Global burden of latent multidrug-resistant tuberculosis: trends and estimates based on mathematical modelling

Abstract

Background: To end the global tuberculosis epidemic, latent tuberculosis infection needs to be addressed. All standard treatments for latent tuberculosis contain drugs to which multidrug-resistant (MDR) Mycobacterium tuberculosis is resistant. We aimed to estimate the global burden of multidrug-resistant latent tuberculosis infection to inform tuberculosis elimination policy.

Methods: By fitting a flexible statistical model to tuberculosis drug resistance surveillance and survey data collated by WHO, we estimated national trends in the proportion of new tuberculosis cases that were caused by MDR strains. We used these data as a proxy for the proportion of new infections caused by MDR M tuberculosis and multiplied trends in annual risk of infection from previous estimates of the burden of latent tuberculosis to generate trends in the annual risk of infection with MDR M tuberculosis. These estimates were used in a cohort model to estimate changes in the global and national prevalence of latent infection with MDR M tuberculosis. We also estimated recent infection levels (ie, in 2013 and 2014) and made predictions for the future burden of MDR tuberculosis in 2035 and 2050.

Findings: 19·1 million (95% uncertainty interval [UI] 16·4 million-21·7 million) people were latently infected with MDR tuberculosis in 2014-a global prevalence of 0·3% (95% UI 0·2-0·3). MDR strains accounted for 1·2% (95% UI 1·0-1·4) of the total latent tuberculosis burden overall, but for 2·9% (95% UI 2·6-3·1) of the burden among children younger than 15 years (risk ratio for those younger than 15 years vs those aged 15 years or older 2·65 [95% UI 2·11-3·25]). Recent latent infection with MDR M tuberculosis meant that 1·9 million (95% UI 1·7 million-2·3 million) people globally were at high risk of active MDR tuberculosis in 2015.

Interpretation: We estimate that three in every 1000 people globally carry latent MDR tuberculosis infection, and prevalence is around ten times higher among those younger than 15 years. If current trends continue, the proportion of latent tuberculosis caused by MDR strains will increase, which will pose serious challenges for management of latent tuberculosis-a cornerstone of tuberculosis elimination strategies.

Funding: UK Medical Research Council, Bill & Melinda Gates Foundation, and European Research Council.

Figure 1

Proportion of new cases of tuberculosis disease accounted for by MDR tuberculosis in the nine countries in the WHO South-East Asia region included in our model The blue lines represent the median proportion from 200 model fits to WHO data (the red datapoints; error bars show 95% CIs). The shaded regions represent the 95% uncertainty intervals. Although we estimated the burden of latent MDR tuberculosis in 2014 (hence the cutoff in this figure) the model trend was fitted to all WHO Drug Resistance Surveillance data. MDR=multidrug resistant.

Figure 2

Prevalence of latent multidrug-resistant tuberculosis infection, by WHO region The red line represents the median from 200 model fits. The shaded region represents the 95% uncertainty interval.

Figure 3

Estimated worldwide prevalence of latent multidrug-resistant tuberculosis infection Countries with no data are shown in grey. Equivalent maps for 2035 and 2050 estimates are in the appendix.

Figure 4

Prevalence of latent multidrug-resistant tuberculosis infection in each age group, by WHO region Error bars show 95% uncertainty intervals. The prevalence of latent infection with drug-susceptible tuberculosis is shown in the appendix.