Additional file 3:Table S1. Data extraction from included studies

Study
author / Walking or cycling / Cohort or study name / Population size and recruitment (by gender if reported) / Mean age at baseline / Follow up length in years / Exposure instrument / Other PA domains assessed / Analysis method reported / Exposure and referent categories / Group size (no. of events) / Reported finding
(95% Confidence interval) / Controlling or adjustment variables
Walking
Johnsen (2013) [16] / Walking / Danish Diet, Cancer and Health Cohort
Denmark / 29861 W
27146 M
33% of women and 37% of men invited enrolled / 50-64yr / Median 15.6 years / Self-administered, interviewer checked questionnaire / Occupational PA and leisure time PA / Cox proportional hazard models / No walking vs. median 3.0hrs/wk / Women:
1744 non-active; 27385 active; 8% mortality
Men:
4811 non-active; 24318 active; 15% mortality / W= 0.92
M=1.01 / 0.80-1.06
0.91-1.13 / Age, follow-up time, occupational activity, leisure time activity, smoking, education, self-reported cardiovascular health-related factors, BMI, blood pressure, serum cholesterol
Wang (2013) [26] / Walking / Shanghai Men’s Health Study
China / 61477 men aged 40-74 years at baseline from urban communities in Shanghai / Mean = 55.4 / Mean 5.48 years / Data collected by trained interviewers using a structured physical activity questionnaire / Participants asked to list other PA. Tai Chi and Jogging picked up as common responses / Adjusted hazard ratio from Cox proportional hazards models / No walking
Walking 5-6hrs/wk / -
7864 (414) / 1
0.84 / -
0.73-0.96 / Age; education; income; occupation; alcohol; smoking; energy intake, red meat, fruits and vegetables; daily physical activity; BMI; and history of disease.
Sabia (2012) [17] / Walking / Whitehall II Study
UK / 7456 participants included in total study sample for paper
6323 participants included in analysis / 35-55yr / Mean 9.6 years (SD 2.7) / Questionnaire / Modified version of Minnesota leisure-time physical activity questionnaire / Cox regression with delayed entry and age as timescale to estimate hazard ratios / <3.5hrs/wk
3.5-5.9 hrs/wk
>6hrs/wk / 189 deaths in walking analysis subgroup (n=6323) / 1
0.83
0.81 / -
0.63-1.08
0.61-1.07 / Other physical activity, age, gender, marital status, socioeconomic status, smoking, alcohol consumption, fruit and vegetable consumption, CHD, stroke, diabetes, self-rated health)
Nagai (2011) [28] / Walking / Ohsaki National Health Insurance Cohort Study
Japan / 54996 eligible individuals
52029 responded
27738 in analytic sample
15521 men, 12217 women / 40-70yr / 13 years / Questionnaire / Sports and physical activity (>=3hr/wk or <3h/wk) / Poisson regression model / Time spent walking
Less than 1hr vs. greater than or equal to 1hr / Men:
7363
8158
Women:
6303
5914 / 1
0.90
1
0.95 / -
0.82-0.98
-
0.82-1.10 / Age groups, smoking status, alcohol drinking, BMI, self-rated health, sports and physical activity, and history of hypertension, diabetes mellitus, cancer, liver disease and kidney disease.
