|Burden of childhood tuberculosis in 22 high-burden countries: a mathematical modelling study|
||Peter J Dodd, Elizabeth Gardiner, Renia Coghlan, James A Seddon
||Lancet, Published online July 9, 2014 http://dx.doi.org/10.1016/S2214-109X(14)70245-1
More than one region
||can result even when children do present to health services. Direct incidence estimates are unavailable, and WHO
estimates build on paediatric notifi cations, with adjustment for incomplete surveillance by the same factor as adult
notifi cations. We aimed to estimate the incidence of infection and disease in children, the prevalence of infection, and
household exposure in the 22 countries with a high burden of the disease.
Methods Within a mechanistic mathematical model, we combined estimates of adult tuberculosis prevalence in 2010,
with aspects of the natural history of paediatric tuberculosis. In a household model, we estimated household exposure
and infection. We accounted for the eff ects of age, BCG vaccination, and HIV infection. Additionally, we tested
sensitivity to key structural assumptions by repeating all analyses without variation in BCG effi cacy by latitude.
Findings The median number of children estimated to be sharing a household with an individual with infectious
tuberculosis in 2010 was 15 319 701 (IQR 13 766 297–17 061 821). In 2010, the median number of Mycobacterium
tuberculosis infections in children was 7 591 759 (5 800 053–9 969 780), and 650 977 children (424 871–983 118) developed
disease. Cumulative exposure meant that the median number of children with latent infection in 2010 was 53 234 854
(41 111 669–68 959 804). The model suggests that 35% (23–54) of paediatric cases of tuberculosis in the 15 countries
reporting notifi cations by age in 2010 were detected. India is predicted to account for 27% (22–33) of the total burden
of paediatric tuberculosis in the 22 countries. The predicted proportion of tuberculosis burden in children for each
country correlated with incidence, varying between 4% and 21%.
Interpretation Our model has shown that the incidence of paediatric tuberculosis is higher than the number of
notifi cations, particularly in young children. Estimates of current household exposure and cumulative infection
suggest an enormous opportunity for preventive treatment.