Skip to main content

Advertisement

Table 11 Studies examining the relationship between physical activity and all-cause mortality.

From: A systematic review of the evidence for Canada's Physical Activity Guidelines for Adults

Publication Country Study Design Quality Score Objective Population Methods Outcome Comments and Conclusions
Blair et al 1989 [7] To study physical fitness (PF) and risk of all-cause mortality in men and women. • n = 13,344 (10,224 men; 3,120 women) Baseline and 8 year follow-up    • 283 deaths Low levels of PF increase the risk for premature mortality.
   • Sex: Men and women   Adjusted risk ratio (RR), 95% confidence interval (CI)  
USA   • Age: 20->60 years (yr) PF assessment: Maximal treadmill exercise test.   
Prospective cohort   • Characteristics: Participants were given a preventative Medicine examination including maximal treadmill exercise test Fitness categorized into quintiles: Men  
D & B score = 12    Q1 = least fit    • Q1 = 3.44 (2.05-5.77)  
    Q2    • Q2 = 1.37 (0.76-2.50)  
    Q3    • Q3 = 1.46 (0.81-2.63)  
    Q4    • Q4 = 1.17 (0.63-2.17)  
    Q5 = most fit    • Q5 = 1.00 (referent)  
     Women  
        • Q1 = 4.65 (2.22-9.75)  
        • Q2 = 2.42 (1.09-5.37)  
        • Q3 = 1.43 (0.60-3.44)  
        • Q4 = 0.76 (0.27-2.11)  
        • Q5 = 1.00 (referent)  
Myers et al 2004 [32] To determine the effects of PF and physical activity (PA) on all-cause mortality. • n = 6,213 Baseline and mean 5.5 ± 2.0 year follow-Up    • 1,256 deaths Being fit or active is associated with >50% reductions in mortality risk.
   • Sex: Men    
USA   • Age: Mean 59.0 ± 11.2 yr   PF Level hazard ratio (HR) (95% CI)  
   • Characteristics: Men referred for exercise testing PF assessment: Treadmill test to measure VO2 peak    • G1 = 1.00 (referent) PF predicted mortality more strongly than PA.
Prospective cohort        • G2 = 0.59 (0.52-0.68)  
        • G3 = 0.46 (0.39-0.55)  
        • G4 = 0.28 (0.23-0.34) Increasing PA (by 1000 kcal/wk or 1 MET) confers a mortality benefit of 20%.
D & B score = 12    PA assessment: Self reported PA divided into 4 groups   
     PA Level HR (95% CI)  
    G1 = Lowest level    • G1 = 1.00 (referent)  
    G2    • G2 = 0.63 (0.36-1.10)  
    G3    • G3 = 0.42 (0.23-0.78)  
    G4 = Highest level    • G4 = 0.38 (0.19-0.73)  
Blair et al 1995 [36] To evaluate the relationship between changes in PF and risk of mortality in men. • n = 9,777 4.9 year mean follow-up    • 223 deaths Men who maintained or increased adequate PF had a reduced risk for all-cause mortality than individuals who were consistently unfit.
   • Sex: Men    
   • Age: 20-82 yr   RR (95% CI)  
USA   • Characteristics: Participants were given a preventative medicine examination including maximal treadmill exercise test PF assessment: Maximal exercise test at baseline and follow-up    • G1 = 1.00 (referent)  
Prospective cohort        • G2 = 0.56 (0.41-0.75)  
        • G3 = 0.52 (0.38-0.70)  
        • G4 = 0.33 (0.23-0.47)  
D & B score = 13    Groups based on changes in PF   
    G1 = unfit to unfit   
    G2 = unfit to fit   
    G3 = fit to unfit   
    G4 = fit to fit   
Bijnen et al 1999 [37] To examine the association of PA at baseline and 5 years • n = 472 1985 and 1990    • 118 deaths Recent levels of PA were more important for mortality risk than PA 5 years previously.
   • Sex: Men    
   • Age: >65 yr PA assessment: Questionnaire, divided into tertiles: Lowest Middle Highest Multivariate adjusted RR (95% CI)  
Netherlands previously with all- cause mortality risk in a cohort of elderly Dutch men. • Characteristics: Mostly independently living elders (~95%)   PA in 1985: Lowest tertile = 1.00 (referent) Middle tertile  
Retrospective cohort   • Zutphen Elderly Study      • Total activity = 1.25 (0.79- 1.99) Becoming or remaining sedentary increased the mortality risk.
D & B score = 12        • Walking = 0.97 (0.60-1.57)  
        • Bike = 0.97 (0.59-1.57)  
        • Gardening = 0.66 (0.39-1.10)  
        • Other = 1.08 (0.66-1.78)  
        • Heavy activity = 0.73 (0.45-1.17)  
        • Non heavy activity = 0.89 (0.57-1.40)  
     Highest tertile  
        • Total activity = 1.25 (0.73-2.12)  
        • Walking = 0.94 (0.58-1.55)  
        • Bike = 1.07 (0.61-1.88)  
        • Gardening = 0.77 (0.42-1.39)  
        • Other = 1.24 (0.74-2.07)  
        • Heavy activity = 0.76 (0.44-1.32)  
        • Non heavy activity = 0.94 (0.58-1.53)  
     PA in 1990:  
     Lowest tertile = 1.00 (referent)  
     Middle tertile  
        • Total activity = 0.56 (0.35-0.89)  
        • Walking = 0.82 (0.51-1.32)  
        • Bike = 0.49 (0.29-0.82)  
        • Gardening = 1.67 (1.00-2.79)  
        • Other = 0.93 (0.53-1.65)  
        • Heavy activity = 1.19 (0.73-1.92)  
        • Non heavy activity = 0.61 (0.38-0.99)  
     Highest tertile  
        • Total activity = 0.44 (0.25-0.80)  
        • Walking = 1.17 (0.70-1.96)  
        • Bike = 0.43 (0.23-0.80)  
        • Gardening = 1.03 (0.55-1.94)  
        • Other = 0.74 (0.44-1.23)  
        • Heavy activity = 0.72 (0.40-1.31)  
        • Non heavy activity = 0.65 (0.40-1.05)  
Gregg et al 2003 [39] To examine the relationship of changes in PA and mortality among older women. • n = 9,518 Baseline (1986-1988) and median 10.6 year follow-up (1992-1994)    • 2,218 deaths Increasing and maintaining PA levels could lengthen life for older women but appears to provide less benefit for women aged at least 75 years and those with poor health status.
   • Sex: Women PA Assessment: Questionnaire, divided into quintiles of PA (kcal/wk)   
   • Age: ≥ 65 yr   Multivariate adjusted HRR  
USA   • Characteristics: White community dwelling participants from 4 US research centres   (95% CI): Quintiles of total  
    Q1= <163 PA  
Prospective cohort    Q2 = 163-503    • Q1 = 1.00 (referent)  
    Q3 = 504-1045    • Q2 = 0.73 (0.64-0.82)  
    Q4 = 1046-1906    • Q3 = 0.77 (0.68-0.87)  
D & B score = 13    Q5 = ≥ 1907    • Q4 = 0.62 (0.54-0.71)  
        • Q5 = 0.68 (0.59-0.78)  
     Walking HRR (95% CI)  
        • Q1 = 1.00 (referent)  
    Quintiles of walking(kcal/wk)    • Q2 = 0.91 (0.81-1.02)  
    Q1 = <70    • Q3 = 0.78 (0.68-0.88)  
    Q2 = 70-186    • Q4 = 0.71 (0.63-0.82)  
    Q3 = 187-419    • Q5 = 0.71 (0.62-0.82)  
    Q4 = 420-897   
    Q5 = 898   
     Multivariate adjusted HRR (95% CI)  
     Change in activity level: Sedentary at baseline  
        • Staying sedentary = 1.00 (referent)  
        • Became active = 0.52 (0.40-0.69)  
     Mod / high active at baseline  
        • Became sedentary = 0.92 (0.77-1.09)  
        • Stayed active = 0.68 (0.56-0.82)  
Wannamethee et al 1998 [40] To study the relationship between heart rate, PA and all- cause mortality. • n = 5,934 Baseline (1978-1980) and 12-14 year follow-up    • 219 deaths Maintaining or taking up light or moderate PA reduces mortality in older men.
   • Sex: Men    
   • Age: Mean 63 yr   Multivariate adjusted RR (95% CI),  
UK   • Characteristics: Healthy, sedentary(4,311 were considered "healthy" in 1992) PA assessment: Questionnaire, split into groups PA  
Prospective cohort   • The British Regional Heart Study      • G1 = 1.00 (referent)  
        • G2 = 0.61 (0.43-0.86)  
        • G3 = 0.50 (0.31-0.79)  
D & B score = 12    PA score    • G4 = 0.65 (0.45-0.94)  
    G1 =   
    Inactive/occasional Regular walking  
    G2 = Light    • G1 = 1.00 (referent)  
    G3 = Moderate    • G2 = 1.15 (0.73-1.79)  
    G4 = Moderately    • G3 = 1.06 (0.75-1.50)  
    vigorous/Vigorous    • G4 = 0.97 (0.65-1.46)  
    Regular walking (min/d)    • G5 = 0.62 (0.37-1.05)  
    G1 = 0 Recreational activity  
    G2 = <20    • G1 = 1.00 (referent)  
    G3 = 21-40    • G2 = 0.95 (0.43-1.07)  
    G4 = 41-60    • G3 = 0.68 (0.43-1.07)  
    G5 = ≥ 60    • G4 = 0.34 (0.35-1.00)  
    Recreational activity, 4 groups Sporting activity  
    G1 = Inactive/fairly Inactive    • G1 = 1.00 (referent)  
    G2 = Average 4 hr/weekend    • G2 = 0.50 (0.25-1.03)  
    G3 = Fairly active >4 h/weekend    • G3 = 0.88 (0.64-1.23)  
    G4 = Very active   
    Sporting activity, 3 Groups   
    G1 = None   
    G2 = Occasional   
    G3 = >1 time/month   
Paffenbarger et al 1986 [63] To examine the PA and life-style characteristics of Harvard alumni for the relationship with all-cause mortality. • n = 16,936 12-16 year follow-up (1962 to 1978)    • 1,413 deaths The findings suggest a protective effect of exercise against all-cause mortality.
   • Sex: Men   Age adjusted RR (95% CI):  
   • Age: 35-74    
USA   • Characteristics: Harvard alumni Records of freshman year physical examinations and records of intercollegiate sport Those who walked  
Prospective cohort        • G1 = 1.00 (referent)  
        • G2 = 0.85  
        • G3 = 0.79  
D & B score = 14     Trend p = 0.0009  
    PA assessment: Mailed questionnaires surveying post college Physical Activity Index (95% CI):  
    PA    • G1 = 1.00 (referent)  
        • G2 = 0.78  
        • G3 = 0.73  
        • G4 = 0.63  
    Exercise reported: Walking (miles/wk) 3    • G5 = 0.62  
    groups    • G6 = 0.52  
    G1 = <3    • G7 = 0.46  
    G2 = 3-8    • G8 = 0.62  
    G3 = ≥ 9   
     Trend p = <0.0001  
    PA index (kcal/wk) 3 groups:   
    G1 = <500   
    G2 = 500-999   
    G3 = 1000-1499   
    G4 = 1500-1999   
    G5 = 2000-2499   
    G6 = 2500-2999   
    G7 = 3000-3499   
    G8 = >3500   
    Cox proportional hazard models   
Schnohr et al 2007 [64] To determine the impact of walking duration and intensity on all-cause mortality. • n = 7,308 (3,204 male; 4,104 female) Baseline and an average of 12 year    • 1,391 deaths The findings indicate that the relative intensity and not duration of walking is the most important in relation to all-cause mortality.
