Introduction: In settings such as Namibia with a high prevalence of human immunodeficiency virus (HIV) and drug-resistant (DR) tuberculosis (TB) co-infection, interactions and adverse events associated with second-line anti-TB and antiretroviral medicines pose a unique challenge in the treatment of both infections.
Objective: The main objective of this study was to compare the absolute risks and risk factors for commonly observed adverse events (occurring in >20 % of patients) during DR-TB treatment in HIV-infected and HIV-uninfected patients.
Methods: This was a retrospective cohort analysis of patients treated for DR-TB between January 2008 and February 2010 at the Kondja DR-TB ward in Walvis Bay, Namibia. Data were anonymously collected from patients' treatment records, using a structured form. The data were then analyzed using descriptive statistics, while 2 × 2 contingency tables stratified by HIV status were employed to examine specific risk factor and adverse event relationships, using Epi Info 3.4.3 statistical software. Eighteen adverse events were studied but, because of the small sample size of patients, only the four most frequent ones (occurring in >20 % of patients) were included in the risk factor analysis. The risk factors were a treatment period of <4 weeks; treatment with any highly active antiretroviral therapy (HAART) regimen; specific treatment with a zidovudine (AZT)-based HAART regimen, a cycloserine-based DR-TB regimen or an amikacin-based DR-TB regimen; female gender; baseline body weight ≤ 45 kg; and age 30 ≥ years.
Results: Of the 57 DR-TB patients who were included in the analysis, 31 (53 %) were co-infected with HIV. When stratified by HIV status, DR-TB patients had similar exposure to specific DR-TB medicines and comparable demographic and clinical characteristics, except for age, as HIV-infected patients were on average 6.5 years older than HIV-uninfected patients (P = 0.007). Of the 18 studied adverse events, tinnitus (40 %), joint pain (26 %), hearing loss (23 %) and nausea (21 %) were the four most commonly observed events. Only for abdominal pain was there a statistically significant difference in the risk of occurrence between HIV-infected patients and HIV-uninfected patients (26 versus 4 %, P = 0.02). The risk ratios (RRs) for the association between treatment with a cycloserine-based DR-TB regimen and occurrence of joint pain did not differ much between HIV-infected and HIV-uninfected patients (RR 4.3 in HIV-infected patients, P = 0.03; RR 5 in HIV-uninfected patients, P = 0.08). Similarly, although some differences in the RRs were observed between the two HIV status groups, the differences were not statistically significant for tinnitus, hearing loss or nausea. In some instances, HIV status appeared to modify the effect of the association of some of the risk factors and adverse event occurrence, but the wide and overlapping confidence intervals were inconclusive.
Conclusion: Generally, the absolute risks and risk factors for adverse events were similar between HIV-infected and HIV-uninfected patients treated for DR-TB in our Namibian cohort of 57 patients. Although our findings of comparable adverse event risks between DR-TB and DR-TB/HIV co-infected patients are encouraging, they are inconclusive because of the low statistical power of our study. We recommend a prospective study with a larger sample size that would increase the power and therefore the confidence in the results.