The role of diabetes on the clinical manifestations of pulmonary tuberculosis


Corresponding Author Daniel Faurholt-Jepsen, Department of Human Nutrition, University of Copenhagen, Rolighedsvej 30, 1958 Frederiksberg C, Denmark. Tel.: +45 35 33 22 29; E-mail:


Objective  Diabetes is associated with pulmonary tuberculosis (TB), possibly due to impaired immunity, and diabetes may exacerbate the clinical manifestations of TB. Our aim was to assess the role of diabetes in the clinical manifestations of TB.

Methods  We studied 1250 patients with pulmonary TB in an urban population in a cross-sectional study in Tanzania. All participants were tested for diabetes and HIV co-infection, and TB culture intensity was assessed. Levels of white blood cells, haemoglobin, acute phase reactants, CD4 count and HIV viral load were measured, and a qualitative morbidity questionnaire was used to identify the prevalence of disease-related symptoms.

Results  Tuberculosis patients with diabetes had a higher neutrophil count (B 0.5 × 109 cells/l, 95% CI 0.2; 0.9, P = 0.001) than non-diabetic TB patients. Serum C-reactive protein (B 18.8 mg/l, CI 95% 8.2; 29.4, P = 0.001) and alpha-1-acid glycoprotein (B 0.2 g/l, CI 95% 0.03; 0.3, P = 0.02) were similarly higher in patients with diabetes. Diabetes did not affect culture intensity or HIV status, but self-reported fever was three times higher among participants with diabetes than in those without diabetes (OR 2.9, CI 95% 1.5; 5.7, P = 0.002).

Conclusion  Diabetes is associated with small changes in the manifestations of TB, but may have little clinical significance.


Objectif:  L’association entre le diabète et la tuberculose (TB) pulmonaire est probablement liée à un déficit immunitaire et le diabète peut aggraver les manifestations cliniques de la TB. Notre objectif était d’évaluer le rôle du diabète dans les manifestations cliniques de la TB.

Méthodes:  Nous avons étudié 1250 patients atteints de TB pulmonaire dans une population urbaine dans le cadre d’une étude transversale en Tanzanie. Tous les participants ont été testés pour le diabète et la coinfection VIH, et le rendement de la culture de TB a étéévalué. Les taux de globules blancs, l’hémoglobine, les réactants de phase aiguë, la numération des CD4 et la charge virale VIH ont été mesurés. Un questionnaire qualitatif sur la morbidité a été utilisé pour déterminer la prévalence des symptômes liés à la maladie.

Résultats:  Les patients TB diabétiques avaient un nombre plus élevé de neutrophiles (B 0,5 x 109 cellules/L, IC95%: 0,2 - 0,9; p = 0,001) que les patients TB non-diabétiques. Le CRP (B 18,8 mg/L, IC95%: 8,2 - 29,4; p = 0,001) et AGP (B 0,2 g/L, IC95%: 0,03 - 0,3; p = 0,02) sériques étaient aussi plus élevés chez les patients diabétiques. Le diabète n’avait pas d’influence sur le rendement de la culture ou le statut VIH, mais la fièvre auto-déclarée était trois fois plus élevée chez les patients atteints de diabète que chez ceux sans diabète (OR: 2,9, IC95%: 1,5 - 5,7; p = 0,002).

Conclusion:  Le diabète est associéà de petits changements dans les manifestations de la TB, mais pourrait n’avoir que peu de signification clinique.


Objetivo:  La diabetes está asociada con la tuberculosis (TB) pulmonar, posiblemente debido a una inmunidad comprometida, y puede exacerbar las manifestaciones clínicas de la TB. Nuestro objetivo era evaluar el papel de la diabetes en las manifestaciones clínicas de la TB.

Métodos:  Hemos estudiado a 1250 pacientes con TB pulmonar durante un estudio crosseccional en población urbana de Tanzania. A todos los participantes se les realizaron pruebas para diabetes y VIH, y se evaluó la intensidad del cultivo de TB. Se midieron los niveles de leucocitos, hemoglobina, reactivos de fase aguda, conteo de CD4 y carga viral de VIH, y se utilizó un cuestionario cualitativo de morbilidad para identificar la prevalencia de los síntomas relacionados con la enfermedad.

