Analysis in Rare Disease - Working with Contradictions

This post was originally published to coincide with Rare Disease Day, which takes place on 29th (or 28th) February

February 28th is Rare Disease Day ,which has got me thinking what it really means to work in rare disease analytics — and why it requires a fundamentally different mindset.

In rare diseases, data isn’t just smaller. It’s more precious, more fragile, and more human.

And it's built on what feels like a paradox. Rare disease analytics is both more precise and more uncertain than in any other disease area. 

Where it's more precise: When patient numbers are tiny, you can’t rely on statistical smoothing or large‑N trends. Every piece of data matters. There is no “noise” — only signals waiting to be understood. A single referral can shift a forecast. A single discontinuation can reshape a trend. A single diagnostic test can reveal a pattern. This forces a level of scrutiny and contextual understanding that simply isn’t required in high‑volume conditions.

But at the same time.... despite this precision, uncertainty is ever‑present. Small datasets mean wider confidence intervals and greater volatility. Most rare diseases have gone decades without robust research or established data sources, meaning traditional indicators like sales trends or large‑scale datasets simply don’t exist to guide decision‑making. It recognises that responsible analytics is not about pretending to know more — it’s about being transparent about what we can know, what we can’t, and what we should watch next.

Rare disease analytics asks us to hold two truths at once:

be more precise than ever, and more comfortable with uncertainty than ever.

As Rare Disease Day approaches, it’s a reminder that our work isn’t just about performance — it’s about purpose, partnership, and protecting the trust placed in us by small but profoundly impacted communities.