Global, regional, and national tuberculosis burden associated with six suboptimal dietary exposures from 1990 to 2021: a comparative risk assessment based on the Global Burden of Disease Study 2021

Introduction

Tuberculosis (TB) remains one of the leading infectious causes of morbidity and death worldwide (1). According to the World Health Organization (WHO), an estimated 10.6 million people developed TB globally in 2022, with an incidence rate of 133 per 100,000 persons. Despite the substantial progress achieved in diagnosis and treatment, the global burden of TB remains considerable, particularly in low- and middle-income countries (2). An estimated 9.9 million people developed TB in 2020, causing 1.5 million deaths, and an estimated 465,000 people developed multidrug-resistant TB or rifampicin-resistant TB in 2019 (3). Certain social and biomedical factors, including human immunodeficiency virus (HIV) infection, undernutrition, diabetes, low body mass index (BMI), and smoking, have been conclusively determined to increase the risk of TB and are well-characterized (4,5). However, although evidence indicates that inadequate nutrition increases the likelihood of progression to active disease and is associated with poorer outcomes, suboptimal dietary exposures remain comparatively underquantified (6,7). Moreover, the contribution of suboptimal dietary exposures to TB-related disease burden has not been systematically quantified at the global or regional level.

Risk factor control has become central to reducing TB burden (8). In parallel, the paradigm of public health nutrition has shifted from a single-nutrient approach to one focused on dietary patterns that can better reflect real-world eating behaviors and immune-metabolic pathways (9). Although undernutrition has long been recognized as a key determinant of TB outcomes, the independent contributions of individual suboptimal dietary exposures to the development of TB—such as diets low in whole grains, fruits, or vegetables or one with a high consumption of sugar-sweetened beverages—remain unclear (10).

Over the past three decades, global dietary patterns have been reshaped by urbanization, globalization of food systems, income growth, agricultural transitions, and the rapid expansion of ultra-processed food markets. Many low- and middle-income countries now face a double burden of malnutrition, in which persistent undernutrition coexists with increased consumption of energy-dense and micronutrient-poor foods, including sugar-sweetened beverages and processed products (11). These shifts may modify TB vulnerability through interacting pathways involving food insecurity, impaired immune function, metabolic dysregulation, and unequal access to health services. Long-term and cross-national analyses are therefore needed to quantify how specific suboptimal dietary exposures are associated with TB burden under a standardized comparative risk assessment (CRA) framework. The Global Burden of Disease (GBD) Study 2021 (GBD 2021) provides a CRA framework that enables systematic estimation of disease burden associated with modifiable suboptimal dietary exposures, offering an opportunity to quantify the contribution of dietary risks to TB burden (12). Leveraging this framework would allow for a consistent quantification of TB burden associated with predefined suboptimal dietary exposures and facilitate comparability across temporal, geographic, sex, and age strata (13).

Unlike broad indicators such as undernutrition or low BMI, the specific suboptimal dietary exposures defined in the GBD study’s CRA framework allow for the attributable burden of distinct dietary components to be compared across populations and over time. Although the nutritional determinants of TB have long been recognized, the global burden associated with specific suboptimal dietary exposures has not been comprehensively quantified according to the more recent GBD 2021 framework. Previous research in this field has largely focused on selected nutritional indicators or regional populations, limiting cross-national comparability and temporal assessment. Leveraging the standardized CRA methods of GBD 2021 across 204 countries and territories, we quantified TB disability-adjusted life years (DALYs) and deaths associated with six suboptimal dietary exposures from 1990 to 2021; assessed temporal trends using estimated annual percentage changes (EAPCs); examined geographic, sociodemographic, sex-specific, and age-specific disparities; and applied age-period-cohort (APC) models to disentangle age, period, and cohort effects. This integrated approach provides a globally comparable and policy-relevant assessment of how modelled dietary exposure-related burden is distributed across populations and time. We present this article in accordance with the GATHER reporting checklist (14) (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1788/rc).

Methods

Data sources and definition of suboptimal dietary exposures

We obtained data on the TB burden associated with suboptimal dietary exposures from the GBD 2021 database, which is publicly accessible through the Global Health Data Exchange (GHDx) platform. The GBD 2021 framework provides standardized estimates of disease burden for 204 countries and territories across 21 regions from 1990 to 2021. Data were accessed from the GBD Results Tool and GHDx for research purposes on June 18, 2025. All data used in this study were publicly available, deidentified, and aggregated at the population level (15).

The GBD 2021 estimates were generated from multiple data sources, including population censuses, household surveys, nutrition and dietary surveys, food availability data, vital registration systems, verbal autopsy data, disease surveillance records, hospital data, and published epidemiological studies. For TB outcomes, the GBD modelling framework integrated case notification data, prevalence surveys, cause-of-death data, and demographic estimates to produce internally consistent estimates of deaths, years of life lost (YLLs), years lived with disability (YLDs), and DALYs. For dietary exposures, GBD 2021 used standardized exposure modelling to estimate age-, sex-, location-, and year-specific intake distributions. Population denominators were derived from GBD demographic estimation and were used to calculate age-specific and age-standardized rates (ASRs) per 100,000 persons.

Data sparsity and heterogeneity across countries and years were addressed within the GBD modelling framework using statistical standardization, covariate adjustment, spatiotemporal smoothing, and uncertainty propagation. This approach was particularly relevant for low- and middle-income countries and earlier calendar years, when dietary surveillance, vital registration, and TB reporting systems were less complete. The 95% uncertainty interval (UI) reflected uncertainty from exposure estimation, relative risk estimation, population denominators, and disease burden modelling. The analysis ended in 2021 because GBD 2021 was the most recent complete GBD cycle available for the exposure-outcome combinations examined in this study at the time of data extraction. Although more recent TB notifications may be available from other surveillance platforms, they have not yet been fully harmonized within the GBD 2021 CRA framework for all six dietary exposures, locations, age groups, sexes, and calendar years.

The primary outcome measure was DALYs, a summary measure of disease burden calculated as the sum of YLLs due to premature mortality and YLDs. We also extracted deaths, age-standardized mortality rates (ASMRs), age-standardized DALY rates (ASDRs), and their corresponding 95% UIs for TB burden associated with suboptimal dietary exposures from 1990 to 2021. Demographic characteristics were stratified by sex (male and female), age (grouped in 5-year intervals: <5, 5–9, 10–14, 15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, 85–89, 90–94, and ≥95 years), calendar year [1990–2021], country or region, and sociodemographic index (SDI) quintile (high, high-middle, middle, low-middle, and low). SDI is a composite measure of fertility among women under 25 years of age, the mean educational attainment among individuals aged 15 years or older, and lag-distributed income per capita (16). SDI regions were categorized as high, high-middle, middle, low-middle, and low.

Using GBD 2021 data from 1990 to 2021, we assessed the modelled population-attributable TB burden associated with six predefined suboptimal dietary exposures within the GBD CRA framework: diets high in processed meat, high in red meat, high in sugar-sweetened beverages, low in fruits, low in vegetables, and low in whole grains. In this study, the term “burden associated with suboptimal dietary exposures” refers to the modelled population-attributable burden estimated within the GBD CRA framework. These estimates were derived from population-level dietary exposure distributions, theoretical minimum risk exposure levels, relative risks from published evidence, population denominators, and disease-specific burden estimates. Therefore, these estimates should be interpreted as population-level comparative risk estimates rather than evidence of a direct causal effect of specific dietary exposures on TB occurrence at the individual level.

This study used publicly available, deidentified, and aggregated population-level data. No individual-level information was used, and ethical approval was not required. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Statistical analysis

Using the GBD 2021 database, we summarized the global, regional, and national burden of TB associated with six suboptimal dietary exposures from 1990 to 2021. As defined within the GBD 2021 CRA framework, these risks included a diet high in processed meat, a diet high in red meat, a diet high in sugar-sweetened beverages, a diet low in fruit, a diet low in vegetables, and a diet low in whole grains. The ASMR and ASDR with corresponding 95% UIs were extracted as primary outcomes (17).

