top of page

Extra-pulmonary TB - and missing millions?

Extrapulmonary TB represented 14% of the 6.4 million incident cases that were notified in 2017, ranging from 8% in the WHO Western Pacific Region to 24% in the WHO Eastern Mediterranean Region.”[i]

This is a direct quote from the last WHO Global TB Report (and another one is due very soon). They published a helpful map to illustrate this identifying where the highest proportions of extrapulmonary TB (EPTB) are known to exist.

It’s a revealing one. You’ll see immediately that, apart from Mongolia and a couple of smaller countries in Central Asia, no higher burden TB countries appear to have found much extrapulmonary TB (EPTB) at all. In fact most have reported 0-19% (and of course we already know from the sentence at the top that the global average of notified cases is 14%).

This is curious – in fact we think that it may be more evidence of an underestimation of disease (and therefore ongoing neglect). We’ll try to explain why.

Firstly, we think it’s curious that there was no estimate anywhere in the 174 pages of Global Report as to how much EPTB might actually be out there. There are many estimates of other aspects of this complex disease – of DS-TB, MDR-TB, childhood TB, HIV-TB - even XDR-TB if you look for it. (In fact in all 174 pages of ‘Main Report’ there are only eleven mentions of EPTB at all.)

At superficial first reading, you might easily think that 14% was an estimated rate of overall TB that is extrapulmonary (meaning that it’s relatively inconsequential in comparison to the wider pandemic). But it’s not inconsequential at all as we’ll see below: the real percentage may be at least twice this, if not higher.

Which takes us to the fact that this percentage is being reported as a proportion of all of the 6.4 million who were notified as having TB and who were actually diagnosed as having EPTB (i.e. 896,000) without any additional commentary. There seem to be several deficiencies as a consequence:

1. there’s no discussion around the invisible implications of this reported percentage;

2. there’s no estimate of the total number of EP cases who may be occurring within the pandemic;

3. there's not even an estimate estimate of the number of EP cases that may be being missed among the cases who were notified;

4. and there's no report of the success rate of those EPTB cases who were notified.

So the rest of this blog offers up a very rough-and-ready attempt to rectify this.

What this may add up to

The last Report was clear enough that it’s believed that globally 3.6 million TB cases were still being missed during 2017, so we might reasonably believe that a similar percentage of this host of missed cases (14%) may have also had EPTB (another 504,000, in other words, making a grand total of 1.4 million EPTB cases occuring in 2017) – but assuming this could be very misleading indeed. In fact we suspect that the number is much higher – and that this may be making for an underestimation in the overall numbers being officially reported. (We will have our own rough and ready stab at some numbers at the end of this blog).

‘So what?’ you may say. You may even do so because you’re an epidemiologist or because you know a lot about TB, and you know that folk with EPTB generally aren’t infectious because they don’t cough and splutter out their infections like pulmonary TB cases are so wont to do. In other words, they don’t contribute to the wider pandemic of infectious TB.

Well, that's true - and maybe that’s exactly why they’re being neglected – because they’re not perceived to be a threat, but simply amount to being unfortunate uncounted casualties who don’t matter. But surely they should be estimated, then found and treated like all TB cases. And besides that, if you're an epidemiologist, you'll also knowthat the prognostic indicators of EPTB are generally even bleaker than for pulmonary TB (including greater frequency of fatal complications, and a more severe decrease in quality of life). In other words, this particular host of neglected souls are basically more likely to die than the millions of other neglected souls who have TB but who are also being missed, and what's more they may also suffer more even if they don’t.

We need to take this apart bit by bit, to explain further as to why we have these concerns.

Where the darker shades in the map are

Firstly, let's take another look at where the darker shades of purple are on the map and review the proportions of disease.

You’ll see that many of them occur over countries which are ‘richer’ (i.e. they have higher incomes, larger GDPs,and better health infrastructures as a result). This has to suggest something important: it implies that generally wherever there is more health resource, there’s a higher proportion of EPTB being found – and this further suggests that it is simply because there’s more capacity to find it (and, of course, to treat it as well).

There’s very clear evidence of this in the ‘Tuberculosis in England 2019 Report’ which has been published very recently by Public Health England.[ii] England, of course, is a far richer country than any country with a high burden of TB (England is currently officially defined as a 'low burden country, in fact.

