Key words: Mobility, muscle strength, obesity, physical activity, postural balance.Correspondence: Janne Sallinen, PhD, VTT Technical Research Centre of Finland, P.O. Box 1199, FIN-70211 Kuopio, Finland.E-mail: janne.sallinen@vtt.fiReceived May 5, 2010; accepted in revised form August 18, 2010.First published ahead of print September 28, 2010 as DOI: 10.3275/7270

Factors associated with maximal walking speedamong older community-living adults

Aging Clinical and Experimental Research

Janne Sallinen1,2, Minna Mänty1, Raija Leinonen1,3, Mauri Kallinen3, Timo Törmäkangas1,Eino Heikkinen1 and Taina Rantanen1

1The Finnish Centre for Interdisciplinary Gerontology, Department of Health Sciences, University ofJyväskylä, 2VTT Technical Research Centre of Finland, Kuopio, 3GeroCenter Foundation for Research andDevelopment, Jyväskylä, Finland

ABSTRACT. Background and aims: The relativecontribution of different domains on walking speed islargely unknown. This study investigated the centralfactors associated with maximal walking speed amongolder people. Methods: Cross-sectional analyses ofbaseline data from the SCAMOB study (ISRCTN07330512) involving 605 community-living ambulatoryadults aged 75-81 years. Maximal walking speed, legextensor power, standing balance and body mass indexwere measured at the research center. Physical activi-ty, smoking, use of alcohol, chronic diseases and de-pressive symptoms were self-reported by standardquestionnaires. Results: The mean maximal walkingspeed was 1.4 m/s (range 0.3-2.9). In linear regressionanalysis, age, gender and body mass index explained11% of the variation in maximal walking speed.Adding leg extensor power and standing balance intothe model increased the variation explained to 38%.Further adjusting for physical activity, smoking statusand use of alcohol increased the variation explained byan additional 7%. A minor further increase in vari-ability explained was gained by adding chronic diseasesand depressive symptoms to the model. In the finalmodel, the single most important factors associatedwith walking speed were leg extensor power, standingbalance and physical activity, and these associationswere similar in men and women and in different bodymass index categories. Conclusions: Lower extremityimpairment and physical inactivity were the central fac-tors associated with slow walking speed among olderpeople, probably because these factors capture theinfluences of health changes and other life-style factors,potentially leading to walking limitations.(Aging Clin Exp Res 2011; 23: 273-278)©2011, Editrice Kurtis

INTRODUCTIONWalking is a fundamental part of many actions of dai-

ly life, and loss of walking ability increases disability riskand threatens independent living in old age (1).

The Disablement Process model is widely used to de-scribe the transition from health conditions to disability (2).The main pathway in the model starts from pathology(e.g., chronic disease or injury), proceeds to the genera-tion of impairment (e.g., decreased muscle strength) andconsequent restriction in functional capacity (e.g., walkinglimitation), and ends in disability. The model also calls at-tention to extrinsic risk factors (e.g., level of physicalactivity) which speed or slow down the disablement pro-cess (2). Measured walking speed is shown to be a verysensitive and reliable tool in detecting even small changesin physical function over time (3, 4).

In compliance with the Disablement Process model (2),earlier studies have identified risk factors of slow walkingspeed to include excess body fatness (5, 6) and chronicconditions and symptoms (6-8), impaired muscle strengthand balance (6-11) and unhealthy lifestyle practices (5, 6).However, the relative contribution of these different do-mains on walking speed is largely unknown.

The purpose of this study was to examine the relativecontribution of chronic conditions, depressive symptomsand body mass index (BMI), leg extensor power (LEP) andstanding balance and life-style practices to differencesin maximal walking speed (MWS) among older commu-nity-living people.

METHODSParticipants and designWe used baseline cross-sectional data on 605 people

for analyses. The target population consisted of all reg-istered individuals aged 75-81 years in the city of

273 Aging Clin Exp Res, Vol. 23, No. 4

Factors associated with walking speed in older adults

Aging Clin Exp Res, Vol. 23, No. 4 274

Jyväskylä, Central Finland in 2003 (n=1310) who hadbeen screened for a randomized controlled trial entitledScreening and Counselling for Physical Activity and Mo-bility in Older People (SCAMOB study, ISRCTN07330512) (12). Suitability for participation was definedas the ability to walk at least 0.5 km without assistance,being only moderately physically active or sedentary,having no severe medical contraindications for physical ac-tivity, having no memory impairment, and a Mini-MentalState Examination score over 21. The selection procedurehas been described in detail elsewhere (12).

