World TB Day 2016 - and 'bending those curves'..
It's World TB Day again, a full 23 years since TB was declared a Global Emergency, and FINALLY we are seeing some heat being turned up on this treatable disease (with the unmistakable handwriting of the inspirational Dr Lucia Ditiu of the Stop TB Partnership evident in almost every initiative). There's good reason to rejoice, but there’s still little reason to believe that the tide in this war has even really turned, let alone ebb away more quickly.
A major new initiative is the Stop TB partnership’s new five year Plan for 2020, complete with its catchy slogan (the ‘90-(90)-90’ target). What's being demanded now is what's being called a ‘paradigm shift’ in thinking about TB. We hate to say this, though, but there's still way too little evidence of new paradigmatic thinking about TB.
The intention now is to “bend the curve” of what is a painfully slow decline in disease, bending it more steeply downwards so that the disease can be finally brought to its knees within a generation rather than festering on for another hundred or more years. It’s been hailed as an ‘ambitious’ strategy and it surely is. It can also succeed if it’s really well managed, but unfortunately we see a fundamental deficiency at its very heart and we fear that this flaw is more reflective of old-style TB thinking rather than of anything resembling the necessary paradigm shift. Highlighting this deficiency today might not be so welcome to the authors of the Plan who want to see everything done to add to the new momentum, but we believe that failing to do so is more likely to help see the Plan derailed than to see it stay on track to its anticipated end.
So here's the fundamental weak spot:
All of the post-2015 targets published so far are talking about TB as if it’s a single disease when it’s not.
What TB is today is an infectious disease which comprises three sub-diseases that are so clinically distinct from each other that we should be seeing them individually – each one requiring different targets, different drugs, different diagnostics, different resources and different strategies. It’s responding to this reality, we think, that's the real paradigm shift thinking that’s now needed.
Here’s what we mean...
If we see the disease as a single whole we can usefully represent it as an ‘all- TB’ pandemic with its associated total prevalence, incidence and mortality rates. Actually this is exactly how the state of disease is most regularly reported (with its most recent numbers being 13 million, 9.5 million and 1.5 million respectively). But what's so easily forgotten is that this ‘all-TB’ pandemic comprises not just the strains which are susceptible to the standard drugs, but also the ones which aren’t.
Meanwhile, at the core of this new End TB Strategy is a single curve, the trajectory of which is currently being reported to be sloping downwards at around 1.6% each year. (Even this far into the Emergency it’s believed that we’re still missing more than a third of all-TB cases each year and largely because of this the pandemic is not dropping as fast as it could be.)
What's now suggested is that this single curving trajectory (of the whole pandemic) can now be bent further downwards by finding those missing millions and then treating them more effectively. We have no issue at all with this idea, but we’re worried that this won’t end the pandemic as is anticipated and we’ll try to explain why.
Here’s the recently published WHO’s graph of how this curve is intended to be bent downwards.
Graph #1: the End TB Plan's graph showing how TB will 'end' by 2035.
Look first at the higher dotted line. This shows the current global trend of the whole all-TB pandemic with a downward trajectory of a little less than 2% a year (which is what’s been being consistently reported for the last ten years). A solid line is planned to split off from it in a year or so and this shows what’s hoped will happen as the new End TB Strategy begins to take effect. The decline should begin to accelerate to around 10% a year, and then accelerate further in 2025 to average an annual decline of 17%.
Viewing the disease in this way very reasonably leads us to believe that the disease can be seen off as a whole by 2035 - largely with some more efficient use of existing interventions in the first ten years, and then with a new vaccine in the second.
But this model completely ignores something important: it ignores the individual dynamics of those three subordinate parts of the ‘all-TB’ pandemic which comprise the whole. If we choose to see the disease without ignoring this then we don’t actually have a single curve to bend downwards at all: we end up with four of them which makes things a lot more challenging.