Two important points can be made from this Report.

Point 1: On page 35 of the Report it states that in 29% of those diagnosed with pulmonary TB cases simultaneously carried a co-infection in another infection site (had concurrent extrapulmonary-pulmonary tuberculosis). This is a type of coinfection that isn’t even mentioned anywhere in the WHO’s Global Report (in fact is barely mentioned anywhere). The risk of this concurrent infection, incidentally, has been found to be higher in HIV coinfections. Of course, these cases should hopefully respond positively to normal TB treatment anyway (and in England will have been treated by appropriate drugs with an 84.7% success rate) – but this won’t have happened for these cases in England if their coinfection was of a different strain to the one in the lungs and is drug-resisistant, of course (which is quite possible).

Point 2: But if we focus only on those who were found to have 'EPTB only (in other words there was no evidence of TB found in the lungs at all) we find that Public Health England is reporting, not 14% of notified cases as having EPTB in England in 2018, but 43% - in other words three times more than the global average.

So is this reported percentage a freak number, something associated with the nature of TB in richer nations, or is it maybe a reflection of what might be typical of a better diagnostic resource?

It certainly seems worth exploring this further.

Is the Public Health England number a freak report?

This is quite possible, and perhaps we should hope that it is because, if 43% of global TB generally is EP, it suggests that an awful lot more TB disease is being missed than is currently being estimated to be the case (and what’s already being estimated to be missed is bad enough).

Thankfully we can find a couple of reports that suggest that the English 43% proportion of TB being EP may indeed not be typical of the overall global epidemic.

The first is an American Report (“Rate of Extrapulmonary TB highest among blacks”).[iv]

This 2010 study identified the fact that the rates of EPTB were significantly higher among black TB patients in the U.S. compared to non-blacks (31.5 percent compared to 24.3 percent). Both these percentages can, of course, be seen to be significantly higher than the 14% quoted in the WHO Global Report, but they’re certainly also both much less than the UK’s 43%.

Then we found a big recent Chinese study which looked at nearly 20,000 hospitalised TB cases which recorded a more modest 33% of them as being EPTB.[v]

So we find ourselves wondering whether this percentage (about a third of all TB cases) is nearer the mark worldwide – which it’s still more than double the 14% proportion of EPTB cases being actually reported worldwide, so suggests both an underestimation of global TB burden and an underestimation of TB mortality, as well as flagging up a missing component in the global response to TB since this percentage isn’t being estimated anywhere.

So who’s most at risk of EPTB? Are some types of patients more prone than others?

We’ve already identified a paper (the American one above) which suggested that blacks may be more vulnerable to EPTB than non-blacks. Whether this is the case worldwide is immediately debatable, but must be worth considering. The American paper proposed a few reasons why the odds may be so much higher if you’re black. These include higher rates of HIV coinfection among black TB patients, and also ‘being of foreign birth’. If we’re thinking globally, then both of these factors surely should raise our eyebrows given that countries with high burdens of HIV and TB with predominantly black populations aren’t reporting much EPTB at all (take another look at the map), and that the same can be said of the typical countries of ‘foreign birth’ that these black TB cases may have typically originated from before arriving in the U.S...

Note - the current proprtion of notified TB cases in the U.S, is between 20-29%.

Of added significance, the authors of the American paper also identified that differential exposure rates to infectious TB plays a potential part, along with access to medical care, and socioeconomic status. Again our eyebrows should surely rise – given that most countries (black or non-black) with higher rates of circulating TB, with poorer access to medical care, and whose citizens typically have lower economic statuses aren’t reporting so much EPTB either. (Just check the map again!).

If this paper’s reasoning is right, then alarm bells should must surely be going off in Geneva regarding the TB epedemic in the African region, because it ticks all of these risk factor boxes. The rate for this region was actually most recently reported to be 16%, incidentally - when it looks likely that it might really be double this or possibly more.

This issue of ‘foreign birth’ as identified in the American paper is further echoed in the Public Health England Report (the one that pegs the rate of EPTB as being 43% among new notified cases). It also records higher rates of EPTB being found among those born outside the UK (a rate of 48.5% in such TB cases) compared with those born in the country (a much lesser 28%).