To evaluate long-term temporal patterns, the ASR was calculated as the weighted average of age-specific rates from a standard reference population. We assumed a log-linear relationship between ASR and calendar year and then fitted the regression models. The calculation of the ASR is as follows:

ASR=∑i=1nriwi∑i=1nwi×100,000[1]

where r_i is specific rate for the i age groups such as prevalence, incidence, or mortality rates; w_iis the standard population weight for the i age group; n is the number of age groups.

To estimate temporal trends, we assumed a log-linear relationship between the ASR and calendar year. The regression model is specified as follows:

Y=α+βX+ε[2]

where Y = ln(ASR), X is the calendar year, β is the estimated slope representing the temporal trend, and ε is the error term. The EAPC is calculated as follows:

EAPC=100×[exp(β)−1][3]

The 95% confidence intervals (CIs) for the EAPC were derived from the fitted regression model. A value greater than 0 for both the EAPC estimate and its lower CI bound indicated an increasing trend, whereas a value less than 0 for both the EAPC estimate and its upper CI bound indicated a decreasing trend; otherwise, the ASR was considered stable (18). ASR was log-transformed and modeled via ordinary least squares regression to estimate the EAPC. All statistical analyses and graphical representations were conducted with R statistical computing software version 4.3.3 (The R Foundation for Statistical Computing, Vienna, Austria). For all analyses, a two-sided P value <0.05 was deemed statistically significant.

We analyzed the spatial distribution of TB burden associated with suboptimal dietary exposures at the global, regional, and national levels. Heatmaps and choropleth maps were generated to visualize the heterogeneity of the ASDR, ASMR, and EAPC. Correlations between national ASR and the SDI were evaluated via Spearman rank correlation coefficients, and corresponding fitted curves were generated to assess the associations between TB burden associated with suboptimal dietary exposures and SDI levels.

APC model

The APC model was used to examine the independent effects of age, period, and birth cohort on the long-term trends in TB mortality and DALY rates associated with suboptimal dietary exposures from 1990 to 2021. The mathematical form of the model is as follows:

ln(Rijk)=ln(yijnij)=u+αi+βj+γk+ε[4]

where i, j, and k are the age group, period, and birth cohort, respectively; R_ijk is the TB mortality or DALY rate of the k-th cohort in the j-th period for the i-th age group; u is the intercept; α_i, β_j, and γ_k are the estimated effects of age, period, and cohort; and ε is the random error assumed to follow a normal distribution.

We applied the R-based APC analysis toolkit developed by the Institute for Health Metrics and Evaluation (IHME) (http://analysistools.cancer.gov/apc/) to implement the APC model. The APC analyses were conducted with the age-specific mortality rates and DALY rates associated with suboptimal dietary exposures, and DALY data associated with suboptimal dietary exposures were grouped into consecutive 5-year age bands and calendar periods. The following estimable functions were derived: net drift, which is the overall log-linear trend across the study period and represents the average annual percentage change in TB mortality and DALY rates; local drift, which is the age-specific annual percentage change and shows the temporal trend for each age group; longitudinal age effect, which is the fitted age-specific rates adjusted for cohort and period deviations and describes the intrinsic pattern of TB burden with age; period effect, which is the period rate ratio (RR) comparing each calendar period with the reference period and reflects the impact of health interventions, dietary transitions, and socioeconomic changes; and cohort effect, which is derived from a comparison of each birth cohort to the reference cohort and reflects the influence of different early-life exposures and generational dietary patterns. By default, the age, period, and cohort were set as reference categories in the APC toolkit. Wald Chi-squared tests were applied to evaluate the statistical significance of net drift, local drift, and deviation functions (19,20). All statistical tests were two-sided, with a P value <0.05 being considered to indicate a statistically significant difference.

Results

Global level

In 2021, dietary risks contributed substantially to the global burden of TB. Among the six dietary factors examined, a diet low in whole grains accounted for the highest burden, while a diet high in sugar-sweetened beverages accounted for the lowest. Overall, most dietary risks showed a declining ASDR and ASMR between 1990 and 2021, although absolute numbers of DALYs or deaths varied across risk categories. A diet high in processed meat was associated with a declining burden. DALYs fell slightly, from 82,125 cases (95% UI: 19,092 to 144,741) in 1990 to 78,892 cases (95% UI: 18,799 to 136,565) in 2021. The ASDR declined from 1.923 per 100,000 persons (95% UI: 0.45 to 3.38) to 0.915 per 100,000 persons (95% UI: 0.22 to 1.58), with an EAPC of −2.88% (95% CI: −3.21% to −2.55%). Deaths decreased from 2,562 cases (95% UI: 612 to 4,475) to 2,362 (95% UI: 571 to 4,104), and the ASMR decreased from 0.06 per 100,000 persons (95% UI: 0.01 to 0.11) to 0.03 per 100,000 persons (95% UI: 0.01 to 0.05), with an EAPC of −3.14% (95% CI: −3.44% to −2.84%) (Table 1 and Figure 1). A diet high in red meat exhibited a modest increase in absolute DALYs, rising by 1.74% from 60,598 cases (95% UI: 0 to 135,334) in 1990 to 61,655 cases (95% UI: 0 to 136,165) in 2021. In contrast, the ASDR decreased from 1.43 per 100,000 persons (95% UI: 0 to 3.18) to 0.71 per 100,000 persons (95% UI: 0 to 1.58), with an EAPC of −2.35% (95% CI: −2.48% to −2.22%). Deaths decreased slightly from 1,952 (95% UI: 0 to 4,317) to 1,854 (95% UI: 0 to 4,078), with the ASMR falling from 0.05 per 100,000 persons (95% UI: 0 to 0.11) to 0.02 per 100,000 persons (95% UI: 0 to 0.05) and an EAPC of −2.77% (95% CI: −2.90% to −2.64%) (Table 1 and Figure 1). A diet high in sugar–sweetened beverages showed a distinctive pattern. Although absolute DALYs increased markedly by 51.9%, from 25,391 cases (95% UI: 12,097 to 39,381) in 1990 to 38,576 cases (95% UI: 18,809 to 59,787) in 2021, and deaths rose by 44.5%, from 739 (95% UI: 355 to 1,133) to 1,068 (95% UI: 512 to 1,644), the standardized rates declined. ASDR dropped from 0.58 per 100,000 persons (95% UI: 0.28 to 0.90) to 0.45 per 100,000 persons (95% UI: 0.22 to 0.70), with an EAPC of −0.94% (95% CI: −1.02% to −0.86%), and ASMR decreased slightly from 0.02 per 100,000 persons (95% UI: 0.01 to 0.03) to 0.012 per 100,000 persons (95% UI: 0.005 to 0.019), with an EAPC of −1.33% (95% CI: −1.41% to −1.24%) (Table 1 and Figure 1). A diet low in fruits also imposed a considerable burden, but this declined over time. DALYs declined from 186,611 (95% UI: 25,894 to 349,586) in 1990 to 164,027 (95% UI: 21,585 to 316,201) in 2021, with an ASDR EAPC of –2.86% (95% CI: –3.01% to –2.71%). Deaths decreased from 6,022 (95% UI: 825 to 11,285) to 5,090 cases (95% UI: 680 to 9,876), and ASMR fell accordingly, with an EAPC of −3.16% (95% CI: −3.32% to −2.99%) (Table 1 and Figure 1). A diet low in vegetables exhibited the steepest decline in standardized rates. DALYs decreased from 50,044 (95% UI: −11,657 to 109,555) in 1990 to 42,728 (95% UI: −10,342 to 91,122) in 2021. The ASDR also decreased, with an EAPC of −2.94% (95% CI: −3.05% to −2.84%). Deaths fell from 1,644 (95% UI: −385 to 3,564) to 1,350 (95% UI: −325 to 2,858), while the ASMR decreased more sharply, with an EAPC of −3.24% (95% CI: −3.36% to −3.12%). The 95% UI for a diet low in vegetables crossed 0, indicating substantial uncertainty and lack of statistical significance (Table 1 and Figure 1). A diet low in whole grains accounted for the greatest burden. Global DALYs decreased from 192,932 (95% UI: 47,457 to 355,523) in 1990 to 177,304 (95% UI: 43,304 to 320,414) in 2021, corresponding to an ASDR decline from 4.56 per 100,000 persons (95% UI: 1.13 to 8.37) to 2.05 per 100,000 persons (95% UI: 0.50 to 3.72), with an EAPC of −2.73% (95% CI: −2.88% to −2.58%). Meanwhile, deaths decreased from 6,288 (95% UI: 1,567 to 11,446) to 5,539 (95% UI: 1,327 to 10,137), and the ASMR fell from 0.16 per 100,000 persons (95% UI: 0.04 to 0.29) to 0.06 per 100,000 persons (95% UI: 0.02 to 0.12), with an EAPC of −3.04% (95% CI: −3.20% to −2.88%) (Table 1 and Figure 1). Taken together, these findings indicate consistent downward trends in the ASDR and ASMR for all six suboptimal dietary exposures from 1990 to 2021, with a diet low in whole grains remaining the dominant contributor, and a diet high in sugar-sweetened beverages showing a rising absolute burden despite declining standardized rates.

