Www.activelivingresearch.org International Studies of Physical Activity & Built Environment James Sallis San Diego State University .

www.activelivingresearch.org

International Studies of Physical International Studies of Physical Activity & Built EnvironmentActivity & Built Environment

James SallisJames SallisSan Diego State UniversitySan Diego State Universitywww.drjamessallis.sdsu.eduwww.drjamessallis.sdsu.edu

www.activelivingresearch.org

Health Statistics and Informatics

Deaths attributed to 19 leading factors,by country income level, 2004

www.activelivingresearch.org

Why Environment & Physical Why Environment & Physical Activity?Activity?

Broad societal & technological developments are believed to be reducing PA in work, transport, & household settings

These developments are happening in all countries

Ecological models of behavior teach that policy & environmental factors have the broadest & longest-lasting impacts

Research on environment & policy aspects of PA is limited in all countries and absent in most

www.activelivingresearch.org

Need for Research on Environment Need for Research on Environment + PA+ PA

WHO global strategy on diet and physical activity emphasizes environment & policy change

National PA plans call for environment & policy change

Country-specific research is needed to inform policy & planning decisions

66

EnvironmentsDiffer!!!!

Atlanta, USA

Ghent, Belgium

7

Adelaide, Australia

88“Walkable”: Mixed use, connected, dense

99

Not “walkable”Not “walkable”

street connectivity and mixed land usestreet connectivity and mixed land use

10

Funded by NIH/NHLBI 2001-2005

www.nqls.org

11

Neighborhood Walkability and Income Are Related to Physical Activity & BMI

James F. Sallis, Brian E. Saelens, Lawrence D. Frank, Donald J. Slymen, Terry L. Conway, Kelli Cain, & James C. Chapman.

San Diego State University; University of Washington; University of British Columbia

12

Primary Aim Investigate whether people who live in

“walkable” communities are more active, after adjusting for SES, than people who live in less walkable communities.

This study was the methodological template for other studies so comparisons could be made

13

NQLS Neighborhood Categories

Walkability

Soc

ioec

onom

ic S

tatu

s Low High

Hig

hLo

w 4 per region

4 per region 4 per region

4 per region

14

GIS-Based Walkability Index

Net Residential Density

Intersection Density (intersections per acre)

Retail floor area ratio (FAR): ratio of retail building square footage to land area

Land Use Mix: evenness of mix across 4 types of uses.

Walkability = sum of z-scores of components

Calculated for census block groups to select N-hoods

15

Participant Selection & Recruitment Adults aged 20-65 recruited from randomly selected

households in target neighborhoods Recruitment by mail & phone 2100 available for analyses; 30% response rate 48% female; 25% non-white

16

Key Measures Actigraph accelerometer

– Worn for up to 14 days– Outcome is mean daily minutes of MVPA

BMI, based on self-report height & weight NEWS: Neighborhood Environment

Walkability Scale

17

Accelerometer-based MVPA Min/day in Walkability-by-Income Quadrants

28.5

33.4

29.0

35.7

0

5

10

15

20

25

30

35

40

MV

PA

min

ute

s p

er d

ay(M

ea

n *

)

Low Income High Income

Low Walk

High Walk

Walkability: p =.0002

Income: p =.36

Walkability X Income: p =.57

* Adjusted for neighborhood clustering, gender, age, education, ethnicity, # motor vehicles/adult in household, site, marital status, number of people in household, and length of time at current address.

18

Percent Overweight or Obese (BMI>25) in Walkability-by-Income Quadrants

63.156.8

60.4

48.2

0

10

20

30

40

50

60

70

% O

verw

eig

ht

or

Ob

ese

Low Income High Income

Low Walk

High Walk

Walkability: p =.007

Income: p =.081

Walkability X Income: p =.26

* Adjusted for neighborhood clustering, gender, age, education, ethnicity, # motor vehicles/adult in household, site, marital status, number of people in household, and length of time at current address.

19 National Centre for Social Applications of GIS

20

PLACE StudyPhysical activity in Localities And Community Environments

Neville Owen, Adrian Bauman, Graeme Hugo, James F Sallis, Eva Leslie, Jo Salmon, Ester Cerin,Tim Armstrong

National Health and Medical Research Council of Australia, 2002–2004

Primary Aim:to investigate whether people who live in ‘walkable’ communities are more physically active, after adjusting for socio-economic status

21

“Walkable”

density, street connectivity, and mixed land use

22

Not “walkable” mixed land use

23

Transport

1030507090

110130

Low-walkable High-walkable

Wee

kly m

inut

es (m

edia

n)

