Articles

Eect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy
I-Min Lee, Eric J Shiroma, Felipe Lobelo, Pekka Puska, Steven N Blair, Peter T Katzmarzyk, for the Lancet Physical Activity Series Working Group*

Summary
Background Strong evidence shows that physical inactivity increases the risk of many adverse health conditions, including major non-communicable diseases such as coronary heart disease, type 2 diabetes, and breast and colon cancers, and shortens life expectancy. Because much of the worlds population is inactive, this link presents a major public health issue. We aimed to quantify the eect of physical inactivity on these major non-communicable diseases by estimating how much disease could be averted if inactive people were to become active and to estimate gain in life expectancy at the population level. Methods For our analysis of burden of disease, we calculated population attributable fractions (PAFs) associated with physical inactivity using conservative assumptions for each of the major non-communicable diseases, by country, to estimate how much disease could be averted if physical inactivity were eliminated. We used life-table analysis to estimate gains in life expectancy of the population. Findings Worldwide, we estimate that physical inactivity causes 6% (ranging from 32% in southeast Asia to 78% in the eastern Mediterranean region) of the burden of disease from coronary heart disease, 7% (3996) of type 2 diabetes, 10% (56141) of breast cancer, and 10% (57138) of colon cancer. Inactivity causes 9% (range 51125) of premature mortality, or more than 53 million of the 57 million deaths that occurred worldwide in 2008. If inactivity were not eliminated, but decreased instead by 10% or 25%, more than 533 000 and more than 13 million deaths, respectively, could be averted every year. We estimated that elimination of physical inactivity would increase the life expectancy of the worlds population by 068 (range 041095) years. Interpretation Physical inactivity has a major health eect worldwide. Decrease in or removal of this unhealthy behaviour could improve health substantially. Funding None.
Lancet 2012; 380: 21929 Published Online July 18, 2012 http://dx.doi.org/10.1016/ S0140-6736(12)61031-9 See Comment page 192 *Members listed at end of paper Division of Preventive Medicine, Brigham and Womens Hospital, Harvard Medical School, Boston, MA, USA (I-M Lee ScD); Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA (E J Shiroma MSc); Global Health Promotion Oce, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA (F Lobelo MD); National Institute for Health and Welfare, Helsinki, Finland (P Puska MD); Department of Exercise Science and Department of Epidemiology/ Biostatistics, Arnold School of Public Health, University of South Carolina, Columbia, SC, USA (S N Blair PED); and Pennington Biomedical Research Center, Baton Rouge, LA, USA (P T Katzmarzyk PhD) Correspondence to: Dr I-Min Lee, Brigham and Womens Hospital, Harvard Medical School, Boston, MA 02215, USA ilee@rics.bwh.harvard.edu

Introduction
Ancient physiciansincluding those from China in 2600 BC and Hippocrates around 400 BCbelieved in the value of physical activity for health. By the 20th century, however, a diametrically opposite viewthat exercise was dangerousprevailed instead.1 During the early 20th century, complete bed rest was prescribed for patients with acute myocardial infarction. And, at the time of the 100th boat race between the Universities of Oxford and Cambridge, UK, in 1954, the senior health ocer of Cambridge University undertook a study to investigate the alleged dangers of exercise by comparing university sportsmen with intellectuals.1 One of the pioneers whose work helped to change that tide of popular opinion was Jerry Morris, who undertook the rst rigorous, epidemiological studies investigating physical inactivity and chronic disease risk, published in 1953.2 Since then, much evidence has clearly documented the many health benets of physical activity (panel 1).35 Despite this knowledge, a large proportion of the worlds population remains physically inactive. To quantify the eect of physical inactivity on the worlds major noncommunicable diseases, we estimated how much of
www.thelancet.com Vol 380 July 21, 2012

these diseases could be averted in the population if inactive people were to become active, as well as how much gain in life expectancy could occur at the population level. We focus on the major non-communicable diseases emphasised by the UN as threats to global health:6 coronary heart disease; cancer, specically breast and colon cancers, which are convincingly related to physical inactivity; and type 2 diabetes.

Methods
Population attributable fraction
The population attributable fraction (PAF) is a measure used by epidemiologists to estimate the eect of a risk factor on disease incidence in a population.7,8 It estimates the proportion of new cases that would not occur, absent a particular risk factor. Thus, it provides policy makers with useful quantitative estimates of the potential eect of interventions to reduce or eradicate the risk factor. PAF is related to prevalence of the risk factor and its associated relative risk (RR). At least two formulae are available to calculate PAF (panel 2). Formula 1 provides an unbiased estimate when there is no confounding of the relation between the risk factor and disease, and
219

Articles

Panel 1: Health benets of physical activity in adults35 Strong evidence of reduced rates of: All-cause mortality Coronary heart disease High blood pressure Stroke Metabolic syndrome Type 2 diabetes Breast cancer Colon cancer Depression Falling Strong evidence of: Increased cardiorespiratory and muscular tness Healthier body mass and composition Improved bone health Increased functional health Improved cognitive function

prevalence estimates of inactivity for cases and adjusted RRs to estimate PAF using formula 2.

Estimation of prevalence of physical inactivity

We dened physical inactivity to be an activity level insucient to meet present recommendations.5 WHO obtains data, by country, for the prevalence of physical inactivity in the population using two similar standardised questionnaires; the latest data are for 2008.9 For calculation of PAFs with RRs adjusted for confounding factors, the prevalence of physical inactivity at baseline in cases of the outcome of interest was needed. These data proved dicult to obtain for countries outside North America and Europe. Further, data for prevalence of inactivity depended on the instrument used for assessment and varied according to whether a study assessed physical activity during leisure only (most commonly), or also included activities in occupation, transportation, or home-based activities. Thus, to estimate the prevalence of inactivity in cases, we contacted several large cohort studies throughout the world using input from the Lancet Physical Activity Series Working Group, attempting particularly to gather data outside North America and Europe. For each study, we obtained the prevalence of physical inactivity in all participants at baseline, and in those eventually developing coronary heart disease, type 2 diabetes, and breast and colon cancer and those who died (appendix). For each outcome, we calculated an adjustment factor, representing the added extent to which physical inactivity occurred in cases compared with the overall population of the cohort study. For example, in the Shanghai Womens Health Study (appendix), the prevalence of inactivity in all women at baseline was 454% versus 516% in women who died, yielding an adjustment factor of 114 (516 / 454 = 114). For each outcome, we calculated the adjustment factor in every study, and averaged this factor across studies. We applied the average adjustment factor to the prevalence of physical inactivity, by country, to estimate the prevalence of inactivity in cases of coronary heart disease, type 2 diabetes, breast and colon cancer, and death from any cause.

Panel 2: Formulae for calculation of population attributable fraction (PAF) Formula 1, using unadjusted relative risk:

See Online for appendix

PAF(%) =

Pe(RRunadj 1) 100 Pe(RRunadj 1) + 1

Where Pe is the proportion of inactive people in the source population, and RRunadj is the relative risk of disease, comparing inactive with active people, unadjusted for confounding factors. Formula 2, using adjusted relative risk:

PAF(%) =

Pd(RRadj 1) 100 RRadj

Where Pd is the proportion of inactive people among cases, and RRadj is the relative risk of disease, comparing inactive with active people, adjusted for confounding factors.