Figure 1 ASDR and ASMR of TB associated with suboptimal dietary exposures globally and by SDI quintiles from 1990 to 2021. (A) ASDR for six suboptimal dietary exposures. (B) ASMR for six suboptimal dietary exposures. ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; DALY, disability-adjusted life year; SDI, sociodemographic index; TB, tuberculosis.

Regional level

From 1990 to 2021, the regional burden of TB associated with suboptimal dietary exposures varied substantially across different SDI strata and geographic regions. Overall, the ASDR and ASMR were consistently highest in low-SDI regions and lowest in high-SDI regions for most suboptimal dietary exposures, with the notable exception of a diet high in sugar-sweetened beverages. For a diet high in processed meat, the ASDR in 2021 reached 3.72 per 100,000 persons (95% UI: 0.83 to 6.54) in low-SDI regions compared with 0.20 per 100,000 persons (95% UI: 0.05 to 0.36) in high-SDI regions, and the corresponding ASMR was 0.13 per 100,000 persons (95% UI: 0.03 to 0.22) and 0.007 per 100,000 persons (95% UI: 0.002 to 0.012), respectively (Table 1 and Figure 1). Similar gradients were observed for a diet high in red meat: the ASDR was 2.03 per 100,000 persons (95% UI: 0 to 4.62) in low-SDI regions and 0.13 per 100,000 persons (95% UI: 0 to 0.29) in high-SDI regions, and the ASMR was 0.07 per 100,000 persons (95% UI: 0 to 0.16) and 0.005 per 100,000 persons (95% UI: 0 to 0.010) in 2021, respectively (Table 1 and Figure 1). A diet low in fruits and vegetables was also strongly associated with excess TB burden in low-SDI regions. In 2021, the ASDR associated with a diet low in fruits was 6.78 per 100,000 persons (95% UI: 0.89 to 12.83) in low-SDI regions and 0.086 per 100,000 persons (95% UI: 0.012 to 0.161) in high-SDI regions, with ASMRs of 0.25 per 100,000 persons (95% UI: 0.03 to 0.48) and 0.004 per 100,000 persons (95% UI: 0.001 to 0.006), respectively (Table 1 and Figure 1). For a diet low in vegetables, the ASDR was 2.15 per 100,000 persons (95% UI: −0.49 to 4.68) in low-SDI regions and 0.015 per 100,000 persons (95% UI: −0.003 to 0.036) in high-SDI regions, with ASMRs of 0.08 per 100,000 persons (95% UI: −0.02 to 0.18) and 0.0006 per 100,000 persons (95% UI: −0.0001 to 0.0014), respectively (Table 1 and Figure 1). A diet low in whole grains consistently conferred the highest burden among all suboptimal dietary exposures. In 2021, its ASDR reached 7.55 per 100,000 persons (95% UI: 1.84 to 14.03) in low-SDI regions and 0.14 per 100,000 persons (95% UI: 0.03 to 0.26) in high-SDI regions, with ASMRs of 0.28 per 100,000 persons (95% UI: 0.07 to 0.52) and 0.005 per 100,000 persons (95% UI: 0.001 to 0.010), respectively (Table 1 and Figure 1). By contrast, the burden associated with a diet high in sugar-sweetened beverages showed a distinct pattern, with the greatest impact observed in low-middle SDI regions, with an ASDR of 0.96 per 100,000 persons (95% UI: 0.46 to 1.49) and an ASMR of 0.03 per 100,000 persons (95% UI: 0.01 to 0.05); meanwhile, the lowest rates were recorded in high-SDI regions, with an ASDR of 0.12 per 100,000 persons (95% UI: 0.054 to 0.195) and an ASMR of 0.003 per 100,000 persons (95% UI: 0.002 to 0.005) (Table 1 and Figure 1).

ASDR trends, expressed as EAPCs, exhibited the steepest declines in regions with a high-SDI or high-middle SDI, reflecting the advanced stages of epidemiological transition. For instance, the ASDR associated with a diet high in processed meat decreased most in high-middle-SDI regions, with an EAPC of −6.62% (95% CI: −7.81% to −5.42%), while the smallest decline was observed in low-middle-SDI regions, with an EAPC of −2.18% (95% CI: −2.36% to −2.00%). Similar gradients were observed for a diet high in red meat, a diet low in fruits, and a diet low in whole grains. In contrast, the burden associated with a diet high in sugar-sweetened beverages declined most in low-SDI regions, with an EAPC of −2.10% (95% CI: −2.19% to −2.02%), and least in high-middle-SDI regions, with an EAPC of −0.93% (95% CI: −1.15% to −0.72%) (Table 1).

Our finding indicated that geographic disparities were also evident. Central Sub-Saharan Africa consistently exhibited the highest ASDR and ASMR across all suboptimal dietary exposures, whereas Australasia, East Asia, Western Europe, and high-income North America had the lowest values. Although ASDRs and ASMRs declined across nearly all 21 GBD regions between 1990 and 2021, the absolute burden remained disproportionately concentrated in Sub-Saharan Africa despite notable reductions over time (table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-1.docx).

National level

The burden of TB associated with suboptimal dietary exposures in 2021 demonstrated substantial variation across 204 countries and territories. For a diet high in processed meat, the ASDR ranged from 0.01 per 100,000 persons in Bermuda to 123.00 per 100,000 persons in the Central African Republic, with the Democratic Republic of the Congo (11.45 per 100,000 persons) and Pakistan (10.51 per 100,000 persons) also ranking high. Correspondingly, the ASMR was highest in the Central African Republic (0.72 per 100,000 persons), followed by Mali (0.35 per 100,000 persons), whereas Bermuda, Jamaica, and Antigua and Barbuda had the lowest ASMR (all <0.001 per 100,000 persons) (Figure 2 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx). For a diet high in red meat, the highest ASDR was observed in the Central African Republic (45.73 per 100,000 persons), the Marshall Islands (15.45 per 100,000 persons), and Kiribati (13.04 per 100,000 persons), while Jamaica and Malta had the lowest, with <0.02 per 100,000 persons. The Central African Republic ranked first in the ASMR (1.43 per 100,000 persons), followed by the Marshall Islands (0.50 per 100,000 persons) and Kiribati (0.42 per 100,000 persons) (Figure 2 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx).

Figure 2 The global distribution of TB burden associated with six suboptimal dietary exposures in 204 countries and territories. (A) ASDR. (B) ASMR. ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; ASR, age-standardized rate; DALY, disability-adjusted life year; TB, tuberculosis.