After screening, a total of 632 participants weredeemed suitable for cross-sectional analysis. We excludeda total of 27 persons with missing information abouttheir BMI (n=1), MWS (n=1), LEP (n=13) and standingbalance (n=2), or who were potentially undernourished orseverely catabolic, with BMI less than 20 kg/m2 (n=10)(13). The final sample consisted of 605 participants withBMI ranging from 20 to 47 kg/m2. The study was ap-proved by the Ethics Committee of the Central FinlandHealth Care District and conducted according to theDeclaration of Helsinki. All participants gave their writteninformed consent to participate.

MeasurementsThe MWS was measured indoors over a distance of 10

meters with an additional 2-3 meters allowed for accel-eration before the start line. Walking time was measuredwith a stopwatch. Participants were asked to walk asquickly as possible without running. Walking speed is ahighly reliable and valid predictor of incident disabilityamong older people (1). In our research center, the re-peatability of walking speed assessment was under 5%(14).

Standing balance was measured in the semi-tandem po-sition with one foot placed one-half of a foot lengthahead of the other, with feet touching. The test wasdemonstrated once and walking aids could not be used.Performance was rated according to the length of timethat the participant was able to hold the position (maxi-mum 30 sec). Balance was considered as: 1) poor if theparticipant was able to hold the position for less than 10sec, 2) moderate if the participant could hold the positionfor 10-20 sec, or 3) good, if the participant could hold theposition for longer than 20 sec.

Unilateral LEP (in watts) was measured on both sideswith a Nottingham Leg Extensor Power Rig (15). Partic-ipants were tested in a seated position with arms folded.The foot to be tested was placed on the pedal, which wasattached to a flywheel, while the other foot rested on thefloor. After a couple of practice trials, participants wereasked to push the pedal as hard and fast as possible. Dur-ing each attempt, verbal encouragement was given toeach participant, to encourage maximal effort. Trialswere repeated until no improvement occurred, but in

any case at least five times. The rest period betweenseparate attempts was 30 sec. Maximal power was de-termined as the best result of the stronger leg. In our lab-oratory, the test-retest coefficient of variation (CV%) in old-er population was 8% (16).

Weight and height were measured at the researchcenter, and BMI was calculated as weight divided byheight squared (kg/m2) (17).

Information on self-reported physical activity, smoking,use of alcohol, chronic diseases and depressive symptomswere collected by standard questionnaires, which werechecked together with a trained nurse on arrival at the re-search center. Overall physical activity was estimated ona 7-point scale (18): (a) mainly resting, (b) most activitiesperformed sitting down, (c) light physical activity 1-2h/wk, (d) moderate physical activity 3 h/wk, (e) moderatephysical activity at least 4 h/wk, and (f) strenuous physi-cal exercise several times a week, and (g) competitivesports several times a week. Participants were defined as:1) physically inactive if they reported mainly resting ormost activities were performed sitting down, 2) moderatelyactive if they engaged in light to moderate physical activity1-3 hours per week, or 3) active if they reported moder-ate to strenuous physical activity at least 4 hours perweek. The short physical activity questionnaire used herehas shown good reliability (Kendall’s tau-b=0.874) amongolder adults when duplicate interviews were carried outtwo weeks apart (12). Smoking status was categorized asnon-smoker (never smoker or former smoker whostopped more than five years ago) or smoker (currentsmoker or former smoker who stopped less than fiveyears ago). Alcohol intake was assessed by asking whetheror not participants used alcohol. Physician-diagnosedchronic conditions included musculoskeletal diseases,such as hip fracture and arthritis (no/yes), cardiovasculardisease, such as ischemic heart disease, coronary arterydisease, hypertension, heart insufficiency, cardiac in-farction or arrhythmia (no/yes), and pulmonary disease,such as asthma, bronchitis or emphysema (no/yes). De-pressive symptoms were assessed with the Center for Epi-demiologic Studies Depression (CES-D) scale (19).

Statistical analysisDifferences in the characteristics between the MWS

quartiles were examined by a Two-Way ANOVA withmain effects of BMI, LEP, standing balance and depres-sive symptoms as well as their interaction with age andgender as covariates and cross-tabulation with the linearby linear association Chi-Square test for categorical vari-ables (Table 1). The Curve Estimation Routine confirmedthat the variation in the MWS was comparably explainedby LEP (R2=28-29%) and standing balance (R2=10%) bylinear, quadratic or cubic curves (i.e., virtually identical es-timates in all three models). For the sake of model par-simony, we therefore used multiple linear regression

J. Sallinen, M. Mänty, R. Leinonen et al.

275 Aging Clin Exp Res, Vol. 23, No. 4

models to assess the explanatory factors of MWS. Thebase model included age, gender and BMI, and was ad-justed in a blockwise manner with factors that correlatedsignificantly with MWS (Table 2). A factor was consideredas a significant predictor of MWS when it had a significanteffect on the model, p<0.05. Analyses were performedwith SPSS version 15.0 (SPSS, Chicago, IL, USA).

