, years of life lost (YLLs), years lived with disability (YLDs) and disability-adjusted life years (DALYs), are six primary indicators in quantifying the epidemiologic status. We derived four secondary indicators from the six indices to assess the quality of care index (QCI), including the prevalence to incidence ratio (PIR), the years of life lost to years lived with disability ratio (YLR), the mortality to incidence ratio (MIR) and the disability-adjusted life years to prevalence ratio (DPR). The YLR assesses the balance between life years lost due to early death and decreased quality of life due to disability, a higher YLR implying that the disease is fatal and shorter survival, while a lower YLR indicates lower quality of life and mortality rate. Similarly, the MIR examines the fatality rate of disease, which is the most direct and critical inverse indicator for evaluating the quality of care, a high rate means that almost all patients with the disease will die, and a low rate reflects the high level of medical care and an effective care system. The DPR measures the overall impact of disease on health relative to its prevalence, with higher DPR suggesting lower prevalence but greater health hazard, and low rate indicates that, despite disease is common but relatively small health loss. Lastly, the PIR evaluates the average survival time of patients after diagnosis, a higher PIR means a large disease population but a relatively small number of new cases per year, and lower PIR usually indicates rapid disease progression and short survival. These four indicators complement each other and together form a multi-dimensional evaluation framework. YLR, DPR, and MIR together reflect the lethality dimension. PIR, on the other hand, provides a separate dimension of chronic degree.

Then, to create an interpretable index called QCI, these four diverse secondary indices were combined by the principal component analysis (PCA), represents the overall quality of care. PCA is a powerful statistical technique used for dimensionality reduction to simplify complex datasets while preserving trends and patterns(15). In most previous studies, the first component derived from PCA, which had the highest score, was considered as QCI(14, 23, 24). However, in our study, YLR, DPR, and MIR contribute similarly to the first principal component (PC1), although PC1 explained most of the variance, the second principal component (PC2), mainly driven by PIR, captured a significant portion of the remaining variation not explained by Dim1, providing new information independent of "lethality." A comprehensive QCI should reflect both the ability of the healthcare system to reduce mortality (represented inversely by PC1) and its ability to manage patients and extend high-quality survival periods (represented positively by PC2). Therefore, to construct a comprehensive, robust, and scientific QCI, we weigh PC1 and PC2 based on the proportion of variance they explain, ensuring that higher-priority dimensions have greater weight in the composite index while no unique important information is overlooked(15). The relevant formula is as follows:$\:\text{YLR}\text{=}\frac{\text{YLLs}}{\text{YLDs}}$

$$\:\text{DPR}\text{=}\frac{\text{DALYs}}{\text{Prevalence}}$$

$$\:\text{MLR}\text{=}\frac{\text{Mortality}}{\text{Incidence}}$$

$$\:\text{PIR}\text{=}\frac{\text{Prevalence}}{\text{Incidence}}$$

PC1 = ω₁·YLR + ω₂·DPR + ω₃·MIR + ω₄·PIR

PC2 = ω₅·YLR + ω₆·DPR + ω₇·MIR + ω₈·PIR

In this formula, ω1, ω2, ω3, ......, ω8 are the weights for each ratio in the PCA process, reflecting their contribution to the component.

$$\:\text{PCAscore}\text{（}\text{X}\text{）}\text{=}\text{（}\frac{\text{Var}\text{（}\text{PC}\text{1}\text{）}}{\text{Var}\text{（}\text{PC}\text{1}\text{）}\text{+}\text{Var}\text{（}\text{PC}\text{2}\text{）}}\text{）}\text{×}\text{PC}\text{1+}\text{（}\frac{\text{Var}\text{（}\text{PC}\text{2}\text{）}}{\text{Var}\text{（}\text{PC}\text{1}\text{）}\text{+}\text{Var}\text{（}\text{PC}\text{2}\text{）}}\text{）}\text{×}\text{PC}\text{2}$$

Here, Var(PC₁) and Var(PC₂) respectively represent the variances that the first principal component and the second principal component can explain.

$$\:\text{QCI}\text{（}\text{x}\text{）}\text{=}\text{（}\frac{\text{PCAscore}\text{（}\text{x}\text{）}\text{-}\text{minPCAscore}\text{（}\text{x}\text{）}}{\text{maxPCAscore}\text{（}\text{x}\text{）}\text{-}\text{minPCAscore}\text{（}\text{x}\text{）}}\text{）}\text{×100\%}$$

Subsequently, the score of QCI was converted to a numerical scale range of 0–100 using the formula above, with higher numbers indicating better nursing quality.

2.4 SHAP Analysis

SHAP analysis, grounded in game-theoretic Shapley value principles, provides a framework for interpreting machine learning model predictions. This approach quantifies the contribution of each input feature to individual prediction outcomes while maintaining additive consistency. Feature importance is evaluated through computation of mean absolute SHAP values, with dependency plots employed to examine nonlinear relationships between features and predictions(25–27). The implementation utilized R "INLA" and "BAPC" packages employing visualization tools including force plots, dependency diagrams, and beeswarm plots to present results.

2.5 Bayesian Age-Period-Cohort (BAPC) Model for Forecasting

The model addresses multicollinearity through simultaneous estimation of age, period, and cohort effects within a constrained parameter space. Second-order random walk priors were employed to capture temporal smoothing of each component, while Bayesian inference enabled probabilistic quantification of parameter uncertainty(28). For projection, period effects were extrapolated using autoregressive integrated moving average techniques, cohort effects were modeled based on demographic transition patterns, and age effects were maintained at current estimated levels. Computational implementation utilized integrated nested Laplace approximation algorithms, with model performance evaluated through deviance information criterion and cross-validation procedures. This methodological framework establishes a robust foundation for long-term disease burden forecasting with complete uncertainty characterization(29).

2.6 Statistical Analysis

All statistical analyses and data visualizations were performed using R (version 4.4.2). Descriptive statistics were generated for all key variables, and results were expressed as means with 95% uncertainty intervals (UIs). For trend analyses, p-values < 0.05 were considered statistically significant.

3.1 Global Burden of Burkitt Lymphoma

During the period from 1990 to 2021, the incidence and mortality of Burkitt lymphoma globally showed significant changes (Table 1). In 2021, the worldwide incident cases was 19,072.93 (95% UI: 9,650.59–32,508.93), a substantial increase compared to 6,194.85 in 1990 (95% UI: 4,466.29–8,140.39). The age-standardized incidence rate (ASIR) increased from 0.127/100,000 (95% UI: 0.094–0.168) in 1990 to 0.236/100,000 (95% UI: 0.121–0.398) in 2021, with an estimated annual percentage change (EAPC) of 2.179 (95% CI: 1.788–2.572), indicating a significant upward trend in incidence (Supplementary Fig.S1A). The number of death cases rose from 3,843.67 in 1990 (95% UI: 2,483.89–5,142.19) to 6,525.62 in 2021 (95% UI: 3,955.21–9,035.38). However, the growth in age-standardized mortality rate (ASDR) was slower, increasing from 0.072/100,000 (95% UI: 0.048–0.095) to 0.085/100,000 (95% UI: 0.052–0.117), with an EAPC of 0.623 (95% CI: 0.449–0.797). This suggests that while the number of death cases increased, the relative increase in mortality after age standardization was relatively small (Supplementary Fig.S1B). Notably, adolescents and elderly patients had higher incidence rates. In the 5–9 age group, the incidence of BL peaks in both males and females, with the highest number of cases in this age group. In terms of gender, males had a higher disease burden than females (Supplementary Fig.S2).

