A newborn’s length and an eight-year-old’s height both go on the growth chart. Do the same genetic differences account for variation at both ages? A Norwegian research team could ask the question twelve times across childhood. Its answer depends on the age of measurement and on which children contributed a measurement at that age.
- The study used 574,580 measurements from 72,704 Norwegian children, covering birth through age eight.
- The researchers estimated that common genetic variants accounted for 16% of the variation at birth, 31% at age one and 41% at age eight.
- These percentages describe differences among children in a population. A forecast for one child’s future height calls for a prediction model and a measure of its error.
- Paper date
- 22 November 2026
- Norwegian cohort
- 72,704 children
- Length and height records
- 574,580 across 12 ages
- Comparison data
- Age-ten recall and measured adult stature in UK Biobank
Original publication: 22 Nov 2026 · The date above refers to this brief.
Share of variation in stature associated with common genetic differences
percent of variation
Read a heritability estimate in three steps
Identify the group
Find the age, sample and type of measurement.
Identify the quantity
The model estimates a share of the variation between people.
Match it to your question
A forecast in centimetres requires predicted values and their errors in other children.
The same tape measure, a changing question
At birth, the researchers obtained length from medical records. Nurses also measured children at routine visits, and parents later supplied measurements through questionnaires. For each age the team estimated how much of the variation in stature across the group tracked with common genetic differences. The estimates at birth, one year and eight years were 0.16, 0.31 and 0.41, respectively. The paper gives standard errors of 0.01, 0.02 and 0.03. The estimate at eight years was 25 percentage points higher than the one at birth.
Source 1 ↗Why 41% does not give a height in centimetres
Imagine reading a chart with each eight-year-old represented by one dot. The estimate describes a share of the spread of those dots under the study’s model. Multiplying one child’s height by 0.41 would give a length with no interpretation in this analysis. Nor can the remaining variation all be assigned to a single cause such as nutrition. The measurements, population and genetic variation captured influence what the model can explain.
Source 1 ↗How many children contributed at each age?
Dividing the total records by the number of children gives about 7.90 measurements per child. That average can hide gaps. There were 72,704 observations at birth and 27,924 at age eight. The latter is 38.4% of the birth count. It tells us the relative sizes of those two analyses. The counts alone cannot explain why a particular measurement was unavailable. The original cohort profile describes how follow-up and registry sources work across a study that started in pregnancy.
Source 1 ↗Source 2 ↗What the comparison with adults adds
The age-eight estimate was close to the adult UK Biobank estimate of 0.38, although the people and settings differed. Similar estimates can still arise from partly different genetic patterns. In fact, the authors report a genetic correlation with adult stature of 0.52 for birth length and 0.73 for height at eight. The age-ten UK Biobank measure was a retrospective judgment about height relative to peers. It cannot fill a missing centimetre measurement for a Norwegian child. The authors highlight European ancestry as a limit on extending their findings to other populations.
Source 1 ↗Follow an estimate from the growth chart to the population
Publisher’s accepted full manuscript. Checked Figure 1, the sample sizes and estimates in Results, the measurement methods and the authors’ limitations. This brief focuses on stature. It does not independently analyse the cohort or its genetics.
01What changed with the age of measurement?
- What was observed
- The estimated share of variation associated with common genetic differences rose across the three displayed childhood ages.
- Where the conclusion stops
- It concerns variation among people and has sampling uncertainty.
Source 1 · Results and Figure 1b
02Were the age-specific samples equally large?
- What was observed
- The birth analysis included 72,704 children and the age-eight analysis included 27,924.
- Where the conclusion stops
- Counts describe analytic coverage. They do not give a reason for each missing record.
