ORIGINAL RESEARCH published: 23 May 2022 doi: 10.3389/fpubh.2022.789952 Subjects With Diabetes Mellitus Are at Increased Risk for Developing Tuberculosis: A Cohort Study in an Inner-City District of Barcelona (Spain) Violeta Antonio-Arques 1,2 , Josep Franch-Nadal 1,3,4*, Antonio Moreno-Martinez 5,6*, Jordi Real 1,4 , Àngels Orcau 6,7 , Didac Mauricio 1,4,8,9 , Manel Mata-Cases 1,4,10 , Josep Julve 4,11 , Elena Navas Mendez 12 , Rai Puig Treserra 1 , Joan Barrot de la Puente 1,13 , Joan Pau Millet 6,7 , Jose Luis Del Val García 12,14 , Bogdan Vlacho 1 and Joan A. Caylà 15 1 DAP-Cat Group, Unitat de Suport a la Recerca Barcelona, Fundació Institut Universitari per a la Recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain, 2 Primary Health Care Center Bordeta Magòria, Gerència d’Atenció Primària Barcelona Ciutat, Institut Català de la Salut, Barcelona, Spain, 3 Primary Health Care Center Raval Sud, Edited by: Gerència d’Atenció Primària Barcelona Ciutat, Institut Català de la Salut, Barcelona, Spain, 4 CIBER of Diabetes and Chiranjib Chakraborty, Associated Metabolic Diseases (CIBERDEM), Instituto de Salud Carlos III (ISCIII), Madrid, Spain, 5 Department of Infectious Adamas University, India Diseases, Hospital Clínic de Barcelona, Barcelona, Spain, 6 CIBER of Epidemiology and Public Health (CIBERESP), Instituto Reviewed by: de Salud Carlos III (ISCIII), Madrid, Spain, 7 Epidemiology Service, Agència de Salut Pública de Barcelona, Barcelona, Spain, Marilia Brito Gomes, 8 Department of Endocrinology and Nutrition, Hospital Universitari de la Santa Creu i Sant Pau, Barcelona, Spain, Rio de Janeiro State University, Brazil 9 Department of Medicine, University of Vic—Central University of Catalonia, Barcelona, Spain, 10 Primary Health Care Center Manojit Bhattacharya, La Mina, Gerència d’Atenció Primària Barcelona Ciutat, Institut Català de la Salut, Barcelona, Spain, 11 Department of Fakir Mohan University, India Biochemistry, Institut de Recerca de l’Hospital de la Santa Creu i Sant Pau, Barcelona, Spain, 12 Fundació Institut Universitari *Correspondence: per a la Recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain, 13 Primary Health Care Center Josep Franch-Nadal Doctor Jordi Nadal, Gerència d’Atenció Primària Girona Ciutat, Institut Catala de la Salut, Salt, Spain, 14 Unitat d’Avaluació, josep.franch@gmail.com Sistemes d’informació i Qualitat, Gerència d’Àmbit d’Atenció Primària Barcelona Ciutat, Institut Català de la Salut, Barcelona, Antonio Moreno-Martinez Spain, 15 Foundation of the Tuberculosis Research Unit of Barcelona, Barcelona, Spain antoniov.moreno@gmail.com Background: Tuberculosis is the leading cause of mortality from lung infectious Specialty section: This article was submitted to disease worldwide in recent years, and its incidence has re-emerged in large cities Clinical Diabetes, in low-incidence countries due to migration and socioeconomic deprivation causes. a section of the journal Diabetes mellitus and tuberculosis are syndemic diseases, with diabetes being Frontiers in Public Health considered a risk factor for developing tuberculosis. Received: 06 October 2021 Accepted: 12 April 2022 Objective: To investigate whether diabetic patients were at increased risk of tuberculosis Published: 23 May 2022 living in an inner-district of a large city of northeastern Spain. Citation: Antonio-Arques V, Franch-Nadal J, Methods: Observational matched retrospective cohort study based on clinical records Moreno-Martinez A, Real J, Orcau À, from the population of the lowest socioeconomic status in Barcelona (Ciutat Vella district). Mauricio D, Mata-Cases M, Julve J, Navas Mendez E, Puig Treserra R, A cohort including patients with type 1 and type 2 diabetes mellitus in 2007 and new Barrot de la Puente J, Millet JP, Del Val cases until 2016 (8004 subjects), matched 1:1 by sex and age with a non-diabetic cohort. García JL, Vlacho B and Caylà JA (2022) Subjects With Diabetes Mellitus Follow-up period was until December 31st 2018. We evaluated the risk of developing Are at Increased Risk for Developing tuberculosis in diabetic patients compared to non-diabetic patients during the follow Tuberculosis: A Cohort Study in an up period. We used time-to-event analysis to estimate the incidence of tuberculosis, Inner-City District of Barcelona (Spain). Front. Public Health 10:789952. and competing risks regression by clusters and conditional Cox regression models to doi: 10.3389/fpubh.2022.789952 calculate the hazard ratio (HR) and its 95% confidence intervals (CI). Frontiers in Public Health | www.frontiersin.org 1 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients Results: Among the 16,008 included subjects, the median follow-up was 8.7 years. The mean age was 57.7 years; 61.2% men and 38.8% women in both groups. The incidence of tuberculosis was 69.9 per 100,000 person-years in diabetic patients, and 40.9 per 100,000 person-years in non-diabetic patients (HR = 1.90; CI: 1.18–3.07). After adjustment for the country of origin, chronic kidney disease, number of medical appointments, BMI, alcoholism and smoking, the risk remained higher in diabetic patients (1.66: CI 0.99–2.77). Additionally, subjects from Hindustan or with a history of alcohol abuse also showed a higher risk of developing tuberculosis (HR = 3.51; CI:1.87–6.57, and HR = 2.73; CI:1.22–6.12 respectively). Conclusion: People with diabetes mellitus were at higher risk of developing tuberculosis in a large cohort recruited in an inner-city district with a high incidence for this outcome, and low socioeconomic conditions and high proportion of migrants. This risk was higher among Hindustan born and alcohol abusers. Keywords: diabetes mellitus, tuberculosis, incidence, diabetes