• No se han encontrado resultados

Capí tulo X. Donde se cuenta la industria que Sancho tuvo para encantar a la sen ora Dulcinea, y de otros sucesos tan

The systematic reviews include literature published up until July 2009 for the social

environment (Carter and Dubois, 2010) and August 2009 for the built environment (Galvez et al., 2010). Ovid was used to search the Medline and Embase databases to identify recent studies (i.e. published since the systematic reviews) which had investigated the association between at least one neighbourhood characteristic (either social or physical/built) and body size in children or adolescents. Thirteen studies were identified; six based in the US, two each in Canada, Australia, and the UK, and one in Ireland. Three had a longitudinal design and height and weight were measured in all but one of the studies. The study outcomes were all BMI/overweight/obesity except for one study which had percentage body fat (Dengel et al., 2009). Sample ethnicity was reported in four of the studies (all US-based), including one which included only Latino children. In order to summarise the recent findings in these 13 papers, I have grouped the neighbourhood characteristics into broad categories (physical activity facilities; walkability; crime/safety; food sources; deprivation; ethnic density; green space; public transport and population density) and these are

discussed in turn. An overall summary is provided in Table 1.3.

.eighbourhood physical activity facilities

Six of the studies included a neighbourhood physical activity (PA) facilities measure. Three used perceived measures (Davidson et al., 2010, Edwards et al., 2009, Nelson and Woods, 2009), two used objective measures (Dengel et al., 2009, Timperio et al., 2010) and one used both (Crawford et al., 2010). Three of the four studies with perceived measures, and one of the three studies with objective measures, found a significant association between perceived physical activity facilities in the neighbourhood and body size.

In the study by Davidson et al. (2010), children of parents who perceived that there were good playgrounds/parks/sidewalks in their neighbourhood had significantly lower BMIs than those whose parents did not (this study grouped sidewalks with parks rather than walkability factors as is done in most other studies) independent of individual SES. In a study of adolescents, those who were overweight or obese reported significantly fewer PA facilities within 5-10mins walk from their home, or on a frequently travelled route (e.g. somewhere they passed on way home from school). Furthermore, the perceived number of PA facilities was negatively associated with overweight/obesity, and the only variable to

remain significant, in fully adjusted models (adjusting for sex, age, population density, parental occupation and other neighbourhood characteristics) (Nelson and Woods, 2009). This study also considered physical activity as a covariate and found that it did not

‘directly influence or mediate the relationship between perceived facilities and weight status’ (p917).

The study by Edwards et al. (2009) was unusual in that the neighbourhood perceptions were those of respondents in a separate study (Health Survey for England 2002); their responses were aggregated and linked to each child’s data at the LSOA level. In this study ‘good access to leisure facilities’ (as perceived by the HSE respondents) was associated with lower levels of obesity regardless of the affluence of the area. This study included two age groups however the sample size was not adequate to test whether the findings differed by age. An important limitation of this study is that it does not appear that any individual-level covariates were adjusted for. One study with perceived measures did not find a significant association; parental perception of availability of sporting venues in the local area was not significantly related to change in BMI SDS (it is unclear which

covariates were adjusted for) (Crawford et al., 2010).

Timperio et al. (2010) used both 800m and 2km buffers around the children’s homes to calculate the density of sport/recreation spaces with no fees or restricted opening hours; the number of sports options for children; the existence of gyms/leisure centres or swimming pools; the length of walking/cycling tracks; and the distance to school. The authors chose the 800m buffer as parents reported that was the walking distance of the younger children (5-6yrs at baseline); the larger 2km buffer size was selected because walking distance increases as children grow older and ‘the scale of neighbourhoods can be further extended via vehicle use’ (p3). In cross-sectional analysis, the number of sports/recreation public open spaces within 800m was inversely related to BMI in the 10-12yr olds but not the 5- 6yr olds. In the longitudinal analysis (change in BMI SDS over 3yrs as outcome) none of these neighbourhood variables was significant.

The Crawford et al. (2010) study used data from the same study as Timperio (2010) but included only the older children (10-12yrs at baseline) and three waves of data rather than the two in the Timperio study. The objective measures are as described for the Timperio study, except that only the 2km buffer was used; none of the measures were significantly related to BMI SDS over 5 years. Dengel et al. (2009) calculated the distance from each child’s home to the nearest park, gym, recreation centre, walking/biking trail, and to

school. There was no significant association between distance to these facilities and % body fat in models adjusted for age, sex and pubertal status or in models stratified by sex.