Resultados:  Los pacientes con TB y diabetes tenían un mayor conteo de neutrófilos (B 0.5 × 109 células/L, 95% CI 0.2;0.9, p = 0.001) que los pacientes no diabéticos-TB. La proteína C-reactiva en suero (B 18.8 mg/L, IC 95% 8.2; 29.4, p = 0.001) y la alfa-1 glicoproteína ácida (B 0.2 g/L, IC 95% 0.03;0.3, p = 0.02) eran igualmente altos en pacientes con diabetes. La diabetes no afectaba la intensidad de cultivo o el estatus de VIH, pero la fiebre autoreportada era tres veces mayor entre los participantes con diabetes que entre aquellos sin diabetes (OR 2.9, IC 95% 1.5; 5.7, p = 0.002).

Conclusión:  La diabetes está asociada con pequeños cambios en las manifestaciones de la TB, pero pueden tener poco significado clínico.


Diabetes (type II) is increasingly common in low-income countries previously dominated by communicable diseases (Mohan et al. 2006). Diabetes is known to increase the risk of infections (Joshi et al. 1999; Muller et al. 2005) possibly due to impaired cellular immunity and neutrophil function (Pozzilli & Leslie 1994; Delamaire et al. 1997; Calvet & Yoshikawa 2001). In low-income populations, highly exposed to tuberculosis (TB), diabetes may contribute considerably to the incidence of TB. Furthermore, hyperglycaemia per se can lead to infectious complications as well as increased mortality (McCowen et al. 2001). Diabetes may also exacerbate the clinical manifestations of TB, although it has only been addressed in few studies from low-income settings and none from the African continent. However, studies from Taiwan and from the US–Mexican border, based on register data, have reported that people with diabetes experience worse clinical manifestations of TB, that is, fever, haemoptysis, lower lung infiltration, higher mycobacterial burden and delayed conversion (Wang et al. 2005, 2009; Alisjahbana et al. 2007; Restrepo et al. 2007).

We recently found that diabetes among urban Tanzanians was associated with a fourfold increased risk of active TB (Faurholt-Jepsen et al. 2011). Based on the data already available on the impaired immunity in patients with diabetes, we therefore hypothesised that TB patients with diabetes initially would present with more severe clinical manifestations compared to TB patients without diabetes. In this article, we report the role of diabetes on the clinical manifestations of TB, as well as on paraclinical parameters including haematology, acute phase response and sputum culture intensity.


Participants and study design

Newly diagnosed patients with pulmonary TB from Mwanza City, Tanzania, were invited to participate in a nutrition study between 2006 and 2008. The patients were invited upon confirmed TB diagnosis. To be included in the study, HIV testing was compulsory. Patients below 15 years of age, pregnant or lactating women, terminally ill (judged unlikely to survive >48 h), suffering from other severe disease, as well as non-residents of Mwanza City were excluded. Participants were treated according to the guidelines of the National Tuberculosis and Leprosy Control Programme of Tanzania. This article presents cross-sectional data from the time of enrolment in the nutritional study.


All enroled patients fullfiled the criteria for active TB and were based on further diagnostic tests diagnosed with either sputum-positive pulmonary TB (PTB+) or sputum-negative pulmonary TB (PTB−). Sputum smear microscopy was performed at the recruiting health facility and at the Zonal TB Reference Laboratory at Bugando Medical Centre. Using the spot-morning-spot procedure (IUATLD 2000), the smear was graded according to the presence of acid fast bacilli (AFB) as either negative (no AFB) or positive: 1–9 in 100 fields, +1 (10–99 AFB in 100 fields), +2 (1–10 AFB per field in >50 fields) or +3 (>10 per field in >20 fields). Additionally, early-morning sputum samples were collected in a sterile universal bottle for the culture of Mycobacterium tuberculosis on Lowenstein Jensen solid media and graded as either negative (no growth of colonies) or positive: 1–19 colonies, +1 (20–100 colonies), +2 (innumerable discrete colonies) or +3 (confluent growth). A positive test result was defined PTB+, and the final TB diagnosis thus relied on culture status; microscopy results were only used if the culture result was missing. The definition PTB− was used for patients with a negative sputum result, but with clinical suspicion of TB, patient history, lack of clinical improvement after treatment with a broad spectrum antibiotic and TB suggestive X-ray based on WHO guidelines (WHO 2009).

