The data used in this study were obtained from the GBD study. Epidemiological data on NHL in children and adolescents from 204 countries and regions worldwide from 1990 to 2021 were downloaded from the GBD website, including indicators such as crude prevalence rate, age-standardized prevalence rate (ASPR), and disability-adjusted life years (DALYs). Corresponding demographic data were also downloaded for the standardized calculation of prevalence rates. The main calculated indicators included: prevalence rate, ASPR, and age-standardized DALY rate (ASDR).

Descriptive analysis of disease burden: Descriptive analysis was performed on the prevalence rate and DALYs of NHL in children and adolescents (0-19 years) globally and in 21 regions from 1990 to 2021. The number of incident cases, crude prevalence rate, ASPR, number of DALYs, ASDR, and the percentage change of these indicators from 1990 to 2021 were calculated for each region and year. Disease burden heatmaps were plotted to show the distribution of NHL in children and adolescents worldwide.

Joinpoint regression analysis: The Joinpoint regression model was applied to analyze the trends in ASPRs of NHL in children and adolescents from 1990 to 2021. The standard errors of ASPRs were calculated using R software, and the data were imported into the Joinpoint software. Time was set as the independent variable, age-standardized rate as the dependent variable, region as the grouping variable, and a maximum of 5 joinpoints were allowed with a 95% confidence interval for parameter testing. The results were expressed as annual percent change (APC).

Age-period-cohort analysis: To explore the temporal trends of NHL in children and adolescents, data from 1992 to 2021 were divided into 5-year periods after excluding data from 1990-1991, and uploaded to the APC website for age, period, and cohort effect analysis. The influence of each effect on the prevalence of NHL was calculated, and graphs of relative risk changes with age, period, and cohort were plotted. The local drifts percentages of prevalence risk in each age group compared to the previous age group were calculated.

Decomposition analysis: To comprehensively assess the driving forces behind changes in the prevalence rate NHL from 1990 to 2021, we applied Das Gupta’s decomposition method. This approach allowed us to quantify the contributions of three major components: population growth, changes in population age structure, and epidemiological changes. The analysis was stratified by global level, sex, and Socio-demographic Index (SDI) regions.

Epidemiological changes were defined as alterations in age-specific and population-standardized prevalence rates over time, after adjusting for demographic changes. Specifically, they reflect shifts in disease risk that are not attributable to population size or age distribution alone—such as improvements in diagnosis, treatment, or exposure to risk factors. For each year y, the total number of cases was calculated using the formula (12):

where a i , y is the proportion of the population in age group i, p y is the total population in year y, and e i , y is the age-specific prevalence rate. The decomposition isolates the effect of each factor by varying one component at a time while keeping the others constant, thereby minimizing the interaction effects among components.

While this method assumes that the three components operate independently, potential interactions (e.g., age structure affecting epidemiological rates) are partially addressed through sequential standardization. Nevertheless, it is important to interpret results in the context of these assumptions.

Frontier analysis: The data envelopment analysis (DEA) method was used to evaluate the relationship between the disease burden of NHL and the SDI. The free disposal hull (FDH) model was used to fit the nonlinear production frontier, and the locally weighted regression scatter plot smoothing (LOWESS) method was used to generate a smoothed frontier. Super efficiency points (points below the frontier boundary) were removed to avoid the influence of outliers.

Health inequality analysis: The slope index of inequality (SII) and concentration index (CI) of the ASPR of NHL were calculated. SII reflects the absolute inequality of NHL, while CI reflects the relative inequality. The SII and CI were calculated for 1990 and 2021, and scatter plots of SII and CI were drawn to compare the changes in health inequalities between the two time points.

Prevalence prediction analysis: The Bayesian Age-Period-Cohort (BAPC) model was used to predict the ASPR of NHL from 2022 to 2036. The BAPC model considers the age, period, and cohort effects, sets prior distributions, infers the three effects a posteriori by combining the sample data, and finally combines the three effects to predict the prevalence rate.

All data analyses and plotting were performed using R 4.2.0, and the BAPC model was implemented using the BAPC package. Statistical tests were considered statistically significant at P<0.05.

In 2021, the global number of incident cases of NHL in children and adolescents was 227,438.038 (95% uncertainty interval [UI]: 189,379.563-276,291.764), and the ASPR was 3.177/100,000 (95% UI: 2.634-3.874). In terms of prevalence number, among the 21 regions, South Asia had the highest number of incident cases in 2021, with 46,144.590 (95% UI: 34,410.725-62,856.516), while Oceania had the lowest number of incident cases in 2021, with 617.558 (95% UI: 346.587-1,046.539). In terms of ASPR, Western Europe had the highest ASPR, reaching 7.494/100,000 (95% UI: 6.275-8.985), while Central Sub-Saharan Africa had the lowest ASPR, at 1.233/100,000 (95%UI:0.721-1.832). The region with the largest annual percentage change increase in ASPR of NHL from 1990 to 2021 was Southern Sub-Saharan Africa, with an increase of 85.469%, while the region with the largest decrease was Central Sub-Saharan Africa, with a decrease of 38.817% ( Table 1 ).

