The relationship between HbA1c control levels and antituberculosis treatment effects: a meta-analysis : Journal of the Chinese Medical Association

Secondary Logo

Journal Logo

Original Articles

The relationship between HbA1c control levels and antituberculosis treatment effects: a meta-analysis

Song, Chaoa; Xie, Wangb; Gong, Lia; Ren, Mina; Pan, Pinhuaa,*; Luo, Bailinga,*

Author Information
Journal of the Chinese Medical Association 82(12):p 915-921, December 2019. | DOI: 10.1097/JCMA.0000000000000205
  • Open



Recently, the global burden of diabetes mellitus (DM) has been rising, and tuberculosis (TB) remains a leading cause of morbidity and mortality throughout the world. Epidemiological studies have elucidated an association between DM and the development of TB disease. According to a recent systematic review among cohort studies, people with DM had approximately three times the risk of developing TB as those without DM.1 The uncontrolled glucose may lead to an increased incidence of various infectious diseases, including TB; moreover, it may also decrease the effectiveness of anti-TB treatments for DM-TB patients. Clearly, it is important for DM-TB patients to monitor their glucose levels. The TB clinical manifestations in DM-TB patients have a close relationship with glucose control levels; however, glucose levels fluctuate wildly and are easily influenced by various factors. TB manifestations may reflect transient glucose levels, but they are ineffective in evaluating diabetes control. The HbA1c is the product of glucose binding to hemoglobin in red blood cells, which is a slow, continuous, and irreversible combining process. It is a better marker of blood glucose control, as it measures blood glucose over a period of 2 to 3 months and is not subject to the rapid swings that can occur with random and fasting blood glucose measurements.2 HbA1c testing is beneficial for diagnosis, treatment, and prognosis of DM. Some studies have suggested that controlled diabetics seem to have lower positive smear rates and higher culture conversion rates after 2 months of anti-TB therapy.3 On the contrary, some studies have reported that there are no differences between treatment time and culture conversion in controlled and uncontrolled diabetes.4–8 A unified view has yet to be reached regarding whether the HbA1c control levels are the most influential factors of anti-TB treatment effects. This meta-analysis was performed to determine whether control status of diabetes influences anti-TB treatment outcomes by monitoring HbA1c.


2.1. Study selection

2.1.1. Inclusion criteria

Studies were included if they met the following criteria: (1) randomized controlled, case-controlled, or cohort studies that discuss HbA1c control status, sputum smear positive rates, and TB focus absorption; (2) All patients were diagnosed with both DM4 and TB, at the same time, HbA1c data, sputum results, and the TB focus absorption condition were available for the study; (3) main outcome measures: controlled and uncontrolled DM were defined as HbA1c <7.0% and HbA1c ≥7.0%, respectively. All patients in both groups received sputum conversion results at 2 and 3 months of anti-TB treatment; secondary outcome measures: the TB focus absorption at the end of the anti-TB intensive treatment phase; (4) TB therapy schedule included the standard TB therapy regime.10

Note that HbA1c provides a measure of blood glucose over a period of 2 to 3 months, and the anti-TB intensive treatments include a 2- and 3-month regime. In general, the sputum will convert after 2 months of anti-TB treatment. Based on these data, we decided to collect sputum conversion results at 2 and 3 months.

2.1.2. Exclusion criteria

We excluded the following: (1) animal experiments, (2) systematic reviews, lectures, or comments; (3) studies repeated and published many times; (4) incomplete medical records, such as studies that did not report at least one of the TB outcomes listed earlier; patients complicated with congenital or secondary immunodeficiency diseases, such as HIV infection; diseases which need long-term use of glucocorticoids, such as autoimmune diseases, organ transplantation, etc; drug-resistant TB and extrapulmonary TB; and (5) pregnant and lactating women. Excluded reports were less than 10 cases.

2.1.3. Languages

Both English and Chinese databases were searched.

2.1.4. Compliance with ethical standards funding

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee, the 1964 Declaration of Helsinki and its later amendments, or comparable ethical standards.

2.1.5. Informed consent

Informed consent was obtained from all individual participants included in the study.

2.2. Publication search

We screened the articles by adopting the combined search methods of computer and manual retrieval. All studies performed were searched from the earliest documents until April 1, 2016, in electronic databases, including Ovid MEDLINE, Embase, Cochrane Library, PubMed, Web of Science, CNKI, SinoMed, WanFang, and VIP. Concurrently, citations in these studies were also searched. We reviewed the retrieved studies and researched the references.

