Evidence on equalisation in health in youth from the West of Scotland

https://doi.org/10.1016/j.socscimed.2003.12.004Get rights and content

Abstract

Many studies report few socioeconomic (SES) differences in health in youth, a pattern contrasting with that of health inequalities in childhood and adulthood. This paper focuses on the child–youth transition to examine the hypothesis of equalisation in health over this period. Specifically, we test two hypotheses: (a) that equalisation is more likely for health state measures (physical and malaise symptoms and accidents) than health status ([limiting] longstanding illness and self-rated health) or health potential (height), and (b) that the patterning of health over this period is similar between occupational (social class) and non-occupational (deprivation, housing tenure and family affluence) SES measures. Data are derived from the West of Scotland 11 to 16 cohort, followed from late childhood (aged 11) through early (13) to mid (15) adolescence. The results showed very little evidence of SES differences in (limiting) longstanding illness at any age for both sexes, while self-rated health exhibited some differentiation, and height (as expected) consistent gradients throughout. By contrast, among males evidence of equalisation was found for both physical and malaise symptoms and pedestrian road traffic accidents (RTAs). Among females, equalisation was confined to specific physical symptoms, pedestrian RTAs, sports injuries and burns/scalds, while for malaise symptoms a reverse gradient at age 11 strengthened with age. These patterns were generally unaffected by the SES measure used. We conclude that while some of the evidence is consistent with the equalisation hypothesis, it needs extending to accommodate patterns of no SES differences, and particularly reverse gradients, in childhood. These patterns may reflect the increasingly pervasive influence of youth culture, suggesting that in the UK the boundary between childhood and youth should be set at an earlier age. This in turn suggests that international comparisons have considerable analytic potential for identifying the conditions under which equalisation does and does not occur.

Introduction

Although largely ignored in policy documents (e.g. Acheson, 1998), the place of adolescence or youth in the health inequalities debate continues to occupy the attention of researchers in the UK and elsewhere. Much of the earlier work aimed to describe patterns of health by socioeconomic status (SES) using cross-sectional datasets, the result of which led one of us to characterise the situation in youth (particularly the mid-teenage years) as one of ‘relative equality’ compared with childhood or adulthood (West, Macintyre, Annandale, & Hunt, 1990). While such descriptive work continues, with particularly interesting results in an international context, more recently attention has been directed towards the documentation of, and possible reasons for, change in SES patterning of health over particular life-course stages (West, 1997; Chen, Matthews, & Boyce, 2002) and the cumulative impact of such ‘critical periods’ for later health and health inequalities (Bartley, Blane, & Montgomery, 1997). One of these is the transition from childhood to youth which we previously conceptualised as about age 11/12, just prior to the onset of puberty and around the primary/secondary school transition in many school systems. On most (but not all) the evidence, the child–youth transition appears to involve a process of equalisation (from health inequality to relative equality) (West, 1997; Chen et al., 2002). The hypothesis of equalisation is the focus of this paper.

Any assessment of available evidence about the SES/health relationship in youth, as compared with childhood, has to consider the extent to which change or stability over this period might vary according to the measures of health and SES used.

While health can be defined in broad holistic terms (WHO, 1984), most health inequalities research has followed a biomedical model, typically measuring ill-health via mortality and various indicators of morbidity. In general, this also applies to studies of children and adolescents. However, a recurring theme in this literature refers to the appropriateness of various health indicators used. One of these refers to mortality which is generally regarded as a poor indicator since death is rare after infancy (Blane et al., 1994). For this reason, it is not considered here. Another view, embracing a more holistic concept of health, suggests the typical morbidity indicators are much too limited and fail to acknowledge the crucial developmental nature of childhood and adolescence. In this view, health is defined not only by reference to current disease or illness but also by reference to age appropriate physical, psychological and social achievements which may anticipate future health. A recent example of this approach is the Child Health and Illness Profile (CHIP) which, in addition to current ‘disorders’, includes satisfaction with health and discomfort together with risks (e.g. smoking), resilience (e.g. family involvement) and educational achievement in the definition of health (Starfield, Riley, Witt, & Robertson, 2002). Useful as this approach may be for certain purposes (e.g. planning child health services), it seems limited as an epidemiological tool since it conflates health with the causes of health, thereby rendering any relationship with SES as tautologous. This is not the approach adopted here.