Table 1 Comparative analysis of Burkitt lymphoma in terms of incidence, deaths, ASIR, and ASDR between 1990 and 2021.

Table 1

The incidence cases, death cases, ASIR and ASDR of Burkitt lymphoma in 1990 and 2021, and its temporal trends from 1990 to 2021.
	Incidence		ASIR			Death		ASDR
	1990 (95% UI)	2021 (95% UI)	1990 (95% UI)	2021 (95% UI)	EAPCs (95% CI)	1990 (95% UI)	2021 (95% UI)	1990 (95% UI)	2021 (95% UI)	EAPCs (95% CI)
Global	6194.847(4466.288 to 8140.393)	19072.931(9650.585 to 32508.929)	0.127(0.094 to 0.168)	0.236(0.121 to 0.398)	2.179(1.788 to 2.572)	3843.674(2483.887 to 5142.187)	6525.615(3955.210 to 9035.376)	0.072(0.048 to 0.095)	0.085(0.052 to 0.117)	0.623(0.449 to 0.797)
Socio-demographic index
High SDI	2229.010(1605.014 to 3375.501)	8834.960(3391.979 to 17415.618)	0.225(0.163 to 0.339)	0.512(0.209 to 0.936)	2.773(2.164 to 3.385)	470.182(340.562 to 710.150)	1092.035(427.978 to 2033.553)	0.047(0.035 to 0.071)	0.062(0.026 to 0.107)	0.912(0.476 to 1.349)
High-middle SDI	890.733(632.951 to 1235.842)	3691.135(1503.982 to 7343.391)	0.089(0.063 to 0.123)	0.228(0.100 to 0.426)	3.451(3.069 to 3.833)	429.805(293.785 to 570.493)	623.071(287.599 to 1071.431)	0.043(0.029 to 0.056)	0.040(0.019 to 0.064)	-0.231(-0.369 to -0.093)
Middle SDI	618.606(424.656 to 812.956)	2143.052(1175.047 to 3112.729)	0.037(0.026 to 0.050)	0.088(0.048 to 0.127)	2.889(2.705 to 3.074)	537.474(371.241 to 684.201)	845.637(462.292 to 1179.997)	0.033(0.023 to 0.042)	0.035(0.019 to 0.049)	0.140(-0.082 to 0.362)
Low-middle SDI	735.891(410.765 to 1029.086)	1527.483(1005.996 to 2019.252)	0.053(0.031 to 0.072)	0.081(0.053 to 0.108)	1.355(1.282 to 1.427)	716.765(392.191 to 1008.250)	1209.548(784.952 to 1568.212)	0.052(0.030 to 0.070)	0.064(0.042 to 0.082)	0.662(0.555 to 0.769)
Low SDI	1715.442(778.405 to 2617.325)	2861.136(1577.263 to 3907.983)	0.265(0.116 to 0.403)	0.224(0.116 to 0.330)	-0.585(-0.633 to -0.536)	1686.439(762.931 to 2583.760)	2750.353(1516.790 to 3779.878)	0.259(0.115 to 0.396)	0.215(0.113 to 0.318)	-0.621(-0.670 to -0.572)
Regions
East Asia	311.472(158.020 to 464.050)	1368.747(625.406 to 1971.910)	0.028(0.015 to 0.043)	0.080(0.038 to 0.112)	3.173(2.900 to 3.448)	250.827(125.291 to 357.126)	251.480(121.735 to 352.119)	0.023(0.012 to 0.033)	0.015(0.007 to 0.020)	-2.433(-2.898 to -1.965)
Oceania	1.271(0.708 to 2.085)	5.184(1.954 to 9.384)	0.023(0.014 to 0.034)	0.045(0.017 to 0.080)	2.397(2.160 to 2.634)	1.193(0.654 to 1.967)	4.717(1.731 to 8.829)	0.022(0.013 to 0.033)	0.041(0.016 to 0.077)	2.195(1.980 to 2.411)
Southeast Asia	76.913(37.360 to 115.461)	292.488(132.359 to 441.703)	0.018(0.009 to 0.025)	0.044(0.020 to 0.067)	2.566(2.330 to 2.803)	68.922(33.145 to 107.983)	151.971(76.512 to 215.580)	0.016(0.009 to 0.023)	0.023(0.011 to 0.032)	0.607(0.256 to 0.959)
Eastern Europe	140.828(84.842 to 214.966)	207.390(82.777 to 354.910)	0.065(0.039 to 0.098)	0.092(0.037 to 0.150)	1.881(1.297 to 2.469)	62.476(38.905 to 91.146)	49.024(19.937 to 84.370)	0.029(0.018 to 0.041)	0.021(0.009 to 0.034)	-0.637(-0.848 to -0.426)
Central Europe	76.694(48.254 to 122.763)	317.393(114.414 to 623.574)	0.058(0.037 to 0.093)	0.194(0.073 to 0.355)	4.236(3.588 to 4.888)	35.952(23.377 to 57.099)	61.191(22.391 to 114.318)	0.027(0.018 to 0.044)	0.036(0.014 to 0.064)	1.096(0.617 to 1.578)
Australasia	48.861(33.963 to 72.105)	196.689(70.818 to 450.093)	0.228(0.159 to 0.337)	0.449(0.173 to 0.965)	2.177(1.556 to 2.802)	8.976(6.275 to 13.177)	19.086(7.193 to 42.315)	0.042(0.029 to 0.061)	0.043(0.017 to 0.089)	-0.005(-0.416 to 0.408)
High-income Asia Pacific	291.573(132.701 to 495.147)	1490.829(550.083 to 2870.538)	0.156(0.072 to 0.262)	0.389(0.151 to 0.689)	3.011(2.424 to 3.601)	59.433(30.053 to 96.260)	161.755(59.836 to 300.491)	0.032(0.017 to 0.051)	0.040(0.016 to 0.070)	0.751(0.376 to 1.127)
Central Asia	18.753(10.449 to 29.250)	19.456(10.865 to 30.591)	0.029(0.015 to 0.045)	0.021(0.012 to 0.033)	-1.056(-1.706 to -0.400)	11.866(7.024 to 17.680)	8.570(5.028 to 13.180)	0.018(0.010 to 0.027)	0.009(0.006 to 0.015)	-2.314(-2.697 to -1.928)
Caribbean	58.050(35.724 to 95.895)	105.036(62.500 to 162.524)	0.173(0.103 to 0.282)	0.211(0.125 to 0.324)	1.572(1.240 to 1.905)	34.090(21.468 to 55.051)	44.348(24.933 to 66.046)	0.101(0.063 to 0.163)	0.092(0.051 to 0.139)	0.443(0.197 to 0.690)
Andean Latin America	30.111(21.014 to 48.127)	125.507(50.265 to 215.204)	0.076(0.054 to 0.122)	0.200(0.079 to 0.345)	3.305(3.173 to 3.438)	28.743(20.082 to 45.680)	52.877(22.579 to 88.349)	0.073(0.052 to 0.116)	0.085(0.035 to 0.144)	0.274(0.059 to 0.490)
Western Europe	958.692(680.850 to 1465.578)	4504.641(1296.676 to 11343.784)	0.205(0.146 to 0.312)	0.618(0.194 to 1.430)	3.723(3.071 to 4.379)	205.789(146.101 to 312.081)	472.334(140.738 to 1130.699)	0.043(0.031 to 0.066)	0.063(0.020 to 0.139)	1.223(0.807 to 1.640)
Tropical Latin America	140.494(101.614 to 203.961)	582.578(263.540 to 949.608)	0.099(0.071 to 0.146)	0.245(0.111 to 0.390)	3.055(2.678 to 3.433)	124.579(90.410 to 178.820)	271.865(127.905 to 421.541)	0.090(0.065 to 0.131)	0.114(0.055 to 0.173)	0.823(0.480 to 1.166)
Southern Latin America	94.693(63.672 to 142.791)	416.356(201.410 to 734.783)	0.195(0.131 to 0.295)	0.537(0.265 to 0.924)	3.544(3.248 to 3.840)	60.838(41.895 to 89.239)	124.611(61.864 to 214.425)	0.126(0.086 to 0.185)	0.159(0.080 to 0.269)	0.943(0.799 to 1.087)
High-income North America	1198.379(836.948 to 1784.134)	3928.149(1698.219 to 6760.968)	0.381(0.269 to 0.560)	0.712(0.330 to 1.157)	2.234(1.607 to 2.864)	230.982(161.541 to 346.305)	534.413(227.125 to 896.260)	0.072(0.051 to 0.107)	0.095(0.043 to 0.150)	1.057(0.570 to 1.546)
Central Latin America	115.944(80.690 to 175.748)	519.166(270.774 to 835.573)	0.079(0.053 to 0.121)	0.208(0.108 to 0.334)	3.230(2.879 to 3.582)	101.369(71.960 to 150.152)	223.304(118.342 to 334.259)	0.071(0.050 to 0.105)	0.090(0.048 to 0.134)	0.748(0.544 to 0.952)
North Africa and Middle East	221.130(132.307 to 341.668)	775.072(443.134 to 1116.633)	0.062(0.037 to 0.094)	0.136(0.076 to 0.200)	2.980(2.767 to 3.193)	189.628(115.839 to 294.800)	248.216(145.959 to 355.619)	0.054(0.032 to 0.083)	0.044(0.025 to 0.062)	-0.606(-0.755 to -0.457)
South Asia	338.315(132.744 to 542.487)	493.060(230.869 to 728.805)	0.025(0.010 to 0.039)	0.027(0.013 to 0.041)	0.125(-0.068 to 0.318)	331.509(130.905 to 534.086)	395.021(187.382 to 587.895)	0.024(0.010 to 0.038)	0.022(0.010 to 0.033)	-0.541(-0.660 to -0.422)
Southern Sub-Saharan Africa	17.706(10.558 to 25.540)	54.663(25.662 to 80.902)	0.032(0.018 to 0.045)	0.072(0.033 to 0.105)	2.946(2.714 to 3.179)	17.049(10.284 to 24.502)	17.049(10.284 to 24.502)	0.031(0.019 to 0.043)	0.061(0.029 to 0.090)	2.471(2.247 to 2.695)
Western Sub-Saharan Africa	736.011(340.330 to 1056.597)	1620.095(903.227 to 2214.900)	0.276(0.128 to 0.406)	0.269(0.142 to 0.388)	0.022(-0.055 to 0.099)	727.909(332.282 to 1050.614)	1457.485(785.865 to 1972.893)	0.273(0.124 to 0.401)	0.242(0.125 to 0.338)	-0.245(-0.361 to -0.129)
Eastern Sub-Saharan Africa	1140.708(490.145 to 1748.061)	1786.597(915.769 to 2589.946)	0.481(0.196 to 0.734)	0.390(0.189 to 0.612)	-0.786(-0.832 to -0.740)	1116.976(481.658 to 1735.024)	1696.028(884.010 to 2404.312)	0.469(0.192 to 0.716)	0.372(0.182 to 0.586)	-0.829(-0.871 to -0.787)
Central Sub-Saharan Africa	178.249(69.040 to 304.391)	263.835(117.705 to 399.796)	0.271(0.102 to 0.443)	0.199(0.078 to 0.311)	-0.997(-1.052 to -0.941)	174.567(69.065 to 299.184)	250.819(112.672 to 378.559)	0.265(0.099 to 0.433)	0.192(0.075 to 0.306)	-1.036(-1.096 to -0.976)

