International Journal of Epidemiology, Volume 33, Number 1, pp. 152-153
IJE vol.33 no.1 © International Epidemiological Association 2004; all rights reserved.
Special Theme: Perinatal and early-life influences on disease |
Commentary: The relation of growth to socioeconomic deprivation
Department of Public Health Sciences, King's College London, 5th Floor Capital House, 42 Weston Street, London SE1 3QD, UK. E-mail: sut.chinn{at}kcl.ac.uk
In this issue of the International Journal of Epidemiology Wright and Parker present data from two cohorts born in Newcastle, England, 40 years apart.1 Birthweight and infant weight decreased with increasing deprivation as measured by the Townsend index in the 1987 cohort, but there was no relation of birthweight or infant weight to social class in the 1947 cohort. This contrast was attributed by the authors to changes in the relation of smoking to social class.
Their other finding, to which they give more prominence, was that of ‘no evidence of a changing influence of socioeconomic deprivation on growth in childhood, despite increases in mean height over a 40-year interval’. The authors described this as puzzling, as they considered a constant relation of height to deprivation to be incompatible with two candidates to explain the secular trend in height. If the secular trend was nutritional in origin they expected to find a reduction in the gradient, and if intergenerational in origin it should be present from birth. They concluded that the relation of height to deprivation largely originated in childhood, and was compatible with a nutritional explanation. As their argument hinges on no change in the relation of height to socioeconomic deprivation the evidence for this deserves close scrutiny.
In the 1947 cohort deprivation was defined by social class, according to the Registrar General's occupational classification. Social class 3 was not divided into manual and non-manual at that time, and the authors grouped classes 1 & 2, and 5 and unclassified, producing four categories. In the 1987 cohort the census-based Townsend index was used, divided into four groups with the same percentages of the population as the four social class groups for the 1947 cohort. The use of the two different measures of deprivation makes a number of assumptions. If each measures the same underlying dimension of socioeconomic deprivation, and the relation of height to absolute level of deprivation is linear or the relation is to relative deprivation, then there is not a problem. However, at an individual level social class and Townsend index cannot measure exactly the same dimension, father's social class being household-based and Townsend index derived at a small area statistics level, and they do not show a close relation when measured on the same individuals.2,3 If the relation of height to absolute deprivation is non-linear, the reduction in absolute deprivation over the 40 years would lead to a different observed gradient of height with deprivation when the equivalent centile groups are compared. If the relation of height is to relative deprivation, rather than absolute deprivation, then the method of analysis is appropriate, but the mechanisms suggested for the gradient are with absolute deprivation.
The Townsend index was used for the 1987 cohort as most parents were not contacted, and this also had the advantage that all children for whom a postcode was known could be ascribed a score. Social class data are rarely complete, and so there are problems in comparing between different studies. Wright and Parker made comparison of their difference between the affluent group, defined in the 1947 cohort as social classes 1 & 2, and the most deprived group, defined as social class 5 plus unclassified, with results from the National Study of Health and Growth (NSHG) inner city sample. The NSHG collected data on ‘representative’ samples of English and Scottish children from 1972 to 1994,4 and inner city English children in alternate years from 1983 to 1993.5 Social class differences were reported several times,6–8 using internally derived standard deviation scores (SDS). No specific comparison was made of the social class gradient between different surveys as it was shown, consistently, that social class differences could be explained by other factors, notably parental heights, and that family size was the socioeconomic variable most strongly and independently related to child's height.
We have calculated height SDS from the UK 1990 reference curves for children in the final inner city NSHG sample surveyed in 1993, and the final representative English sample in 1994. The difference in height SDS between social classes 1 & 2, and 5 with unclassified, for white children, was 0.62 SDS in the inner city sample, and 0.17, approximately 1 cm, in the representative sample. The NSHG children were aged 4–11 years, but there was no interaction of age group with social class in the analysis. In Wright and Parker's analyses the difference between the affluent and very deprived group in 9 year old children was 0.61 SDS in the 1947 cohort and 0.78 in the 1987 cohort. It is not clear that is appropriate to compare Newcastle with the inner city sample, which was selected to contain the most deprived wards in England.5,7 Although the 23-year span of the NSHG is less than that of the Newcastle cohorts, it does provide a firmer comparison over time, and not just for one selected area. In 1972 the difference between social classes 1 & 2 and 5 was 0.57 SDS for the white representative sample,6 which compared with the comparable figure in 1994 of 0.29 SDS suggests that there has been a decrease in the gradient. Hence it seems likely that Wright and Parker's conclusion of no change in the gradient of height with deprivation is a result of the use of two different measures of deprivation and the consequent method of analysis.
Apart from this difficulty, the meaning of social class has changed over time.9 Social class 5 has decreased, as determined from NSHG data, from around 7% of those classified in 1972 to about 3.5% in 1994. We wonder whether there is any definition of deprivation, that can be used on data 40 years apart, that has not changed in its interpretation.
 |
References |
|---|
 Top References |
|---|
1 Wright CM, Parker L. Forty years on: the effect of deprivation on growth in two Newcastle birth cohorts. Int J Epidemiol 2004;33:147–52.
2 Joyce R, Webb R, Peacock J. Social class and census-based deprivation scores: which is the best predictor of stillbirth rates? Paediatr Perinatal Epidemiol 1999;13:269–77.[CrossRef][Web of Science][Medline]
3 Shohaimi S, Luben R, Wareham N et al. Residential area deprivation predicts smoking habit independently of individual educational level and occupational social class. A cross sectional study in the Norfolk cohort of the European Investigation into Cancer (EPIC-Norfolk). J Epidemiol Community Health 2003;57:270–76.
4 Hughes JM, Li L, Chinn S, Rona RJ. Trends in growth in England and Scotland, 1972 to 1994. Arch Dis Child 1997;76:182–89.
5 Chinn S, Hughes JM, Rona RJ. Trends in growth and obesity in ethnic groups in Britain. Arch Dis Child 1998;78:513–17.
6 Rona RJ, Swan AV, Altman DG. Social factors and height of primary school children in England and Scotland. J Epidemiol Community Health 1978;32:147–54.
7 Rona RJ, Chinn S. National Study of Health and Growth: social and biological factors associated with height of children from ethnic groups living in England. Ann Hum Biol 1986;13:453–71.[CrossRef][Web of Science][Medline]
8 Gulliford MC, Chinn S, Rona RJ. Social environment and height: England and Scotland 1987 and 1988. Arch Dis Child 1991;66:235–40.
9 Marmot M. From Black to Acheson: two decades of concern with inequalities in health. A celebration of the 90th birthday of Professor Jerry Morris. Int J Epidemiol 2001;30:1165–71.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||