Consideration of broader concepts of health does however highlight its multidimensional nature and implies that an adequate test of the equalisation hypothesis requires that different dimensions are represented. Following a distinction between health status (a relatively long-term property of individuals) and health state (a shorter term property), Blaxter (1989), Blaxter (1990) identified four dimensions: disease/impairment and ‘fitness’ (based on physiological measurements), reflecting different dimensions of health status; and psychosocial malaise and illness (symptoms), reflecting health state. The distinction between health status and health state is particularly useful because it suggests the relationship between health status and the social environment would be more fixed over the child–youth transition, while health state would be more likely to fluctuate with changing social circumstances. Thus, a pattern of inequalities in disease/impairment would be expected to persist into youth because it continues to incorporate (SES) influences from the earliest years. A similar expectation would pertain for some physiological measures such as height which, as an indicator of health potential (Macintyre, 1988), is patterned by SES from the earliest years. By contrast, the SES relationship with measures of health state such as psychosocial malaise or physical symptoms might be expected to change since they are subject to additional influences associated with youth which have the potential to cut across those associated with SES. Such influences might result from the mixing of young people from different social backgrounds in (secondary) school, and/or exposure to youthful influences associated with the school, peer group or youth culture (West, 1997; Siahpush & Singh, 2000). Similar influences might impact on the SES relationship with accidents/injuries.

To date, very few studies have included measures on all these dimensions to enable even a partial test of the hypothesis. Of the relevant research, by far the most usual measures are those relating to disease/impairment, typically involving questions about (limiting) longstanding illness together with subjective assessments of general health. Much less data have been collected on health state measures such as physical or psychological symptoms though a number of studies have incorporated the General Health Questionnaire (GHQ) (Goldberg & Williams, 1988) to measure psychological distress. In contrast, the present study has a broad range of health measures corresponding to Blaxter's four dimensions, which allows us to test the equalisation hypothesis both by reference to measures of health status and health state.

The second consideration refers to the conceptual and methodological adequacy of the SES measures used. Based on the assumption that SES in childhood and youth is ascribed rather than (as in adulthood) achieved, most research has utilised measures relating to the family unit such as parental occupation, education or household income which reflect different levels of material and/or cultural resources. However, the extent to which such measures adequately reflect SES in youth has been questioned precisely because of the intermediate position youth occupies between childhood and adulthood. Thus, Piko and Fitzpatrick (2001) argue that while SES in childhood is an ascribed (family) status, SES measures for young people should capture their trajectory between social origins and (probable) destinations, which they believe is best measured by subjective assessments of social status. This is an interesting development, but because it combines the resources available to an individual (ascribed SES) with attributes such as educational achievement (and health) which impact on subsequent achieved SES, it too conflates cause with effect, and in our view is not a substitute for objective SES measures.

Of the methodological problems associated with objective measures of SES, the main concern relates to the question of the validity and reliability of measures of social class based on parental occupation, with the suggestion that the reported pattern of ‘relative equality’ in youth might simply be an artefact of the measure used. Thus, Judge and Benzeval (1993) showed how conclusions about the class patterning of mortality in youth were affected by the exclusion of cases (largely lone parent families) with ‘no (occupational) class’, their inclusion in a ‘low’ social class category changing the pattern to one of health inequalities. One solution to this problem of reliance on a measure of SES that excludes an important category of poorer people is to utilise non-occupational measures such as household income, deprivation or housing tenure. When this was done with the dataset on which the original findings of ‘relative equality’ in youth were based, very little evidence of differentiation on any non-occupational SES measure was found, consistent with the ‘relative equality’ thesis (Macintyre & West, 1991).