High SES Low SES

Recreation

1030507090

110130

Low-walkable High-walkable

High SES Low SES

Walking for Transport and Recreation in Low- and High-Walkable Communities*

* Preliminary Analyses: unadjusted for confounders

24

Differences in PA behaviour in Belgian adults living in ‘high

walkable’ versus ‘low walkable’

neighbourhoods. Belgian Environmental Physical Activity

Study (BEPAS)

Ghent University – BELGIUMFaculty of Medicine and Health SciencesDepartment of Movement and Sports Sciences

Delfien Van Dyck

Ilse De Bourdeaudhuij

Greet Cardon

Benedicte Deforche

Preventive Medicine, 2010

25

BEPAS: Accelerometer-based MVPA Min/day in Walkability-by-Income Quadrants

Walkability: β(SE)= .095(.030) p <.001

Income: β(SE)= -.026(.029) p =0.18

Walkability X Income: β(SE)= -.014(.040) p =.36

0

5

10

15

20

25

30

35

40

45

low income high income

32,9530,78

41,13

36,14

CSA

MV

PA m

in/d

ay

low walk

high walk

Adjusted for neighborhood clustering, gender, age, education, working status

26

BEPAS: Transport Walking Min/week in Walkability-by-Income Quadrants

Walkability: β(SE)= .746(.157) p <.001

Income: β(SE)= -.360 (.155) p <.05

Walkability X Income: β(SE)= .027(.220) p =.45

0

20

40

60

80

100

120

140

160

low income high income

50,3

25,27

151,16

83,85

min

/wee

k tr

ansp

ort

wal

king

low walk

high walk

Adjusted for neighborhood clustering, gender, age, education, working status

27

BEPAS: Transport Cycling Min/week in Walkability-by-Income Quadrants

Walkability: β(SE)= .447(.105) p <.001

Income: β(SE)= .029(.102) p =.39

Walkability X Income: β(SE)= -.051(.144) p =.36

0

10

20

30

40

50

60

70

80

90

low income high income

40,5647,25

80,95 83,67

min

/wee

k tr

ansp

ort

cycl

ing

low walk

high walk

Adjusted for neighborhood clustering, gender, age, education, working status

28

BEPAS: Percent Overweight or Obese (BMI>25) in Walkability-by-Income Quadrants

Walkability: β(SE)= -.870(.182) p <.001

Income: β(SE)= -.197(.167) p =.12

Walkability X Income: β(SE)= .910(249) p <.001

Adjusted for neighborhood clustering, gender, age, education, working status

0

5

10

15

20

25

30

35

40

45

low income high income

44,7

39,4

24,8

39,7

% o

verw

eigh

t or

obe

se

low walk

high walk

29

BEPAS: Conclusions

• Living in high walkable neighbourhoods: • 80 min/week more walking for transport• 40 min/week more cycling for transport• 20 min/week more walking for recreation• 35 min/week less motor transport• 50 min/week more MVPA (accelerometer)

– Very similar to US results

30

Relationships between environmental attributes and walking for various

purposes among Japanese adults 　

Shigeru Inoue, MD, PhDDepartment of Preventive Medicine

and Public HealthTokyo Medical University

31

Residential area Shopping street

Nakano area of Tokyo, Japan

32

Methods Study designA Cross-sectional mail survey in

four Japanese cities (Tsukuba, Koganei, Shizuoka and Kagoshima).

TsukubaArea: 284km2

Population: 208,985Density: 736 /km2

Shizuoka1,388km2

710,854512 /km2

Koganei11km2

113,43310,312 /km2

Kagoshima547km2

604,4311,105 /km2

33

Summary of the results from NEWS

• All environmental variables were related to specific types of walking behavior in expected direction.

• Environmental variables related to walking for leisure and walking for daily errands were different

• Relationship between walking and environment was especially strong in women’s walking for daily errands

○ ： Significant relationship

total leisure dailyerrands

total leisure dailyerrands

total leisure dailyerrands

R es idential dens ity ○ ‐ ○ ○ ‐ ○ ○ ‐ ○L and use mix-divers ity ○ ‐ ○ ○ ‐ ‐ ‐ ‐ ○L and use mix-access ○ ‐ ○ ○ ‐ ○ ○ ‐ ○S treet connectivity ‐ ‐ ○ ‐ ‐ ‐ ‐ ‐ ○Walking/cycling facilities ○ ○ ○ ○ ○ ‐ ○ ‐ ○Aesthetics ○ ○ ○ ‐ ○ ‐ ○ ○ ○T raffic safety ○ ○ ‐ ‐ ‐ ‐ ‐ ‐ ‐C rime safety ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ○

All Men Women

34

Built environment correlates of physical activity behaviours in a developing city:

The case of Bogota, Colombia

Olga Lucia Sarmiento and team

Universidade de los Andes

35photo: O.L. Sarmiento

36

Main Results• Walking for transport (30 min/day for at least 5 days/week) was positively associated with:

– Street density (POR 1.71, 95% CI 1.19-2.46)– Street connectivity (POR 2.21, 95% CI 1.40-3.49)– Bus Rapid Transit stations in the neighborhood (POR

1.71, 95% CI 1.19-3.47)

• Biking for transport was positively associated with:– Street density (POR 1.99, 95% CI 1.24-3.19)

• Leisure time physical activity (30 min/day for at least 5 days/week) was positively associated with:

– Park density (POR 2.05, 95%CI 1.13-3.72)– Bus Rapid Transit stations in the neighbohood (POR

1.27, 95% CI 1.07-1.50)

37

Built Environments & Physical Activity: An 11-Country Study

James F. Sallis, USAHeather Bowles, AustraliaAdrian Bauman, Australia Barbara E. Ainsworth, USA

Fiona C. Bull, UKMichael Sjostrom, Sweden

Cora Craig, CanadaEt al.

Am J Prev Med, 2009

38

Rationale• Each country has limited range of variation

in built environment variables, so multi-country studies are needed to understand full impact of environments.

• Most studies of environmental correlates of PA analyze each environment variable separately, so the cumulative effects are not clear.

• Built environment survey measure was added to International PA Prevalence Study

39

PANES: Physical Activity Neighborhood Environment Survey

• Perceived social & physical environment items adapted from published surveys (Kirtland, 2003; Saelens, 2003).

• Standard methods of translation and cultural adaptation.

• Reliability confirmed in Sweden, Nigeria, US

• Attributes rated on 4-point scale, but dichotomized as “agree” vs “disagree”

40

Sample Sizes by Country

• Belgium, 1425• Brazil, 951• Canada, 856• Colombia, 2699• Hong Kong, 1225• Japan, 1001

• Lithuania, 2099• New Zealand, 1298• Norway, 1131• Sweden, 998• United States, 4711

41

Associations Between Individual Environmental Characteristics and HEPA/Minimal Activity Among Respondents who Live in Cities with Population ≥ 30,000

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Single FamilyHouses

Shops NearHome

T ransit StopNear Home

SidewalksP resent

Facilit ies toBicycle

Low Cost RecFacilit ies

Unsafe to Walkdue to Crime

'Agre e ' wi th Environm e ntal C haracte ristic('Disagre e ' i s re fe re nt)

Odd

s R

atio

HE

PA

/Min

imal

Act

ivit

y

42

Dose Response between Number of Environmental Characteristics and HEPA/Minimal Activity

(Pooled City Sample)

0.60

1.00

1.40

1.80

2.20

2.60

3.00

1 2 3 4 5 6

Total Num be r of Environm e ntal C haracte ristics(Ze ro i s re fe re nt)

Odd

s R

atio

HE

PA

/Min

imal

ly A

ctiv

e

43

Started at ICBM in Mainz Germany in 2004 by:Sallis & Kerr, USOwen, AustraliaDeBourdeaudhuij, Belgium

Studies in 3 countries indicated that a common study design and measures were feasible, so the goal was to apply methods to other countries, improving on IPS study

44

Why do we need the IPEN study?• National organizations and WHO

recommend environment & policy changes to increase PA– Need international evidence

• Full variability in environments requires research in multiple countries

• If data are to be pooled, common measures & design are needed

• More detailed measures will provide more specific policy guidance

45

Maximizing within and between country variance (illustration)

USAus

HKJapan

UK

NZ

Sweden

Belgium

Czech

Columbia

Walkability

Wal

kin

g

BUT relationship between walking and walkability may not be linear

46

Main NCI Study Aims

• IPEN study funded by NCI 2009—2013

• Main aims:1. Support countries to collect or enhance data

according to common protocol

2. Transfer data to central dataset

3. Study co-ordination, quality control, & pooled analyses

4. Support the network more widely

5. Advance science through pooled analyses

6. Use results to inform policy internationally

47

IPEN participating countries (so far)

–Australia–Belgium–Brazil–Canada (2)–Denmark–Columbia–Czech Republic

–Hong Kong

–Mexico

–Spain

–Sweden

–UK

–New Zealand

–USA

48

Policy relevance

• If studies show stronger relationship between activity & environment, then policy makers more likely to support & fund environmental change

• Examples from other countries with unique environments can inform built environment changes (without expensive experiments)

• National data needed in each country to convince national policy makers

49

IPEN investigators in Toronto 2010: www.ipenproject.org

50

Conclusions• Environments Seem to Matter Around the World• The international databaseIs expanding rapidly• We need to use research to Drive policy change

www.ipenproject.org