Research in context
Evidence before this study
All ten published randomised trials of BCG were reviewed for evidence for the duration of BCG efficacy against tuberculosis by Sterne and colleagues in 1998. This review was complemented by a comprehensive systematic review by Abubakar and colleagues in 2012, which also included all observational studies. Abubakar and colleagues searched for articles in electronic medical databases up to May 31, 2009 (including MEDLINE, Embase, Cochrane Central Register, and others), and in trial registers and grey literature sources. Search terms for disease were “TB”, “tuberculosis”, “tubercle bacill*”, “M. tuberculosis complex”, “M. bovis”, “M. africanum”, “M. canetti”, “M. microti”, and “M. tuberculosis”, and for intervention were “BCG vaccine”, “BCG”, “BCG vacc*”, “BCG imm*”, and “Bacillus Calmette”. Details of all databases searched are published in the report. We repeated the search strategy in MEDLINE, Embase, Cochrane Central, and Web of Knowledge, but identified no new additions. Authors of the most recent review suggested that BCG is, on average, effective against tuberculosis if given to individuals not already infected with Mycobacterium tuberculosis or sensitised by environmental mycobacteria, and the vaccine can protect for 10–15 years. Pooled VE estimates (against all forms of tuberculosis disease) from trials were 60% (95% CI 37–74) for 0 to less than 5 years, 56% (17–76) for 5 to less than 10 years, and 46% (18–64) for 10–15 years. Seven of the ten trials provided some data for follow-up beyond 15 years after vaccination, but investigators of only one noted evidence of protection, whereas the others had too few events for meaningful estimates. Abubakar and colleagues also identified 22 relevant observational studies (consisting of five cohort, five case-population, nine case-control, and three cross-sectional studies), of which only four had some data for BCG effectiveness up to 20 years after vaccination. Authors of three of these studies suggested decreasing but persisting protection 15–20 years after vaccination. Overall, evidence is consistent with significant BCG-derived protection against tuberculosis for 10–15 years after vaccination, with waning over time; however, the vaccine effect beyond that period is uncertain.
Added value of this study
Long-term follow-up of participants in the Native American and Alaska Natives BCG trial suggested that BCG could protect against tuberculosis for up to 60 years. Our study is, to our knowledge, only the second (and the first from western European countries) in which some of these findings are replicated. Our results suggest a BCG protection of about 50% during 40 years, with some evidence of about 40% effectiveness 30–40 years after vaccination. The consistency between results from these two settings strengthens the hypothesis that BCG-derived immunity could persist for much longer than 10–15 years as previously assumed.
Implications of all the available evidence
A longer duration of protection than that currently thought would imply that BCG is potentially more cost effective and beneficial than was previously estimated. This finding could be relevant if countries revise their BCG vaccination policies in response to changing tuberculosis epidemiology, especially in low-incidence countries. The potentially long-lived effect of BCG should be investigated and taken into account in development of new tuberculosis vaccines, especially in view of our low understanding of immunity to M tuberculosis. The duration and changes in levels of BCG-derived protection would also be relevant to scheduling of vaccination if the new family of BCG-booster vaccines was successfully developed and introduced.