2.3. Search strategy

The following search terms were applied: tuberculosis, Mycobacterium tuberculosis, tuberculosis infection, tuberculosis disease and diabet*, HbA1c, and glycated hemoglobin. We restricted our search to publications in English and Chinese. All titles and abstracts were read to select the appropriate studies addressing the impact of HbA1c on TB outcomes. Unpublished sources of data were not included, as the current large number of DM-TB studies make the presence of publication bias improbable. All search keywords were used in conjunction with the free words.

2.4. Data extraction and study quality assessment

Two reviewers independently collected data from the selected studies using a data extraction form designed before starting the study, and the methodological qualities of the studies were assessed independently. Differences were resolved by consensus and discussion.

The quality of the randomized controlled trials (RCTs) was assessed using the risk of bias assessment tools of the Cochrane systematic review. Cohort and case–control studies were evaluated in accordance with the Newcastle–Ottawa Scale (NOS). The Cochrane systematic review risk of bias assessment tools includes six aspects: random method, allocation concealment, blinding method, incomplete outcome data, selective reporting, and other bias. NOS includes four entries of the research object: comparability between the groups is one entry, and outcome measures have three entries. The total score is nine points, and scores less than six are rated as B-class, while the others are rated as A-class.11

2.5. Statistical analysis

The RevMan 5.3 software of the Cochrane Collaboration was used to conduct the meta-analysis. The odds ratio (OR) was used as the statistic of the count data (sputum conversion rate [SCR] or TB focus absorption), and the effect size was performed using a 95% CI. Z (U) test was used to test the OR values, and p < 0.05 was considered statistically significant. Inconsistency (statistical heterogeneity) among studies was assessed by the conventional Chi-square-based Q test for heterogeneity, When p > 0.05 or I2 < 50% in the Q test, it indicated that good homogeneity existed, and a fixed-effect model (Mantel–Haenszel method) was used; otherwise, it showed there was some heterogeneity, and a random-effect model (DerSimonian–Laird method) was used. Sensitivity and subgroup analysis were used to evaluate potential sources of homogeneity. Sensitivity analysis was conducted by excluding studies successively to check the stability of the results, and publication bias was assessed by funnel plots.


3.1. Literature search and quality assessment

According to the predetermined retrieval scheme, a total of 1106 studies were screened; we excluded duplicate studies, consulted the abstracts, and finally 12 studies (Fig. 1) were included. The general information is shown in Table 1, and all of the studies were conducted in Asia, which included nine RCTs (Table 2), two cohort studies, and one case–control study (Table 3). Eight of the nine RCTs were assessed as having a moderate bias risk, and one study had a high risk of bias. Two studies of the non-RCTs (NRCT) were assessed as A-grade, while another study was B-grade.

Table 1:
General information of included studies
Table 2:
The methodological quality assessment of the RCT
Table 3:
The methodological quality assessment of the NRCT
Fig. 1.:
The literature screening processes and outcomes.

