The increasing prevalence of overweight and obesity has been linked to increasing physical inactivity and changes in eating patterns [6]. There has been an increase in the consumption of foods outside the home, and increases in portion size in out-of-home outlets (particularly 'fast food' outlets) [7-9]. In the UK in 2002–3, 27% of expenditure on food and drinks (excluding alcohol) was spent on consumption outside the home [10].

Foods eaten outside the home are often higher in energy and fat than foodstuffs prepared at home ; and it has been suggested that:

'the high energy densities of many fast foods challenge human appetite control systems with conditions for which they were not designed. Among regular consumers this is likely to result in the accidental consumption of excess energy and hence to promote weight gain and obesity' [7] p187.

Studies of fast food restaurant use have shown positive associations with intake of total energy and percent fat, and negative associations with intakes of fibre [11,12]. A longitudinal study found that people who ate meals from fast-food restaurants more than twice a week, at both baseline and 15 year follow up, gained 4.5 kilos more weight and had 104% greater increase in insulin resistance than people who ate less than one meal at a fast-food restaurant each week [13]. One study in the USA found, at the state level, a correlation between density measures of fast-food restaurants (per resident and per square mile) and obesity rates [14], although another study found that childhood overweight was not associated with proximity to fast-food restaurants in Cincinnati [15].

Dietary intake among poorer socio-economic groups is less likely to meet current nutritional guidelines for fruit and vegetable consumption [16], and more likely to be high in fat, salt and sugar (typically features of fast food) [17]. In the UK this applies to socio-economic status (SES) whether measured by household occupational social class or area-based measures of deprivation [18,19]. In developed countries obesity rates in women rise linearly with decreasing SES; the pattern for men is less straightforward and there are less steep, and in some places no or non linear, SES gradients in men [18-21]. However, area deprivation has consistently been shown to be related to overweight and obesity, even in models which take individual socio-demographic characteristics into account [22-24].

This observation has led to researchers to hypothesise that deprived areas may have fewer outlets selling healthy foods at affordable prices [25], but be better supplied by fast food outlets selling foods that are high fat and energy dense [26]. Evidence on the first point is equivocal, some studies finding poorer access in more deprived areas to supermarkets selling foods recommended in current dietary guidelines [27-30], but other studies finding either no socio-economic differences in shop locations or more supermarkets in poorer areas [31-37].

There is less available evidence on the second point. A study in Melbourne, Australia, found that there was a dose-response relation between SES and the density of fast food outlets, with people living in the poorest SES areas having 2.5 times greater exposure to fast food outlets than people in the wealthiest areas [26]. A study in New Orleans found that people living in low income and predominantly black areas had significantly more exposure to fast food restaurants [38]. Cummins et al found that in England and Scotland there was a higher density of McDonald's restaurants per thousand population in more deprived areas [39]. Although these are the only existing empirical studies on this issue, it is commonly suggested that residents of poorer neighbourhoods are more exposed to fast food outlets, e.g. in a frequently cited comprehensive review of obesity it is stated that:

'Poorer neighbourhoods tend to have fewer recreation amenities, be less safe, and have a higher concentration of fast food outlets' [20]p133.

Table ​Table22 outlines the results from the logistic regression analysis. Using Q2 as the reference category there was a statistically significant reduced probability of having any restaurants in data zones within Q3 (OR 0.45, 95% CI 0.24 to 0.82), Q4 (OR 0.21, 95% CI 0.10–0.43), and Q5 (OR 0.18, 95% CI 0.09–0.40). The probability of restaurants being present in data zones in Q1 was also lower but confidence intervals included 1 (OR 0.77, 95% CI 0.45–1.33). Confidence intervals were wide for fast food chain restaurants given their small number and estimates are thus unreliable. For cafes, the probability of data zones having any outlets was lower in each quintile than in the reference category, but only in Q4 did the confidence interval for the estimated odds ratio not include 1 (OR 0.40, 95% CI 0.22–0.73). Takeaways showed no clear pattern with the only statistically significant result being for Q1 (OR 0.49, 95% CI 0.29–0.81). Finally, combining all types of outlet, the probability of data zones having any outlet was lower in quintiles 1, 3, 4 and 5 than 2, though for Q3 the difference was non-significant (95% CI 0.56–1.42). There was no evidence of a linear relationship between any of the categories of outlet and deprivation. Overall there was no evidence that poorer areas in Glasgow were more likely to contain any out-of-home eating outlets.

