Green space

The most common measure in large studies is the Normalised Difference Vegetation Index (NDVI), a satellite score of how much live vegetation lies within a set distance of each home. It counts any vegetation, including private gardens, farmland and roadside verges, and cannot separate a public park from greenery that residents cannot enter. The studies pooled in the 2019 meta-analysis all used NDVI within 500 m or less of the home [1], and the Nurses' Health Study used NDVI within 250 m and 1,250 m [3].

A second approach uses land-use maps of public green spaces. The London study measured the percentage of green area, the number of access points (entrances) per square kilometre and the distance to the nearest access point within a 1,000 m walking network around each home, and it separated parks by type, including pocket parks (areas for rest and recreation under 0.4 hectares) [2].

A third approach records the time people spend in natural settings. A survey of 19,806 adults in England found that, compared with no nature contact in the previous week, 120 to 179 minutes of contact was associated with higher odds of good self-reported health (odds 1.59 times as high) and high wellbeing (1.23 times), with no further gain beyond 200 to 300 minutes a week; it made no difference whether the time came from one long visit or several shorter ones [5]. This measure concerns self-reported health rather than mortality.

The 2019 meta-analysis identified more than 9,000 studies and pooled nine long-term studies from seven countries with 8,324,652 participants. Seven of the nine found significantly lower mortality with greater surrounding greenness and two found no association. The pooled hazard ratio (the death rate in one group relative to another; 0.96 means 4% lower) was 0.96 (95% CI 0.94 to 0.97) per 0.1-unit rise in NDVI, although the results varied widely between studies [1].

The Nurses' Health Study followed 108,630 US women from 2000 to 2008, during which 8,604 died. After adjustment for age, race, smoking and personal and neighbourhood income, women in the greenest fifth of homes had a 12% lower rate of non-accidental death than those in the least green fifth (hazard ratio 0.88, 95% CI 0.82 to 0.94) [3]. A broader 2018 review of 103 observational and 40 intervention studies reported lower all-cause mortality (0.69), cardiovascular mortality (0.84) and type 2 diabetes (0.72) with more green space exposure, but noted that several of its pooled results were limited by poor study quality and large differences between studies [6].

The London study linked 2011 census records for 4,645,581 adults to death registrations over a mean 8.4 years. Mortality did not differ with overall green coverage (hazard ratio 1.0004, 95% CI 0.9996 to 1.0012) and rose very slightly with more access points (1.0076). Each 1 percentage point more pocket-park coverage was associated with lower mortality (0.9441, 95% CI 0.9213 to 0.9675), and ten more pocket-park access points per square kilometre with lower respiratory mortality (0.9164). The authors stated that the mechanisms behind these associations need further research [2].

Full deep dive: green space and longevity.

The association between greenness and death differs by cause. In the Nurses' Health Study, the greenest fifth of homes had 34% lower respiratory mortality and 13% lower cancer mortality than the least green fifth, but no clear difference in deaths from coronary heart disease and a borderline difference for stroke [3]. The London study likewise found its clearest cause-specific result for respiratory deaths [2].

A 2016 expert workshop grouped the proposed pathways into three functions of green space: reducing harm, such as exposure to air pollution, noise and heat; restoring capacities, such as attention and recovery from psychological stress; and building capacities, such as encouraging physical activity and social cohesion [7]. The workshop recommended that future studies measure the presence, access and use of green space separately and use designs that can test cause and effect [7].

The Nurses' Health Study tested four possible pathways statistically. Lower rates of depression (physician-diagnosed or treated with antidepressants) explained an estimated 30.6% of the association between greenness and mortality, and social engagement explained 19.1%. Lower fine-particle air pollution (PM2.5) explained 4.4% and physical activity 2.1% [3]. Each pathway was estimated separately, the confidence intervals were wide, and all four together explained an estimated 27.1%, leaving most of the association unexplained [3].

A Dutch study of 10,089 residents supports the social pathway: people with less green space within 1 km and 3 km of home reported more loneliness and a greater perceived shortage of social support, and these partly accounted for the link between green space and self-rated health [8]. The study measured everything at one point in time, so the order of cause and effect cannot be determined.

The biodiversity hypothesis proposes that contact with a wider range of environmental microorganisms shapes the bacteria that live on the body and trains the immune system. In a sample of Finnish adolescents, those with allergic sensitisation had lower environmental biodiversity around their homes and a lower diversity of one group of skin bacteria (gammaproteobacteria) [9]. The study concerned allergy in teenagers rather than mortality. Physical activity outdoors is covered under walking and daily steps and sports and active hobbies.

A study of the whole population of England below retirement age, 40,813,236 people with 366,348 deaths from 2001 to 2005, tested whether green space changed the link between income and death. In the least green areas, the most income-deprived quarter had 1.93 times the all-cause death rate of the least deprived (95% CI 1.86 to 2.01); in the greenest areas the ratio was 1.43 (1.34 to 1.53). For circulatory disease the ratios were 2.19 and 1.54 [4].

The same pattern did not appear for lung cancer or intentional self-harm, causes of death that green space would not be expected to affect [4]. This check makes it less likely that the result simply reflects differences between the kinds of people who live in greener areas, although it does not rule that out.

The main studies measure different exposures. Satellite greenness, public park maps and time spent in nature are not interchangeable, and a finding for one does not transfer to the others [2, 5]. The pooled estimate for satellite greenness showed very large differences between studies [1], and the largest single study found no association with overall green coverage [2].

All of the mortality evidence is observational. People who live in greener areas differ in income, health and habits, and although studies adjust for many of these factors, unmeasured differences can remain. The pathway estimates in the Nurses' Health Study come from a single group of female nurses in the United States [3], and the social and biodiversity studies measured exposure and outcome at the same time [8, 9]. No randomised trial has tested whether adding green space near homes changes death rates.