Abstract
Overlapping risk factors may explain observed association between HIV and drug resistance in TB patients http://ow.ly/DdRGT
To the Editor:
Recent meta-analyses have indicated that, on average, new HIV-positive tuberculosis (TB) patients are at increased risk of multidrug-resistant (MDR)-TB compared with HIV-negative patients, while this risk is less clear for previously treated TB patients [1, 2]. In a recent issue of the European Respiratory Journal, Dean et al. [3] reported an association between HIV infection and MDR-TB disease based on aggregated data reported annually to the World Health Organization. Out of 24 countries where ⩾75% of TB patients had a HIV test result and at least one HIV-positive MDR-TB case was reported, 11 showed a significant positive association between HIV and multidrug resistance among TB patients, mostly in countries with a high prevalence of multidrug resistance. One of these countries was Kazakhstan, which added data for 40 975 out of the 104 781 patients included in their report.
The authors pointed out that there are likely to be risk factors common to HIV-positive and MDR-TB patients that could not be explored in their analysis. They refer to a study from Moldova, where detailed analysis of 2007–2010 surveillance data showed that the positive association observed among new patients remained after adjustment for potential confounders [4]. We would like to refer to a similar analysis performed in Kazakhstan [5]. Using routine data from the electronic, national TB register from 2007–2011, we assessed patient characteristics associated with MDR-TB and/or HIV. In Kazakhstan, all patients are tested for HIV and TB drug resistance. HIV test results were available for 97% of TB patients and drug susceptibility test results for 93% of culture-positive patients. The proportion of TB patients with HIV increased from 0.6% in 2007 to 1.5% in 2011. The proportion of TB patients with multidrug resistance was high but did not change over time; on average, it was 36.1% for new and retreatment patients combined. In a crude analysis, among all 50 589 TB patients with HIV and drug susceptibility test results available, we observed a 20% higher prevalence of MDR-TB among HIV-positive patients than among HIV-negative patients (OR 1.2, 95% CI 1.02–1.4). Our results revealed that risk factors among TB patients for HIV and multidrug resistance were largely overlapping; in a univariate analysis, both included male sex, young adult age, urban residence, a history of incarceration, homelessness and drug abuse. Moreover, after adjustment for these overlapping risk factors in addition to demographics and treatment history, this association disappeared completely (adjusted OR 1.0, 95% CI 0.86–1.2). In addition, at the regional level, no clear association was observed between the prevalence of multidrug resistance and HIV (fig. 1). On average, 0.6% of the TB patients were infected both with HIV and MDR-TB Mycobacterium tuberculosis, but this risk varied greatly. TB patients at highest risk of being infected both with HIV and MDR-TB M. tuberculosis were those using drugs (12.5%) and those with a history of imprisonment (3.4%). These findings indicate that in Kazakhstan, the dual epidemic of HIV and MDR-TB is converging in specific socially vulnerable groups, and that enhanced efforts are necessary to provide (access to) diagnosis, TB/HIV treatment and care to these groups. This is important as both HIV and MDR-TB put patients at increased risk of treatment failure and mortality, and their combination is even more dangerous [6, 7]. Determinants of HIV and multidrug resistance are at least partly setting-specific and whether there is an independent association between them also may be situational. Therefore, it is important that more countries assess whether specific determinants may explain an apparent crude association between HIV and MDR-TB. This will require additional data collection beyond that available in routine surveillance registers in most settings.
Prevalence of multidrug resistance (MDR) and HIV among a) new and b) previously treated tuberculosis patients in Kazakhstan, 2007–2011.
Footnotes
Conflict of interest: None declared.
- Received July 18, 2014.
- Accepted August 12, 2014.
- Copyright ©ERS 2015