Impact of diabetes and smoking on mortality in tuberculosis

Abstract

Background: Diabetes mellitus is a risk factor for tuberculosis (TB) disease. There is evidence that diabetes also influences TB severity and treatment outcomes but information is incomplete and some published results have been inconsistent.

Methods: A longitudinal cohort study was conducted at the National Masan Tuberculosis Hospital in the Republic of Korea. Subjects presenting with a first episode of TB or for retreatment of TB were followed from enrollment through completion of treatment. Demographic, clinical, and microbiological variables were recorded, along with assessment of outcomes. Results were compared in TB patients with and without diabetes or smoking history. Data were adjusted for gender, age, cohort, educational level and alcohol consumption.

Results: The combined cohorts comprised 657 subjects. Diabetes was present in 25% and was associated with greater radiographic severity and with recurrent or relapsed TB. Diabetes and cigarette smoking independently increased the risk of death in the first 12 months after enrollment. Estimating the combined impact of diabetes and smoking yielded a hazard ratio of 5.78. Only 20% of diabetic subjects were non-smokers; 54% smoked ≥1 pack daily. In this cohort, the impact of diabetes on mortality was greater in patients younger than 50 years, compared to older patients.

Conclusions: In this cohort of Korean patients, diabetes exacerbated the severity of TB disease. Diabetic subjects who smoked ≥1 pack of cigarettes daily were at particularly high risk of death from TB. Strategies to improve TB outcomes could productively focus resources for patient education and TB prevention on the vulnerable population of younger diabetics, particularly those who also smoke.

Figure 1. Non-linear association of diabetes and age.

Lowess smoothing curve fit to diabetes prevalence vs. age. A spline fit model (knot at age = 50) is significantly different from a linear fit (p<0.001).

Figure 2. Survival estimates.

Unadjusted Kaplan-Meier survival curve for all causes of death in diabetic subjects (red lines) and non-diabetic subjects (blue lines) grouped by age ≥50 (dashed lines) or age <50 (solid lines). Separate curves by age (dashed vs. solid) and diabetes (red vs. blue) illustrate the interaction of diabetes and age on survival. For age <50 the separation between diabetic vs. non-diabetic subjects is larger than for age ≥50.