Denmark   • Sex: Male and female follow-up Multivariate adjusted HR (95% CI):  
   • Age: 20-93 yr PA assessment: Questionnaire, 4 durations and 3 intensities   
Prospective cohort   • Characteristics: Participants with no history of CHD, stroke or cancer and who had no difficulty in walking   Men  
D & B score = 12   • The Copenhagen City Heart Study      • G1 = 1.00 (referent)  
        • G2 = 0.38 (0.25-0.58)  
        • G3 = 0.38 (0.18-0.79)  
    Duration (hours/day)    • G4 = 0.69 (0.44-1.07)  
    1 = <0.5    • G5 = 0.37 (0.26-0.54)  
    2 = 0.5-1    • G6 = 0.33 (0.18-0.61)  
    3 = 1-2    • G7 = 0.78 (0.50-1.23)  
    4 = >2    • G8 = 0.41 (0.29-0.59)  
        • G9 = 0.33 (0.20-0.54)  
    Intensity    • G10 = 0.43 (0.22-0.82)  
    Slow intensity (SI)    • G11 = 0.42 (0.29-0.60)  
    Average intensity (AI)    • G12 = 0.28 (0.16-0.48)  
    Fast intensity (FI)   
     Women  
    12 groups    • G1 = 1.00 (referent)  
    G1 = 1 and SI    • G2 = 0.82 (0.52-1.29)  
    G2 = 1 and AI    • G3 = 0.78 (0.27-2.21)  
    G3 = 1 and FI    • G4 = 1.22 (0.82-1.81)  
    G4 = 2 and SI    • G5 = 0.74 (0.52-1.05)  
    G5 = 2 and AI    • G6 = 0.56 (0.33-0.96)  
    G6 = 2 and FI    • G7 = 0.94 (0.60-1.47)  
    G7 = 3 and SI    • G8 = 0.87 (0.61-1.23)  
    G8 = 3 and AI    • G9 = 0.48 (0.28-0.83)  
    G9 = 3 and FI    • G10 = 0.88 (0.40-1.88)  
    G10 = 4 and SI    • G11 = 0.64 (0.44-0.95)  
    G11 = 4 and AI    • G12 = 0.38 (0.21-0.69)  
    G12 = 4 and FI   
Kushi et al 1997 [65] To evaluate the association between PA and all-cause mortality in postmenopausal women. • n = 40,417 7 year follow-up • 2,260 deaths The results demonstrate a graded inverse association between PA and all-cause mortality in postmenopausal women.
   • Sex: Women    
   • Age: 55-69 yr PA assessment: Questionnaire for frequency of moderate and vigorous LTPA Multivariate adjusted Frequency of moderate PA per week RR (95% CI):  
USA   • Characteristics: Postmenopausal Iowa women    
Prospective cohort      
     • G1 = 1.00 (referent)  
     • G2 = 0.71 (0.63-0.79)  
D & B score = 13    Divided by frequency/week • G3 = 0.63 (0.56-0.71)  
     • G4 = 0.59 (0.51-0.67)  
    G1 = Rarely/never Trend p = <0.001  
    G2 = 1 time/week to a few times/month   
     Frequency of vigorous PA per week  
    G3 = 2-4 times/week   
    G4 = >4 times/week • G1 = 1.00 (referent)  
     • G2 = 0.83 (0.69-0.99)  
     • G3 = 0.74 (0.59-0.93)  
    Activity index • G4 = 0.62 (0.42-0.90)  
    G1 = Low Trend p = 0.009  
    G2 = Medium   
    G3 = High   
     • G1 = 1.00 (referent)  
     • G2 = 0.77 (0.69-0.86)  
     • G3 = 0.68 (0.60-0.77)  
     Trend p = <0.001  
Paffenbarger et al 1993 [67] To analyze changes in the lifestyles of Harvard College alumni and the association of these changes with mortality. • n = 10,269 Baseline (1977) and 8 year follow-up (1985) • 476 deaths Beginning moderately vigorous sports activity was associated with lower rates of death from all causes among middle aged and older men.
   • Sex: Men    
   • Age: 45-84 yr (in 1977)   Beginning moderate sports activity was associated with 23% lower risk of death (95% CI 4%-42%, p = 0.015) than those not taking up moderate activity  
USA   • Characteristics: Participants with no reported life- threatening disease PA Assessment: Questionnaire -- blocks walked daily, stairs climbed daily and type, frequency and duration of weekly sports and recreational activities   
Prospective cohort      
D & B score = 13      
    Physical activity index (kcal/wk)   
    Sports and recreational activities   
    Light <4.5 METs   
    Moderate >4.5 METs   
    Weekly lists of deaths were obtained from the Harvard college alumni office   
    Proportional hazard models with Poisson regression methods   
Katzmarzyk and Craig 2002 [154] To quantify the relationship between musculoskeletal fitness and all-cause mortality. • n = 8,116 (3,933 male; 4,183 female) Baseline (1981) and • 238 deaths Some components of musculoskeletal fitness are predictive of mortality.
    13 year follow-up   
   • Sex: Men and women   RR (95% CI) adjusted for age, smoking status, body mass and VO2max  
Canada    Musculoskeletal fitness (sit ups, push ups, grip strength, sit and reach) measures divided into quartiles   
   • Age: 20-69 yr Q1 = lowest Sit ups  
Prospective cohort   • Characteristics: Participants who had musculoskeletal fitness measurements taken Q2 Men  
    Q3 • Q1 = 2.72 (1.56-4.64)  
    Q4 = highest • Q2 = 1.32 (0.73-2.41)  
D & B score = 11     • Q3 = 1.61 (0.90-2.87)  
     • Q4 = 1.00 (referent)  
   • Canadian Fitness Survey    
    Cox proportional hazard ratio model Women  
     • Q1 = 2.26 (1.15-4.43)  
     • Q2 = 2.24 (1.07-4.67)  
     • Q3 = 1.27 (0.59-2.72)  
     • Q4 = 1.00 (referent)  
     Push-ups  
     Men  
     • Q1 = 1.25 (0.77-2.05)  
     • Q2 = 1.17 (0.71-1.90)  
     • Q3 = 0.94 (0.55-1.62)  
     • Q4 = 1.00 (referent)  
     Women  
     • Q1 = 0.61 (0.32-1.17)  
     • Q2 = 0.81 (0.45-1.47)  
     • Q3 = 0.87 (0.48-1.58)  
     • Q4 = 1.00 (referent)  
     Grip strength (kg)  
     Men  
     • Q1 = 1.49 (0.86-2.59)  
     • Q2 = 1.42 (0.82-2.45)  
     • Q3 = 1.59 (0.95-2.68)  
     • Q4 = 1.00 (referent)  
     Women  
     • Q1 = 1.08 (0.58-1.99)  
     • Q2 = 0.62 (0.44-1.56)  
     • Q3 = 1.25 (0.70-2.23)  
     • Q4 = 1.00 (referent)  
     Sit and reach (cm)  
     Men  
     • Q1 = 1.06 (0.64-1.74)  
     • Q2 = 1.01 (0.61-1.66)  
     • Q3 = 1.20 (0.74-1.95)  
     • Q4 = 1.00 (referent)  
     Women  
     • Q1 = 1.18 (0.66-2.10)  
     • Q2 = 1.07 (0.60-1.91)  
     • Q3 = 0.77 (0.44-1.46)  
     • Q4 = 1.00 (referent)  
Andersen et al 2000 [163] To evaluate the relationship between levels of OPA, LTPA, cycling to work and sports participation and all-cause mortality. • n = 30,640 (17,265 men; 13,375 women) 14.5 year follow-up • 8,549 deaths LTPA was inversely associated with all-cause mortality in both men and women in all age groups.
    PA assessment: Questionnaire for LTPA, divided into: Incidence of all-cause mortality and PA  
Denmark   • Sex: Men and women    
Prospective cohort   • Age: 20-93 years (yr)   Multivariate adjusted RR (95% CI)  
    G1 = Low   
   • Characteristics: Participants of the Copenhagen City Heart Study, Glostrup Population Study and Copenhagen Male Study G2 = Moderate   
D & B score = 13    G3 = High Age 20-44 yr  
     Men  
     • G1 = 1.00 (referent)  
     • G2 = 0.73 (0.56-0.96)  
     • G3 = 0.74 (0.55-1.01)  
     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.75 (0.54-1.04)  
     • G3 = 0.66 (0.42-1.05)  
     Age 45-64 yr  
     Men  
     • G1 = 1.00 (referent)  
     • G2 = 0.75 (0.67-0.84)  
     • G3 = 0.75 (0.67-0.85)  
     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.73 (0.65-0.83)  
     • G3 = 0.66 (0.56-0.77)  
     Age >65 yr  
     Men  
     • G1 = 1.00 (referent)  
     • G2 = 0.62 (0.53-0.73)  
     • G3 = 0.60 (0.50-0.72)  
     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.52 (0.45-0.61)  
     • G3 = 0.49 (0.39-0.61)  
     All age groups  
     Men  
     • G1 = 1.00 (referent)  
     • G2 = 0.72 (0.66-0.78)  
     • G3 = 0.71 (0.65-0.78)  
     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.65 (0.60-0.71)  
     • G3 = 0.59 (0.52-0.67)  
Barengo et al 2004 [164] To investigate whether moderate or high LTPA are associated with reduced CVD and all-cause mortality, independent of CVD risk factors and other forms of PA in men and women. • n = 31,677 (15,853 men; 16,824 women) 20 year follow-up HRR (95% CI) Moderate and high levels of LTPA and OPA are associated with reduced premature all-cause mortality.
   • Sex: Men and women PA assessment: Questionnaire self administered to measure OPA, LTPA and commuting activity LTPA  
Finland   • Age: 30-59 yr      • 1.00 (referent) = low  
   • Characteristics: Participants from eastern and south-western Finland      • 0.91 (0.84-0.98) = mod, Men  
Prospective cohort      
        • 0.79 (0.70-0.90) = high, Men  
D & B score = 14        • 0.89 (0.81-0.98) = mod, women  
        • 0.98 (0.83-1.16) = high, women  
     OPA  
        • 1.00 (referent) = low  
        • 0.75 (0.68-0.83) = mod, men  
        • 0.77 (0.71-0.84) = active, men  
        • 0.79 (0.70-0.89) = mod, women  
        • 0.78 (0.70-0.87) = active, women  
Bath 2003 [165] To examine differences between older men and women on the self-rated health mortality relationship. • n = 1,042 (406 men; 636 women at baseline) Baseline, 4 and 12 years post Number of deaths: At 4 years 242 (106 men; 136 women) The self-rated health-mortality relationship can be explained by health and related factors among older men and women.
UK   • Sex: Men and women   • At 12 years 665 (287 men; 378 women)  
Prospective cohort   • Age: >65 yr    
   • Characteristics: Community-dwelling Elderly General physical health 14-item health index (Ebrahin et al 1987) scoring from 0-14 (no health problems -- multiple health problems) Multivariate adjusted HR (95% CI)  
D & B score = 11      
   • The Nottingham Longitudinal Study of Activity and Ageing    
     Men after 4 years  
     • High = 1.00 (referent)  
     • Med = 1.19 (0.61-2.33)  
    PA assessment: Self-rated health surveys, divided into 3 levels of PA: • Low = 1.51 (0.75-3.03)  
    High Women after 4 years  
    Medium • High = 1.00 (referent)  
    Low • Med = 1.03 (0.58-1.82)  
     • Low = 1.51 (0.86-2.67)  
     Men after 12 years  
    Cox proportional hazards regression Models • High = 1.00 (referent)  
     • Med = 1.28 (0.94-1.74)  
     • Low = 1.13 (0.82-1.55)  
     Women after 12 years  
     • High = 1.00 (referent)  
     • Med = 1.20 (0.90-1.61)  
     • Low = 1.23 (0.93-1.62)  
Bijnen et al 1998 [166] To describe the association between PA and mortality (CVD, stroke, all-cause) in elderly men. • n = 802 10 year follow-up • 373 deaths PA may protect against all- cause mortality in elderly men
   • Sex: Men    
   • Age: 64-84 yr PA assessment: Questionnaire, divided into groups: Multivariate adjusted RR (95% CI)  
Netherlands   • Characteristics: Retired Dutch men    
     • G1 = 1.00 (referent)  
Prospective cohort    G1 = Lowest • G2 = 0.80 (0.63-1.02)  
    G2 = Middle • G3 = 0.77 (0.59-1.00)  
    G3 = Highest p = 0.04  
D & B score = 12      
Blair et al 1993 [167] To evaluate the relationship of sedentary living habits to all-cause mortality in women. • n = 3,120 Baseline and 8 year follow-up • 43 deaths There is a graded inverse relationship between PF and all-cause mortality in women.
   • Sex: Women    
   • Age: Not available   Age adjusted death rates (per 10,000 person years) by fitness  
USA   • Characteristics: Participants were given a preventative medicine examination PF assessment: PF measured via maximal treadmill exercise test;   
Prospective        • Low Fitness = 40 The lack of relationship between PA and death rate was believed to be due to an inadequate assessment of PA.