.eighbourhood walkability/road safety

The definitions of ‘walkability’ differed by study; I have taken walkability to include any items related to: the presence of sidewalks/pavements; the layout/density/type of roads; traffic/road/pedestrian safety; and street lighting. Studies solely examining overall safety/crime are not included here as they are summarised in a later section; however it is acknowledged that general crime levels may be an important component of walkability.

Six studies examined walkability. Three used perceived measures (Davidson et al., 2010, Elder et al., 2010, Nelson and Woods, 2009), two used objective measures (Dengel et al., 2009, Timperio et al., 2010) and one used both (Crawford et al., 2010). Two studies reported significant associations (one perceived and one objective). The study with

perceived measures which found a significant association combined parental reports of the existence of sidewalks on most streets with existence of good playgrounds/parks

(Davidson et al., 2010). This variable had a negative association with BMI (i.e. the more sidewalks/playgrounds/parks the lower the BMI); however it is not possible to know whether it was the sidewalks (which are arguably directly related to neighbourhood walkability) or playgrounds/parks (which are perhaps more ‘destinations’) or both which are driving this negative association.

In the one study which included adolescent’s perceptions of walkability, overweight/obese adolescents were more likely to report poorer pedestrian safety in their neighbourhoods however there were no differences in the other walkability characteristics (e.g. street connectivity, street lighting, presence of hills) by weight status. Furthermore, in fully adjusted models none of the walkability characteristics were associated with

overweight/obesity (Nelson and Woods, 2009). Two studies found no association between parental perceptions of walkability and their child’s BMI SDS (Crawford et al., 2010, Elder et al., 2010). In the study of Latino children from relatively deprived families, neighbourhood walkability was the last variable to be added to a model which already included many potential confounders (child characteristics; home characteristics; parental characteristics; and school characteristics) (Elder et al., 2010). Objective measures of road connectivity (cul-de-sacs, intersection density, total length of access paths) and traffic exposure (total length of busy roads) were included in the Australian studies (Crawford et

al., 2010, Timperio et al., 2010). In the Timperio et al. (2010) study, in cross-sectional analysis more access paths were associated with lower BMI SD scores for both younger (800m and 2km buffer) and older children (800m buffer only). The length of local roads in 2km buffer was also negatively related to BMI SDS for older children. In longitudinal analysis (change in BMI SDS over 3yrs), the number of 4-way intersections was negatively associated with change in BMI SDS for younger children (800m buffer). A greater length of access paths was associated with a greater increase in BMI SDS in older children (800m buffer). In the Crawford et al. (2010) study none of these measures were related to BMI SDS. A study with objective measures of street patterns, pedestrian infrastructure, and distance to and density of transit neighbourhood (all within 1600m buffer) found none to be associated with % body fat (Dengel et al., 2009).

.eighbourhood crime/safety/disorder

Six studies examined the impact of neighbourhood safety on body size; all used perceived measures. Three of the studies found significant results. In the study by Bacha et al. (2010), mothers reported their perceptions of neighbourhood safety when their children were in 3rd Grade (8-9yrs), then the height and weight of the children was measured in 5th Grade (10-11yrs). Girls, but not boys, living in the least safe tertile in 3rd Grade were significantly more likely to be obese and to have higher mean BMI SD scores in 5th Grade than children in the safest tertile (after adjustment for sex, race and household SES). Time spent outdoors, television watching, and pubertal status did not explain these significant relationships between perceived safety and body size. The Davidson study (2010) also included 5th graders and parents safety perceptions; children living in areas perceived to be safer had significantly lower BMIs than those in less safe areas. Edwards et al. (2009) only included one neighbourhood safety item (from HSE 2002); ‘perception of problem with teenagers hanging around’. This variable was positively associated with child obesity in the most affluent census ward in the study but not in the medium or low SES wards.

Adolescents who rarely or never felt safe in their neighbourhoods were more likely to be overweight or at risk of overweight but results were not statistically significant in fully adjusted models (controlling for ethnicity, school grade, and clustering by school) (Duncan et al., 2009). Those who said they never or rarely felt safe in their neighbourhoods were more likely to have reported the presence of gang violence and to have seen someone attacked with a weapon. In a further study of adolescents, personal safety (streets well lit at night; crime rate) and aesthetics (litter; trees along the streets) were not associated with

overweight/obesity in models adjusted for age, sex, SES and clustering by school (Nelson and Woods, 2009). Similarly, no significant association with BMI SDS were found for the safety measure used by Elder et al. (2010) (already described in the previous section on walkability as it is an aggregate of items on perception of crime and lights and vehicle exhaust).