Abbreviations: ASIR, age-standardized incidence rate; ASDR, age-standardized death rate; SDI, socio-demographic index; UI, uncertainty interval EAPC; Estimated annual percentage change. The rates for all indicators are expressed per 100,000 population.

3.2 Regional Burden of Burkitt Lymphoma

There are significant differences in the disease burden across different socio-demographic index (SDI) regions (Fig. 1 and Table 1). In 2021, the ASIR was highest in high SDI regions (0.512/100,000), while it was lowest in low SDI regions (0.224/100,000). Notably, from 1990 to 2021, the ASIR in high SDI and middle-high SDI regions showed a significant upward trend (EAPC of 2.773 and 3.451, respectively), while the ASIR in low SDI regions showed a downward trend (EAPC: -0.585). In terms of mortality rates, the ASDR was lowest in high SDI regions (0.062/100,000), and its EAPC was positive (0.912), indicating an increase in mortality rates. Conversely, the ASDR was highest in low SDI regions (0.215/100,000), but showed a downward trend (EAPC: -0.621). This suggests that although the mortality burden in low SDI regions remains heavy, there has been improvement in recent years (Supplementary Table S1).

Significant heterogeneity exists in the disease burden among regions (Table 1 and Supplementary Table S1). In 2021, the ASIR was highest in Western Europe (0.618/100,000) and lowest in South Asia (0.027/100,000). In terms of mortality, the ASDR was highest in Eastern Africa (0.372/100,000) and lowest in East Asia (0.015/100,000). Notably, although the ASIR in East Asia increased (EAPC: 3.173), the ASDR significantly decreased (EAPC: -2.433), indicating significant progress in the treatment and management of Burkitt lymphoma in the region. Conversely, both the ASIR and ASDR in Eastern and Central Africa showed declining trends, but these regions remain the most heavily burdened globally.

