Let’s be honest - there’s comfort in orthodoxy.  Health technology assessment (HTA) is built on a foundation of tried-and-tested conventions: lifetime horizons, a 'reference case' untainted by productivity, and a good old-fashioned faith in 'intention to treat' (ITT) analysis corrected by placebo control. But occasionally, a study comes along that quietly taps HTA on the shoulder and says, “What if we did things a bit differently?” We’ve just published a piece that does exactly that. It doesn’t blow up the methodological playbook - but it does underline the fact that the rules we follow are sometimes more a matter of habit than of logic.

Here are four ways the analysis dares to diverge:

1. A 12-Month Time Horizon Isn’t Heresy

We’re all conditioned to think that a “proper” economic evaluation must stretch to full patient lifetimes - especially for chronic conditions/treatments.  But what if a lifetime perspective is not only unnecessary, but actively unhelpful?

In this case, we opted for a 12-month time horizon, and no, it wasn’t because we were trying to pull the wool over anyone's eyes, quite the opposite in fact.  It was because the relevant costs and effects really do materialise within a year. This is particularly true for interventions with a short-term mechanism of action like daridorexant, that has a half-life around 8-hours.  To get it's therapeutic effect you need to take it every day and that is clear from the clinical results.  The clinical studies are a three month phase III registration study and a 40 week extension study.  Hence we opted for a 12-month time horizon - sure we can model beyond that time period, but uncertainty explodes beyond the window of the observed data.  When the future is murky and the present is policy-relevant, the “short-run” can offer the most meaningful long-term insight.

2. Placebo Correction - Handle With Care

It’s a familiar refrain: “Subtract the placebo effect to find the real treatment impact.” But this assumes that placebo responses are neatly separable, predictable, and static - like they’ve been pre-measured in a Petri dish.

Our approach leans into the idea that placebo responses are part of the real-world effect. If patients improve because they believe they’re being treated, does that not matter? The temptation to surgically remove the placebo effect often leads to assumptions about uniformity and persistence that don’t hold in practice. Our approach argued that after three months, any remaining improvements were not due to increasing efficacy (via a mechanism for which there is no biological basis) but rather due to selective attrition.  Since placebo patients would receive no treatment in real-life we aregue that placebo correction beyond three months was not appropriate.  A controversial idea for a community indoctrinated into 'evidence based medicine' and ITT analyses.

3. Selective Attrition Isn’t Just Noise

Missing data is annoying. Selective attrition is more than that—it’s a structural bias hiding in plain sight. If certain types of patients are more likely to drop out, and if that dropout correlates with treatment or outcome, then your ITT estimates start to wobble.

We tackled selective attrition as a methodological challenge, not a nuisance. Accounting for attrition selectively, revealed a very different narrative about the intervention’s cost-effectiveness. It turns out, who sticks around in your study can tell you much about the outcomes those subjects experience.

4. Productivity Gains: the thorn in the side of the 'Reference Case'

Productivity gains in HTA are a bit like carbs in a trendy diet: everyone pretends they don’t matter, but most secretly think they do.

Yes, the reference case says productivity should be excluded (at least in the UK). But in many real-world scenarios - especially those affecting working-age populations - the ability to function at work is a major benefit. We included productivity impacts not as an act of rebellion, but as an act of realism. If the intervention helps people get back to work, that’s an economic benefit, not an nuisance and certainly should not be ignored when decision-making for populations.

The Bottom Line

Methodology is never neutral. Every assumption—about time, bias, comparators, or value - shapes the story we tell about a technology. This study doesn’t claim to overthrow the HTA establishment. But it does ask a few uncomfortable questions, and maybe that’s exactly what HTA needs.

So if you’re tired of evaluations that follow the rules but miss the point, come take a look. Sometimes, breaking the mould is the most methodologically sound thing you can do.