        • Mod Fitness = 16  
D & B score = 14        • High Fitness = 7  
    PA assessment: Questionnaire   
     No difference between levels of PA  
Blair et al 1996 [168] To review the association of PF to all-cause and CVD mortality. • n = 32,421 (25,341 men; 7,080 women) Baseline and average 8 year follow-up (range 0.1-19.1 years) • 601 deaths in men The study observed a steep inverse gradient of death rates across low, moderate and high PF levels. The association was strong and remained after adjustment for potential confounding factors.
     • 89 deaths in women  
   • Sex: Men and women    
USA   • Age: 20-80 yr (mean 43 yr)   RR (95% CI) in low PF vs.  
    PF assessment: Treadmill test; duration was used to assign participants to sex specific groups: high PF  
Prospective cohort   • Characteristics: Participants were excluded if they did not reach 85% of their age predicted maximal heart rate on the maximal exercise treadmill test   Men  
     • 1.52 (1.28-1.82)  
     Women  
D & B score = 14     • 2.10 (1.36-3.26)  
    Low (least fit 20%) Adjusted deaths per 10,000 person years according to PF  
    Moderate (next 40%)   
    High (most fit 40%) Men  
   • Aerobics Center Longitudinal Study Proportional hazard modeling • Low = 49  
     • Med = 27  
     • High = 23  
     Women  
     • Low = 29  
     • Med = 13  
     • High = 14  
Boyle et al 2007 [169] To examine the association between PA and the risk of incident disability, including impairment in activities of daily living and instrumental activities of daily living in community based older persons free from dementia. • n = 1,020 2.6 year follow-up • 156 deaths The risk of death decreased 11% with each hour of PA/wk.
   • Sex: Men and women    
   • Age: 54-100 yr PA assessment: Questionnaire, hr/wk of PA Incidence of all-cause mortality HR for all-cause mortality  
USA   • Characteristics: Participants from 40 retirement communities across Chicago   The risk of death was 11% lower for each hr/wk of PA  
Prospective cohort      
D & B score = 13   • Rush Memory and Aging Project    
Bucksch et al 2005 [170] To examine the effect of moderately intense PA on all-cause mortality. • n = 7,187 (3,742 men; 3,445 women) Baseline (1984-1986) and 12-14 yr follow-up (1998) • 943 deaths Participants who achieved recommended amounts of MPA or VPA were at a significantly lower risk of death than their sedentary counterparts.
   • Sex: Men and women   RR (95% CI) for achieving recommended PA vs. not achieving recommendation  
Germany   • Age: 30-69 yr    
Prospective cohort   • Characteristics: Participants were healthy and physically active during leisure time PA assessment: Questionnaire (Minnesota Leisure Time Physical Activity questionnaire) divided into groups based on: Achieving recommended amount of MPA (30 min, 5 d/wk (≥2.5 h/wk))   
     Women  
     • MPA = 0.65 (0.51-0.82)  
D & B score = 13     • VPA = 0.78 (0.57-1.08)  
     • MPA or VPA = 0.60 (0.47-0.75)  
     Men  
     • MPA = 0.90 (0.77-1.01)  
     • VPA = 0.74 (0.61-0.90)  
     • MPA or VPA = 0.80 (0.68-0.94)  
    Achieving recommended amount of VPA (20 min, 3 d/wk (≥ 1 h/wk))   
     RR (95% CI) for volume of lifestyle activities (kcal/kg/wk)  
    Volume of lifestyle activities (kcal/kg/wk) Women  
    G1 = 0 • G1 = 1.00 (referent)  
    G2 = <14 • G2 = 0.79 (0.57-1.08)  
    G3 = 14-33.5 • G3 = 0.68 (0.50-0.94)  
    G4 = ≥ 33.5 • G4 = 0.57 (0.41-0.79)  
     p < 0.001  
     Men  
     • G1 = 1.00 (referent)  
     • G2 = 0.98 (0.76-1.17)  
     • G3 = 0.80 (0.63-1.00)  
     • G4 = 0.91 (0.74-1.13)  
     p = 0.20  
     Adjusted for age, other recommendation, social class, smoking, BMI, cardio risk factor index, alcohol intake, chronic disease index and dietary factors  
Bucksch and Helmert 2004 [171] To examine LTPA and premature death in the general population of former West Germany. • n = 7,187 (3,742 men; 3,445 women) Baseline (1984-1986) and 12-14 year follow-up (1998) • 943 deaths LTPA is inversely associated with all-cause mortality in men and women.
   • Sex: Men and women   RR (95% CI)  
   • Age: 30-69 yr   Men, LTPA  
Germany   • Characteristics: Participants were selected on the basis of the German Cardiovascular Prevention Study PA assessment: Questionnaire (Minnesota Leisure Time Physical Activity questionnaire) divided into groups based on: LTSA (h/wk) • G1 = 1.00 (referent)  
     • G2 = 0.85 (0.78-0.93)  
Prospective cohort     • G3 = 0.64 (0.50-0.82)  
     • G4 = 0.70 (0.54-0.91)  
     p < 0.001  
D & B score = 14   • The National Health Survey of the German Federal Institute of Population Research (1984-1998)   Men, LTPA index  
    G1 = 0 • G1 = 1.00 (referent)  
    G2 = <1 • G2 = 0.92 (0.70-1.23)  
    G3 = 1-2 • G3 = 0.89 (0.69-1.17)  
    G4 = >2 • G4 = 0.61 (0.44-0.84)  
     p <0.01  
    The LTSA-index (kcal/kg/wk)   
    G1 = 0 Women, LTPA  
    G2 = 1-10 • G1 = 1.00 (referent)  
    G3 = 10-25 • G2 = 0.93 (0.82-1.04)  
    G4 = >25 • G3 = 0.69 (0.48-0.98)  
     • G4 = 0.57 (0.35-0.94)  
    Mortality -- Records from the mandatory population registries p < 0.01  
     Women, LTPA index  
     • G1 = 1.00 (referent)  
    Cox proportional hazard regression model • G2 = 0.68 (0.45-1.01)  
     • G3 = 0.79 (0.51-1.21)  
     • G4 = 0.46 (0.25-0.85)  
     p < 0.01  
     Adjusted for age, social class, smoking, BMI, cardio risk factor index, alcohol intake, chronic disease index and dietary factors  
Carlsson et al 2006 [172] To investigate the association between PA and mortality in post-menopausal women. • n = 27,734 Baseline (1997) and 2-7 year follow-up (1999-2004) • 1,232 deaths The study indicates that even fairly small amounts of activity will reduce mortality in older women.
   • Sex: Women    
   • Age: 51-83 yr   RR (95% CI) adjusted for lifestyle and medical problems  
Sweden   • Characteristics: Women who participated in a population based Screening programme in 1987    
Prospective cohort    PA assessment: Questionnaires for: METs/day, different PA (walking/biking), LTPA, OPA, household PA, TV watching and reading   
     PA (METs/day)  
     • >50 = 1.00 (referent)  
D & B score = 12     • 45-50 = 1.05 (0.77-1.42)  
   • The Swedish Mammography Cohort   • 40-45 s = 1.09 (0.81-1.46)  
     • 45-40 = 1.26 (0.94-1.70)  
     • <35 = 2.56 (1.85-3.53)  
    Mortality -- Records from the National Population Register   
     Different PA  
     Walking/biking (min/d)  
     • > 90 = 1.00 (referent)  
     • 60-90 = 1.01 (0.76-1.34)  
     • 40-60 = 0.92 (0.70-1.20)  
     • 20-40 = 0.96 (0.75-1.23)  
     • <20 = 1.16 (0.90-1.50)  
     • Almost never = 1.94 (1.51-2.50)  
     LTPA (hr/wk)  
     • >5 = 1.00 (referent)  
     • 4-5 = 0.95 (0.74-1.22)  
     • 2-3 = 1.02 (0.83-1.26)  
     • 1 = 1.09 (0.88-1.36)  
     • <1 = 1.91 (1.56-2.35)  
     OPA  
     • Heavy manual labour = 1.00 (referent)  
     • Walking/lifting/ a lot carrying = 0.96 (0.55-1.70)  
     • Walking/lifting/ not a lot carrying = 1.00 (0.60-1.68)  
     • Mostly standing = 0.91 (0.52-1.61)  
     • Seated 50% of time = 0.97 (0.58-1.62)  
     • Mostly sedentary = 1.93 (1.15-3.25)  
     Household work (hr/d)  
     • >8 h/d = 1.00 (referent)  
     • 7-8 = 0.68 (0.49-0.93)  
     • 5-6 = 0.66 (0.51-0.87)  
     • 3-4 = 0.83 (0.64-1.06)  
     • 1-2 = 0.89 (0.69-1.15)  
     • <1 = 1.73 (1.30-2.32)  
     Adjusted for age  
Crespo et al 2002 [173] To study the relationship between PA and obesity with all- cause mortality in Puerto Rican men. • n = 9,136 (1962-1965) Baseline and 12 year follow-up • 1,445 deaths Some PA is better than none in protecting against all-cause mortality. The benefits are independent of body weight.
Puerto Rico   • Sex: Men PA assessment: Questionnaire, divided into 4 groups based on METs Multivariate OR (95% CI) adjusted for age  
    G1 = low   
    G2   
    G3   
    G4 = high   
Prospective cohort   • Age: 35-79 yr Multivariate logistic function model • C1 = 1.00 (referent)  
D & B score = 12   • Characteristics: Participants with no known coronary heart disease   • C2 = 0.67 (0.57-0.78)  
   • The Puerto Rico Heart Health Program   • C3 = 0.63 (0.54-0.74)  
     • C4 = 0.54 (0.46-0.64)  
     p < 0.0001  
     Multivariate adjusted OR (95% CI)  
     • C1 = 1.00 (referent)  
     • C2 = 0.68 (0.58-0.79)  
     • C3 = 0.63 (0.54-0.75)  
     • C4 = 0.55 (0.46-0.65)  
     p < 0.0001  
Davey Smith et al 2000 [174] To examine the relationship of PA and various causes of death. • n = 6,702 (at baseline) Baseline (1969-1970) and 25 year follow-up • 926 deaths In the study, an inverse association of both LTPA and walking pace with mortality from all-causes was seen.
UK   • Sex: Men PA assessment: Questionnaire with 3 groups for walking pace (Slower, same, faster) and 3 groups for LTPA (inactive, moderately active, active) Age adjusted RR (95% CI) for walking pace  
Prospective cohort   • Age: 40-64 yr   • Slower = 2.47 (2.2-2.8)  
D & B score = 13   • Characteristics: Participants from rural northern Japan   • Same = 1.35 (1.2-1.5)  
   • Whitehall study   • Faster = 1.00 (referent) p < 0.001  
     Fully adjusted RR (95% CI) for walking pace  
     • Slower = 1.87 (1.6-2.1)  
     • Same = 1.21 (1.1-1.3)  
     • Faster = 1.00 (referent) p < 0.001  
     Age adjusted RR (95% CI) for LTPA  
     • Inactive = 1.44 (1.3-1.6)  
     • Mod = 1.13 (1.0-1.2)  
     • Active = 1.00 (referent) p < 0.001  
     Fully adjusted RR (95% CI) for LTPA  
     • Inactive = 1.20 (1.1-1.3)  
     • Mod = 1.07 (1.0-1.2)  
     • Active = 1.00 (referent) p < 0.001  
Eaton et al 1995 [175] To determine whether self-reported PA predicts a decreased rate of CHD and all- cause mortality in middle aged men. • n = 8,463 21 year follow-up • 2,593 deaths Baseline levels of self- reported LTPA predicted a decreased rate of CHD and all-cause mortality.
Europe, Israel, mid eastern Asia, Northern Africa   • Sex: Men PA assessment: Questionnaire for LTPA Age adjusted RR (95% CI) LTPA  
Prospective cohort   • Age: ≥40 yr G1 = Sedentary • G1 = 1.00 (referent)  
D & B score = 12   • Characteristics: Government employees without known CVD G2 = Light • G2 = 0.84 (0.74-0.94)  
    G3 = Light daily • G3 = 0.81 (0.73-0.90)  
    G4 = Heavy • G4 = 0.84 (0.72-0.98)  
     OPA  
    Questionnaire for OPA • G1 = 1.00 (referent)  
    G1 = Sitting • G2 = 0.99 (0.88-1.12)  
    G2 = Standing • G3 = 1.09 (0.99-1.20)  
    G3 = Walking • G4 = 1.16 (1.03-1.30)  
    G4 = Physical labour   
Fang et al 2005 [176] To assess the association of exercise and CVD outcome among persons with different blood pressure status. • n = 9,791 (3,819 men; 5,972 women) 17 year follow-up Incidence of all-cause mortality and PA A significant effect of exercise on mortality in normotensive subjects was not found.
