br There is growing recognition that children bear a

There is growing recognition that children bear a substantial burden of tuberculosis morbidity and mortality. However, in regions where the diagnosis is contingent on smear microscopy, the burden attributable to children, who usually have paucibacillary disease, is grossly underestimated. The downstream ramifications of inaccurate estimates include inadequate attention and funding for childhood tuberculosis surveillance, diagnosis, treatment, prevention, and research. In the absence of microbiologically based diagnoses, estimates of tuberculosis in children have been based on extrapolation from adult data. In this issue of , Peter Dodd and colleagues provide new estimates of the burden of childhood tuberculosis in the 22 high-burden countries. By Imiloxan hydrochloride cost with standard estimates that are contingent on paediatric case reporting, which varies widely between countries, these estimates used country-specific household-based and population-based data. WHO first disaggregated paediatric data in 2012 and has not included children with HIV infection in the estimates of mortality. The estimates by Dodd and colleagues differ substantially from WHO estimates in terms of incidence rates and the proportion of disease in children. The median number of incident cases in the high-burden countries (650 997, IQR 424 871–983 118), although almost 25% higher than the WHO global estimate of 530 000 paediatric cases in 2012, is more consistent with other estimates. Jenkins and colleagues used data from a systematic review to estimate that, in 2010, almost 1 million children developed the disease. These higher estimates seem to reflect global population structure more accurately; roughly 25% of the world population is estimated to be younger than 15 years, with a range of 15–49% (median 31%) in high-burden countries. Dodd and colleagues also provided estimates of the burden of tuberculosis in HIV-infected children, which infrequently has been disaggregated from adult tuberculosis and HIV co-infection. The overall estimated case detection rate of 35% was substantially lower than the WHO estimate of 66% case detection in children. This finding has profound implications for resource allocation and, combined with the estimated disparities in the burden of disease borne by children in the high-burden countries, prompts questions about how to make paediatric diagnosis and reporting easier. Although molecular methods such as Xpert MTB/RIF offer improved sensitivity over smear microscopy, their role is restricted to settings in which respiratory specimens can be obtained from younger children and to situations in which tuberculosis is suspected. Findings of a Peruvian study showed that a negative tuberculin skin test strongly predicted death in children admitted to hospital with tuberculosis, possibly because the diagnosis was considered initially but discarded on the basis of this insensitive test. Additionally, Dodd and colleagues modelled tuberculosis infection and estimated a prevalence of 50 million infected children. Because infection is not reportable in most settings, few population-based infection estimates exist. The new estimates allow for future benchmarking of preventable disease cases (eg, young children who are household contacts of cases, who did not receive isoniazid preventive therapy, and who progressed to disease). Although several studies have reported poor use of isoniazid preventive therapy as per WHO guidelines in young exposed children, broader population-based estimates of infection would encourage countries to alter resource allocation to improve secondary prevention. The study by Dodd and colleagues provides a strong foundation for future studies attempting to model the burden of childhood tuberculosis. The model incorporates both social and epidemiological variables. It would be intriguing to apply the model to low-incidence nations, where more complete paediatric data are available. If the model could be validated in these countries, confidence in its accuracy might be increased. Also of interest would be another sensitivity analysis changing the estimates of adult prevalence to see the effect on estimates in children. Variables that could be included in future models include modelling risk of tuberculosis exposure outside the home, especially in high-incidence settings where up to 40% of exposures stem from non-household contacts. Finally, there is growing evidence that only a minority of adults with culture-confirmed pulmonary tuberculosis produce culturable aerosols, which might be a better marker of infectiousness than smear positivity. Integration of these data into future models could result in more accurate estimates.