Article Text
Statistics from Altmetric.com
Wang et al1 are right when they underline the need to adjust for confounding variables when analysing the results of a tuberculin skin test (TST) survey, but they are oversimplistic when, using surveys done in many countries worldwide, they study only the effect of the BCG vaccination on the TST results. The authors did not discuss the rate of atypical mycobacterial infection in a population, an important confounding factor that drives up the rate of tuberculous infection, because atypical mycobacteria crossreact with Mycobacterium tuberculosis. Unlike Wang et al, at least one TST survey included in the meta-analysis, done in Kenya,2 adjusted for the rate of atypical mycobacterial infection. A study performed in the Netherlands on army recruits which double tested for both M tuberculosis and atypical mycobacteria showed that 48% of the reactors with indurations in the range 10–15 mm and 16% with indurations ≥10 mm were classified as false positive.3 There was no interaction between BCG and TST as long as the recruits were not BCG vaccinated. False positives were then excluded from the latent tuberculosis infection treatment.
Another confounding factor in this meta-analysis is pooling together TST results from surveys performed in populations during the post-war period of starvation4 with those from developed countries without considering the effect of a low calorie diet on the TST results. Only 3–4 months after BCG vaccination the rate of apparent TST conversion in malnourished children may be as low as 23% compared with 85% in well nourished children.5
The rate of atypical mycobacterial infection might therefore have been confused for the TST positivity due to BCG vaccination, while the effect of BCG vaccination on the rate of TST positivity was underestimated in starving populations.
The author did not need financial support to write this letter and does not have any material interest in the subject.
References
Authors' reply
We would like to thank Dr Jalba for the thoughtful comments on our paper. As the prevalence of atypical mycobacterial infection was not systematically evaluated in the studies we reviewed, we could not integrate this into our analysis. From a practical point of view both BCG and non-BCG groups are likely to have similar exposures to atypical infection, and clinicians would not be able to assess routinely for such infections. Similarly, information on nutrition was not systematically available but, as a surrogate, we have looked at the impact of BCG as one moves further from the equator and found no differences in its impact. On the basis that malnutrition would be a greater confounder in studies done closer to the equator, this suggests that nutrition is not a significant factor in the sample size we have generated.