PT - JOURNAL ARTICLE AU - Richard Mitchell AU - Thomas Astell-Burt AU - Elizabeth A Richardson TI - A comparison of green space indicators for epidemiological research AID - 10.1136/jech.2010.119172 DP - 2011 Oct 01 TA - Journal of Epidemiology and Community Health PG - 853--858 VI - 65 IP - 10 4099 - http://jech.bmj.com/content/65/10/853.short 4100 - http://jech.bmj.com/content/65/10/853.full SO - J Epidemiol Community Health2011 Oct 01; 65 AB - Background The potential for natural environments to be salutogenic has received growing interest from epidemiologists, but there has been no critical examination of the extent to which associations between green space and health might vary according to the indicator of green space coverage used.Methods Three different indicators of green space coverage were derived for a set of 268 small areas in four cities within Britain. The indicators had different origins and provided a spectrum of sensitivity from larger spaces only, through to ambient greenery. Two indicators reproducible for anywhere in Europe were included. Agreement between the indicators on the quantity of green space in a small area, and their independent association with measures of mortality and self-reported morbidity, were compared.Results Overall, the indicators showed relatively close overall agreement (all r2>0.89, p<0.001). However, agreement varied by level of area socioeconomic deprivation (p<0.001). The indicator that detected larger spaces only found less green space in areas of socioeconomic deprivation than the other two. Despite this difference, all indicators showed similar protective associations with the risk of mortality and self-reported morbidity suggesting that larger green spaces may be more important for health effects than smaller spaces.Conclusions Associations between green space indicator and health were not sensitive to indicator origin and type. This raises the possibility of trans-European epidemiological studies. Larger green spaces may be the most important for health effects, but may also be less prevalent in more deprived areas.