Epidemiology of latent tuberculosis infection among patients with and without diabetes mellitus

Transcription

1 Family Practice, 2017, Vol. 34, No. 5, doi: /fampra/cmx017 Advance Access publication 23 March 2017 Epidemiology Epidemiology of latent tuberculosis infection among patients with and without diabetes mellitus Yogarabindranath Swarna Nantha*, Anurupa Puri, Siti Zubaidah Mohamad Ali, Poopathy Suppiah, Siti Aishah Che Ali, Bhavani Ramasamy and Intan Masni Ibrahim Seremban Primary Care Clinic, Jalan Rasah, Seremban, Negeri Sembilan Darul Khusus, Malaysia. *Correspondence to Yogarabindranath Swarna Nantha, Non-Communicable Disease Clinic, Klinik Kesihatan Seremban, Jalan Rasah, Seremban, Negeri Sembilan Darul Khusus, Malaysia; Abstract Background. Individuals with type 2 diabetes mellitus (DM) are at a greater risk of tuberculosis (TB) reactivation. There is a paucity of information about the risk factors associated with latent tuberculosis infection (LTBI) in patients with diabetes. Objective. We conducted an observational study to compare the prevalence and risk factors associated with LTBI factors in Malaysian adults with and without DM. Methods. Four hundred and four patients with DM and 359 patients with non-dm at a regional primary care clinic were recruited as participants in this case control study. The tuberculin sensitivity test (TST) was performed. The presence of LTBI was defined by a TST value of 10 mm in DM patients and 10 mm in the non-dm group. A logistic regression model was used to identify variables associated with LTBI. Results. There was no statistical significant difference in the prevalence rates seen between the DM and non-dm group of the study. LTBI prevalence among patients with DM was 28.5%. The proportion of patients in the non-dm group with LTBI was 29.2%. When a critical cut-off of 8 mm was used, the adjusted odds ratio of LTBI in DM patients was 1.88 (95% confidence interval: ). Smoking was an independent risk factor for LTBI regardless of DM status. HbA1c levels or anthropometric measurements were not associated with LTBI in diabetic patients. Conclusions. There is no significant risk of contracting LTBI in DM patients using the standard 10- mm TST cut-off. Nonetheless, using lower cut-offs in a DM population appear valid. Smoking is an important predictor of LTBI. Key words: Diabetes, latent tuberculosis, primary care, screening, smoking, tuberculin test. Introduction In recent years, there has been a steady increase in the incidence of type 2 diabetes mellitus (DM) in Malaysia (1,2). A parallel rise was also witnessed in the number of tuberculosis (TB) cases in the country (3). These trends have been attributed to the accelerated reactivation of latent TB infection (LTBI) (4). However, there is a scarcity of information on the prevalence rates of LTBI to guide specific intervention, especially within high-risk groups such as individuals with diabetes. Considerable research effort has been undertaken to establish the link between DM and active TB (5). On the other hand, little emphasis has been given to the definitive determination of the influence of DM on LTBI (5). Several studies report high prevalence rates of LTBI among patients with DM (6 8). There is also evidence showing The Author Published by Oxford University Press. All rights reserved. For permissions, please journals.permissions@oup.com. 532

2 Prevalence and comparison of risk factors for latent tuberculosis infection 533 significant correlation between LTBI and hyperglycaemia (6 8). From a regional perspective, a study in Singapore has verified the prevalence rates of LTBI among individuals with DM at a tertiary care centre (9). However, the findings from these studies remain inconclusive. In the absence of a control group (low-risk population in the community), little meaningful comparison can be drawn from these results (5). Routine screening and treatment is recommended in patients with a concurrent diagnosis of diabetes mellitus and LTBI in developed nations (where the burden of TB is low). This primary preventive measure eradicates the possibility of the emergence of active TB in high-risk groups (10). In Malaysia (a country with an intermediate burden of TB in the population), the detection and treatment of patients with LTBI remain controversial (4). In tandem with the rise in the burden of diabetes, a research-orientated approach is needed to gauge the feasibility of prophylactic treatment of vulnerable populations in the society (4). The objective of this study is 2-fold. The principal aim of the study is to determine the LTBI prevalence rates in a DM and non- DM population at a community level. The second objective is to evaluate the strength of various factors related to the reactivation of LTBI. The outcome of this study could provide better insight into the magnitude of the hidden reservoir of LTBI in a population who are at a