complications, Mycobacterium tuberculosis infection, alcohol abuse INTRODUCTION Diabetes mellitus (DM) is considered by the WHO as one of the most prevalent chronic diseases and one of the main Tuberculosis (TB) is a contagious airborne disease that has risk factors for developing active TB (1). The global prevalence become the leading cause of mortality worldwide in recent of DM is around 9.3% in adults between 20 and 79 years- years. The poorest countries and the most disadvantaged old, affecting about 463 million people worldwide. According populations in developed countries are especially affected. to recent estimates, the incidence of DM will continue growing According to the World Health Organization (WHO), in 2020, (6), especially in areas with low and medium incomes and 10 million people developed this disease, causing 1,3 million a high burden of TB, such as the Middle East, East Pacific, deaths, 214,000 additional deaths in HIV-infected people. Two- Southeast Asia and Africa, that represent 80% of the global thirds of TB cases worldwide are concentrated in 8 countries: burden of DM (6). In terms of notification, prevalence of DM India, China, Indonesia, the Philippines, Pakistan, Nigeria, was estimated to be around 13.8%, about half of that was Bangladesh and South Africa (1). At a global level, it is unknown (7). very difficult to calculate the number of cases with sub- This increased risk of TB in DM patients may be, at least diagnosis or with under notification. However, in the last partly, explained by hyperglycemia-induced disruption of WHO TB Annual Report, the data showed a substantial fall immunological mechanisms (8). Currently, there are new (18%) between 2019 and 2020, from 7.1 million TB cases hypotheses about the role that hormonal and metabolic to 5.8 million due to the impact of COVID. This set back, changes that take place in the context of diabetes would coupled with continued disruptions in 2021, mean that the have in immunity against M. tuberculosis (9). Socioeconomic United Nations high-level meeting target of treating 40 million deprivation increases the risk of both conditions. Diabetic people diagnosed with TB in the 5-year period 2018–2022 is patients are much more frequently exposed to potentially off-track (1). contagious patients due to a higher rate of visits to healthcare Regarding the situation of TB in Europe, high notification facilities (10). In turn, TB can also become a risk factor rates have been reported in big cities from low-incidence for developing DM. Due to the inflammation induced by countries, including Birmingham, London, Brussels and the infection itself, a transient hyperglycemia has been Barcelona (2). In recent years, migratory movements to high- described in infected people at diagnosis (11), which income countries from low-income countries with a high TB can further be related to a greater risk of progression to burden have changed the epidemiological pattern of this disease DM (12). in recipient countries, mainly in disadvantaged urban areas of Some studies have observed that the prevalence of latent large cities (3). Additionally, these areas commonly have a high TB infection (LTBI) is two times higher in diabetic patients risk of overlapping health and social problems, contributing (13). Also, diabetic subjects show an approximately three to TB incidence: homelessness, residents living below the times higher incidence of active TB compared to non-diabetic poverty line and under overcrowded conditions, alcohol and subjects (14), more severe clinical presentation, with worse drug abuse, and history of imprisonment (4, 5). An example chest radiographs, more adverse effects to anti-TB drugs, a of an inner district that meets these characteristics is Ciutat greater need for admission at diagnosis (15), and a higher risk Vella in Barcelona, showing the highest incidence of TB in of relapse after TB treatment (16, 17). Moreover, the risk of the city. having sputum cultures remaining positive 2–3 months after Frontiers in Public Health | www.frontiersin.org 2 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients starting the treatment would be approximately double. A two- Variables fold higher risk of multidrug-resistant TB has also been observed The primary outcome was defined as the occurrence and date among DM patients. Patients with DM and TB would have a of a TB diagnosis during the follow-up period in DM and non- 1.88 times higher risk of mortality (17, 18), especially those DM patients. who are older or have more comorbidities or a lower albumin We obtained the following information on the subject level (19). at baseline: (1) demographic characteristics, smoking habit The aim of our study was to determine the risk of TB in and excessive alcohol consumption (defined by the recorded diabetic patients compared to a non-diabetic population in an International Classification of Disease 10 (ICE.10) F10.1), (2) inner-city district with a high TB incidence. clinical variables, (3) laboratory data, (4) history of comorbidities, (5) history of immunosuppressive or corticosteroids treatment, MATERIALS AND METHODS (6) history of sexually transmitted infections. As Ciutat Vella is a district with a high percentage of Study Design migrants (50.1%), mostly from South-East Asia (Pakistan 13.6%, We designed an observational matched retrospective cohort Bangladesh 8.1% and India 3.8%) (20), we grouped subjects study (subjects with either type 1 or type 2 DM, and matched coming from these countries in one category: Hindustani origin. non-diabetic subjects). The recruitment period was between Hindustan is an Asian peninsula formed by India, Pakistan, January 1st 2007 and December 31st 2016, with a follow-up