Estimation of RRs associated with physical inactivity

We searched electronic databases (Medline and Embase) using keywords related to physical activity (physical activity, motor activity, energy expenditure, walking, and exercise) and the outcomes of interest (breast cancer, breast carcinoma, colon cancer, colorectal cancer, colon carcinoma, colorectal carcinoma, diabetes, type 2 diabetes, all-cause mortality, mortality, cardiovascular disease, coronary heart disease, and heart disease) for peer-reviewed reviews of adults published in English, selecting the most recent one as of June 30, 2011. For all outcomes apart from breast cancer, published meta-analyses of the pooled RR were available.1013 For breast cancer, no comprehensive
www.thelancet.com Vol 380 July 21, 2012

requires knowledge of the prevalence of the risk factor in the population and the RR not to be adjusted for confounders (crude RR). Formula 2 is preferred when there is confounding;8 it requires knowledge of the prevalence of the risk factor in people eventually developing the disease (cases) and the adjusted RR. Because some confounders (eg, hypertension in coronary heart disease, overweight in diabetes) are exacerbated by inactivity, formula 2 might overadjust, whereas formula 1 can add perspective. Thus, we sought prevalence estimates of inactivity for the whole population and unadjusted RRs to estimate PAF using formula 1, and
220

Articles

meta-analysis was found (one of only case-control studies is available14), so we selected the most recent qualitative review15 and did a meta-analysis of their primary studies. All the meta-analyses calculated only pooled RRs adjusted for potential confounders (generally selecting maximally adjusted RRs from individual studies); no pooled estimates of crude RRs were reported. Thus, we obtained the primary papers to identify the crude RRs. For most papers, this information was not reported; for several, data were provided that allowed its calculation. When the crude RR was unavailable or could not be calculated, the age-adjusted RR was often available. Thus, to obtain a pooled estimate of the crude RRs, we used either crude RRs or age-adjusted RRs, calling this value the unadjusted RR. This method enabled use of data from a larger number of studies, and a closer parallel between studies used to calculate the pooled unadjusted and adjusted RRs. In sensitivity analyses that compared results using only crude RRs with those using both crude and age-adjusted RRs, estimates were generally similar; thus, bias using unadjusted instead of crude RRs is unlikely. We used simple, random-eects meta-regression to account for heterogeneity across studies, using MIX 2.0.

eliminated, we assumed that the age-specic death rates for a country would be decreased by an amount equal to this PAF (calculated using the adjusted RR) if inactivity were eliminated. Studies of physical activity and all-cause mortality have mainly been in people aged 40 years and older, with few data available for those aged 80 years and older, which also suggest benet.3 Thus, we conservatively decreased age-specic death rates by the PAF only for ages 4079 years, and calculated the revised life expectancy from birth, by country. In a sensitivity analysis, we did parallel analyses that decreased agespecic death rates for all ages 40 years and older.

Role of the funding source

No funding organisation had any role in the writing of the report or the decision to submit for publication. The corresponding author had full access to all data in the study and nal responsibility for the decision to submit for publication.

Results
We estimated the prevalence of physical inactivity in cases of the outcomes studied, by country, using adjustment factors of 120 (SE 003) for coronary heart disease, 123 (005) for type 2 diabetes, 105 (009) for breast cancer, 122 (008) for colon cancer, and 122 (007) for all-cause mortality. The highest prevalence was noted in people who went on to develop type 2 diabetes, followed by those who died and those who developed colon cancer, coronary heart disease, and breast cancer (table 1, appendix). Table 1 summarises the RRs associated with physical inactivity, unadjusted and adjusted for confounders, for the outcomes studied. Sattelmair and colleagues10 investigated the dose-response relation between leisure-time energy expenditure and incidence of coronary heart disease. The pooled RR associated with energy expenditure that fullled present recommendations compared
Breast cancer* Colon cancer All-cause mortality

Calculation of PAFs and gains in life expectancy

We calculated PAFs for each outcome, by country, and used Monte Carlo simulation techniques (10 000 simulations) to estimate 95% CIs. We assumed normal distributions for physical inactivity prevalence and the log of the RRs. We used life-table analysis to estimate gains in life expectancy that could be expected if physical inactivity were eliminated, using life tables published by WHO that provide age-specic death rates, by country; the latest data are for 2009.16 Since the country-specic PAF for all-cause mortality estimates how much premature mortality can be removed from the population if physical inactivity were
Coronary heart disease Prevalence of inactivity in population (%) Prevalence of inactivity in people eventually developing the outcome (%) RR, unadjusted RR, adjusted PAF with unadjusted RR (%) PAF with adjusted RR (%) 352% (223405)

Type 2 diabetes

352% (223405) 388% (233443)

352% (223405) 352% (223405) 429% (234571) 429% (234571) 138 (131145) 132 (123139) 118% (68151) 104% (57138) 147 (138157) 128 (121136) 142% (83180) 94% (51125)

422% (230562) 432% (236576) 407% (225567) 133 (118149) 116 (104130) 104% (72134) 58% (3278) 163 (127211) 120 (110133) 72% (3996) 134 (125143) 133 (126142) 101% (56141)

181% (108228) 116% (68155)

Physical inactivity was dened as insucient physical activity to meet present recommendations. RR=relative risk. PAF=population attributable fraction. *Women only. Data are overall median (range of medians for WHO regions); details of country-specic values for the population are available from reference 9; country-specic values for people eventually developing these diseases are provided in the appendix. Data are RR (95% CI); for details of calculation of unadjusted RRs, see appendix; the unadjusted RRs pooled both crude and age-adjusted RRs, since the crude RR was often unavailable; the adjusted RR of coronary heart disease was obtained from Sattelmair and colleagues,10 for type 2 diabetes from Jeon and colleagues,11 for breast cancer and all-cause mortality see appendix, and for colon cancer from Wolin and co-workers.12 Data are overall median (range of medians for WHO regions); details of country-specic values calculated with unadjusted RRs are provided in appendix; country-specic values calculated with adjusted RRs are shown in table 2.