greater risk for reactivation of TB. Research design and methods Study population The study involves the assessment of the reaction to tuberculin sensitivity test (TST) in a randomly selected sample receiving treatment at an urban primary care clinic between October 2014 and December A case control study design was adopted, and a purposive sampling technique was employed to identify eligible patients at the non-communicable disease clinic at the Seremban Primary Health Clinic, a regional primary care clinic in the state of Negeri Sembilan, Malaysia. The sample comprised an adult population aged 18 and older seeking consultation at the primary care clinic. Based on the annual clinic appointment list for the year 2014 and 2015, a total of 842 records of patients (all from the district of Seremban) were reviewed. Consequently, a nominal list of eligible participants was generated to allow the random selection of patients with and without diabetes. Patients who refused to participate were randomly replaced with other available participants on the nominal list. The participants of the study were classified into two distinct categories the DM and non-dm group. The DM arm encompasses DM patients seeking treatment at the primary care setting with documented HbA1c levels. The non-dm arm consisted of patients who had been screened to rule out pre-dm and DM states. Participants in both arms were excluded if they (i) have received previous Mantoux testing, (ii) are diagnosed having active TB infection or previous known TB infection, (iii) have underlying cognitive impairment, (iv) recent exposure to TB, (v) are health care workers, (vi) are diagnosed with Hepatitis B and C markers, (vii) received immunoglobulins, corticosteroids or immunosuppressants, (viii) have been diagnosed with chronic kidney disease/end stage renal failure, (ix) HIV infection, (x) are undergoing chemotherapy for cancer and (xi) are pregnant. Study design The case control study involved the concurrent recruitment of both DM and non-dm patients from the non-communicable disease department at the clinic (Fig. 1). They were then randomly selected to undergo tuberculin testing. Once informed consent was obtained, the investigator interviewed and examined the participants to acquire information on socio-demographic data, history of present medical illness, known previous history of TB, the presence of Bacille Calmette-Guerin (BCG) vaccination scar, history of smoking and history of TB contact. This cursory step was designed to rule out the presence of active TB and other potential risk factors for reactivation. The medical records of each patient were also reviewed to ensure the accuracy and the completeness of the information given. Anthropometric measurements that included height (in centimetre) and weight (in kilogram) were assessed with the participants wearing light clothing with shoes. Body mass index (BMI) was calculated as weight (in kilogram) divided by height (in metre squared). Two trained personnel performed all measurements. Details relevant to mean blood glycosylated haemoglobin (HbA1c) and serum creatinine levels were also obtained from these records. Measurements for serum creatinine and fasting/random blood sugar levels were performed in the Cobas 6000 Analyzer Series (Roche Diagnostic GmbH, Germany). Mean blood glycosylated haemoglobin (HbA1c level was determined by the D-10 HbA1c test (Bio-Rad Laboratories). The TST was carried out by a research nurse trained in the administration of the test. The patients were injected with purified protein derivative (PPD) of 0.1 ml (Tuberculin PPD RT 23 SS2 T.U/0.1 ml, Statens Serum Institut Denmark) in accordance to local guidelines (11) on the volar surface of the arm below the antecubital fossa. The injected area was marked with a red waterproof marker. The reaction to the TST was read at hours as millimetre of induration using the ballpoint pen method (12). The induration was measured perpendicular to the long axis of the forearm. The principal investigator was responsible for reading all TST results. In order to minimize bias, administration and the reading of TST in all patients were standardized in accordance to CDC guidelines in the form of a wall chart in the TST administration/interpretation room (13). An induration that read 10 mm was considered a positive tuberculin reaction for both DM and non-dm participants (11). All patients were informed of the consequence of the TST results. Patients with negative tests were instructed to continue their routine follow-up at the clinic. They were informed that they do not require further assessment as their test results were normal. Patients with a positive test were informed of the results and were recommended to perform a chest X-ray and sputum acid-fast bacilli (if symptomatic) to rule out the possibility of underlying active TB. All chest X-rays were reviewed and reported by a team