Lead
You come home from the 18-month checkup — a universal pediatric visit in Japan — and the pediatrician's words are still in your head: "Your child's weight is tracking toward the lower end of the curve." Now you're staring at the growth chart trying to work it out. "Is the 50th percentile the average? My child is at the 10th percentile — does that mean she's below average? Is something wrong?"
That question contains several overlapping concepts. Mean, median, percentile, and the proper way to read a growth curve are related but distinct. Learning to distinguish them lets you strip the unnecessary anxiety out of a checkup number and extract only the information that actually matters.
Three "midpoints" — and they are not the same
The mean is the sum of all measurements divided by the number of children measured. It is intuitive, but it is pulled by outliers. If a few exceptionally heavy children are included in a sample, the mean shifts upward — which means many children in the group end up below the mean, even when nothing is wrong with them.
The median is the value that sits in the middle when you rank everyone from smallest to largest. Half of all children fall above it, half below. It resists outliers. For biometric measures like height and weight, the median and the 50th percentile are nearly identical.
The percentile is a measure of relative position within a group. The 10th percentile means the child's measurement is larger than 10 percent of the reference population — or equivalently, smaller than the other 90 percent. The lines labeled "3rd percentile" and "97th percentile" on a growth chart mark the outer edges of the band that contains 94 percent of the reference population.
The current international standard is the WHO Multicentre Growth Reference Study, published in 2006, which followed approximately 8,500 infants in six countries — Brazil, Ghana, India, Norway, Oman, and the United States. These standards were constructed from children raised under optimal nutrition and environmental conditions and are explicitly designed as normative: describing an ideal or prescriptive standard of healthy development, rather than a descriptive average of what actually occurs benchmarks: how children should grow, not merely how they do grow on average [1].
The LMS method: how growth curves are built
Knowing a little about how growth curves are constructed changes how you read them.
The LMS method: a statistical technique using three smoothed parameters — L (skewness), M (median), S (coefficient of variation) — to construct normalized growth percentile curves across age, developed by Timothy J. Cole and published in 1990, is the modern standard for constructing growth curves for height and weight [2]. It summarizes the distribution of measurements at each age using three parameters — L (skewness), M (median), and S (coefficient of variation) — and smooths those parameters across the age axis to produce continuous, smooth percentile curves.
What this tells you: a growth curve is not an "ideal path." It is a cross-sectional snapshot of a reference population's distribution. The curve labeled "10th percentile" is derived from data collected across many children at each age. It does not mean that a child who measures at the 10th percentile today will track along that line at every subsequent checkup.
Cross-sectional reference versus longitudinal trajectory
This is one of the most important distinctions for interpreting checkup results.
Growth curves are built from cross-sectional: data collected from different individuals at a single point in time, producing a population snapshot rather than individual trajectories data — measurements from different children at each age, stitched together. A child's position on the curve at any one checkup tells you where she sits relative to the reference population at that moment.
Longitudinal observation — tracking your own child's measurements over time — carries different and often richer information. Whether a child has tracked stably at the 10th percentile for a year, or whether she has dropped sharply from the 30th percentile to the 10th over three months, are two very different stories. The latter is clinically more informative than the absolute percentile position.
The 2000 CDC growth charts, developed by Kuczmarski and colleagues, were designed with this in mind: crossing two or more major percentile lines in a short period is considered a more meaningful clinical signal than the absolute position at any single visit [3]. The 2022 revision of the CDC developmental milestone guidelines by Zubler and colleagues similarly emphasized change patterns over single-point snapshots [4].
"My child is at the low end" — and the cognitive frame that distorts it
When parents receive the information that their child is at the 10th percentile, the most common cognitive frame is: "My child is below average." But the 10th percentile is not abnormal. It is a position within the normal range — toward the lighter or shorter end, but within it.
The situations that warrant concern are different: falling below the 3rd percentile or above the 97th percentile, or showing a sharp crossing of major percentile lines across visits. A child who comfortably occupies the 10th percentile and does so consistently is not giving any signal that requires a clinical response.
When a pediatrician says "let's watch this," the intention is almost always to verify at the next visit whether the position is stable — not to imply that the current number is a problem in itself. Interpreting a single checkup data point in isolation, without the trajectory context, is what generates false alarms.
How ongoing records enable longitudinal understanding
If you have recorded your child's height and weight month by month, you already have what you need for a longitudinal view. "Stable at the 10th percentile for 12 months" is a different and more reassuring story than "moved from the 3rd to the 10th percentile over the last six months."
One of the practical values of a parenting-record app like Memori is that it places a single checkup number inside your child's own time series. A figure that is handed to you in a 10-minute appointment looks very different once you can see it against six months of prior data — whether it represents a sudden change or simply a consistent position.
Summary
- Mean, median, and percentile are different things. The figure referenced at a checkup is a percentile position, and "below 50" does not mean deficient or abnormal [1, 2].
- Growth curves are built from cross-sectional data. The pattern of change across visits is often more clinically meaningful than position at any single point [3, 4].
- The cognitive frame of "my child is the only one who's low" misrepresents what a percentile actually measures.
- Ongoing records create the foundation for longitudinal observation — the kind of reading that turns a single data point into a comprehensible story.
Numbers tell you where things stand right now. What they mean only becomes clear when you read them alongside the trajectory that led there.
References
- WHO Multicentre Growth Reference Study Group; de Onis M. WHO Child Growth Standards based on length/height, weight and age. Acta Paediatr Suppl. 2006;450:76–85. doi:10.1111/j.1651-2227.2006.tb02378.x. PMID: 16817681.
- Cole TJ. The LMS method for constructing normalized growth standards. Eur J Clin Nutr. 1990;44(1):45–60. PMID: 2354692.
- Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC growth charts for the United States: methods and development. Vital Health Stat 11. 2002;(246):1–190. PMID: 12043359.
- Zubler JM, Wiggins LD, Macias MM, et al. Evidence-Informed Milestones for Developmental Surveillance Tools. Pediatrics. 2022;149(3):e2021052138. doi:10.1542/peds.2021-052138. PMID: 35132439.
- Cole TJ. The development of growth references and growth charts. Ann Hum Biol. 2012;39(5):382–394. doi:10.3109/03014460.2012.694475. PMID: 22780429.