.eighbourhood food outlets

Five studies investigated whether the presence of food outlets in the neighbourhood was associated with body size. The type of food outlets assessed varied between studies. One out of the two perceived studies, and one out of the three objective studies, found a

significant result. In one study, perceived access to supermarkets (from HSE 2002) had a strong negative association with BMI SDS independent of the ward deprivation level Edwards et al. (2009). However in the other study to use perceived measures, parents’ perceived access to stores to purchase vegetables and fruits was combined with two other measures (‘like my neighbourhood’ and ‘access to sport/recreation programs’) to make an overall neighbourhood satisfaction variable; this was not associated with BMI SDS (Davidson et al., 2010).

Children living in deprived areas of Leeds (a multi-ethnic UK city) had more fast food outlets in their neighbourhoods than children in more affluent areas, and the distance from the home to the nearest fast food outlet was shorter for those in deprived areas. In models adjusted for age, sex, and neighbourhood deprivation the density of fast food outlets in an area was positively associated with overweight/obesity but not with BMI SDS. Proximity to nearest fast food outlet was not associated with either overweight/obesity or BMI SDS (Fraser and Edwards, 2010). In a US study of young children (5-8yrs), neither the number of restaurants or grocery stores within a 1-mile radius of the school was significantly related to BMI SDS (Elder et al., 2010) (this study used school neighbourhood as a proxy for residential neighbourhood).

Dengel et al. (2009) report a significant negative association between the distance to the nearest fast food outlet and % body fat, and a significant positive association between % body fat and the density of both small and large grocery stores within 1600m (20min walk) of the home in unadjusted analysis. However after adjustment for age, sex and pubertal status there were no significant associations. It was not possible to determine which of these factors explained the association as adjusted models were not presented.

.eighbourhood green space

Two studies, both using objective measures, examined the relationship between green space and body size. No significant results were found. In one study, percent land use for ‘parks and recreation’ within 1600m (20min walk) of home, and distance from home to park, were unrelated to % body fat (Dengel et al., 2009). In the second study, four measures of green space were examined: the number of parks/green space per 10,000 residents in an area; the proportion of land in an area which was parks/green space; mean distance to nearest park/green space in each community; proportion of each area within walking distance (800m) of park/green space. No significant associations with

overweight/obesity were found for any of these measures in analysis controlled for age, sex, neighbourhood deprivation and ethnic density (Potestio et al., 2009).

.eighbourhood land use - other

Land-use ‘other’ refers to neighbourhood land use which is not otherwise covered in other sections (such as green space, road networks, food outlets). The study by Dengel et al. (2009) explored the proportion of land use within a 1600m buffer which was residential, and the proportion which was vacant. Neither was associated with % body fat.

.eighbourhood Ethnic density

Two studies (both US) investigated the impact of neighbourhood ethnic density on body size. Non-White ethnic density was positively associated with obesity levels in analysis adjusted for age, sex and individual SES (health plan) and spatial clustering; however this relationship was completely attenuated by further adjustment for neighbourhood

deprivation (Grow et al., 2010). Similarly, ‘% visible minority’ in a neighbourhood had no significant association with overweight/obesity in models adjusted for sex and

neighbourhood deprivation (Potestio et al., 2009). A weakness of both these studies is that individual level measures of ethnicity were not included.

.eighbourhood deprivation

The association between neighbourhood deprivation and body size was examined in four studies (2 US, one UK, one Canada); three found deprivation to be positively associated with overweight/obesity. In the Canadian study, neighbourhood deprivation was measured by community education level (% with Bachelor’s degree). Children in less educated areas

had higher the rates of overweight/obesity independent of family-income (which was also a significant correlate) and % visible minority in the neighbourhood (Potestio et al., 2009).

Similarly, in Leeds neighbourhood deprivation (Index of Multiple Deprivation at LSOA level) was significantly associated with overweight/obesity (but not with BMI SDS) in models adjusted for age, sex, and density of fast food outlets in the neighbourhood. When ‘distance to’ rather than ‘density of’ fast food outlets was included in the models, the association between neighbourhood deprivation and obesity remained significant but not the association with overweight including obesity (Fraser and Edwards, 2010).