Table 1: Summary of estimates of the prevalence of physical inactivity, RRs, and PAFs for coronary heart disease, type 2 diabetes, breast cancer, colon cancer, and all-cause mortality associated with physical inactivity

www.thelancet.com Vol 380 July 21, 2012

221

Articles

Coronary heart disease Type 2 diabetes Africa Algeria Benin Botswana Burkina Faso Cameroon Cape Verde Chad Comoros Congo (Brazzaville) Cte dIvoire Democratic Republic of the Congo Eritrea Ethiopia Gabon Ghana Guinea Kenya Madagascar Malawi Mali Mauritania Mauritius Mozambique Namibia Niger So Tom and Prncipe Senegal Seychelles Sierra Leone South Africa Swaziland The Gambia Zambia Zimbabwe Median for region Latin America and Caribbean Argentina Barbados Brazil Colombia Dominica Dominican Republic Ecuador Guatemala Jamaica Mexico Paraguay Saint Kitts and Nevis Uruguay Median for region 113% (31 to 210) 78% (28 to 130) 82% (15 to 164) 73% (09 to 156) 40% (17 to 166) 99% (17 to 166) 71% (10 to 146) 27% (03 to 65) 79% (14 to 161) 62% (08 to 132) 68% (10 to 142) 63% (23 to 105) 56% (21 to 94) 71% 140% (48 to 247) 96% (48 to 150) 101% (24 to 189) 90% (13 to 182) 50% (27 to 199) 123% (27 to 199) 87% (15 to 172) 33% (05 to 78) 98% (22 to 185) 77% (10 to 158) 85% (14 to 168) 79% (39 to 122) 70% (35 to 109) 87% 67% (24 to 112) 15% (05 to 25) 58% (21 to 97) 26% (02 to 61) 67% (10 to 139) 34% (13 to 57) 41% (00 to 93) 14% (03 to 36) 80% (14 to 164) 54% (06 to 119) 75% (28 to 124) 67% (24 to 112) 32% (01 to 76) 61% (05 to 133) 29% (11 to 48) 20% (02 to 50) 27% (04 to 68) 39% (14 to 64) 17% (06 to 28) 35% (01 to 80) 73% (15 to 143) 64% (07 to 135) 12% (04 to 20) 97% (23 to 189) 49% (18 to 80) 31% (11 to 52) 38% (00 to 88) 37% (13 to 61) 33% (11 to 54) 87% (31 to 145) 114% (32 to 215) 41% (15 to 67) 29% (03 to 70) 39% (00 to 90) 39% 83% (42 to 129) 19% (09 to 29) 72% (36 to 113) 32% (03 to 72) 83% (18 to 162) 42% (21 to 66) 50% (02 to 110) 17% (04 to 42) 100% (22 to 191) 67% (08 to 139) 93% (47 to 145) 83% (42 to 128) 40% (01 to 89) 75% (06 to 155) 36% (18 to 57) 25% (02 to 56) 34% (06 to 79) 48% (24 to 75) 21% (10 to 33) 43% (02 to 95) 90% (22 to 169) 79% (13 to 159) 15% (07 to 23) 120% (35 to 222) 60% (30 to 93) 39% (20 to 60) 47% (01 to 101) 46% (23 to 72) 41% (20 to 63) 107% (54 to 168) 142% (47 to 251) 50% (25 to 78) 35% (05 to 82) 49% (01 to 107) 48%

Breast cancer

Colon cancer

All-cause mortality

128% (59 to 200) 29% (13 to 46) 115% (54 to 180) 43% (10 to 95) 126% (18 to 238) 77% (35 to 118) 68% (08 to 144) 28% (09 to 64) 138% (28 to 249) 96% (07 to 190) 136% (65 to 211) 143% (67 to 224) 58% (09 to 127) 121% (07 to 236) 54% (25 to 84) 47% (07 to 102) 47% (11 to 106) 74% (35 to 115) 34% (16 to 53) 62% (07 to 134) 124% (33 to 218) 102% (08 to 200) 19% (09 to 30) 170% (46 to 295) 89% (42 to 141) 69% (31 to 106) 67% (07 to 142) 58% (26 to 92) 62% (29 to 95) 147% (67 to 231) 188% (59 to 324) 75% (35 to 117) 50% (12 to 113) 67% (08 to 143) 71% 185% (59 to 317) 145% (68 to 227) 134% (23 to 247) 125% (12 to 239) 90% (42 to 261) 164% (42 to 261) 126% (15 to 237) 44% (12 to 101) 134% (26 to 248) 100% (08 to 198) 109% (10 to 212) 125% (59 to 195) 105% (48 to 164) 125%

120% (68 to 172) 27% (15 to 39) 104% (59 to 151) 46% (08 to 99) 120% (23 to 221) 61% (34 to 89) 73% (03 to 151) 25% (08 to 57) 144% (30 to 262) 97% (07 to 185) 134% (73 to 194) 120% (67 to 171) 57% (06 to 121) 108% (06 to 215) 52% (29 to 75) 36% (06 to 77) 49% (10 to 109) 69% (39 to 100) 30% (17 to 44) 62% (07 to 131) 130% (30 to 230) 114% (12 to 216) 21% (11 to 31) 173% (50 to 299) 87% (48 to 126) 56% (31 to 82) 68% (03 to 137) 66% (37 to 96) 59% (33 to 85) 155% (88 to 224) 204% (73 to 337) 73% (41 to 105) 51% (10 to 114) 70% (04 to 146) 70% 202% (68 to 335) 139% (76 to 201) 146% (29 to 261) 130% (13 to 248) 72% (42 to 275) 178% (42 to 275) 126% (21 to 235) 48% (11 to 106) 141% (25 to 255) 112% (11 to 213) 122% (16 to 229) 113% (64 to 165) 101% (56 to 146) 126%

108% (86 to 131) 24% (19 to 30) 94% (75 to 114) 41% (01 to 86) 109% (34 to 186) 55% (43 to 68) 65% (04 to 129) 22% (05 to 50) 130% (44 to 220) 88% (20 to 161) 121% (96 to 146) 108% (86 to 130) 52% (02 to 105) 98% (18 to 182) 47% (37 to 57) 32% (01 to 67) 44% (03 to 93) 62% (49 to 75) 27% (21 to 33) 56% (02 to 111) 117% (45 to 192) 103% (26 to 182) 19% (15 to 24) 156% (70 to 248) 78% (62 to 95) 51% (40 to 62) 62% (05 to 120) 60% (47 to 73) 53% (42 to 65) 140% (111 to 169) 184% (94 to 277) 65% (52 to 79) 46% (02 to 96) 64% (06 to 124) 63% 182% (95 to 277) 125% (99 to 151) 132% (48 to 217) 117% (28 to 210) 65% (53 to 230) 160% (53 to 230) 114% (35 to 197) 43% (03 to 92) 128% (46 to 216) 101% (25 to 182) 110% (30 to 191) 102% (81 to 124) 91% (72 to 111) 114% (Continues on next page)