Venous blood was drawn in EDTA tubes at local health facilities and transported to the research laboratory, whereupon serum was collected and kept at −80 °C until analysed. For HIV diagnosis, two rapid tests, Determine HIV 1/2 (Inverness Medical Innovations, Inc., Delaware, USA) and Capillus HIV-1/HIV-2 (Trinity Biotech Plc., Wicklow Ireland) were used. If the tests were discordant, HIV diagnosis was based on ELISA results. Cluster of differentiation 4 (CD4) counts was determined by flow cytometry after CD4 immuno-flourochrome staining of the leucocytes (Partec FACS, Partec GmbH., Germany), and haemoglobin levels (g/dl) and white blood cell (109/l) count including differentials were carried out at the research laboratory at the National Institute for Medical Research in Mwanza. Anaemia was defined as haemoglobin <12.0 g/dl in women and <13.0 g/dl in men. Serum concentrations of the acute phase reactants C-reactive protein (CRP) and alpha-1-acid glycoprotein (AGP) were determined at the Department of Clinical Biochemistry, Aalborg University Hospital, Denmark. AGP above 1.2 g/l was used to define elevated acute phase response (Dati et al. 1996).

Fasting blood glucose was determined using point-of-care diagnostic instruments (HemoCue 201+ Glucose System, HemoCue, Sweden) and those with a fasting blood glucose between 5.1 and 11.0 mm (capillary whole blood levels) completed a standard 2-h oral glucose tolerance test (OGTT) using 75 g of glucose. The range of the fasting blood glucose to proceed with an OGTT was expanded from the normal 5.6–6.0 mm, because the participants all had an infectious disease possibly leading to impaired glucose metabolism, namely stress-induced hyperglycaemia. The test was performed between 8.00 and 10.00 am. All participants were instructed to be fasting from midnight (>8 h), and only water was allowed prior to test. Participants with fasting blood glucose >6 mm or post-OGTT blood glucose level >11 mm were diagnosed with diabetes in accordance with the most recent WHO classification (WHO 2006). As the diagnosis of diabetes was for epidemiological purposes only, we did not repeat the test in those with values suggestive of diabetes. Participants already known to have diabetes were only classified as such in the study, if the diagnosis could be reproduced by our methods. The diabetes testing was performed as soon as possible (few days) after the initiation of TB treatment to eliminate the role of adverse drug effects.

Recall data on known symptoms indicative of TB or diabetes were collected through interviews using a questionnaire; the patients were asked whether they, during the last month, had experienced fever, polyuria, polydipsia, skin ulcerations, visual impairment, weight loss, cough, haemoptysis or night sweat. Furthermore, the HIV-infected participants were asked about the use of anti-retroviral treatment (ART). Their medical records, if any, were checked for preceding events of TB, to allow a distinction between new and relapsed TB infection to be made. All participants underwent baseline examination of anthropometry as reported elsewhere (PrayGod et al. 2011a). Handgrip strength was determined to the nearest 0.1 kg using a digital dynamometer (Takei Scientific Instruments, Japan), and the mean from the two maximum results (one in each hand) was reported.

Statistical analysis

Data were double entered and analysed using Stata 11.2 (StataCorp LP, College Station, USA). Continuous variables assessed were age, haemoglobin level, white blood cell count, serum CRP, serum AGP, CD4 count and handgrip strength. Categorical variables assessed were sex (female/male), TB status (sputum culture negative/positive) and culture intensity (+1, +2 and +3), HIV status (negative/positive), anaemia (no/yes), reported symptoms and medical history. For each of the possible predictors, linear or logistic regression models adjusted for age and sex were used to test for differences between TB patients with and without diabetes. P-values <0.05 were considered significant.

