Tuberculosis (TB) is a major public health issue that threatens global health, which caused 1.3 million deaths in 2020. The burden of TB is further aggravated by the growing prevalence of acquired immunodeficiency syndrome (AIDS), diabetes mellitus (DM) and kidney disease [1,2,3], as they may contribute to the TB risk and affect treatment outcomes [4,5,6]. With the changes in people’s lifestyles, the global burden of DM is continuously increasing. It is estimated that 693 million people worldwide will suffer from DM by 2045 [7]. The epidemic of DM will further aggravate the burden of TB, especially in low- and middle-income countries, DM and impaired glucose regulation were risk factors for TB in South Africa, which ORs were 2.4 (95% CI: 1.3–4.3) and 2.3(95% CI: 1.6–3.3), TB-DM patients also had higher odds of death(OR = 2.86,95%CI:1.08–7.62) in Italy [8, 9].

Multidrug-resistant TB (MDR-TB) is at least resistant to isoniazid and rifampicin, which may result from primary infection and treatment. MDR-TB is a serious threat for global TB control, and there are about 500,000 new cases of MDR-TB in each year all around the word [10]. According to the World Health Organization (WHO) estimated, there were 157,903 multidrug -resistant (MDR) TB cases reported in 2020, nearly 69% of cases were not diagnosed and treated in time [11]. Drug-resistant TB (DR-TB) and MDR-TB make controlling TB more challenging [12]. Patients afflicted with both DR-TB and DM will face worse treatment outcomes [13, 14], Some studies had shown that DM patients have a large bacterial load, which results in longer time to culture conversion and lengthen treatment. DM also can affect the absorption and metabolism for anti-TB drugs [15]. However, there were few systematic analyses to clarify and quantify the association between DM and DR/MDR-TB outcomes. Given the increasing burden of TB among people with DM, we performed a meta-analysis to systematically assess the association between DM and the treatment outcomes of DR/MDR-TB.

Materials and methods

Search strategy and study selection

We completed the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guideline for this study. This systematic review has been registered with the International Prospective Register of Systematic Reviews (PROSPERO) ( ID = CRD42022350214; registration number: CRD42022350214). We conducted a systematic search of the electronic database, including PubMed, Excerpta Medica Database (EMBASE), Web of Science, ScienceDirect and Cochrance Library by July 2022. We used the following search terms: (“Tuberculosis” or “Tuberculosis’s” or “Multidrug” or “Drug-resistant tuberculosis” or “Drug resistant tuberculosis” or “Multidrug-resistant tuberculosis” or “Multidrug resistant tuberculosis”) AND (“Diabetes mellitus” or “Diabetes insipidus” or “Diabetes” or “Mellitus” ) AND (“Treatment(s) outcome(s)”or “Treatment(s)”or “Outcome(s)”). The EndNote X9.0 software was used to manage records, screen, and exclude duplicates.

The inclusion criteria were as follows: (1) The study was designed as a cohort, case-control or cross-sectional study;(2)We did not set any specific exclusion criteria for the type of diabetes in DR/MDR-TB patients; (3) TB cases could provide whether there was a history of DM; (4) TB cases were diagnosed as DR/MDR-TB; (5) Treatment outcomes of TB cases were recorded, and the exclusion criteria were as follows: (1) No DM patients were involved in the treatment; (2) Only TB treatment outcomes; (3) Reviews/meta-analysis; (4) Treatment outcomes information only included sputum culture and/or smear; (5) Did not have enough outcomes to extract the value; (6) Other reasons for exclusion.

Data extraction

Two reviewers extracted data independently and subsequently met to resolve discrepancies. In case of continued disagreement, a third reviewer made the final disposition. We extracted data on demographic characteristics, study design, location of the population, number of participants in each study, drug-resistant type, type of DM, score of quality assessment, adjusted odds ratio (OR) and relevant covariates (Table 1).

Table 1 Characteristics of the studies included for meta-analysis

Treatment outcomes definitions

Treatment outcomes were divided into six categories, namely cured, treatment completed, treatment failed, death, lost to follow-up, and not evaluated. Cured and completed treatment were considered successful, and the rest were deemed unsuccessful in accordance with the WHO guidelines [16] (Table 2).

Table 2 Definitions of treatment outcomes for drug-resistant tuberculosis patients [16]

Quality assessment

The quality of the included studies was evaluated using a modified version of the Newcastle-Ottawa Scale for cohort and case-control studies [17]. Studies were classified as having low (≥ 7 stars), moderate (5–6 stars), and high risk of bias (≤ 4 stars) with an overall quality score of 9 stars [18] (Table 1). For cross-sectional studies, we assigned each item of the AHRQ checklist a score of 1 (answered “yes”) or 0 (answered “no” or “unclear”). The high, moderate, and low risk of bias were identified as having a score of 0–3, 4–7, and 8–11, respectively (Table 1).