The DALYs caused by NHL were 962,473.898 (95% UI: 769,887.139-1,207,937.694), and the ASDR was 13.535/100,000 (95% UI: 10.772-17.052). Among the other 21 regions, South Asia had the highest DALYs caused by NHL in 2021, at 229,744.212 (95% UI: 175,628.271-310,401.205), while Australasia had the lowest DALYs caused by NHL in 2021, at 727.547 (95% UI: 582.512-896.601). In terms of ASDR, the Eastern Sub-Saharan Africa region had the highest ASDR, at 33.526/100,000 (95% UI: 23.028-48.271), while High-income North America had the lowest ASDR, at 3.039/100,000 (95% UI: 2.819-3.267). The Southern Sub-Saharan Africa region had the largest annual percentage change increase in ASDR from 1990 to 2021, at 55.546%, while the East Asia region had the largest decrease, at 70.117% ( Table 2 ).

The results of Joinpoint regression analysis showed that during 1990 to 2021, the global ASPR of NHL exhibited an overall fluctuating downward trend, with five significant turning points detected in 1995, 1999, 2002, 2013, and 2019 ( Figure 1 ). Specifically, from 1990 to 1995, the global ASPR of NHL showed an upward trend, with an annual percent change (APC) of 1.05% (P<0.05). From 1996 to 1999, the prevalence rate turned to decline, with an APC of -0.77% (P<0.05). From 2000 to 2002, the prevalence rate continued to decline, and the magnitude of decline further increased compared to the previous period, with an APC of -2.28% (P<0.05). From 2003 to 2013, the prevalence rate showed an upward trend again, with an APC of 0.62% (P<0.05). From 2014 to 2018 and from 2019 to 2021, the prevalence rate turned to a downward trend again, with the rate of decline gradually accelerating, and the APCs were -0.77% (P<0.05) and -3.45% (P<0.05), respectively.

Further analysis revealed that the trends in the standardized prevalence rate of NHL in males from 1990 to 2021 were similar to those in the overall population, showing a fluctuating downward trend, while the prevalence rate in females showed an upward trend. For males, the APCs were 0.75% P<0.05), -0.91% (P<0.05), -2.19% (P<0.05), 0.38% (P<0.05), -1.24% (P<0.05), and -3.08% (P<0.05) for the periods of 1990-1994, 1995-1999, 2000-2002, 2003-2013, 2014-2017, and 2018-2021, respectively. For females, the APCs were 1.48% (P<0.05), -1.45% (P<0.05), 1.24% (P<0.05), 0.51% (P<0.05), -2.83% (P<0.05) for the periods of 1990-1996, 1997-2003, 2004-2011, 2012-2018, and 2019-2021, respectively.

There were significant age effects, period effects, and cohort effects on the prevalence risk of NHL. In terms of age effects, the prevalence risk of NHL gradually decreased with increasing age, reaching the lowest level at 12.5 years old (Rate=8.486/100,000) and slightly increasing to 8.761/100,000 at 17.5 years old ( Figure 2A ). This suggests that younger children may be more susceptible, with 12.5 years old being a critical turning point for prevalence risk, and there may be a slight rebound in late adolescence (17.5 years old).

The period effect analysis found that the prevalence risk of NHL showed an overall decreasing and then increasing trend from 1994 to 2019. Specifically, it showed a rapid decline from 1994 to 2004, followed by a slow increase after 2004, but the prevalence risk in 2019 was still lower than that in 1994 ( Figure 2B ). Taking 2004 as the reference, the relative risk (Rate Ratio, RR) of NHL in 1994 was 1.128, and it decreased to 1.082 by 2019. This indicates that the impact of period effects on the risk of NHL showed an overall weakening trend, and although there was a slight increase in recent years, it did not exceed the historical high point.

The cohort effect analysis results showed that there were significant differences in the prevalence risk of NHL among children an d adolescents in different birth cohorts, indicating a clear cohort effect ( Figure 2C ). Taking the 1992-1997 birth cohort as the reference (RR=1), the RR of the 1972-1977, 1977-1982, 1982-1987, and 1987-1992 birth cohorts increased from 0.880 to 1.00. After 1997, the RR continued to decrease, reaching 0.802 in the 2012-2017 birth cohort. This suggests that there are generational differences in the decline of prevalence risk of NHL born in different eras.

