Updates to estimates in Global TB Report 2015
Each year, new data become available for the estimation of TB disease burden. Periodically, methods for using surveillance and survey data as well as other sources of information to estimate TB disease burden are reviewed and updated. This box provides a summary of updates that were made in 2014. Updates for specific countries that are expected in the near future, pending the finalization of results from ongoing national prevalence surveys, are also highlighted.
By August 2015, final results became available from national TB prevalence surveys implemented between 2012 and 2014 in Ghana, Malawi, Soudan, the Republic of Tanzania, Zambia and Zimbabwe. In Indonesia, burden estimates have been revised upwards (see Box a fix reference). Given the size of Indonesia’s population and TB burden, this upward revision in burden estimates affects global estimates of the absolute number of TB cases and to a lesser extent TB deaths, but not overall trends. In the other 5 countries with new prevalence survey results, updated estimates are either higher (Ghana, Malawi, Zambia) or lower (Sudan, Zimbabwe) than previous estimates, with overlap in uncertainty intervals except for Malawi.
Prevalence surveys result in estimates of prevalence that are generally more precise and more accurate than previous estimates indirectly derived from estimated incidence and heavily influenced by expert-opinion (see Figure 1). Details are provided in the online technical appendix.
There are some relatively small changes to estimates of TB incidence, mortality and prevalence for many countries that reflect new VR data reported to WHO between mid-2014 and mid-2015, updated WHO estimates of the total number of deaths from all causes (that provide overall mortality envelopes), updates to estimates of the burden of HIV-associated TB and new TB notification data including corrections made to historical data and updates to population sizes from the United Nations Population Division. In most instances, changes are well within the uncertainty intervals of previously published estimates of TB burden and time trends are generally consistent. Newly-reported data are the reason for small changes to estimates of the number of TB deaths among women and children and to the number of incident MDR-TB cases.
In Indonesia, estimates of TB mortality (HIV-negative) are for the first time obtained from the sample vital registration system, after adjustment for incomplete coverage and ill-defined causes of death. In South Africa, estimates of TB mortality (HIV-negative) were derived from the national vital registration system after adjustment for widespread miscoding of HIV and TB causes of deaths and obtained from IHME (Murray 2014). The interpretation of death statistics in South Africa cannot be made on face value as a large proportion of deaths caused by HIV infection are misclassified(Groenewald 2005).
In March 2015, the WHO Task Force on TB impact measurement reviewed methods for estimating TB burden (see Box b fix reference). The meeting recommended for the 2015 targets assessment that current WHO methods for estimation of TB incidence, mortality and prevalence should be used. The methods should not be changed now – the consensus was to “finish the cycle with established methods”. New methods should only be introduced, if available and proven to be better than current methods, after 2015.
The meeting recommended that WHO updates methods to estimate childhood TB incidence by combining with an ensemble approach estimates derived from case notifications adjusted for under-detection and under-reporting(Jenkins 2014) and estimates derived from dynamical modelling(Dodd 2014). Estimates of childhood TB incidence presented in this report reflect the recommendation. An additional recommendation from the meeting was for the estimation of HIV-positive TB mortality in children using a similar approach to that for disaggregating TB/HIV mortality by gender. This new estimate of HIV-positive TB mortality in children is incorporated for the first time in this year’s report.
Updates to burden estimates have drawn on new analyzes undertaken as part of in-depth epidemiological reviews. Epidemiological reviews are also helping to identify performance gaps in TB surveillance and form the basis of detailed and costed monitoring and evaluation plans. Estimates for Angola were revised based on discussions with experts from the National TB Programme and partners and should be considered preliminary pending the findings of an ongoing epidemiological review. Estimates for Kazakhstan reflect findings from a reassessment mission in close collaboration with the Ministry of Health.
Updates to estimates of disease burden are expected in several countries with ongoing prevalence surveys. These include: Bangladesh, Mongolia, Uganda. Future updates will be made available in online country profiles and associated datasets.
There is increase country interest in inventory studies to directly measure under-reporting of cases (i.e. the number of cases that is detected but not reported to NTPs). If such studies are implemented, they may result in updates to estimates of TB burden, especially TB incidence.