Stamatakis (2009) [18] / Walking / Scottish Health Survey Participants
UK / 16144 respondents across 1995, 1998 and 2003 samples
13726 individual included in the all-cause mortality analysis
5896 men, 7325 women / 35-64yr / Mean 8.4 years (SD 3.4)
Total of 114838 person-years of risk / Questionnaire / Housework, heavy DIY activities, and any leisure-time sports and exercises in 1998 and 2003 questionnaires. / Cox proportional hazards model with months as the timescale / Men
None
Medium
High
Women
None
Medium
High / Cases (total)
294 (3157)
87 (1206)
118 (1533)
266 (3675)
88 (1751)
91 (1899) / 1
1.10
0.91
1
0.89
0.92 / -
0.86-1.14
0.73-1.13
-
0.69-1.14
0.72-1.18 / Socio-economic group, age, marital status, BMI category, longstanding illness, self-rated health, family disease history, smoking, other types of physical activities
Besson (2008) [32] / Walking / European Prospective Investigation into Cancer
Germany / 25639 identified from participating GP lists
14903 completed follow up investigation (EPAQ-2) / Mean = 63 / Median 7 years / Self-completed questionnaire
EPAQ2 / PA at home, work and recreation / Associations (HR) from Cox regression analysis / Non-walker
Walking up to 90 min per week
Walking over 90 min per week / 376*
399*
353*
*events only presented / 1
0.96
0.89 / -
0.80-1.15
0.73-1.09 / Age, sex, smoking, alcohol, social class, cancer history, CVD or stroke history, diabetes history, and PA for home, sport, exercise, work
Matthews (2007) [27] / Walking / Shanghai Women’s Health Study
China / All females (81170) in 7 selected communities approached
75221 enrolled (92.7%)
67,143 for analysis / 40-70yr / 5.7 years / Interviews / Exercise, household activities, occupational type / Hazard ratio from Cox proportional hazard / 0 MET-hours/day (ref)
0.1-3.4
3.5-10.0
>10.0 / 20.9% (297)
36.3% (422)
21.2% (197)
21.7% (175) / 1
0.94
0.83
0.86 / -
0.81-1.09
0.69-1.00
0.71-1.05 / Age, marital status, education, household income, smoking, alcohol drinking, pregnancies, contraception, menopausal status, other PA, chronic medical conditions
Schnohr (2007) [19] / Walking / Copenhagen City Heart Study
Denmark / Random sample of 19329 men and women aged 20-93 years
5698 women and 4437 men participated
4104 women and 3204 men in analytic sample / 20-93 yr / 10-13 years / Self-administered questionnaire / Details on physical activity questions not given / Cox proportional hazard regression analysis with age as the underlying timescale and delayed entry accordingly / Duration of walking (hr/day)
Women:
<0.5
0.5-1
1-2
>2
Men:
<0.5
0.5-1
1-2
>2 / Not given separately for duration and intensity variables / Ref
1.00
1.04
0.80
Ref
0.87
0.95
0.89 / -
0.77-1.30
0.80-1.36
0.59-1.10
-
0.68-1.10
0.75-1.21
0.69-1.14 / Age, number of sports activities, BMI, blood pressure, antihypertensive medication, cholesterol, HDL cholesterol, smoking, education, income, alcohol intake, and diabetes
Smith (2007) [22] / Walking / Rancho Bernado California Community Cohort
USA / Cohort of 1664 for this study, of whom 1317 had normal glucose tolerance, and 347 were diabetic
All were Caucasian, middle class, and community dwelling.
42.5% male / Mean = 69.1 / 10 years / Questionnaire / Exercise >= 3 times per week. Further details not given. / Cox proportional hazards models / Daily walking
Non-walker
<1mile walker
>=1mile walker / Not given / 1.00
0.98
0.89 / -
0.76-1.25
0.67-1.18 / Sex, age (continuous), smoking, BMI, average drinks per day, exercise, hypertension, triglycerides, HDL cholesterol, and CHD
Lee (2000) [23] / Walking / Harvard Alumni Health Study
USA / 17835 alumni returned a 1977 questionnaire eligible (all men)
13486 for analytic sample
83.4% non-smokers / Mean = 57.5