3.2. Results of the meta-analysis

The overall effect magnitude of this study is reflected in the ORs, and the weight of each independent research is shown in Figures 2 and 3. The SCR after 2 months of anti-TB therapy: five studies were included,5–9 in which 348 were uncontrolled diabetes cases (HbA1c ≥7.0), 698 were controlled (HbA1C <7.0), and the treatment regime was 3HRZE/6HR. According to the tests for heterogeneity (p < 0.05), we intended to choose the random-effects model. The random-effects OR of the SCR between controlled and uncontrolled diabetes was 2.14 (95% CI: 0.84-5.43), the 95% CI contained the case of OR equaling 1, and Z = 1.6, p = 0.11 > 0.05; therefore, there was no statistically significant difference between these two groups. This indicated that after 2 months of anti-TB therapy, we could not believe that the SCRs of the well-controlled group were higher than that of the uncontrolled group (77.64% vs 73.63%) (Fig. 2). When analyzing the characteristics of the literature, we successively included the studies and conducted cumulative analyses by referring to the quality assessment. As a result, the OR point estimates and 95% CI were unstable and waved violently; however, there were two of the five studies that showed the existence of a statistically significant difference between these two groups.14,16 According to the sensitivity analysis, except for the small sample size and the ambiguous description of the treatment regime, there was no correlation between the basic characteristics of the selected studies and the SCRs. We subsequently extracted the two articles successively, and the pooled results did not change. Next, we excluded these two studies, and the pooled OR was 1.30 (95% CI: 0.44-3.87), which was consistent with the previous results and indicated that the results were reliable. In addition, one article13 in the five studies showed that the 2 months SCR of the uncontrolled group was less than that of the controlled group, but there was no statistical difference (OR was 0.57, 95% CI: 0.31-1.03). Sensitivity analysis was carried out, and the results of another four studies were statistically significant (pooled OR was 3.17, 95% CI: 2.01-4.99). Moreover, the four studies were homogeneous, and the results were not consistent with the meta-analysis. As for the study,13 except for the large sample size disparity (82 cases were in the controlled group, 428 were in the uncontrolled group), some patients had no sputum results during follow-up, and there was no correlation between the basic characteristics of the selected articles and the SCRs. In addition, the number of uncontrolled group participants in this study was more than half of the whole uncontrolled group number (428/698), which was obviously higher than that of other studies. Thus, the difference between the sample size and the incomplete data may be the source of heterogeneity. The SCR after 3 months of anti-TB therapy: seven articles conform to the requirements,10–16 among which, 205 were controlled and 201 were uncontrolled groups. According to the heterogeneity test (p = 0.99), the fixed-effect model was selected for meta-analysis. The SCRs of the controlled group were higher than that of the uncontrolled group (82.93% vs 58.71%), as the fixed pooled OR was 3.39 (95% CI: 2.12-5.43); that is to say the 3 months SCRs between these two groups were statistically significant (p < 0.05). Regarding the seven studies, the SCRs of the controlled groups were all higher than that of the uncontrolled groups, and four of these seven articles had no statistical difference.16,18,19,22 Sensitivity analysis was performed, and the study was successively excluded; however, the results remained the same and proved that the analysis results were reliable (Fig. 3).The TB focus absorption at the end of anti-TB intensive treatment (2-3 months) was: there were seven studies comparing the absorption of TB focus between the controlled and uncontrolled groups,12,14–19 with 275 controlled and 360 uncontrolled, and the results were not statistically different (p = 0.99), so the fixed-effect model was used. The results were statistically significant as the fixed pooled OR was 2.69 (95% CI: 1.91-3.79) (Fig. 4), which showed that the absorption of the TB focus in the HbA1c controlled group was better than the uncontrolled group. In addition, there were three of the seven studies comparing the 2 months anti-TB treatment:12,14,15 170 were controlled, and 242 were uncontrolled. Studies were not heterogeneous (p = 0.59), so the fixed-effect model was used, and the pooled OR was 2.64 (95% CI: 1.71-4.08). TB focus absorption between the two groups had a statistically significant difference (Fig. 5). Another four studies compared the results of the 3 months of anti-TB therapy:16–19 105 were controlled, 118 were uncontrolled, and a fixed-effect model was used, as the results among the studies had no statistical heterogeneity (p = 0.77), and the fixed pooled OR was 2.76 (95% CI: 1.57-4.86) (Fig. 6). The sensitivity analysis of the abovementioned results was carried out, and the results showed no change, which proved that the results were reliable. The assessment of publication bias: We did not find evidence for publication bias by funnel plot for the SCR after 3 months of anti-TB treatments (Fig. 7).

Fig. 2.:
The meta-analysis of the SCR between the controlled group and uncontrolled group after 2 months of anti-TB treatment. SCR, sputum conversion rate; TB, tuberculosis.
Fig. 3.:
The meta-analysis of the SCR between the controlled group and uncontrolled group after 3 months of anti-TB treatment. SCR, sputum conversion rate; TB, tuberculosis.
Fig. 4.:
The meta-analysis of the TB focus absorption at the end of anti-TB intensive treatment phase. TB, tuberculosis.
Fig. 5.:
The meta-analysis of the TB focus absorption after 2 months of anti-TB treatment. TB, tuberculosis.
Fig. 6.:
The meta-analysis of the tuberculosis focus absorption after 3 months of anti-TB treatment. TB, tuberculosis.
Fig. 7.:
The funnel plots of publication bias after 3 months of anti-TB treatment. TB, tuberculosis.