Two further developments merit consideration. The first refers to an alternative ‘objective’ measure of SES, the ‘family affluence scale’ (FAS) (Currie, Elton, Todd, & Platt, 1997), the perceived need for which arose from the large volume of missing data associated with child and youth reports of parental occupation in the WHO Health Behaviour in School-aged Children (HBSC) study (Currie, Hurrelmann, Settertobulte, Smith, & Todd, 2000). The FAS is a simple, three item index originally comprising car and phone ownership and (child's) own bedroom, a more recent version replacing phone ownership with number of family holidays in the previous year. Although associated with very little missing data, correlations with parental social class measures are low (e.g. 0.25 with father's occupation), raising questions about what the FAS actually measures (Currie et al., 2000). The second development concerns ‘subjective’ measures of SES which are based on perceptions of (family) social position in society. These have variously referred directly to ‘socioeconomic status’ (Piko & Fitzpatrick, 2001), perceived affluence (Currie et al., 2000) and general social status as measured by position on a ‘ladder’ (Goodman et al., 2001). Correlations between these subjective measures and those based on parental characteristics are as low or lower than that reported for FAS, suggesting they too might measure another aspect of SES, or possibly another dimension altogether. Interestingly, in the HBSC study, in most (but not all) countries with relevant data, the strongest evidence for health inequalities in youth occurs on the subjective measure, followed by the FAS, the least variation being found by reference to father's social class.

In the present study, in cognizance of criticisms of occupational based measures of social class, additional measures (deprivation and housing tenure) were included and an approximate FAS measure is also possible, thus enabling a test of the equalisation hypothesis by reference to several different SES measures. There is no subjective SES measure available.

Almost all the evidence to date is cross-sectional, thereby limiting conclusions about equalisation since change can only be inferred in the comparison between observed patterns in different samples of children and youth, respectively. As noted by others (e.g. Sacker, Schoon, & Bartley, 2002), this might be erroneous for a number of reasons including sample representativeness and age/period interactions. Nevertheless, much of the evidence is consistent with the hypothesis, and particularly so when the distinction between health status and health state is taken into account. We consider the UK evidence first.

As indicated earlier, health status is most commonly measured via questions about (limiting) longstanding illness. There is now evidence from several UK studies of little or no SES variation in longstanding illness in youth, a pattern distinguished from that in childhood (see West, 1997). With respect to limiting longstanding illness, the picture is less clearcut, our ‘Twenty-07’ study for example finding a class gradient among 15-year-old males (but not females) (West et al., 1990). Furthermore, evidence from the 1991 British Census revealed that the most severe chronic illness (2–3% with ‘long-term limiting illness’) was class patterned in both sexes from infancy onwards, a finding which necessitated some revision of the earlier conclusion of relative equality in youth (West, 1997). Thus, when health status is defined by reference to more severe or restricting disorders, the evidence is less supportive of equalisation, suggesting instead the persistence of SES based influences.

Evidence in relation to another indicator of health status, self-rated health, is limited because most studies such as the UK General Household Survey (GHS) (Office of National Statistics (ONS), 2000) only collect data on those aged 16+, thus excluding both childhood and (early) youth. The findings such as they are present a mixed picture. In one study of 15–16-year-olds, no relationship was found between self-rated health and SES using occupational and non-occupational measures (Glendinning, Love, Hendry, & Shucksmith, 1992), while in our study of the same age group some evidence of health inequalities was observed (again for males but not females) (West et al., 1990). In the HBSC study, among Scottish 15-year-olds ‘feeling healthy’ was unrelated both to (father's) social class and FAS, but was related to the subjective indicator, perceived family wealth (Mullan & Currie, 2000). Thus, the evidence is inconclusive and consideration of whether SES patterns remain stable or change over the child–youth transition is not possible due to the absence of data in childhood.

By contrast, the evidence in respect of health state measures is suggestive of a change in SES patterning between childhood and youth. Thus for (non-fatal) accidents, several studies (e.g. Williams, Currie, Wright, Elton, & Beattie, 1997) have found no relationship between any accident/injury and SES in youth in contrast to the pattern of inequalities widely observed in childhood (West, 1997; Chen et al., 2002). However, variation in SES patterning is observed for particular types of accident, higher rates of sports injuries for example being reported by young people from higher social backgrounds (Williams et al., 1997). Changes in SES patterning with age may also occur for symptoms of physical ill-health, the limited data on youth showing little or no relationship with several SES measures. In the HBSC study, for example ‘daily symptoms’ (headache, stomach-ache or backache) were unrelated to father's social class, FAS and perceived family wealth among Scottish 15-year-olds (Currie et al., 2000). This pattern of relative equality contrasts with that (at least for some symptoms) usually found in childhood (West, 1997). Finally, it may apply to various aspects of mental health, a number of studies finding no relationship in youth between SES and both malaise symptoms (e.g. Mullan & Currie, 2000) and ‘psychological distress’ as measured by the GHQ (e.g. McMunn, Bost, Nazroo, & Primatesta, 1998). For more severe mental health problems, as with severe chronic illness, there does appear to be stronger evidence of SES differences. Thus, in a recent UK study (Meltzer, Gatward, Goodman, & Ford, 2000), rates of psychiatric disorder, particularly ‘conduct disorder’, increased with falling SES using both occupational and non-occupational measures, a pattern indicative of continuity throughout childhood and youth.

