Article Text


Tuberculin testing, BCG and tuberculosis today
  1. J Moore-Gillon
  1. Correspondence to:
    Dr J Moore-Gillon
    Department of Respiratory Medicine, St Bartholomew’s and Royal London Hospitals, London EC1A 7BE, UK;

Statistics from

Little progress has been made in the last 50 years

Tuberculosis is not simply a problem hanging over from the past—it is a growing challenge—but what are the risks of acquiring tuberculosis infection in childhood, and how should those risks be assessed?

In this issue of Thorax Leung and colleagues1 analyse the results of tuberculin skin testing in over 21 000 children in Hong Kong aged 6–9 years between October 1999 and February 2000. 99% of the children had received BCG vaccination in the neonatal period. The authors used three different approaches to the data to estimate the subsequent annual risk of tuberculous infection. In broad terms, one approach used the size of the tuberculin response to assess whether infection had occurred, using a cut off point of ⩾10 mm induration after skin testing with one unit of tuberculin. They then used age (which equals the number of years each child had been at risk of infection) to estimate the annual risk. This is the “classic” method. The second approach estimated the annual risk of infection by comparing the rates of tuberculin positivity in the 8–9 year old age group with those in the 6–7 year old group. The third estimated the prevalence of infection by locating a secondary peak in the tuberculin reaction distribution curve and assuming that, among those with tuberculosis infection, there would be a symmetrical distribution of reaction sizes around this peak. Age can then be used to calculate the annual risk of infection.

The first and second of these methods are conventionally used to estimate the annual risk of tuberculous infection, and such estimates are necessary as part of the calculations made about the cost effectiveness of BCG programmes. Leung et al show, however, that in their population the first and second methods gave estimates for the annual risk of infection which were some three times higher than the estimate produced by the third method. It would be easy to conclude that the third (“secondary peak”) method is not a valid approach which underestimates the risk. It is, however, this third non-conventional method which gives an estimate for the annual risk of infection which agrees best with that which would be expected from the prevalence of active tuberculous disease (in contrast to tuberculous infection) in children.

If these findings are applicable to other populations, what is their importance? This paper refers to a population with prior (neonatal) BCG vaccination, and it is possible that the tuberculin skin test is a poor indicator of infection in the presence of BCG. Certainly, it makes it more difficult to interpret the test in the clinical context. The principles by which the results have been analysed appear, however, to be equally applicable to a population which has not been near universally vaccinated, and the results suggest that the actual risk of tuberculous infection may be lower than that implied by “conventional” methods of its calculation. This may influence decisions about the cost effectiveness of mass BCG vaccination programmes, such as that currently offered to 13–14 year olds in the UK.

But there are wider issues here. The authors have used the very best available methods of testing for tuberculous infection. These tests are part of the screening programme which is carried out before offering BCG vaccination—the vaccine which is again the very best available. It is, however, a poor reflection on the priority which has been accorded tuberculosis that exactly the same study could have been carried out decades ago, because neither the best method of diagnosing tuberculous infection nor the best vaccine against tuberculosis have changed at all in that time. Quite clearly, this is not a criticism of these authors: they have carried out a first class study with the latest fully validated tools available. The trouble is that, in this area, “latest” means 50 or more years old. Investigation of new diagnostic techniques for tuberculous infection (rather than disease) and the development of new vaccines have indeed been going on, largely by a limited number of research groups who have struggled to find funding for their programmes. It is a reflection on funding bodies and political priorities—not the quality of the researchers—that, unlike probably any other disease, the “gold standard” for this study as carried out in 2000 was the same as it would have been had it been carried out in 1950.

How good is that gold standard? Not that good. The value of tuberculin testing as an indicator of tuberculous infection is hindered by prior BCG vaccination and by exposure to environmental mycobacteria, hence the complex assumptions about the relationship between tuberculin dose and reaction size which have to be made by these and all other authors. The efficacy of BCG vaccination varies widely in different populations throughout the world, and even at its very best it does not approach the levels of protection achieved with vaccines against other diseases.

Progress is being made in finding better ways of identifying latent tuberculous infection. The identification of antigen specific T cells in individuals infected with M tuberculosis may be achieved by demonstrating the release of interferon-γ when they are appropriately stimulated.2,3 If the antigens ESAT-6 or CFP10 are used, discrimination is possible between individuals who have been infected with M tuberculosis and those who have received prior BCG vaccination4,5—a clear advantage over tuberculin testing.

Turning to vaccine development, matters are less far forward. Many approaches have been tried in animal models but few are at the stage of even phase 1 clinical trials. Possible new vaccines are usually modified forms of BCG, are mutant strains of M tuberculosis, or are prepared from subunits of M tuberculosis.6 New vaccines now have to overcome the hurdle of being safe to use in at least the great majority of HIV positive individuals if their benefits are to outweigh their risks in many populations at risk of TB. A greater understanding of the mycobacterial genome undoubtedly helps in identifying possible ways forward,7 but at the present rate of research progress a replacement for BCG is some way off.

There is, then, much to learn from the study carried out by Leung and colleagues. At the most superficial level, it reminds us that tuberculosis remains a threat to children; it ought not to be. It then suggests that the methodology usually used for estimating the annual risk of tuberculous infection in children may be inappropriate, with consequences for screening and BCG programmes. Finally, the nature of the tools that were available to the authors shows that there has been a comprehensive failure on the part of the political and medical establishment to invest adequately in research into one of the leading infectious threats to global health.

Little progress has been made in the last 50 years


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.