Taking the birth cohort of 1992-1997 as the reference (RR = 1), the RR of the birth cohorts of 1972-1977, 1977-1982, 1982-1987, 1987-1992, and 1992-1997 increased from 0.880 to 1.00. After 1997, the RR continued to decline, and by the birth cohort of 2012-2017, the RR decreased to 0.802. This suggests that there are intergenerational differences in the decline of the risk of NHL among children and adolescents in different birth years, and such differences may be related to generational characteristics such as improvements in lifestyle and changes in environmental factors.

Further analysis of the local drift percentages of different age groups, which refers to the percentage change in disease risk for each age group compared to the previous age group. The results showed significant differences in the local drift percentage between different age groups. The local drift percentage in the age group of 2.5 to 7.5 years was negative, indicating a declining trend in the incidence of NHL. This decline gradually slowed with increasing age. However, after 12.5 years, the local drift percentage became positive and continued to rise, suggesting a rebound in disease risk after puberty ( Figure 2D ).

To investigate the epidemiological characteristics of NHL among children under five years old from 1990 to 2021, we conducted an in-depth analysis using data for the 12-23 months age group, which was the only available age group in the GBD database. The results demonstrated a fluctuating downward trend in NHL prevalence among children in this age group during the study period ( Figure 3 ), decreasing from 20.25/100,000 in 1990 to 15.86/100,000 in 2021, representing a reduction of 21.68%.

Gender-stratified analysis revealed that while both boys and girls followed the overall declining trend, significant gender disparities persisted. Specifically, the prevalence among boys decreased from 24.81/100,000 in 1990 to 17.79/100,000 in 2021, showing a reduction of 28.30%. For girls, the prevalence declined from 15.40/100,000 to 13.80/100,000, with a reduction of 10.37%. Notably, throughout the entire study period, the prevalence among boys consistently remained significantly higher than both the overall average and the prevalence among girls, and although the gender gap narrowed, it continued to persist.

Globally, population growth was the primary factor driving the increase in the prevalence rate of NHL, contributing 179.03% of the prevalence rate change. In contrast, population aging (-8.03%) and epidemiological factors (-71.00%) had negative effects on the increase in prevalence rate ( Figure 4A ). This indicates that although population aging and disease prevention factors have reduced the prevalence risk to a certain extent, the expansion of population size is still the main reason for the increase in the number of incident cases.

By gender, the prevalence rate of NHL in male children and adolescents increased slightly, with the contribution of population growth as high as 3936.56%, while population aging (-171.50%) and epidemiological factors (-3665.05%) largely offset some of the growth. The trend of prevalence rate change in females was similar to that in the overall population, with the contributions of population growth, population aging, and epidemiological factors being 58.97%, -2.89%, and 43.92%, respectively ( Figure 4A ).

Further analysis by SDI showed that as the SDI level increased, the prevalence rate of NHL showed a downward trend, and the contribution of population aging to prevalence rate changes also varied ( Figure 4B ). Among them, the contribution of population aging was lowest in low SDI regions (-12.24%), followed by low-middle SDI regions (-11.08%). In low SDI and low-middle SDI regions, population factors were the most important factors driving the increase in prevalence rate, with contribution rates as high as 170.67% and 92.80%, respectively. In middle SDI regions, population factors had a major negative effect on the increase in prevalence rate (-71.53%), and in this region, the contribution of epidemiological factors to prevalence rate reached 180.6%, suggesting that changes in population structure in this region actually favored a decrease in disease prevalence, but possibly due to significant influences of epidemiological factors such as improved disease diagnostic capabilities, improved surveillance systems, changes in environmental factors, and changes in lifestyle, the prevalence rate ultimately increased. In high SDI and high-middle SDI regions, population factors were the most important factors for the decrease in prevalence rate, and the contribution increased with increasing SDI, which may be due to the continued decline in birth rates in high SDI regions and a more significant decrease in the population base of children and adolescents.

The results of DEA showed that the DALY rate of NHL was non-linearly negatively correlated with SDI. As SDI increased, the DALY rate of NHL showed an overall downward trend, but the rate of decline gradually slowed down ( Figure 5A ). Specifically, when SDI was low (<0.2), the DALY rate declined relatively rapidly; after SDI exceeded 0.2, the rate of decline in DALY rate slowed down; when SDI reached above 0.4, the DALY rate had basically stabilized at a lower level. This result indicates that improving SDI level, especially in socioeconomically underdeveloped regions, may help significantly reduce the health loss caused by NHL.

The DEA frontier line reflects the theoretically achievable lowest DALY rate at a given SDI level, providing an important reference for evaluating the efficiency of disease prevention and control in each country. For the inefficient points above the DEA frontier line, their potential for improvement in controlling the disease burden of NHL can be found through horizontal comparison. Taking the 2021 data as an example ( Figure 5B ), the actual DALY rates of some countries in the coordinate graph are higher than the lowest DALY rates shown by the frontier line at the same SDI level, indicating that these countries still have room for reduction in the disease burden of NHL under the current socioeconomic conditions. Among them, countries such as Malawi, Uganda, South Sudan, and Niue have significantly higher DALY rates than the frontier line. Measures including early diagnosis and improving healthcare accessibility are needed to curb rising inequality in these low-income countries.

