Sex differences in weight in infancy and the British 1990 national growth standardsBMJ 1996; 313 doi: http://dx.doi.org/10.1136/bmj.313.7056.513 (Published 31 August 1996) Cite this as: BMJ 1996;313:513
- Charlotte M Wright, first assistanta,
- Sally S Corbett, research psychologistb,
- Robert F Drewett, senior lecturerb
- a Department of Child Health, University of Newcastle upon Tyne, Gateshead NE8 3EB,
- b Department of Psychology, University of Durham, Durham DH1 3LE
- Correspondence to: Dr Wright.
- Accepted 21 June 1996
Objectives: To determine if there is a sex difference in infancy in the new British national standards for weight (based on data from 1990).
Design: Weight data in a birth cohort were compared with the 1990 standards and Tanner and Whitehouse (1966) standards up to age 12 months.
Setting: Newcastle upon Tyne.
Subjects: 3418 term infants.
Results: Our cohort showed a mean difference in standard deviation scores of 0.42 between boys and girls (P<0.0001) when compared with the 1990 standards. Two and a half times as many girls as boys had weights below the 3rd centile during the first year, with an equivalent excess of boys above the 97th centile (P<0.0001). Similar results were found with Tanner and Whitehouse standards.
Conclusions: These differences could result in substantial sex bias in the identification of poor growth in early childhood. The standards need modification.
Growth standards for the United Kingdom, based on data from 1990,1represent a major advance in the process of weight monitoring in Britain, as older standards show major discrepancies with contemporary patterns of weight gain.2 However, an unexpected problem with the weight standards in infancy has emerged. The problem was uncovered by the observation that there were nearly three times as many girls as boys in a proposed study population of children with early failure to thrive; in population based studies, failure to thrive has previously shown an even sex divide.3 This led us to explore the possibility that the new growth standards were inadequately standardised for sex.
We hold a longitudinal dataset of weights collected routinely between birth and 24 months from a birth cohort of 3418 term infants born between 1987 and 1988.4 The weights nearest to 6 weeks and 3, 9, and 12 months were identified and transformed into standard deviation scores in comparison with the 1990 standards1 as well as with the previous national standard,5 by using the Castlemead Growth Package,6 and means for boys and for girls were compared. The crude weight data were also grouped in monthly age bands up to 15 months. The median values for this monthly grouping were compared directly with the 1990 curves for median weight.
The mean standard deviation score for weight of boys was 0.42 higher than that for girls after the age of 3 months (table 1), and twice as many girls as boys fell below the 3rd weight centile during the first year. There was an equivalent excess of boys above the 97th centile (table 2)
The boys' weight curves rose slightly above the British median while those for girls fell substantially below the median (fig 1). Comparing the means for boys and girls in Newcastle with the Tanner and Whitehouse standards showed mean differences in standard deviation scores of 0.34-0.37 after the age of 3 months.
The differences observed in this study are not trivial, being equivalent to over half an intercentile space and resulting in major imbalances in the number of boys and girls who fall outside the normal range for weight. Although the new standards present problems, these were present to a similar order in the standard they have replaced.
Although it is puzzling to find that children in Newcastle show such striking discrepancies with the two growth standards, it should be remembered that these standards are based, in infancy, on small numbers of children (160 children for the Tanner and Whitehouse standard; 252 for the 1990 standard). The Newcastle dataset comprises 3418 children, with 60-90% of the birth cohort represented in each 3 month age band. This observation thus cannot be easily explained by sampling error or bias. A smaller cohort of children studied in Glasgow showed similar differences.7 These findings imply that there is either a substantial and previously unrecognised regional variation or a more general problem with the standards.
It is a matter of urgency to discover whether similar sex differences are observable in southern Britain. If they are, the current weight curves are not representative and require modification. In the meantime, researchers using weight centiles or standard deviation scores as selection criteria should investigate the possibility of sex differences in their study populations so that they avoid inadvertently recruiting a study sample with a major sex imbalance.
Funding This work was supported by the Wellcome Trust.
Conflict of interest None.