222

www.thelancet.com Vol 380 July 21, 2012

Articles

Coronary heart disease Type 2 diabetes (Continued from previous page) North America Canada USA Median for region Eastern Mediterranean Iran Iraq Kuwait Lebanon Libya Pakistan Saudi Arabia Tunisia United Arab Emirates Median for region Europe Austria Belgium Bosnia and Herzegovina Bulgaria Croatia Cyprus Czech Republic Denmark Estonia Finland France Georgia Germany Greece Hungary Ireland Italy Kazakhstan Latvia Lithuania Luxembourg Malta Netherlands Norway Poland Portugal Romania Russia Serbia Slovakia Slovenia Spain Sweden Turkey 58% (06 to 121) 71% (12 to 147) 56% (04 to 121) 44% (01 to 99) 39% (00 to 89) 92% (20 to 179) 41% (03 to 93) 58% (06 to 124) 29% (02 to 69) 63% (08 to 132) 54% (19 to 89) 37% (13 to 62) 46% (01 to 104) 26% (02 to 62) 43% (01 to 95) 88% (20 to 174) 91% (19 to 180) 52% (04 to 115) 53% (03 to 115) 37% (13 to 63) 79% (13 to 160) 119% (33 to 223) 30% (01 to 71) 73% (12 to 153) 46% (02 to 104) 84% (17 to 173) 64% (07 to 136) 34% (01 to 81) 113% (31 to 209) 37% (01 to 85) 50% (03 to 111) 83% (17 to 167) 73% (12 to 151) 93% (21 to 183) 71% (10 to 145) 88% (19 to 170) 69% (08 to 141) 55% (01 to 117) 48% (01 to 106) 114% (30 to 210) 51% (02 to 109) 72% (08 to 147) 35% (04 to 81) 78% (13 to 156) 67% (33 to 103) 46% (23 to 71) 57% (02 to 124) 32% (04 to 74) 53% (00 to 114) 109% (29 to 202) 112% (30 to 209) 65% (06 to 134) 66% (05 to 136) 46% (23 to 72) 98% (19 to 191) 147% (53 to 260) 37% (03 to 83) 91% (17 to 179) 57% (03 to 120) 105% (26 to 202) 79% (12 to 158) 43% (02 to 95) 140% (47 to 246) 46% (00 to 99) 62% (04 to 129) 103% (25 to 195) 91% (19 to 174) 115% (32 to 210) 61% (22 to 102) 97% (35 to 158) 107% (39 to 177) 78% (29 to 129) 76% (28 to 125) 67% (10 to 140) 114% (42 to 188) 59% (07 to 126) 103% (26 to 199) 78% 76% (38 to 118) 120% (60 to 187) 132% (66 to 207) 96% (47 to 149) 94% (47 to 147) 83% (14 to 163) 141% (71 to 219) 74% (10 to 150) 128% (38 to 236) 96% 56% (05 to 122) 67% (25 to 111) 62% 70% (08 to 144) 83% (42 to 129) 76%

Breast cancer

Colon cancer

All-cause mortality

92% (02 to 186) 124% (58 to 192) 108% 122% (58 to 189) 141% (66 to 219) 188% (88 to 292) 109% (51 to 169) 142% (66 to 219) 125% (19 to 236) 199% (92 to 306) 105% (10 to 202) 180% (59 to 309) 141% 102% (08 to 201) 117% (20 to 215) 96% (04 to 192) 75% (04 to 156) 55% (06 to 116) 163% (42 to 288) 58% (04 to 123) 92% (02 to 184) 49% (09 to 109) 91% (02 to 184) 97% (46 to 151) 61% (28 to 95) 74% (03 to 155) 38% (11 to 87) 67% (02 to 137) 152% (35 to 273) 156% (42 to 280) 81% (02 to 163) 94% (01 to 187) 65% (29 to 101) 119% (12 to 230) 191% (60 to 324) 40% (11 to 92) 117% (15 to 222) 82% (02 to 170) 142% (31 to 259) 120% (12 to 232) 49% (09 to 107) 191% (69 to 323) 55% (09 to 121) 88% (02 to 175) 138% (26 to 255) 115% (16 to 217) 163% (40 to 289)

100% (07 to 195) 120% (67 to 174) 110% 109% (62 to 158) 173% (97 to 251) 191% (106 to 277) 138% (76 to 200) 136% (74 to 195) 120% (16 to 220) 204% (113 to 293) 106% (10 to 203) 185% (52 to 317) 138% 103% (11 to 194) 126% (21 to 231) 99% (07 to 194) 79% (01 to 161) 70% (05 to 143) 164% (43 to 288) 74% (00 to 148) 104% (10 to 199) 51% (09 to 112) 112% (11 to 212) 96% (54 to 138) 66% (36 to 96) 83% (03 to 167) 46% (09 to 101) 77% (02 to 156) 157% (41 to 276) 162% (39 to 281) 93% (05 to 184) 95% (02 to 186) 67% (37 to 98) 141% (26 to 258) 213% (75 to 353) 54% (06 to 114) 131% (22 to 241) 82% (01 to 164) 151% (36 to 270) 114% (13 to 219) 62% (06 to 131) 202% (70 to 335) 66% (03 to 137) 89% (00 to 176) 149% (31 to 266) 131% (21 to 241) 166% (42 to 290)

91% (18 to 166) 108% (86 to 131) 99% 99% (79 to 119) 156% (125 to 188) 172% (137 to 208) 125% (99 to 151) 122% (97 to 148) 108% (32 to 188) 184% (147 to 221) 96% (24 to 171) 167% (73 to 262) 125% 93% (23 to 165) 114% (37 to 195) 90% (18 to 165) 72% (09 to 136) 63% (05 to 124) 148% (62 to 241) 67% (09 to 126) 94% (22 to 171) 46% (02 to 94) 101% (26 to 178) 87% (69 to 105) 60% (47 to 72) 75% (09 to 145) 42% (02 to 87) 69% (07 to 132) 142% (60 to 229) 146% (58 to 237) 84% (15 to 156) 85% (14 to 158) 60% (47 to 75) 127% (42 to 218) 192% (98 to 289) 49% (00 to 98) 118% (35 to 202) 74% (10 to 139) 136% (52 to 226) 103% (26 to 184) 56% (02 to 110) 182% (94 to 276) 59% (05 to 115) 80% (12 to 152) 134% (49 to 224) 118% (38 to 202) 150% (62 to 239) (Continues on next page)

www.thelancet.com Vol 380 July 21, 2012

223

Articles

Coronary heart disease Type 2 diabetes (Continued from previous page) Ukraine UK Median for region Southeast Asia Bangladesh Bhutan Burma India Indonesia Maldives Nepal Sri Lanka Thailand Median for region Western Pacic Australia Cambodia China Cook Islands Federated States of Micronesia Japan Kiribati Laos Malaysia Marshall Islands Mongolia Nauru New Zealand Papua New Guinea Philippines Samoa Solomon Islands Tonga Vietnam Median for region Overall Median 58% 72% 63% (08 to 131) 19% (07 to 31) 51% (19 to 85) 119% (44 to 197) 110% (41 to 183) 100% (24 to 190) 77% (27 to 128) 31% (02 to 75) 102% (38 to 169) 82% (30 to 137) 16% (06 to 26) 77% (29 to 129) 79% (29 to 131) 32% (12 to 53) 39% (01 to 90) 85% (31 to 141) 72% (27 to 119) 69% (25 to 115) 25% (02 to 62) 72% 78% (10 to 154) 23% (11 to 36) 64% (32 to 98) 148% (75 to 234) 136% (68 to 211) 123% (36 to 225) 96% (48 to 150) 39% (04 to 87) 126% (63 to 196) 102% (51 to 157) 19% (10 to 30) 95% (48 to 148) 98% (49 to 152) 40% (19 to 62) 49% (01 to 107) 105% (53 to 164) 90% (45 to 140) 86% (42 to 134) 31% (04 to 73) 90% 08% (03 to 13) 87% (18 to 173) 21% (08 to 35) 26% (10 to 42) 49% (18 to 82) 65% (07 to 139) 26% (03 to 62) 43% (16 to 71) 32% (12 to 52) 32% 10% (05 to 15) 107% (26 to 203) 26% (13 to 41) 32% (16 to 50) 61% (30 to 95) 80% (10 to 161) 32% (04 to 74) 53% (27 to 83) 39% (20 to 61) 39% 31% (02 to 73) 105% (40 to 173) 55% 38% (04 to 85) 130% (64 to 202) 68%

