The burden of tuberculosis disease in children

Abstract

In The Lancet, Helen Jenkins and colleagues reported a new global estimate of nearly 1 million incident cases (999 792; 95% CI 937 877–1 055 414) of tuberculosis in children in 2010, which was almost double the latest WHO estimate of 530 000 cases in 2012 (95% CI 510 000−550 000). The new estimate is flawed because it is greatly aff ected by the estimated proportion of tuberculosis cases that have smear-positive pulmonary disease in children (aged 0−14 years) and in six other age groups, with parameter values assumed to be the average of those in notifi cation data for Norway in 1951–72 and the USA 1985−87. Such historical data cannot be considered globally representative nowadays. Although the authors cite five recent studies to support their use of the historic values, these studies show wide variability in the proportion of children with smear-positive pulmonary tuberculosis (6−16% compared with 8·6% in Norway in 1951–72 and 8·9% in the USA in 1985−87) as do recent notifi cation data reported to WHO, with regional values of 4·2−22·9% and a global average of 15·2% (table, appendix). For example, if the methods of Jenkins and colleagues are applied using 2007−12 notifi cation data for high-income countries including Norway and the USA, then the result is a global estimate of about 630 000 (95% CI 470 000−830 000) incident cases in children, far outside the authors’ own uncertainty range and closer to the WHO estimate. The new estimate of nearly 1 million incident cases of tuberculosis in children in 2010, and the derived estimate of 31 948 cases of multidrug-resistant tuberculosis, are thus biased and the uncertainty is clearly understated. As suggested by this new estimation attempt, available methods are imperfect. The ultimate goal is to measure childhood tuberculosis incidence reliably through surveillance systems that capture most cases, with diagnosis made, whenever possible, on the basis of mycobacteriological confirmation. Until then, much uncertainty about the burden of disease in children will persist. WHO and partners are pursuing innovative ways to improve burden estimates, including better quantification of under-reporting and use of dynamic models to triangulate estimates from notifi cation data.