Furthermore, in a US study all four measures of census tract deprivation (median

household income, % home ownership, % females with ≤ high school education, % single parent household) were significantly associated with obesity when analysed independently in models adjusted for age, sex and SES (health plan). For each variable, more deprivation was associated with higher levels of obesity. When entered into a model together, along with % non-White, only median income and home ownership remained significant. However in this model a decrease in home ownership was associated with lower obesity levels (i.e. the opposite direction of relationship to that seen in the univariate analysis) (Grow et al., 2010).

In contrast, another US study did not find an association between neighbourhood SES and BMI (Voorhees et al., 2009). This study, which included only girls, measured

neighbourhood SES by the Townsend Index (based on % employed, % owner occupation, mean number persons per household, and % households with no vehicle). Neighbourhood was defined by a 0.5 mile buffer around the home, and measures of individual SES

(highest parental education, whether the child was in receipt of a free school lunch) and school SES (% in school receiving free lunch) were adjusted for. None of the SES measures (individual, neighbourhood or school) were significantly related to BMI. Ethnicity was significant independent of the SES measures; Blacks and Hispanics had higher BMIs than Whites. This was one of only two studies included in this review of recent papers which presented any results stratified by ethnicity/ethnic-specific effects sizes (the other being the Duncan (2009) study).

Public transport

Two studies included perceived measures of public transport. There was no association between BMI SDS and parental reports of whether ‘public transport is limited in my area’ (Crawford et al., 2010). Similarly, in the study which used neighbourhood perceptions from the HSE 2002, ‘quality of public transport links’ was unrelated to children’s obesity levels (Edwards et al., 2009).

Population Density

Four studies included a measure of population density. Two studies classified areas based on population (e.g. urban, town, rural); Davidson (2010) found that those in towns and rural areas had higher BMIs than those in urban areas (although detailed results are not reported). However, Nelson and Woods (2009) found no relationship between area type and overweight/obesity. An alternative density measure, population per unit land area in a 1600m buffer of each home was unrelated to % body fat (Dengel et al., 2009). Similarly, number of households per acre was unrelated to obesity (Grow et al., 2010). Overall there was little evidence of an association between population density and body size.

School neighbourhoods

Young people who walk or use public transport to get to school, and those who leave their school grounds at lunchtime, will spend time in the neighbourhood surrounding their school. School neighbourhoods could impact on obesogenic behaviours through

encouraging unhealthy diets or discouraging active transport. Five studies examined the association between school neighbourhoods and body size; one UK-based (Drummer et al., 2005), one US-based (Powell et al., 2007), and three Canadian (Janssen et al., 2006,

Merchant et al., 2007, Seliske et al., 2009) (of which the Seliske and Janssen papers used data from same study). Only Seliske (2009) explicitly stated an interest in the effects of school neighbourhood on obesity. In the Merchant et al. (2007) study, which used perceived rather than objective measures, the neighbourhood and school neighbourhoods are conceptualised as being one in the same due to children living close to school. In the other studies, school neighbourhood is a proxy of residential neighbourhood (perhaps due to individual postcodes not being available). A sixth study included neighbourhood characteristics at both the school and neighbourhood level (Sturm and Datar, 2005). However as results were almost the same the school results were not presented in the paper (this study is included in the systematic reviews described previously).

Two of the studies examined the food environment surrounding schools and body size. The Canadian study found that pupils with more food retailers in their environment actually had a lower risk of overweight/obesity than those with less food retailers; they concluded that ‘limiting the number and type of food retailers within the school environment may not be an effective strategy for the prevention and reduction of overweight and obesity in youth’ (Seliske et al., 2009). In contrast, the US-based study found a significant negative association between density of chain supermarkets and BMI/overweight status independent of the presence of other food outlets, food price indices, and area-level SES (Powell et al., 2007). Conversely, the density of convenience stores was positively associated with BMI and overweight although much of this association was explained by neighbourhood SES. This study stratified analysis by ethnicity and mother’s work status; the association between supermarket density and BMI was substantially stronger for African Americans than Whites and Hispanics, and stronger for those whose mothers worked full-time compared to those whose mothers did not work.

Three studies investigated school neighbourhood deprivation and body size. The UK- based study found no association between electoral ward deprivation (IMD) and overweight/obesity (Drummer et al., 2005). In contrast, a Canadian study found that school neighbourhood deprivation (as measured by unemployment rate) was significantly associated with obesity independent of family-level SES. However the other measures of area-level deprivation in this study (education levels and income) were not associated with obesity. This study used a relatively large 5k buffer around schools; however they report that analysis was also conducted with a 1k buffer and results were the same (Janssen et al., 2006). In another Canadian study, the two schools were located in areas which differed in

Outline

Documento similar