SD = 8.9 / Followed from 1977 through 1992 / Questionnaire / Also reported sports or recreational activities / Proportional hazards models / Walking (km/wk)
<5
5-<10
10-<20
>=20 / Deaths
850
514
658
517 / Ref
0.91
0.92
0.84 / -
0.82-1.02
0.83-1.01
0.75-0.94 / Age, Quetelet’s index (BMI), cigarette smoking, alcohol, and early (<65yrs) parental death and light, moderate and vigorous activities
Bath (1998) [20] / Walking / Nottingham Longitudinal Study of Activity and Ageing
UK / 1042 baseline (1985); 690 re-interviewed in 1989; 410 re-interviewed in 1993
“Representative sample” closely resembling England and Wales / 65+ / 12 years / Detailed activity inventory administered by trained interviewer / Outdoor activities, indoor activities, shopping, leisure, strength activities, and joint flexibility activities / Adjusted hazard ratio from Cox regression models / >= 10min walking per day
<10min walking per day / [Total of 647 deaths notified]
559 dead (334 censored) / 1
1.33 / -
1.12-1.58 / Age group, health status, smoking status and weight category as measured at baseline
Hakim (1998) [24] / Walking / Subset of Cooperative Lipoprotein Phenotyping Study
People of Japanese ancestry living in Hawaii / 1379 members before exclusions
707 (retired) men for study sample / Average = 68.9
SD=5.1 / 12 years / Questionnaire / Overall measures of activity with questionnaires similar to those of the Framingham and Puerto Rico heart studies / Proportional hazards regression models / Distance walked (miles/day)
0.0-0.9 vs. 2.1-8.0
0.0-0.9 vs. 1.0-2.0
1.0-2.0 vs. 2.1-8.0 / Deaths (not reported in this analysis type / 1.8
1.5
1.1 / 1.2-2.7
1.1-2.1
0.8-1.7 / Age, total and HDL cholesterol, hypertension, diabetes, alcohol use, the overall physical activity index, and preference for a Japanese diet (% of foods)
Wannamethee (1998) [21] / Walking / British Regional Heart Study
UK / 7735 men aged 40-59 in original cohort
5567 sample for analysis in this study (complete PA data in 1992 questionnaire) / No other baseline characteristics given (reported elsewhere) / 4 years / Questionnaire / Usual patterns of PA / Cox proportional hazards models / Regular walking (min/day)
0
<=20
21-40
41-60
>60 / Deaths (age-adjusted mortality rate)
92 (14.3)
25 (14.0)
52 (12.1)
33 (10.9)
17 (7.5) / 1.00
1.15
1.06
0.97
0.62 / -
0.73-1.79
0.75-1.50
0.65-1.46
0.37-1.05 / Age, smoking, BMI, social class, and self-perception of health, and for other activity types (recreational activity, sporting activity)
La Croix (1996) [25] / Walking / Random sample of men and women aged 65 or older enrolled in Group Health Cooperative of Puget Sound
USA / 6050 contacted
2178 completed questionnaire
1645 in analytic sample / Baseline characteristics given by number of hours walked per week / 4 to 5 years (average 4.2 years) / Questionnaire / Specific activities not listed but frequency and duration of a list of 50 activities was asked / Cox proportional hazards models / Time spent walking
< 1hr/wk
1-4hrs/wk
>4hrs/wk / Deaths (total number)
513 (53)
571 (37)
561 (38) / 1
0.83
0.91 / -
0.53-1.29
0.58-1.42 / Age, sex, functional status, smoking status, BMI, chronic disease score, self-rated health, and alcohol consumption
Cycling
Johnsen (2013) [16] / Cycling / Danish Diet, Cancer and Health Cohort
Denmark / 29861 W
27146 M
33% of women and 37% of men invited enrolled / 50-64yr / Median 15.6 years / Self-administered, interviewer checked questionnaire / Occupational PA and leisure time PA / Cox proportional hazard models / No cycling vs. median 2.0hrs/wk / Women:
8730 non-active; 20339 active; 8% mortality
Men:
9287 non-active; 17289 active; 15% mortality / W= 0.79
M=0.91 / 0.73-0.85