USA   • Sex: Men and women PA assessment: Questionnaire with 3 groups Multivariate adjusted HR (95% CI)  
Prospective cohort   • Age:25-74 yr G1 = Least exercise • G1 = 1.00 (referent)  
D & B score = 12   • Characteristics: Non- institutionalized participants G2 = Moderate exercise • G2 = 0.75 (0.53-1.05)  
    G3 = Most exercise • G3 = 0.71 (0.45-1.12)  
Fried et al 1998 [177] To determine the disease, functional and personal characteristics that jointly predict mortality. • n = 5,886 5 year follow-up • 646 deaths PA was a predictor of 5-year mortality.
USA   • Sex: Men and women PA assessment: Self reported exercise (5 groups) Incidence of all-cause mortality and PA  
Prospective cohort   • Age: ≥65 yr MPA or VPA (kJ/wk) Multivariate adjusted RR (95% CI)  
D & B score = 11   • Characteristics: Community dwelling elders G1 = ≤282 • G1 = 1.00 (referent)  
    G2 = 283-1789 • G2 = 0.78 (0.60-1.00)  
    G3 = 1790-4100 • G3 = 0.81 (0.63-1.05)  
    G4 = 4101-7908 • G4 = 0.72 (0.55-0.93)  
    G5 = >7908 • G5 = 0.56 (0.43-0.74) p < 0.005  
Fujita et al 2004 [178] To examine the relationship between walking duration and all-cause mortality in a Japanese cohort. • n = 41,163 (20,004 men; 21,159 women) Baseline (1990) and 11 year follow-up (2001) • 1,879 deaths Time spent walking was associated with a reduced risk for all-cause mortality.
Japan   • Sex: Men and women PA assessment: Questionnaire Walking, 3 levels: Age and sex adjusted RR (95% CI) for time spent walking (hr/d)  
    G1 = ≤30 min   
    G2 = 30 min to 1 hr   
    G3 = ≥1 hr   
Prospective cohort   • Age: 40-64 yr Cox proportional hazard model Whole group  
D & B score = 13   • Characteristics: Healthy, sedentary   • G1 = 1.22 (1.09-1.35)  
     • G2 = 1.09 (0.95-1.22)  
     • G3 = 1.00 (referent) p < 0.001  
     Men only  
     • G1 = 1.14 (1.00-1.30)  
     • G2 = 1.03 (0.90-1.19)  
     • G3 = 1.00 (referent p = 0.061  
     Women only  
     • G1 = 1.40 (1.16-1.68)  
     • G2 = 1.23 (1.01-1.49)  
     • G3 = 1.00 (referent) p < 0.001  
     RR (95% CI) for time spent walking (hr/d) (adjusted for age, education, marital status, past history of diseases, smoking, drinking, BMI and dietary variables)  
     Whole group  
     • G1 = 1.17 (1.04-1.31)  
     • G2 = 1.06 (0.93-1.20)  
     • G3 = 1.00 (referent) p = 0.011  
     Men  
     • G1 = 1.08 (0.94-1.25)  
     • G2 = 0.98 (0.84-1.14)  
     • G3 = 1.00 (referent) p = 0.318  
     Women  
     • G1 = 1.38 (1.12-1.70)  
     • G2 = 1.24 (1.00-1.54)  
     • G3 = 1.00 (referent) p < 0.001  
Glass et al 1999 [179] To examine any association between social activity, productive activity and PA and mortality in older people. • n = 2,761 (1,169 men; 1,143 women) 13 year follow-up Incidence of all-cause mortality by fitness activity quartile More active elderly people were less likely to die than those who were less active.
USA   • Sex: Men and women PA assessment: Interview, Amount of activity 13 yr mortality by amount of activity  
Prospective cohort   • Age: ≥ 65 yr G1 = Low • G1 = 74.0  
D & B score = 12   • Characteristics: Healthy elders G2 = Low-medium • G2 = 69.8  
    G3 = Medium-high • G3 = 62.4  
    G4 = High • G4 = 55.2  
Gulati et al 2003 [180] To determine whether exercise capacity is a predictor for all-cause mortality in asymptomatic women. • n = 5,721 Baseline (1992) and 8 year follow-up (2000) • 180 deaths This study confirmed that exercise capacity is an independent predictor of death in asymptomatic women, greater than what has been previously established among men.
USA   • Sex: Women PF Assessment: Treadmill stress test Exercise capacity (METs) G1 = <5 G2 = 5-8 G3 = >8 For every 1 MET increase there was a reduced death risk of 17% (p < 0.001)  
Prospective cohort   • Age: Mean 52 ± 11 yr   Age-adjusted RR  
D & B score = 11   • Characteristics: Asymptomatic women   • G1 = 2.0 (1.3-3.2)  
   • St James Women Take Heart Project   • G2 = 1.6 (1.1-2.4)  
     • G3 = 1.00 (referent)  
     Adjusted for Framingham Risk Score  
     • G1 = 3.1 (2.1-4.8)  
     • G2 = 1.9 (1.3-2.9)  
     • G3 = 1.00 (referent)  
Haapanen et al 1996 [181] To examine the association between LTPA and all-cause mortality. • n = 1,072 Baseline and a 10 yr 10 month follow-up • 168 deaths Low PA is a risk factor for all-cause mortality.
Finland   • Sex: Men PA assessment: Self-reported LTPA, divided into 4 groups by EE (kJ/wk) G1 = 0-3349 G2 = 3350-6279 G3 = 6280-8791 G4 = >8791 RR (95% CI) according to EE group  
Prospective cohort   • Age: 35-63 yr Mortality--National Death Index search • G1 = 2.74 (1.46-5.14)  
D & B score = 14   • Characteristics: Healthy, sedentary Cox proportional HR • G2 = 1.10 (0.55-2.21)  
     • G3 = 1.74 (0.87-3.50)  
     • G4 = 1.00 (referent)  
Hakim et al 1998 [182] To examine the association between walking and mortality in retired men. • n = 707 Baseline and 12 yr follow-up • 208 deaths The findings in older physically capable men indicate that regular walking is associated with a lower overall mortality rate.
USA   • Sex: Men   RR (95% CI) according to distance walked  
Prospective cohort   • Age: 61-81 yr   Adjusted for age  
D & B score = 12   • Characteristics: Retired non-smoking men who were physically capable of participating in low intensity activities on a daily basis PA assessment: Questionnaire Distance walked (miles/day) • G1 vs. G3 = 1.9 (1.3-2.9)  
    G1 = 0.0-0.9 • G1 vs. G3 = 1.6 (1.2-2.2)  
    G2 = 1.0-2.0 • G2 vs. G3 = 1.2 (0.8-1.7)  
    G3 = 2.1-8.0 Trend p = 0.002  
   • Honolulu Heart Program    
     Adjusted for risk factors  
     • G1 vs. G3 = 1.8 (1.2-2.7)  
     • G1 vs. G2 = 1.5 (1.1-2.1)  
     • G2 vs. G3 = 1.1 (0.8-1.7)  
     Trend p = 0.01  
Hillsdon et al 2004 [183] To examine whether VPA is associated with all-cause mortality. • n = 10,522 (4,929 men; 5,593 women) >10 year follow-up • 825 deaths Questionnaire respondents who reported engaging in VPA less than twice a week experienced a 37% reduced risk of all-cause mortality compared with respondents who reported a lower frequency of VPA.
   • Sex: Men and women PA assessment: Questionnaire for frequency of VPA Age and sex adjusted RR (95% CI)  
UK   • Age: 35-64 yr G1 = Never, <1 time/month   
   • Characteristics: Healthy, sedentary G2 = <2 times/wk • G1 = 1.00 (referent)  
Prospective Cohort   • OXCHECK study G3 = >2 times/wk • G2 = 0.57 (0.42-0.79)  
     • G3 = 0.72 (0.54-0.95)  
D & B score = 11     Fully adjusted RR (95% CI)  
     • G1 = 1.00 (referent)  
    Mortality -- Recorded from the Office of National Statistics • G2 = 0.63 (0.45-0.89)  
     • G3 = 0.81 (0.60-1.09)  
    Cox proportional HR   
Hu et al 2005 [184] To examine the association of PA and BMI and their combined effect with the risk of total, CVD and cancer mortality. • n = 47,212 (22,528 men; 24,684 women) 17.7 year follow-up • 7,394 deaths Regular PA is an important indicator for decreased risk of all-cause mortality. PA has a strong independent effect on mortality.
   • Sex: Men and women    
Finland   • Age:25-64 yr PA assessment: Questionnaire for PA level, divided into 3 groups Adjusted HR (95% CI)  
   • Characteristics: Participants from eastern Finland   Men  
Prospective cohort     • G1 = 1.00 (referent)  
     • G2 = 0.74 (0.68-0.81)  
    G1 = Low • G3 = 0.63 (0.58-0.70)  
D & B score = 12    G2 = Moderate Trend p = <0.001  
    G3 = High   
     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.64 (0.58-0.70)  
     • G3 = 0.58 (0.52-0.64)  
     Trend p = <0.001  
Hu et al 2004 [185] To examine the association of BMI and PA with death. • n = 116,564 Baseline (1976) and • 10,282 deaths Reduced PA is a strong and independent predictor of death.
   • Sex: Women 24 year follow-up   
   • Age: 30-55 yr   Multivariate RR (95% CI) by PA (hr/wk)  
USA   • Characteristics: Females free of known CVD and cancer PA assessment: Questionnaire for PA level, divided into 3 groups (hr/week) • G1 = 1.00 (referent)  
    G1 = ≥ 3.5 • G2 = 1.18 (1.10-1.26)  
Prospective cohort    G2 = 1.0-3.4 • G3 = 1.52 (1.41-1.63)  
D & B score = 11    G3 = <1.0 Multivariate RR (95% CI) by PA adjusted for BMI  
     • G1 = 1.00 (referent)  
    BMI (kg/m2) • G2 = 1.14 (1.06-1.22)  
    G1 = <25 • G3 = 1.44 (1.34-1.55)  
    G2 = 25-29   
    G3 = 30   
    Cox proportional HR   
Kampert et al 1996 [186] To examine PF and PA in relation to all-cause and cancer mortality. • n = 32,421 (25,341 men; 7,080 women) Baseline (1970) and ~8 year follow-up (1989) • 690 deaths The data support the hypothesis that an active and fit way of life delays death.
   • Sex: Men and women   Adjusted RR (95% CI) by quintiles of activity  
USA   • Age: 20-88 yr (mean ~43)    
Prospective cohort   • Characteristics: Predominantly white and from the middle and upper socioeconomic strata PA assessment: Questionnaire, divided into quintiles of activity (min/wk) Men  
     • Sedentary = 1.00 (referent)  
     • C1-2 = 0.71 (0.58-0.97)  
D & B score = 13     • C3 = 0.83 (0.59-1.16)  
    Male activity categories • C4 = 0.57 (0.30-1.08)  
     • C5 = 0.92 (0.29-2.88)  
    Sedentary = 855 Trend p = 0.011  
    C1-2 = 1,072   
    C3 = 1,292 Women  
    C4 = 1,453 • Sedentary = 1.00 (referent)  
    C5 = 1,601 • C1-2 = 0.68 (0.39-1.17)  
     • C3 = 0.39 (0.09-1.65)  
    Females activity categories • C4-5 = 1.14 (0.27-4.80)  
    Sedentary = 605 Trend p = 0.217  
    C1-2 = 792   
    C3 = 979   
    C4-5 = 1,158   
    Cox proportional HR   
Kaplan et al 1996 [187] To assess LTPA and its association with all cause mortality. • n = 6,131 (3298 men; 2833 women) 28 year follow-up • 1,226 deaths The data provide further support for the importance of PA and indicate that the protective effect of PA is a robust one.
