The MDR-TB data were sourced from the GBD 2019 database (http://ghdx.healthdata.org/gbd-2019), a comprehensive, multinational research platform providing epidemiological insights into 204 territories or countries, encompassing all WHO member states. These countries or regions are grouped into 21 GBD regions based on geographic location, including Sub-Saharan Africa, Oceania, Asia, the Middle East, Europe, the Caribbean, Latin America, Australasia, High Income North America, and High Income Asia-Pacific (10).

The GBD 2019, covering 369 disorders and 87 associated risk factors (11), is an authoritative resource for comprehending global health challenges. The GBD employs a robust set of statistical models to estimate disease burden. Data sources include published literature, official national health department websites, WHO, demographic health surveys, and national and regional disease surveillance systems. To address variations in data quality, comparability, accuracy, and completeness, statistical methods like misclassification correction, rubbish code reassignment, and data noise reduction were applied to enhance data uniformity and comparability. Specifically, the analysis for tuberculosis included 3,613 distinct datasets.

GBD 2019 defined the following terms:

Multidrug-resistant tuberculosis without extensive drug resistance: the absence of extensive drug resistance, among individuals unaffected by HIV, represents a particular clinical scenario where the disease does not respond to the two most potent first-line antibiotics for treating tuberculosis. However, this form of TB does not exhibit resistance to any fluoroquinolone or any of the second-line injectable drugs.

Alcohol use: any alcohol consumption.

High fasting plasma glucose: a serum fasting plasma glucose level exceeding 4.8–5.4 mmol/L.

Smoking: currently engages in daily or periodic use of smoked tobacco products.

Disability-adjusted life years (DALYs): the total of years lost to premature death and years lived with disability.

The SDI, ranging from 0 to 1, is a composite average reflecting the per capita income, educational attainment, and birth rates across all regions included in the GBD study. The SDI quintiles, as depicted in Data 1, are categorized into five tiers: low, low-middle, middle, high-middle, and high, indicating progressively better socio-demographic conditions from the lowest to the highest tier.

The study conducted on GBD 2019 utilized the Comparative Risk Assessment framework to measure the disease burden caused by specific risk factors (11). To achieve this, the study followed several methodological steps. Initially, the study determined the relative risks (RR) as a measure of exposure to the risk factors. Subsequently, the Bayesian meta-regression model DisMod-MR 2.1 and the spatio-temporal Gaussian process regression model (ST-GPR) were utilized to approximate the mean exposure to each risk factor across all relevant data sources. Then, the Theoretical Minimum Risk Exposure Level (TMREL) was determined as the lowest exposure level to a risk factor observed in cohort studies and experimental trials. Moreover, the Population Attributable Fraction (PAF) was computed using age, sex, location, and time as variables, with the calculation method documented in earlier scholarly works (10). Finally, based on the data from these calculations, the estimates of disease caused by each risk factor were determined.

In this study, we extracted data on incidence, mortality, and DALYs related to MDR-TB from GBD 2019. We conducted a thorough analysis by organizing the data according to gender, age, etiology, geographical region, and SDI level. Additionally, we utilized Joinpoint software, the Age-Period-Cohort model, and the Nordpred tool (with specific methodologies detailed subsequently) to evaluate temporal trends in age-standardized DALYs and mortality rates.

As previously described in the literature (12), the age-period-cohort model (APCM) was used to analyze and project trends in incidence and mortality for MDR-TB. The input dataset comprised a granular distribution of 20 age brackets, ranging from birth to 95+ years, in increments of 5 years, and covered six distinct time frames, from 1990–1994 to 2015–2019. In addition, the dataset integrated 25 distinct birth cohorts, extending from 1890–1899 to 2010–2019 (Supplementary Table S1). In this study, control groups for ages 40, the 2000–2004 period, and the 1960 cohort. Based on the APCM, longitudinal age-specific rates were generated, which were adjusted for period bias using the reference cohort to reflect the natural history of age-related changes (i.e., the age effect). Simultaneously, the model yielded relative risks (RR) of incidence and mortality for each period (represented by different cohorts), capturing the period-specific influences on disease. The RR is defined as the ratio of the age-specific rate in each period (cohort) to the reference period (cohort). Furthermore, rooted in the APCM, we also employed the Nordpred tool in R to project MDR-TB incidence and mortality from 2020 to 2030.