In sum, these (predominantly cross-sectional) findings suggest that without taking into account the distinction between health status and state, the equalisation hypothesis could be mis-specified. It is unlikely, however, that this is the only dimension of relevance. Much of the evidence adduced in support of equalisation derives from this body of UK research, and it is plausible that the patterning of health in youth by SES might vary between countries. In the US, where there has recently been considerable interest in the issue, the findings are mixed. In one review (Brooks-Gunn, Duncan, & Britto, 1999), the main conclusion was that among elementary and high school pupils there was little or no relationship between SES (primarily family income) and various physical and mental health measures, in contrast with the large effect observable for IQ and educational outcomes. By contrast, among 11–21-year-olds in the ‘Add-Health’ study, self-rated health and depressive affect were related to three different measures of SES (Call & Nonnemaker, 1999; Goodman, 1999). However, while this finding also extended to obesity, no relationship was found either with asthma or with two ‘acute conditions’ (suicide attempt and STD), leading the investigator to emphasise the inconsistency of the SES/health gradient in youth (Goodman, 1999). In a third study of 11–17-year-olds, using the CHIP, a relationship between SES was found with a composite ‘health profile’ index (Starfield et al., 2002). However, this did not apply to a ‘disorders’ domain (including major and minor symptoms) which was not included in the index.

Elsewhere than the US the evidence is again mixed. In respect of the health status measure, ‘long-term illness’, a weak relationship with SES (education and income) was found in a combined Nordic sample of 2–17-year-olds, a pattern also extending to the health state measure, ‘psychosomatic complaints’ (Berntsson & Kohler, 2001). Unfortunately, no finer age breakdown is provided. By contrast, among secondary school pupils in Hungary, no relationship was found between self-rated health, psychological well-being, psychosomatic symptoms and SES as measured by parental occupation (Piko & Fitzpatrick, 2001). Among 16-year-olds in the Australian Youth Survey, there was also no relationship between social class and either self-rated health or psychological health as measured by the GHQ, the latter suggesting a reverse class gradient (Siahpush & Singh, 2000). A similar result was found on a range of health measures, including symptoms and psychosomatic complaints, in two groups of Swiss adolescents (Vuille & Schenkel, 2001).

In the HBSC study (Mullan & Currie, 2000), considerable between country variation in the SES/health relationship was found among 15-year-olds. For example, while self rated health and father's social class were unrelated in five countries (including Scotland), in four others a weak relationship was observed, while in two it was quite strongly class-related. More evidence of SES variation occurred with the FAS (7 of 11 countries) and subjective SES measures (all countries). Similarly, in respect of daily symptoms, no relationship was found with social class in five countries while in six others a relationship was found, the FAS and subjective measures again exhibiting more variation. The contrast is particularly pronounced between 15-year-olds in Scotland, where symptoms were not differentiated on any SES measure, and their counterparts in Latvia, the US and Denmark where they were. More extensive analysis of the Danish HBSC dataset, using the (occupational) social class measure, shows that this applies to both physical and psychological symptoms and appears to be a consistent pattern at ages 11, 13 and 15, largely explained by different (class based) family orientations to school (Due, Lynch, Holstein, & Modvig, 2003). Thus, even on health indicators like symptoms, hypothesised to be more likely candidates for equalisation, there is evidence demonstrating it is unlikely to be a universal phenomenon, but rather one responsive to different constellations of class and youth influences in different societies.

As originally specified, the equalisation hypothesis involves a change in SES patterning from one in childhood characterised by health inequalities to one in youth characterised by relative equality which could occur ‘when one or more influences associated with age (the school, peer group or youth culture) cut across those associated with class (the family, home background and neighbourhood), the net effect of which is to reduce, remove or even reverse social class differences in a characteristic present in the earlier period of childhood’ (West, 1997, p. 836). In light of the distinction between health status and health state, we hypothesise that this is least likely in respect of severe chronic illness and height, and most likely in respect of symptoms of physical and mental health together with accidents/injuries. We further hypothesise that this will hold across occupational (social class) and non-occupational (deprivation, housing tenure and FAS) SES measures.

