Childhood eczema: disease of the advantaged?BMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6937.1132 (Published 30 April 1994) Cite this as: BMJ 1994;308:1132
- Hywel C Williams, Wellcome research fellow in dermatoepidemiologya,
- David P Strachan, senior lecturer in epidemiologya,
- Roderick J Hay, professor of cutaneous medicinea
- a St John's Institute of Dermatology, United Medical Schools of Guy's and St Thomas's Hospitals, Guy's Hospital, London SE1 9RT
- Department of Public Health Sciences, St George's Hospital Medical School, London SW17 0RE
- Correspondence to: Dr H C Williams, Department of Dermatology, Queen's Medical Centre, University Hospital, Nottingham NG7 2UH.
- Accepted 16 March 1994
Objective: To determine whether the increased prevalence of childhood eczema inadvantaged socioeconomic groups is due to increased parental reporting.
Design: Comparison of parental reports of eczema with visible eczema recorded by medical officers during a detailed physical examination.
Setting: National birth cohort study.
Subjects: 8279 children from England, Wales, and Scotland born during 3-9 March 1958 and followed up at the ages of 7, 11, and 16.
Main outcome measures: Prevalence of eczema according to parental report compared with medical officer's examination at the ages of 7, 11, and 16.
Results: Prevalence of both reported and examined eczema increased with rising social class at the ages of 7, 11, and 16 years. The point prevalence of examined eczema at age 7 was 4.8%, 3.6%, 3.6%, 2.4%, 2.2%, and 2.4% in social classes I, II, III non-manual, III manual, IV, and V respectively (X2 value for linear trend 12.6, P<0.001). This trend persisted after adjustment for potential confounders such as region and family size and was not present for examined psoriasis or acne.
Conclusions: Eczema is more prevalent among British schoolchildren in social classes I and II than those in lower classes. Exposures associated with social class are probably at least as important as genetic factors in the expression of childhood eczema.
Eczema seems to be more common in children from higher social classes than in those from lower classes, but it is not clear whether this is due to parental overreporting
In this study eczema based on examinations by school medical officers was twice as common in social classes I and II as in IV and V
The proportion of cases of parental reported eczema that were validated by examination was higher in higher social classes
Environmental factors seem to be as important as genetic factors in determining childhood eczema
Social class should be included as a possible confounder in all studies of the prevalence of eczema
One of the most striking findings of the analysis of risk factors for childhood eczemain the 1970 British national cohort study was an increased prevalence of reported eczema in advantaged socioeconomic groups.1 2 Similar trends in prevalence of reported eczema were found withother measures of socio-economic advantage such as parental education, less overcrowding,and more household possessions.
One important question arising from these results was whether children from higher social classes really had a higher risk of eczema or whether affluent parents were just morelikely to report it. Parents from social classes I and II may use the term eczema more freely for children with minor skin conditions,3 4 and recall of eczema in early life might vary by social class. Affluent parents might be more likely to take their children to a doctor and acquire the label of eczema for borderline conditions such as dry skin more readily than those in less advantaged socioeconomic groups.5
We sought to determine whether the social class trend for reported eczema was also present for eczema found by examination by analysing data from the national child development study. Unlike the other two large national cohort studies in Britain6 7 the child development study substantiated the presence of eczema by medical examination.8
Subjects and methods
The national child development study developed from the national perinatal mortality survey,8 which gathered detailed information on over98% of babies born in England, Wales, and Scotland during the week of 3-9 March 1958. These babies have been followed up at the ages of 7, 11, 16, 23, and 33 years. Of the 9518 children for whom information was gathered at the ages of 7, 11, and 16, only those with complete responses on the presence or absence of visible eczema at these ages were included in this study (n = 8279, 87%).
When the children were aged 7 parents were asked by health visitors using a structuredquestionnaire whether their child had had eczematous rashes during the first year of lifeor at any time after the first year. When the children were aged 11 or 16, parents were sked whether they had had eczematous rashes in the past 12 months. We have used the term examined eczema to refer to visible eczematous dermatoses recorded by experienced school medical officers during a complete physical examination at the ages of 7, 11, and 16. Examination of the skin was part of a more general examination including physical and mental development. No hypothesis on skin disease was being tested by the study. We derived social class using the registrar general's classification according to the father's occupation when the childen were aged 7.9
We used the EpiInfo statistical software to analyse results.10 Adjusted effects were explored by using unconditional logistic regressiotechniques with the EGRET software package,11 exploring each potential confounder individually and in combination. Categorical data were analysed with contingency tables, hypothesis testing being based on the χ2 test for linear trend for social classes I through to V. Results are expressed in terms of prevalence with 95% confidence intervals.
The prevalence of reported eczema was 1.5-2 times higher in social classes I and II than in classes IV and V at each follow up point (table I). The social class trend was strongest when the children were aged 7 and under.
Prevalences for examined eczema showed similar social class gradients to those for reported eczema at all ages in the follow up study (table II), with a roughly twofold difference in prevalence between the two highest and two lowest social classes. Estimates of prevalence of examined eczema for the small number of single parent families were generally similar to those for social class V.
The point prevalence estimates for examined eczema were roughly half the annual period prevalences of reported eczema at the same age. The proportion of reported cases of eczema validated by a doctor's examination was higher in social classes I and II. When the presence of reported eczema at 11 and 16, or both, was validated against examined eczema at these ages, the proportion of validated cases was 59% (17/29), 51% (42/82), 53% (21/40), 36% (76/213), 42% (31/73), and 38% (10/26) in social classes I, II, III non-manual, III manual, IV, and V respectively (χ2 for linear trend 4.57). This significant (P = 0.03) trend is in the direction opposite to that proposed in the prior hypothesis-- that is, that overreporting of eczema occurs in higher social classes. The proportion of children with examined eczema who did not have reported eczema showed no trend with social class.