The results of the health inequality analysis showed that the SII of NHL was 4.1 in 1990 and increased to 9.4 in 2021 ( Figure 6A ). The increase in the absolute value of SII indicates an exacerbation of the absolute inequality of NHL between high and low SDI regions. This means that the burden of NHL has become more concentrated in low SDI regions, while high SDI regions have a relatively lighter burden.

From the SII scatter plot, it can be seen that the slope index of the global ASPR of NHL from 1990 to 2021 shows an overall increasing trend year by year, and the fitted regression line indicates that the long-term trend of the slope index is tilted upward, i.e., the degree of absolute inequality is intensifying ( Figure 6B ).

The CI increased from 0.09 in 1990 to 0.15 in 2021 ( Figure 6C ), indicating that the degree of inequality in the distribution of health resources and disease burden is intensifying. The increase in CI is close to twofold, indicating that health inequalities have significantly worsened over the past 30 years. Both the CI in 1990 and 2021 were positive values, indicating that the prevalence of NHL were mainly concentrated in high SDI regions. The CI was closer to 0 in 1990, indicating that the distribution of prevalence among different SDI regions was more balanced in 1990. The increase in CI further corroborates the trend of exacerbation of relative inequality in NHL.

In summary, both SII and CI showed a clear upward trend, indicating that the distribution of prevalence rates of NHL among different SDI regions is becoming increasingly unbalanced, and health inequalities are becoming more prominent. In addition, the disease burden is mainly concentrated in high SDI regions, and its distribution inequality has significantly worsened compared to 1990.

The BAPC model predicted that the global ASPR of NHL will continue to decrease in the next 15 years ( Figure 6D ). Specifically, the predicted prevalence rate will decrease from 3.177/100,000 in 2021 to 2.048/100,000 in 2036, a decrease of 35.56%. This result indicates that on the basis of existing disease prevention and control measures and medical technology levels, the global burden of NHL is expected to be further controlled and reduced. It should be noted that although the prediction results show a downward trend in future prevalence rates, NHL remains an important disease that threatens the health of children and adolescents.

Comparing the global distribution maps of crude prevalence rates of NHL between 1990 and 2021, it can be found that the regional distribution pattern of NHL disease burden in children and adolescents has changed to some extent over the 30 years. In 1990, the crude prevalence rates of NHL were higher in regions such as the Republic of San Marino, Republic of Estonia, and Principality of Andorra ( Figure 7A ), while the crude prevalence rates were lower in regions such as the Republic of Mauritius, Republic of Kiribati, and Republic of Cabo Verde. In terms of the trend of ASPRs, the ASPRs of NHL in countries such as the Republic of San Marino, Republic of Estonia, and Principality of Andorra remained high in 1990 ( Figure 7C ), while the ASPRs in island countries such as the Republic of Mauritius, Republic of Kiribati, and Republic of Cabo Verde were at a lower level worldwide.

In 2021, in countries such as the Republic of Mozambique, Republic of Honduras, and Republic of Kiribati, which are mostly located in underdeveloped regions of Africa, Central America, or the central Pacific, the number of patients reported with NHL was relatively small ( Figure 7B ). From the global distribution map of ASPRs in 2021, the ASPRs in low-income or lower-middle-income countries such as the Republic of Mozambique, Republic of Honduras, and Republic of Kiribati were relatively low ( Figure 7D ).

Comprehensively comparing the results of 1990 and 2021, it can be found that the regional distribution pattern of NHL disease burden in children and adolescents has changed to some extent over the 30 years. Overall, the prevalence rates in some African, Central American, and Pacific island countries were relatively low. It is worth noting that the lower prevalence of NHL in less developed countries could be partially attributed to under-reporting or misclassification due to the limitations of healthcare infrastructure, diagnostic facilities, treatment capabilities, and reporting systems in these regions (13).

To further explore the relationship between SDI and the disease burden of NHL, this study conducted a correlation analysis of SDI and ASPRs in different regions and countries ( Figures 8A, B ). From a regional perspective, the ASPRs of NHL exhibited distinct wave-like patterns as SDI levels improved, characterized by an initial decline, followed by an increase, and subsequently another decrease ( Figure 8A ). This nonlinear relationship suggests that there is a complex interaction between SDI and the prevalence risk of NHL, which may be influenced by multiple factors. From a national perspective, the ASPRs of NHL in middle SDI, high-middle SDI, and high SDI countries were generally higher than the global average level and showed a continuous upward trend ( Figure 8B ). This result indicates that developed regions still face great challenges in the prevention and control of NHL.