0.85-0.97 / Age, follow-up time, occupational activity, leisure time activity, smoking, education, self-reported cardiovascular health-related factors, BMI, blood pressure, serum cholesterol
Sahlqvist (2013) [29] / Cycling / European Prospective Investigation into Cancer
UK / 25633 individuals (784 missing data)
22450 for analysis (55% female) / 40-79yr
Mean = 58 / Mean 15.3 years (SD 3.3) / PA questionnaire / Household, work leisure and commuting / Hazard ratios from Cox proportional hazard regression model / 0 min/wk
1-59 min/wk
60+ min/wk / 17305 (3,686)
926 (100)
4280 (612) / 1
0.96
0.91 / -
0.73–0.85
0.85-0.97 / Age, sex, education, social class, smoking, medication, history cancer or CVD and other PA
Andersen (2011) [30] / Cycling / Copenhagen City Heart Study
Denmark / 14976 from 20094 invited W = 6510
M = 8466 / 20-93yr / 16 years / Questions about commuter cycling / Other exercise, walking, jogging, sports and other PA / Relative risks from Cox proportional hazard regression analysis / No cycling
0-3 hr/wk
3-7 hr/wk
7+ hr/wk / 8167
2473
2316
2020
3787 deaths
total / 1
0.78
0.76
0.70 / -
0.69-0.88
0.68-0.85
0.62-0.78 / Age, sex, BMI, smoking, education, BP, cholesterol and other PA
Schnohr (2010) [31] / Cycling / Copenhagen City Heart Study
Denmark / 5106 cyclists from 10135 respondents to 3rd follow up from 16563 invited (61.2%)
Original 19698 random sample
M = 2398
W= 2708 / 21-90yr / Average 18 years / Self-administered questionnaire checked by staff / Other exercise, walking, jogging, sports and other PA / Relative risks from Cox proportional hazard regression analysis / 0-0.5 hr/day
Slow
Average
Fast
0.5-1 hr/day
Slow
Average
Fast
>1hr/day
Slow
Average
Fast / 120 (52)
1125 (252)
223 (19)
95 (43)
1378 (347)
518 (36)
55 (32)
1117 (330)
475 (61) / 1
0.67
0.54
0.87
0.70
0.44
0.85
0.71
0.68 / -
0.49-0.92
0.31-0.94
0.57-1.33
0.51-0.95
0.28-0.69
0.53-0.1.35
0.52-0.97
0.46-1.01 / Age, gender, sports, blood pressure and medication, HDL cholesterol, smoking, income, alcohol
Besson (2008) [32] / Cycling / European Prospective Investigation into Cancer
Germany / 25639 identified from participating GP lists
14903 completed follow up investigation (EPAQ-2) / Mean = 63 / Median 7 year / Self-completed questionnaire
EPAQ2 / PA at home, work and recreation / Associations (HR) from Cox regression analysis / Non-cyclist
Cycling up to 30 min per week
Cycling over 30 min per week / 990*
77*
61*
*events only presented / 1
1.02
1.01 / -
0.77-1.35
0.76-1.36 / Age, sex, smoking, alcohol, social class, cancer history, CVD or stroke history, diabetes history, and PA for home, sport, exercise, work and walking
Matthews (2007) [27] / Cycling / Shanghai Women’s Health Study
China / All females (81170) in 7 selected communities approached
75221 enrolled (92.7%)
67,143 for analysis / 40-70yr / 5.7 years / Interviews / Exercise, household activities, occupational type / Hazard ratio from Cox proportional hazard / 0 MET-hours/day (ref)
0.1-3.4
3.5+ / 50693 (1,001)
12959 (73)
3559 (17) / 1
0.79
0.66 / -
0.61-1.01
0.40-1.07 / Age, marital status, education, household income, smoking, alcohol drinking, pregnancies, oral contraception, menopausal status, other PA, chronic medical conditions
Andersen (2000) [33] / Cycling / Glostrup Population Study and Copenhagen Male Study
Denmark / 6954 asked if used bike for transport (6171 men and 783 women)
Total cohort 17265 men and 13375 women / M: age = 49.7 (11.2)
W: age = 50.4 (12.7) / M: 15.1 years (SD 6.6)
W: 13.8 years (SD 5.0) / Question Use a bike to work (yes/no) / PA for work and leisure time also assessed / Relative risks from Cox proportional hazards / 0 min per week
3 hours per week / 2291 deaths
Group sizes not reported / 1
0.72 / -
0.57-0.91 / Age, sex, educational level, LTPA, BMI, blood lipid levels, smoking and blood pressure