For a diet high in sugar-sweetened beverages, Kiribati (7.01 per 100,000 persons), Lesotho (6.56 per 100,000 persons), and the Marshall Islands (5.58 per 100,000 persons) had the highest ASDR, whereas Andorra and Antigua and Barbuda had the lowest (<0.01 per 100,000 persons). The ASMR ranged from 0.00005 per 100,000 persons in Bermuda to 0.21 per 100,000 persons in Kiribati, with Lesotho also showing relatively high values (0.20 per 100,000 persons) (Figure 2 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx). For a diet low in fruit, the highest ASDR was found for the Central African Republic (41.80 per 100,000 persons), Kiribati (24.40 per 100,000 persons), and the Democratic Republic of the Congo (21.05 per 100,000 persons), while Andorra and Israel had the lowest ASDR (<0.01 per 100,000 persons). The ASMR was 1.39 per 100,000 persons in the Central African Republic, 0.82 per 100,000 persons in Kiribati, and 0.71 per 100,000 persons in Somalia, while that of Bermuda was low, at 0.0002 per 100,000 persons (Figure 2 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx). For a diet low in vegetables, the ASDR ranged from 0.0002 per 100,000 persons in Armenia to 15.24 per 100,000 persons in the Central African Republic, with Kiribati (9.19 per 100,000 persons) and the Marshall Islands (7.64 per 100,000 persons) also ranking high. The Central African Republic had the highest ASMR (0.51 per 100,000 persons), followed by Kiribati (0.32 per 100,000 persons) and the Marshall Islands (0.26 per 100,000 persons). Of note, China had a relatively low ASDR and ASMR for this risk factor compared with many other countries (<0.01 per 100,000 persons) (Figure 2 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx). For a diet low in whole grains, the highest ASDR was found for the Central African Republic (48.52 per 100,000 persons), Kiribati (34.78 per 100,000 persons), and the Marshall Islands (27.06 per 100,000 persons), whereas Andorra and Malta had the lowest (<0.02 per 100,000 persons). The Central African Republic had the highest ASMR at 1.62 per 100,000 persons, followed by Kiribati (1.19 per 100,000 persons) and the Marshall Islands (0.93 per 100,000 persons) (Figure 2 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx).

From 1990 to 2021, there was marked heterogeneity in trends across 204 countries and territories according to the EAPC. Russia exhibited substantial declines in both ASDR and ASMR for a diet high in processed meat (ASDR EAPC: −7.46, 95% CI: −9.14 to −5.75; ASMR EAPC: −7.33, 95% CI: −8.94 to −5.68), a diet high in red meat (ASDR EAPC: −3.39, 95% CI: −4.74 to −2.03; ASMR EAPC: −3.56, 95% CI: −4.84 to −2.27), a diet low in fruit (ASDR EAPC: −2.23, 95% CI: −3.64 to −0.8; ASMR EAPC: −2.42, 95% CI: −3.74 to −1.09), and a diet low in whole grains (ASDR EAPC: −1.86, 95% CI: −3.26 to −0.44; ASMR EAPC: −2.02, 95% CI: −3.33 to −0.7); however, it showed an increasing EAPC for a diet high in sugar-sweetened beverages (ASDR EAPC: 0.6, 95% CI: −0.61 to 1.82; ASMR EAPC: 0.39, 95% CI: −0.74 to 1.54) and a diet low in vegetables (ASDR EAPC: 0.05%, 95% CI: −1.01 to 1.12; ASMR EAPC: −0.17, 95% CI: −1.14 to 0.81) (Figure 3 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx). Several African countries, including Zimbabwe and Lesotho, exhibited an increase in EAPC for both DALYs and deaths across multiple dietary risks, indicating worsening burdens. By contrast, the Maldives, Spain, Japan, and Taiwan (province of China) demonstrated consistent and pronounced declines, particularly for ASMR. In addition, countries such as Guatemala also reported substantial reductions (Figure 3 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx). These findings highlight wide disparities, with some high-burden countries continuing to face escalating risks and others achieving substantial reductions in TB burden associated with suboptimal dietary exposures (Figure 3 and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-2.docx).

Figure 3 EAPC in TB burden associated with six suboptimal dietary exposures in 204 countries and territories. (A) ASDR. (B) ASMR. ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; DALY, disability-adjusted life year; EAPC, estimated annual percentage change; TB, tuberculosis.

Global burden of TB associated with six suboptimal dietary exposures by age and sex

In 2021, the ASDR of TB associated with the six suboptimal dietary exposures was higher in the 25- to 29-year age group and increased with advancing age. Among men, DALY rates reached the highest levels in the 90- to 94-year age group for a diet high in red meat and similarly peaked in the 90- to 94-year age group for other dietary risks. For diets high in processed meat, red meat, and sugar-sweetened beverages, respectively, the DALYs were highest in the 50- to 54-year age group among both men and women. By contrast, for diets low in fruit, vegetables, and whole grains, respectively, the DALYs were highest in the 55- to 59-year age group among men and the 50- to 54-year age group among women. Overall, both DALY counts and ASDRs were consistently higher in men than in women. The largest burden was observed for a diet low in whole grains, with 15,014.64 DALYs (95% UI: 3,658.9 to 29,375.3) in men and 7,690.41 DALYs (95% UI: 1,755.87 to 14,944.35) in women (Figure 4A and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-3.docx).

Figure 4 Burden of TB associated with six suboptimal dietary exposures by age and sex in 2021. (A) Number of DALYs and DALY rates per 100,000 population. (B) Number of deaths and mortality rates per 100,000 population. Bars represent counts, lines represent rates, and error bars and shaded areas indicate 95% uncertainty intervals. DALY, disability-adjusted life year; TB, tuberculosis.

The highest ASMRs of TB associated with suboptimal dietary exposures were observed in older age groups, with rates increasing from age 40 years onward. Among men, mortality rates associated with diets high in processed meat, red meat, and sugar-sweetened beverages, respectively, reached their highest at ages 95 years and older, whereas mortality associated with a diet low in fruit, vegetables, and whole grains, respectively, peaked at ages 90–94 years. Meanwhile, for a diet high in processed meat, the absolute number of deaths was highest at ages 55–59 years among men and women. For a diet high in sugar-sweetened beverages, male deaths peaked at ages 55–59 years, whereas female deaths peaked at ages 50–54 years. For a diet high in red meat and diets low in fruit, vegetables, and whole grains, respectively, deaths peaked at 60–64 years among men; among women, deaths peaked at ages 65–69 years for diets low in fruits and low whole grains, respectively, and at ages 60–64 years for a diet low in vegetables. Overall, the number of deaths and mortality rate were consistently higher in men than in women. The greatest burden was observed for a diet low in whole grains, with 429.96 deaths (95% UI: 100.25 to 781.07) in men and 227.01 deaths (95% UI: 56.29 to 424.3) in women (Figure 4B and table available at https://cdn.amegroups.cn/static/public/jtd-2026-1788-3.docx).

Association with the SDI

At the regional level, we observed a significant negative correlation between the SDI and both the ASDR and ASMR of TB associated with suboptimal dietary exposures from 1990 to 2021. Overall, the ASDR and ASMR generally decreased with increasing SDI. From 1990 to 2021, several regions showed burdens higher than expected for their level of SDI. For example, Central and Southern Sub-Saharan Africa consistently had a higher ASDR than that predicted across all suboptimal dietary exposures. For a diet high in processed meat, Eastern Europe had a higher expected ASDR relative to its SDI level, whereas for a diet high in red meat, Oceania, East Asia, and Central Asia exhibited higher-than-expected burdens. For diets low in fruit, vegetables, and whole grains, South Asia, Southeast Asia, and the global aggregate all showed ASDR values exceeding those predicted based on SDI. By contrast, the burden of TB associated with suboptimal dietary exposures was lower than expected in Tropical Latin America, Eastern Sub-Saharan Africa, North Africa and the Middle East, Central Latin America, and Western Sub-Saharan Africa across the study period (Figure 5).