The incidence, mortality and DALY rates of Burkitt lymphoma in 2021 showed a significant gradient distribution among different SDI regions (Fig. 1 and Supplementary Table S1). The incidence is highest in areas with high SDI and gradually decreases as SDI decreases. However, mortality and DALY rates were higher in low and low moderate SDI areas, indicating poor disease outcomes in these regions despite low incidence, reflecting an uneven distribution of healthcare resources and quality of care.

3.3 Trends and Future Projections of the Disease Burden of Burkitt Lymphoma

The overall incidence of Burkitt lymphoma globally has shown a consistent increasing trend from 1990 to 2021. The period from 1990 to 1993 was a mild growth phase (APC = 2.883%), 1993 to 2001 was a rapid growth phase (APC = 5.025%), and 2001 to 2007 saw a slowdown in growth rate (APC = 2.993%), but still maintained a stable increase. From 2007 to 2021, it was a plateau phase (APC = 0.037%), with the incidence rate remaining almost unchanged during these 14 years. Predictions indicate that this trend will continue until 2035 (Fig. 2A and Supplementary Fig.S3A). Among all indicators, the mortality rate has shown the most significant and sustained improvement. The overall trend from 1990 to 2035 is as follows: 1990–2004: a period of slow increase (APC = 1.662%); 2004–2012: a period of significant decline (APC = -0.084%); 2012–2016: a period of slight recovery (APC = 0.548%), with a temporary rebound in the mortality trend; 2016–2021: a period of accelerated decline (APC=-1.234%), which is the most important stage, with mortality entering a rapid and significant downward channel, this rapid decline trend has been consistently maintained until 2035 (Fig. 2C and Supplementary Fig. S3B).

The trends of YLDs rates and prevalence rates are highly similar, with the overall trend being a significant increase followed by stabilization, entering a high plateau phase since 2016. Predictions indicate that the prevalence rates and YLDs rates will remain at this historically high level for the next decade or so, without significant growth (Fig. 2B and 2D). From 1990 to around 2016, the YLLs rate showed an overall upward trend, indicating that during this period, the increase in incidence may have offset some of the benefits from improved survival rates, leading to an increase in the total number of years of life lost due to premature death. After 2016, a decisive turning point occurred, with the YLLs rate beginning to decline sharply and continuously due to the dual effects of decreasing mortality and a beginning decrease in incidence. Predictions indicate that this positive downward trend will be the main melody in the future (Fig. 2E). The DALY rate, which combines the burden of premature death and disability, has a trend highly similar to mortality and has also experienced the process of "increase → deceleration of increase → near stability → significant decrease" (Fig. 2F), which is consistent with the characteristics of Burkitt lymphoma as a highly lethal cancer. During the period from 2016 to 2021, the DALY rate began a rapid downward trend, with an average annual decrease of 1.50% (Supplementary Fig.S3C), which is highly consistent with the accelerated decrease in mortality during the same period (APC=-1.234%) (Supplementary Fig. 3B), clearly indicating that the overall disease burden of Burkitt lymphoma globally is being effectively reduced. Predictions indicate that as a core indicator of the total disease burden, the DALYs rate will continue to decline in the future (Fig. 2F and Supplementary Fig.S3C).

3.4 Construction and verification of quality of care index

We successfully constructed the QCI for Burkitt lymphoma. Figure 3B shows that YLR is strongly positively correlated with DPR, while MIR shows weak correlation with other indicators such as PIR. This demonstrates that each indicator provides an independent and complementary perspective, evaluating with any single indicator alone would lead to biased conclusions. Principal component analysis (PCA) provides a foundation for integrating this information (Fig. 3C). The first principal component (Dim1, 91.8%) is a global disease burden core dimension contributed by all indicators. The second principal component (Dim2, 7.6%) is a disease pattern differentiation dimension defined by PIR (positive load, representing chronication) and MIR (negative load, representing high lethality). Overall, the construction of the QCI has a solid statistical foundation. It successfully integrates indicators measuring different aspects into a unified framework through PCA, encompassing both the overall burden intensity (Dim1) and key pattern differences (Dim2), thereby forming a comprehensive and unbiased comprehensive evaluation tool.

To further explore the key factors influencing QCI, we constructed a machine learning model to fit QCI data (Fig. 4), which showed high correlation between predicted and actual values on the test set, indicating good model performance (Fig. 4A). Global analysis based on SHAP values (Fig. 4B) revealed that age is the most important factor affecting QCI, with the widest distribution range of SHAP values, suggesting its greatest impact on QCI, followed by year and gender. To further investigate the age effect pattern, we plotted SHAP dependence plots (Fig. 4C), which revealed a complex and non-monotonic relationship between age and QCI. The most striking finding was that in the young and adolescent population aged 0 to 35, their SHAP values were consistently negative, indicating that the 'young' feature was associated with lower predicted care quality (QCI) in the model, which may be related to the heavier global burden of disease among children and adolescents (Supplementary Fig. S2). Subsequently, SHAP values enter the positive region after about age 35, indicating that the age characteristics of middle and old age are associated with a higher predictive QCI. Gender analysis shows that after controlling for other variables, the SHAP values for female gender as a feature are more often distributed in the negative region, suggesting it is more commonly associated with a lower predictive QCI, i.e., the care quality for female patients may systematically be lower than that for male patients. We employed SHAP waterfall plots for local interpretability analysis, taking the case of a 52.5-year-old male patient from 1990 (Fig. 4D). His baseline predicted QCI was 56.8, with age being the strongest positive driving factor, contributing a SHAP value of + 9.68, significantly improving the prediction. Year was a strong negative driving factor, with a SHAP value of -5.84. Gender had a slight positive impact on the prediction, contributing + 1.67, collectively resulting in his final predicted QCI being only 62.4.

3.5 Analysis of global QCI distribution and influencing factors

We compared the QCI across global and different SDI regions, the results showed a clear rank correlation between QCI levels and the degree of social development (Fig. 3A). The median QCI was highest in high SDI regions (78.50%), indicating that the overall quality of care in these regions was optimal, followed by medium-high SDI regions (63.70%). In contrast, the median QCI was lowest in low SDI regions among all groups (21.60%), while the median QCI in medium SDI (39.10%) and medium-low SDI regions fell between these two extremes, showing a clear gradient of social and economic development: the higher the SDI rank, the higher the median QCI. Additionally, the box plots in high SDI regions were shorter and closer to the top of the chart, indicating that the QCI values across different regions within this group were relatively concentrated with higher consistency. In comparison, the box plots in medium-high SDI or medium SDI regions were longer, suggesting greater variability in QCI among different regions within these groups.