   • Sex: Men and women PA assessment: Three questions about PA, with scores 0 (never), 2 (sometimes) or 4 (often). Incidence of all-cause mortality and PA  
USA   • Age: 16-94 yr    
   • Characteristics: Northern Californian adults    
Prospective cohort     Death rates/1000 person years  
     Men  
D & B score = 13     • T1 = 24.68  
    Tertiles of PA score • T2 = 11.37  
    T1 = 0-2 • T3 = 7.59  
    T2 = 4-6 Women  
    T3 = 8-12 • T1 = 18.03  
     • T2 = 7.66  
     • T3 = 3.88  
Khaw et al 2006 [188] To examine the relationship between PA patterns over 1 year and total mortality. • n = 22,191 (9,984 men; 12,207 women) 8 year follow-up • 1,553 deaths Even very moderate levels of usual PA are associated with reductions in mortality.
   • Sex: Men and women PA assessment: Questionnaire, divided into 4 groups of PA Incidence of all-cause mortality and PA  
UK   • Age: 45-79 yr   Adjusted RR (95% CI)  
   • Characteristics: Community living participants   All  
Prospective cohort    G1 = Inactive • G1 = 1.00 (referent)  
    G2 = Moderately inactive • G2 = 0.83 (0.73-0.95)  
D & B score = 13     • G3 = 0.68 (0.58-0.80)  
    G3 = Moderately active • G4 = 0.68 (0.57-0.81)  
    G4 = Active Age <65  
     • G1 = 1.00 (referent)  
     • G2 = 1.01 (0.78-1.31)  
     • G3 = 0.81 (0.62-1.07)  
     • G4 = 0.82 (0.62-1.09)  
     Age >65  
     • G1 = 1.00 (referent)  
     • G2 = 0.77 (0.66-0.91)  
     • G3 = 0.65 (0.53-0.79)  
     • G4 = 0.64 (0.50-0.80)  
Kohl et al 1996 [189] To determine the association of maximal exercise hemodynamic responses with risk of all-cause mortality. • n = 26,621 (20,387 men; 6,234 women) Average 8.1 year follow-up • 348 deaths in men and 66 in women The results suggest an exaggerated SBP or an attenuated heart rate response to maximal exercise may indicate an elevated risk for mortality.
   • Sex: Men and women    
USA   • Age: Male mean 42.2 yr; female mean 41.9 Yr   Adjusted RH (95% CI) by maximal exercise test HR  
Prospective cohort     Men  
   • Characteristics: Apparently healthy patients of a preventive medicine centre PF assessment: Maximal exercise test HR (bpm), divided into 4 Groups: • Q1 = 1.00 (referent)  
    G1 = <171 • Q2 = 0.61 (0.44-0.85)  
D & B score = 12    G2 = 171-178 • Q3 = 0.69 (0.51-0.93)  
    G3 = 179-188 • Q4 = 0.60 (0.41-0.87)  
    G4 = >188 Trend p<0.05  
     Women  
     • Q1 = 1.00 (referent)  
     • Q2 = 1.23 (0.65-2.32)  
     • Q3 = 0.69 (0.30-1.63)  
     • Q4 = 0.71 (0.22-2.24)  
     Trend p>0.05  
Kujala et al 1998 [190] To investigate LTPA and mortality in a cohort of twins. • n = 15,902 (7,925 men; 7,977 women) Baseline 1975 and death outcome from 1977-1994 • 1,253 deaths LTPA is associated with reduced mortality, even after genetic and other familial factors are taken into account.
   • Sex: Men and women   HR (95% CI)  
Finland   • Age: 25-64 yr    
   • Characteristics: Healthy, Finnish same sex twins PA assessment: Questionnaire, quintiles of fitness in MET hours/day Adjusted for age and sex  
Prospective cohort     • Sedentary = 1.00 (referent)  
     • OE = 0.71 (0.62-0.81)  
   • The Finnish Twin Cohort   • CE = 0.57 (0.45-0.74)  
D & B score = 13    Q1 = <58 Trend p = 0.001  
    Q2 = 59-1.29   
    Q3 = 1.30-2.49 Adjusted for age, sex, smoking  
    Q4 = 2.50-4.49   
    Q5 = >4.50 • Sedentary = 1.00 (referent)  
     • OE = 0.76 (0.67-0.87)  
    Categorized into: • CE = 0.68 (0.53-0.88)  
    -Sedentary   
    -Occasional exerciser (OE) Trend p = 0.001  
    -Conditioning exerciser (CE) Adjusted for age, sex, smoking, occupational group, alcohol  
     • Sedentary = 1.00 (referent)  
     • OE = 0.80 (0.69-0.91)  
     • CE = 0.76 (0.59-0.98)  
     Trend p = 0.002  
     HR (95% CI) among 434 same sex twin pairs compared with sedentary category in 1975  
     • Sedentary = 1.00 (referent)  
     • OE = 0.66 (0.46-0.94)  
     • CE = 0.44 (0.23-0.83)  
     Trend p = 0.005  
     Adjusted for smoking  
     • Sedentary = 1.00 (referent)  
     • OE = 0.70 (0.48-1.01)  
     • CE = 0.56 (0.29-1.09)  
     Trend p = 0.04  
     Adjusted for smoking, occupational group, alcohol  
     • Sedentary = 1.00 (referent)  
     • OE = 0.73 (0.50-1.07)  
     • CE = 0.56 (0.29-1.11)  
     Trend p = 0.06  
     OR (95% CI) in quintiles among 434 same sex twin pairs compared with sedentary category in 1975  
     • Q1 = 1.00 (referent)  
     • Q2 = 0.85  
     • Q3 = 0.72  
     • Q4 = 0.68  
     • Q5 = 0.60  
LaCroix et al 1996 [191] To determine whether walking is associated with a reduced risk of CVD hospitalization and death in older adults. • n = 1,645 (615 men; 1030 women) 4.2 year follow-up RR (95% CI) by category of walking Walking more than 4 hr/wk was associated with a reduced risk of mortality from all-causes.
   • Sex: Men and women PA assessment: Questionnaire for walking h/wk, divided into 3 groups   
USA   • Age: ≥65 yr G1 = <1 hr/week Men  
   Characteristics: Participants from a group health co-operative G2 = 1-4 hr/week • G1 = 1.00 (referent)  
Prospective cohort    G3 = >4 hr/week • G2 = 0.78 (0.43-1.45)  
     • G3 = 0.89 (0.49-1.62)  
D & B score = 12     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.50 (0.28-0.90)  
     • G3 = 0.48 (0.25-0.83)  
     Age 65-74 yr  
     • G1 = 1.00 (referent)  
     • G2 = 0.81 (0.40-1.61)  
     • G3 = 1.13 (0.60-2.15)  
     Age ≥75 yr  
     • G1 = 1.00 (referent)  
     • G2 = 0.63 (0.37-1.08)  
     • G3 = 0.46 (0.25-0.84)  
     High functioning  
     • G1 = 1.00 (referent)  
     • G2 = 0.73 (0.38-1.41)  
     • G3 = 0.89 (0.48-1.65)  
     Limited functioning  
     • G1 = 1.00 (referent)  
     • G2 = 0.60 (0.34-1.05)  
     • G3 = 0.51 (0.28-0.92)  
Lam et al 2004 [192] To investigate the relationship LTPA and mortality in Hong Kong. • n = 24,079 cases (13,778 men; 10,301 women); 10 years prior Multivariate adjusted OR (95% CI) by LTPA The data confirm and extend previous findings in Caucasian populations on the association between LTPA and longevity.
    PA assessment: Men  
Hong Kong   • n = 13,054 controls (3,918 men; 9,136 women) Questionnaire for LTPA, divided into 3 groups • G1 = 1.00 (referent)  
     • G2 = 0.60 (0.54-0.67)  
Case-Control     • G3 = 0.66 (0.60-0.73)  
   • Sex: Men and women G1 = <1 times per month   
D & B score = 12   • Age: ≥35 yr   Women  
   • Characteristics: All ethnic Chinese G2 = 1-3 times per month • G1 = 1.00 (referent)  
     • G2 = 0.81 (0.74-0.88)  
    G3 = ≥4 times per month • G3 = 0.71 (0.66-.077)  
Lan et al 2006 [193] To investigate the relationship between exercise and all-cause mortality. • n = 2,113 (1,081 men; 1,032 women) Baseline and 2 year follow-up • 197 deaths Older persons are recommended to expend at least 1000 kcal/wk through regular exercise for mortality reduction.
   • Sex: Men and women   HR (95% CI) by LTPA frequency  
Taiwan   • Age: ≥65 yr PA assessment: Questionnaire for LTPA (frequency/wk)   
Prospective cohort   • Characteristics: Non-institutionalized elders   Adjusted for age and sex Protection of exercise against death also increases with the number of activities.
    G1 = Sedentary • G1 = 1.00 (referent)  
   • Taiwan National Health Interview Survey G2 = 1 time/wk • G2 = 0.49 (0.36-0.67)  
D & B score = 13    G3 = ≥2 times/wk • G3 = 0.20 (0.09-0.46)  
     Trend p = <0.001  
    Questionnaire for EE (kcal/wk), divided into 5 groups: Multivariate adjusted  
     • G1 = 1.00 (referent)  
    G1 = Sedentary • G2 = 0.70 (0.50-0.98)  
    G2 = <500 • G3 = 0.35 (0.15-0.82)  
    G3 = 500-999 Trend p = 0.014  
    G4 = 1000-1999   
    G5 = ≥2000   
     HR (95% CI) by EE  
     Adjusted for age and sex  
     • G1 = 1.00 (referent)  
     • G2 = 0.64 (0.41-1.01)  
     • G3 = 0.55 (0.35-0.85)  
     • G4 = 0.30 (0.17-0.53)  
     • G5 = 0.24 (0.12-0.48)  
     Trend p <0.001  
     Multivariate adjusted  
     • G1 = 1.00 (referent)  
     • G2 = 0.80 (0.49-1.30)  
     • G3 = 0.74 (0.46-1.17)  
     • G4 = 0.50 (0.27-0.90)  
     • G5 = 0.43 (0.21-0.87)  
     Trend p = 0.043  
Laukkanen et al 2001 [194] To examine the relationship between maximal oxygen uptake and overall mortality. • n = 1,294 Baseline and 10.7 year follow-up • 124 deaths PF has a strong, graded, inverse association with overall mortality.
   • Sex: Men   Adjusted RR (95% CI) by quartile  
Finland   • Age: 42.0-61.3 yr (mean 52.1)    
   • Characteristics: Men free from CVD, COPD, and cancer at baseline PF assessment: Exercise tolerance test, 4 groups by maximal oxygen uptake (ml/kg/min)   
Prospective cohort     Maximal oxygen uptake  
     • G1 = 1.00 (referent)  
     • G2 = 1.47 (0.71-3.01)  
D & B score = 14     • G3 = 2.79 (1.44-5.39)  
    G1 = >37.1 • G4 = 3.85 (2.02-7.32)  
    G2 = 32.3-37.1 Linear trend p = <0.001  
    G3 = 27.6-32.2   
    G4 = <27.6 Test duration  
     • G1 = 1.00 (referent)  
    Test duration (min) • G2 = 2.22 (1.08-4.55)  
    G1 = >11.2 • G3 = 2.23 (1.11-4.49)  
    G2 = 9.6-11.2 • G4 = 3.94 (2.01-7.74)  
    G3 = 8.2-9.5 Linear trend p<0.001  
    G4 = <8.2   
Lee and Paffenbarger 2000 [195] To compare various levels of PA with mortality. • n = 13,485 Baseline and 15 year follow-up • 2,539 deaths The study provides some support for recommendations that emphasize MPA. A benefit of VPA is also evident.
   • Sex: Men    
   • Age: Mean 57.5 yr   RR (95% CI)  
   • Characteristics: Men who matriculated as undergraduates in 1916-1950 PA assessment: • G1 = 1.00 (referent)  
USA    Questionnaires for LTPA index (including walking, stair climbing, sports and recreational activity), • G2 = 0.80 (0.72-0.88)  
     • G3 = 0.74 (0.65-0.83)  
Prospective cohort     • G4 = 0.80 (0.69-0.93)  
   • The Harvard Alumni Health Study   • G5 = 0.73 (0.64-0.84)  
     Trend p = <0.001  
D & B score = 12    5 groups (kJ/wk)   
    G1 = <4200   
    G2 = 4200-8399   
    G3 = 8400-12599   
    G4 = 12600-16799   
    G5 = ≥ 16800   
Lee et al 1995 [196] To examine the independent association of vigorous and non-vigorous PA with longevity. • n = 17,321 Follow-up 22-26 years • 3,728 deaths There is a graded inverse relationship between PA and mortality. Vigorous, but not non-vigorous activities are associated with longevity.