Bangladesh, Bhutan, Nepal, Sri Lanka and Maldivas. period of two more years, until December 2018. We identified diabetic patients if they had an ICE 0.10 The inclusion date was January 1st 2007 for prevalent diagnostic code for DM in their medical record (E10–E14), or diabetic participants, and the date of diagnosis for incident if they were taking any anti-diabetic drug other than metformin diabetic subjects during the recruitment period. For each (sometimes used in other conditions). We also recorded glycated DM case, a non-DM control matched by sex, year of birth, hemoglobin, DM duration, treatment (lifestyle management, oral and time of inclusion was selected (ratio = 1:1), using a medication and/or insulin), and microvascular complications: density sampling approach (Figure 1) see more information neuropathy, nephropathy or retinopathy. about the methodology in https://jrealgatius.github.io/TBC_ A person was considered to have TB if they had a ANALISIS/codi/shiny/DashBoard_TB.html. maintained anti-TB drug prescription according to the Barcelona TB Program; the program has an active epidemiological Study Population and Sample surveillance system with a good follow-up of patients. TB cases The study was carried out in Ciutat Vella, a district in Barcelona and their characteristics such as diagnostic procedure (smear (Northeastern Spain), which has been recently characterized by a observation, culture, tuberculin skin test), location (pulmonary high percentage of immigrants (50,1%) and a low socioeconomic or extrapulmonary), treatment and socioeconomic deprivation level, with a population of 108,000 inhabitants (20). index (MEDEA) were collected. This index classifies the subject All diabetic patients over the total attended population in by living area: values have an average of 0 and a standard Ciutat Vella were selected, and finally 8,004 of them were deviation of 1, and higher values indicate a more unfavorable matched (see the flow-chart in Figure 2). socioeconomic situation (21). Inclusion and Exclusion Criteria Statistical Methods In the DM cohort, patients aged over 18 years, fulfilling criteria Baseline characteristics of both groups were evaluated to examine for a DM diagnosis were included. The diabetic subjects were homogeneity in comorbidities and demographic and clinical matched for age and sex with non-DM sample (controls). characteristics, and were described by frequencies (n) and People younger than 18 years old or without any appointment percentages (%), continuous and count variables were described registered in their Primary Health Care Centre during the study using mean and standard deviation (SD), using compareGroups period were excluded. R Package (22). A complete data analysis was performed without imputation of missing data. If the diagnosis was not recorded, we Data Sources assumed that the patient did not have the given condition. The All data was treated as confidential according to the ethical main variables, DM and TB diagnoses, were based on clinical principles of the Helsinki Declaration of 1964, revised by records. Continuous variables that were included in the models the World Medical Organization in Edinburgh in 2000, and were categorized in tertiles, and the missing category was added. the Spanish Organic Law 3/2018 of Data Protection. Clinical We conducted a time-to-event analysis to estimate the characteristics and specific data on DM were extracted from the incidence of TB by group, and to examine the association Primary Care medical record database from Barcelona. Specific between DM condition and risk of TB. To examine time- data on TB were obtained from the Barcelona TB Prevention related incidence curves and differences between groups, we and Control program. Both sources were linked through an performed competing risks regression for clustered data using anonymized unique identifier. Because data was collected on crrSC R package version 1.1 (23). Risk functions and hazard a routine basis following the Spanish law about mandatory ratios (HR) unadjusted and adjusted, with their 95% confidence notification diseases, no informed consent was required. The data interval (CI), were estimated. CIs and p-values were computed collection took place during 2019. with robust standard errors to account for the matched sample. Frontiers in Public Health | www.frontiersin.org 3 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients FIGURE 1 | Sampling and follow-up scheme: an example of s pairs of individuals by age and sex. FIGURE 2 | Flow chart of the cohorts within the study population. Frontiers in Public Health | www.frontiersin.org 4 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients A sensitivity analysis of different fitted models was performed smoking status, alcohol use, homelessness, unemployment, including different adjusted variables using a Cox proportional death, site of TB and tuberculin skin test, did not significantly hazard model by clusters. The models were adjusted by origin differ between both groups. However, patients with TB from and clinical characteristics with known clinical association with the two groups had a different socioeconomic deprivation TB and DM. Statistical significance was established as a p- index according to the Medea classification (Table 4). value < 0.05. Regarding the location of the infection, we observed that Finally, we performed a descriptive analysis of the cases of TB in the DM group, 64.55% of the patients that developed in DM and non-DM subjects. Data management and analysis was TB had pulmonary involvement (with 6.25% of them with done with the R version 3.6.3 package (R Core Team 2020). The extrapulmonary involvement too), while in the non-DM group, R data analysis scripts