Similarly, there are significant differences in the geographical distribution of the average QCI across various regions of the world (Fig. 5A). The Northern Europe region has the highest QCI scores (falling within the 84.41–89.96% range), followed by high-income areas such as the Persian Gulf and Eastern Mediterranean, which also exhibit a higher level of QCI (ranging from 65.73–84.41%). These areas are high SDI regions (Fig. 3A), boasting advanced high-income economies and substantial investments in medical infrastructure, consistently performing well in terms of quality of life and healthcare indices. On the contrary, the QCI scores in regions such as Western Sub-Saharan Africa and Southeast Asia are in the lower range (13.50–33.64%), indicating that these areas face significant challenges in terms of healthcare accessibility and quality of care. These regions are the geographical cores of the global "medium and low" and "low" SDI categories (Fig. 3A). The QCI of the Balkan Peninsula and the Caribbean and Central America is in the "medium" to "high" range (33.64–55.99%), represented on the map in neutral to light warm tones, corresponding to the "medium and high" and "medium" SDI categories (Fig. 3A). Overall, the distribution of QCI levels shows a clear spatial consistency with the level of social and economic development. The QCI of the Balkan Peninsula and the Caribbean and Central America is in the range of "medium" to "high" (33.64–55.99%), represented on the map by neutral to light warm tones, corresponding to the "medium-high" and "medium" SDI zones (Fig. 3A). Regarding countries, the United Kingdom, Cyprus, and United States of America rank in the top 3 for QCI, while the Republic of Benin, located in central-western Africa, has the lowest QCI (Supplementary Table S2). In general, the distribution of high and low QCI shows a clear spatial consistency with the level of socio-economic development.

3.6 Long-term evolution trend of QCI

Over the past three decades, the landscape of Burkitt lymphoma care quality has undergone fundamental shifts across both age and gender dimensions. In 1990, the QCI value steadily increased with age, reaching a peak in the elderly group (Fig. 6A). By 2021, this pattern was completely reversed, with the QCI value being highest among children and adolescents (Fig. 6B), followed by a significant decline as age increased. Similarly, the quality of care between different genders has also changed. In 1990, male patients had higher QCI values than female patients in almost all age groups (Fig. 6A), indicating that the healthcare system at the time may have been more favorable to male patients. However, in 2021, female patients' QCI values surpassed or even significantly exceeded those of male patients in most age groups (Fig. 6A). This reversal suggests that the progress made in the past three decades in promoting health equity and gender equality has successfully reversed the quality of care gap that previously favored male patients and has begun to benefit female patients. At the same time, compared to 1990, the absolute QCI values for all age and gender groups in 2021 were significantly improved (Fig. 6), indicating substantial progress in global healthcare levels, with the youngest and female populations benefiting the most, while older male and middle-aged male populations experienced relatively smaller gains, leading to reshaping the age and gender landscape.

3.7 QCI forecast trends from 2022 to 2035

Based on historical data from 1990 to 2021, we have predicted the trends of Burkitt lymphoma QCI from 2022 to 2035 (Fig. 7). The analysis reveals a sustained long-term growth in global QCI, while the gender disparity pattern of QCI will complete a fundamental reversal and form a new norm. In the historical trend (1990–2021), the global QCI value began a steady climb from a lower baseline in 1990 (less than 25%) and reached a higher level by 2021 (approximately 75–85%), marking a significant leap in diagnostic and treatment technologies as well as healthcare accessibility. The predictive model shows that this positive upward trend will continue until 2035, after 2021, the growth rate gradually slowed and entered a high-level plateau around 2035, at which point the median global QCI is expected to stabilize above 90%. This indicates that the medical management strategies for Burkitt lymphoma will tend to mature and optimize in the coming decade. Additionally, the gender disparity pattern has undergone a complete transformation, with the advantage in nursing quality for women continuing to expand and stabilize. Looking back at historical data, the QCI values for male patients have consistently been higher than those for female patients. Predictions show that the QCI values for female patients will historically surpass those for male patients around 2025. By 2035, the advantage in nursing quality for female patients will not only continue to exist but also the gap may further consolidate and slightly widen, forming a new steady state.

Based on GBD 2021 data, we analyzed 21 global health regions encompassing 195 countries and 5 SDI quintile regions, systematically assessing the global burden of disease and quality of care for BL. We found that many countries are currently experiencing epidemiological transitions as a result of rapid advancements in medical diagnosis and treatment technologies, the acceleration of population aging, and shifts in risk factors (12). The rise in global ASIR and ASDR is a cause for concern, although the growth in ASDR is relatively slow. The increase in incidence may be attributed to improved and sensitive disease surveillance and diagnostic systems(3). Despite continuous optimization of treatment strategies, the concurrent rise in mortality suggests that the survival benefits gained from these advancements may be offset by the absolute increase in the number of cases, especially in regions with limited healthcare resources. Time trend analysis shows that since 2016, global ASDR and DALY rates for BL have begun to decline, indicating progress in treatment modalities and care systems, such as the application of rituximab in BL, where rituximab in combination with high-intensity chemotherapy significantly improves the cure rate and survival duration for Burkitt lymphoma patients (30, 31).

The speed and quality of diagnosis and treatment have a critical impact on long-term patient outcomes, and diagnostic delays may lead to greater disease burden and more complex clinical situations for patients. In this study, high SDI regions had higher QCI, the finding consistent with multiple global cancer care quality studies, benefiting from their well-developed medical infrastructure, high healthcare spending, easier access to high-intensity chemotherapy regimens, supportive care, and professional experience in managing chemotherapy toxicities(7). Conversely, low-SDI regions, which are popular areas for localized BL, still bear a heavy disease burden due to insufficient diagnostic capacity (32), poor treatment accessibility, weak supportive care, and issues like P. falciparum infections, resulting in low-quality care (33, 34). This indicates that the level of disease burden is no longer primarily determined by biological factors, but more by socioeconomic and healthcare system levels (35).

Age and gender influence on QCI revealed another dimension of nursing quality. In the SHAP analysis, the 0 to 35 age group was associated with lower predicted QCI, and additionally, the QCI values in 1990 steadily increased with age. The underlying reasons and the age pattern of BL incidence are related. BL rates showed a bimodal age pattern with pediatric and elderly peaks in all regions (3), which is consistent with our findings. There are reports that between 1973 and 2005 in the US, BL rates have a trimodal pattern in the 0-14-year-old pediatric age group with an early peak, and the other two peaks occur at the 40-year-old and 70-year-old age groups (1, 36). The pattern reflects the association with different age-specific EBV infection rates (37). In regions of Africa with extremely scarce overall medical resources, the high incidence of BL in children aged 0–14 is most closely associated with the risk of infection with P. falciparum (4, 38). In sub-Saharan Africa, treatment abandonment often occurs in families that need to borrow money for diagnosis and treatment, accounting for about two-thirds (35); in East Africa, the rate of treatment delays in children and young adults with lymphoma is relatively high (32), leading to poor treatment outcomes and failures. There are differences in treatment approaches, children with BL typically receiving extremely intensive chemotherapy regimens, which are highly effective in resource-rich areas. However, in resource-limited regions, due to the lack of sufficient supportive treatments such as anti-infection, nutritional support, component blood transfusions, and the use of cytokines, the mortality rate related to treatment is extremely high (33, 39), thus lowering the average QCI of the entire pediatric population. Therefore, when global data is analyzed as a whole, the significant disease burden in economically underdeveloped regions may completely overshadow the high cure rates of children with BL in high-income areas, presenting a general pattern of youth disadvantage. However, in 2021, the situation was completely reversed, with the QCI value being the highest among adolescents and young adults, gradually decreasing with age, which may be related to the optimization of treatment regimens and the control of malaria, among other factors, leading to improved prognosis (30). This finding strongly warns us that the survival advantage of pediatric cancer globally is not a given and is highly dependent on the underlying healthcare systems (34), it is particularly important to take individualized therapy for low-income areas (40).