   • Sex: Men    
   • Age: Mean 46 yr PA assessment: Questionnaires for EE (kJ/wk), quintiles RR (95% CI) by EE (kJ/wk)  
USA   • Characteristics: Harvard University alumni, without self-reported physician diagnosed cardiovascular disease, cancer or chronic obstructive pulmonary disease   Q1= 1.00 (referent)  
     • Q2 = 0.94 (0.86--1.04)  
Prospective cohort    Q1 = ≤ 630 • Q3 = 0.95 (0.86--1.05)  
    Q2 = 630-1680 • Q4 = 0.91 (0.83 - 1.01)  
    Q3 = 1680-3150 • Q5 = 0.91 (0.82-1.00)  
D & B score = 12    Q4 = 3150-6300   
    Q5 = >6300 RR (95% CI) by EE (Vigorous activity, kJ/wk)  
     • Q1 = 1.00 (referent)  
   • The Harvard Alumni Health Study   • Q2 = 0.88 (0.82-0.96)  
     • Q3 = 0.92 (0.82-1.02)  
     • Q4 = 0.87 (0.77-0.99)  
     • Q5 = 0.87 (0.78-0.97)  
Lee et al 2004 [197] To investigate the effect of various PA patterns on all-cause mortality. • n = 8,421 Baseline 1988 and follow-up 1993 • 1,234 deaths The results suggest that regular PA generating 1000 kcal/wk or more should be recommended for lowering mortality rates. Among those with no major risk factors, even 1-2 episodes per week generating 1000 kcal or more can postpone mortality.
   • Sex: Men    
   • Age: Mean 66 yr   Age adjusted RR (95% CI) by PA pattern  
USA   • Characteristics: Participants free of major chronic disease PA assessment: Questionnaire for PA (kcal/wk), 4 groups   
     • G1 = 1.00 (referent)  
Prospective cohort     • G2 = 0.75 (0.63-0.90)  
    G1 = <500 • G3 = 0.82 (0.63-1.07)  
   • The Harvard Alumni Health Study (Sedentary) • G4 = 0.61 (0.53-0.69)  
D & B score = 11    G2 = 500-999   
    (Insufficiently active) Multivariate adjusted  
    G3 = ≥ 1000   
    (Weekend warrior) • G1 = 1.00 (referent)  
    G4 = Regularly active • G2 = 0.75 (0.62-0.91)  
     • G3 = 0.85 (0.65-1.11)  
     • G4 = 0.64 (0.55-0.73)  
Leitzmann et al 2007 [198] To examine PA guidelines in relation to mortality. • n = 252,925 (142,828 male; 110,097 women) Baseline and 6 month follow-up • 7,900 deaths Following PA guidelines is associated with lower risk of death. Mortality benefit may also be achieved by engaging in less than recommended activity levels.
USA   • Sex: Men and women PA assessment: Questionnaire for MPA and VPA, 5 groups each MPA (h/wk) Multivariate adjusted RR (95% CI) according to activity  
   • Age: 50-71 yr   MPA  
Prospective cohort   • Characteristics: Participants free of CVD, cancer or emphysema   • G1 = 1.00 (referent)  
   • The National Institute of Health-American Association of Retired Persons   • G2 = 0.85 (0.79-0.93)  
     • G3 = 0.79 (0.74-0.85)  
D & B score = 13    G1 = sedentary • G4 = 0.76 (0.71-0.82)  
    G2 = <1 • G5 = 0.68 (0.63-0.74)  
    G3 = 1-3 Trend p = <0.001  
    G4 = 4-7 VPA  
    G5 = >7   
    VPA (frequency/wk) • G1 = 1.00 (referent)  
    G1 = inactive • G2 = 0.77(0.71-0.83)  
    G2 = <1 • G3 = 0.77 (0.72-0.82)  
    G3 = 1-2 • G4 = 0.68 (0.63-0.73)  
    G4 = 3-4 • G5 = 0.71 (0.66-0.77)  
    G5 = ≥ 5 Trend p = <0.001  
    Cox proportional HR   
Leon et al 1997 [199] To examine the long-term association of LTPA and risk of death from coronary heart disease and all-causes. • n = 12,138 16 year follow-up • 1,904 deaths The data suggest that a relatively small amount of daily moderate intensity LTPA can reduce premature mortality in middle-aged and older men at high risk for CHD.
   • Sex: Men    
   • Age: 35-57 yr PA assessment: Minnesota LTPA questionnaire, categorized by frequency/month and average duration, deciles (min/d) Multivariate adjusted RR (95% CI) by deciles of LTPA  
USA   • Characteristics: Men who at entry to the study were free of clinical evidence of CHD or other serious medical problems but were at the upper 10%-15% of a CHD probability score distribution derived from the FHS data    
Prospective cohort     • D1 = 1.00 (referent)  
     • D2-4 = 0.85 (0.73-0.99)  
     • D5-7 = 0.87 (0.75-1.02)  
D & B score = 12     • D8-10 = 0.83 (0.71-0.97)  
    D1 = 4.9   
    D2-4 = 22.7   
    D5-7 = 53.9   
    D8-10 = 140.4   
   • Multiple Risk Factor Intervention Trial Cox proportional HR   
Lissner et al 1996 [200] To examine the relationship of OPA and LTPA on all-cause mortality in women. • n = 1,405 Baseline and 20 year follow-up • 277 deaths Decreases in PA as well as low initial levels are strong risk factors for mortality.
   • Sex: Women    
   • Age: 38-60 yr   RR (95% CI) by LTPA  
Sweden   • Characteristics: Free from major disease at baseline PA assessment: Questionnaire for OPA and LTPA, 3 groups   
     20 year follow-up  
Prospective cohort     LTPA during age 20-38 years  
   • The Gothenburg Prospective Study of Women   • Low = 1.00 (referent)  
    G1 = Low • Med = 0.66 (0.34-1.26)  
D & B score = 10    G2 = Medium • High = 0.46 (0.21-1.01)  
    G3 = High   
     LTPA during age 39-60 years  
    Proportional hazard regression • Low = 1.00 (referent)  
     • Med = 0.56 (0.35-0.90)  
     • High = 0.44 (0.22-0.91)  
     LTPA during the past 12 months  
     • Low = 1.00 (referent)  
     • Med = 0.56 (0.39-0.82)  
     • High = 0.45 (0.24-0.86)  
     20 year follow-up  
     OPA during age 20-38 years  
     • Low = 1.00 (referent)  
     • Med = 0.59 (0.18-1.87)  
     • High = 0.50 (0.16-1.58)  
     OPA during age 39-60 years  
     • Low = 1.00 (referent)  
     • Med = 0.66 (0.21-2.08)  
     • High = 0.47 (0.14-1.52)  
     OPA during the past 12 months  
     • Low = 1.00 (referent)  
     • Med = 0.28 (0.17-0.46)  
     • High = 0.24 (0.14-0.43)  
Manini et al 2006 [201] To determine whether energy expenditure is associated with all-cause mortality in older adults. • n = 302 (150 men; 152 women) Mean follow-up of 6.15 years • 55 deaths Free-living activity EE was strongly associated with lower risk of mortality.
   • Sex: Men and women   HR (95% CI) by tertiles of PA EE  
USA   • Age: 70-82 yr PA assessment: Questionnaire, divided into tertiles of PA EE (kcal/d)   
Prospective cohort   • Characteristics: High-functioning community dwelling elders   Adjusted for age, sex, race and study site  
    T1 = <521 • T1 = 1.00 (referent)  
D & B score = 13    T2 = 521-770 • T2 = 0.63 (0.29-1.18)  
    T3 = >770 • T3 = 0.37 (0.15-0.76)  
     Trend p = 0.009  
     Adjusted for age, sex, race, study site, weight, height, percent body fat and sleep duration  
     • T1 = 1.00 (referent)  
     • T2 = 0.57 (0.30-1.09)  
     • T3 = 0.31 (0.14-0.69)  
     Trend p = 0.004  
     Adjusted for age, sex, race, study site, self rated health, education, smoking, CVD, lung disease, diabetes, hip or knee osteoarthritis, osteoporosis, cancer and depression  
     • T1 = 1.00 (referent)  
     • T2 = 0.65 (0.33-1.28)  
     • T3 = 0.33 (0.15-0.74)  
     Trend p = 0.007  
Matthews et al 2007 [202] To determine the effects of exercise and non-exercise PA on mortality. • n = 67,143 Baseline and an average of 5.7 year follow-up • 1,091 deaths Overall PA levels are an important determinant of longevity.
   • Sex: Women    
   • Age: 40-70 yr   RR (95% CI)  
China   • Characteristics: Women without heart disease, stroke or cancer    
    PA assessment: Interview to report (MET h/d), 4 groups Overall activity Multivariate adjustment  
Prospective cohort     Overall activity (MET hr/d)  
     • G1 = 1.00 (referent)  
   • The Shanghai Women's Health Study   • G2 = 0.81 (0.69-0.96)  
D & B score = 12    G1 = ≤ 9.9 • G3 = 0.67 (0.57-0.80)  
    G2 = 10.0-13.6 • G4 = 0.61 (0.51-0.73)  
    G3 = 13.7-18.0 Trend p = 0.000  
    G4 = ≥ 18.1   
     Adult exercise (MET hr/d)  
    Adult exercise • G1 = 1.00 (referent)  
    G1 = 0 • G2 = 0.84 (0.74-0.96)  
    G2 = 0.1-3.4 • G3 = 0.77 (0.59-0.99)  
    G3 = 3.5-7.0 • G4 = 0.64 (0.36-1.14)  
    G4 = ≥ 7.1 Trend p = 0.008  
    Cox proportional hazard models   
Menotti and Seccareccia 1985 [203] To investigate the relationship between OPA and all-cause mortality. • n = 99,029 Baseline and 5 year follow-up • 2,661 deaths The results suggest that PA may play a role in the prediction of fatal events.
   • Sex: Men    
   • Age: 40-59 yr    
   • Characteristics: Men employed on the Italian railway system PA assessment: Questionnaire Men at risk classified by 3 levels of PA and 3 levels of job responsibility, combined to create 8 groups of PA-job responsibility Age adjusted death rates per 1000 over 5 years classified by PA only  
Italy     • Sedentary = 26.20  
Prospective cohort     • Moderate = 27.05  
     • Heavy = 27.35  
D & B score = 12     Age adjusted death rates per 1,000 over 5 years classified by PA and job responsibility  
    G1 = sedentary -- low • G1 = 30.00  
    G2 = sedentary -- med • G2 = 25.20  
    G3 = sedentary -- high • G3 = 25.80  
    G4 = moderate -- low • G4 = 26.30  
    G5 = moderate -- med • G5 = 28.50  
    G6 = moderate -- high • G6 = 25.80  
    G7 = heavy -- low • G7 = 26.90  
    G8 = heavy -- med • G8 = 30.80  
Mensink et al 1996 [204] To compare various indices for PA and their association with cardiovascular risk factors as well as total and CVD mortality. • n = 15,436 (7,689 men; 7797 women) 5-8 year follow-up Incidence of all-cause mortality and PA An inverse relation of PA and total mortality.