Figure 5 Association of SDI with the ASR of TB associated with six suboptimal dietary exposures for 21 GBD regions from 1990 to 2021. Thirty-two points are plotted for each region and show the observed ASR from 1990 to 2021 for the given region. Expected values, based on SDI and disease rates in all locations, are shown as a solid line. Regions above the solid line indicate a higher-than-expected ASR, whereas regions below the line indicate a lower-than-expected ASR. (A) ASDR. (B) ASMR. ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; ASR, age-standardized rate; DALY, disability-adjusted life year; GBD, Global Burden of Disease; SDI, sociodemographic index; TB, tuberculosis.

At the national level, a significant negative correlation was observed between the SDI and both the ASDR and ASMR for TB associated with suboptimal dietary exposures. For a diet high in sugar-sweetened beverages, the ASDR and ASMR showed an inverted U-shaped pattern, first increasing and then declining with higher levels of socioeconomic development. By contrast, for other suboptimal dietary exposures, both ASDR and ASMR decreased steadily as SDI increased. In 2021, several countries, particularly those in Africa, exhibited higher-than-expected burdens relative to their SDI, including the Central African Republic, the Democratic Republic of the Congo, and Mali. Conversely, countries such as Nicaragua, Sudan, and Yemen had lower-than-expected burdens in the 1990–2021 period (Figure 6).

Figure 6 Association of SDI with the ASR of TB associated with six suboptimal dietary exposures in 204 countries and territories from 1990 to 2021. Each point represents an observed value for a given country between 1990 and 2021. Expected values, based on SDI and disease rates across all countries, are shown as a solid line. Countries above the solid line indicate a higher-than-expected ASR, whereas those below the line indicate a lower-than-expected ASR. (A) ASDR. (B) ASMR. ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; ASR, age-standardized rate; DALY, disability-adjusted life year; SDI, sociodemographic index; TB, tuberculosis.

APC analysis for DALYs and mortality rate of TB

Globally, the net drifts were negative across all suboptimal dietary exposures for both DALY and mortality rates, indicating long-term declines in TB burden from 1990 to 2021. For DALYs, net drifts again indicated consistent decreases, with greatest reductions observed for a diet high in processed meat (−3.27%; 95% CI: −3.42% to −3.12%) and a diet low in vegetables (−3.08%; 95% CI: −3.12% to −3.00%), with the smallest reduction for a diet high in sugar-sweetened beverages (−1.22%, 95% CI: −1.34% to −1.10%). All DALY outcomes exhibited significant heterogeneity across age groups (all P values <0.05), indicating that age-specific declines deviated from the overall trend to varying degrees (Table 2, Figure 7A, and Figure S1A).

Figure 7 APC analysis of TB burden associated with a diet high in processed meat and a diet low in vegetables from 1990 to 2021. (A) Age, period, and cohort effects for DALY rates. (B) Age, period, and cohort effects for mortality rates. APC, age-period-cohort; DALY, disability-adjusted life year; TB, tuberculosis.

For deaths, the steepest decline was observed for a diet high in processed meat (−3.33% per year, 95% CI: −3.53% to −3.12%), followed by a diet low in vegetables (−3.14%, 95% CI: −3.41% to −2.86%) and a diet low in fruit (−3.10%, 95% CI: −3.25% to −2.96%). Declines were more modest for a diet high in red meat (−2.74%, 95% CI: −2.98% to −2.50%) and smallest for a diet high in sugar-sweetened beverages (−1.34%, 95% CI: −1.70% to −0.97%). Wald tests indicated variability in age for a diet high in processed meat, a diet high in red meat, a diet low in fruit, a diet low in vegetables, and a diet low in whole grains (all P values <0.05), whereas deaths associated with a diet high in sugar-sweetened beverages showed no significant discrepancy between the local drifts and the net drift (P>0.05) (Table 2, Figure 7B, and Figure S1B). The statistically significant negative net drift reflects a steady long-term decline in age-adjusted mortality associated with high sugar-sweetened beverage consumption; the absence of significant period and cohort deviations further indicates that this downward trajectory remained uniform, with no notable non-linear perturbations across eras or generational cohorts.

Discussion

In this systematic analysis of the GBD 2021, we quantified the TB burden associated with certain suboptimal dietary exposures across 204 countries and territories from 1990 to 2021. Globally, the ASDR associated with suboptimal dietary exposures declined consistently over the study period, with the EAPC ranging from −2.94% for a diet low in vegetables to −0.94% for a diet high in sugar-sweetened beverages per year across exposures. A similar declining pattern was observed for the ASMR, with all major suboptimal dietary exposures demonstrating a negative EAPC over three decades. Among individual suboptimal dietary exposures, the steepest annual reduction was observed for a diet low in vegetables, with an EAPC of −3.24%, whereas the slowest decline was noted for a diet high in sugar-sweetened beverages, with an EAPC of −1.33%. Despite sustained declines in the ASR, the absolute number of TB deaths associated with a diet high in sugar-sweetened beverages increased by approximately 44.5% between 1990 and 2021, and the corresponding DALYs increased by about 51.9%, reflecting the dominant impact of population growth and demographic aging. In 2021, ASMR associated with a diet low in fruit in low-SDI regions (0.25 per 100,000 persons) was approximately 62.5 times higher than that in high-SDI regions (0.004 per 100,000 persons). Although all SDI strata demonstrated declining trends, the annual reduction in high-SDI regions (EAPC =−4.85) was nearly twice as rapid as that observed in low-SDI settings (EAPC =−2.69), resulting in persistent and widening inequality in TB burden associated with suboptimal dietary exposures. The wide UI likely reflects the weakness of the epidemiological evidence linking vegetable intake to TB risk in the current CRA framework (21).

The overall decline in ASR is consistent with the global downward trend in TB incidence and mortality reported by the WHO, largely associated with expanded diagnostic capacity, improved treatment regimens, and intensified public health efforts under the WHO End TB Strategy (22). However, the persistence of substantial absolute burden associated with suboptimal dietary exposures suggests that structural determinants, especially undernutrition and suboptimal dietary patterns, continue to influence TB susceptibility and outcomes (23). Dietary inadequacy may contribute to TB pathogenesis through several mechanisms, such as impaired cell-mediated immunity—particularly in the context of micronutrient deficiencies—reduced macrophage and T-cell function—which compromises host defense against Mycobacterium tuberculosis—and systemic inflammation and metabolic dysregulation—which may influence disease progression (24,25). The coexistence of infectious and nutritional transitions in many low-SDI settings may partly explain why declines in the ASR have not translated into proportional reductions in absolute burden (26).

Marked inequalities across SDI strata suggest that the TB burden associated with suboptimal dietary exposures is not merely a matter of individual dietary choice but is embedded within structural determinants of health. Low-SDI regions often face concurrent food insecurity, limited dietary diversity, higher background TB transmission, delayed diagnosis, and constrained access to high-quality TB care. These factors may interact biologically and socially: inadequate intake of nutrient-dense foods may weaken cell-mediated immunity, whereas poverty, crowded living conditions, and health-system barriers may increase exposure to M. tuberculosis and delay treatment initiation. Therefore, the high burden observed in low-SDI settings likely reflects the combined effects of nutritional deprivation, health-system limitations, and broader socioeconomic inequities. In 2021, low-SDI regions had the highest ASDR and ASMR associated with multiple suboptimal dietary exposures. For example, the ASDR associated with a diet low in whole grains was 7.55 (95% UI: 1.84 to 14.03) per 100,000 persons in low-SDI regions but only 0.14 (95% UI: 0.03 to 0.26) per 100,000 persons in high-SDI regions, representing an approximately 54-fold disparity and corresponding 56-fold disparity in the ASMR. Similarly, for a diet low in fruits, the ASDR in low-SDI regions was 6.78 (95% UI: 0.89 to 12.83), while it was just 0.086 (95% UI: 0.012 to 0.161) in high-SDI regions, representing a nearly 79-fold difference. These magnitudes far exceed the typical interregional variation observed for many biomedical suboptimal dietary exposures and underscore the central role of structural food access, poverty, and nutritional deprivation in shaping TB vulnerability (27). The observed SDI gradient is biologically and epidemiologically plausible. Undernutrition has long been recognized as a major risk factor for TB incidence and mortality, impairing cell-mediated immunity, and heightened susceptibility to active disease progression (28). In settings characterized by food insecurity and limited dietary diversity, the interaction between chronic nutritional deficits and sustained community transmission may amplify TB burden beyond what would be expected from infection pressure alone (29).

