Sports facilities, socio-economic context and overweight among the childhood population in two southern European cities: a cross sectional study

Prevalence of overweight and obesity in children and adolescents is increasing in developed countries such as Spain (31.8%) [1]. This in turn implies future health problems, such as obesity itself (70% of obese children and adolescents will be obese adults), [2] various chronic processes, [3] and, in addition, an increase in the risk of mortality due, in particular, to ischaemic heart disease, [4].

Several factors have been proposed as determinants of childhood weight and one of them is the socioeconomic context of the area of residence of individuals. The findings of a wide variety of studies in children and adolescents show that the prevalence of overweight is greater in areas with greater material deprivation or less wealth, regardless of household socio-economic position, [5‐9].

Furthermore, the principal cause of weight gain is known to be an imbalance between energy intake and expenditure, [3], where regulation of food intake in sedentary individuals is different to that in active individuals, [10]. In this equilibrium there is a threshold of physical activity: under it, body weight increases even at low calorie intakes; above it, small changes in exercise will suffice to maintain weight, even when calorie consumption increases, [4].

Different epidemiological studies on children have reported a negative association between physical activity and body mass index (BMI), [5, 11, 12], as well as a direct relationship between the presence of neighbourhood sports facilities and doing physical exercise, [9, 13, 14]. The presence of neighbourhood sports facilities might thus explain the relationship between exercise and BMI: even so, the results are controversial. Whereas some authors have found an association between sports facilities or parks and a decreased risk of excess weight, [15‐17], other studies have failed to observe this association, [4, 14, 18‐21].

Most of the above evidence comes from research carried out in the USA, Canada, New Zealand and the United Kingdom, and yet few studies have been conducted in other European countries, where the influence exerted by socio-economic environment on weight might well differ to that seen in English-speaking countries, [2]. Accordingly, we sought to study the relationship between socio-economic environment, as reflected by three indicators, and overweight in the cities of Madrid and Barcelona, the two towns with most inhabitants in Spain, adjusting for possible confounding factors, such as age, sex, household socio-economic position, physical inactivity and availability of sports facilities.

The characteristics of the subjects in the two cities, according to the three variables of area of residence, are shown in Table 1. The two cities showed a similar magnitude in the socio-economic context indicators. For example, the unemployment rate was 6.0% Madrid and 6.5% in Barcelona.

Overweight was related, in the case of Madrid, to the percentages of population with a tertiary education (p = 0.008) and second home (p = 0.015) respectively, and in the case of Barcelona to the unemployment rate (p = 0.034), higher education (p = 0.029) and a second home (p = 0.019). The areas with higher proportion of the population with higher education and with a greater proportion of the population with a second home show a negative relationship with overweight. In addition, in Barcelona the areas with higher unemployment rate showed a positive relationship with overweight.

Insofar as physical inactivity was concerned, in Madrid this was related to the unemployment rate (p = 0.025) and percentage of a second home (p = 0.007), whereas in Barcelona there was no relationship with any of the 3 environmental variables.

Table 2 shows the prevalence of overweight in relation to the personal variables, variables of household socio-economic position and variables of physical inactivity, for the two cities. In Madrid and Barcelona alike, the prevalence of overweight was highest among the youngest children (p = 0.000 and p = 0.000 respectively), and in Madrid, in addition, among males (p = 0.013).

Table 3 lists the ORs with their confidence intervals for overweight according to the three area indicators, for both cities. In Madrid, after adjustment for sex, age, educational level and occupational class, the ORs for the districts situated in the tertile with the worst socio-economic context were: 1.17 (0.74–1.86) by reference to the unemployment rate; 1.53 (1.00–2.32) by reference to the percentage population with a university education; and 1.57 (1.02–2.42) by reference to the percentage of the population with a second home. Adjustment for physical inactivity did not greatly reduce the magnitude of the OR. Adjustment for availability of sport facilities similarly failed to reduce the magnitude of the OR, except when the indicator of socio-economic context was the unemployment rate, in which case the magnitude decreased by 58% (from 1.17 to 1.07). In Barcelona, after adjusting for sex, age, educational level and occupational category, the ORs for the districts situated in the tertile having the worst socio-economic context were: 1.80 (1.01–3.22) by reference to the unemployment rate; 1.80 (1.00–3.23) by reference to the percentage of the population with a university education; and 1.86 (1.04–3.33) by reference to the percentage of the population with a second home. Adjustment for physical inactivity did not greatly reduce the magnitude of the OR. Adjustment for availability of sport facilities reduced the magnitude of the OR in the tertile having the worst socio-economic context: by 14% when the indicator was the unemployment rate (from 1.80 to 1.69); by 10% when the indicator was the percentage of the population with a university education (from 1.80 to 1.72); and by 9% when the indicator was the percentage of the population with a second home (from 1.80 to 1.79).

This study sought to assess the influence exerted on overweight by socio-economic context, measured by reference to the unemployment rate, and the percentages of the population having a tertiary education and second home respectively, in view of the fact that earlier studies have shown that, regardless of household characteristics, children and adolescents living in deprived neighbourhoods tend to have a higher body weight, [5, 26].