Germany   • Sex: Men and women PA assessment: Questionnaire Total activity, 3 groups Adjusted RR (95% CI)  
   • Age: 25-69 yr    
Prospective cohort   • Characteristics: Participants from communities in Western Germany   Total activity, men  
    G1 = Low • G1 = 1.00 (referent)  
    G2 = Moderate • G2 = 0.56 (0.30-1.04)  
D & B score = 12    G3 = High • G3 = 0.78 (0.42-1.44)  
     Total activity, women  
    LTPA, 3 groups • G1 = 1.00 (referent)  
    G1 = Low • G2 = 1.24 (0.60-2.58)  
    G2 = Moderate • G3 = 1.29 (0.58-2.85)  
    G3 = High   
    Conditioning activity, 3 groups LTPA, men  
     • G1 = 1.00 (referent)  
    G1 = No activity • G2 = 0.61 (0.35-1.05)  
    G2 = Moderate • G3 = 0.79 (0.48-1.31)  
    G3 = High LTPA, women  
     • G1 = 1.00 (referent)  
    Sports activity, 4 groups • G2 = 0.94 (0.51-1.75)  
     • G3 = 0.81 (0.44-1.49)  
    G1 = no sports   
    G2 = <1 hour Conditioning activity, men  
    G3 = 1-2 hours • G1 = 1.00 (referent)  
    G4 = >2 hours • G2 = 0.76 (0.44-1.34)  
     • G3 = 0.67 (0.36-1.25)  
     Conditioning activity, women  
     • G1 = 1.00 (referent)  
     • G2 = 0.38 (0.13-1.06)  
     • G3 = 0.80 (0.42-1.54)  
     Sports Activity, men  
     • G1 = 1.00 (referent)  
     • G2 = 0.49 (0.26-0.95)  
     • G3 = 0.57 (0.30-1.09)  
     • G4 = 0.36 (0.16-0.79)  
     Sports activity, women  
     • G1 = 1.00 (referent)  
     • G2 = 0.38 (0.12-1.23)  
     • G3 = 0.52 (0.23-1.17)  
     • G4 = 0.28 (0.07-1.17)  
Morgan and Clarke 1997 [205] To assess the value of broadly based customary PA scores in predicting 10-year mortality in elderly people. • n = 1,042 (407 men; 635 women) 10 year follow-up Incidence of all-cause mortality and PA A wide range of customary or habitual PA, can provide indices showing both cross sectional and predictive validity for 10 year mortality.
   • Sex: Men and women PA assessment: Questionnaire for PA, 3 groups   
UK   • Age: ≥65 yr   HR (95% CI)  
   • Characteristics: British elders   Men  
Prospective cohort    G1 = Low • G1 = 1.59 (1.12-2.25)  
   • Nottingham Longitudinal Study of Activity and Aging G2 = Intermediate • G2 = 1.35 (0.96-1.89)  
    G3 = High • G3 = 1.00 (referent)  
D & B score = 12     Women  
     • G1 = 2.07 (1.53-2.79)  
     • G2 = 1.53 (1.12-2.09)  
     • G3 = 1.00 (referent)  
Myers et al 2002 [206] To compare PF and PA levels with all-cause mortality. • n = 6,213 Baseline and mean 6.2 ± 3.7 year follow-up • 1,256 deaths Exercise capacity is a more powerful predictor of mortality among men than other established risk factors for CVD.
   • Sex: Men    
   • Age: Mean 59 ± 11 yr   Age adjusted RR (95% CI) by quintile  
USA   • Characteristics: Participants with a normal exercise test result (n = 2,534) and participants with an abnormal exercise test or CVD or both (n = 3,679)    
    PF assessment: Treadmill test for VO2 peak, divided into quintiles (METs) • Q1 = 4.5 (3.0-6.8)  
Prospective cohort     • Q2 = 2.4 (1.5-3.8)  
     • Q3 = 1.7 (1.1-2.8)  
     • Q4 = 1.3 (0.7-2.2)  
D & B score = 12    Q1 = Lowest level • Q5 = 1.00 (referent)  
    1.0-5.9   
    Q2   
    Q3   
    Q4   
    Q5 = Highest level   
    ≥13.0   
Ostbye et al 2002 [207] To analyze the effect of smoking and other modifiable risk factors on ill health, defined in a multidimensional fashion. • n = 12,956 6 year follow-up • 782 deaths Quitting smoking and increasing exercise levels are the lifestyle interventions most likely to improve overall health.
   • Sex: Men and women    
   • Age: 50-60 yr PA assessment: Questionnaire for PA, 4 groups Incidence of all-cause mortality and PA  
USA   • Characteristics: Participants from the Health and Retirement Study (HRS) only    
Prospective cohort    G1 = Sedentary Death rates (95% CI) per 1000 population/yr  
    G2 = Light   
    G3 = Moderate • G1 = 20.6 (17.8-24.0)  
D & B score = 13    G4 = Heavy • G2 = 9.1 (8.1-9.5)  
     • G3 = 8.3 (7.5-9.2)  
     • G4 = 4.4 (3.5-5.6)  
Paffenbarger et al 1994 [208] To study the adoption or maintenance of PA and other optional lifestyle patterns for their influence on mortality rates of Harvard College alumni. • n = 14,786 Follow-up between • 2,343 deaths Adopting a physically active lifeway delays mortality and extends longevity.
   • Sex: Men 1977 and 1988   
   • Age: 45-84 yr (in 1977)   RR (95% CI) of mortality according to PA  
USA    PA assessment: Questionnaire for blocks walked daily, stairs climbed daily and type, frequency and duration of weekly sports and recreational activities   
   Characteristics: Harvard College alumni    
Prospective cohort     Physical activity index (kcal/wk)  
     • G1 = 1.00 (referent)  
D & B score = 14     • G2 = 1.13 (1.01-1.26)  
     • G3 = 0.72 (0.64-0.82)  
     • G4 = 0.77 (0.69-0.85)  
    Physical activity index (kcal/wk) Sports and recreational activities were scored according to intensity and duration Walking (km/wk)  
     • G1 = 1.00 (referent)  
     • G2 = 1.21 (1.08-1.35)  
     • G3 = 0.94 (0.83-1.07)  
     • G4 = 0.89 (0.78-1.01)  
     Moderately vigorous sports play (METs)  
    Light < 4.5 METs   
    Moderate ≥ 4.5 METs   
     • G1 = 1.00 (referent)  
     • G2 = 1.11 (0.93-1.33)  
     • G3 = 0.73 (0.65-0.81)  
     • G4 = 0.72 (0.64-0.80)  
     Adjusted for potential confounding influences  
Richardson et al 2004 [209] To investigate the impact of a sedentary lifestyle on all-cause mortality. • n = 9,611 (4,642 men; 4,969 women) Baseline (1992) and 8 year follow-up • 810 deaths A sedentary lifestyle is associated with a higher risk of death in pre- retirement aged adults.
   • Sex: Men and women   OR (95% CI)  
USA   • Age: 51-61 yr PA assessment: Questionnaire for PA, 3 groups: • G1 = 1.00 (referent)  
Prospective cohort   • Characteristics: Participants born between 1931-1941 and who not institutionalized in 1992   • G2 = 0.64 (0.52-0.81)  
    G1 = Sedentary • G3 = 0.62 (0.44-0.85)  
    G2 = occasional or light p = 0.01  
D & B score = 13    G3 = Regular MVPA   
   • Health and Retirement Study    
Rockhill et al 2001 [210] To determine the association between recreational PA and mortality in women. • n = 80,348 Baseline (1980) and follow-up between 1982-1996 • 4,871 deaths People who are more physically active are at reduced mortality risk relative to those who are less active.
   • Sex: Women    
   • Age: 30-55 yr   Multivariate adjusted RR (95% CI) by (hr/wk)  
USA   • Characteristics: Free from CVD or cancer at baseline    
   • Nurses Health Study PA assessment: Questionnaire in 1980 and up-dated every 2- 4 years, 5 groups of PA (hr/wk) • G1 = 1.00 (referent)  
Prospective cohort     • G2 = 0.82 (0.76-0.89)  
     • G3 = 0.75 (0.69-0.81)  
     • G4 = 0.74 (0.68-0.81)  
D & B score = 11     • G5 = 0.71 (0.61-0.82)  
     p<0.001  
    G1 = <1   
    G2 = 1-1.9   
    G3 = 2-3.9   
    G4 = 4-6.9   
    G5 = ≥7   
Rosengren and Wilhelmsen 1997 [211] To investigate the effect of OPA and LTPA on risk of death. • n = 7,142 Baseline (1970-1973) and 20 year follow-up • 2,182 deaths The study demonstrates the protective effect of LTPA on mortality.
   • Sex: Men    
   • Age: 47-55 yr   Unadjusted RR (95% CI)  
   • Characteristics: Without symptomatic CHD PA assessment: Postal questionnaires, 3 groups: • G1 = 1.00 (referent)  
Sweden     • G2 = 0.74 (0.68-0.82)  
     • G3 = 0.73 (0.68-0.79)  
Prospective cohort    G1 = Sedentary   
    G2 = Moderately active Multivariate adjustment  
    G3 = Regular exercise • G1 = 1.00 (referent)  
D & B score = 13     • G2 = 0.84 (0.77-0.93)  
     • G3 = 0.83 (0.77-0.90)  
Schnohr et al 2003 [212] To assess the associations of regular LTPA and changes in LTPA with risk of death. • n = 7,023 (4,471 men; 5,676 women) 18 year follow-up • 2,725 deaths Maintaining or adopting a moderate or high degree of PA was associated with lower risk of death.
   • Sex: Men and women PA assessment: Questionnaire, 9 groups Incidence of all-cause mortality and PA and changes in PA  
Denmark   • Age: 20-79 yr    
   • Characteristics: Participants from the Copenhagen City Heart Registered Population    
Prospective cohort    G1 = Low--low   
    G2 = Low--moderate Adjusted RR (95% CI)  
    G3 = Low--high Men  
D & B score = 12    G4 = Moderate- low • G1 = 1.00 (referent)  
    G5 = Moderate-Moderate • G2 = 0.64 (0.49-0.83)  
     • G3 = 0.64 (0.47-0.87)  
    G6 = Moderate-high • G4 = 0.73 (0.56-0.96)  
    G7 = High-low • G5 = 0.71 (0.57-0.88)  
    G8 = High-moderate • G6 = 0.64 (0.51-0.81)  
    G9 = High-high • G7 = 1.11 (0.76-1.62)  
     • G8 = 0.66 (0.51-0.85)  
     • G9 = 0.61 (0.48-0.76)  
     Women  
     • G1 = 1.00 (referent)  
     • G2 = 0.75 (0.57-0.97)  
     • G3 = 0.72 (0.50-1.05)  
     • G4 = 0.70 (0.54-0.91)  
     • G5 = 0.64 (0.52-0.79)  
     • G6 = 0.58 (0.45-0.73)  
     • G7 = 0.72 (0.48-1.07)  
     • G8 = 0.61 (0.47-0.80)  
     • G9 = 0.66 (0.51-0.85)  
Schnohr et al 2004 [213] To examine whether the relationship between established risk factors and mortality differs with socioeconomic status as measured by level of education. • n = 30,635 (16,236 men; 14,399 women) 16 year follow-up • 10,952 deaths The study shows the strong predictive effect of PA on mortality is independent of education level.
   • Sex: Men and women Socioeconomic status assessment: level of education Incidence of all-cause mortality and PA stratified by years of education  
Denmark   • Age: 20-93 yr    
   • Characteristics: Participants from the Copenhagen City Heart Registered Population    
Prospective cohort    PA assessment: Questionnaire Deaths <8 years of education  
D & B score = 12     Men  
    4 groups of PA G1 = 916  
    G1 = none or very little G2 = 1693  
    G2 = 2-4 h/wk of LPA G3 = 1012  
    G3 = >4 h/wk of LPA or 2-4 h/wk of high level activity G4 = 67  
    G4 = Competition level or >4 h/wk of hard level activity Women  
     • G1 = 872  
     • G2 = 1298  
     • G3 = 346  
     • G4 = 10  
     8-11 years of education  
     Men  
     • G1 = 432  
     • G2 = 1040  
     • G3 = 616  
     • G4 = 33  
     Women  
     • G1 = 363  
     • G2 = 852  
     • G3 = 268  
     • G4 = 10  
     >11 years of education  
     Men  
     • G1 = 104  
     • G2 = 302  
     • G3 = 182  
     • G4 = 11  
     Women  
     • G1 = 48  
     • G2 = 129  
     • G3 = 61  
     • G4 = 3  
Schnohr et al 2006 [214] To investigate the association between LTPA and mortality. • n = 4,894 (2,136 men; 2,758 women) Baseline (1976) and start of follow-up in 1981-1983 (to 2000) • 1,787 deaths Long-term moderate or high PA was associated with significantly lower mortality in men and women.
