According to a 2011 (news) article in Science

More than a third of the world's population is estimated to be infected with Mycobacterium tuberculosis. Most people's immune system can keep the bacteria in check, but there is a lifetime chance of 1 in 10 that the dormant infection will progress to TB; the disease still kills 4000 people every day.

N.B. according to Wikipedia (citing the WHO)

People with latent TB do not spread the disease.

Also, Wikipedia mentions that latent TB infections are diagnosed through tuberculin skin test. But looking at the article Wikipedia links for the latter, it mentions quite a few caveats for this test:

Tuberculin skin testing

This test measures a patient's immune response to M. tuberculosis antigens (tuberculin). A small amount of tuberculin is injected intradermally and the skin reaction is measured two or three days later.

The test is very sensitive for detecting tuberculosis in healthy individuals if 5 mm induration is used to define a positive reaction. However, many conditions result in false negative reactions, including active tuberculosis. Bacillus Calmette-Gurin (BCG) vaccination and exposure to environmental nontuberculous mycobacteriosis cause intermediate size reactions. Sensitivity is often sacrificed by choosing larger indurations to define a positive reaction based on the incidence of tuberculosis and the extent of non-specific cross-reactivity in the population being tested. Box 2 provides general recommendations for categorising skin reactions, but regional tuberculosis control units should be consulted for local guidelines.

Box 2: Criteria for defining a tuberculin skin testing reaction as positive *

≥5 mm – in people with recent exposure (within 2 years) to tuberculosis + high risk for progression to active disease (e.g. <5 years of age, HIV infection, other immunosuppressive illness; see Table 1)

≥10 mm – in people with recent exposure to tuberculosis, regardless of BCG vaccination status; all non-BCG vaccinated people except for those with both low lifetime risk for tuberculosis infection and residence in geographical areas where exposure to environmental nontuberculous mycobacteriosis is common

≥15 mm – in all people regardless of BCG vaccination status

* This refers to the induration produced by an intradermal injection of purified protein derivative (PPD) equivalent to 5 units of PPD-S. These criteria are meant as suggestions only. Local tuberculosis control units should be consulted for local guidelines.

So it seems rather non-trivial to assess the tuberculin skin test results, i.e. there are quite a few confounding factors and the criteria for judging a test positive [at least in that paper] are influenced by other factors linked to the potential risk of developing active disease (e.g. HIV infection).

So, given all those complications, going back to the first quote (from Science), how well is the percentage of latent TB infections bracketed/known? I.e. what's the confidence interval for true latent TB infections?

  • A non-COVID question at last! ;-)
    – Carey Gregory
    Commented May 16, 2020 at 22:53
  • @CareyGregory: yeah, but still a respiratory illness... Commented May 16, 2020 at 22:55
  • Still, at least it's a different respiratory illness.
    – Carey Gregory
    Commented May 16, 2020 at 23:01
  • I wouldn't use a tst. I'd use an IGRA. Commented May 17, 2020 at 5:50

1 Answer 1


Yes, estimating the prevalence of latent tuberculosis infection using the TST is not trivial.

Use only of the TST to define latent tuberculosis infection requires specification of a cut-point for positivity. There is a trade-off between sensitivity and specificity when selecting this cut-point. In prevalence surveys, this is usually >10 mm induration.

More important, the TST is not the only test that can be used to identify latent tuberculosis infection. This can also be done using interferon-gamma release assays (IGRA). There are two widely available IGRAs-- QuantiFERON-gold-in-tube (QFT-GIT) and T-SPOT. The IGRA’s also have associated sensitivity and specificity in identifying latent tuberculosis infection.

A published 2019 systematic review and meta-analysis reported estimates of the prevalence of latent tuberculosis infection based on studies that used TST and IGRA results published between 2005 and 2018. The review identified 88 studies from 36 countries with 41 IGRA estimates (n=67, 167 people) and 67 TST estimates (n=284,644 people). Based on these studies, the authors estimated that the global prevalence of latent tuberculosis infection was 24.8% (95% CI: 19.7–30.0%) based on IGRA and 21.2% (95% CI: 17.9–24.4%) for the TST using a 10 mm cut-off for positivity.


Based on the systematic review, it seems reasonable to say that the range of estimates of the prevalence of latent tuberculosis infection is 17% to 30%. But this is not a “confidence interval” in a statistical sense. Based on a systematic review, perhaps it could be stated that “with reasonable certainty, the world prevalence of latent tuberculosis infection is not less than 17% and not more than 30%.” This needs to be associated with a time period—2005-2018.

A 2016 mathematical modeling study estimated that “in 2014, the global burden of LTBI was 23.0% (95% uncertainty interval [UI]: 20.4%-26.4%).”


The model based estimate may be more valid than the estimate based on studies. The model formally incorporates uncertainty.

WHO now (7/31/2020) estimates that “About one-quarter of the world's population has latent TB.” The website does not explain how this estimate was derived and does not mention uncertainty in the estimate.

https://www.who.int/news-room/fact-sheets/detail/tuberculosis/ accessed 7/31/2020

Cohen A, Mathiasen VD, Schön T, Wejse C. The global prevalence of latent tuberculosis: a systematic review and meta-analysis. Eur Respir J. 2019 Sep 12;54(3). pii: 1900655. doi: 10.1183/13993003.00655-2019. Print 2019 Sep. PubMed PMID: 31221810.

Houben RM, Dodd PJ. The Global Burden of Latent Tuberculosis Infection: A Re-estimation Using Mathematical Modelling. PLoS Med. 2016 Oct 25;13(10):e1002152. doi: 0.1371/journal.pmed.1002152. PMID: 27780211; PMCID: PMC5079585.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.