DM and TB are chronic wasting diseases, and these two conditions can influence and promote each other. Relevant studies at home and abroad have found that the impact of diabetes on TB patients may be related to the following factors: (1) high blood glucose or deficiency of insulin secretion has provided a good environment for the survival of Mycobacterium tuberculosis; therefore, it is favorable to the colonization of the TB bacilli but unfavorable to the control of TB. (2) Abnormal immune function, impaired monocyte chemotaxis, damaged alveolar macrophage activation, decreased interferon gamma levels, and the change of type I cells live hormone expression can decrease the patients’ ability to resist and remove M. tuberculosis, and as a result, it spreads in the body, which is not conducive to the control of TB.23,24 (3) The interaction of rifampicin and hypoglycemic drugs reduced the absorption of rifampicin in patients with DM, which can influence the treatment effects of TB. There is evidence that rifampicin effects depend on the peak concentration of drug exposure time, which was lower in TB patients with DM than that of TB patients without DM. This directly affects the effectiveness of TB treatments and may lead to the production of acquired resistance.25 (4) The high incidence of drug resistance in patients with DM complicated with TB makes it difficult to use the short-course chemotherapy for controlling the disease. It is known from the factors earlier that immune system disorders and respiratory system immunity decline may occur in patients when DM and TB are combined, which leads to a vicious cycle, since the therapeutic drugs for these two diseases are mutually negatively influenced. As one of the testing indexes of diabetes, the correlation between HbA1c and the severity of TB indirectly reflects the influence of diabetes on the treatment of TB. This study shows that in the first 2 months of treatment for diabetes and TB, the SCR of TB and the severity of TB show no obvious correlation, yet as the treatment continues, the control level of HbA1c exerts a direct influence upon the prognosis of TB; moreover, in cases of more than 2 months of anti-TB therapy, the focal absorption of those patients whose level of HbA1c is well controlled within a low level performs better than those whose level of HbA1c is poorly controlled; hence, maintaining the level of HbA1c at a low level is beneficial for therapeutic effects and outcomes. It can thus be seen that, as for the treatment of diabetes complicated with pulmonary TB, the control over diabetes, especially the level of HbA1c, is helpful for the improvement of patients’ metabolism, thus adjusting the functions of immune system and enhancing the immunity of patients. The time of anti-TB therapy should be appropriately extended, which is positively significant for its treatment. In short, it is crucial for diabetes patients to control their blood glucose level, since it is directly related to the emergence, development, treatment, and prognosis of TB.2 Meanwhile, as the testing index of the control over diabetes, HbA1c directs the therapeutic schedule’s making of diabetes and TB. During the anti-TB therapy, HbA1c is monitored regularly (1-3 months) to evaluate the control level of diabetes for timely adjustment of the blood glucose reduction schedule.

This study had some limitations. First, it included a small sample size, and most were derived from domestic research, making the study quality uneven. Second, the study of random methods and distribution of hidden and blind methods were not clear. Moreover, due to the long treatment period for TB, tests may be affected by more confounding factors, and patient compliance was also an important factor in response to the results. A survey26 on TB medication compliance showed that compliance rates in 1 year were only 54%, as 46% of the patients failed to adhere to the standard treatment. More importantly, these 12 studies were not describing compliance rates. Furthermore, when making research plans, this study intended to include articles from different countries and regions, but eventually 10 of the included 12 studies were Chinese. The two English studies were conducted in Taiwan and South Korea, and such regional differences may affect the authenticity of the research. In a word, the abovementioned factors restrict the popularization and application of this research.

In conclusion, recent evidence suggests that the treatment principle should adhere to treating the two diseases at the same time. Not only should we provide regular anti-TB treatment, but we should also pay attention to the control of blood glucose in patients with DM via HbA1c: monitoring blood glucose or glycosylated hemoglobin values to make the diabetic blood glucose and HbA1c control stable over a long term. At the end, the two diseases have been effectively treated and controlled. However, due to the limitations of the methodology in the research, the reliability of the meta-analysis evaluation results is decreased. We look forward to higher quality study to further verify the effect of HbA1c levels on the treatment of pulmonary TB.


We would like to thank Editage ( for English language editing.