Table III shows the adjusted risks for examined eczema at the ages of 7, 11, or 16 according to social class. Several potential confounders were explored in the regression analysis, including region of residence, ethnic group, maternal or paternal smoking when the children were aged 16, sex of child, breast feeding, and family size. Only region of residence and family size were significantly associated with both social class and examined eczema. The size of the odds ratio for examined eczema within each social class remained virtually unchanged despite adjusting for these factors. A history of hay fever or asthma was not considered as a potential confounder. Similar social class trends for examined eczema were seen within the unlinked data at each follow up point (data not shown).
We also examined household tenure, another possible indicator of socioeconomic status, with respect to eczema. The prevalence of reported eczema at any age was 12.3% (438/3555), 10.1% (90/892), 10.6% (308/2893), and 6.6% (13/197) in privately owned, privately rented, council rented, and rent free properties respectively (χ2 value for privately owned versus council rented property 4.2, P = 0.04). The prevalence of examined eczema at any age was 6.1% (222/3622), 5.7% (52/907), 4.5% (146/3254), and 2.0% (4/201) in privately owned, privately rented, council rented, and rent free properties respectively (χ2 value for owned versus council rented property 8.8, P = 0.003).
To explore whether medical officers had a general tendency to overestimate visible skin disease in social classes I and II we also examined two other common inflammatory skin diseases recorded in the study -psoriasis and acne--with respect to social class (table IV). Visible psoriasis was recorded only at the ages of 11 and 16 and acne only at the age of 16. Prevalences for these two diseases showed no evidence of a strong social class gradient.
Prevalences of reported hay fever were also higher in social clases I and II, and the difference in prevalence between social classes increased from the age of 7 to 16 (table V). Prevalence of reported asthma or wheezy bronchitis, or both, in the past year did not show such trends.
Other studies that have considered the possible relation between social class and eczema have given conflicting results.*RF 12-16* Our results suggest that the increased prevalence of reported eczema in advantaged socioeconomic groups is genuine. It is consistent at each stage of the follow up, regardless of how eczema is defined, and is also seen with housing tenure, another measure of socioeconomic advantage. The trend in social class for examined eczema was virtually unchanged after adjustment for potential confounders, and the proportion of reported cases validated by medical officer's examination increased rather than decreased in higher socioeconomic groups. Similar trends were not observed for psoriasis and acne.
Hay fever, but not asthma or wheezy bronchitis, showed the same social class gradient as eczema. Eczema and hay fever have also been shown to have similar associations with household size17 and region of residence,18 whereas asthma or wheezy bronchitis showed a different epidemiological pattern in this cohort.19 A genuine social class gradient for allergic diseases such as eczema is supported by the findings of a similar socioeconomic gradient for more objective measures of allergic disease such as positive results in skin prick tests.*RF 20-23*
One of the strengths of this study is its low potential for observer bias because medical officers were unaware of the hypotheses, which were suggested many years later, and information on social class was collected independently by an interviewer. A possible limitation is the reliance on medical officers' diagnoses of eczema. At the time the study was carried out eczema was used synonymously with atopic eczema for children in Britain. The prevalence of examined eczema is similar to that in other studies conducted at that time in developed countries,24 25 and the difference in prevalence between reported and examined eczema is entirely compatible with the twofold difference between the point and one year period prevalences for eczema found in other studies.26 27 Inclusion of cases of scabies and nickel dermatitis due to earrings would be likely to reduce rather than exaggerate social class trends shown in this study.28 29
Reporting bias could still have occurred if articulate upper middle class children were more likely to tell medical officers that they had eczema or if medical officers examined well turned out children more thoroughly. This seems unlikely as children were undressed at the time, and the skin examination was only a small part of a more detailed physical assessment. The lack of a social class trend for acne or psoriasis or for cases of examined eczema not validated by parental report lends weight to this argument. Social class differences of eczema may reflect factors that determine chronicity rather than incidence of disease. If this was the case examined eczema at the age of 7 should show a much stronger social class trend than reported eczema from birth to 7. This was not the case.
EXPLANATION OF THE TREND
There are many possible reasons for the social class trend for examined eczema shown in this study. Correlates of educational status, such as positive health related behaviour--for example, increased uptake of immunisation or differing rates of exposure to topical corticosteroids-- could be risk factors for eczema.2 Other factors such as differences in the use of carpets and heating (which could influence house dust mite populations),30 overuse of showers or soaps, decreased exposure to ultraviolet light, increased close contact with pets, and prenatal exposures correlated with higher social class such as higher maternal age and maternal diet also need to be considered.2 31
Our findings relate to British children in the 1960s and 1970s, and social class differences may have altered since. In developing countries a social class gradient in the opposite direction may be seen for eczema because of increased secondary infections, increased exposure to primary irritants through child labour, and reduced access and provision of medical care in less privileged groups.32
Atopic eczema causes much suffering and has a lifetime prevalence of around 12% at the age of 5 in the United Kingdom.1 The twofold variation in prevalence of eczema between the highest and lowest social classes represents a considerable burden of potentially preventable disease. Since environmental determinants of eczema are largely unknown, exploration of correlates with social class is a useful starting point. At the least, social class should be considered as a potential confounder for comparative studies of eczema prevalence.
We thank the Economic and Social Research Council Data Archive and the National Child Development Study User Support Group for making the data available for study and all those involved in the original study for recording information on skin disease. We also thank Barbara Butland for help with data processing. HCW is supported by the Wellcome Trust.