Breast cancer

Colon cancer

All-cause mortality

43% (08 to 93) 179% (85 to 278) 93% 17% (08 to 27) 166% (42 to 290) 39% (19 to 61) 48% (23 to 74) 73% (33 to 115) 108% (05 to 214) 44% (10 to 101) 87% (42 to 135) 56% (25 to 87) 56% 104% (09 to 202) 29% (13 to 45) 84% (40 to 130) 191% (90 to 298) 194% (91 to 305) 161% (39 to 291) 143% (69 to 223) 55% (10 to 121) 171% (80 to 266) 145% (69 to 225) 25% (11 to 39) 130% (62 to 203) 131% (62 to 203) 56% (25 to 88) 68% (07 to 147) 170% (81 to 267) 129% (61 to 202) 135% (64 to 211) 41% (11 to 94) 130% 101%

54% (08 to 116) 187% (105 to 271) 98% 14% (08 to 20) 155% (38 to 276) 38% (21 to 54) 46% (26 to 66) 88% (49 to 128) 115% (10 to 221) 46% (08 to 102) 77% (42 to 111) 57% (32 to 82) 57% 112% (14 to 212) 33% (18 to 48) 92% (52 to 131) 213% (118 to 309) 196% (110 to 286) 178% (50 to 309) 138% (77 to 200) 56% (08 to 120) 182% (102 to 265) 147% (81 to 212) 28% (15 to 40) 138% (76 to 198) 141% (79 to 203) 57% (31 to 83) 70% (05 to 144) 151% (85 to 218) 129% (71 to 188) 124% (68 to 180) 45% (08 to 98) 129% 104%

49% (00 to 101) 169% (136 to 203) 88% 13% (10 to 15) 140% (53 to 228) 34% (27 to 41) 42% (33 to 50) 80% (63 to 97) 104% (26 to 185) 41% (02 to 87) 69% (55 to 83) 51% (41 to 62) 51% 101% (28 to 180) 30% (24 to 36) 83% (66 to 100) 192% (153 to 231) 177% (141 to 213) 161% (72 to 254) 125% (98 to 151) 50% (01 to 103) 164% (130 to 197) 132% (106 to 160) 25% (20 to 30) 124% (98 to 149) 127% (102 to 154) 52% (41 to 63) 63% (04 to 124) 136% (109 to 164) 117% (92 to 140) 112% (88 to 136) 41% (02 to 84) 117% 94%

Data are PAF (95% CI). PAF=population attributable fraction. *PAFs calculated with unadjusted relative risks are shown in the appendix.

Table 2: Estimated PAFs, calculated with adjusted relative risks,* for coronary heart disease, type 2 diabetes, breast cancer, colon cancer, and all-cause mortality associated with physical inactivity, by WHO region and country

with no leisure activity, adjusted for potential confounders, was 086 (95% CI 077096). With increasing energy expenditure, coronary heart disease incidence fell further, in a curvilinear fashion. For this report, we used the RR corresponding to an activity level that met minimum present recommendations (086). Taking the inverse of this number to obtain the adjusted RR for physical inactivity yielded 116 (95% CI 104130). Although these data are from only North America and Europe (ie, studies with sucient information to investigate dose response),
224

they are congruent with ndings from the INTERHEART study17 undertaken in 52 countries worldwide, in which the adjusted odds ratio for myocardial infarction associated with physical inactivity was identical (116, 95% CI 103132). We did a parallel meta-analysis to obtain the corresponding pooled unadjusted RR (ie, pooling crude and age-adjusted RRs), which was 133 (95% CI 118149; appendix). Crude RRs were available for only four studies and pooling of these yielded a value of 154 (95% CI 125192); thus, the pooled unadjusted RR is conservative.
www.thelancet.com Vol 380 July 21, 2012

Articles

Gain in life expectancy (years) Africa Algeria Benin Botswana Burkina Faso Cameroon Cape Verde Chad Comoros Congo (Brazzaville) Cte dIvoire Democratic Republic of the Congo Eritrea Ethiopia Gabon Ghana Guinea Kenya Madagascar Malawi Mali Mauritania Mauritius Mozambique Namibia Niger So Tom and Prncipe Senegal Seychelles Sierra Leone South Africa Swaziland The Gambia Zambia Zimbabwe Median for region Latin America and Caribbean Argentina Barbados Brazil Colombia Dominica Dominican Republic Ecuador Guatemala Jamaica Mexico Paraguay Saint Kitts and Nevis Uruguay Median for region 139 (071 to 214) 091 (072 to 111) 108 (038 to 181) 082 (019 to 150) 051 (041 to 186) 128 (041 to 187) 080 (024 to 141) 035 (002 to 074) 101 (036 to 174) 076 (018 to 140) 085 (023 to 151) 077 (061 to 095) 070 (055 to 085) 082 (Continues in next column) 079 (062 to 096) 019 (015 to 024) 081 (064 to 099) 031 (001 to 066) 085 (026 to 149) 046 (035 to 056) 046 (003 to 093) 017 (004 to 040) 106 (035 to 187) 072 (016 to 137) 089 (070 to 109) 089 (070 to 108) 043 (002 to 089) 081 (015 to 155) 040 (032 to 049) 025 (001 to 053) 037 (002 to 079) 052 (041 to 063) 021 (017 to 026) 040 (014 to 081) 095 (035 to 160) 090 (022 to 162) 014 (011 to 017) 145 (062 to 239) 057 (045 to 069) 036 (028 to 044) 049 (004 to 097) 051 (040 to 063) 038 (030 to 046) 126 (099 to 153) 156 (076 to 245) 052 (041 to 063) 036 (002 to 077) 056 (005 to 113) 051 (Continued from previous column) North America Canada USA Median for region Eastern Mediterranean Iran Iraq Kuwait Lebanon Libya Pakistan Saudi Arabia Tunisia United Arab Emirates Median for region Europe Austria Belgium Bosnia and Herzegovina Bulgaria Croatia Cyprus Czech Republic Denmark Estonia Finland France Georgia Germany Greece Hungary Ireland Italy Kazakhstan Latvia Lithuania Luxembourg Malta Netherlands Norway Poland Portugal Romania Russia Serbia Slovakia Slovenia Spain Sweden Turkey

Gain in life expectancy (years)

