Well the WHO’s Global Report has been out for over a week, and it turned out to contain pretty much the sort of numbers we were anticipating. Unfortunately we’re still struggling a little to make coherent sense out of all of them. While we wait for their expert response to some specific questions we’ve asked relating to what appear to be serious anomalies regarding some of the numbers relating specifically to South Africa, we’ve decided to share some of our other confusions.
To set the scene we’ve selected some specific individual sentences from the Report itself, and then compared them with some similarly selected sentences from the sobering statement made by MSF last week in response to the Report’s publication (‘Global TB report reveals rate of diagnosis for multi-drug resistant TB cases is heading in the wrong direction’).
We’ve chosen these sentences very deliberately, and make no apology for the fact that those of WHO which we’ve selected are not necessarily representative of the wider Report. We’ve chosen them because we suspect that they contain some discrepancies in statistical representation and in language choice which may merely be the tip of an epidemiological iceberg, the most dangerous part of which is unseen as we will further explain at the end of this particular first piece. As such we end this piece by suggesting that something rather important may be being missed – and naturally we welcome comments accordingly if others see otherwise.
In each instance below the first selected text is from the WHO’s Report, and it’s then followed immediately with a comparative text from MSF. Both of these are then followed by our own commentary which is intended to add context to each selection. Here is the first one.
WHO: “The advances are major: TB mortality has fallen 47% since 1990, with nearly all of that improvement taking place since 2000, when the MDGs were set.”
MSF: “Yet another year of disheartening statistics, such as TB’s persistent annual 1.5 million death toll, should serve as a wake-up call that enormous work still needs to be done to reduce the burden of this ancient, yet curable disease.”
Our own commentary: First of all we have to admit that we strongly suspect that the annual toll of 1.5 million premature TB deaths may be an underestimation. But be that as it may, we simply can’t accept the claim that “TB mortality has fallen by 47%” anyway. This is because it deliberately omits from the body count all of those who died from TB but who were co-infected with HIV/AIDS at the time they died. For some curious reason best known to the folk in Geneva these 400,000 dead people just don’t get counted in this particular equation which relates to one of the targets set by the Stop TB Partnership back in 2006 – though they do get counted in the WHO’s estimated 1.5 million annual death toll referred to by MSF. If they were added into the falling-mortality equation as we believe they always should have been then the number of deaths that could be said to be directly caused by TB last year was actually not 47% – but about 9% less, and we can even use a WHO graph to demonstrate this.
You can see from this graph that two curves are separately tracked – one of deaths among HIV-negative people which peaked around 2000, the other among HIV-positive ones which peaked around 2005. You can also see that back in 1990 there were estimated to be 1.5 million deaths in the former and about 0.15 million deaths in the latter (totalling 1.65 million). In 2014 there were estimated to be 1.1 million deaths in the former, and 0.4 million in the latter (totalling 1.5 million). And this, any which way you may choose to look at it, represents a reduction in deaths of 9%, and not 47%.
So to the next selected sentences:
WHO: “More TB patients were tested for drug-resistance in 2014 than ever before.”
MSF: “When it comes to the deadlier forms of the disease – such as multidrug-resistant TB – the news is particularly bleak. Despite progress in rolling out better diagnostics such as rapid molecular tests, fewer people were detected with MDR-TB in 2014 than in 2013, even though the estimated number of new cases remained steady.”
Our commentary: Here we have examples of carefully selected words intending to promote quite different conclusions – with both statements being apparently correct. Fewer people were detected with MDR last year than the year before (something which MSF sees as being “particularly bleak”), and at the same time more TB patients were tested than ever before (which the WHO reports as being encouraging). If we accept both statements as they stand, however, and then put them together we can even draw a further conclusion still – and it’s one that also seems hopefully optimistic. If more patients were tested and in the same period fewer MDR cases were detected it suggests that the total number of cases of MDR disease hanging out among the majority host of drug-susceptible TB must somehow be on the wane. And if this were the case then it should surely have been the headline of the whole Report, since the WHO’s Director General herself stated a couple of years ago that the world is sitting on a “powder keg” of drug-resistant TB. So collating these two statements together offers a hopeful sign that the powder keg’s sizzling taper might actually be showing signs of fizzling out.
There was no such big headline however – the best the Report was prepared to offer was that they the proportion of DR-TB within the pandemic last year was “essentially unchanged” (on page 56) and/or that it has “changed little in recent years” (on page 2).
The trouble with this more sober reckoning though is that, as much as we might hope it might be true it really doesn’t make sense. If it really was such a struggle to see the rate of the whole TB pandemic (the vast majority of which is susceptible to the standard DOTS regimen of drugs) stall and then begin to decline (now dropping at a rate of about 1.5% a year) with the majority of cases getting treatment and the vast majority of these being successfully treated, then how can the same thing possibly be happening with MDR cases when everyone agrees that the vast majority of cases never see treatment at all and that those who do only complete treatment successfully half of the time?
So let’s try another selection, this one also relating to MDR-TB.
WHO: “The ratio of patients enrolled in treatment to patients newly notified as having MDR-TB … was 90% globally …”
MSF: “We’re losing ground in the battle to control drug-resistant forms of TB, and without considerable corrective action, the vast majority of people with MDR-TB won’t ever be diagnosed, put on treatment, or cured. Today, a person with multidrug-resistant TB has worse than a one-in-four chance of being properly diagnosed.”