The analysis draws on a longitudinal study of young people in one social context, the West of Scotland, the design enabling a proper assessment of stability or change in SES patterning of health over this period since it refers to the same cohort and not, as in cross-sectional studies, to different samples of children and young people. In addition, although not specified in the original hypothesis, because influences associated with adolescence (e.g. puberty) and/or youth (e.g. youth culture) might plausibly impact differently (perhaps earlier) on females than males we report the findings separately by sex.

Section snippets

Background and sample

Data are derived from the West of Scotland 11 to 16 Study, located in the Central Clydeside Conurbation, a predominantly urban area centred around Glasgow city. One among several aims of 11 to 16 was to test the hypothesis of equalisation in health between late childhood and mid adolescence (West & Sweeting, 1996). Reflecting this, it is a longitudinal, school-based, study of a cohort first surveyed in their final year of primary school aged 11 (1994) and followed up in their secondary schools

Health status—longstanding illness and self-rated health

Table 1 shows rates of (limiting) longstanding illness, and self-rated health, by social class at each age and for each sex. In respect of longstanding illness, there are no significant class variations for either males or females at any age. For limiting illness, there is some evidence of a class gradient among males at ages 11 and 15, though neither of the comparisons reaches conventional levels of statistical significance. Among females, there is no evidence of any significant difference

Conclusion

We have examined the extent to which the pattern of relative equality in health in youth, found in many studies, might in part result from processes of equalisation between childhood and youth which reduce, remove or even reverse SES differences observed in the earlier period. It was hypothesised that equalisation would be more likely to occur for measures of health state (physical and mental health symptoms and accidents/injuries) than health status (longstanding illness and self-rated health)

Acknowledgements

The authors would like to thank Sally Macintyre and two anonymous referees for very helpful comments on this paper, and the young people, teachers, schools, nurse interviewers and all colleagues from the MRC Social & Public Health Sciences Unit involved in the study. Patrick West and Helen Sweeting are supported financially by the Medical Research Council of Great Britain.

References (50)

  • L.T Berntsson et al.

    Long-term illness and psychosomatic complaints in children aged 2–17 years in the five Nordic countriesComparison between 1984 and 1996

    European Journal of Public Health

    (2001)
  • M Blaxter

    Self-reported health

  • M Blaxter

    A comparison of measures of inequality in morbidity

  • M Blaxter

    Health and lifestyles

    (1990)
  • J Brooks-Gunn et al.

    Are socioeconomic gradients for children similar to those for adults? Achievement and health of children in the United States

  • K.T Call et al.

    Socioeconomic disparities in adolescent healthContributing factors

    Annals of the New York Academy of Sciences

    (1999)
  • V Carstairs et al.

    Deprivation and health in Scotland

    (1991)
  • Census Microdata Unit. (1993). A user guide to the SARs. Manchester: University of Manchester Census Microdata...
  • E Chen et al.

    Socioeconomic differences in children's healthHow and why do these relationships change with age?

    Psychological Bulletin

    (2002)
  • C.E Currie et al.

    Indicators of socioeconomic status for adolescentsThe WHO health behaviour in school-aged children survey

    Health Education Research

    (1997)
  • Currie, C., Hurrelmann, K., Settertobulte, W., Smith, R., & Todd, J. (Eds.) (2000). Health and health behaviour among...
  • P Due et al.

    Socioeconomic health inequalities among a nationally representative sample of Danish adolescentsThe role of different types of social relations

    Journal of Epidemiology and Community Health

    (2003)
  • Ecob, R., Sweeting, H., West, P., Mitchell, R. (1996). The West of Scotland 11 to 16 Study: Schools, sample design and...
  • Glendinning, A., Love J. G., Hendry, L. B., Shucksmith J. (1992) Adolescence and health inequalities: Extensions to...
  • D Goldberg et al.

    A user's guide to the General Health Questionnaire

    (1988)
  • Cited by (178)

    View all citing articles on Scopus
    View full text