Importantly, the SDI gradient persisted even for exposures commonly associated with dietary transition, such as processed and red meat intake. This suggests that TB vulnerability is shaped not only by caloric intake but by overall dietary quality and systemic inequities in food systems. A distinct pattern was observed for a diet high in sugar-sweetened beverages, with the highest burden occurring in low-middle–SDI regions rather than in the lowest SDI regions. This pattern is consistent with the nutrition transition, in which countries undergoing economic development experience rapid shifts toward energy-dense but micronutrient-poor diets before achieving full epidemiological transition (30). The relatively slow decline and increasing absolute burden associated with high sugar-sweetened beverage intake deserve particular attention. In many transitioning economies, urbanization and market expansion have accelerated the availability of inexpensive ultra-processed foods and sugar-sweetened beverages. This pattern may partly explain why the burden associated with sugar-sweetened beverages did not decline as rapidly as the burden associated with other dietary exposures. These conditions may generate a double burden of malnutrition, with persistent undernutrition coexisting with metabolically unfavorable dietary patterns, which may differentially influence TB susceptibility and outcomes (31).

Temporal trends further suggested unequal progress in the TB burden associated with suboptimal dietary exposures. Declines in ASR were generally steeper in high-SDI regions than in low-SDI regions. For example, for the ASDR of a diet low in fruit, the EAPC was −4.85 in high-SDI regions and −2.66 in low-SDI regions. This widening gradient implies that improvements in diet-related TB burden have proceeded more rapidly in settings with stronger health systems, social protection mechanisms, and food security infrastructure (32). From a policy perspective, these findings are directly relevant to the WHO End TB Strategy, which emphasizes prevention, patient-centered care, social protection, and action on social determinants (33). However, nutrition-sensitive interventions remain insufficiently operationalized in many TB control programs (34). Our results support embedding nutritional assessment, dietary counselling, and targeted food or cash support into TB care pathways, particularly in low-SDI and high-burden settings. Potential policy measures include subsidizing affordable whole grains, fruits, and vegetables; integrating nutrition support into community-based TB care; strengthening food security programs for TB-affected households; and aligning TB control with agricultural, antipoverty, and social protection policies (35). Ultimately, the burden associated with suboptimal dietary exposures appears not merely to be an individual behavioral issue but a manifestation of structural inequity operating through the social determinants of health.

The age-specific and sex-specific analyses revealed a clear demographic stratification in the TB burden associated with suboptimal dietary exposures. In 2021, DALY rates generally increased with advancing age, whereas DALY counts peaked in middle age. Mortality rates increased markedly from the 40- to 44-year age group onward, whereas death counts peaked in middle-to-older age groups. Among men, DALY rates peaked at 90–94 years for the six suboptimal dietary exposures, while the mortality rate associated with a diet high in processed meat, a diet high in red meat, and a diet high in sugar-sweetened beverages peaked at ages ≥95 years. In contrast, the DALYs and number of deaths were higher at a younger age, between 50 and 59 years. This difference between rate peaks and absolute burden peaks reflects population structure dynamics, as middle-aged groups account for the larger proportion of the population despite having a lower per-capita risk (36). Nutritional inadequacy may further exacerbate immune dysfunction, amplifying susceptibility to active TB and mortality in older adults (37). The progressive age gradient observed in our study is therefore consistent with life-course accumulation of dietary exposure and immunological vulnerability. Across all six suboptimal dietary exposures, DALY counts, DALY rates, death counts, and mortality rates were consistently higher in men than in women. For example, in the 55- to 59-year age group, DALYs associated with a diet low in whole grains reached 15,014.6 (95% UI: 3,658.9 to 29,375.3) in men but was only 7,612.62 (95% UI: 1,752.53 to 14,483.06) in women, while the number of deaths was 414.6 (95% UI: 101.38 to 822.62) and 208.33 (95% UI: 47.66 to 395.26), respectively. This approximately twofold higher burden among males aligns with global TB epidemiology, as men consistently exhibit higher incidence and mortality rates than do women, with possible explanations for this including higher baseline exposure to infection, behavioral cofactors such as smoking and alcohol use, differential health-seeking behavior, and sex-based immunological differences (38). Interestingly, the peak age for deaths differed slightly between sexes under certain dietary exposures. For instance, for diets low in fruits or whole grains, male deaths peaked earlier at 60–64 years, whereas female deaths peaked at 65–69 years, indicating a modest sex difference in the age distribution of TB mortality associated with suboptimal dietary exposures. Taken together, our findings demonstrate that TB burden associated with suboptimal dietary exposures is structured by a three-dimensional demographic gradient: rates increase sharply with advanced age, absolute numbers peak in middle age, and the prevalence is higher among men. Integrating age- and sex-sensitive nutritional strategies into TB control frameworks may therefore enhance both the equity and effectiveness of intervention programs, in line with the WHO End TB Strategy’s emphasis on addressing social determinants of health (39).

APC modelling further clarified the temporal dynamics underlying TB burden associated with suboptimal dietary exposures. Net drift estimates were uniformly negative across all dietary exposures (40). For DALYs, net drift ranged from −3.271 to −1.218, with the steepest annual decline observed for a diet high in processed meat and the slowest decline for a diet high in sugar-sweetened beverages. A similar pattern was observed for mortality, for which the net drift ranged from −3.325 to −1.335, again with the most pronounced reduction for a diet high in processed meat exposure and the most attenuated decline for a diet high in sugar-sweetened beverages. These consistently negative net drift estimates indicate a sustained log-linear annual decline in age-specific TB DALYs and mortality rates across calendar periods after adjustments for age, period, and cohort effects. However, the magnitude of decline varied substantially across exposures, suggesting variability in the contributions of dietary patterns to long-term TB burden reduction. For DALYs, local drift, period deviations, and cohort deviations revealed significant heterogeneity for the six suboptimal dietary exposures (P<0.05), indicating that age-specific, period-specific, and cohort-specific trends deviated from the overall net drift. In contrast, for deaths associated with a diet high in sugar-sweetened beverages, tests for local drift, period deviations, and cohort deviations were not statistically significant (P>0.05), suggesting a comparatively more uniform log-linear decline without pronounced age-specific, period-specific, or cohort-specific divergence.

From a population perspective, the observed temporal patterns may also reflect broader nutrition transitions occurring in many low-income and middle-income settings. The rapid shift from traditional diets to energy-dense but micronutrient-poor food environments has generated a double burden of malnutrition, characterized by the coexistence of undernutrition and diet-related metabolic disorders (41). These transitions may modify TB susceptibility both through persistent undernutrition and through the metabolic dysregulation associated with overnutrition. Our findings support the notion that dietary risks operate within a wider framework of social determinants of health, including poverty, food insecurity, and health system capacity (42).

The WHO End TB Strategy emphasizes social protection and management of comorbidities but does not explicitly prioritize nutritional risk quantification within routine TB surveillance systems (43). Our findings suggest that integrating dietary risk assessment into TB control frameworks could offer additional benefits. Previous GBD-based analyses have consistently identified undernutrition and related nutritional deficiencies as leading suboptimal dietary exposures for TB incidence and mortality worldwide. However, most of the earlier studies on this subject focused primarily on single exposures, such as low BMI, or were conducted on earlier GBD cycles, without incorporating the most recent post-2019 estimates (23). We have extended this work by leveraging the updated estimates from GBD 2021—incorporating a continuous time series through 2021—applying APC modeling, systematically evaluating multiple dietary risks within the CRA framework, and explicitly disentangling age, period, and cohort effects. This integrated methodological approach allows for a more refined interpretation of long-term temporal dynamics and age-specific vulnerabilities in TB burden associated with suboptimal dietary exposures, particularly in the context of demographic transition and evolving global nutrition patterns. This expansion to the methodology provides a more nuanced understanding of long-term trends and age-specific vulnerabilities in TB burden associated with suboptimal dietary exposures, particularly in the context of demographic transition and evolving global nutrition patterns.

