Setting: Karonga district, northern Malawi.
Objective: To assess the sensitivity and repeatability of BCG scar reading, and factors affecting scar size.
Design: Follow-up of individuals aged > 3 months who were recruited into a BCG vaccine trial (1986-1989), and of infants vaccinated in health centres (1989-1991), who were examined for presence and size of BCG scars in subsequent years. All examinations were carried out blind of information on true vaccination status or the results of previous examinations.
Results: For trial individuals who were considered scar negative at recruitment and received BCG, the sensitivity of scar reading was > or = 93%, repeatability was > or = 94% for those < 60 years old at vaccination, and only around 1% were assessed as having > 1 BCG scar post-vaccination. For infants vaccinated when < 1 month old in health centres, the proportion who still had recognisable scars 4 years later was < 80%. Scars were larger in individuals with a prior BCG vaccination, and for those aged 15-59 at vaccination the scars were approximately 1 mm larger for males than for females.
Conclusions: A BCG scar is a highly sensitive and repeatable indicator of vaccination status when the vaccine is properly handled, delivered appropriately, and given at over 3 months of age, but not for vaccinations given within 1 month of birth. Given that most vaccinations in the world are given soon after birth, this low sensitivity will lead to both vaccine coverage and vaccine efficacy being underestimated in studies in which vaccination status is inferred from the presence/absence of a distinctive BCG scar. Age-sex patterns identified for scar size show important similarities to those found with skin test responses to tuberculin.