Influenza virus infection is associated with increased risk of death amongst patients hospitalized with confirmed pulmonary tuberculollsis in South Africa, 2010-2011

Background: Data on the association between influenza and tuberculosis are limited. We describe the characteristics of patients with laboratory-confirmed tuberculosis, laboratory-confirmed influenza and tuberculosis-influenza co-infection. Methods: Patients hospitalized with severe respiratory illness (acute and chronic) were enrolled prospectively in four provinces in South Africa. Naso/oropharyngeal specimens were tested for influenza virus by real time reverse transcriptase polymerase chain reaction. Tuberculosis testing was conducted as part of clinical management. Results: From June 2010 through December 2011, 8032 patients were enrolled and influenza testing was conducted on 7863 (98%). Influenza virus was detected in 765 (10%) patients. Among 2959 patients with tuberculosis and influenza results, 2227 (75%) were negative for both pathogens, 423 (14%) were positive for tuberculosis alone, 275 (9%) were positive for influenza alone and 34 (1%) had influenza and tuberculosis co-infection. On multivariable analysis amongst individuals with symptoms for = 7 days, tuberculosis influenza co-infection was associated with increased risk of death, (adjusted relative risk ratio (aRRR) (6.1, 95% confidence interval (CI) 1.6-23.4), as compared to tuberculosis only infection. This association was not observed in individuals with symptoms for < 7 days (aRRR. 0.8, 95% CI 0.1-7.0). Conclusion: Tuberculosis and influenza co-infection compared to tuberculosis single infection was associated with increased risk of death in individuals with symptoms >= 7 days. The potential public health impact of influenza vaccination among persons with laboratory-confirmed tuberculosis should be explored.