055 (011 to 102) 078 (062 to 094) 066 071 (057 to 087) 130 (103 to 159) 112 (089 to 137) 095 (075 to 116) 093 (077 to 118) 085 (025 to 152) 151 (119 to 184) 064 (016 to 116) 111 (048 to 178) 095 058 (014 to 103) 073 (023 to 126) 062 (012 to 116) 058 (007 to 111) 045 (004 to 091) 080 (033 to 133) 048 (006 to 092) 064 (015 to 119) 038 (002 to 078) 066 (017 to 117) 055 (044 to 067) 052 (041 to 063) 047 (006 to 092) 023 (001 to 049) 061 (006 to 118) 087 (036 to 142) 080 (031 to 132) 079 (014 to 150) 077 (012 to 146) 053 (041 to 065) 083 (027 to 145) 112 (056 to 171) 029 (000 to 059) 068 (020 to 118) 060 (008 to 114) 086 (033 to 145) 087 (021 to 158) 052 (002 to 105) 150 (102 to 233) 046 (004 to 092) 054 (008 to 105) 078 (028 to 132) 067 (022 to 116) 106 (043 to 174) (Continues in next column)

www.thelancet.com Vol 380 July 21, 2012

225

Articles

Gain in life expectancy (years) (Continued from previous column) Ukraine UK Median for region Southeast Asia Bangladesh Bhutan Burma India Indonesia Maldives Nepal Sri Lanka Thailand Median for region Western Pacic Australia Cambodia China Cook Islands Federated States of Micronesia Japan Kiribati Laos Malaysia Marshall Islands Mongolia Nauru New Zealand Papua New Guinea Philippines Samoa Solomon Islands Tonga Vietnam Median for region Overall Median 068 056 (015 to 100) 024 (019 to 029) 061 (048 to 073) 157 (124 to 191) 145 (114 to 177) 091 (040 to 146) 127 (099 to 155) 043 (001 to 090) 135 (106 to 165) 133 (104 to 163) 024 (019 to 029) 121 (095 to 147) 076 (061 to 093) 043 (034 to 052) 052 (003 to 104) 117 (092 to 142) 090 (070 to 109) 103 (080 to 126) 031 (001 to 064) 090 010 (008 to 012) 115 (042 to 195) 027 (022 to 033) 034 (027 to 041) 065 (051 to 080) 075 (018 to 137) 033 (002 to 071) 051 (040 to 061) 041 (032 to 049) 041 046 (000 to 097) 107 (085 to 129) 063

Data in parentheses are 95% CI. Uncertainty interval calculated on the basis of the lower and upper bounds of the 95% CI of the adjusted population attributable fraction for all-cause mortality.

Table 3: Estimated gains in life expectancy if physical inactivity were eliminated, by WHO region and country

For the association of type 2 diabetes incidence with physical activity, Jeon and co-workers11 reported a pooled RR of 083 (95% CI 076090), adjusted for several confounders including body-mass index (BMI). Taking the inverse to obtain the adjusted RR for inactivity, we calculated an RR of 120 (95% CI 110133). We calculated the corresponding, pooled unadjusted RR, which was 163 (95% CI 127211; appendix). This magnitude of risk increase was similar
226

to that pooling only the crude RRs, which yielded 158 (95% CI 111226). We used the primary papers in the qualitative review by Friedenreich and colleagues15 to undertake a metaanalysis of the pooled adjusted and unadjusted RRs for breast cancer incidence (appendix). The adjusted RR for physical inactivity, including adjustment for BMI, was 133 (95% CI 126142). This result was little dierent from the unadjusted RR of 134 (95% CI 125143; similar to that pooling only crude RRs, yielding 135, 95% CI 126145). For the association of colon cancer incidence and physical activity, Wolin and colleagues12 reported a pooled adjusted RR of 076 (95% CI 072081). Taking the inverse of these results gave an adjusted RR of 132 (95% CI 123139) for inactivity. Our calculation of the pooled unadjusted RR for colon cancer was 138 (95% CI 131145; appendix); the pooled crude RR was similar (137, 95% CI 129146). Lollgen and colleagues13 reported a meta-analysis of RRs for all-cause mortality associated with moderate and high levels of physical activity, qualitatively dened. Investigators reported separate estimates for studies in which participants were classied into three, four, or ve levels of activity. The adjusted RRs for moderate levels compared with a low level ranged from 053 to 078; for high levels, from 052 to 080. We used their primary papers to do a meta-analysis to obtain one pooled RR that compared low with moderate physical activityie, a conservative estimate of the eect of inactivity. Our pooled adjusted RR was 128 (95% CI 121136), whereas the pooled unadjusted RR was 147 (95% CI 138157) and similar to the pooled crude RR of 146 (95% CI 134160; appendix). For coronary heart disease, median PAFs associated with physical inactivity, calculated with adjusted RRs, ranged from 32% (in southeast Asia) to 78% (in the eastern Mediterranean region), with an overall median of 6% (tables 1, 2). These results suggest that 6% of the burden of disease worldwide due to coronary heart disease can be eliminated, if all inactive people become active. The burden of disease was 7% for type 2 diabetes (ranging from 39% to 96%), 10% (56141) for breast cancer, and 10% (57138) for colon cancer. Removal of physical inactivity had the largest eect on colon cancer, and the smallest on coronary heart disease, in terms of percentage reduction. However, with respect to the number of cases that can potentially be averted, coronary heart disease would have a far larger eect than would colon cancer because of its higher incidence. Although the worldwide incidence of coronary heart disease is not readily available, deaths from coronary heart disease can be viewed against colorectal cancer deaths to provide some perspective; in 2008, 725 million people worldwide died from coronary heart disease versus 647 000 from colorectal cancer.18 Applying the median PAFs, we estimated that 15 000 deaths from
www.thelancet.com Vol 380 July 21, 2012

Articles

Years 100 075099 050074 025049 <025 No data

Figure: Estimated gains in life expectancy worldwide with elimination of physical inactivity

coronary heart disease in Africa could have been averted in 2008 by removal of physical inactivity. 60 000 could have been avoided in the Americas, 44 000 in the eastern Mediterranean region, 121 000 in Europe, 59 000 in southeast Asia, and 100 000 in the western Pacic region. With respect to deaths from breast cancer, 3000 could have been averted in Africa, 11 000 in the Americas, 4000 in the eastern Mediterranean, 14 000 in Europe, 5000 in southeast Asia, and 10 000 in the western Pacic; for deaths from colorectal cancer, these numbers were 1000, 14 000, 2000, 24 000, 4000, and 24 000, respectively. For all-cause mortality, the overall median PAF was 9%. Applying this gure to the 57 million deaths worldwide in 2008,18 we estimated that more than 53 million deaths (ranging from 525 000 in the eastern Mediterranean to 15 million in the western Pacic region) could be averted every year if all inactive people become active. Because physical inactivity is unlikely to be completely eliminated, we further estimated potential deaths averted when assuming a decrease of inactivity prevalence by 10% or 25% with eective public health interventions, instead of 100% (elimination). These alternate scenarios resulted in more than 533 000 and 13 million deaths potentially avoided worldwide each year. Using an alternate classication of countries by income (data not shown), we calculated median PAFs for all-cause mortality of 4% for countries with low incomes, 8% for lower-middle incomes, 10% for upper-middle incomes, and 11% for high incomes (with number of deaths averted ranging from 409 000 in countries with low incomes to 25 million in those with lower-middle incomes). This analysis yielded estimated numbers of deaths potentially averted in 2008 from coronary heart disease of 15 000, 184 000, 96 000, and 98 000; from breast cancer of 2000, 16 000, 10 000, and 20 000; and from colorectal cancer of 1000, 19 000, 13 000, and 37 000, respectively. We estimated that the median years of life potentially gained worldwide with elimination of physical inactivity
www.thelancet.com Vol 380 July 21, 2012

