Child Growth Percentiles: A Rank, Not a Grade, and the Trend Is What Counts

A growth percentile is a rank. A girl on the 70th percentile for height is taller than 70 out of 100 girls her age and shorter than the other 30. That is the whole definition, and most of the worry parents bring to it comes from reading it as a score.

The 90th is not better than the 30th. Healthy children sit on every line. What a doctor looks at is whether the child stays near the same line over time. A child who has tracked the 40th since birth and is still there at five is growing as expected. A child who slides from the 70th to the 20th in a few months is the one who gets a closer look.

I read trend lines for a living, and this is the same discipline: one point tells you where something is, a series tells you where it is going. This post covers what gets measured, which chart it is read against (the Dutch one is different from the American one), how to read the curve, and the handful of patterns that are worth a call.

One practical note first: the curves move fast in the early years, so the age has to be exact. The Age Calculator gives it in years, months and days; a month out and you are reading the wrong part of the chart.

Five measurements, and the chart they are read against

A well-child visit records some or all of these:

• Weight for age. The one every visit has.
• Length or height for age. Lying down until two, standing after, which is why a child can appear to lose a centimetre at the second birthday.
• Weight for length, under two. Whether the weight fits the length. 80th for both is a proportionally big baby; 80th for weight against 30th for length is a different conversation.
• BMI for age, from two upward. The same idea with age built in. The BMI Calculator does the division, but a child's BMI is read against the curve for their age and sex, never against the adult cut-offs; a number that is normal at 35 can be high at 5.
• Head circumference, under three. Tracks brain growth, and most of that happens in the first year.

Which chart the numbers land on depends on the country. In the United States the WHO charts are used under two and the CDC charts from two to twenty. In the Netherlands the consultatiebureau (the child health clinic that sees every child from birth to four) and the school doctor after that use the Dutch TNO charts, built from measurements of Dutch children, who are among the tallest in the world. The same child sits on a lower line on the Dutch chart than on the American one. Those charts also mark standard deviation lines rather than percentiles: minus 2 SD is roughly the 2nd percentile, plus 2 SD roughly the 98th. Different labels, same reading.

How to read the curve, and why the 50th is not a target

Age runs along the bottom, the measurement up the side, and the curved lines are the percentiles: 3rd, 10th, 25th, 50th, 75th, 90th, 97th on most charts. Find the age, go up to the measurement, and see which two curves the point falls between. That is the percentile.

Three things to hold on to.

The 3rd to the 97th is the normal band. A point outside it deserves a look. It is not, by itself, a problem.

The 50th is the middle of the distribution, not the goal. Half of all healthy children are above it and half below, by definition. Parents' heights set where a child is likely to sit; feeding a child toward the 50th is not a plan.

The line through several points beats any single point. A flat track along the 15th is fine. A flat track along the 85th is fine. A point that has moved two curves since the last visit is what starts the conversation, and a point that moved one curve is usually the normal wobble between visits, especially in the first two years.

If you keep your own log, keep it in centimetres and kilograms with the exact date, and let the clinic do the plotting. Home measurements of a wriggling toddler are not accurate enough to read a percentile off, and a log of them mostly generates worry.

The patterns that are worth a call

Children grow in bursts, and a shift of ten or twenty percentile points between two visits is ordinary. These are the patterns that are not:

• Crossing two major lines, in either direction. 75th to 25th over a few months is the standard trigger for a closer look.
• Falling below the 3rd. Some children are small and healthy, and the doctor will want to rule out nutrition, illness and hormones before settling on that.
• Weight climbing past the 97th while height does not follow. Rapid gain that outruns length is worth a conversation about feeding, and occasionally about metabolism.
• Height and weight moving apart. One up while the other is down points at something specific.
• A newborn who does not regain birth weight. Losing up to 7% in the first days is normal and most babies are back at birth weight by two weeks. The Percentage Calculator turns two weights into that percentage; beyond 10%, or no regain by day fourteen, the clinic wants to know that day.
• No weight gain over a couple of months in the first year.
• No signs of puberty by 13 in a girl or 14 in a boy.

All of these are reasons to ask, not diagnoses. Most children who cross a line are healthy and the conversation ends there. The point is that the question gets asked early, when it is cheap to answer.

What drives growth, in order

Parents' heights, first and by a distance. Genes set the range a child can reach; everything else decides where inside that range they land.

Food: enough of it, with enough protein, iron, zinc and vitamin D. A shortage of any of those slows height. A surplus of calories adds weight and nothing else.

Sleep: growth hormone is released mostly in deep sleep, and children who are chronically short of it grow more slowly. The usual targets are 12 to 16 hours for infants, 11 to 14 for toddlers, 9 to 12 at school age and 8 to 10 for teenagers.

Movement: ordinary running about builds bone and keeps weight in step with height. Very intense training in young athletes can slow growth for a while, which is a known effect in gymnastics and distance running and one to raise with a coach.

Illness: coeliac disease, inflammatory bowel disease, an underactive thyroid and growth hormone deficiency all show up on the chart before they show up anywhere else. That is why an unexplained slowdown gets a blood test.

Medication: long courses of corticosteroids, common for asthma, can slow growth. A child on daily medication should have growth on the checklist at every review.

Prematurity: a baby born at 32 weeks is plotted at corrected age (counted from the due date, not the birth date) for the first two years. Without that correction the chart shows a small child who is in fact growing exactly on schedule.

FAQ

How often should growth be measured?

The clinic schedule is enough: several visits in the first year, then every few months until three, then yearly. In the Netherlands the consultatiebureau schedule does this for you until four and the school health service takes over after. Measuring at home every week adds noise, not information.

From the 80th to the 50th, is that a problem?

It depends on how long it took and how old the child is. In the first two years children commonly move off the line set by their birth weight, which owes a lot to the pregnancy, and onto the line set by their genes. A shift over a year in a toddler is usually that. The same shift in three months in a six-year-old is a question for the doctor.

WHO, CDC or Dutch charts?

Whichever your clinic uses; do not mix them. The WHO charts describe how healthy, breastfed children grow across several countries. The CDC charts describe how American children did grow. The Dutch TNO charts describe Dutch children, who run taller. A child compared against the wrong one reads a line lower or higher than they are.

Can the percentile predict adult height?

Roughly, and only as a range. The mid-parental method: add both parents' heights in centimetres, divide by two, then add 6.5 for a boy or subtract 6.5 for a girl. The real outcome lands within about 8.5 cm of that on either side, which is a wide enough band that I would not plan anything around it.