As a first step, a descriptive study was conducted to measure the relationship between individual and socio-economic factors. Since a strong association was found between the latter indicators and the number of neighbourhood sports facilities (Table 1, p < 0.000), it was decided to include these in the adjustments in the regression analysis. While previous studies undertaken in the USA, Canada and the United Kingdom [14, 27] have shown that people living in underprivileged areas are more inactive, it is equally likely that they have fewer sports facilities and green areas, [14, 28], and moreover, that this availability varies from city to city, [14].

Overweight in Madrid displayed a risk gradient, except vis-à-vis the unemployment rate, which was highest in the intermediate tertile (Table 3). In Barcelona, this increased risk was again in evidence, especially in the case of the population that had a second home (Table 3). Physical inactivity would not account for this risk of overweight in either of the two cities: in Madrid, the availability of sports facilities reduced this risk by 58% for the upper tertile of the unemployment rate; and in Barcelona, the availability of sports facilities accounted for around 10% of overweight with respect to all three indicators.

A previous study, undertaken in Canada, [5], reported a relationship between a socio-economic environmental indicator, i.e., the unemployment rate, and weight, an association also reported for other indicators by a number of studies, [5, 29‐31]. Although it is reasonable to assume that one of the mechanisms whereby socio-economic level influences overweight would be greater physical inactivity, [13, 14, 27], a certain degree of controversy nevertheless surrounds this contention, [5]. In this regard, while previous studies have found that the number of neighbourhood sports facilities is related to lower weight, [15, 16, 32] and that the most underprivileged neighbourhoods have fewer centres at which people can do physical exercise, [33], other studies have failed to observe this relationship, [34]. It should be borne in mind here that a range of factors, including the availability of foot or cycle-paths leading to such sports facilities, [35], the distance to be covered, [17], their ease of access, [21, 36], the quality of the sports centres themselves, and the safety of the neighbourhoods in which they are situated, [3, 37‐39], could exert an influence, giving rise to these differences in results.

It should be stressed that, in general, socio-economic factors were better indicators of risk in the case of Barcelona than in that of Madrid. This might be due to the administrative reorganisation of the city of Barcelona, implemented in 2007, which divided it into neighbourhoods created on the basis of sociological criteria, [40]. In Madrid, in contrast, districts are administrative divisions of the city; hence, rather than being constructs devised for the purpose of grouping together neighbouring areas with similar characteristics, they are units having a certain degree of internal homogeneity, [40].

Furthermore, it should be borne in mind that that socio-economic environmental determinants have less influence than do individual determinants on risk of overweight [5], for the simple reason that, because they are drawn from census data, they probably yield an underestimate of the real effects, due to a smaller degree of variability in the population and less opportunity for differential misclassification bias.

Insofar as the strong points of this study are concerned, mention should be made, firstly, of the fact that the data were drawn from 2 random samples of the two largest cities in Spain, and that use was made of routine census data collected by government authorities for public health purposes. Moreover, the study assessed both the individual factors (age, sex, socio-economic position and physical inactivity), and the environmental factors of wealth and poverty (unemployment rate, population with tertiary education, and population having a second home), and integrated the relationship between the two types of factors and the weight of children and adolescents, as well as the presence of neighbourhood sports facilities, by means of multilevel analysis.

A potential limitation of our study is its cross-sectional design, which does not take the temporal relationship among the variables into account, though it is highly unlikely that overweight among children would be determined by the neighbourhood in which they resided. It should likewise be noted that the BMI data were reported by the interviewees, who might not have measured the weight accurately, [41]. Various studies on obesity in children have used weight and height referred by the person responsible for the child [42, 43] and in other countries [44, 45]. In a study conducted in Spain, the BMI with information declared by the parents showed a sensitivity of 78% and a specificity of 96% [46] and in a study conducted in Germany the authors obtained sensitivities between 78 and 85% and specificities close to 100%. In any case, we cannot rule out an information bias in the measurement of overweight in our study, although there is no reasons to think that misclassification of subjects be differential with respect to the socioeconomic context indicators used.

In our study, the questionnaire was validated in 50 interviewees in the Madrid City Health Survey and in 18% of those interviewed in the Barcelona Health Survey, and it was found that information on weight and height had high reliability [22, 23] Similarly, the possibility of there being unknown errors of measurement in the collection of data on physical inactivity should not be ruled out; even so, there is no need to suppose that such errors would differ in terms of neighbourhood of residence. The use of a comprehensive population source, such as the census, for the socioeconomic characterization of the population, has the advantage that there is less probability of an information bias in this exposure measure. On the other hand, being able to reflect the great socioeconomic heterogeneity may have led to an underestimation of the effect.

Finally, as in other studies, future research is called for to study the working mechanisms through which neighbourhood social environment exerts its effect on BMI, [47].

We study the relationship socio-economic environment, and overweight in two southern European cities: overweight displayed a risk gradient in Madrid and Barcelona. That is, the areas of these cities with the best socioeconomic indicators show the lowest prevalence of overweight and the areas with the worst socioeconomic indicators showed the highest prevalence of overweight among the childhood population. Although in our study we have not shown that physical inactivity is a determining factor in overweight, there is a decrease in the risk of overweight in relation to the availability of sports facilities in the two cities.