Ethical considerations

Ethical permission was obtained from the Ethics Committee of the National Institute for Medical Research in Tanzania, and The Danish National Committee on Biomedical Research Ethics approved the study. Written and oral information were presented to all eligible participants before written consent was obtained. Written consent was obtained from parents/legal guardians of any participants between 15 and 18 years of age. Pre-HIV test counselling was compulsory, and post-HIV test counselling was offered to all HIV-positive participants. Patients with diagnosed HIV or diabetes were referred for further management.


Of 1250 patients with pulmonary TB recruited, 1205 (96.4%) had complete blood glucose data and were included in the analyses. Mean (standard deviation) age and BMI were 36.6 years (13.0) and 18.8 kg/m2 (3.0), respectively, and 493 (40.9%) were women. HIV infection was present in 60.6% and 43.8% of women and men (P < 0.001), respectively, and 12.6% of HIV-infected patients used ART prior to the study. Nearly 803 (66.6%) were diagnosed as PTB+, and among those with a PTB+ diagnosis, 735 (92%) were culture-positive and 68 (8%) were diagnosed as PTB+ using smear microscopy owing to either a contaminated or missing culture sample. The overall prevalence of anaemia was 78.1%. The mean (standard deviation) total white blood cell count was 6.5 × 109 (2.6 × 109) cells/l, of which 4.2 × 109 (2.2 × 109) cells/l (64.6%) were neutrophil granulocytes and 1.9 × 109 (0.9 × 109) cells/l (29.9%) were lymphocytes, and 89.4% had an acute phase response elevated above the reference level. Of the 1205 patients with TB, 444 (36.9%) had impaired glycaemia, including 197 (16.4%) with diabetes. Demographic characteristics of the population are shown in Table 1.

Table 1.   Background characteristics of 1205 patients with tuberculosis stratified by diabetes status*
 Non-diabetes (n = 1008)Diabetes (n = 197)P
  1. *Data are mean (SD) or %.

Female sex (%)
Age (year)36.3 (12.8)38.0 (13.6)0.08
Ethnic group (%)
Marital status
Occupation (%)
Religion (%)
HIV co-infection (%)49.750.80.79

There was no difference in Mycobacterium tuberculosis culture status (OR 1.2, CI 95% 0.9; 1.8, P = 0.26) and HIV status (OR 1.0, CI 95% 0.7; 1.4, P = 0.98) across diabetes status. The associations between diabetes and clinical and paraclinical parameters were assessed using linear regression models adjusted for age and sex (Table 2). Patients with diabetes had 0.4 × 109 cells/l higher white blood cell count (95% CI 0.02; 0.8, P = 0.04) and 0.5 × 109 cells/l higher neutrophil count (95% CI 0.2; 0.9, P = 0.001), whereas no difference was found for the level of lymphocytes. There was an interaction between diabetes and the TB diagnosis (PTB+ vs. PTB−) for the acute phase reactants CRP (P = 0.004) and AGP (P = 0.04); the association between diabetes and acute phase reactants was present among PTB+ patients only. PTB+ patients with diabetes had 18.8 mg/l (CI 95% 8.2; 29.4, P = 0.001) and 0.2 g/l (CI 95% 0.003; 0.3, P = 0.02) higher levels of CRP and AGP, respectively, compared to PTB+ patients without diabetes. Diabetes did not affect the levels of CD4 count and HIV viral load.

Table 2.   Clinical and paraclinical status of 1205 patients with pulmonary TB with and without diabetes
 Patients without diabetes
(n = 1008)
Patients with diabetes
(n = 197)
Difference, crudePDifference, adjusted*P
  1. WBC, white blood cell count; CD4, cluster of differentiation 4; TB, tuberculosis; PTB+/−, sputum positive/negative pulmonary TB.

  2. Data are mean (standard deviation) or difference (95% confidence interval).

  3. *Adjusted for age and sex.

  4. †C-reactive protein: diabetes – TB diagnosis interaction (P = 0.004).