of radiologists who were blinded to the TST results. This step was done to ensure accuracy of the interpretation of the chest X-ray films. Once active TB has been excluded by performing these tests, the patients were followed up via scheduled appointments for 5 10 years. This surveillance strategy allows both the principal investigator and a family medicine specialist to monitor for the progression of LTBI to active TB among the study participants. Data analysis The sample size was calculated to achieve 95% power at the 5% significance level. An appropriate sample size of 314 patients was calculated based on the prevalence of latent TB in DM patients (28.2%) (9). All statistical analyses were done with 95% confidence interval (CI), and the level of significance was set at P < The statistical interpretation of the data was performed using Statistical Package for Social Sciences (SPSS version 20; IBM Corp.). Continuous data were described as mean, median and standard deviation (SD). All descriptive data were reported as mean and percentages. The chi-square and independent t-tests were carried out to compare variables between the DM and non-dm population. The

3 534 Family Practice, 2017, Vol. 34, No. 5 Figure 1. Flow chart of the study design and process. definition of a positive tuberculin reaction was 10 mm for the DM group and 10 mm for the non-dm group (10). The association between the induration size and the clinical variables affecting TST was assessed using Pearson and Spearman-rank order correlation analysis. Multivariate logistic regression was used to identify the predictors of a positive tuberculin reactivity. The positive predictive value (PPV) of the TST was determined through the utilization of a TST/interferon-γ-release assays (IGRA) interpreter (14). Results A total of 842 patients were recruited between December 2014 and August The characteristics of the population with and without diabetes are summarized in Table 1. Out of the 415 DM participants of the study, 404 patients were chosen. The drop-out rate was 3.1% (n = 13). In contrast, the final number of participants in the non-dm group was 359, with a larger drop-out rate of 16.3% (n = 68). The reasons for the patient drop out were (i) refusal to participate in the

4 Prevalence and comparison of risk factors for latent tuberculosis infection 535 study (n = 4, 4.9%), (ii) failure to return for TST interpretation/reading (n = 6, 7.4%), (iii) failure to attend TST appointments (n = 18, 22.2%) and (iv) failure to meet the eligibility criteria (n = 53, 65.4%). The mean age for both groups was 61.7 and 63.0, respectively. The mean BMI for both groups were 27.1 and 25.7 kg/m 2. There were a higher number of males in the non-dm population (57.4%) and more females in the DM group (58.2%). Majority of the patients in both groups were from Chinese descent (DM: 43.8%, non-dm: 68.8%) followed by Indians (DM: 40.8%, non-dm: 16.2%) and Malays (DM: 15.1%, non-dm: 13.4%). The distribution of BCG vaccination in both groups (evidenced by visible vaccination scars) was >98% as all these patients were born in the period after the introduction of BCG vaccination in Malaysia. No participants had either recent contact with patients with TB or history of known previous TB infection. About 9.4% (n = 38) of the DM population and 16.7% (n = 60) of the non-dm population were current smokers. The mean duration of treatment of diabetes was 9.5 years. The majority of the DM population (60.6%, n = 245) were treated with oral anti-hypoglycaemics (OHA). None of these patients had any significant changes on the chest X-ray suggestive of active TB. Taking into account the factors that influence TST reactivity, the PPV of the TST for both groups was >80%. The characteristics of DM, non-dm and the LTBI population are shown in Tables 2 4. There were 115 cases of LTBI among the 404 members of the DM group. The prevalence of LTBI in the DM group was 28.5% compared to 29.2% in the non-dm sample (P = 0.812). The adjusted odds ratio (aor) for both groups was not statistically significant (95% CI: ) with a P-value of A TST critical value of 8 mm in the DM group was sufficient enough to cause a statistically significant difference when compared with the non-dm group (P = 0.002). The aor among patients with diabetes using this threshold was 1.88 (95% CI: ). In the non-dm group, the proportion of the male population with a positive TST reaction was greater compared to females (35.0% versus 21.6%) with a P-value of A greater percentage of men were also diagnosed with LTBI in the DM group (31.4% versus 26.4%,). However, this finding failed to achieve statistical significance (P = 0.274). There was a significant inverse relationship between age and tuberculin reactivity in both persons with diabetes and without diabetes (P < 0.05). Ex-smokers and current smokers were more likely to have a positive TST than never smokers in both