Our commentary: Once again we can see some selective phrasing being used by both parties. MSF’s language in this instance is a little more emotive. It’s also quite explicit. At best only one-in-four cases are being properly diagnosed. (Whilst this proportion is taken directly from the data in the WHO Report, we have to add that we also think that this number is pretty optimistic). Meanwhile the WHO confirms from their data that a lucky 90% (or nine-out-of-ten) of these one-in-four MDR cases who get diagnosed will also get approved treatment (and do please note that by inference this computes to only nine-out-of-every-forty of all estimated MDR cases getting approved treatment). But the WHO also offers data in their Report that indicate that only half of these complete treatment successfully. So we have to factor this in as well – and so actually it looks like only nine out of every 80 MDR cases were successfully treated last year – or just over 10%. Under such a scenario it does look like MSF are right: ground must be being lost even though 90% of newly notified patients may have been being enrolled on treatment programs, and “considerable corrective action” does need to be implemented.
So let’s try a fourth and final comparison.
WHO: “Globally in 2014, 123,000 patients with MDR-TB or rifampicin resistant tuberculosis (RR-TB) were notified … This was equivalent to 41% of the 300,000 estimated TB patients who were estimated to have MDR-TB in 2014.”
MSF: “Worse, the total number of people diagnosed with MDR-TB globally in 2014 was actually lower than the previous year (123,000 in 2014 vs 136,000 in 2013), although the total estimated number of people who developed MDR-TB remained the same.”
Our commentary: Here we have two statements which use the same number (123,000 patients either notified or diagnosed as being MDR by national TB programs) to infer vastly different things. Broadly speaking one infers that this 123,000 tells us that some things are still going backwards; the other that it gives grounds for optimism.
But in relation to the WHO’s statement above there was one important word that was missed out from the WHO’s summary quoted above which does influence our possible interpretation of it. This was the word ‘pulmonary’. The WHO suggests above that there were 300,000 patients who were estimated to have MDR-TB in 2014 – when this was not actually their best guess for all cases of MDR-TB at all – it was only their best guess for new cases of pulmonary MDR-TB. On an entirely different page their estimate of the total burden of all types of new cases of MDR TB was actually 480,000 – a full 60% more. In fact on another page of the Report they even compute this number against the 123,000: “Of the 480,000 estimated to have occurred in 2014, only about a quarter of these – 123,000 – were detected and reported”. So 41% of “estimated TB patients” getting notified or 25% of them? You can take your pick depending on which section you read, on which page you open the document, or on which conclusion you prefer to draw, more positive or more negative.
The missing measure
But we have one ore thing to add to this blog and we believe it to be very important indeed. The number that’s never getting estimated (and we think it’s a rather important one) is the total number of MDR cases who might just be ‘out there’ now – not just the estimated number of new cases or the estimated annual number of missed MDR cases (if it’s even possible to remotely accurately estimate this given the terrible dearth of diagnostic surveillance). In order to estimate the number of living-breathing MDR cases that have to be out there we need to add in the number of missed cases from the year before who have survived, and also the number from the year before that who are still surviving as well, along with the number of MDR cases who fail treatment each year but who survive and remain infectious until they either spontaneously cure (as some do) or die – or continue to survive permanently infectious until they die of other causes.
A survey in South Africa has suggested that XDR patients who fail all treatment are then discharged and survive for a median time of 19 months before dying; a similar one in Cambodia suggests that they may last for an average of three years. Such periods of time are far from epidemiologically insignificant. In three years a single infectious patient can be anticipated to infect a further forty patients with untreatable (albeit initially latent) disease. Of these forty latent cases, however, epidemiologists would only expect between two and four to re-activate and become infectious themselves in the course of time, but that’s certainly enough to both perpetuate and potentially increase a pandemic of functionally untreatable disease. (And these sorts of number would certainly be expected to be higher if there was much HIV in the community which is without question the case in South Africa).
In epidemiological parlance these unlucky folk are called ‘prevalent’ cases, and with a chronic and widely-disseminated disease like tuberculosis it’s expected that there will invariably be a higher number of ‘prevalent’ cases in any given period (let’s say during the 12 months of 2014) than there will be newly emerging ‘incident’ cases in the same year – if only because the incident cases have to be included amongst them. (With some significant misgivings, however, we note once again in this year's Report that several of the reported national numbers suggest the opposite for last year, including three of the 22 so-called ‘High Burden Countries’ where particular focus has been exerted on their epidemics for well over a decade, with a further four of the 22 improbably reporting prevalence:incidenceratios that were nearly equal at 1:1).
The WHO’s numbers for the whole pandemic support a more probable ratio, however: they estimate that there were 13 million prevalent cases in 2014, but that there were 9.6 million new incident ones. The year before it was 12 million and 9 million but the ratio is almost exactly the same – at 4:3.
But the WHO believes that there were 480,000 new incident cases of MDR in 2014, a number which we know can only represent a proportion of the prevalent MDR pandemic even if MDR-TB incidence really “has changed little in recent years”. So what about this slowly accumulating host of prevalent MDR cases (those who struggle on for two to three years before expiring – or even longer)? There’s no mention of them anywhere in the Report despite the fact that they possibly pose the biggest threat of all to the WHO’s best efforts to end the pandemic, not just because they are drug-resistant, but also because they are bound to also include untreatable infectious cases.
In the next blog we will try and explain why the general prevalence ratio of 4:3 for the TB pandemic has to be much higher in the MDR and XDR sub-pandemics, and we’ll try putting a few rough numbers together in the process of doing so.