The strengths of this study are as follows: First, we employed standardized and globally comparable estimates from GBD 2021, enabling consistent cross-national comparisons within the CRA framework (44). Second, the three-decade time span allowed for the evaluation of sustained temporal patterns, and APC modeling provided additional insight into age-specific and generational dynamics, with comprehensive geographic coverage further enhancing the generalizability of the study.

However, several limitations to the study should also be acknowledged. First, the GBD CRA framework estimates population-level attributable burden and does not establish individual-level causality. The estimated burden represents the expected reduction in TB deaths or DALYs under counterfactual exposure scenarios, assuming the validity of exposure distributions, theoretical minimum risk exposure levels, relative risks, and modelling assumptions. Therefore, our findings should not be interpreted as supporting individual-level clinical diagnosis or causal attribution of TB to any specific dietary exposure. Another important limitation is the potential disruption of TB diagnosis, care, and surveillance during the coronavirus disease 2019 (COVID-19) pandemic. Lockdowns, reallocation of healthcare resources, reduced access to diagnostic services, delayed care-seeking, and diagnostic backlogs may have altered TB case detection and reporting in 2020 and 2021. Therefore, estimates for the final years of the study period should be interpreted cautiously, as observed trends may reflect both true epidemiological changes and pandemic-related surveillance disruptions. Although the GBD modelling framework attempts to harmonize and correct data inconsistencies, residual bias related to COVID-19-related interruptions in TB services cannot be fully excluded.

As with all GBD analyses, the estimates were based on modeling assumptions and the uncertainty propagation inherent to the GBD framework and may be affected by data sparsity in some regions. Moreover, the ecological design precluded individual-level causal inference. Dietary exposure estimates may not capture within-country heterogeneity, and residual confounding by socioeconomic or comorbidity-related factors could not be excluded. In addition, population-attributable fractions assume independent risk effects, which may not fully reflect complex interactions between nutrition and other TB determinants (45).

From a public health nutrition perspective, our findings have important implications for policy making. The substantial contribution of suboptimal dietary exposures to the global TB burden highlights the need to integrate nutritional strategies into TB control programs, particularly in low- and middle-income countries with a high disease burden (46). Interventions targeting key suboptimal dietary exposures, such as improving protein intake, reducing micronutrient deficiencies, and promoting balanced dietary patterns, may offer a complementary approach to conventional TB prevention and treatment strategies.

Notably, the observed sociodemographic inequalities suggest that nutrition-related interventions should be tailored to regional contexts and prioritize populations in lower SDI levels, where both undernutrition and TB burden remain disproportionately high. Strengthening food systems, enhancing access to affordable nutritious foods, and integrating nutrition screening into TB care pathways could potentially reduce disease burden at the population level.

Future research should clarify the causal pathways linking dietary risks and TB by examining interactions with metabolic comorbidities and assessing subnational heterogeneity in high-burden settings. Evaluating the effectiveness of nutrition-focused interventions in improving TB outcomes is also essential. Integrating GBD-based population modeling with individual-level prospective cohort data would facilitate causal inference and support the development of targeted, evidence-informed nutritional strategies within TB control strategies. Future studies incorporating external validation with independent datasets or cohort data are warranted to further confirm the robustness of these findings.

Conclusions

Although the ASR of TB burden associated with suboptimal dietary exposures has declined globally since 1990, substantial absolute burden persists, particularly in low-SDI regions and older age groups. Suboptimal dietary exposures remain associated with a substantial, potentially modifiable population-level TB burden. Integrating nutrition-focused strategies into TB control frameworks may help accelerate progress toward global TB elimination targets.

Acknowledgments

We would like to thank the IHME and the authors of GBD 2021 for making the GBD estimates publicly available through the GHDx. We are also thankful to all data providers and institutions who contributed to the underlying data sources that informed the GBD 2021 CRA framework.

Footnote

Reporting Checklist: The authors have completed the GATHER reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1788/rc

