Short, plain-spoken pieces on what longitudinal data shows — and what it doesn't, yet.

Picture two children. Both are screened on the same day, at the same clinic, by the same health worker. Both have a MUAC reading of 128mm. By every standard a single screening can apply, they look identical — the same status, the same number on the same form.

One of them has held that reading for months. The other has lost several millimetres since the last time anyone measured them, and is several weeks into a decline that simply hasn't reached the threshold yet.

A single number can't tell you which child you're looking at. Only the line connecting that number to the ones before it can.

This isn't a criticism of screening — screening does exactly what it's designed to do, which is check a child's status against a threshold, right now. The problem is that "right now" is the only thing a single measurement can ever tell you. It has no memory of what came before, and no way to flag what might come next.

A trajectory is different. It doesn't need to be dramatic to be informative — a slow, steady dip across three or four routine visits is often a clearer signal than any single alarming reading, simply because a trend rules out coincidence in a way one data point never can.

This is the entire premise longitudinal monitoring is built on. Not better measurement instruments, not more frequent visits for their own sake — just linking the readings that already happen into something that can be read as a line, not a dot.

It's one of the more uncomfortable patterns in child nutrition, and one of the least studied: two children, raised in the same household, sharing the same caregiver, the same food, the same roof — and one of them repeatedly becomes malnourished while the other doesn't.

It's uncomfortable because it complicates a tidy explanation. If food availability alone explained malnutrition, siblings in the same home should track together. They often don't.

We don't have a confident answer for why. We think that's worth saying plainly, rather than implying we do.

Part of the reason this question stays open is structural: most nutrition surveys are built to compare across households, not within them. A survey that captures "this household is food insecure" has no natural way to also capture "and within it, this particular child is faltering while their sibling isn't." The unit of analysis is usually the household or the population — rarely the sibling pair.

Longitudinal, child-level monitoring changes that, almost by accident. If you're already tracking every child under five in a household over time, sibling divergence stops being invisible — it shows up automatically, as two trajectories that should move together and don't.

Whether the explanation turns out to be caregiving allocation, illness exposure, individual resilience, or something else entirely, we don't yet know. It's one of the open questions we think this kind of data is best positioned to actually answer — not because we have a theory we're trying to confirm, but because right now, almost no one is even collecting the data that would let the question be asked properly.