Tag Archives: cea

treatment tailored to genes will kill economic evaluation

At least, kill economic evaluation as currently practised in much of Europe and Canada/NZ/Australia.

I am going to start by sharing an anecdote. A senior collaborator revealed that a submission to NICE, the body that recommends whether medical interventions are cost-effective and should be reimbursed by the British NHS, contained an error. The modelling (decision analytic markov chain monte carlo models) has reached such complexity that there are fewer than 10 people in the country qualified to check any such model. The error was spotted by the author before NICE met to make its final decision – and in any case the error (having a plus where a minus should be), thankfully would not have changed the recommendation. But it could have. And it highlighted the extreme vulnerability here – crucial funding decisions that save lives might be wrong, because there isn’t oversight to check and re-check these models. This is truly scary.

Now these models are being expanded to attempt to quantify whether it is “worth” getting additional information to continue considering a drug or whether we should cut our losses and consider something else.

However, there is a BIG assumption here. Namely, that the average cost and average effectiveness data are correct. (Also that the distributions surrounding these parameters are approximately correct). Unfortunately these assumptions are about to collapse quite spectacularly due to the rapidly falling cost of complete genetic sequencing.

Soon, there will be no such thing as “chemotherapy X for breast cancer” upon which to quantify effectiveness and population cost. Because the population of breast cancer patients no longer get it. Depending on their gene mutations, there may be 10, 20 or 50 different treatment regimes, tailored to whether you have one or both of the two BRCA breast cancer mutations. Effectiveness will skyrocket for people with those genes. Indeed the average effectiveness rate will skyrocket since patients without the genes won’t be given the treatment. So the demoninator of the cost-effectiveness ratio will change radically. The numerator too, potentially, since a bunch of people who are now known not to be helped by it won’t be given it, and people with the gene will be prioritised.

As personalised genetic medicine becomes more common, the number of economic evaluations required will increase exponentially. Uncertainty around estimates will cease to be a nice scatter of points and skew towards zero and one. And whether you are 0 or 1 will be determined by your genes.

How will health economists deal with that? Rates for economic evaluation were recently unilaterally cut by the Australian Federal govt to health economists doing such economic evaluation. Expect this to become more and more common.

Perhaps it is time to simply have a national debate as to which conditions do not deserve more funding as they are currently incurable. Or a re-evaluation according to my post on MMT as to what conditions need more real resources.

Because the incremental approach is going to collapse, it is only a matter of time.