Study population and data collection

The children in the study sample were selected from two pregnancy and birth cohort studies in Greenland. The first was the Greenland Inuit Child Cohort (Ivaaq) of 400 children founded by the National Institute of Public Health, Center for Health Research in Greenland, Denmark, in 1999 11. The second was the Climate Changes, Environmental Contaminants and Reproductive Health Cohort (CLEAR) of 600 children founded by the European Commission's 7th Framework Programme in 2002 12. All 383 children who were included in these two cohort studies and came from the capital of Nuuk or the town of Ilulissat were identified at the age of 6–10 years in 2012. Of these, 21 could not be identified with certainty and were excluded and 51 children participated in both cohorts, leaving 311 unique children eligible for the analysis. The anthropometric measurements, demographic characteristics and medical histories of these children were retrieved from the medical records of the Queen Ingrid's Hospital and the primary healthcare clinic in Nuuk, from the Regional Hospital in Ilulissat and from the records of visiting nurses in both Nuuk and Ilulissat. Records from the child psychiatry healthcare system were not included. The following data were collected: town of residence, gender, gestational age, length measured lying down or height measured standing upright, weight and head circumference at birth and at each following contact visit, number of hospitalisations and outpatient consultations in the secondary healthcare sector and visits to primary healthcare clinics between birth and 15 October 2012. The type of feeding was not registered in the records, as data were not systematically recorded. For each hospitalisation and outpatient consultation, a diagnosis according to the WHO International Classification of Diseases 10th edition (ICD‐10) was retrospectively recorded. Primary healthcare visits were categorised as illness‐related, recommended preventive health check‐ups or immunisation visits. We selected a random sample of every sixth child listed in the study sample and retrospectively coded each illness‐related contact with primary healthcare using the ICD‐10 classification. Based on all hospitalisations, outpatient consultations and primary healthcare visits, eight children with diseases known to affect growth such as chronic anaemia, cardiac or lung diseases and endocrine disorders were excluded, leaving 303 children in the study sample. None of the children had a diagnosis of precocious puberty, defined as breast Tanner stage 2+ in girls before eight years, or genital stage 2+ or testicles volume of 4 mL+ in boys before nine years. Finally, in accordance with the WHO criteria 4, 24 children born before a gestational age of 37, or after a gestational age of 42 weeks, were excluded, leaving a final sample of 279 children. Nationwide population data on Greenlandic children were obtained from governmental statistics 13.

To avoid overrepresentation of measurements from children with frequent healthcare contacts, we only included one measurement of length, weight and head circumference per month during the first year and only one measurement for every six months from the age of one year onwards in the analyses. Due to the retrospective study design, it was not possible to assess inter‐observer or intra‐observer measurement variations or to document the equipment used to measure length, height and weight. A common plastic centimetre tape measure was commonly used for head circumference measurements.