can be consulted at: https://github.com/ 48% of the patients had pulmonary involvement, 12% with jrealgatius/TBC_ANALISIS. extrapulmonary implication. RESULTS DISCUSSION General Characteristics Initially we started from a potential population from Ciutat Vella Our work showed a 1.90-fold higher risk of developing TB (Barcelona) of 85,919: 10,065 diabetic subjects (prevalent and in DM patients. After adjusting for country of origin, chronic incidents during the follow up) and 75,854 potential non-diabetic kidney disease, number of medical visits, BMI, alcohol abuse subjects. Of these, 8,004 diabetics and 8,004 controls of the same and smoking, the risk remained higher at 1.66 for DM patients. age and sex were matched (Figure 2). By analyzing other variables related to TB, our analysis revealed Regarding the sample demographic characteristics (Table 1), that subjects of a Hindustani origin and history of alcohol the average age was 57.7 years (SD = 14.2), and 61.2% were abuse were at higher risk for developing TB. As the use of men and 38.8% women in each group. Twice as many patients alcohol is really frequent in our context (24), we considered in the DM cohort were of Hindustani origin (13.5 vs. 6.4% in the using abuse of alcohol as a risk factor for TB, instead of use. non-DM cohort). Diabetic patients had a higher frequency of TB- When we classify the subjects under Hindustani origin, we associated risk factors including alcohol abuse (5.0 vs. 3. 9% in the refer to a geographical aspect. We consider that their country non-DM cohort), smoking (7.3 vs. 5.7%), chronic renal failure of origin, together with their migration process and living (3.2 vs. 1.1), corticosteroids treatment (1.1 vs. 0.6%), number of conditions in Barcelona, have a significant impact on their health medical appointments (86.3 vs. 74.5) and a higher BMI (30.1 status. Noteworthy, a higher proportion of DM subjects with vs. 28.9). TB had not been previously prescribed with any specific anti- In the DM cohort, the mean DM duration was 3.2 years (SD DM treatment. We determined that the incidence rate of TB 5.8), and 45.3% (n = 1,505) had a glycated hemoglobin value reached 56.24 cases/100,000 person-years, and most of the cases above 7% (Table 1). were diagnosed in foreign-born subjects (44 cases vs. 29 in Spanish-born patients). The number of subjects with TB was Risk of TB in DM Patients significantly elevated in foreign-born in both DM and non- At follow-up (Table 2), 73 new cases of TB had been reported DM cohorts. Interestingly, the incidence of DM was two-times (median follow-up = 8.7 years). The overall incidence rate was higher in migrants coming from Hindustan compared to other 56.2 cases per 100,000 inhabitants per year. Specific rates were origins. On the other hand, regarding the incidence of TB, DM 69.9 cases of TB per 100,000 person-years in the DM group, patients had more pulmonary involvement compared to the and 40.8 cases per 100,000 person-years in the non-DM group. non-DM group. The unadjusted risk ratio observed was 1.90 (CI: 1.18–3.07) In the model 4, when adjusting by all covariables, the (Figure 3). confidence interval remains in the limit of statistical significance, The result of sensitivity analysis taking into account different probably due to a lower statistical power or a risk of adjusting variables and different approaches (competing risk overadjustment. Hindustan origin and alcohol abuse are expected models and Cox proportional hazard model by clusters) did not risk factors for developing TB because are also related to materially change the results in terms of the direction of the risk poor socioeconomic conditions in this context, and they are (Table 3). considered risk factors independently of having or not diabetes. The forest plot for different HRs of each analyzed variable The inner-city district of Ciutat Vella is characterized by the related to TB (DM, country of origin, number of medical visits, lowest socioeconomic level in Barcelona and a high percentage chronic kidney disease, BMI, alcoholism and smoking) is shown of immigrants. Half of them are migrants from Asia (Pakistan in Figure 4. 13.6%, and Bangladesh 8.1%), Oceania (Philippines 8,5%) and Central and South America (20). In 2017, the incidence of TB Differences Between DM and Non-DM in this city district was estimated to be 43.8 cases per 100,000 Patients in TB Localization inhabitants/year, while the global incidence in the city was Finally, we analyzed the differences between diabetic and 16.5 (25). non-diabetic patients who presented with a first episode of TB. Our results showing a higher risk of developing TB in DM Demographic and clinical data, including country of origin, patients are in line with several other studies performed in Frontiers in Public Health | www.frontiersin.org 5 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients TABLE 1 | Baseline characteristics of the study variables between diabetic and non-diabetic cohorts. Variable DM (n = 8,004) Non-DM (n = 8,004) p-values Age: mean (SD) 57.7 (14.2) 57.7 (14.2) 0.921 Gender: n (%) 0.998 Men 4,911 (61.1%) 4,914 (61.1%) Women 3,127 (38.9%) 3,127 (38.9%) Origin <0.001 Spain/high-income countries 5,949 (74.3%) 6,661 (83.2%) Hindustan 1,083 (13.5%) 508 (6.35%) Other 972 (12.1%) 835 (10.4%) Alcohol abuse 401 (5.01%) 311 (3.89%) 0.001 Smoking habit 588 (7.35%) 452 (5.65%) <0.001 BMI 30.1 (5.30) 28.9 (5.02) <0.001 SBP 137 (17.9) 134 (16.1) <0.001 DBP 79.2 (11.2) 78.1 (10.8) <0.001 Total cholesterol 207 (48.0) 212 (41.7) <0.001 HDL cholesterol 48.1 (13.5) 54.4 (14.1) <0.001 LDL cholesterol 124 (36.3) 133 (35.0) <0.001 Triglycerides 106 (27.2) 99.2 (27.5) <0.001 Hemoglobin 13.5 (1.25) 13.5 (1.20) 0.671 Leukocytes 11.5 (2.46) 12.0 (6.52) 0.501 Platelets 259 (71.5) 251 (69.7) 0.013 VSG 23.0 (19.4) 20.5 (16.7) 0.042 Chronic renal failure 254 (3.16%) 89 (1.11%) <0.001 Kidney transplant 12 (0.15%) 5 (0.06%) 0.145 Peripheral arteriopathy 227 (2.84%) 94 (1.17%) <0.001 Coronary heart disease 644 (8.05%) 280 (3.50%) <0.001 Stroke 371 (4.64%) 177 (2.21%) <0.001 Autoimmune disease* 80 (1.00%) 71 (0.89%) 0.513 Immunosuppressive treatment 36 (0.45%) 22 (0.27%) 0.087 Corticosteroids treatment 90 (1.12%) 49 (0.61%) 0.001 STI 51 (0.64%) 48 (0.60%) 0.840 AIDS 80 (1.00%) 158 (1.97%) <0.001 Flu vaccination 3,039 (38%) 2,193 (27.4%) <0.001 Number of medical visits 86.3 +/- 80.7 74.5 +/-77.6 <0.001 Diabetic characteristics Glycosylated hemoglobin: mean (SD) 7.41% (1.85) <7%: n (%) 1,818 (54.7%) >7%: n (%) 1,505 (45.3%) Diabetes treatment No treatment n (%) 4,824 (60.3%) Only oral medication n (%) 2,627 (32.8%) Insulin n (%) 553 (6.91%) Diabetes complications Diabetic nephropathy n (%) 95 (1.19%) Diabetic neuropathy n (%) 97 (1.19%) Diabetic retinopathy n (%) 297 (3.71%) Data are presented as number (percentage), mean ± SD (standard deviation); n, sample number; P value ≤ 0.05 were considered significant. AIDS, acquired immune deficiency syndrome; BMI, body mass index; DBP, diastolic blood pressure; DM, diabetes mellitus; HDL, high density lipoprotein; LDL, low density lipoprotein; SBP, systolic blood pressure; SD, standard deviation; STI, sexual transmitted disease; VSG, erythrocyte sedimentation rate. Autoimmune disease*, Behçet, lupus, polyarthritis, rheumatoid arthritis, sarcoidosis, scleroderma, Sjögren, vasculitis. different countries, with elevated estimated risk ratios being 1.77 addition, a systematic review of 13 observational studies from times (CI: 1.41–2.24) for active TB in DM patients in China different countries published in 2008 reported 3-fold increased (26), 1.9 (CI: 1.7–2.1) in Denmark in 2015 (27), and 6 times (CI: risk in subjects with DM (14). Co-existence of DM and TB has 5–7.2) in southern Mexico in 2004 (28), to mention some. In been described in studies done worldwide. In a report from Perú, Frontiers in Public Health | www.frontiersin.org 6 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients TABLE 2 | Incidence of tuberculosis in diabetic and non-diabetic subjects. Number of subjects Total person time (years) Number of TB cases Incidence Unadjusted risk Adjusted hazard density rate ratio ratio* (95%CI) (95% CI) Non-DM subjects 8,004 61,198.45 25 40.85 Ref Ref (26.44–60.33) DM subjects 8,004 68,605.36 48 69.97 1.90 1.66 (51.59–92.76) (1.18–3.07 (0.99–2.77) *Adjusted by: country of origin, chronic kidney disease, number of medical visits, BMI, alcohol abuse and smoking. FIGURE 3 | Incidence of tuberculosis in diabetic and non-diabetic patients. TABLE 3 | Estimated hazard ratios according to different approaches. Variables in model Competitive risk Cox model HR (95% CI) HR (95% CI) Model 1 Group DM, origin 1.68 (1.03–2.75) 1.50 (0.92–2.45) Model 2 Group, origin, number of visits 1.68 (1.03–2.74) 1.50 (0.92–2.45) Model 3 Group, origin, number of visits, CKD; BMI 1.68 (1.01–2.79) 1.49 (0.90–2.46) Model 4 Group DM, Origin, CKD, number of visits, BMI, 1.66 1.47 alcohol, smoke (0.99–2.77) (0.89–2.44) BMI, body mass index; CKD, chronical kidney disease; HR, Hazard Ratio. Frontiers in Public Health | www.frontiersin.org 7 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients FIGURE 4 | Forest plot of HR of each TB risk factor in the multivariate model. DM was described as the most frequent comorbidity in patients related to a central distribution of adiposity in these people. This with TB (29). In another work done in Ethiopia, DM had a high is consistent with our results related to the higher incidence of prevalence between patients with TB (30). DM found in Hindustani subjects. The higher incidence of TB in immigrant populations may be Our data also showed that DM patients with TB had more due to migrants coming from countries with a high incidence pulmonary involvement. This result is consistent with previous of TB. However, socioeconomic deprivation (i.e., living below reports (26, 34). In line with this, other presentations, such as the poverty line) and living conditions (i.e., overcrowded) may osteoarticular, lymphatic and digestive involvement, have been also increase exposure to TB. Indeed, in a study in people with highly associated with people of Indian subcontinent origin (35). TB from across Barcelona (2000–2013), autochthonous people Because of the large number of subjects affected worldwide by had a higher risk of developing DM (15). In contrast, our both diseases, and that they exacerbate each other and increase work involved people from a deprived district where almost half the burden when they appear together, we can probably define of the inhabitants are migrants coming from high TB burden this as a syndemic. Their interaction is a priority issue in global countries. According to another study, migration from outside public health. TB is often related to poor living conditions and the European Union (EU) does contribute to the TB burden, overcrowding, poverty and immigration. At the same time, these being one in five cases of TB notified in the EU originated in an conditions are also related to poor nutritional habits, obesity and, extra-EU country between 2007 and 2013 (31). in case of DM, poor glycemic control. Interestingly, an independent study conducted in 2008 in the In contrast to DM, the contribution of obesity to TB is same district, showed that Hindustani young people had up to controversial. Although it is considered a risk factor for DM, it 3 times more cases of type 2 DM than autochthon population, has been reported as a protective factor against TB