was 068 years (ranging from 041 years in southeast Asia to 095 years in the eastern Mediterranean region; table 3, gure). When we classied countries by income, the median gains were 037 years for countries with low incomes, 065 for lower-middle incomes, 080 for uppermiddle incomes, and 068 years for high incomes. In a sensitivity analysis that decreased age-specic death rates by the PAF for all ages 40 years or older (instead of only ages 4079 years), the new estimate of years gained worldwide increased to a median of 092 years (range 049125). Finally, we used an example to illustrate gains under less stringent assumptions. A recent study of Taiwanese people aged 20 years and older comparing most with least active groups reported an RR for all-cause mortality of 135.19 Applying this RR to China for people aged 20 years and older, we calculated a PAF of 98% and gain in life expectancy of 103 years, versus 83% and 061 years obtained under the standard assumptions of tables 2 and 3.

Discussion
Worldwide, we estimated that physical inactivity causes 610% of the major non-communicable diseases of coronary heart disease, type 2 diabetes, and breast and colon cancers. Furthermore, this unhealthy behaviour causes 9% of premature mortality, or more than 53 of the 57 million deaths in 2008.18 With elimination of physical inactivity, life expectancy of the worlds population might be expected to increase by 068 years. These ndings make inactivity similar to the established risk factors of smoking and obesity. The added years of life need to be interpreted correctly: they seem low because they represent gains in the whole population (including inactive and active people), rather than in inactive people who become active. Because all the gain accrues to people who move from inactive to active, the increase in life expectancy in the inactive group alone is
227

Articles

greater. For perspective, other research done in the USA estimated that inactive people would gain 1337 years from age 50 years by becoming active.20,21 In an east Asian population, life expectancy from age 30 years in active people was 2642 years greater than that in inactive people.19 How does physical inactivity compare with other risk factors for poor health? Although risk factors are classied on dierent scales (thus, the proportion at risk varies across risk factors), it is nonetheless informative to look at two established risk factors targeted for government action worldwide: smoking and obesity. Smoking was estimated to cause about 5 million deaths worldwide in 2000.22 The proportion of deaths attributable to smoking in China, one of the top ve cigaretteconsuming countries, has been estimated at 31% for women and 129% for men.23 By elimination of smoking, life expectancy at age 50 years was estimated to increase by 2325 years in the US population and 1122 years in the populations of nine other high-income countries.24 At an individual level, Beijing never-smokers aged 55 years and older had a life expectancy 4288 years longer than that of present smokers.25 As for obesity, if all obese people in the USA were to attain normal weight, life expectancy in the population was estimated to increase by 0711 years at birth in one analysis26 and 0507 years at age 50 years in another.24 Thus, physical inactivity seems to have an eect similar to that of smoking or obesity. The present analysis updates information from a 2004 WHO report27 and additionally estimates added years of life expectancy in the population. In the WHO report, because of unavailability of data needed for the preferred PAF formula, the incorrect formula (formula 1) was used. Their PAFs ranged from 10% for breast cancer to 22% for coronary heart diseasesimilar to the present estimate for breast cancer, but larger for coronary heart disease. In part, this dierence is because the RR of breast cancer for physical inactivity is not confounded by other variables (unadjusted RR 134, adjusted RR 133), whereas that for coronary heart disease is (unadjusted RR 133, adjusted RR 116). Further, we conservatively used a pooled RR for coronary heart disease that compared physical inactivity with the minimum recommended activity level using recently published data,10 whereas WHO used data available at the time of their analysis that compared extreme activity categories, yielding RRs of larger magnitude. Our estimates are likely to be very conservative. First, the RRs were almost always based on self-reported physical activity levels,28 which are imprecise. In prospective studies in which self-reports cannot be biased by the outcomes studied (since they have not yet occurred at the time of reporting), random reporting errors result in underestimation of the RRs. Some studies of physical tnessa related measure to physical activity that is more objectively measuredshow stronger magnitudes
228

of association with non-communicable diseases29 (which also might reect inherited physiological and metabolic characteristics related to both tness and a favourable risk prole). Second, the pooled RRs were derived from maximally adjusted RRs in the primary studies. Often, these RRs were adjusted for characteristics such as blood pressure, lipid prole, and glucose or insulin sensitivity. These could be overadjustments, since physical activity reduces risk of coronary heart disease and premature mortality partly through benecial eects on these variables (a recent analysis suggested an attenuation of about 10% in the RR10). For type 2 diabetes, we used RRs adjusted for BMIalso conservative, since physical activity plays an important part in weight management.3 Third, we used the same RR to calculate PAFs for all countries, based on data mainly from North America and Europe. Whether physical inactivity has similar eects in other populations is unclear. For example, we used a pooled adjusted RR for coronary heart disease of 116; however, a study in India reported a larger adjusted RR (227, 95% CI 141370).30 However, our pooled adjusted RR for all-cause mortality, 128, is similar to that of 125 (95% CI 118133) in an east Asian study, comparing inactive people with those meeting minimum physical activity recommendations.19 Fourth, we assumed physical activity to reduce all-cause mortality rates only at ages 4079 years. In a sensitivity analysis that extended the benet to age 40 years and older, larger gains in life expectancy were obtained. Fifth, we used one RR instead of a range of RRs to reect the dose-response relation between physical inactivity and risks of non-communicable disease because sparse data are available for the dose-response relation.10 In an illustrative example using China, we showed that applying less stringent assumptions increased PAF by 18% (98% vs 83%) and life expectancy by 69% (103 vs 061 years) compared with calculations made under standard assumptions. Limitations of this study include the use of an adjustment factor to estimate the prevalence of physical inactivity in cases. This adjustment factor was mainly based on populations in North America and Europe, and one study each from China and India; how applicable this factor might be to other countries such as those in Africa or with low incomes is unclear. Also, successful interventions will probably increase activity levels across the board, instead of shifting people across a binary divide of inactive to active assumed in our calculations, potentially yielding greater benets. We examined only the major non-communicable diseases and all-cause mortality, and not other disorders aected by physical inactivity (panel 1) or disability resulting from non-communicable diseases. Finally, not all physically inactive people choose to be so; some might be physically incapable. This summer, we will admire the breathtaking feats of athletes competing in the 2012 Olympic Games. Although only the smallest fraction of the population will attain these heights, the overwhelming majority of us are
www.thelancet.com Vol 380 July 21, 2012