So here’s a similar simple graph to illustrate this. It’s directly reflective of the first one, but incorporates four separate lines, each one reflective of a component estimable part of the pandemic.
Graph #2: disaggregating the pandemic into its drug-susceptible and drug-resistant parts
Exactly as in the first graph the top line is reflective of 'all-TB' (the dotted line of the first graph). It’s the existing decline of the ‘all-TB’ pandemic dropping at just under 2% a year.
The second line reflects the largest proportion of this wider pandemic. This is the disease which is sensitive or susceptible to the standard DOTS drugs regimen (in other words that’s curable by the standard one-size-fits all approach) – we can call this ‘drug-susceptible’, or DS-TB for short. This is currently estimated to comprise 95% of the existing pandemic (though we’re sure it must be really be a bit less than this), and it’s curable in at least nine out of ten cases.
We can be confident that this second DS-TB curve is already be dropping downwards a little faster than the top ‘all-TB’ line – that one which represents the whole pandemic. This is because we know that the first curve of ‘all-TB’ in the graph above must inevitably contain the estimated 5% of the pandemic that's now MDR (a very significant majority of which is neither detected nor treated).
We need to acknowledge something at this point. We are NOT epidemiologists, nor are we expert mathematical modellers. Such expertise relies on complex algebraic calculations which were not attempting to emulate here. Because we're far from sure that these experts are presenting the whole diverse picture of this huge global health threat, we're simply applying what we consider to be simple logic and a little basic maths. And please be assured - we will be only too relieved if the results of our projections can be successfully challenged because of what they suggest.
Because you will see that we also have a third (red) line on our graph. This one represents the ‘other’ 5% - the proportion of the disease that’s resistant to at least the two strongest DOTS drugs (and which therefore isn’t curable by standard treatment nor is 'normally' diagnosable). At least 75% of these MDR cases are never diagnosed (and in we believe that it’s probably substantially more), and of those that are diagnosed and put on treatment less than half of them currently enjoy successful outcomes. It’s a very different story indeed o that of the line immediately above it.
Given the low detection rates and the poor treatment outcomes (using the WHO’s own published data) we’re sure that this line must be rising. The WHO, for reasons that we can't figure out, reckon that it's not. We're simply no longer prepared to accept this as a logical possibility, and actually we suggest that this proportion of the pandemic could logically be rising at around 10% a year (something which we discuss further shortly). So this is how we’ve pitched its trajectory in the graph above.
And finally we have a fourth line - the proportion of the disease that's resistant to even more of the anti-TB drugs, including at least two that we might first want to use in any alternative drug regimen to DOTS. This is XDR-TB - what Mario Raviglione, the WHO’s TB czar has called “the worst thing we could have imagined”. It’s currently being officially estimated that this part of the pandemic comprises just under 10% of the MDR component of the pandemic – which effectively means that it’s currently believed comprise about 0.5% of ‘all TB’. This may not sound like much at all, but it’s important to bear in mind that this type of TB is not just vastly more challenging to treat, it’s also vastly more dangerous, with treatment reported to only succeed in one-in-every five patients. Not only that, but the diagnostic capacity to actually diagnose and notify XDR-TB in most TB-endemic countries (let alone treat it) is in terrifyingly short supply, so logically it’s a certainty that this part of the pandemic is also rising, and it’s probable that it’s rising faster than MDR-TB. In our graph above we’ve pitched it as incrementally rising at a modest 0.5% more than the 10% rise we’re reckoning for MDR-TB. Because it starts from a much smaller number this makes it look a lot less threatening but, given everything we know about XDR-TB, we know that we should remain as alert to the trajectory of this curve as to any of the others.
We should add that everything about XDR-TB numbers are currently terribly uncertain and we’ve included this line really only because it’s so important to show that it exists.
What we wanted to particularly focus on on this World TB Day is the way that the blue and the red lines converge around 2034, each one with an incident rate at this point of time of about 30/100,000. According to our graph we can suggest that if there are no other successful interventions there’s a likelihood that after this there will be more drug-resistant disease within the pandemic than there is disease that is drug-sensitive.
You may well think that our tentative extrapolations are extreme. They’re certainly unapologetically tentative, but we’re not sure that they’re that extreme. David Dowdy, who is an associate professor at John Hopkins Bloomberg School of Public Health and an expert in MDR-TB epidemiology, is on record as saying that “scenarios in which MDR-TB rates double in 15-20 years are a reasonable expectation”. In the light of his expert opinion our extrapolation does look a little over-egged because, by using a 10% cumulative rise, our rates pretty much quadruple (not double) in a twenty year span. In the previous blog in this series, however, we put on record that six years ago the WHO itself was reckoning that MDR-TB was rising at around 9% a year. Not just that, but between 2000 and 2008 this sort of rise was being estimated to be pretty much the norm. Furthermore published data from South Africa has showed MDR-TB increasing more than four-fold in a period of just six years, with XDR-TB increasing a very scarey eighteen-fold in the same period. There is appalling uncertainty here. In fact, even if we stuck with David Dowdy’s scenario, these two lines would still converge before 2040, so unless anyone is able to confidently give us real numbers (which no-one can) we’re sticking with our extrapolations simply because they comprise an exercise in precautionary principles which we are convinced the precarious situation warrants.
Taking this further, we believe that the best way to prevent this convergence happening (in fact probably the only way to prevent it if the current successful trends in treating DS-TB are maintained and accelerated ) is to begin aggressively attacking this MDR portion of the pandemic immediately wherever we can to keep this third red line from rising. As such, prioritising and funding distinct MDR-TB programs should be an absolute fundamental part in any post-2015 strategy. What worries us is that we don’t see this in the current plan, not least because there are no distinct targets for TB yet set and the existing budgets won’t really allow for them anyway.
There are plenty of things that could change this of course. The most probable is that parts of the the End TB Strategy are successfully implemented, including “optimising current tools”, and “pursuinguniversal health coverage and social protection” as we saw above – but if this isn’t done without successfully and aggressively attacking MDR-TBin a way which hasn’t yet happened nor is yet planned, then actually we could even see the convergence happening even more quickly.
The next very simple graph attempts to illustrate this and unfortunately it shows both good news and bad..
Graph #3: Extrapolation of MDR-TB pandemic if the curve of 'all-TB is successfully bent downwards.
In this graph we’ve left out the XDR-TB line for simple reasons of clarity, and we’ve bent the top blue line of ‘all-TB’ downwards to reflect the anticipated curve of the WHO’s graph (the solid line in the very first graph in this blog). This puts it in line with the Plan, with the number of new cases dropping from 9 million each year to around 2 million by 2030. Similarly to what we saw in our own second graph, the line of DS-TB now curves downwards even faster too, in fact it even looks like DS-TB could peter out into epidemiological insignificance sometime around 2030 (which by happy coincidence happens to be the SDG target date for ‘ending’ TB).
But we've not changed our MDR-TB line. It's the same as in the graph above, rising at 10% each year and so the convergence point at which drug-sensitive and drug-resistant TB become essentially equal in incidence will have already occurred around three years before DS-TB might peter out - in 2027. After this (because the DS-TB component will now be becoming so insignificant) the line of ‘all-TB’ can be expected to essentially follow that of MDR-TB and so will begin to rise again for the first time since around 2005.
In other words by around 2030 ‘all TB’ will have essentially become a single MDR pandemic – at an incidence rate of around 25/100,000 (or roughly four times what MDR-TB is today, but about a quarter of what ‘all-TB’ is).
Now this needn’t happen, and it won’t do if we anticipate its possibility and aggressively attack MDR disease at the same time as we attack the drug-susceptible pandemic. But we’re quite sure that something like this will happen if we don’t attack MDR-TB really vigorously starting today.
And this definitely can't happen unless we begin to confront the probable numerical realities of MDR-TB.