development albeit having a lower average BMI (32). The notion that Asian (36). An integrated approach, combining the management of the Indians are more susceptible to develop type 2 DM is widely two diseases, could lead to a better health service for people in accepted (33), being probably due to a higher insulin resistance disadvantaged areas. Frontiers in Public Health | www.frontiersin.org 8 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients TABLE 4 | Demographic and clinical differences in TB patients according to their Finally, there might have been missing data regarding evolution DM status. and treatment of TB, toxic habits or traveling. DM (48) Non-DM (25) p-value Our work has also a number of strengths. Among these, the population database for a city area with an extremely high Origin incidence of TB, and a long follow-up period of 10 years. Both Spanish-born 17 (35.4%) 12 (48%) 0.324 sources of information are large and they provided important Foreign-born 31(64.6%) 13 (52%) high quality information. Moreover, our results were consistent Smoking 15 (31.2%) 9 (36%) 0.883 with the literature published elsewhere. Alcohol abuse 8 (16.7%) 4 (16%) 1.000 Further studies are required to investigate the interaction Homeless 3 (6.25%) 3 (12%) 0.406 of both diseases, the contribution of optimization of glycemic Unemployment 15 (32.6%) 4 (16.7%) 0.072 control, and to consider whether screening for both diseases in MEDEA (social risk) 1.55 (0.61) 1.88 (0.61) 0.048 specific areas could be cost effective. Death 7 (14.6%) 1 (3.85%) 0.247 Localization Pulmonary 28 (58.3%) 9 (36%) 0.169 CONCLUSION Extrapulmonary 17 (35.4%) 13 (52%) Both 3 (6.25%) 3 (12%) This large retrospective cohort study showed that in an inner- TST city district with high TB incidence and a high percentage of Unknown 1 (2.1%) 0 (0.0%) 0.844 migrants, DM was associated with an increased risk of developing Undone 25 (52.1%) 15 (60%) 5 (10.4%) 2 (8%) the disease. It is possible that the higher incidence was coupled Negative Positive 17 (35.4%) 8 (32%) with a higher TB transmission. Besides, TB in migrants can also be due to the reactivation of old infections, especially in people TST, tuberculin skin test. coming from Hindustan, where this disease is highly prevalent. The percentages are referred to the total of each column. Moreover, alcohol abuse was also identified as a risk factor for TB. In view of these results, we advocate for screenings to control diabetes-associated TB in the inner districts of high-income countries where the proportion of migrant people is high and We believe that our work may be important to study the characteristics such as low socioeconomic level further contribute association of the two diseases and make specific strategic plans in to increase the risk. areas of high TB concentration even in low-incidence countries. Prevention of DM and TB are two of the WHO’s main As with chronic complications of the disease, glycemic control objectives. Changes in lifestyle, unhealthy diets, globalization, a in DM patients probably influences both the risk (36) and the greater access to diagnosis and a longer life expectancy could prognosis of TB (37). More studies are needed to analyse how probably explain the increase of DM in countries where the the metabolic situation of DM can play a role in the different burden of TB is greater. presentation and localization of TB, in treatment outcomes We strongly recommend TB screening in DM patients in areas including mortality in patients with both conditions. with a high TB incidence. According to the CDC’s guidelines Some limitations can be found in our study. One of them for the diagnosis of LTBI (38), Tuberculin skin test (TST) or is the retrospective and observational nature of the study TB blood test can be used indistinctly in diabetic patients. TB design. All studies done with databases depend directly on the blood tests are recommended in people who are not likely to quality of the data record, which may be more worrisome in return for TST reading, have received the BCG vaccine, people retrospective studies. In addition, retrospective studies may have who are likely to be infected with M. tuberculosis and are at a more difficulty establishing a correct temporal relationship, and low to intermediate risk of progression to TB disease or people they have more selection and information biases, but we have who have a low probability of being infected with TB. In case of tried to minimize these biases with a careful methodology and using the TST in DM patients, a result of 5 mm or more should creating dynamic cohorts. Working with a population at risk be considered positive. of social exclusion may imply a risk of underreporting in key clinical factors. The 60.3% of DM without anti-DM treatment could be explained by a missing data in the databases or a lack DATA AVAILABILITY STATEMENT of adherence to treatments. Moreover, some of the cases were incident DM; therefore, with the exception of cases with a highly Publicly available datasets were analyzed in this study. This data elevated glycated hemoglobin, diet and exercise alone would have can be found here: https://jrealgatius.github.io/TBC_ANALISIS/ been recommended. However, the incidence rate shown was codi/shiny/DashBoard_TB.html. The data analysis code is shown similar to the rates reported in independent studies (25). The at https://github.com/jrealgatius/TBC_ANALISIS. measurement for glycated hemoglobin was sometimes lacking, as this is a retrospective study based in habitual medical practice. ETHICS STATEMENT Apart from this, subjects’ nationality, registered according to the Social Security data in the Primary Care medical record database, The studies involving human