  5. ‡Alpha-1-acid glycoprotein: diabetes – TB diagnosis interaction (P = 0.04).

Haemoglobin (g/dl)
 Women10.1 (2.3)9.9 (2.5)−0.2 (−0.8; 0.3)0.43−0.3 (−0.8; 0.2)0.27
 Men11.2 (2.4)11.3 (2.8)0.04 (−0.5; 0.5)0.890.04 (−0.5; 0.6)0.88
WBC total (×109/l)6.4 (2.6)6.8 (2.6)0.4 (−0.04; 0.8)0.080.4 (0.02; 0.8)0.04
 Neutrophils (×109/l)4.1 (2.2)4.6 (2.3)0.5 (0.2; 0.8)0.0040.5 (0.2; 0.9)0.001
 Lymphocytes (×109/l)2.0 (0.9)1.9 (0.8)−0.1 (−0.2; 0.1)0.26−0.1 (−0.2; 0.1)0.27
C-reactive protein (mg/l)
 PTB−51.4 (57.4)41.6 (45.2)−9.7 (−24.8; 5.3)0.20−7.5 (−22.4; 7.3)0.32
 PTB+88.3 (55.6)106.1 (61.3)17.8 (7.2; 28.4)0.00118.8 (8.2; 29.4)0.001
Alpha-1-acid glycoprotein (g/l)‡
 PTB−1.9 (0.8)1.8 (0.9)−0.1 (−0.4; 0.1)0.31−0.1 (−0.3; 0.2)0.50
 PTB+2.6 (0.7)2.8 (0.7)0.2 (0.02; 0.3)0.030.2 (0.03; 0.3)0.02
CD4 count (cells per μl)
 HIV-negative553.5 (339.8)540.4 (335.1)−13.1 (−86.8; 60.7)0.73−12.3 (−86.2; 61.6)0.74
 HIV-positive292.2 (206.4)271.8 (0.37)−20.5 (−64.8; 23.9)0.37−14.9 (−59.1; 29.2)0.51
HIV viral load (RNA copies/ml) (log10)4.3 (1.39)4.1 (1.5)−0.2 (−0.5; 0.1)0.28−0.2 (−0.5; 0.1)0.29
Handgrip strength (kg)25.8 (8.3)24.5 (7.8)−1.3 (−2.6; −0.03)0.04−0.8 (−1.9; 0.4)0.18

The distribution of symptoms typically indicative of TB and diabetes is shown in Table 3. In a logistic regression model adjusted for age and sex, the proportion of self-reported fever was almost three times (OR 2.9, CI 95% 1.5; 5.7, P = 0.002) higher among PTB+ participants with diabetes than among PTB+ participants without diabetes, whereas no association was found for PTB− participants (interaction, P = 0.02). The association between diabetes and fever was not confounded by the levels of acute phase reactants and white blood cells (data not shown).

Table 3.   Reported morbidity among 1205 patients with pulmonary TB with and without diabetes
 Patients without diabetes
(n = 1008)
Patients with diabetes
(n = 197)
Difference, crude
OR (95% CI)
PDifference, adjusted*
OR (95% CI)
  1. TB, tuberculosis; ARV, anti-retroviral treatment; PTB+/−, sputum positive/negative pulmonary tuberculosis.

  2. Data are % or odds ratio (95% confidence interval).

  3. *Adjusted for age and sex.

  4. †Fever: diabetes – TB diagnosis interaction (P = 0.02).