Source 1 · Results, Figure 1a
Numbers you can inspect
| Measure | Value & unit | Origin & method |
|---|---|---|
| Norwegian children at birth | 72704 children | Reported Birth-length sample. Source 1 · Results, Figure 1a |
| All Norwegian stature records | 574580 measurements | Reported Across twelve ages. Source 1 · Results and study overview |
| Children in the age-eight analysis | 27924 children | Reported Age-eight height sample. Source 1 · Results, Figure 1a |
| Measurements per child across the series | 7.9 records per child | Calculated 574,580 divided by 72,704, rounded to two decimals. Source 1 · Results, sample description |
| Age-eight count relative to birth count | 38.4 percent | Calculated 27,924 divided by 72,704 times 100, rounded to one decimal. Source 1 · Results, Figure 1a |
| Birth estimate | 0.16 share of variance | Reported SNP-based heritability, standard error 0.01. Source 1 · Results, Figure 1b |
| One-year estimate | 0.31 share of variance | Reported SNP-based heritability, standard error 0.02. Source 1 · Results, Figure 1b |
| Eight-year estimate | 0.41 share of variance | Reported SNP-based heritability, standard error 0.03. Source 1 · Results, Figure 1b |
| Difference between the age-eight and birth estimates | 25 percentage points | Calculated Subtract 0.16 from 0.41 and multiply by 100. Descriptive difference of point estimates. Source 1 · Results, Figure 1b |
| Adult height estimate | 0.38 share of variance | Reported Adult UK Biobank comparison, standard error 0.02. Source 1 · Results, Figure 1b |
| Genetic correlation with adulthood at birth | 0.52 genetic correlation | Reported Birth length and adult height. Source 1 · Results, Figure 1c |
| Genetic correlation with adulthood at age eight | 0.73 genetic correlation | Reported Height at eight years and adult height. Source 1 · Results, Figure 1c |
Compare the actual experiments
These studies answer different questions. Read the unit and endpoint before comparing results.
| Study | Unit & setting | Readout | Interpretation boundary |
|---|---|---|---|
| Original MoBa cohort profile Source 2 · Cohort profile update on recruitment and follow-up | Pregnancy-based recruitment with follow-up in Norway | Questionnaires and registry data over time | Cohort membership and availability of height at each age answer different questions. |
| Norwegian birth cohort and UK Biobank Source 1 · Results, Figure 1 and Methods | Childhood measurements, retrospective relative height at ten, measured adult height | Age-specific estimates and genetic correlations | The comparison spans different ages, cohorts and methods of measurement. |
The export includes claims, available numbers, methods and source locations. It contains our reading notes and published summaries; it is not raw participant data or an independent reanalysis.
Evidence update · 22 Nov 2026
First publication. Any calculations in this brief use reported figures. The reading notes identify possible follow-up questions. AI source check; no human editorial or clinical review.
Three ages, three estimates of variation
| Age in the Norwegian cohort | Common-variant estimate | Standard error |
|---|---|---|
| Birth | 0.16 | 0.01 |
| One year | 0.31 | 0.02 |
| Eight years | 0.41 | 0.03 |
Figure 1 and Results in the accepted manuscript. Standard errors describe uncertainty in the population estimates.
Your questions, answered
Does 41% mean that 41% of one child’s height is genetic?
The denominator is variation between children of that age in the population studied. The estimate cannot divide one child’s body length into genetic and environmental parts.
Did every child have twelve measurements?
Twelve ages were analysed. Dividing 574,580 records by 72,704 children gives 7.90 records per child on average.
Did a UK participant recall an exact height at ten?
The comparison used recalled childhood height relative to peers. The adult height comparison used measured stature.
Could an eight-year-old’s height determine future diabetes risk?
The paper also examines metabolic outcomes using genetic analyses and extra assumptions. An individual risk estimate cannot be read from these height-variation figures.
Limits of this interpretation
- The number of observations changes with age and the reasons for individual missing measurements are not available from the totals.
- Common-variant estimates depend on the population and method. The authors describe their analytic sample as being of European ancestry.
- The UK age-ten and adult data come from a different cohort. Childhood comparative height there was recalled.
Sources & transparency
- Fragoso-Bargas et al. (2026) — The genetic and proteomic determinants of pediatric stature development and their link to adult height and type 2 diabetes
Publisher’s accepted full manuscript. Checked Figure 1, the sample sizes and estimates in Results, the measurement methods and the authors’ limitations. This brief focuses on stature. It does not independently analyse the cohort or its genetics. · Accessed 22 Nov 2026
DOI: 10.1038/s41467-026-79465-x - Magnus et al. (2016) — Cohort Profile Update: The Norwegian Mother and Child Cohort Study (MoBa)
Cohort investigators’ profile of the pregnancy-based study. Checked recruitment and follow-up through questionnaires and registry data. · Accessed 22 Nov 2026
DOI: 10.1093/ije/dyw029
Prepared and source-checked with AI for Press-news Team. The source notes identify the material available during preparation. The article received no human editorial or clinical review. The source list supports the reported results and the calculations shown.
Source check: AI source check — primary research and stated source limits
Clinical review: Not applicable to this educational guide
Suggest a correction