It is also interesting that in this study, men were associated with a higher QCI, which contrasts with previous studies where women typically had higher nursing quality (13, 14). Men's BL rate is two to four times higher than that of women, and male dominance is a consistent feature of BL across all age groups and geographic regions (3). In this study, it was also observed that male patients had a heavier disease burden, manifested as higher incidence, mortality, and DALY rates compared to women. This difference may be related to sex chromosome differences that affect cancer susceptibility, making women's immune systems more adept at monitoring and clearing malignant cells. A similar pattern has been observed in other cancers, and male patients often have more severe outcomes (37, 41). Additionally, gender differences in sex hormones, gut microbiome composition, and the interplay of environmental and behavioral factors (42) may contribute to this. HIV infection in male homosexuality, which can lead to immune deficiency, may also be a factor contributing to the increase in BL (43, 44), but the exact mechanisms still need further exploration. Socio-cultural factors may also play a significant role in this, families being more willing to invest in the health of male children, seeking medical attention more promptly, and thus receiving earlier and more effective treatment. Predictive analysis indicates that the QCI values of female patients will historically surpass those of male patients around 2025. This long-term trend reversal may result from the combined effects of multiple factors, such as increased global research and development investment in women's health, the implementation of targeted public health policies, and the advantages brought by improved awareness of women's health, leading to earlier diagnosis and better adherence to treatment. This new landscape also suggests the need to be vigilant about the possibility of male patients becoming a new relatively vulnerable group, and targeted intervention measures should be developed to ensure that all patients can equally benefit from future medical advancements, ultimately achieving the goal of universal health coverage.

This study predicts that by 2035, the global QCI will continue to improve and gradually enter a plateau phase, a trend suggesting that health systems need to begin focusing on long-term follow-up of patients in the recovery phase, complication management, and quality of life optimization (8). Of course, social psychological support should also be given full attention and efforts should be made in this regard (45). Additionally, future efforts should focus on the unmet needs in the treatment of elderly, with central nervous system involvement, and relapsed/refractory BL patients (1, 46). Predictions for future trends indicate that while global nursing quality is expected to improve slowly, existing health inequalities will persist or even worsen without targeted interventions. This calls for global action: in low SDI regions, strengthen laboratory and pathological diagnostic capabilities(32); promote adapted and effective treatment strategies through training and support (39); enhance supportive care to prevent and manage complications such as infections (7, 34); and persist in primary prevention measures for infectious diseases like malaria (47). For high SDI countries, continue to focus on etiology and prevention research, developing and studying efficient and low-toxicity treatment strategies (e.g., immunotherapy, targeted therapy) (31, 48).

The strength of this study lies in using GBD, an authoritative, standardized, and comparable global data source, and constructing a more comprehensive and interpretable QCI. Limitations mainly include: the inability to perform stratified analysis for the three subtypes of BL; BAPC predictions rely on the continuation of historical trends, which may not accurately predict the impact of future public health emergencies (such as pandemics) or treatment breakthroughs; Although SHAP analysis can provide interpretations of important features from a machine learning model, it is subject to model-specific biases which misrepresent the relationships between variables (26); failure to promptly diagnose and register all cases in some areas, affecting the accuracy of GBD data.

In summary, this study reveals that the global burden of BL disease has not decreased over the past three decades and that there is a severe global inequality in the quality of care. The disparities in care quality not only stem from macro-level socioeconomic factors but are also closely related to the demographic characteristics of individual patients. Future global health policies and priorities should focus on narrowing the gap in healthcare quality, calling on the international community and national governments to increase investment in research on effective and low-toxicity solutions, improving treatment accessibility, and building healthcare infrastructure to achieve equitable improvements in the quality of BL care.

Conflicts of Interest

The authors have no competing interests to declare that are relevant to the content of this article.

Acknowledgements

We express our sincere appreciation to the Institute for Health Metrics and Evaluation (IHME) for making the GBD data publicly available.

Data availability statement

The data utilized in this study are publicly accessible from the following sources: the Global Burden of Disease Results Tool of the Global Health Data Exchange, this data can be found at: https://vizhub.healthdata.org/gbd-results/.

Funding Information

No funding was received for conducting this study.

Ethics approval

Not applicable.

Consent to Participate s

Not applicable.

Author Contributions

MY and LJY conceptualized the study. MY、LF and PHJ developed the study protocol. PHJ、CYX was responsible for statistical analyses and interpretations of the data. SYJ、CYX and LJY performed the literature search. MY and SYJ drafted the manuscript, which was critically revised by other authors. LF and PHJ accessed and validated the data. All authors reviewed and approved the final version of the manuscript.