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1788/prf

Funding: This work was supported by the Henan Provincial Medical Science and Technology Key Joint Project (No. SBGJ202301003).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1788/coif). All authors report that this work was supported by the Henan Provincial Medical Science and Technology Key Joint Project (No. SBGJ202301003). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study used publicly available, deidentified, and aggregated population-level data; no individual-level information was used; therefore, ethical approval was not required. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Franco JV, Bongaerts B, Metzendorf MI, et al. Undernutrition as a risk factor for tuberculosis disease. Cochrane Database Syst Rev 2024;6:CD015890. [Crossref] [PubMed]
2. Jiang F, Li X, Qiao Q, et al. Global, regional, and national burden of tuberculosis, 1990-2050: a systematic comparative analysis based on retrospective cross-sectional of GBD 2021 and WHO surveillance systems. Int J Surg 2026;112:250-69. [Crossref] [PubMed]
3. Dean AS, Tosas Auguet O, Glaziou P, et al. 25 years of surveillance of drug-resistant tuberculosis: achievements, challenges, and way forward. Lancet Infect Dis 2022;22:e191-6. [Crossref] [PubMed]
4. Naidoo A, Naidoo K, Padayatchi N, et al. Use of integrase inhibitors in HIV-associated tuberculosis in high-burden settings: implementation challenges and research gaps. Lancet HIV 2022;9:e130-8. [Crossref] [PubMed]
5. McQuaid CF, Clark RA, White RG, et al. Estimating the epidemiological and economic impact of providing nutritional care for tuberculosis-affected households across India: a modelling study. Lancet Glob Health 2025;13:e488-96. [Crossref] [PubMed]
6. Sinha P, Bhargava M, Carwile ME, et al. A roadmap for integrating nutritional assessment, counselling, and support into the care of people with tuberculosis. Lancet Glob Health 2025;13:e967-73. [Crossref] [PubMed]
7. Li H, Chee CBE, Geng T, et al. Joint Associations of Multiple Lifestyle Factors With Risk of Active Tuberculosis in the Population: The Singapore Chinese Health Study. Clin Infect Dis 2022;75:213-20. [Crossref] [PubMed]
8. Burden of 375 diseases and injuries, risk-attributable burden of 88 risk factors, and healthy life expectancy in 204 countries and territories, including 660 subnational locations, 1990-2023: a systematic analysis for the Global Burden of Disease Study 2023. Lancet 2025;406:1873-922.
9. Yang G, Du X, Wang J, et al. Unveiling the Roles of Immune Function and Inflammation in the Associations Between Dietary Patterns and Incident Type 2 Diabetes. J Am Nutr Assoc 2025;44:59-67. [Crossref] [PubMed]
10. Lu X, Lu X, Jiang M, et al. Global burden of tuberculosis attributable to diet low in whole grains from 1990 to 2021, with projection to 2045. Front Nutr 2025;12:1679569. [Crossref] [PubMed]
11. Hong T, Sun F, Wang Q, et al. Global burden of diabetes mellitus from 1990 to 2019 attributable to dietary factors: An analysis of the Global Burden of Disease Study 2019. Diabetes Obes Metab 2024;26:85-96. [Crossref] [PubMed]
12. Wang L, Wang H, Wang Y, et al. The global burden of tuberculosis attributable to diet high in processed meat from 1990 to 2021: findings from the Global Burden of Disease Study 2021. Front Nutr 2025;12:1666550. [Crossref] [PubMed]
13. Qiu L, Zhang Y, Yan K, et al. Global burden and temporal trends of tuberculosis attributable to high sugar-sweetened beverage consumption: insights from the Global Burden of Disease Study 2021. Front Nutr 2025;12:1638390. [Crossref] [PubMed]
14. Stevens GA, Alkema L, Black RE, et al. Guidelines for Accurate and Transparent Health Estimates Reporting: the GATHER statement. Lancet 2016;388:e19-23. [Crossref] [PubMed]
15. Global burden and strength of evidence for 88 risk factors in 204 countries and 811 subnational locations, 1990-2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet 2024;403:2162-203.
16. Jiang Q, Shu Y, Jiang Z, et al. Burdens of stomach and esophageal cancer from 1990 to 2019 and projection to 2030 in China: Findings from the 2019 Global Burden of Disease Study. J Glob Health 2024;14:04025. [Crossref] [PubMed]
17. Global burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020;396:1223-49.
18. Sha R, Kong XM, Li XY, et al. Global burden of breast cancer and attributable risk factors in 204 countries and territories, from 1990 to 2021: results from the Global Burden of Disease Study 2021. Biomark Res 2024;12:87. [Crossref] [PubMed]
19. Li H, Kong W, Liang Y, et al. Burden of osteoarthritis in China, 1990-2019: findings from the Global Burden of Disease Study 2019. Clin Rheumatol 2024;43:1189-97. [Crossref] [PubMed]
20. Hu X, Guo C. Temporal trends and cohort variations of gender-specific major depressive disorders incidence in China: analysis based on the age-period-cohort-interaction model. Gen Psychiatr 2024;37:e101479. [Crossref] [PubMed]
21. Xu X, Yan P, Chen W, et al. The global burden of disease attributable to suboptimal fruit and vegetable intake, 1990-2021: a systematic analysis of the global burden of disease study. BMC Med 2025;23:456. [Crossref] [PubMed]
22. Fukunaga R, Glaziou P, Harris JB, et al. Epidemiology of Tuberculosis and Progress Toward Meeting Global Targets - Worldwide, 2019. MMWR Morb Mortal Wkly Rep 2021;70:427-30. [Crossref] [PubMed]
23. Saunders MJ, Cegielski JP, Clark RA, et al. Body mass index and tuberculosis risk: an updated systematic literature review and dose-response meta-analysis. Int J Epidemiol 2025;54:dyaf154. [Crossref] [PubMed]
24. VanValkenburg A, Kaipilyawar V, Sarkar S, et al. Malnutrition leads to increased inflammation and expression of tuberculosis risk signatures in recently exposed household contacts of pulmonary tuberculosis. Front Immunol 2022;13:1011166. [Crossref] [PubMed]
25. David E, Zhu M, Bennett BC, et al. Undernutrition and Hypoleptinemia Modulate Alloimmunity and CMV-specific Viral Immunity in Transplantation. Transplantation 2021;105:2554-63. [Crossref] [PubMed]
26. Wagnew F, Alene KA, Kelly M, et al. Undernutrition increases the risk of unsuccessful treatment outcomes of patients with tuberculosis in Ethiopia: A multicenter retrospective cohort study. J Infect 2024;89:106175. [Crossref] [PubMed]
27. Abou Jaoude GJ, Garcia Baena I, Nguhiu P, et al. National tuberculosis spending efficiency and its associated factors in 121 low-income and middle-income countries, 2010-19: a data envelopment and stochastic frontier analysis. Lancet Glob Health 2022;10:e649-60. [Crossref] [PubMed]
28. Dodd PJ, Yuen CM, Jayasooriya SM, et al. Quantifying the global number of tuberculosis survivors: a modelling study. Lancet Infect Dis 2021;21:984-92. [Crossref] [PubMed]
29. Richterman A, Saintilien E, St-Cyr M, et al. Food Insecurity at Tuberculosis Treatment Initiation Is Associated With Clinical Outcomes in Rural Haiti: A Prospective Cohort Study. Clin Infect Dis 2024;79:534-41. [Crossref] [PubMed]
30. Baker P, Machado P, Santos T, et al. Ultra-processed foods and the nutrition transition: Global, regional and national trends, food systems transformations and political economy drivers. Obes Rev 2020;21:e13126. [Crossref] [PubMed]
31. Bhargava A, Bhargava M, Meher A, et al. Nutritional support for adult patients with microbiologically confirmed pulmonary tuberculosis: outcomes in a programmatic cohort nested within the RATIONS trial in Jharkhand, India. Lancet Glob Health 2023;11:e1402-11. [Crossref] [PubMed]
32. Richterman A, Steer-Massaro J, Jarolimova J, et al. Cash interventions to improve clinical outcomes for pulmonary tuberculosis: systematic review and meta-analysis. Bull World Health Organ 2018;96:471-83. [Crossref] [PubMed]
33. Reeve E, Thow AM, Huse O, et al. Policy-makers' perspectives on implementation of cross-sectoral nutrition policies, Western Pacific Region. Bull World Health Organ 2021;99:865-73. [Crossref] [PubMed]
34. Bhargava A, Bhargava M, Meher A, et al. Nutritional supplementation to prevent tuberculosis incidence in household contacts of patients with pulmonary tuberculosis in India (RATIONS): a field-based, open-label, cluster-randomised, controlled trial. Lancet 2023;402:627-40. [Crossref] [PubMed]
35. Yang Y, Cai J, Wang X, et al. Nutritional supplementation during tuberculosis treatment to improve clinical symptoms: a double-blinded placebo-controlled randomized trial. Food Funct 2025;16:102-11. [Crossref] [PubMed]
36. Global, regional, and national age-specific progress towards the 2020 milestones of the WHO End TB Strategy: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Infect Dis 2024;24:698-725.
37. Teo AKJ, Morishita F, Islam T, et al. Tuberculosis in older adults: challenges and best practices in the Western Pacific Region. Lancet Reg Health West Pac 2023;36:100770. [Crossref] [PubMed]
38. Rickman HM, Phiri MD, Feasey HRA, et al. Sex differences in the risk of Mycobacterium tuberculosis infection: a systematic review and meta-analysis of population-based immunoreactivity surveys. Lancet Public Health 2025;10:e588-98. [Crossref] [PubMed]
39. Auld SC, Barczak AK, Bishai W, et al. Pathogenesis of Post-Tuberculosis Lung Disease: Defining Knowledge Gaps and Research Priorities at the Second International Post-Tuberculosis Symposium. Am J Respir Crit Care Med 2024;210:979-93. [Crossref] [PubMed]
40. Zou Z, Liu G, Hay SI, et al. Time trends in tuberculosis mortality across the BRICS: an age-period-cohort analysis for the GBD 2019. EClinicalMedicine 2022;53:101646. [Crossref] [PubMed]
41. Muharram FR, Tjandra S, Madani NJ, et al. Trends in the double burden of malnutrition among Indonesian adults, 2007 to 2023. Sci Rep 2025;15:34883. [Crossref] [PubMed]
42. Wu CY, Ku CC, McQuaid CF, et al. Estimating the impact of nutritional transition and ending hunger on tuberculosis in 12 high-burden countries: a model-based scenario analysis. BMJ Glob Health 2025;10:e018839. [Crossref] [PubMed]
43. Ockenga J, Fuhse K, Chatterjee S, et al. Tuberculosis and malnutrition: The European perspective. Clin Nutr 2023;42:486-92. [Crossref] [PubMed]
44. Tan SCW, Zheng BB, Tang ML, et al. Global Burden of Cardiovascular Diseases and its Risk Factors, 1990-2021: A Systematic Analysis for the Global Burden of Disease Study 2021. QJM 2025;118:411-22. [Crossref] [PubMed]
45. 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 2024;403:2133-61.
46. Sinha P, Lönnroth K, Bhargava A, et al. Food for thought: addressing undernutrition to end tuberculosis. Lancet Infect Dis 2021;21:e318-25. [Crossref] [PubMed]

(English Language Editor: J. Gray)