participants were reviewed and did not always correspond to the real origin of the subjects. approved by Ethics Committee of the Primary Healthcare Frontiers in Public Health | www.frontiersin.org 9 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients University Research Institute IDIAP-Jordi Gol (P16/023). FUNDING Written informed consent for participation was not required for this study in accordance with the national legislation and the This work was supported by two grants given by the institutional requirements. Instituto de Salud Carlos III PI16/01751 (Spanish Ministry of Economy) and the Institut Universitari per a la Recerca AUTHOR CONTRIBUTIONS a l’Atenció Primària de Salut Jordi Gol i Gurina (Catalan Health Institute) PREDOC_ECO-19/2. JJ was recipient of a VA-A, JC, JF-N, JR, ÀO, and AM-M participated in the study Miguel Servet contract (CPII18/00004) (Instituto de Salud design. ÀO, JD, and JR worked on data collection. JR, VA-A, EN, Carlos III). RP, and JF-N performed all statistical work. VA-A, JF-N, and JC were major contributors in writing the manuscript. MM-C, DM, ACKNOWLEDGMENTS JJ, JM, JB, and AM-M reviewed and corrected the manuscript. BV contributed to prepare the manuscript according to the journal The authors thank Amanda Prowse for a very helpful policies. All authors contributed to the article and approved the grammatical revision of the manuscript, and Susana Marcos for submitted version. the anthropological advice. REFERENCES 2011–2012. Diabetes Res Clin Pract [Internet]. (2018) 139:366–79. doi: 10.1016/j.diabres.2018.03.022 1. World Health Organization. Global Tuberculosis Report 2021. Available 14. Jeon CY, Murray MB. Diabetes mellitus increases the risk of active online at: https://www.who.int/teams/global-tuberculosis-programme/tb- tuberculosis: A systematic review of 13 observational studies. PLoS Med. reports/global-tuberculosis-report-2021 (accessed May 15, 2021). (2008) 5:1091–101. doi: 10.1371/journal.pmed.0050152 2. de Vries G, Aldridge RW, Caylã JA, Haas WH, Sandgren A, van Hest NA, 15. Moreno-Martínez A, Casals M, Orcau, Gorrindo P, Masdeu E, Caylà JA. et al. Epidemiology of tuberculosis in big cities of the European union Factors associated with diabetes mellitus among adults with tuberculosis in and European economic area countries. Eurosurveillance. (2014) 19:1–8. a large European city, 2000-2013. Int J Tuberc Lung Dis. (2015) 19:1507–12. doi: 10.2807/1560-7917.ES2014.19.9.20726 doi: 10.5588/ijtld.15.0102 3. Khan K, Hirji M, Miniota J, Hu W, Wang J, Gardam M, et al. Domestic impact 16. Baker MA, Harries AD, Jeon CY, Hart JE, Kapur A, Lönnroth K, of tuberculosis screening among new immigrants to Ontario, Canada. CMAJ. et al. The impact of diabetes on tuberculosis treatment outcomes: a (2018) 187:473–81. doi: 10.1503/cmaj.150011 systematic review. BMC Med [Internet]. (2011) 9:81. doi: 10.1186/1741- 4. Millet JP, Moreno A, Fina L, Baño L. Del, Orcau A, De Olalla PG, et al. Factors 7015-9-81 that influence current tuberculosis epidemiology. Eur Spine J. (2013) 22:8. 17. Huangfu P, Ugarte C, Pearson F, Golub J, Critchley J. The effects of doi: 10.1007/s00586-012-2334-8 diabetes on tuberculosis treatment outcomes: an updated systematic review 5. van Hest NA, Aldridge RW, de Vries G, Sandgren A, Hauer B, Hayward A, and meta-analysis. J Epidemiol Community Health. (2016) 70:A50.2-A51. et al. Tuberculosis control in big cities and urban risk groups in the European doi: 10.1136/jech-2016-208064.93 Union: a consensus statement. Eurosurveillance [Internet]. (2014) 19:1–13. 18. Salindri AD, Kipiani M, Kempker RR, Gandhi NR, Darchia L, Tukvadze N, doi: 10.2807/1560-7917.ES2014.19.9.20728 et al. Diabetes reduces the rate of sputum culture conversion in patients with 6. International Diabetes Federation. IDF Diabetes Atlas. 10th ed. International newly diagnosed multidrug-resistant tuberculosis. Open Forum Infect Dis. Diabetes Federation (2021). p. 1-150. Available online at: www.diabetesatlas. (2016) 3:1–10. doi: 10.1093/ofid/ofw126 org 19. Zhang S, Tong X, Wang L, Zhang T, Huang J, Wang D, et al. 7. Soriguer F, Goday A, Bosch-Comas A, Bordiú E, Calle-Pascual A, Clinical Characteristics and Prognostic Analysis of Patients With Carmena R, et al. Prevalence of diabetes mellitus and impaired glucose Pulmonary Tuberculosis and Type 2 Diabetes Comorbidity in China: regulation in Spain: the diabetes study. Diabetologia. (2011) 55:88–93. A Retrospective Analysis. Front Public Heal [Internet]. (2021) 9:1216. doi: 10.1007/s00125-011-2336-9 doi: 10.3389/fpubh.2021.710981 8. Ayelign B, Negash M, Genetu M, Wondmagegn T, Shibabaw 20. Perfil de la població estrangera a Ciutat Vella i els seus barris. Gener 2020 T. Immunological Impacts of Diabetes on the Susceptibility of [Internet]. Available online at: https://ajuntament.barcelona.cat/estadistica/ Mycobacterium tuberculosis. J Immunol Res [Internet]. (2019) 2019:1–8. catala/Estadistiques_per_temes/Poblacio_i_demografia/Documents_ doi: 10.1155/2019/6196532 relacionats/pobest/a2020/part1/nt27.htm (accessed September 15, 2021). 