Lopez C, Burkhardt B, Chan JKC, Leoncini L, Mbulaiteye SM, Ogwang MD et al (2022) Burkitt lymphoma. Nat Rev Dis Primers 8(1):78. https://doi.org/10.1038/s41572-022-00404-3
Panea RI, Love CL, Shingleton JR, Reddy A, Bailey JA, Moormann AM et al (2019) The whole-genome landscape of Burkitt lymphoma subtypes. Blood 134(19):1598–1607. https://doi.org/10.1182/blood.2019001880
Mbulaiteye SM, Devesa SS (2022) Burkitt Lymphoma Incidence in Five Continents. Hemato 3(3):434–453. https://doi.org/10.3390/hemato3030030
Broen K, Dickens J, Trangucci R, Ogwang MD, Tenge CN, Masalu N et al (2023) Burkitt lymphoma risk shows geographic and temporal associations with Plasmodium falciparum infections in Uganda, Tanzania, and Kenya. Proc Natl Acad Sci U S A 120(2):e2211055120. https://doi.org/10.1073/pnas.2211055120
Atallah-Yunes SA, Murphy DJ, Noy A (2020) HIV-associated Burkitt lymphoma. Lancet Haematol 7(8):e594–e600. https://doi.org/10.1016/S2352-3026(20)30126-5
Crombie J, LaCasce A (2021) The treatment of Burkitt lymphoma in adults. Blood 137(6):743–750. https://doi.org/10.1182/blood.2019004099
Dlouhy I, Viegas D, Coelho I, Ionita A, Carvalho S, Cabecadas J et al (2025) Intensive Treatment in Adult Burkitt Lymphoma with Lymphome Malin B (LMB) Regimen: Excellent Outcomes Despite Substantial Toxicity and Supportive Care Demands. Cancers (Basel) 17(17). https://doi.org/10.3390/cancers17172914
Johnson PC, Markovitz NH, Yi A, Horick N, Newcomb RA, Cronin K et al (2023) Clinical Outcomes and Health Care Utilization in Patients With Burkitt Lymphoma. JCO Oncol Pract 19(9):759–766. https://doi.org/10.1200/OP.23.00146
Nussbaum DP, Rushing CN, Sun Z, Yerokun BA, Worni M, Saunders RS et al (2021) Hospital-level compliance with the commission on cancer's quality of care measures and the association with patient survival. Cancer Med 10(11):3533–3544. https://doi.org/10.1002/cam4.3875
Keykhaei M, Masinaei M, Mohammadi E, Azadnajafabad S, Rezaei N, Saeedi Moghaddam S et al (2021) A global, regional, and national survey on burden and Quality of Care Index (QCI) of hematologic malignancies; global burden of disease systematic analysis 1990–2017. Exp Hematol Oncol 10(1):11. https://doi.org/10.1186/s40164-021-00198-2
Metekoua C, Ruffieux Y, Mwansa-Kambafwile J, Kellett P, Egger M, Muchengeti M et al (2025) Patterns of incident Burkitt lymphoma during the HIV epidemic among the Black African and White population in South Africa. Br J Cancer 132(5):462–468. https://doi.org/10.1038/s41416-024-02937-8
Access GBDH, Quality C. Electronic address cue, Access GBDH, Quality C (2017) Healthcare Access and Quality Index based on mortality from causes amenable to personal health care in 195 countries and territories, 1990–2015: a novel analysis from the Global Burden of Disease Study 2015. Lancet. ;390(10091):231 – 66. https://doi.org/10.1016/S0140-6736(17)30818-8
Pan Y, Liu Q, Zhang N, Peng S, Li X, Zhou F (2025) Global assessment of leukemia care quality: insights from the quality of care index (QCI) from 1990 to 2021. EClinicalMedicine 79:102996. https://doi.org/10.1016/j.eclinm.2024.102996
Geng J, Zhao J, Fan R, Zhu Z, Zhang Y, Zhu Y et al (2024) Global, regional, and national burden and quality of care of multiple myeloma, 1990–2019. J Glob Health 14:04033. https://doi.org/10.7189/jogh.14.04033
Uyeda JC, Caetano DS, Pennell MW (2015) Comparative Analysis of Principal Components Can be Misleading. Syst Biol 64(4):677–689. https://doi.org/10.1093/sysbio/syv019
Elhaik E (2022) Principal Component Analyses (PCA)-based findings in population genetic studies are highly biased and must be reevaluated. Sci Rep 12(1):14683. https://doi.org/10.1038/s41598-022-14395-4
Diseases GBD, Injuries C (2024) Global incidence, prevalence, years lived with disability (YLDs), disability-adjusted life-years (DALYs), and healthy life expectancy (HALE) for 371 diseases and injuries in 204 countries and territories and 811 subnational locations, 1990–2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet 403(10440):2133–2161. https://doi.org/10.1016/S0140-6736(24)00757-8
Murray CJL (2022) The Global Burden of Disease Study at 30 years. Nat Med 28(10):2019–2026. https://doi.org/10.1038/s41591-022-01990-1
Haagsma JA, James SL, Castle CD, Dingels ZV, Fox JT, Hamilton EB et al (2020) Burden of injury along the development spectrum: associations between the Socio-demographic Index and disability-adjusted life year estimates from the Global Burden of Disease Study 2017. Inj Prev 26(Supp 1):i12–i26. https://doi.org/10.1136/injuryprev-2019-043296
Global Burden of Disease, Cancer C, Kocarnik JM, Compton K, Dean FE, Fu W, Gaw BL et al (2022) Cancer Incidence, Mortality, Years of Life Lost, Years Lived With Disability, and Disability-Adjusted Life Years for 29 Cancer Groups From 2010 to 2019: A Systematic Analysis for the Global Burden of Disease Study 2019. JAMA Oncol 8(3):420–444. https://doi.org/10.1001/jamaoncol.2021.6987
Cao F, Liu YC, Ni QY, Chen Y, Wan CH, Liu SY et al (2023) Temporal trends in the prevalence of autoimmune diseases from 1990 to 2019. Autoimmun Rev 22(8):103359. https://doi.org/10.1016/j.autrev.2023.103359
Cao G, Liu J, Liu M, Global (2022) Regional, and National Incidence and Mortality of Neonatal Preterm Birth, 1990–2019. JAMA Pediatr 176(8):787–796. https://doi.org/10.1001/jamapediatrics.2022.1622
Nejadghaderi SA, Roshani S, Mohammadi E, Yoosefi M, Rezaei N, Esfahani Z et al (2022) The global, regional, and national burden and quality of care index (QCI) of colorectal cancer; a global burden of disease systematic analysis 1990–2019. PLoS ONE 17(4):e0263403. https://doi.org/10.1371/journal.pone.0263403
Ghamari SH, Yoosefi M, Abbasi-Kangevari M, Malekpour MR, Saeedi Moghaddam S, Shahin S et al (2022) Trends in Global, Regional, and National Burden and Quality of Care Index for Liver Cancer by Cause from Global Burden of Disease 1990–2019. Hepatol Commun 6(7):1764–1775. https://doi.org/10.1002/hep4.1910
Qurat Ul Ain S, Islam Rather KU (2025) Integrated statistical modeling and machine learning techniques with SHAP for epidemiological data analysis. Ann Epidemiol 108:85–91. https://doi.org/10.1016/j.annepidem.2025.06.012
Takefuji Y (2025) Reevaluating feature importance in machine learning: concerns regarding SHAP interpretations in the context of the EU artificial intelligence act. Water Res 280:123514. https://doi.org/10.1016/j.watres.2025.123514
Baptista ML, Goebel K, Henriques EMP (2022) Relation between prognostics predictor evaluation metrics and local interpretability SHAP values. Artif Intell 306. https://doi.org/10.1016/j.artint.2022.103667
Riebler A, Held L (2017) Projecting the future burden of cancer: Bayesian age-period-cohort analysis with integrated nested Laplace approximations. Biom J 59(3):531–549. https://doi.org/10.1002/bimj.201500263
Lin K, Shao J, Cao Y, Lu L, Lei P, Chen X et al (2024) The trend of lymphoma incidence in China from 2005 to 2017 and lymphoma incidence trend prediction from 2018 to 2035: a log-linear regression and Bayesian age-period-cohort analysis. Front Oncol 14:1297405. https://doi.org/10.3389/fonc.2024.1297405
Mawalla WF, Achola C, Nabalende H, Otim I, Legason ID, Ogwang MD et al (2025) Rituximab for children with EBV-positive Burkitt lymphoma in East Africa. Blood Adv 9(10):2393–2401. https://doi.org/10.1182/bloodadvances.2024015234
Ribrag V, Koscielny S, Bosq J, Leguay T, Casasnovas O, Fornecker LM et al (2016) Rituximab and dose-dense chemotherapy for adults with Burkitt's lymphoma: a randomised, controlled, open-label, phase 3 trial. Lancet 387(10036):2402–2411. https://doi.org/10.1016/S0140-6736(15)01317-3
Mawalla WF, Morrell L, Chirande L, Achola C, Mwamtemi H, Sandi G et al (2023) Treatment delays in children and young adults with lymphoma: a report from an East Africa lymphoma cohort study. Blood Adv 7(17):4962–4965. https://doi.org/10.1182/bloodadvances.2022009398
Chapman H, Ntemi PS, Gisiri M, Vasudevan L, Kashaigili HJ, Schroeder K (2024) Retrospective analysis of pediatric patients with Burkitt lymphoma treated in Tanzania following the implementation of the 2016 National Treatment Guidelines: Poor outcomes to current standard-of-care therapy. Pediatr Blood Cancer 71(9):e31145. https://doi.org/10.1002/pbc.31145
Israels T, Ribeiro RC, Molyneux EM (2010) Strategies to improve care for children with cancer in Sub-Saharan Africa. Eur J Cancer 46(11):1960–1966. https://doi.org/10.1016/j.ejca.2010.03.027
Chagaluka G, Afungchwi GM, Landman L, Njuguna F, Hesseling P, Tchintseme F et al (2021) Treatment abandonment: A report from the collaborative African network for childhood cancer care and research-CANCaRe Africa. Pediatr Blood Cancer 68(12):e29367. https://doi.org/10.1002/pbc.29367
Mbulaiteye SM, Anderson WF, Bhatia K, Rosenberg PS, Linet MS, Devesa SS (2010) Trimodal age-specific incidence patterns for Burkitt lymphoma in the United States, 1973–2005. Int J Cancer 126(7):1732–1739. https://doi.org/10.1002/ijc.24934
Richter J, John K, Staiger AM, Rosenwald A, Kurz K, Michgehl U et al (2022) Epstein-Barr virus status of sporadic Burkitt lymphoma is associated with patient age and mutational features. Br J Haematol 196(3):681–689. https://doi.org/10.1111/bjh.17874
Rochford R, Cannon MJ, Moormann AM (2005) Endemic Burkitt's lymphoma: a polymicrobial disease? Nat Rev Microbiol 3(2):182–187. https://doi.org/10.1038/nrmicro1089
Ozuah NW, Lubega J, Allen CE, El-Mallawany NK (2020) Five decades of low intensity and low survival: adapting intensified regimens to cure pediatric Burkitt lymphoma in Africa. Blood Adv 4(16):4007–4019. https://doi.org/10.1182/bloodadvances.2020002178
Gopal S, Gross TG (2018) How I treat Burkitt lymphoma in children, adolescents, and young adults in sub-Saharan Africa. Blood 132(3):254–263. https://doi.org/10.1182/blood-2018-04-844472
Dunford A, Weinstock DM, Savova V, Schumacher SE, Cleary JP, Yoda A et al (2017) Tumor-suppressor genes that escape from X-inactivation contribute to cancer sex bias. Nat Genet 49(1):10–16. https://doi.org/10.1038/ng.3726
Zhang X, Zhong H, Li Y, Shi Z, Ren H, Zhang Z et al (2021) Sex- and age-related trajectories of the adult human gut microbiota shared across populations of different ethnicities. Nat Aging 1(1):87–100. https://doi.org/10.1038/s43587-020-00014-2
Serraino D, Salamina G, Franceschi S, Dubois D, La Vecchia C, Brunet JB et al (1992) The epidemiology of AIDS-associated non-Hodgkin's lymphoma in the World Health Organization European Region. Br J Cancer 66(5):912–916. https://doi.org/10.1038/bjc.1992.384
Centers for Disease C (1982) Diffuse, undifferentiated non-Hodgkins lymphoma among homosexual males–United States. MMWR Morb Mortal Wkly Rep 31(21):277–279. https://doi.org/, https://www.ncbi.nlm.nih.gov/pubmed/6808345
Westmoreland K, Reeve BB, Amuquandoh A, van der Gronde T, Manthalu O, Correia H et al (2018) Translation, psychometric validation, and baseline results of the Patient-Reported Outcomes Measurement Information System (PROMIS) pediatric measures to assess health-related quality of life of patients with pediatric lymphoma in Malawi. Pediatr Blood Cancer 65(11):e27353. https://doi.org/10.1002/pbc.27353
Atallah-Yunes SA, Habermann TM, Khurana A (2024) Targeted therapy in Burkitt lymphoma: Small molecule inhibitors under investigation. Br J Haematol 204(6):2165–2172. https://doi.org/10.1111/bjh.19425
Schmit N, Kaur J, Aglago EK (2024) Mosquito Bed Net Use and Burkitt Lymphoma Incidence in Sub-Saharan Africa: A Systematic Review and Meta-Analysis. JAMA Netw Open 7(4):e247351. https://doi.org/10.1001/jamanetworkopen.2024.7351
Samples L, Sadrzadeh H, Frigault MJ, Jacobson CA, Hamadani M, Gurumurthi A et al (2025) Outcomes among adult recipients of CAR T-cell therapy for Burkitt lymphoma. Blood 145(23):2762–2767. https://doi.org/10.1182/blood.2024026831