Statistical analysis

Data was extracted using the Excel 2019 software, and further analyzed by Stata/se17.0. Heterogeneity between studies was assessed using the I2 statistic described by Higgins et al [19]. The pooled effects were estimated with fixed or random effect models: I2 ≤ 50% and P > 0.10 representing insignificant heterogeneity, using fixed-effects models; I2 ≥ 50% and P < 0.10 representing significant heterogeneity, using random-effects models [20].

The pooled effects of DM on DR/MDR-TB treatment outcomes were described by forests plots, quantified by OR (besides case-control studies, cross-sectional studies and cohort studies were also estimated by OR) and the corresponding 95% confidence interval (CI). P < 0.05 was considered as statistically significant. The publication bias was assessed through funnel plot and Egger’s test. All analyses were performed using the STATA 17.0 software (Texas, USA).


Study selection and characteristics

We searched 9,918 papers by titles, abstracts and keywords and then excluded 9,416 papers without TB treatment outcomes. Among 502 articles under full-text reading, 477 articles were excluded for lacking targeted data or imperfect data (Fig. 1). Finally, we involved twenty-five eligible studies in the meta-analysis (Table 1) [13, 14, 21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43], including nine cohort studies, fourteen case-control studies and two cross-sectional studies. These studies were published from 2005 to 2022. Eleven studies were identified as having a low risk of bias, and fourteen studies had moderate risk of bias (Table 1). Twenty studies were conducted in Asian populations, four were in Europe populations, three were in African populations, and one was in American populations. The total sample size of subjects was 16,905 DR-TB patients, of which 10,124 (59.89%) participants were MDR and 1,952 (11.54%) had DM (DM+).

Fig. 1
figure 1

Flowchart of the study selection

Unsuccessful treatment outcomes

Twenty-three studies analyzed the risk of DM on unsuccessful treatment outcomes in patients with DR-TB and twenty studies analyzed the risk of DM on unsuccessful treatment outcomes in patients with MDR-TB. DM patients were more likely to have unsuccessful treatment outcomes in DR-TB (OR = 1.56, 95% CI: 1.24–1.96) (Table 3; Fig. 2A) and MDR -TB patients (OR = 1.57, 95% CI: 1.20–2.04) (Table 3; Fig. 2B). Sensitivity analysis showed that four studies contributed the main heterogeneity [21, 27, 32, 35], which might be attributed to the inclusion of extensively drug-resistant (XDR-TB) [32, 35]. Figure 3 A and Fig. 3B illustrated the funnel plots of involved studies for DR- TB and MDR-TB patients with DM. We did not find the evidence for publication bias in DR-TB treatment outcomes (P = 0.086) and MDR-TB treatment outcomes (P = 0.365) by Egger’s test (Table 3).

Table 3 Pooled effects odds ratio (95% confidence interval), Heterogeneity test and Egger’s test for publication bias
Fig. 2
figure 2

Forest plots for the association of diabetes mellitus with unsuccessful treatment outcomes for DR-TB (A) and MDR-TB (B)

Fig. 3
figure 3

Funnel plot of the studies based on the association between DM and unsuccessful treatment outcomes for DR-TB (A) and MDR-TB (B)


We further compared the risk of death for DR/MDR-TB patients with and without DM. The random-effects model was used to estimate the pooled effects, as there was a significant heterogeneity for DR-TB studies (I2 = 53.3%, P = 0.029) and MDR-TB studies (I2 = 59.2%, P = 0.016) (Table 3). The pooled OR was 1.32 (95% CI: 0.97–1.82) and 1.33 (95% CI: 0.85–2.07), respectively (Table 3; Fig. 4A and B). There was no evidence for publication bias by Egger’s test (P = 0.929 in DR-TB; P = 0.940 in MDR-TB) (Table 3).

Fig. 4
figure 4

Forest plots for the association of diabetes mellitus with death treatment outcomes for DR-TB (A) and MDR-TB (B). # The number in this study was zero


DR/MDR-TB patients without DM were more likely to be cured (DR-TB: OR = 0.64, 95% CI: 0.44–0.94 (Table 3; Fig. 5A); MDR-TB: OR = 0.55, 95% CI: 0.35–0.87) (Table 3; Fig. 5B). The random-effects model was used as there was significant heterogeneity (DR-TB: I2 = 75.7%, P = 0.001; MDR-TB: I2 = 66.5%, P = 0.018). The Egger’s test suggested that there was no publication bias (P = 0.062 in DR-TB and P = 0.263 in MDR-TB).

Fig. 5
figure 5

Forest plots for the association of diabetes mellitus with cured treatment outcomes for DR-TB (A) and MDR-TB (B)

Treatment completed

DR/MDR-TB patients without DM were more likely to complete treatment (DR-TB: OR = 0.63, 95% CI: 0.46–0.86 (Table 3; Fig. 6A); MDR-TB: OR = 0.66, 95% CI: 0.46–0.93) (Table 3; Fig. 6B). There was no evidence for heterogeneity (DR-TB: I2 = 0.00%, P = 0.660; MDR-TB: I2 = 0.00%, P = 0.559). There was no evidence for publication bias by Egger’s test (P = 0.221 in DR-TB and P = 0.192 in MDR-TB).