groups (P < 0.001). In ever smokers, it was not clear if there was a trend towards a dose relationship between the proportion with TST reactivity and the number of pack-years smoked in both populations. On the other hand, there is some evidence that smoking may affect the possibility of being diagnosed with LTBI in both DM and non-dm group (P = 0.05). BMI did not have a significant association with TST reactivity in both groups (P = 0.816). In DM individuals, there was no significant difference in HbA1c levels between positive and negative Mantoux results (P = 0.787). Likewise, there was no difference in the positive TST reactions among OHA and insulin users in the DM population (P = 0.994). A stepwise multiple logistic regression analysis was carried out to identify the significant determinants of tuberculin reactivity. The model contained six independent variables (gender, age, ethnicity, BMI, DM status and smoking habit). The full model containing all predictors was statistically significant, χ 2 (5, N = 552) = 51.0, P < The important predictors of LTBI status were age and smoking habit (P < 0.001). It was found that being a smoker (ex and current in both groups) increased the risk of a reactive TST by 4-fold (P < 0.001). Conversely, factors such as BMI, age group, ethnicity and gender were poor predictors of LTBI (P > 0.38). When an 8-mm TST cut-off was used for the diabetic group, both DM status and smoking habit increased the risk of a positive TST reaction by 2- and 4-folds, respectively (P > 0.003). Discussion In Malaysia, efforts have been made to gauge the prevalence of LTBI in several high-risk groups. The prevalence rates among these groups (except health care workers) were high, ranging between 53% and 88% (4). Nevertheless, there is limited data on the prevalence estimates of LTBI specifically targeted to persons with diabetes. In our study, the prevalence of LTBI among patients with diabetes (28.5%) was almost similar to another study undertaken at a tertiary care centre (28.2%) (9). However, the researchers in the study by Leow et al. (9) did not recruit a low-risk sample to allow any comparison to be made with regards to the prevalence rates found in persons with and without diabetes. Recent guidelines suggest that TST reactions in persons vaccinated with BCG should be interpreted using the same criteria for those who were not BCG vaccinated (15). When a TST definition of 10 mm was used for both DM and non-dm group, there was no difference in LTBI rates in both groups (Table 4) (10,15). Conversely, when a TST definition of 8 mm in individuals with diabetes and 10 mm in the non-dm population was used, we found higher rates of LTBI in the DM population. These results indicate people with diabetes do have a certain degree of immunosuppression. They are also likely to have smaller responses to the TST as well as a higher risk of progression in active TB. Therefore, using a lower cut-off for persons with diabetes is valid. In this study, smoking was found to be an independent risk factor associated with LTBI. Smokers also had an almost equal risk of contracting LTBI regardless of DM status. These results are in line with findings described in other large population-based studies (16,17). Also, studies have shown that the risk of reactivation of LTBI in a smoking adult is 2-fold that of a never smoker (16). Comparison with literature Two factors were of particular interest in our study in terms of clinical parameters. In tandem with the findings of a similar study (18), BMI status in DM patients was not associated with an increased risk of LTBI. Second, HbA1c levels among persons with diabetes in our study poorly correlated to LTBI. On the other hand, a study conducted in Mexico discovered that HbA1c levels were associated with LTBI in DM patients (18). In that study, the mean HbA1c was recorded at 7.8%, comparable to the HbA1c levels found among DM participants in our study (7.7%) (18). In line with previous studies (19,20), our study found that age was inversely correlated with TST measurements. While the mean age of participants in the study performed by Matinez-Aguilar et al. (18) was 53.4 years, the participants in our study were much older with a mean age of 61.7 years. Thus, age differences among DM participants in both studies could be a possible explanation for the discrepancy between HbA1c and LTBI status. Strengths and limitations This analysis would be the first in Malaysia designed to determine the prevalence rate of LTBI between two populations in the community a DM and non-dm group. This study utilized a group of participants that represent LTBI in patients without diabetes as a benchmark for valid comparative assessment of the prevalence rates of LTBI found in persons with diabetes. Meticulous effort was made to control or ensure the baseline population was devoid of other risk factors related to reactivation of LTBI.