Articles

able to be physically active at very modest levels eg, 1530 min a day of brisk walkingwhich bring substantial health benets.35,19 We must explore all avenues and support all eorts to reduce physical inactivity worldwide.
Contributors I-ML and PTK designed the study and other authors provided critical input. EJS and PTK undertook data analyses. I-ML, EJS, and FL extracted data from the primary studies used in the meta-analyses. I-ML drafted the report. EJS, FL, PP, SNB, and PTK critically reviewed the report. Lancet Physical Activity Series Working Group Jasem R Alkandari, Lars Bo Andersen, Adrian E Bauman, Steven N Blair, Ross C Brownson, Fiona C Bull, Cora L Craig, Ulf Ekelund, Shifalika Goenka, Regina Guthold, Pedro C Hallal, William L Haskell, Gregory W Heath, Shigeru Inoue, Sonja Kahlmeier, Peter T Katzmarzyk, Harold W Kohl 3rd, Estelle Victoria Lambert, I-Min Lee, Grit Leetongin, Felipe Lobelo, Ruth J F Loos, Bess Marcus, Brian W Martin, Neville Owen, Diana C Parra, Michael Pratt, Pekka Puska, David Ogilvie, Rodrigo S Reis, James F Sallis, Olga Lucia Sarmiento, Jonathan C Wells. Conicts of interest We declare that we have no conicts of interest. Acknowledgments The ndings and conclusions in this report are those of the authors and do not necessarily represent the ocial position of the US Centers for Disease Control and Prevention. I-ML was supported in part by grant CA154647 from the US National Institutes of Health. EJS was supported in part by grant HL007575 from the US National Institutes of Health. PTK was supported in part by the Louisiana Public Facilities Authority Endowed Chair in Nutrition. We thank Kenneth E Powell, Shane A Norris, and Beverly J Levine for reviewing a previous draft of the report and providing critical input. We thank several people for providing data to calculate the adjustment factor: David Batty, Kennet Harald, Duck-chul Lee, Charles E Matthews, Martin Shipley, Emmanuel Stamatakis, Xuemei Sui, and Nicholas J Wareham. We thank Jacob R Sattelmair and Kathleen Y Wolin for assisting with meta-analyses. References 1 Rook A. An investigation into the longevity of Cambridge sportsmen. BMJ 1954; 1: 77377. 2 Blair SN, Davey Smith G, Lee IM, et al. A tribute to Professor Jeremiah Morris: the man who invented the eld of physical activity epidemiology. Ann Epidemiol 2011; 20: 65160. 3 US Department of Health and Human Services. 2008 Physical Activity Guidelines Advisory Committee report. http://www.health. gov/paguidelines/ (accessed Sept 1, 2011). 4 Warburton DE, Charlesworth S, Ivey A, Nettlefold L, Bredin SS. A systematic review of the evidence for Canadas Physical Activity Guidelines for Adults. Int J Behav Nutr Phys Act 2010; 7: 39. 5 WHO. Global recommendations on physical activity for health. Geneva: World Health Organization, 2010. 6 Beaglehole R, Bonita R, Alleyne G, et al. UN High-Level Meeting on Non-Communicable Diseases: addressing four questions. Lancet 2011; 378: 44955. 7 Powell KE, Blair SN. The public health burdens of sedentary living habits: theoretical but realistic estimates. Med Sci Sports Exerc 1994; 26: 85156. 8 Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable fractions. Am J Public Health 1998; 88: 1519. 9 WHO. Global status report on noncommunicable diseases 2010. Geneva: World Health Organization, 2011.

10

11

12 13

14 15

16

17

18 19

20

21

22 23 24

25

26

27

28

29

30

Sattelmair JR, Perman J, Ding EL, Hohl HW III, Haskell WL, Lee I-M. Dose-response between physical activity and risk of coronary heart disease: a meta-analysis. Circulation 2011; 124: 78995. Jeon CY, Lokken RP, Hu FB, van Dam RM. Physical activity of moderate intensity and risk of type 2 diabetes: a systematic review. Diabetes Care 2007; 30: 74452. Wolin KY, Yan Y, Colditz GA, Lee IM. Physical activity and colon cancer prevention: a meta-analysis. Br J Cancer 2009; 100: 61116. Lollgen H, Bockenho A, Knapp G. Physical activity and all-cause mortality: an updated meta-analysis with dierent intensity categories. Int J Sports Med 2009; 30: 21324. Monninkhof EM, Elias SG, Vlems FA, et al. Physical activity and breast cancer: a systematic review. Epidemiology 2007; 18: 13757. Friedenreich CM. Physical activity and breast cancer: review of the epidemiologic evidence and biologic mechanisms. Recent Results Cancer Res 2011; 188: 12539. WHO. Life tables for WHO member states. 2011. http://www. who.int/healthinfo/statistics/mortality_life_tables/en/ (accessed Sept 1, 2011). Yusuf S, Hawken S, unpuu S, et al, on behalf of the INTERHEART Study Investigators. Eect of potentially modiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet 2004; 364: 93752. WHO. Global Health Observatory Data Repository. 2011. http:// apps.who.int/ghodata/ (accessed Oct 4, 2011). Wen CP, Wai JPM, Tsai MK, et al. Minimum amount of physical activity for reduced mortality and extended life expectancy: a prospective cohort study. Lancet 2011; 378: 124453. Franco OH, de Laet C, Peeters A, Jonker J, Mackenbach J, Nusselder W. Eects of physical activity on life expectancy with cardiovascular disease. Arch Intern Med 2005; 165: 235560. Paenbarger RS Jr, Hyde RT, Wing AL, Hsieh CC. Physical activity, all-cause mortality, and longevity of college alumni. N Engl J Med 1986; 314: 60513. Ezzati M, Lopez AD. Estimates of global mortality attributable to smoking in 2000. Lancet 2003; 362: 84752. Gu D, Kelly TN, Wu X, et al. Mortality attributable to smoking in China. N Engl J Med 2009; 360: 15059. Crimmins EM, Preston SH, Cohen B. Chapters 3 and 5. Explaining divergent levels of longevity in high-income countries. Washington, DC: National Academies Press, 2011. Tian X, Tang Z, Jiang J, et al. Eects of smoking and smoking cessation on life expectancy in an elderly population in Beijing, China, 19922000: an 8-year follow-up study. J Epidemiol 2011; 21: 37684. Olshansky SJ, Passaro DJ, Hershow RC, et al. A potential decline in life expectancy in the United States in the 21st century. N Engl J Med 2005; 352: 113845. Ezzati M, Lopez AD, Rodgers A, Murray CJ, eds. Comparative quantication of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization, 2004. Lee I-M, Paenbarger RS Jr. Design of present-day epidemiologic studies of physical activity and health. In: Lee I-M, ed. Epidemiologic methods in physical activity studies. New York, NY: Oxford University Press, 2009: 10023. Kodama S, Saito K, Tanaka S, et al. Cardiorespiratory tness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: a meta-analysis. JAMA 2009; 301: 202435. Rastogi T, Vaz M, Spiegelman D, et al. Physical activity and risk of coronary heart disease in India. Int J Epidemiol 2004; 33: 75967.

www.thelancet.com Vol 380 July 21, 2012

229