Reply to Ouattara et al.: past history of tuberculosis is not a risk factor for incident tuberculosis during antiretroviral treatment in South Africa.