Disease-related symptoms (%)
 Fever (yes)†
  PTB−78.477.81.0 (0.5; 1.8)0.911.0 (0.5; 1.9)0.97
  PTB+80.992.52.9 (1.5; 5.7)0.0022.9 (1.5; 5.7)0.002
 Ulcers, leg/feet (yes) (0.2; 1.1)0.070.4 (0.2; 1.1)0.06
 Visual impairment (yes)21.823.21.1 (0.8; 1.6)0.661.0 (0.7; 1.5)0.94
 Polydipsia (yes)50.948.20.9 (0.7; 1.2)0.490.9 (0.7; 1.2)0.53
 Polyuria (yes)59.752.30.7 (0.5; 1.0)0.0530.7 (0.5; 1.0)0.06
 Weight loss (yes)80.777.20.7 (0.4; 1.1)0.110.7 (0.4; 1.1)0.14
 Cough (yes)96.696.50.9 (0.4; 2.2)0.900.9 (0.4; 2.1)0.88
 Haemoptysis (yes)22.321.30.9 (0.7; 1.4)0.761.0 (0.7; 1.4)0.89
 Night sweat (yes)75.479.21.2 (0.9; 1.8)0.261.2 (0.9; 1.8)0.25
Medical history (%)
 TB relapse (yes) (0.9; 2.6)0.101.6 (0.9; 2.6)0.09
 Relatives with diabetes (yes) (0.8; 2.7)0.171.5 (0.8; 2.7)0.17
 ARV treatment for HIV (yes) (0.9; 2.8)0.101.5 (0.9; 2.6)0.16

None of the other self-reported diabetes-related symptoms were different between diabetes and non-diabetes TB patients. However, the seemingly higher proportion with TB relapse, and lower proportion with weight loss, ulcers on the extremities and polyuria in those with diabetes were not statistically significant. Except for the above mentioned, no other interactions between diabetes and HIV as well as between diabetes and TB diagnosis were found for paraclinical parameters or symptoms.


Among patients with pulmonary TB, the presence of diabetes was associated with minor alterations in clinical and paraclinical manifestations. Although TB patients with diabetes had slightly higher levels of neutrophil granulocytes and acute phase response and more often had fever, most symptoms and paraclinical tests were similar to TB patients without diabetes. In particular, there was no difference in culture intensity and HIV viral load.

It has previously been reported that TB patients with diabetes may have higher mycobacterial burden, higher TB culture intensity and altered radiographic patterns (Hendy & Stableforth 1983; Mboussa et al. 2003; Wang et al. 2005, 2009). We did indeed find a higher acute phase response, including elevated levels of neutrophil granulocytes, among TB patients with diabetes. In a study from Indonesia, leucocyte count and CRP levels were reported, but there was no statistical difference between patients with and without diabetes co-morbidity, but in contrast to our data, the median level of both measures seemed to be lower in the diabetes group (Alisjahbana et al. 2007). The TB diagnosis (PTB−vs. PTB+) affected the level of the acute phase reactants, causing a higher level of CRP and AGP among the PTB+ participants, in which group those with diabetes had the highest levels. The higher acute phase response in PTB+ patients with diabetes indicates that diabetes does affect the clinical manifestations of TB, but the differences are so small that they may not be clinically relevant. Also, the higher acute phase response seen in the diabetes group could possibly be associated with stress hyperglycaemia, indicating that the glucose levels in patients with more severe TB could be elevated to such extent that the patients fullfiled the criteria for diabetes. We did not see this association among patients with PTB−, but the PTB− diagnosis embrace a group of patients, which is rather heterogeneous, and may include a large proportion of severely ill patients not capable of producing an adequate acute phase response. Finally, our data do not support higher culture intensity among patients with diabetes, and neither was the proportion of PTB− and PTB+ affected by diabetes. This finding is in contrast to other studies, where diabetes has been associated with more severe bacterial status on sputum microscopy and culture (Restrepo et al. 2007; Wang et al. 2009).