5 536 Family Practice, 2017, Vol. 34, No. 5 Table 1. Baseline characteristics of the study population Subject characteristics Type 2 diabetes mellitus sample (n = 404) Non-type 2 diabetes mellitus sample (n = 359) Gender, n (%) Male 169 (41.8) 206 (57.4) Female 235 (58.2) 153 (42.6) Mean age in years (SD) 61.7 (9.78) 63.0 (10.0) Age categories, n (%) <35 3 (0.7) 1 (0.3) (20.3) 68 (18.9) (69.3) 237 (66.0) (9.2) 51 (14.2) Unspecified 2 (0.5) 2 (0.6) Mean BMI (kg/m 2 ) BMI categories, n (%) Underweight 2 (0.5) 9 (2.5) Normal 96 (23.8) 102 (28.4) Overweight 135 (33.4) 107 (29.8) Obese 66 (16.3) 36 (10.0) Unspecified 105 (26%) 105 (29.2) Ethnicity, n (%) Malay 61 (15.1) 48 (13.4) Chinese 177 (43.8) 247 (68.8) Indian 165 (40.8) 58 (16.2) Others 1 (0.2) 6 (1.7) Previous BCG vaccination, n (%) Scar present 397 (98.3) 357 (99.4) Not present 7 (1.7) 2 (0.6) History of recent tuberculosis contact, n (%) No 404 (100) 359 (100) Yes 0 (0) 0 (0) Comorbidities, n (%) Hypertension 335 (82.9) 343 (95.5) Dyslipidaemia 272 (67.3) 169 (47.1) Chronic kidney disease 1 (0.25) 1 (0.29) Ischaemic heart disease 9 (2.2) 17 (4.7) Cerebrovascular accident 0 (0) 5 (1.39) Chronic obstructive airway disease 2 (0.50) 1 (0.29) Gout 3 (0.74) 6 (1.67) Bronchial asthma 4 (1.0) 6 (1.67) Osteoarthritis 1 (0.25) 1 (0.29) Thyroid disorder 4 (1.0) 8 (2.23) Atrial fibrillation 2 (0.50) 0 (0) Fatty liver disease 0 (0) 1 (0.29) Hepatitis B infection 0 (0) 1 (0.29) Benign prostatic hyperplasia 0 (0) 7 (1.95) Type of diabetic treatment, n (%) Diet control 5 (1.2) OHA a 245 (60.6) Insulin only 15 (3.7) OHA + insulin 100 (24.8) Unspecified 39 (9.7) Mean duration of treatment for diabetes in years (SD) 9.5 (6.7) Smoking status, years Mean pack years Mean creatinine levels in µmol/l (SD) b 80.0 (52.0) Mean egfr levels (SD) b 81.7 (22.4) Mean HbA1c levels in % (SD) a 7.7 (1.65) egfr = estimated glomerular filtration rate. a Missing n = 7. b Missing n = 5. LTBI in the study population was solely detected using TST cut-offs in agreement with international and locally accepted diagnostic guidelines (10,11). The decision to use various TST cut-offs in the diagnosis of LTBI is influenced by false-positive reactions related to BCG vaccinations and non-tb mycobacterium infection (15). Therefore, in this study, an estimation of the PPV of TST readings was performed using

6 Prevalence and comparison of risk factors for latent tuberculosis infection 537 Table 2. Comparison of latent tuberculosis infection in non-diabetic and diabetic patients using various tuberculin sensitivity test cut-offs Variable Population with latent tuberculosis (all) 8/ 10 a 10/ 10 b Latent tuberculosis in population with type 2 diabetes b Latent tuberculosis in population without type 2 diabetes b Prevalence (%) 256 (33.6) 220 (28.8) 115 (52.3) 105 (47.7) Gender, n (%) Male 144 (56.2) 125 (56.8) 53 (46.1) 72 (68.6) Female 112 (43.8) 95 (43.2) 62 (53.9) 33 (31.4) Age groups, years <35 2 (1.1) 2 (0.9) 2 (1.7) 0 (0.0) (24.4) 65 (28.1) 29 (25.2) 25 (23.8) (65.9) 143 (61.9) 72 (62.6) 71 (67.6) (7.4) 18 (7.8) 10 (8.7) 8 (7.6) Unspecified 2 (1.1) 3 (1.3) 2 (1.7) 1 (1.0) BMI (kg/m 2 ) Underweight 3 (1.7) 4 (1.7) 2 (1.7) 1 (1.0) Normal 46 (26.1) 61 (26.4) 29 (25.2) 31 (29.5) Overweight 48 (27.2) 61 (26.4) 33 (28.7) 27 (25.7) Obese 27 (15.3) 34 (14.7) 21 (18.3) 13 (12.4) Unspecified 52 (29.5) 71 (30.7) 30 (26.1) 33 (31.4) Smoking status c Never smoker 124 (70.5) 149 (67.7) 92 (80.0) 57 (54.3) Ex-smoker 7 (4.0) 12 (5.5) 1 (0.9) 21 (20.0) Current smoker 43 (24.4) 57 (25.9) 20 (17.4) 27 (25.7) Mean Mantoux reading (mm) 13.7 (3.9) 14.6 (3.5) 14.1 (3.7) 15.1 (3.7) a Diagnostic TST criteria for LTBI shown: 8 mm in patients with diabetes and 10 in patients without diabetes. b Diagnostic TST criteria for LTBI shown: 10 mm in patients with diabetes and patients without diabetes. c Missing n = 9. Table 3. Logistic regression predicting likelihood