The implications from both America and England are worrisome. They suggest that two countries which are reporting low rates of TB but higher rates of EPTB are finding the EP disease among people who originate in the countries that are reporting lower rates of EPTB but which have far higher TB burdens. A logical conclusion is that a significant amount of EPTB must be being missed in high burden countries if so much less EPTB is being found in these same countries.

And if this really is true, we feel it right to identify that this deficiency isn’t even being suggested, let alone addressed, in the current global strategies to end TB.

According to the WHO Report, incidentally, the WHO region with the lowest reported rate of EPTB among notified TB cases, was the Western Pacific region - reported at a lowly 8%. Given that this region includes China (with the report referenced above reporting 33% of 20,000 hospitalised cases being EPTB), it suggests that there may well be a big problem in this region as well, suggesting that the vast majority of EPTB cases in the Western Pacific region may be being missed completely and left untreated.

HIV co-infection

The HIV risk factor is certainly significant because it’s universally accepted that HIV co-infection increases the risk of EPTB, although it’s also worth adding that this risk will be dependant on whether or not the HIV is being treated with ARV drugs before TB reactivation or afterwards. In fact EPTB has been historically reported as occurring in “more than 50% of patients with concurrent AIDS and tuberculosis”[vi] - a proportion that is without question the highest percentage we’ve encountered anywhere.

There’s something else that worrying, however, which is that a higher proportion of MDR-TB is found in EPTB cases than in pulmonary TB ones. The good news about this is that the disease won’t be infectious so shouldn’t add to the MDR crisis. The bad news is that this certainly won’t make EPTB any easier to treat when the global authorities get around to it.


As just mentioned, a higher proportion of MDR-TB is being found in EPTB cases than in pulmonary ones. This could be challenged, however, because this correlation hasn’t been found in studies from countries with low TB incidence, but it was certainly the case with the Chinese study.

In fact the authors of the study reviewed this factor carefully, adding that it is ‘well documented’ that ‘Beijing’ genotype strains of TB are strongly associated with drug-resistance. They suggested, therefore, that the higher frequency of MDR-TB which they found in the EPTB cases they surveyed could be reasonably explained by the current epidemic of the Beijing genotype in China. ‘Several studies’ they added, have identified a higher than average rate of Beijing genotype in EPTB cases when compared to pulmonary TB cases, so if this is generally the case, then it suggests a deadly double whammy: the cases who are already the most difficult to find and treat (MDR-TB) may have a significant sub-group within them who is even harder to find and treat (MDR-EPTB).

In this respect, it’s also worth bearing in mind that this phenomenon may not be confined to China because the Beijing strains have been associated with drug-resistance in the U.S., Russia and South Africa as well as in China itself.

So how reliable is the reporting?

One of the reasons that’s been suggested for any incorrect estimations of extrapulmonary TB is simply down to differences in terminology. The WHO defines EPTB as TB of any organs other than the lungs. These organs may be the lymph nodes, kidneys, genitals, breast, liver, skin, bones and joints, and pleura - in fact it’s been reckoned that TB can affect just about any organ or tissue other than the hair. Meanwhile, guidelines in the Russian Federation have apparently been defining pleural TB as pulmonary TB and not EPTB. The logic behind these Russian guidelines is that because the pleura cover the lungs they essentially belong to the respiratory system and thus can’t be considered as an extrapulmonary organ. If this sort of confusion is occurring at any scale at all, it suggests that it may be impossible to make any proper estimation of incidence.

A Russian paper has taken a closer look at this[vii] – reviewing 175 cases of EPTB in Novosibirsk in 2014. The authors found that 67 of them had pleural TB (in other words in nearly 40% of EPTB cases the disease was in the pleura). Were these cases notified as EPTB cases or not? The paper implies that they may not have been. Now check out that map again.

Russia is shaded very lightly indeed (less than 9.9%) – much lighter in comparison with nearly all of its neighbours except China (which we've already touched on).

Sites of EPTB

Of course this high proportion of pleural TB isn’t necessarily typical anyway (though it has also been found in Poland and Romania which are also lightly shaded). The Chinese paper, for instance, found pleural TB to be the second most frequent site (second to skeletal TB); in many other countries, the lymph nodes are found to be the most common sites of disease. In Public Health England’s Report, for example, pleural TB was found in only 9% of EPTB cases; TB of the lymph nodes, meanwhile, was reported in 40% of EPTB cases.