In accordance with previous studies from the US–Mexican border and Taiwan (Restrepo et al. 2007; Wang et al. 2009), we found that the proportion of reported fever was three times as high in those with diabetes as in those without diabetes. The elevated neutrophil count and acute phase reactants seen in the patients with diabetes could be mediated by cytokines also responsible for fever (Exton 1997), and the cytokines responsible for fever are also associated with reduced appetite, which we reported elsewhere (PrayGod et al. 2011b). Because of these circumstances, it is likely that people with diabetes experience accelerated weight loss caused by the TB disease and subsequently delayed weight gain during the course of the anti-TB treatment. Some of the patients diagnosed with diabetes may have stress hyperglycaemia; a state common in severely ill patients (i.e. trauma, surgery and intensive care unit) (Başoğlu et al. 1999; Gearhart & Parbhoo 2006). The mechanisms behind stress hyperglycaemia are primarily caused by the activation of the hypothalamic–pituitary–adrenal axis, leading to insulin resistance and hyperglycaemia (Gearhart & Parbhoo 2006). Severe TB infection could potentially lead to hyperglycaemia in this population. Therefore, the association between diabetes and reported fever might partly be mediated by stress hyperglycaemia, and only a repeated diabetes diagnosis after 6 months of anti-TB treatment could reveal to what extent stress hyperglycaemia was present in this population. We do know that the prevalence of diabetes in a comparable non-TB population from the same city was very high (9%; Faurholt-Jepsen et al. 2011), and because there is growing evidence of the association between habitual diabetes and increased risk of incident TB, non-diabetic stress hyperglycaemia is likely to represent a minority of the diabetes diagnosed in the present study. Furthermore, owing to the inclusion criteria from the nutritional study, wherefrom the participants were recruited, patients with TB with severe or terminal illness were excluded from the study; thus, those at highest risk of stress hyperglycaemia were already excluded. Nevertheless, interpretation of glucose parameters, without taking stress-induced hyperglycaemia in patients with infectious diseases into account, should be carried out with precaution, and the diabetes diagnosis should preferably be confirmed after the elimination of the infection. Also, the two diseases may share risk factors, and socio-economic circumstances in early life could lead to a higher risk of both diabetes and TB in later life, thus reducing the biological interaction between the two diseases as they co-occur.

HIV is the strongest risk factor for TB (Lönnroth et al. 2010) and interferes markedly with the clinical presentation of TB causing more extra-pulmonary and disseminated forms of TB. Despite a higher prevalence of diabetes among HIV uninfected compared to HIV-infected (Faurholt-Jepsen et al. 2011) groups, no interaction was found between HIV infection and diabetes in relation to the manifestations of TB. One of eight HIV-infected patients received ART, which potentially could cause metabolic dysfunction (Butt et al. 2009), but no differences were identified between the users of ART and ART-naïve HIV-infected participants (data not shown). Because this study is the first to be conducted in a TB population heavily co-infected with HIV, it is encouraging that we did not experience a worsened clinical status among patients with both diabetes and HIV co-morbidity. However, the nature of the cross-sectional data does not allow commenting upon prognosis, but because diabetes (Young et al. 2009) and HIV (Swaminathan et al. 2010) individually impairs treatment outcome of TB, the combination of diabetes and HIV co-morbidity may further worsen TB treatment outcome.

It is unfortunate that we did not have access to detailed descriptions of X-rays, because other studies found altered radiographic patterns among patients with diabetes co-morbidity (Hendy & Stableforth 1983; Wang et al. 2005, 2009), but we chose a broad spectrum of paraclinical parameters as well as a qualitative morbidity interview to investigate the impact of diabetes on the clinical manifestations of TB. Even though the definitions for normal and elevated blood glucose levels are widely used in non-TB patient populations, the group with low fasting blood glucose is quite heterogeneous and may not necessarily identify the healthiest group of patients seen from a metabolic point of view. Also, many of the patients diagnosed with diabetes had blood glucose levels in the lower end of the range for diabetes, which could underestimate the role of diabetes on the clinical manifestations, because the manifestations are likely to be more pronounced in patients with advanced diabetes. More biomarkers related to inflammation, insulin, glycosylated haemoglobin measurements as well as a repeated diabetes screening after TB recovery are highly needed. Also, this being a cross-sectional study, we may have failed to show that people with diabetes have severe clinical presentation of TB, because those with severe manifestation died or failed to seek treatment at the hospitals, or were excluded from the study owing to the exclusion criteria about severe disease.

In conclusion, although diabetes is a risk factor for pulmonary TB in an urban Tanzanian population, it does not seem to considerably increase the severity of TB.


The authors would like to thank all the health staff and study participants involved in the study. This study was supported by the Danish Council for Independent Research, by Danida, through the Consultative Research Committee for Development Research, the University of Copenhagen through the Cluster in International Health and Augustinus Fonden.