We found an overall prevalence of DM of 5.0% among TB patients in Denmark. The prevalence among Danish-born TB patients was 6.1%. Young Danish-born TB patients seemed to have a higher prevalence of DM compared to an age and sex matched Danish background population and a high proportion, up to 39%, of older migrant TB patients had DM. Age over 40 years and being from Africa, Europe outside Denmark, or Asia were risk factors for having DM, whereas being from Greenland was not. Factors related to low socio-economic status were associated with not having DM.
The DM prevalence ranged from 0% among patients from Greenland, 6.1% among patients from Denmark to 7.7% among patients from Europe, outside Denmark. The prevalence of 6.1% among Danish-born was slightly higher than the estimated prevalence of 4.9% among the Danish background population . The risk of TB among patients with DM in Denmark has been addressed in a register-based population study  showing a high rate ratio (RR) of TB (RR: 1.9, 95% CI [1.7–2.1]) among DM patients compared to non-DM patients. In contrast to our study, they found a low RR (0.5) for having TB among African DM patients when stratifying for ethnicities. The apparent differences could be explained by differences of age, study design and study period.
The DM prevalence among older migrants was as expected higher compared to the Danish patients. This is in line with studies from Africa, Asia, and countries such as Mexico, reporting prevalences of DM among TB patients ranging from 5% to up to 50% [20,21,22,23]. Interestingly, these studies found that the DM prevalence among TB patients was not only high, but also higher than in the background population. The high prevalence not only reflects the high burden of DM in the areas, it also underlines the possible interaction between the two diseases where DM and poor glycaemic control compromise innate and adaptive immunity and enhance susceptibility to TB [23, 24].
The finding that the prevalence of DM among migrant TB patients was highest in the old age groups could be related to the fact that DM is associated both with age and the fact that many patients migrated from DM high-endemic regions and therefore had a higher risk of DM compared to the Danish patients. Also, older TB patients also had more comorbidities and therefore also had more contact with the health-care system allowing them to be diagnosed with DM .
In contrast the lower than expected prevalence in the young migrant TB patients could be due to differences in health seeking behaviour explained by cultural differences, language barriers or social problems such as homelessness and substance abuse or the fact that they were newly immigrants to Denmark. Patients from Greenland were included in the migration-group, in the assessment of age, sex and DM in Fig. 2.
In our study, younger Danish-born women (35–44 years of age) and men (35–54 years of age) had almost two times higher prevalence of DM compared to an age matched background population. This was an unexpected finding, and even though the numbers are small and the two cohorts are not directly comparable, one should be aware of the possible interactions in younger individuals, and patients with undiagnosed DM and pre-DM may be at risk of complications to DM if undiagnosed and untreated. Together these findings emphasize the need for DM screening of TB patients even in a low incidence country as Denmark. Danish guidelines for TB [25, 27] focus on this matter by recommending screening of TB patients for DM and suggest treatment of latent TB infection in individuals from high TB endemic regions with compromised immunity including DM.
In Denmark, TB is frequently seen among individuals from Greenland where the incidence of TB is high . We found no patients with known DM among TB patients from Greenland. This is surprising, since other studies show a high (and increasing) prevalence of DM in Greenland. One study  found a DM prevalence of between 15 and 18% among the Inuit population in Greenland, when using an oral glucose test, and between 6.2 and 6.8%, when using HbA1c. Additionally, a relative risk for TB among DM patients in Greenland of 11.7 has been reported, and DM in TB patients from Greenland would be expected to be common . One reason for the low prevalence in patients from Greenland could be due to lack of knowledge of their DM status; it has previously been estimated that 70% of Greenlanders with DM were unaware of their DM status . This estimate, however, is almost 20 years old and awareness of DM both in the Greenlandic public and health care system has increased in recent years .
The patients from Greenland in this study had a high prevalence of homelessness, tobacco use and alcohol use. This was also true for patients from Denmark. Homelessness, tobacco use and alcohol use is associated with low socioeconomic status. These patients often experience social barriers and therefore have altered health seeking behaviour [33,34,35,36], which could be a reason why these patients were not screened for DM. Danish studies among socially marginalised individuals show a high use of both primary and secondary health care services, but only in relation to critical illness [34, 35, 37].
Thus, the low prevalence of known DM among TB patients from Greenland could be explained by low socio-economic status and result in an underestimation of the DM prevalence in Greenlanders. Since we suspect that the patients from Greenland were underdiagnosed with DM, we believe that the prevalence of DM in the migration-group was biased towards a lower prevalence than the estimated prevalence.
As a consequence of this study and the increased awareness of the potential severe outcome of TB in DM patients, we are now routinely screening all TB patients for DM with HbA1c and blood glucose at TB diagnosis in line with screening for HIV and Hepatitis. Preliminary findings showed that 1.8% were newly diagnosed with DM and that 27.3% had pre-DM with blood glucose of > 39 mmol/mol . These patients are referred for treatment of DM and are informed about risk reduction and advised a control by their general practitioner.
Strengths and limitations
The major strength of this study was that we were able to identify all patients diagnosed with TB in Denmark between 2009 and 2014 due to the close TB monitoring performed by SSI and the Department of Infectious Epidemiology and Prevention.
In Denmark, TB treatment is free of charge and all patients are treated regardless of their residence status. We can thus assume that no patients opted out of treatment or were not treated due to financial reasons.
The main limitation of this study is the retrospective study design, with information gathered only from patient records and therefore we were only able to include information on DM registered in a Danish hospital medical record at the time of TB diagnosis. Thus, we have no information on DM diagnosis made after the start of treatment or in other settings such as by a primary physician and our estimations of DM prevalence are most likely underestimated. Also, we did not have access to information on family status, lifestyle or how long patients had had their DM diagnosis prior to TB diagnosis, which could be a confounding issue. Since DM diagnosis was not uniformly registered, this may have resulted in the fact that some individuals may have been misclassified resulting in underestimation of the DM prevalence. We were not able to analyse the prevalence of pre-DM or dysglycemia since we did not have HbA1c levels or blood glucose on all patients. To explore the true magnitude of the DM problem among TB patients in Denmark, there is a call for prospective studies including a follow up period, where different aspects of DM, pre-diabetes and dysglycemia can be assessed.