The Chinese paper went on to suggest two reasons for such heterogeneity.

One is that the site of infection may be influenced by prior vaccination with BCG because this vaccination provides differential protection to different forms of TB. Some countries have used the vaccine heavily for decades; others (the USA, for instance) have never used it – so logic dictates that this alone might significantly affect sites of infection.

The second reason is because some surveillance studies have shown that lymphatic TB is more frequently observed in children who are less than 15 years of age, so the highest proportions may also heavily depend on the proportion of childhood TB that is being found in any country. This is known to be highly variable, and largely dependent on contact tracing which is still desperately deficient in high burden countries.

So one reasonable conclusion is that it’s very difficult to draw any reasonable conclusions.

So what conclusions can we draw?

There’s certainly plenty of uncertainties, but it seems highly probable that a sizeable tranche of disease is remaining invisible and still undiagnosed - not just within the TB that is already recognised to be being missed (which suggests that this portion of the pandemic is underestimated), but also within the TB that is actually being notified (ditto). No-one can be certain of how much this might be, but it certainly looks very likely to be sizeable.

And what does this mean if it’s true? As already mentioned, it suggests that the current estimates for incidence of global TB are underestimated, but it also means that mortality rates are underestimated; and that rates of MDR-TB are also probably under-estimated as well.

It’s been a habit in these blogs to put some numbers together on these sorts of discrepancies – as wayward as we accept they may be.

In 2017, 6.4 million TB cases were notified globally. By applying the definitive 14% proportion to them we can conclude that something around 900,000 cases of EPTB were found that year. But let’s assume for a moment that 33% of all TB was EPTB instead of allowing ourselves to be fooled by that 14% proportion of notified cases. If the global proportion were indeed 33% (significantly less than the percentage found in England but would include a higher proportion of HIV coinfections, and would constitute a cohort a host of individuals with much higher exposure to infectious TB than those incident cases in the darker shaded countries), it suggests that 2.1 million EP cases should have emerged among those 6.4 million TB cases who were notified, (This would amount to a 1.2 million shortfall which should surely then be added to the total global estimated TB burden, making it 11.2 million).

But then we still won’t have included the proportion of EP among 3.6 million cases who were believed to have been missed in the same year. Have any of them been estimated to have been EPTB, and if they have at what proportion? The truth is that we don’t know, but we very much doubt that many of them will have been. (Our evidence for this is simply such a lack of reference to EPTB in the Global Reports).

If there’s been no allowance for them, it suggests that perhaps another 1.2 million (and maybe even more) TB cases could be added to the global estimated incident TB burden adding to the grand total.

So is it fair of us to suggest that the global estimate for TB may be falling 10-20% short of the mark? We think it could be – particularly given that these cases basically don’t figure in the End TB Strategy[viii], and moreover failed to get a single mention in the UN High Level Meeting’s Political Declaration.

If this is remotely true the mortality rates should be being hiked proportionately higher too, if only because EPTB is known to have higher mortality outcomes but also because there must be more TB out there overall, and more of it appears to be being missed. Might mortality rates, therefore, be as much as 30% higher than currently estimated?

Maybe they could be…

Finally, what sort of percentage of these EP cases might be being missed? Well, if 3.3 million should have been found and only 900,000 were notified we can suggest that 73% of EPTB cases may have been missed in 2017.

So let’s see what appears in the WHO’s imminent Global Report for 2019.


[ii] (see page 35)

[iii] “A much higher proportion of people with TB born outside the UK had extra-pulmonary disease only (48.5%, 1,591/3,282), compared with those born in the UK (27.8%, 360/1,294).”





[viii] There is one single reference in the main text relating to improving diagnostics: “For PLHIV with bacteriologically negative or extrapulmonary TB, diagnostic algorithms should be in place to ensure early identification and timely treatment of diagnosed or presumed TB.”

Featured Posts
Check back soon
Once posts are published, you’ll see them here.
Recent Posts
Search By Tags
No tags yet.
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square
bottom of page