1. Jeon CY, Murray MB. Diabetes mellitus increases the risk of active tuberculosis: a systematic review of 13 observational studies.Plos Med20085e152
2. Kong ZS, Huang ML, Liu RM, Gao MQ. Clinical features of the diabetes combined with primary pulmonary tuberculosis.Chin J Diabetes Mellitus20146329–331[In Chinese, English abstract]
3. Magee MJ, Kempker RR, Kipiani M, Gandhi NR, Darchia L, Tukvadze N, et al. Diabetes mellitus is associated with cavities, smear grade, and multidrug-resistant tuberculosis in Georgia.Int J Tuberc Lung Dis201519685–92
4. Mi F, Tan S, Liang L, Harries AD, Hinderaker SG, Lin Y, et al. Diabetes mellitus and tuberculosis: pattern of tuberculosis, two-month smear conversion and treatment outcomes in Guangzhou, China.Trop Med Int Health2013181379–85
5. Wang JY, Lee LN, Hsueh PR. Factors changing the manifestation of pulmonary tuberculosis.Int J Tuberc Lung Dis20059777–83
6. Dooley KE, Tang T, Golub JE, Dorman SE, Cronin W. Impact of diabetes mellitus on treatment outcomes of patients with active tuberculosis.Am J Trop Med Hyg200980634–9
7. Ponce-De-Leon A, Garcia-Garcia MML, Garcia-Sancho MC, Gomez-Perez FJ, Valdespino-Gomez JL, Olaiz-Fernandez G, et al. Tuberculosis and diabetes in southern Mexico.DIABETES CARE2004271584–90
8. Park SW, Shin JW, Kim JY, Park IW, Choi BW, Choi JC, et al. The effect of diabetic control status on the clinical features of pulmonary tuberculosis.Eur J Clin Microbiol Infect Dis2012311305–10
9. 2014 ADA Diagnosis and Classification of Diabetes Mellitus.Diabetes Care201437S81–S90
10. CSTBTuberculosis diagnosis and treatment guidelines.CHIN PRACT J RURAL DOCT2013207–11
11. Liu M. Systematic Review, meta-analysis, design and implementation method20111st edBeijingPeople’s Medical Publishing House[In Chinese]
12. Song L. Effect of dietary intervention on type 2 diabetes mellitus patients with pulmonary Tuberculosis2014HenanZhengzhou University[In Chinese]
13. Chiang CY, Bai KJ, Lin HH, Chien ST, Lee JJ, Enarson DA, et al. The influence of diabetes, glycemic control, and diabetes-related comorbidities on pulmonary tuberculosis.PLOS ONE201510e121698
14. Li P, Han YH. Analysis of the effect and safety of early insulin therapy in patients with pulmonary tuberculosis complicated with diabetes.CONTEMP MED FORUM201419258–9[In Chinese]
15. Jiang XH. Impact of glycosylated hemoglobin values after recovering from tuberculosis and diabetes.CHI MED PHARM201413157–8[In Chinese]
16. Wang XH, Xu AX. Observation on the effect of short-term insulin intensification therapy for newly-diagnosed type 2 diabetes mellitus complicated with pulmonary tuberculosis.J CLIN RE2009261897–9[In Chinese, English abstract]
17. Zhang P. Influence of BCG polysaccharide nucleic acid on peripheral blood T lymphocyte subsets in pulmonary tuberculosis patients complicating type 2 diabetes.J CLIN PULM MED201571199–201[In Chinese]
18. Huang HZ, Zhong GC, Ji JP. The curative effect of type 2 diabetes complicated with pulmonary tuberculosis by rosiglitazone combined with insulin.CHIN J PRIM MED PHARM2007121992–4[In Chinese]
19. Ge ZK. Curative Effect Observation of insulin therapy in type 2 diabetes mellitus complicated with newly-treated pulmonary tuberculosis.J ANHUI HEALTH VOCAT TECH COLL20140161–2[In Chinese]
20. Li HH, Wen YY, Xiao H, Huang ZJ. Study on the effect of out-hospital follow-up nursing on self-care awareness and condition improvement of pulmonary tuberculosis with diabetic patients.PRACT PREV MED201308998–9[In Chinese]
    21. Zhen GQ, Lu J, Mai CJ, Wang HL, Deng YC, Gu MH. Study on out hospital follow up nursing on selfcare theory of pulmonary tuberculosis with diabetic patients.NURS PRACT RES2011164–6[In Chinese]
      22. Wang CM, Li GX, Xu LD, Jiang MF, Li XM. Evaluation of effect on nutrition support in patients with pulmonary tuberculosis and type 2 diabetes.J CLIN PULM MED201404685–7[In Chinese]
      23. Nijland HM, Ruslami R, Stalenhoef JE, Nelwan EJ, Alisjahbana B, Nelwan RH, et al. Exposure to rifampicin is strongly reduced in patients with tuberculosis and type 2 diabetes.Clin Infect Dis200643848–54
      24. Restrepo BI, Fisher-Hoch SP, Pino PA, Salinas A, Rahbar MH, Mora F, et al. Tuberculosis in poorly controlled type 2 diabetes: altered cytokine expression in peripheral white blood cells.Clin Infect Dis200847634–41
      25. Martens GW, Arikan MC, Lee J, Ren F, Greiner D, Kornfeld H. Tuberculosis susceptibility of diabetic mice.Am J Respir Cell Mol Biol200737518–24
      26. Zhao LY, Xie XY, Shen YH, Shao YX, Guo Y. Research analysis on medication compliances of tuberculosis patients.CHIN PHARM2005171311–3[In Chinese]

      Diabetes mellitus; Hemoglobin A; Meta-analysis sputum; Tuberculosis

      Copyright © 2019, the Chinese Medical Association.