To delineate the trajectory of MDR-TB burden, comparisons of annual percentage change (APC) values for age-standardized DALY rate (ASDR) and age-standardized mortality rate (ASMR) across varying periods were conducted. In this study, regression analyses were performed utilizing the Joinpoint tool, provided by the National Cancer Institute of the United States of America (13). The APC and average annual percentage change (AAPC) were employed as primary indicators to gauge the progression of MDR-TB. Additionally, AAPC values were scrutinized across gender, etiologies, SDI, and GBD regions from 1990 to 2019. An APC or AAPC value above zero signifies an upward trend, whereas a value below zero indicates a declining trend.

Statistical analysis was performed using R 4.3.2, Joinpoint Regression Program (version 4.9.0.0), and GraphPad Prism 9. The Wald χ² test was used for parameter estimation. Test level α = 0.05 (two-sided), P < 0.05 (two-tailed) was defined as significant.

From 1990 to 2019, the rate of DALYs attributable to MDR-TB has seen a startling escalation. The Joinpoint analysis exhibited an AAPC of 4.57 [95% confidence interval (CI): 3.89–5.26], signifying a substantial rise from 13.9 [95% uncertainty interval (UI): 4.9–32.5] per 100,000 individuals in 1990 to 52.4 (22.6–97.6) in 2019 (Table 1). The collective DALYs tallied for MDR-TB reached a staggering 4.2 million (1.8–7.8 million) by 2019 (Table 1). It is noteworthy that males were disproportionately affected compared to females. When regarding the SDI, the data revealed an inverse correlation between SDI levels and the number of MDR-TB DALYs. Over the past three decades, there has been a marked and persistent increase in DALYs from MDR-TB in low and low-middle SDI regions, with an AAPC exceeding 7% (Table 1).

Regionally, the greatest burden of MDR-TB fell on South-East Asia and Sub-Saharan Africa in 2019. South Asia alone accounted for 2.4 million DALYs (range: 0.6–5.5 million), constituting 56.4% of the global total. The following regions also had significant burdens: Eastern Sub-Saharan Africa at 13.7%, Western Sub-Saharan Africa at 8.0%, Southeast Asia at 4.6%, Central Sub-Saharan Africa and Eastern Europe at 4.0%, Southern Sub-Saharan Africa at 2.8%, and East Asia at 2.4%. Significantly, the top three GBD regions for ASR of DALYs were Eastern Sub-Saharan Africa at 204.2 (88.9–390.1) per 100 000 population, Central Sub-Saharan Africa at 177.9 (38.7–571.7), and Southern Sub-Saharan Africa at 157.7 (56.4–329.4) (Table 1). Within the ranks of the top three GBD regions grappling with a high MDR-TB burden, the ASDRs have continued to escalate. The AAPC was 10.43 (9.63–11.24) in South Asia, 9.92 (9.24–10.59) in Eastern Sub-Saharan Africa, and 5.49 (4.58–6.41) in Western Sub-Saharan Africa (Table 1).

Specifically, in 2019, India (South Asia), Russia (Eastern Europe), and China (Eastern Asia) led in the number of DALYs across 204 countries and territories. India topped the list with 220,000 cases, ahead of Russia with 31,000 and China with 30,000 cases (Figure 1A; Data 1, p. 15–19). Additionally, the top 10 countries by case count include South Asian nations: Pakistan (28,000 cases), Bangladesh (8,000 cases); Western sub-Saharan Africa's Nigeria (12,000 cases); Eastern European: Ukraine (12,000 cases); and South-East Asian: Philippines (9,000 cases), Viet Nam (6,000 cases), along with Eastern sub-Saharan Africa's Ethiopia (7,000 cases). In sub-Saharan Africa, Eswatini, Somalia, and Comoros exhibited the highest ASDRs of 33.9 (8.1–83.2), 30.4 (8.7–87.5), and 26.6 (3.4–87.0) per 100,000 population in the same period, respectively (Figure 1B; Data 1, p. 20–23). Additionally, the analysis of DALYs associated with MDR-TB exhibits a comparable pattern, as illustrated in Supplementary Figure S1 (Data 1, p. 87–95).