The map displays the location of all outlets and shows that out-of-home food outlets tend to be located in the city centre (which is a retail centre, theatre/concert hall/cinema/club/pub area, Central Business District, and transport hub containing the main bus station and two train stations); along arterial highways (Great Western Road, Maryhill Road, London Road); and in cosmopolitan areas which are busy with workers during the day and have active night-time economies (e.g. Byres Road, Pollokshaws Road). From maps overlain with quintiles of area deprivation it appears that all types of outlet tend to be more common in City Centre and West End type areas, and to be less common in the peripheral deprived social housing areas such as Easterhouse, Castlemilk, Drumchapel, and Pollok, and in the most affluent housing areas in the City (maps not shown, available from authors).

These findings may be unique to Glasgow and its particular history of urban development and planning, but we suspect the pattern we observed may be common in other UK cities. This is because restaurants and takeaways are likely to be located where there is most potential custom both during the day and evenings, and such demand is higher in retail, transport and commercial centres, areas with high density of entertainment facilities such as cinemas, theatres and pubs, and along arterial highways with much passing traffic. 'Gentrification' has involved the movement of socio-economically advantaged individuals into Glasgow City centre (e.g. 'the Merchant City') and the West End. Deprived areas in Glasgow are primarily residential and have often been noted to be lacking a whole range of local amenities (such as public transport, schools etc), especially the peripheral public housing estates (Castlemilk, Eaterhouse, Pollok, Drumchapel; see map) which were built immediately after the second world war to alleviate appalling housing conditions in the inner city. Thus the location of out-of-home eating outlets may reflect rational responses on the part of owners or franchisees to principles of supply and demand. This pattern may be different from those observed in the USA, where deprived populations may be concentrated more in the inner city and wealthier people dispersed to suburbs [34,45], and from Australia, where it has been argued by some that there is less spatial segregation along social and economic lines [46]. The different measures of deprivation used in the New Orleans, Melbourne, and Glasgow studies may also contribute to different findings.

Despite the fact that the location of out-of-home eating outlets may be a rational response to likely demand, there seems to be a prevailing assumption that such outlets, particularly fast food outlets, are targeted at deprived communities and are part of the reason for poorer diets and higher obesity levels in poor places. In an article in The Observer, a national Sunday newspaper, David Smith wrote about Shettleston, a relatively deprived area in Glasgow, which has one the worst life expectancy records in the UK. He gave considerable emphasis to the density of fast food outlets:

'At the front of Celtic FC's Parkhead Stadium, children queue for a burger bar at 10 in the morning....(and) Bridgeton, within the Shettleston constituency, is possibly the alcohol and fast food capital of Britain. Within a radius of just 200 yards around the metro station there are nine pubs, an off licence and seven takeaways'

and he quotes several local people similarly emphasising fast food availability:

"The children go to Burger Kings or McDonalds, and there's nothing you can do"...."There is a fast food shop at every corner. Going to those places becomes a habit" [47].....

While such stereotypes are common, they are not supported by the findings of this study. Though there are a number of out-of-home outlets along Shettleston Road and London Road (see map) their presence there is no more dense than along other main roads including ones in more middle class areas such as the West End. Just as perceptions on the part of the mass media, policymakers and food activists to the effect that deprived communities have poorer access to healthy foods at affordable prices may not be borne out by the empirical evidence, [34] perceptions held about a greater exposure among deprived communities to unhealthy diets may not necessarily be empirically substantiated.

There are clearly some limitations to our study. It is restricted to one city (as are those undertaken in Melbourne and New Orleans). By definition, we have not been able to map the location of mobile fast food outlets (e.g. vans selling pies, burgers etc) and it may be that these target poorer residential areas. It should also be noted that although takeaways and the chain restaurants primarily serve what is conventionally called fast food (i.e. high in fat, energy, and salt), they, and other restaurants and cafes, may serve healthier options instead of, or as well as, fast foods. Our interpretation of the maps is descriptive rather than using a more sophisticated GIS approaches.

Our findings may not appear consistent with those from another study in which we assessed the location of McDonald's restaurants in England and Scotland in relation to deprivation in 2005, and found that these restaurants were significantly more likely to be found in more deprived neighbourhoods[39]. Large global chains such as McDonald's have sophisticated marketing systems that allow them to pin-point with some accuracy the 'optimum' location based on geo-demographics, distance from headquarters, distribution points and main transport links and intersections, and sales figures in existing outlets [9,48] whereas the locational strategies of independent outlets may be more locally based. It may also be that individual chains avoid areas where others are located, so that if one chain tends to be located in more deprived neighbourhoods, others will locate in less deprived ones and so balance each other out.

The association between area level social deprivation and the availability of foods prepared outside the home, in particular fast foods, must be investigated in the context of the local availability of land, the price of real estate and the ease of obtaining planning permission. Differences in these factors may explain the differences between the findings reported here and those reported nationally for a single large global chain, and in other countries.