Introduction
The prevalence of diabetes worldwide has increased by about 20% in the past three decades with particularly large rates in low and middle income countries [
1,
2]. The International Diabetes Federation (IDF) predicts further increase in prevalence of DM especially in developing countries in the foreseeable future due to changes in life style, eating habits and other risk factors [
3]. Indeed countries in sub Saharan Africa (SSA), are going through an epidemiological transition with rising prevalence of non-communicable diseases including type two DM (T2DM) [
4]. For example between the 1980s and 2010 the prevalence of DM in rural Tanzania increased from 0.9 to 5% [
5,
6]. Furthermore data from the city of Dar- es -salaam indicates that the prevalence of DM among the general population increased from < 2 to 9.1% during the period of 1980–2012 [
5,
6]. A bi-directional interaction between DM and Tuberculosis is well established [
7]. People with DM are three times at risk of developing TB disease compared to non-diabetics [
8]. Glucose intolerance or transient hyperglycemia has been observed in up to 49% of patients with active TB [
9]. Patients with DM have impaired cellular immunity and ciliary function that predisposes them to TB while stress response to TB may result to insulin resistance. TB disease of the pancreas can lead to endocrine hypo-function with consequent DM [
7]. In addition DM is an independent risk factor for poor TB treatment response as well as death [
10,
11]. In 2015 the IDF reported that 66.7% of people with diabetes in SSA are unaware that they have the disease [
3].
Tanzania is a high TB endemic country with reported prevalence rates of 293 cases per 100,000 for individuals aged 15–64 years and 709 cases per 100,000 for individuals above 65 years of age [
12]. Hence the rising prevalence of DM is likely to have significant negative impact on TB control and has been likened to the impact of HIV on TB control [
13].
Consequently, screening TB patients for DM and vice versa should result in better control of both diseases, as this would aid in early detection and treatment that would result in better outcomes. Indeed, routine screening of TB patients for DM using HbA1c in India by Balachrishnan et al. found that over 40% had diabetes [
14]. However, Balachrisnan and colleagues did not advocate routine DM screening using HbA1c but recommended further operational research be performed to determine the most cost- effective ways of diabetes screening.
We aimed at determining the prevalence of DM among TB patients and characterizing TB patients with DM in order to identify factors associated with TB-DM dual disease among patients attending TB treatment clinics in Dar es Salaam, Tanzania.
Methods
Ethics statement
Ethical approval was obtained from the Muhimbili University of Health and Allied Science institutional review board approval number MU/PGS/SAEC/Vol.XVI/. All recruited participants were informed about the study details prior to enrollment and verbal consent was obtained from each study participant prior to enrollment in the study. Patients who were newly diagnosed with diabetes during this study were referred to a diabetic clinic to receive standard of care treatment for diabetes.
Study design and population
This descriptive cross-sectional study was conducted in four TB clinics in Dar es Salaam, Tanzania from September 2016 to January 2017. Each clinic represents one of the three districts in Dar es Salaam and receives about 30 TB patients per day. Consecutive patients on treatment for tuberculosis aged 18 years or above attending the four clinics at the time of the study were eligible and consenting patients were enrolled into the study. Recruited participants were either newly diagnosed TB patients or patients who were continuing with TB treatment.
Data collection
Data was collected using an interviewer based structured questionnaire. A research assistant conducted face to face interviews and physical examination. Information collected included socio-demographic characteristics, symptom screening for diabetes, factors associated with type 2 diabetes including alcohol consumption, cigarette smoking and a past medical history related to diabetes. Smoking was categorized as ever smoked in a lifetime or never smoked, Alcohol intake was categorized as ever taken alcohol or never taken alcohol in their lifetime.
We also collected information on sputum smear status, TB anatomical site, TB treatment category, and HIV status from the participants TB treatment cards.
We measured the participant’s weight and height using standard equipment based at the TB clinics thereafter computed body mass index (BMI). BMI was computed as the participant’s weight in kilograms divided by the participant’s height in meters squared. BMI was grouped into four categories as follows; BMI of < 18.5 was regarded as underweight, 18.5–24.9 normal weight, 25.0–29.9 overweight and a BMI > 20.9 was regarded as obese [
15,
16].
A capillary fingertip blood sample was collected from each patient for fasting blood glucose. All participants with impaired fasting blood glucose were given 75 mg of oral glucose that was followed by a two-hour post-prandial blood glucose measurement using standard ACCU-CHECK blood glucose meter. The American Diabetic Association (ADA) diagnostic criteria were used for diagnosis of diabetes [
17]. Diabetes was defined as either a fasting blood glucose level of > 6.9 mmol/L or a 2-h postprandial glucose level of ≥11.1 mmol/L while impaired fasting glucose was defined as glucose levels of 5.6 mmol/L- 6.9 mmol/L [
18].