No competing interests reported.

SupplementaryFigure1.pdf
Supplementary Figure 1 The disease burden of Burkitt lymphoma by age group and sex in 2021
SupplementaryFigure2.pdf
Supplementary Figure 2 Temporal trends in age-standardized incidence rate (ASIR)，age-standardized death rate (ASDR) and age-standardized DALY rate from 1990 to 2035
SupplementaryFigure3.pdf
Supplementary Figure 3 Supplementary Figure 3. Trends in incidence, mortality, and DALY rate from 1990 to 2021, analyzed using joinpoint regression
SupplementaryTable1.docx
Supplementary Table 1 DALYs and temporal trends of Burkitt lymphoma in 1990 and 2021 at global, different SDI levels, and regions. Abbreviations: DALYs, disability-adjusted life years.
SupplementaryTable2.docx
Supplementary Table 2 The average QCI value of 204 countries worldwide.

Global assessment of the quality of care index for Burkitt Lymphoma from 1990 to 2021

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

1. Introduction

2. Methods

2.1 Data Sources and Collection

2.2 Time Trend Analysis

2.3 Quality of care index

2.4 SHAP Analysis

2.5 Bayesian Age-Period-Cohort (BAPC) Model for Forecasting

2.6 Statistical Analysis

3. Results

3.1 Global Burden of Burkitt Lymphoma

3.2 Regional Burden of Burkitt Lymphoma

3.3 Trends and Future Projections of the Disease Burden of Burkitt Lymphoma

3.4 Construction and verification of quality of care index

3.5 Analysis of global QCI distribution and influencing factors

3.6 Long-term evolution trend of QCI

3.7 QCI forecast trends from 2022 to 2035

4. Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1