Objective: To study the relationship between smoking and tuberculosis in Hong Kong.
Method: Indirect sex and age adjustment was used to compare the prevalence of ever smokers between a sample of 851 patients from the 1996 tuberculosis notification registry and the general population. The clinical characteristics of smokers and non-smokers were compared by stratified univariate analysis and multiple logistic regression.
Results: Tuberculosis patients were more likely to have smoked than population controls. The respective odds ratios for ever smoking between tuberculosis patients and population controls were 2.44 and 2.08 for males and females aged 16-64 (Mantel-Haenszel weighted OR = 2.40, P < 0.001), and 2.09 and 2.83 for males and females aged > or = 65 (Mantel-Haenszel weighted odds ratio = 2.19, P < 0.001). Male sex, age > or = 65, working at onset of illness, regular alcohol use, drug abuse and absence of contact history were associated with ever smokers (all P < 0.05). Ever smokers were more likely to have cough (OR 1.69), dyspnoea (OR 1.84), upper zone involvement (OR 1.67), cavity (OR 1.76), miliary lung involvement (OR 2.77), positive sputum culture (OR 1.43), but less isolated extrathoracic involvement (OR 0.31), even after controlling for the confounding background variables (all P < 0.05).
Conclusion: There was a consistent association between smoking and tuberculosis. More aggressive lung involvement was also found among ever smokers.