Active tuberculosis (TB) can provoke metabolic derangements, including hyperglycaemia and insulin resistance, which may increase the risk of developing type 2 diabetes mellitus and worsen TB severity. Infection with Mycobacterium tuberculosis disrupts cytokine and hormonal balance, increasing pro-inflammatory mediators that impair pancreatic β-cell function and insulin signalling. Anti-TB drugs such as rifampicin and isoniazid have also been implicated in raising blood glucose, particularly during the intensive phase. Hyperglycaemia can weaken host immunity, alter drug absorption and metabolism, and has been associated with unfavourable TB outcomes. Despite this, longitudinal data describing the dynamics of TB-associated hyperglycaemia in individuals not previously diabetic are limited, especially in high TB-burden, low- and middle-income settings. This study aimed to describe the prevalence, temporal patterns, and clinical impact of TB-associated hyperglycaemia among newly diagnosed PTB patients without prior diabetes.
This was a five-month prospective cohort study conducted between February and November 2024 at two TB-specialized public facilities in Nairobi, Kenya: Mbagathi County Hospital and Rhodes Chest Clinic. Adults aged >18 years with bacteriologically confirmed pulmonary TB by Xpert MTB/RIF Ultra were enrolled prior to initiating TB treatment. Individuals with known prediabetes or diabetes, pregnancy, lactation, multidrug- or rifampicin-resistant TB, or other chronic diseases (e.g., cancer, cardiovascular disease) were excluded. Participants were followed with visits at baseline (before starting TB therapy), month two, and month five.
At each visit investigators collected socio-demographic data, smoking and alcohol use, anthropometry (weight, height, BMI), vital signs, and TB-related clinical symptoms. HIV status was ascertained by self-report and confirmatory testing per national guidelines. Sputum specimens were obtained and tested according to routine diagnostic algorithms; TB treatment outcomes were classified bacteriologically as favourable (smear negative) or unfavourable (smear positive) at months two and five.
Approximately 4 mL of venous blood was drawn in a lavender-top vacutainer for HbA1c testing using a fluorescence immunoassay sandwich immunodetection method (Finecare™). Blood for baseline hyperglycaemia testing was collected before treatment initiation; follow-up samples were taken at months two and five.
Glycaemic classification followed American Diabetes Association thresholds. Participants with HbA1c 4.0–5.6% were considered normoglycaemic; HbA1c 5.7–6.4% was classified as prediabetes (pre-DM); and HbA1c ≥6.5% as diabetes mellitus. Longitudinal HbA1c trajectories across the three time points were categorized as persistent normoglycaemia, persistent hyperglycaemia (HbA1c ≥5.7% at all time points), transient hyperglycaemia (HbA1c ≥5.7% at one or two time points only), and incident hyperglycaemia (normoglycaemic at baseline, hyperglycaemic at month two or five). After follow-up participants were further grouped as having glycaemic stability (HbA1c consistently 4.0–5.6%) or glycaemic variability (fluctuating between normoglycaemic and hyperglycaemic ranges).
Analyses were performed using SPSS v26 and GraphPad Prism v9. Normality was tested with Shapiro-Wilk. Continuous variables were summarized as medians and interquartile ranges; categorical variables as counts and percentages. Group differences used Mann-Whitney U, χ2, or Fisher’s exact tests as appropriate. The Friedman test assessed overall differences in HbA1c across the three time points in a complete-case subset; significant findings prompted pairwise Wilcoxon signed-rank tests with Bonferroni correction.
Longitudinal associations between hyperglycaemia and clinical symptoms or sputum outcomes were modelled with Generalized Estimating Equations (GEE) using a binomial distribution and log link, exchangeable correlation structure, and robust standard errors. Multivariable models adjusted a priori for age, BMI, HIV status, alcohol use, and smoking. Sensitivity analyses addressed symptom overlap. Missing data patterns were examined; comparability of baseline characteristics between complete and incomplete follow-up groups supported the plausibility of the missing completely at random assumption used for GEE available-case analysis.
The study enrolled 87 newly diagnosed PTB patients. During follow-up seven participants died and several were lost to follow-up or transferred out, yielding 66 participants tested at month two and 70 at month five after accounting for deaths, transfers, and re-engagements. The cohort was predominantly male (76%), not married (53%), HIV-negative (84%), with a median age of 37 years (IQR 29–44).
At baseline the point prevalence of hyperglycaemia (HbA1c ≥5.7%) was 75% (95% CI 65–83), composed of 45% prediabetes and 30% diabetes by HbA1c thresholds. Across follow-up, the proportion with prediabetes increased from 24% (16/66) at month two to 30% (21/70) at month five, while the proportion classified as diabetes decreased from 21% (14/66) at month two to 9% (6/70) at month five. Overall, HbA1c levels differed significantly across the three time points (p < 0.001), indicating dynamic glycaemic changes during anti-TB therapy, with both persistent and transient patterns observed.
In GEE analysis hyperglycaemic participants were more likely to report classic diabetes-related symptoms: frequent micturition (adjusted RR 1.76, 95% CI 1.12–2.77), increased thirst (aRR 2.37, 95% CI 1.38–4.05), unintentional weight loss (aRR 1.84, 95% CI 1.24–2.75), and blurred vision (aRR 1.95, 95% CI 1.20–3.18). After adjustment for age, BMI, HIV status, alcohol use, and smoking, hyperglycaemia was independently associated with persistent sputum positivity (aRR 4.22, 95% CI 2.09–8.52). Glycaemic variability was significantly associated with unfavourable bacteriological TB treatment outcomes (p < 0.001).
Findings indicate that new-onset hyperglycaemia is common among adults newly diagnosed with PTB and that glycaemic abnormalities frequently persist through the fifth month of treatment. The observed association between hyperglycaemia and classic diabetic symptoms supports the clinical relevance of screening. The independent association with persistent sputum positivity and the link between glycaemic variability and unfavourable outcomes suggest that metabolic monitoring and, where appropriate, management should be integrated into routine TB follow-up to potentially improve bacteriological response and clinical outcomes. The study adds longitudinal data to prior cross-sectional reports and highlights the heterogeneity of TB-associated glycaemic trajectories.
The study was approved by the University of Nairobi–Kenyatta National Hospital Ethics and Research Committee (P553/06/2023) and received institutional permissions. Participants provided written informed consent. The study was funded by the NIH/Fogarty International Center (grant 5D43TW011817); funders did not influence study conduct or reporting. The underlying datasets are not publicly available due to ethical and privacy constraints; qualified researchers may request access subject to review and approval by the University of Nairobi–Kenyatta National Hospital ERC.