BCG vaccination is effective against forms of tuberculosis that occur commonly in childhood that have a high morbidity and mortality, such as tuberculous meningitis.^1,2 Current guidelines for the United Kingdom recommend tuberculin skin testing before BCG vaccination for all children older than 3 months.³ The evidence base for this recommendation is unclear—no randomised controlled trials have been conducted to compare outcome of BCG vaccination with and without prior tuberculin skin testing, and the resulting two or three stage procedure seriously compromises uptake.

The BCG vaccine was not available in the United Kingdom from 9 August 2002—when all stocks from the monopoly supplier, Evans Vaccines, were declared potentially ineffective by the Medicines Control Agency and the Department of Health—until supplies from Denmark's State Serum Institute became accessible during December 2002. The currently available BCG vaccine is for intradermal use only. From November 1998 to July 2001 a similar loss of supply of percutaneous BCG vaccine, which is used especially by midwives for neonatal vaccination, and tuberculin occurred. The logistical difficulties we experienced to recover the ground lost by this gap in supply made us investigate the literature to see whether tuberculin testing is required before BCG vaccination in children.

Tuberculin testing before BCG vaccination might be recommended on the grounds that it will detect tuberculosis and will prevent complications due to pre-existing immunity to mycobacterial antigens. Protection from tuberculosis with the BCG vaccine does not correlate with tuberculin sensitivity.⁴ Normally BCG vaccination in people without a BCG scar gives a papule with induration at around two weeks, followed by an ulcer and healing with a scar at six weeks. In people with tuberculosis, an earlier reaction to BCG vaccine occurs, which is termed the Koch phenomenon.⁵ This early reaction characteristically gives a papule with induration that is more than 5 mm within 24-48 hours of injection, enlargement in size to form a pustule by three to five days, an ulcer at seven days, and a scab within the next three to five days. Variations include an accelerated reaction with induration at 6-12 hours of injection, a pustule on the third day, and a scab by five to six days, and a delayed reaction, although still with pustule formation at five days. All three reactions with early pustule formation indicate concurrent tuberculosis. BCG can therefore itself be used as a diagnostic test for tuberculosis and has been recommended as such by the World Health Organization,⁶ especially in children.

Udani was the first to compare the BCG test with tuberculin testing and observed that the BCG test was positive in all children with tuberculosis, whereas 28.8% of tuberculin skin tests were negative.⁷ Subsequent studies in children up to 12 years of age confirmed the greater sensitivity of the BCG test (81-92% sensitivity, 82.5-100% specificity), compared with tuberculin (sensitivities 45.5-52.3%, with specificities comparable to the BCG test).^8,9,10 The BCG test was especially helpful in children who were malnourished or who had tuberculous meningitis or miliary disease. Naturally treatment for tuberculosis will also kill live BCG and therefore avoid any serious adverse reaction from vaccination in these circumstances.

Complications in using the BCG vaccine without prior tuberculin skin testing were largely due to subcutaneous administration and were no more common in children with tuberculosis than in healthy children; seven of 1129 vaccinated developed regional lymphadenitis.^8,9,10 A review of published and unpublished data, including a survey sponsored by the International Union against Tuberculosis, recorded 10 371 complications following almost 1.5 billion BCG vaccinations in adults and children.¹¹ Disseminated BCG (3 per million) and death (0.02 per million) due to primary immunodeficiencies were the most serious complications, but would not have been detected by prior tuberculin testing. BCG vaccination is not recommended in children who are HIV positive. Most studies included individuals with positive tuberculin responses, some with indurations greater than 16 mm.¹² Necrotic reactions (282) were most common in people who had been vaccinated before, rather than in those who were tuberculin positive without tuberculosis. Subcutaneous abscesses and regional lymphadenitis were not distinguished by prior tuberculin status. Tuberculous skin lesions were more common in people over 15 years or with revaccination.

BCG vaccination in children with tuberculosis is not associated with more complications, but rather an earlier reaction. BCG vaccination without prior tuberculin skin testing in young children would ensure a greater uptake, reduce clinic visits by parents and their children, reduce the workload of clinic staff, and identify contact with tuberculosis with a greater sensitivity than the tuberculin skin test. The requirement for tuberculin skin testing after the age of 3 months before BCG vaccination is not supported by the literature and seriously hampers the effective implementation of the neonatal BCG programme. We think that tuberculin skin testing is not required before BCG vaccination in young children.