Statistical methods
The total number of participants recruited to estimate the prevalence of diabetes was based on the sample size estimation formula with an estimated prevalence of diabetes among TB patients to be 16.7% [
19], type I error at 5 and 3% precision that amounted to 594 participants.
Data was transferred from the questionnaires and entered into SPPS version 23.0 database for cleaning and analysis. Continuous variables were summarized and presented as mean and standard deviation while categorical variables were summarized as proportions. Univariate and multivariate logistic regression analysis was performed to examine for factors that were associated with the outcome (diabetes mellitus) using Pearson’s chi square test. All covariates with a
p value of < 0.2 in univariate analysis, and confounders such as age, were entered in the multivariate analysis model based on literature [
20]. As the symptoms for diabetes such as polyuria, nocturia, polyphagia and polydipsia were correlated we summarized these factors into one composite variable that was labeled any symptom for diabetes. Statistical significance was set at a
p value < 0.05. For the multivariate and univariate model, risks were calculated and summarized as odds ratios and 95% confidence intervals significance level was set as a
p value of < 0.05 in the multivariate analysis model.
Discussion
This descriptive cross-sectional study was conducted among patients attending routine clinics for TB, in Dar es Salaam. We found that about 10% of patients receiving TB treatment had DM which is higher than the estimated prevalence of 3.5 and 4.3% among adults aged 20–79 years in the general population [
3,
21]. Similar prevalence rates of DM among TB patients have been reported in recent studies conducted in sub Saharan Africa [
22‐
27] and elsewhere. Indeed, a review of several studies and a meta-analysis study on screening for DM among TB patients concluded that DM was associated with increased risk of TB [
8].
Noteworthy however, is the fact that two thirds (60%) of our study participants with DM did not know they had diabetes (diagnosis of DM was a result of this study). These findings are in keeping with reports from several other studies [
4,
21,
27,
28] and lend support to the recommendation that TB patients be screened for diabetes as T2DM is often asymptomatic and frequently presents initially with complications. Indeed, reports from the IDF show that more than two thirds of adults with T2DM are not aware they have the disease [
3]. In spite of these reports, DM screening is not standard of care among TB patients in most settings in sub Saharan Africa and elsewhere.
Findings from this study show that TB patients aged 44 years and above had four-fold increased risk of diabetes. Likewise, TB patients with family history of DM had a threefold increased risk of diabetes. Findings similar to these have been reported elsewhere [
24]. Our study findings do not advocate DM screening for all TB patients but instead suggest that TB patients aged 44 years and above and those with positive family history, should be screened for diabetes mellitus.
Untreated DM is known to be a risk factor for poor TB outcomes that include treatment failure, relapse and death. In addition, published data indicate that diabetic patients are likely to remain sputum smear positive for AFB for 2 to 3 months following treatment for TB [
29]. In another study 22% of the diabetic TB patients remained sputum culture positive after 6 months treatment course with TB medications [
22].
Screening of TB patients for DM should therefore identify patients who need extra attention and care for better TB treatment outcomes and prevention of complications related to untreated DM.
Over 23% of TB patients in this study had impaired fasting glucose (IFG) for which we are not able with certainty to explain the cause. However, IFG may be indicative of stress- induced hyperglycemia that would disappear after TB treatment [
30‐
32] but it may also be indicative of high risk to developing DM.
We did not follow up the study participants with impaired glucose tolerance to determine if the impaired glucose tolerance detected was TB induced hyperglycemia or was an indicator of later development of DM. Due to the study design we are not able to establish what started first if it was DM or TB. Nonetheless, results of this study indicate that about a quarter (23%) of TB patients had IFG that may have been stress-induced, by the TB disease or an indicator of developing DM in the future. Such patients would benefit from regular blood sugar monitoring to detect early onset DM as well as appropriate changes in their lifestyle to prevent/delay onset of DM.
We observed that HIV sero-negative TB patients were more likely to have DM compared to those who were HIV sero-positive. Similar findings have been reported in other studies in SSA [
19,
26,
33]. However, published reports on association between HIV and diabetes have not been consistent. A nationwide population based cohort study in Denmark found that HIV infected people not on HAART were not at increased risk of developing DM [
34]. On the other hand a study from the USA reported increased risk of DM among HIV infected persons on treatment [
35]. A meta-analysis on the relationship between HIV and DM in SSA concluded that there was no association between the two conditions [
36].
Further studies are needed to examine the effects of CD4 cell count level HIV antiretroviral regimens, and obesity among TB HIV co-infection and DM.
There was a trend that indicated that overweight and obese individuals had an increased risk for development of DM, however due to small numbers of participants with overweight and obesity statistical significance could not be achieved.
The results of this study indicate that screening for DM among patients attending public TB clinics in Dar es Salaam was possible and led to detection of patients with unknown DM and impaired glucose intolerance. Furthermore, DM screening should be targeted to TB patients aged 44 years or more and those with a family history of DM. Nonetheless, similar studies need to be done to confirm our findings before policy change can be considered.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.