In two recent excellent articles, Lawn and colleagues [1,2] reported the incidence and risk factors for active tuberculosis among HIV-infected adults receiving antiretroviral therapy (ART) in South Africa. In both studies, they found contradictory results regarding the association between the baseline World Health Organization (WHO) clinical stage and the occurrence of incident tuberculosis during follow-up, and contradictory trends towards an association between a past history of tuberculosis at enrolment and a lower (first study) or higher (second study) incidence of tuberculosis during follow-up.
We wonder if these contradictions might be a result of their multivariate analysis approach.
In their first study (346 patients, 14% with a past history of tuberculosis, median baseline CD4 cell count 247 cells/μl), Lawn and colleagues  used a multivariate Cox regression model to look for factors associated with incident tuberculosis. The model included sex, age, past history of tuberculosis, baseline WHO clinical stage, baseline CD4 cell count and baseline viral load as independent variables. They found that three factors were significantly associated with a higher incidence of tuberculosis after initiation of ART: having a baseline CD4 cell count of less than 100 cells/μl [hazard ratio (HR) 2.38, P = 0.04], being classified at WHO clinical stage 3 or 4 (HR 3.60, P = 0.01), or being aged under 33 years (HR 2.86, P = 0.01). In addition, they observed a trend towards a negative association between a past history of tuberculosis and the occurrence of incident tuberculosis, but this trend did not reach statistical significance (HR 0.31, P = 0.07).
In their second study (944 patients, 52% with a past history of tuberculosis, median baseline CD4 cell count 96 cells/μl), Lawn and colleagues  used a multivariate Poisson regression model, including sex, age, past history of tuberculosis, baseline WHO clinical stage, prevalent tuberculosis at enrolment, current CD4 cell count, and follow-up viral load as independent variables. Of note is the fact that in that study they included in the analysis patients with prevalent tuberculosis at enrolment, but they do not clarify in their article whether a potential interaction between the ‘tuberculosis past history’ variable and the ‘prevalent tuberculosis at enrolment’ variable was looked for before including those patients in the analysis. They found that the current CD4 cell count was the only variable significantly associated with the incidence of tuberculosis after the start of ART [incidence rate ratio (IRR) 0.75 for 100 CD4 cells/μl increase, P = 0.007]. In addition, they found a trend towards a positive association between a past history of tuberculosis and the occurrence of incident tuberculosis (IRR 1.46, P = 0.14) and a trend towards a negative association between prevalent tuberculosis at enrolment and the occurrence of incident tuberculosis (IRR 0.59, P = 0.14). Both trends were non-significant. In their second study, they did not find any significant association between the baseline WHO clinical stage and the occurrence of incident tuberculosis during follow-up.
Patients with a past history of tuberculosis are by definition classified at WHO stage 3 or 4, with the exception of patients with pulmonary tuberculosis diagnosed in more than 2 years. Therefore, including both the ‘past history of tuberculosis’ variable and the ‘WHO clinical stage’ variable in the same multivariate model of analysis for tuberculosis risk factors may lead to misleading results regarding the association of both variables with the outcome. We suggest that further studies looking for factors associated with incident tuberculosis in patients receiving ART would combine the ‘tuberculosis past history’ and the ‘WHO clinical stage’ into one variable only, e.g. a variable with three groups: WHO stage 1 or 2; WHO stage 3 or 4 without tuberculosis past history; and WHO stage 3 or 4 with tuberculosis past history. This may help to further the still controversial issue of whether patients with a past history of tuberculosis might or might not be at a higher risk of developing active tuberculosis after the initiation of HAART [3–5].
1. Lawn SD, Badri M, Wood R. Tuberculosis among HIV-infected patients receiving HAART: long term incidence and risk factors in a South African cohort. AIDS 2005; 19:2109–2116.
2. Lawn S, Myer L, Bekker L, Wood R. Burden of tuberculosis in an antiretroviral treatment programme in sub-Saharan Africa: impact on treatment outcomes and implications for tuberculosis control. AIDS 2006; 20:1605–1612.
3. Seyler C, Toure S, Messou E, Bonard D, Gabillard D, Anglaret X. Risk factors for active tuberculosis following antiretroviral treatment initiation in Abidjan. Am J Respir Crit Care Med 2005; 172:123–127.
4. Lawn S, Badri M, Wood R. Risk factors for tuberculosis among HIV-infected patients receiving antiretroviral treatment. Am J Respir Crit Care Med 2005; 172:1348, author reply 1348–1349.
5. Seyler C, Toure S, Messou E, Anglaret X. Risk factors for active tuberculosis in adults on highly active antiretroviral therapy in Africa. AIDS 2006; 20:1463, author reply 1463–1465.