   • Sex: Men and women   RR (95% CI)  
Denmark   • Age: 20-79 yr    
   • Characteristics: Healthy males and women   Unadjusted  
Prospective cohort    PA assessment: Survey for LTPA, 3 groups: • G1 = 1.00 (referent)  
     • G2 = 0.64 (0.56-0.73)  
   • The Copenhagen City Heart Study   • G3 = 0.56 (0.48-0.65)  
D & B score = 13    G1 = Low Trend p < 0.001  
    G2 = Mod   
    G3 = High Multivariate adjustment  
     • G1 = 1.00 (referent)  
     • G2 = 0.78 (0.68-0.89)  
     • G3 = 0.75 (0.64-0.87)  
     Trend p = 0.001  
Schooling et al 2006 [215] To examine how a Comprehensive assessment of baseline health status affects the relationship between obesity or PA and mortality. • n = 54,088 (17,849 men; 36,239 women) 4.1 year follow-up • 3,819 deaths PA, which normally has a negative relationship with adiposity, had the largest impact on survival for the health states, with the strongest inverse relationship between BMI and mortality.
   • Sex: Men and women PA assessment: Interview for PA min/d, 3 groups Incidence of all-cause mortality and PA  
Hong Kong   • Age: ≥ 65 yr    
Prospective cohort   • Characteristics: Chinese elders G1 = None Adjusted HR (95% CI)  
    G2 = ≤ 30 min/d • G1 = 1.00 (referent)  
    G3 = ≥ 30 min/d • G2 = 0.83 (0.76-0.91)  
D & B score = 13     • G3 = 0.73 (0.67-0.80)  
     Trend p<0.001  
Sundquist et al 2004 [216] To study the association between varying levels of PA and all-cause mortality in the elderly. • n = 3,206 (1,414 men; 1,792 women) Baseline (1988-1989) and follow-up in 2000 • 1,806 deaths Even occasional PA decreases the risk of mortality among elderly people.
   • Sex: Men and women PA assessment: Questionnaire for PA, 5 groups Age-adjusted HR (95% CI)  
Sweden   • Age: ≥65 yr   Men  
   • Characteristics: Non-institutionalized elders   • G1 = 1.00 (referent)  
Prospective cohort     • G2 = 0.74 (0.62-0.87)  
    G1 = none • G3 = 0.57 (0.44-0.73)  
   The Swedish Annual Level-of-Living Survey (Statistics Sweden) G2 = occasionally • G4 = 0.51 (0.41-0.64)  
D & B score = 12    G3 = once per week • G5 = 0.60 (0.44-0.82)  
    G4 = twice per week Women  
    G5 = vigorously at least twice per week • G1 = 1.00 (referent)  
     • G2 = 0.70 (0.59-0.82)  
     • G3 = 0.59 (0.46-0.77)  
    Cox proportional HR • G4 = 0.47 (0.35-0.62)  
     • G5 = 0.54 (0.31-0.94)  
     Men and women  
     Multivariate adjustment  
     • G1 = 1.00 (referent)  
     • G2 = 0.72 (0.64-0.81)  
     • G3 = 0.60 (0.50-0.71)  
     • G4 = 0.50 (0.42-0.59)  
     • G5 = 0.60 (0.46-0.79)  
Talbot et al 2007 [217] To investigate how changes in LTPA affect all-cause mortality. • n = 2,092 (1,316 men; 776 women) Baseline in 1958 for males and in 1978 for females and an average follow-up of 21.2 ± 9.4 years for men and 10.2 ± 5.6 years for women • 628 deaths (538 male; 90 female) Greater declines in total and high-intensity LTPA are independent predictors of all-cause mortality.
   • Sex: Men and women    
USA   • Age: 19-<90 yr   RR (95% CI) for standard deviation of rate of change in LTPA  
Prospective cohort   • Characteristics: Community residents, generally with above average income, high education and with good or excellent self related health   (If RR is <1 then a SD increase is associated with decrease mortality. If RR is >1, then a SD increase is associated with increase in mortality)  
D & B score = 13    PA assessment: Questionnaire for LTPA (METs min/24 h), 3 groups   
   The Baltimore Longitudinal Study of Aging G1 = low   
    G2 = medium Multivariate adjustment  
    G3 = high Men <70 years  
    Rate of change (ROC) • G1 = 0.96 (0.84-1.08)  
     • G2 = 0.91 (0.79-1.04)  
     • G3 = 0.42 (0.33-0.53)  
     • ROC low = 0.90 (0.80-1.01)  
     • ROC med = 1.01 (0.90-1.14)  
     • ROC high = 0.78 (0.65-0.94)  
     Men >70 years  
     • G1 = 0.95 (0.82-1.10)  
     • G2 = 0.89 (0.76-1.05)  
     • G3 = 0.78 (0.62-0.97)  
     • ROC low = 1.07 (0.93-1.24)  
     • ROC med = 1.13 (1.00-1.27)  
     • ROC high = 0.91 (0.75-1.12)  
     Women <70 years  
     • G1 = 0.75 (0.53-1.07)  
     • G2 = 0.61 (0.36-1.03)  
     • G3 = 0.80 (0.50-1.30)  
     • ROC low = 1.02 (0.74-1.40)  
     • ROC med = 1.38 (0.86-2.28)  
     • ROC high = 0.90 (0.63-1.27)  
     Women >70 years  
     • G1 = 0.85 (0.63-1.15)  
     • G2 = 0.78 (0.39-1.59)  
     • G3 = 0.62 (0.32-1.22)  
     • ROC low = 1.10 (0.85-1.42)  
     • ROC med = 0.96 (0.46-2.03)  
     • ROC high = 0.70 (0.40-1.22)  
Trolle-Lagerros et al 2005 [218] To quantify the effect of PA on overall mortality in younger women and to assess the effect of past versus current activity. • n = 99,099 11.4 year follow-up • 1,313 deaths Current PA substantially reduces mortality among women. The association is observed even with low levels of PA and is accentuated with increased PA.
   • Sex: Women    
   • Age: 30-49 yr PA assessment: Questionnaire using a 5 point scale, 5 groups Incidence of all-cause mortality and PA past and current  
Sweden and Norway   • Characteristics: Participants from Norway and one region of Sweden    
Retrospective cohort    G1 = Sedentary Adjusted HR (95% CI)  
    G2 = Low PA at enrolment  
    G3 = Moderate • G1 = 1.00 (referent)  
D & B score = 13    G4 = High • G2 = 0.78 (0.61-1.00)  
    G5 = Vigorous • G3 = 0.62 (0.49-0.78)  
     • G4 = 0.58 (0.44-0.75)  
     • G5 = 0.46 (0.33-0.65)  
     Trend p<0.0001  
     PA at age 30 yr  
     • G1 = 1.00 (referent)  
     • G2 = 0.79 (0.55-1.15)  
     • G3 = 0.90 (0.64-1.28)  
     • G4 = 0.98 (0.68-1.42)  
     • G5 = 0.96 (0.65-1.44)  
     Trend p = 0.22  
     PA at age 14 yr  
     • G1 = 1.00 (referent)  
     • G2 = 0.95 (0.66-1.38)  
     • G3 = 0.96 (0.69-1.34)  
     • G4 = 0.88 (0.62-1.25)  
     • G5 = 1.06 (0.75-1.51)  
     Trend p = 0.62  
Villeneuve et al 1998 [219] To examine the relationship between PF, PA and all-cause mortality. • n = 14,442 (6,246 men; 8,196 women) Baseline (1981) and 7 year follow-up RR (95% CI) by EE, multivariate adjustment There was a reduction in mortality risk associated with even modest participation in activities of low intensity.
   • Sex: Men and women    
Canada   • Age: 20-69 yr PA assessment: Questionnaire for EE (kcal/kg/day), 5 groups LTPA, men  
   • Characteristics: Asymptomatic for CVD   • G1 = 1.00 (referent)  
Prospective cohort     • G2 = 0.81 (0.59-1.11)  
     • G3 = 0.79 (0.54-1.13)  
   Canadian Fitness Survey G1 = 0-<0.5 • G4 = 0.86 (0.61-1.22)  
D & B score = 11    G2 = 0.5-<1.5 • G5 = 0.82 (0.65-1.04)*  
    G3 = 1.5-<3.0   
    G4 = ≥ 3.0 Non vigorous LTPA, men  
    G5 = ≥ 0.5 • G1 = 1.00 (referent)  
    PF levels: • G2 = 0.81 (0.56-1.17)  
    Recommended • G3 = 0.70 (0.44-1.13)  
    Minimum • G4 = 0.82 (0.53-1.27)  
     • G5 = 0.78 (0.59-1.04)*  
    Undesirable Refusal   
     LTPA, women  
    Multivariate Poisson regression analysis • G1 = 1.00 (referent)  
     • G2 = 0.94 (0.69-1.30)  
     • G3 = 0.92 (0.64-1.34)  
     • G4 = 0.71 (0.45-1.11)  
     • G5 = 0.88 (0.68-1.04)*  
     Non vigorous LTPA, women  
     • G1 = 1.00 (referent)  
     • G2 = 0.97 (0.69-1.36)  
     • G3 = 0.87 (0.57-1.33)  
     • G4 = 0.72 (0.43-1.21)  
     • G5 = 0.89 (0.67-1.17)*  
     RR (95% CI) by fitness levels, adjusted for age, sex and smoking Recommended = 1.00 (referent)  
     • Minimum = 1.02 (0.69-1.51)  
     • Undesirable = 1.52 (0.72-3.18)  
     • Refusal = 1.04 (0.45-2.39)  
Weller and Corey 1998 [220] To study the relationship between PA and mortality in women. • n = 6,620 Baseline and 7 year follow-up • 449 deaths PA is inversely associated with risk of death in women.
   • Sex: Women    
   • Age: ≥;30 yr   OR (95% CI)  
Canada   • Characteristics: Without known heart disease PA assessment: Questionnaires for: EE (kcal/kg/d), quartiles   
   • Canadian Fitness Survey   EE (kcal/kg/d)  
Prospective cohort     • Q1 = 1.00 (referent)  
     • Q2 = 0.91 (0.66-1.25)  
    Q1 = lowest • Q3 = 0.94 (0.72-1.23)  
D & B score = 11    Q2 = • Q4 = 0.89 (0.67-1.17)  
    Q3 =   
    Q4 = highest LTPA levels  
    LTPA, 3 groups • G1 = 1.00 (referent)  
    G1 = Sedentary • G2 = 0.63 (0.46-0.86)  
    G2 = Mod • G3 = 0.76 (0.59-0.98)  
    G3 = High   
     Walking  
    Walking, 3 groups • G1 = 1.00 (referent)  
    G1 = < half the time • G2 = 0.64 (0.49-0.82)  
    G2 = half the time • G3 = 0.64 (0.47-0.86)  
    G3 = > half the time   
Yu et al 2003 [221] To examine the relationship between LTPA and all-cause mortality. • n = 1,975 Baseline and 10 year follow-up • 252 deaths The study found a strong inverse association between heavy LTPA and all-cause mortality.
UK   • Sex: Men    
   • Age: 49-64 yr   Age adjusted HR (95% CI)  
   • Characteristics: Without a history of CHD at baseline PA assessment: Questionnaire (Minnesota LTPA index, kcal/d), 3 group • G1 = 1.00 (referent)  
     • G2 = 0.73 (0.54-0.99)  
Prospective cohort     • G3 = 0.74 (0.55-1.04)  
     Trend p = 0.046  
D & B score = 11    G1 = Light to no activity Multivariate adjusted  
    G2 = Moderate activity • G1 = 1.00 (referent)  
    G3 = Heavy activity • G2 = 0.79 (0.58-1.08)  
     • G3 = 0.76 (0.56-1.04)  
     Trend p = 0.083  
  1. D & B score, Downs and Black quality score; PF, physical fitness; YR, years; RR, risk ratio; 95% CI, 95% confidence interval; PA, physical activity; VO2 peak, peak oxygen consumption; HR, hazard ratio; min/d, minutes per day; kcal/wk, kilocalories per week; LTPA, leisure-time physical activity; MET, metabolic equivalent; VO2 max, maximal oxygen consumption; OPA, occupational physical activity; CVD, cardiovascular disease; hr/wk, hours per week; MPA, moderate physical activity; kcal/kg/wk, kilocalories per kilogram per week; kJ/wk, kilojoules per week; EE, energy expenditure; G, groups; EE, energy expenditure; BMI, body mass index; C, class; kg/m2, kilogram by meters squared; HR, heart rate; BPM, beats per minute; MVPA, moderate to vigorous physical activity; OR, odds ratio; Q, quartile or quintile; RCT, randomized clinical trial; T, tertiles; TPA, total physical activity; VPA, vigorous physical activity; mL/kg/min, milliliters per kilogram per minute.