of reporting latent tuberculosis infection for different tuberculin sensitivity test cut-offs Predictor Tuberculin sensitivity test cut-off 8/10 mm Tuberculin sensitivity test cut-off 10/10 mm β Standard error B Adjusted odds ratio β Standard error B Adjusted odds ratio Gender (male) Age Ethnicity Malay Chinese Indian BMI DM * Smoker (current + ex-smoker) * Constant χ d.f. 1 1 d.f., degrees of freedom. *P < P < Table 4. Percentages and adjusted odds ratio for different latent tuberculosis infection diagnostic criteria Tuberculin sensitivity test cut-offs N (%) Adjusted odds ratio P-value (95% confidence interval) a Type 2 diabetes mellitus Non-type 2 diabetes mellitus 10 mm for DM and 10 for non-dm 115 (52.3) 105 (47.7) 1.23 ( ) mm for DM and 10 mm for non-dm 151 (59.0) 105 (41.0) 1.88 ( ) <0.001 a Adjusted for gender, age, ethnicity, smoking status and BMI.

7 538 Family Practice, 2017, Vol. 34, No. 5 a cut-off of 10 mm to validate the TST results in our setting. The PPV for TST in this study (>80% for both groups with and without diabetes) was higher when compared to a similar study conducted among healthy non-bcg vaccinated participants (PPV of 75%) (21). In areas of intermediate to higher endemicity such as Malaysia, commercial IGRA seems to be a more accurate method of LTBI detection in high risk or immunosuppressed individuals (22). Thus, one important limitation in our study was the absence of concurrent testing of patients utilizing IGRA, which would have allowed diagnostic comparisons to be made. Nevertheless, it is widely known that both TST and IGRA are acceptable but imperfect tests in the detection of LTBI (23,24). Furthermore, IGRA is not a cost-effective tool in a nationwide screening intervention involving high-risk groups. An economically viable option for the screening and treatment of LTBI in high-risk groups would be to perform IGRA to confirm a positive tuberculin result (22,25,26). In this study, we confirmed the statistical significance of our results by comparing the LTBI levels found in the non-dm group (Table 3). However, the outcome of both of studies was gauged based on a one-step TST testing among DM patients. Since the waning reactivity in both DM and control group population could be recalled using repeat tuberculin testing (two-step testing), the prevalence rates in this study could likely be an underestimation of the actual prevalence of LTBI in the study population (15,27). Conclusion This study is a preliminary analysis of a more protracted longitudinal study. The next phase of the study involves the active surveillance of these patients over the next 5 10 years. The data from the analysis will allow a better estimation of the conversion rates of LTBI to clinically overt TB. The results could influence the need to undertake an interventional study assessing the efficacy of anti-tb prophylaxis in DM patients. Subsequently, the outcome of the study would be able to provide justification for or against recommending the initiation of anti-tb prophylaxis among DM patients. Acknowledgements We thank Dr Anna Ralph at Menzies School of Health, Darwin, Australia, for her insightful feedback that helped render this article in its current form. We also thank the medical and administrative staff at the Seremban District Health Office, namely Dr Paid Yusof, Dr Adnan Ananth and Ms Sivaneswari for their help in conducting TST testing as a part of disease surveillance and case-finding opportunity. We also convey our special thanks to Assoc. Prof. Lim Khean Ghee of the International Medical University who was instrumental in raising awareness about the issue discussed in this article. Declaration Funding: departmental resources. Ethical approval: Malaysian Medical Research Ethics Committee (NMRR ). Conflict of interest: none. References 1. National Health and Morbidity Studies. Prevalence of Diabetes Mellitus in Malaysia 2006: Results of the 3rd National Health and Morbidity Survey (NHMS III). Putrajaya, Malaysia: Ministry of Health Malaysia, Letchuman GR, Wan Nazaimoon WM, Wan Mohamad WB et al. Prevalence of diabetes in the Malaysian National Health Morbidity Survey III Med J Malaysia 2010; 65: World Health Organization (WHO). Global Health Observatory Data Repository: Tuberculosis, Mortality and Prevalence. WHO apps.who.int/ghodata/# (accessed on 31 January 2014). 