Fig. 6
figure 6

Forest plots for the association of diabetes mellitus with completed treatment outcomes for DR-TB(A) and MDR-TB(B). # The number in this study was zero

Treatment failure

DR/MDR-TB patients with DM were more likely to have treatment failed outcomes (DR-TB: OR = 1.28, 95% CI: 1.03–1.58 (Table 3; Fig. 7A); MDR-TB: OR = 1.37, 95% CI: 1.08–1.75) (Table 3; Fig. 7B). There was no evidence for heterogeneity (DR-TB: I2 = 22.7%, P = 0.256; MDR-TB: I2 = 19.7%, P = 0.284). There was no publication bias by Egger’s test (P = 0.263 in DR-TB and P = 0.275 in MDR-TB).

Fig. 7
figure 7

Forest plots for the association of diabetes mellitus with failed treatment outcomes for DR-TB (A) and MDR-TB (B)


This study systematically reviewed the impact of DM on the treatment outcomes of DR/MDR-TB patients. We demonstrated the negative effect of DM on the prognosis of TB, which was consistent with the findings by Meghan and Sanju et al. [44, 45]. In this kinds topic research, previous systematic review and meta-analysis were focused on the treatment outcomes of TB and MDR-TB with DM, such as Huangfu and Tegegne et al. on treatment outcomes of TB and MDR-TB [9, 46]. Our study included treatment outcomes for both DR and MDR-TB patients with DM.

The prevalence of DM in TB patients was 11.54% (95% Cl: 11.06–11.93) in this study, which was lower than the global level (15.4%, 95% Cl: 14.1–16.6), and marginally higher as compared to the prevalence in Africa (9%, 95% Cl: 6.0–12.0) and China(7.8%, 95%CI:1.6–30.5)in Asian [47,48,49]. This result was most likely due to a higher proportion (88.0%) of African and Asian countries in our studies. The reason for this result is the difference of income in different countries and regions, for example, the study of Maier W al. show regional income plays a significant part in the explanation of diabetes prevalence [50].

DM can induce abnormalities in innate and adaptive immune responses, increasing the risk of the activation, complication, and outcomes of TB [51]. TB patients with DM have a rapidly progressive infection and a higher bacterial burden [52]. Coincident DM modulates Th1-, Th2-, and Th17-cell responses in latent TB in an IL-10- and TGF-β-dependent manner [53]. TB patients with DM had an increased risk of death and late culture transformation [54, 55].

The possible hypothesis of delay in the time of clearance and treatment failure of TB among DM patients is related to higher bacterial burden at diagnosis, which could be related to slower kinetics in the immune response in DM patients and altered pharmacokinetics of anti-TB drugs [55,56,57,58]. A pharmacokinetic study noted that plasma levels of rifampicin were 53% lower in TB patients with DM [59]. Depressed production of IFN-γ in DM patients is related to a decreased immune response to TB infection. The reduced IL-12 response to mycobacterial stimulation in leukocytes from TB with DM suggests a compromise of the innate immune response [60]. Roger et al. showed that TB patients with prediabetes or DM were more likely to have unsuccessful treatment outcomes in Peru, with an OR of 6.1 (95% Cl: 1.9–19.6) [61]. Siti et al. reported that TB patients with DM were three times more likely to have an unsuccessful treatment outcome than those without DM in Kelantan state, Malaysia [62]. MDR-TB is a type of TB, Therefore, the effect of glycemic control on treatment outcomes in TB patients with DM can also be applied to MDR-TB patients. Blood glucose control had a positive effect on the treatment outcome of TB patients with DM, An Indian study reported 30% fewer unsuccessful treatment outcomes (aOR = 0.72, 95% CI: 0.64–0.81) and 2.8 times higher odds of ‘no recurrence’ (aOR = 2.83, 95% CI: 2.60–2.92) among patients with optimal glycemic control at baseline [63]. Magee MJ et al. from Lima, Peru found reported faster culture conversion among those with glycemic control(aHR = 2.2,95% CI:1.1,4) [64]. There are some limitations to this study, Firstly, most of the included studies were from developing countries Asia and Africa and none were randomized controlled trials (RCT), which may have biased our research results. There were many factors that affected the severity of tuberculosis such as income level, temperature, and presence of other comorbidities. However, we found that a lot of relevant information could not be extracted in the original study, which may affect the generalization of finding.

In conclusion, DM is a risk factor for adverse outcomes in DR-TB or MDR-TB patients. Controlling hyperglycemia may contribute to a favorite prognosis of TB. Given the increasing burden of TB among people with DM, especially in areas with highly prevalent TB. It is needed to control glucose and therapeutic monitoring during the treatment of DR-TB /MDR-TB patients.