Alcohol was the primary cause of DALYs from MDR-TB, surpassing smoking and high fasting plasma glucose levels. In 2019, alcohol-related DALYs from MDR-TB reached a staggering 0.7 million, with an AAPC of 4.43 (3.55–5.30) percent (Table 1).

Compared to females, males were predominantly affected by alcohol consumption. Additionally, the worldwide burden of DALYs caused by alcohol consumption was considerably higher for men in 2019 than in 1990, as shown in Figure 2. In 2019, respiratory infections and tuberculosis linked to alcohol use were the leading causes of DALYs among men aged 45–49, 50–54, and 40–44, with respective tallies of 926 (650–1,163), 890 (629–1,137), and 888 (619–1,126) thousand (Figure 2A; Data 1, p. 24–34). These figures were eight to ten times higher than those for women in the same age brackets. Likewise, the ASDR associated with alcohol consumption was also markedly higher in men than in women and increased with age (Figure 2B; Data 1, p. 35–45). Furthermore, Supplementary Figure S2 illustrates the corresponding mortality data (Data 1, p. 96–115).

Analysis via Joinpoint regression revealed a mounting trend in the ASDR for MDR-TB until 2002, which was then followed by a decline from 2002 to 2019, with an APC of −3.06 (Figure 3A; Data 1, p.46–47). The turning point for this downward trend occurred earliest in the high SDI regions in 1994, while in the low SDI regions, it appeared until 1999. Nevertheless, the decline in DALYs for regions with low SDI levels plateaued after 2010 (Figure 3B; Data 1, p. 48–51). Moreover, Supplementary Figure S3 presents the associated mortality data (Data 1, p. 116–121).

Figure 4 shows the age-period-cohort effects from the APCM analysis, categorized by SDI quintile. Age curves present age effects, period effects display the relative risk of incidence by period, and cohort effects show the relative risk of incidence by cohort. Globally, there has been a consistently lower incidence of MDR-TB across all age groups. Nonetheless, it is concerning that the risk of developing MDR-TB increases with age, especially in men aged 40 and above in regions with lower SDI, in stark contrast to regions with higher SDI, where the risk is diminishing (Figure 4A; Data 1, p. 52–59).

Between 1990 and 2000, there was a troubling increase in the risk of MDR-TB incidence. However, the tide began to turn after 2000, and over the past decade, we have seen a sustained reduction in the incidence risk. The decline in high-SDI countries started earlier, and this downward trend was mirrored in the middle, high-middle, and high-SDI zones. In contrast, low-middle and low-SDI regions have shown little to no improvement in the incidence rates over the last decade, suggesting a stagnation in efforts to combat the disease (Figure 4B; Data 1, p. 60–61).

Worldwide, individuals born after 1960 have been increasingly susceptible to MDR-TB. The impact of declining birth cohorts has been more pronounced in higher-SDI countries. These countries have witnessed a progressive improvement in the incidence of MDR-TB for those born after the 1960s, while the risk in low and low-middle SDI countries has conversely mounted, with a particular increase observed in men. For individuals born in 2015, the risk ranged from 13.85 (95% CI 12.54–15.30) in low-middle SDI countries to a significantly lower 0.26 (0.18–0.38) in high SDI locations, in comparison to those born in 1960 (Figure 4C; Data 1, p. 62–70). In addition, Supplementary Figure S4 displays the correlative mortality data (Data 1, p. 122–141).

According to the Nordpred model, the number of people affected by MDR-TB is expected to decrease slightly between 2020 and 2030. The age-standardized incidence rate (ASIR) is projected to drop from 5.6 in 2019 to 4.8 per 100,000 population by 2030 (Figure 5A; Data 1, p. 71–85). However, the forecasting for 2030 indicates that the low-middle and low SDI regions may see a significant rise in new cases, with an estimated 261,473 and 97,419 individuals affected, respectively. This translates to an ASIR of 10.0 and 9.5 per 100,000 individuals (Figure 5B; Data 1, p. 71–85). Additionally, Supplementary Figure S5 illustrates the related death data (Data 1, p. 142–156).