4. Swarna Nantha Y. A review of tuberculosis research in Malaysia. Med J Malaysia 2014; 69: Baghaei P, Marjani M, Javanmard P, Tabarsi P, Masjedi MR. Diabetes mellitus and tuberculosis facts and controversies. J Diabetes Metab Disord 2013; 12: Vega RA, Conde JG, Díaz M. Prevalence of tuberculin reactivity and prevalence of risk factors for the development of active tuberculosis in a nursing home in Puerto Rico. P R Health Sci J 1996; 15: Mansilla Bermejo MJ, Sanz Gil MJ, Moraleda Velasco P et al. Tuberculin test in diabetic patients in a health center. Aten Primaria 1995; 16: Nwabudike LC, Ionescu-Tîrgovişte C. Intradermal reactions to purified protein derivative in patients with diabetes mellitus. Rom J Intern Med 2005; 43: Leow MK, Dalan R, Chee CB et al. Latent tuberculosis in patients with diabetes mellitus: prevalence, progression and public health implications. Exp Clin Endocrinol Diabetes 2014; 122: CDC. Latent Tuberculosis Infection: A Guide for Primary Health Care Providers, 1st edn. Atlanta, GA: CDC, Ministry of Health Malaysia. Clinical Practice Guidelines Management of Tuberculosis, 3rd edn. Kuala Lumpur, Malaysia: Ministry of Health Malaysia, Sokal JE. Editorial: measurement of delayed skin-test responses. N Engl J Med 1975; 293: Centers for Disease Control and Prevention. Mantoux Tuberculin Skin Test Wall Chart. Atlanta, GA: CDC, (accessed on 25 October 2014). 14. Menzies D, Gardiner G, Farhat M et al. The Online TST/IGRA Interpreter (accessed on 15 November 2016). 15. Centers for Disease Control and Prevention. National Center for HIV/ AIDS, Viral Hepatitis, STD and TPD of TE (eds). Core Curriculum on Tuberculosis: What the Clinician Should Know, 6th edn. Atlanta, GA: Centers for Disease Control and Prevention, Lin HH, Ezzati M, Murray M. Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and meta-analysis. PLoS Med 2007; 4: e Horne DJ, Campo M, Ortiz JR et al. Association between smoking and latent tuberculosis in the U.S. population: an analysis of the National Health and Nutrition Examination Survey. PLoS One 2012; 7: Martínez-Aguilar G, Serrano CJ, Castañeda-Delgado JE et al. Associated risk factors for latent tuberculosis infection in subjects with diabetes. Arch Med Res 2015; 46: Weiskopf D, Weinberger B, Grubeck-Loebenstein B. The aging of the immune system. Transpl Int 2009; 22: Rajagopalan S, Yoshikawa TT. Tuberculosis in the elderly. Z Gerontol Geriatr 2000; 33: Berkel GM, Cobelens FG, de Vries G, Draayer-Jansen IW, Borgdorff MW. Tuberculin skin test: estimation of positive and negative predictive values from routine data. Int J Tuberc Lung Dis 2005; 9: Diel R, Nienhaus A, Loddenkemper R. Cost-effectiveness of interferongamma release assay screening for latent tuberculosis infection treatment in Germany. Chest 2007; 131: Campion EW, Getahun H, Matteelli A et al. Latent Mycobacterium tuberculosis infection. N Engl J Med 2015; 372: Pai M, Denkinger CM, Kik SV et al. Gamma interferon release assays for detection of Mycobacterium tuberculosis infection. Clin Microbiol Rev 2014; 27: Diel R, Nienhaus A, Lange C, Schaberg T. Cost-optimisation of screening for latent tuberculosis in close contacts. Eur Respir J 2006; 28: Wrighton-Smith P, Zellweger JP. Direct costs of three models for the screening of latent tuberculosis infection. Eur Respir J 2006; 28: Menzies D. Interpretation of repeated tuberculin tests. Boosting, conversion, and reversion. Am J Respir Crit Care Med 1999; 159: