Determining the societal value of a prospective drug for ME/CFS in Germany, 2024, Afschin Gandjour

Determining the societal value of a prospective drug for ME/CFS in Germany
Afschin Gandjour

BACKGROUND
Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) imposes a substantial societal and economic burden. The aim of this study is to ascertain the optimal level of public research and development (R&D) investment in Germany for a prospective drug, given the pressing need for effective treatments.

METHODS
This study calculates the societal value from a German perspective by integrating health and economic outcomes in the context of public R&D investment for ME/CFS. It considers factors such as direct medical costs, productivity loss, and the effectiveness of a prospective drug.

RESULTS
The anticipated introduction of a prospective drug is estimated to yield a quality-adjusted life year (QALY) gain of approximately 29,000 and a societal value of about €2.6 billion. The optimal R&D investment in Germany is estimated at €676 million, which represents about a quarter of the total investment required to bring a significant drug to market, considering diminishing returns and market constraints. Results were confirmed in the sensitivity analysis.

CONCLUSIONS
The study concludes that a coordinated international approach is imperative to address the funding and market size limitations effectively in developing treatments for ME/CFS and to realize the substantial societal and economic benefits.

Link | PDF (PLOS ONE) [Open Access]

 

They argue that the optimal R&D investment in Germany was €676 million, which represents only a quarter of the total investment needed to bring a drug to market. They therefore say that international collaboration is necessary.

A closer look at their assumptions:

- The data on quality of life was taken on the Hvidberg et al. 2015 study that recruited patients from the Danish ME/CFS Association. The minimally important difference (MID) was posited as 15% of the range the EQ-5D3L questoinnaire.

- They assumed the drug would not work for everybody, only for a subgroup. To estimate these they used data on non-orphan drugs that underwent the German early benefit assessment of new drugs. Table 1 says that the average disease population is 1,322,051 and the Average target population 375,331, so that would make a target of 28%?

- They estimated only 17% uptake of the drug at year 4 post-market launch because or biases of physicians against ME/CFS (I suspect this will change once an effective drug for ME/CFS is found). I think patients will make sure that they get access to the drug.

- The direct medical costs of ME/CFS were considered basically to be 0 except for rehabilitation where they used estimates from Long Covid: €3210 per case. That brings the total estimate for medical costs at only €13 million.

- For indirect costs and productivity loses were much higher at €142 million assuming a national employment rate of 76.9% and work disability rate for ME/CFS or 60%, and an average national employee income per year of €46,700.

- The anticipated quality of life gains however were even one magnitude higher: 29,000 QALY gains over the relevant lifetime, translating to a monetary value of approximately €2.5 billion.​

 

This is odd. I guess they must only mean the cost of treatments/drugs. Which still isn't zero, but is probably fairly low with the lack of treatments. The cost of rehabilitation appears too low as well, although anything that has zero effectiveness is too expensive.

However the costs of the dysfunctional diagnostic process is definitely not zero. It contrasts harshly with the constant whining about how expensive we are.

And that doesn't even take into account the increased costs of subsequently ignored/delayed health problems, but that's impossible to calculate accurately.

 

Can anyone help me understand the impact of such studies? Does anyone know whether this approach is so well established (even if it only applies to one country) that it is usually included in the discussion about the distribution of funds?

 

It is something that does need to be discussed. Because leaving it open as a theoretical idea it does get abused

Bit worried with some of these itemised listings when reading it with a cynical eye of how bits might be cherry-picked.

Because something has gone very wrong in the next stage part of things in the past: the bit of 'we must do / offer something' and that ending up focusing on cherry-picking of the data to claim 'cost-savings' eg by forcing people to stay full-time instead of going part-time for a few years (and having a chance of being less ill in the long-term).

Part of the problem I suspect eg with the old NIce ones was that it probably was designed based on the drug results, where there were at least licensing regulations for trials and yellow card systems. Then the behavioural stuff was just slotted in as if it was the same thing, when it refuses to do any long-term or harm-based checks.

I mean really no treatment that can only claim at 6mnths 20% better (which turned out to be wrong too) for a long-term condition that isn't end of life would have been given any credence - if it wasn't for 'nothing else' being the alternative they'd wangled it to. Truth was that zero effectiveness divided by any number is actually asymptotic cost, but eg IAPT LTC presentations found a way to twist that.

Plus they removed any proper clinical care, so there could be no clinical-based management and monitoring or development. And that seems to be where the 'mind the gap' is. And boy haven't certain professions capitalised on their competitive advantage (weaker research regs) and that issue. Instead of their blaggy treatments being chucked out if they haven't used long-term controlled objective criteria, they seem to get away with finding somewhere they can sell a need to and I guess play the game on costs.

Then there is the twisting of the focus from a decent health improvement meaning that the lost productivity is prevented in future generations vs claims such as of 'reducing appointments [elsewhere 'maybe', whilst the CBT or physio rakes it in with 12 sessions on their books]'.

So in essence we need to watch out for opening the door further to opportunism selling 'cheap interim solutions' when the number and infrastructure involved (clinical stuff for ME/CFS would need to be put in place, whereas it seems like rehab has a lot of people supply-led just waiting for patients) seem to be a number to stomach.

And these professions seem to act like deluded teflon at the idea they've had it all their own way for the last few decades so that could act as 'ground zero cost' for what is lost productivity with their help, but just blag that 'maybe it just needed higher intensity or gimme another go cos others weren't tailored enough' BS.

 

It is an interesting one when you consider how poorly ME/CFS has actually been acknowledged in most countries nevermind reliably diagnosed and prognosis and impacts actually noted ie I'd imagine the accurate data wouldn't be there (because of attitudes to ME/CFS historically).

I'm slightly making notes as I have a read through and see where these are from and if they are caveated before I forget any points.

Does anyone have any idea what Germany's attitude was to ME/CFS, certainly prior to Long Covid?

I'm interested to note the following, given Germany's population is approx 84m vs UK 68m (and 250,000 has been cited for years as the number of ME/CFS)

I was intrigued that the author was from a school of finance and management and thought if it was just this then it was an interesting non-medical background approach.

But have looked him up further and he has all sorts of qualifications:

Prof. Dr. Dr. Afschin Gandjour | Frankfurt School (frankfurt-school.de)

I'm completely making assumptions here, but going by this CV makes me think that he might have some insight into the sorts of heuristics that are used in decisions like this in different countries like Germany and the USA.

I'm heartened reading the introduction to read the description of the illness (and the historical issues), particularly contextualised in relation to the methodological issues that need to therefore borne in mind when he notes 'several factors contribute to this lack of approved treatments'. Which include covering the subtle nature of changes and underestimation of severity, along with the fluctuation pattern of the condition etc. which I would have initially noted.

This particular point I wouldn't have picked up on myself, and found interesting:

particularly when combined with:

 

OK so this is also interesting:

I've flagged this because I don't know whether it is suggesting that all absentees for a certain time, whether near retirement or not, are no longer seen as contributing to productivity losses. I'd be interested to understand more detail on why this is used - although can imagine perhaps its a quirk of how 'losses' are calculated, or maybe it is conservatism to not assume those who've been ill for 20yrs can step back into where they would have been even if they are only 35yrs old.

 

Anyway so it looks like at first they are looking at equations where they are calculating value for 'SV health' - a health improvement value for each patient, and then one for the costs (what is saved in health costs). These are also combined with the above, which is the productivity losses.

The SV health seems to be the one that includes productivity gains/prevention of productivity losses. based on MID x 'max willingness to pay', and then multiplying that by number of people affected x each year affected (with some assumption of there being a discount over time). Which basically reflects the chronic nature of the condition.

When it is combining all 3 into 'overall society value' it is basically taking the 'max willingness to pay' - 'current health costs (that would no longer therefore be needed)' x each year, and adding it to 'rehab costs' - 'savings on productivity losses'(see above) x each year. And then multiplying this by 'MID'


MID is 'minimally important difference' - which I assume to be a conservative suggestion for what would 'count' as a 'good drug' % improvement-wise in the eyes of someone taking it. I guess this provides some sort of 'break even' theoretically as an average for each illness.

This seems to be being backed up in the equation by using 'maximum willingness to pay for a health outcome', I'd guess to differentiate between things that could be differentiated from placebo but hardly enough that it's really worth it vs it makes the difference between me having a life or not.

BUT that measure seems to be quite political, here is a link where it talks about this measure and the QALY: How Should Willingness-to-Pay Values of Quality-Adjusted Life-Years Be Updated and According to Whom? | Journal of Ethics | American Medical Association (ama-assn.org)

I'm now riffing here using my non-expertise to suggest this isn't a bad idea theoretically depending on how that is calibrated (if it is individuals then you've an issue regarding income probably being related to severity), as I assume that some of that could be due to ME/CFS having such varied symptoms we are always asking the question of how would you define an improvement (and what would it look like).

BUT, as usual, whilst it might be the 'best fit' of the measures that are well-acknowledged by governments etc, it could be somewhat complex for ME/CFS, particularly given I'm not even sure those from different severities of ME/CFS would automatically know or get the intricacies of another (nevermind laypersons, so how good would the info be, and how specific to severity would it be?).


I know for example when you are getting near to very severe then even a 5% improvement could be worth having given the threshold paradigm means that could make all the difference between being able to stay within it given your living situation/essential needs vs not and going downhill. 5% when mild is perhaps less 'make-or-break' to whether you can avoid a constant deterioration situation (but again its never that easy because currently those in the workplace etc might be at their limits too).

Which is the sad thing about not having clinician-led clinics (or good measures), where moments of truth could be identified as 'urgent points' and that 5% can be translated to the difference between having your teeth brushed ever or not. Which some of the surveys for example that I see fail to get near to accommodating or representing the importance of as a task, but would be an off-the-scale workload in itself vs someone's threshold.

 

Yes in particular the following sound like problematic 'best-fit proxies' which could perhaps have been bettered data-wise:

Why Danish, when the rest is Germany? what are 'Danish population preference weights'? and is 112 and a response rate of 35% anywhere near appropriate to ensure representation of the full spectrum of the condition - particularly given I'm assuming the questionnaire might be long enough it isn't easily accessible to the very severe?

This seems to have been used to inform the QALYs, but I'm not totally clear on how. Yet the way the other calculations ended up adding it accounts for 2.5Bn of the 2.7Bn

As the productivity gains are so low due to using a model that only looks at whether someone could return before their job was filled ie basically long-term sickness pay costs, not once someone has had to leave their job or reduce their hours etc. And the health costs savings are apparently low because not much was spent before.

So really those latter parts were always 'filler/window-dressing' for a long term illness, even if they were using 't' ie for each year in the equation, because those two savings (productivity losses being only the friction costs if someone was 'off sick' and returned before lost job, and health costs) could only be saved once.

Which makes that QALY - if that turns out to be the 'max willingness to pay' in the SV Health equation - the absolutely most important number in the whole sum.

It was always going to be multiplied by the max per yr the health system would pay, and they set the Minimally important difference arbitrarily at 15%

But then have based what they've multiplied it by based on a sample of 112? That they can't even be sure is representative of the different severities in the population?

I mean it is a start, and we sort of need this modelling.

But I'd probably have been more interested if they'd been able to model out the numbers for 'MID' being 15%, 20%, 25% and so on, all of which seem as eminently possible as 15% does from a potential drug?

And I'm not convinced how much this adds on top of the Nice type calcs which just look at QALY/cost.

ie this isn't really adding in what we think it is on productivity etc. ?

And it seems this is slightly side-stepping the issue of actually touching on real lost productivity given how long-term the illness is?

I mean you realise how hard this stuff is when you then have to think whether 15% vs 25% difference would mean difference in amount of care needed or if some could work or do different hours etc?

And that they've put in the assumption that the drug would keep its efficacy long-term, but we don't know whether that means noone would then deteriorate any further (which is quite different actually to only having 15% efficacy - imagine getting diagnosed mild then 'getting back' 15%, and it keeping you there whatever happens including overwork or a nasty virus?! - which then of course reminds me of the potential ceiling effect if their survey used for QALY data was predominantly less severe pwme)

 

So this is further explained, suggesting it really is focusing on quite short-term (for ME/CFS) 'friction costs':