9. Segura-Cerda CA, López-Romero W, Flores-Valdez MA. Changes in host 21. Domínguez-Berjón MF, Borrell C, Cano-Serral G, Esnaola S, Nolasco A, response to mycobacterium tuberculosis infection associated with type 2 Pasarín MI, et al. Construcción de un índice de privación a partir de datos diabetes: beyond hyperglycemia. Front Cell Infect Microbiol. (2019) 4:342. censales en grandes ciudades españolas (Proyecto MEDEA). Gac Sanit. (2008) doi: 10.3389/fcimb.2019.00342 22:179–87. doi: 10.1157/13123961 10. Lee MR, Huang YP, Kuo YT, Luo CH, Shih YJ, Shu CC, et al. Diabetes mellitus 22. Subirana I, Sanz H, Vila J. Building Bivariate tables: The compareGroups and latent tuberculosis infection: a systemic review and metaanalysis. Clin package for R. J Stat Softw. (2014) 57:1–16. doi: 10.18637/jss.v057.i12 Infect Dis. (2017) 64:719–27. doi: 10.1093/cid/ciw836 23. Zhou B, Latouche A, Rocha V, Fine J. Competing risks regression 11. Yorke E, Atiase Y, Akpalu J, Sarfo-Kantanka O, Boima V, Dey ID. The for stratified data. Biometrics [Internet]. (2011) 67:661–70. bidirectional relationship between tuberculosis and diabetes. Tuberc Res Treat. doi: 10.1111/j.1541-0420.2010.01493.x (2017) 17:1–6. doi: 10.1155/2017/1702578 24. Situación actual del consumo de alcohol en España | Embarazadas: cero 12. Pearson F, Huangfu P, McNally R, Pearce M, Unwin N, Critchley alcohol. Available online at: https://www.sanidad.gob.es/campannas/ JA. Tuberculosis and diabetes: Bidirectional association in a UK campanas10/embarazadasCeroAlcohol/situacion_actual_del_consumo_de_ primary care data set. J Epidemiol Community Health. (2019) 73:142–7. alcohol.html doi: 10.1136/jech-2018-211231 25. Àngels O, Carmen GC, Joan A. Caylà i Buqueras CR i G, Recull. La tuberculosi 13. Barron MM, Shaw KM, Bullard KMK, Ali MK, Magee MJ. Diabetes a Barcelona. Informe. (2017). Available online at: https://www.aspb.cat/wp- is associated with increased prevalence of latent tuberculosis infection: content/uploads/2020/10/tuberculosi-barcelona-2017.pdf (accessed accessed Findings from the National Health and Nutrition Examination Survey, May 15, 2021). Frontiers in Public Health | www.frontiersin.org 10 May 2022 | Volume 10 | Article 789952 Antonio-Arques et al. Tuberculosis Risk in Diabetic Patients 26. Leung CC, Lam TH, Chan WM, Yew WW, Ho KS, Leung GM, et al. Diabetic 35. Sotgiu G, Falzon D, Hollo V, Ködmön C, Lefebvre N, Dadu A, control and risk of tuberculosis: a cohort study. Am J Epidemiol. (2008) et al. Determinants of site of tuberculosis disease: An analysis of 167:1486–94. doi: 10.1093/aje/kwn075 European surveillance data from 2003 to 2014. PLoS One. (2017) 12:1–14. 27. Kamper-Jørgensen Z, Carstensen B, Norredam M, Bygbjerg IC, Andersen PH, doi: 10.1371/journal.pone.0186499 Jørgensen ME. Diabetes-related tuberculosis in Denmark: Effect of ethnicity, 36. Lee PH, Fu H, Lee MR, Magee M, Lin HH. Tuberculosis and diabetes in low diabetes duration and year of diagnosis. Int J Tuberc Lung Dis. (2015) 19:1169– and moderate tuberculosis incidence countries. Int J Tuberc Lung Dis. (2018) 75. doi: 10.5588/ijtld.14.0932 22:7–16. doi: 10.5588/ijtld.17.0329 28. Ponce-De-Leon A, Garcia-Garcia MDL, Garcia-Sancho MC, Gomez- 37. Song C, Xie W, Gong L, Ren M, Pan P, Luo B. The relationship between Perez FJ, Valdespino-Gomez JL, Olaiz-Fernandez G, et al. Tuberculosis HbA1c control levels and antituberculosis treatment effects: A meta-analysis. J and diabetes in Southern Mexico. Diabetes Care. (2004) 27:1584–90. Chinese Med Assoc. (2019) 82:915–21. doi: 10.1097/JCMA.0000000000000205 doi: 10.2337/diacare.27.7.1584 38. Centers for Disease Control and Prevention. Latent Tuberculosis Treatment 29. Ugarte-Gil C, Curisinche M, Herrera-Flores E, Hernandez H, Rios Guidelines: 2020 Update (2020). Available online at: https://www.cdc.gov/tb/ J. Situation of the tuberculosis-diabetes comorbidity in adults in publications/ltbi/pdf/LTBIbooklet508.pdf Peru: 2016–2018. Rev Peru Med Exp Salud Publica. (2021) 38:254–60. doi: 10.17843/rpmesp.2021.382.6764 Conflict of Interest: The authors declare that the research was conducted in the 30. Gezahegn H, Ibrahim M, Mulat E. Diabetes mellitus and tuberculosis absence of any commercial or financial relationships that could be construed as a comorbidity and associated factors among bale zone health institutions, potential conflict of interest. Southeast Ethiopia. Diabetes, Metab Syndr Obes Targets Ther. (2020) 13:3879–86. doi: 10.2147/DMSO.S248054 Publisher’s Note: All claims expressed in this article are solely those of the authors 31. Ködmön C, Zucs P, van der Werf MJ. Migration-related and do not necessarily represent those of their affiliated organizations, or those of tuberculosis: Epidemiology and characteristics of tuberculosis the publisher, the editors and the reviewers. Any product that may be evaluated in cases originating outside the European Union and European this article, or claim that may be made by its manufacturer, is not guaranteed or Economic Area, 2007 to 2013. Eurosurveillance. (2016) 21:1–10. doi: 10.2807/1560-7917.ES.2016.21.12.30164 endorsed by the publisher. 32. Nadal JF, Peinado RM, Calafell NR, Aparicio CC, Martos AMM, Griera JM. Diabetes mellitus en inmigrantes indostanos jóvenes. Un estudio descriptivo Copyright © 2022 Antonio-Arques, Franch-Nadal, Moreno-Martinez, Real, Orcau, Endocrinol y Nutr. (2008) 55:454–8. doi: 10.1016/S1575-0922(08)75842-X Mauricio, Mata-Cases, Julve, Navas Mendez, Puig Treserra, Barrot de la Puente, 33. Mohan V, Jaydip R, Deepa R. Type 2 diabetes in Asian Indian youth. Millet, Del Val García, Vlacho and Caylà. This is an open-access article distributed Pediatr Diabetes [Internet]. (2007) 8:28–34. doi: 10.1111/j.1399-5448.2007. under the terms of the Creative Commons Attribution License (CC BY). The use, 00328.x distribution or reproduction in other forums is permitted, provided the original 34. Antony SJ, Harrell V, Christie JD, Adams HG, Rumley RL. Clinical differences author(s) and the copyright owner(s) are credited and that the original publication between pulmonary and extrapulmonary tuberculosis: a 5-year retrospective in this journal is cited, in accordance with accepted academic practice. No use, study. J Natl Med Assoc. (1995) 87:187–92. distribution or reproduction is permitted which does not comply with these terms. Frontiers in Public Health | www.frontiersin.org 11 May 2022 | Volume 10 | Article 789952