Low weight baby - bf to schedule

[TeacupTempest]

DD has dropped below bottom centile. She is 7 weeks. We had been bf on demand.

After being checked out at hospital they could find nothing wrong with her or my milk supply.

We have been sent home with instructions to stop feeding on demand and to start feeding every 3 hours for a max of 20 mins...

They say on one hand she isn't getting enough calories as they say she falls asleep on the breast ( she does as she also comfort suckers but this is at the end of a feed or an additional visit to the breast) on the other hand they say she is expending too much energy sucking for so long.

Seems very contradictory as an explanation.

The advice seems counter to how I want to parent. I feel sick having to take my tiny baby away from the boob and for so long.

I have been to a local group and they said my latch and her feeding were good.

Lost.

Thanks for that Mama very, very interesting.

yes, that's really interesting mama. I am surprised that the sample was so small actually. Can you explain the bit about the tails a bit more, I'm not certain that I got all the finer points, not being a statistician myself...

pease, I'll have a go.

To get the percentiles, you want an estimate of where to draw the line so that 1% (or whatever) of the population will lie under it, and 99% above it. But you only have a sample of 454 babies. The sample reflects the population, but it is not exactly the same. So you make a histogram of your sample, then try and draw a curve of predetermined shape through it. IF your curve is the right shape, then the curve will be a better estimate of the population than your histogram. If you've guessed nearly right, then the curve will probably be fine in the middle (around 50%) because you've got so many data points they will fix the curve in the right place. But in the tails - under 10%, over 90% - you haven't got many sample points, so you're relying a lot on the shape of your curve being right. It probably isn't, because we know most curves are inadequate fits to data in the tails, and the further into the tail you go (towards 0.4% for example), the less real data you have and the more you rely on the curve.

They did lots of work to try and pick the best shaped curve, but the tails will be estimated with lots of uncertainty even so.

To put it another way. If you picked another 454 babies and ran the whole exercise again, the middle centiles would probably be about the same. But the extremes would almost certainly be different, and the more extreme, the more different they would be likely to be.

I'm a geeky statistician, not a teacher. Is that any clearer than mud?

that's perfect, thank you!!

The sample is not really all that small, but it's widely accepted that removing 'outliers' - those babies who are at the far ends of the spectrum and beyond it - may distort the distribution. It is necessary to exclude them, even so, but it's true that they may include perfectly healthy individuals who are growing physiologically but just at their own, er, individual rate :)

I think I am right in saying that the researchers deliberately excluded outliers (not sure if it says so in the methodology). The babies may also have excluded themselves not just because small babies may well have switched to formula because they scared the bejabers out of their HCPs, but also because large babies sometimes have the same effect. Mothers are told all sorts of daft things like their milk is too rich so they must switch, or they restrict their bf and then their supply drops and they switch to ff as a result of that.

The charts we now have are based on good data, though, certainly the best data we have for the normal growth of healthy infants, but their use as a single diagnostic tool is necessarily limited. They have a valuable role to play but they need careful interpretation when it comes to individual babies.

Actually low birthweight babies (

I meant the real outliers - the ones who would show above or below the charts as they are written now. I was at a presentation where we were shown a slide where some outliers were included - represented by dots on a graph. There were some real oddities - babies of a few months old weighing 10 kg, or 2 kg, sort of thing. Sometimes it was thought this was just badly recorded data - human error hitting the wrong key on the keyboard, sort of thing. Some of them would prob be healthy and very unusual individuals.

Thanks for the interpretation MamaChocoholic, that's brilliant.

So in other words, it is quote possible, though not very very common, for a baby to lie below the 0.4 centile line, but be perfectly healthy.

If all checks show all is well and baby is alert and happy and gaining weight (albeit rather slowly), then probably the parents of that baby should relax a little?

yes, exactly. :)

Oh I see tiktok

TT, in any group of healthy individuals, someone is always going to be the smallest. And if the 0.4%ile is calculated accurately, then 4 in every 1000 healthy babies will be below it, by definition. Not very, very common, but not incredibly unusual either :)

tiktok, no, the sample is not that small for estimating the middle of the distribution. But I was surprised by how small it was, given that HV etc, in common with most people I imagine, will not understand the greater uncertainty involved in the estimates of the extreme %iles. certainly the health professionals involved in this story appear to be focused on the number rather than the baby.

They didn't remove low weight babies, and there appears to be very little drop out after 1 week which is encouraging. Despite this, it is good statistical practice to remove extreme and obvious outliers, as these are almost always data errors, particularly in longitudinal data, where you can check an individual baby's weight over time.