RESULTSThe mean MWS was 1.4 m/s (SD 0.4, range 0.3-2.9)

and the mean BMI 28 kg/m2 (SD 4, range 20-47). In theage-adjusted linear regression model, gender by BMI in-teraction on MWS was non-significant (p=0.177), andtherefore men and women were included in the sameanalyses.

Table 1 - Characteristics of participants according to maximal walking speed quartiles (n=605).

Quartiles of maximal walking speedTotal p-value*

I II III IV

Age (yrs) 77.9 (2.0) 77.8 (2.0) 77.5 (1.9) 77.1 (1.8) 77.5 (1.9) 0.002Body mass index (kg/m2) 29.3 (4.5) 28.8 (4.4) 28.1 (3.8) 27.3 (3.6) 28.3 (4.1) <0.001Leg extensor power (W) 81 (46) 95 (45) 112 (54) 123 (54) 104 (53) <0.001Leg extensor power (W/kg) 1.1 (0.6) 1.3 (0.6) 1.5 (0.6) 1.7 (0.7) 1.4 (0.7) <0.001Semi-tandem balance (s) 26.0 (8.5) 29.6 (3.2) 29.8 (1.6) 29.9 (0.8) 28.9 (4.7) <0.001Physical activity

Inactive 51 21 18 7 23 <0.001Insufficiently active (max. 3 hrs/wk) 42 59 54 48 50Sufficiently active (min. 4 hrs/wk) 8 20 28 45 26

Non-smoker 89 97 94 98 94 0.012Non-drinker 47 45 42 31 41 0.006Musculoskeletal disease 68 51 44 40 50 <0.001Cardiovascular disease 63 78 69 59 67 0.212Pulmonary disease 28 15 12 14 17 0.001Depressive symptoms (CES-D scale) 12.5 (8.2) 10.5 (7.3) 9.5 (7.6) 7.7 (5.9) 9.9 (7.5) <0.001

Values are means (standard deviations) for continuous variables and prevalences as percentages for categorical variables. BMI: Body Mass Index, CES-D scale:Center for Epidemiologic Studies Depression Scale. In men, cut-points for walking speed quartiles were 1.3 m/s, 1.5 m/s and 1.8 m/s. Among women, cor-responding cut-points were 1.1 m/s, 1.3 m/s and 1.5 m/s. *Age and gender-adjusted two-way ANOVA for continuous variables and cross-tabulation with lin-ear by linear association chi-square test for categorical variables.

Model 1 (Base model) Model 2 Model 3 Model 4 (Full model)

β p R2 β p R2 β p R2 β p R2

Age (yr) -0.144 <0.001 0.114 -0.086 0.008 0.383 -0.065 0.036 0.449 -0.073 0.018 0.474Female gender -0.236 <0.001 0.149 0.001 0.131 0.001 0.151 <0.001Body mass index (kg/m2) -0.172 <0.001 -0.169 <0.001 -0.122 <0.001 -0.126 <0.001Leg power (W) 0.580 <0.001 0.488 <0.001 0.466 <0.001Standing balance (3 levels) 0.229 <0.001 0.198 <0.001 0.182 <0.001Physical activity (3 levels) 0.249 <0.001 0.211 <0.001Smoking -0.071 0.023 -0.064 0.037Use of alcohol 0.068 0.029 0.064 0.040Musculoskeletal disease -0.089 0.005Pulmonary disease -0.072 0.018Depressive symptoms -0.131 <0.001

All models were statistically significant, p<0.001. β= standardized regression coefficient, R2= coefficient of determination. Balance categories for semi-tandemstand were: poor (<10 s), moderate (10-20 s) or good (>20 s). Physical activity categories were: inactive, insufficiently active (max. 3 hours/wk) or sufficientlyactive (min. 4 hours/wk). Depression symptoms were assessed by CES-D scale. Model 1 (Base model): adjusted for body mass index, age and gender;Model 2: Model 1 + leg extensor power and semi-tandem balance; Model 3: Model 2 + physical activity, smoking and use of alcohol; Model 4 (Full model): Mod-el 3 + musculoskeletal- and pulmonary diseases and depressive symptoms.

Table 2 - Factors associated with maximal walking speed among community-living older people (n=605).

Factors associated with walking speed in older adults

Aging Clin Exp Res, Vol. 23, No. 4 276

Table 1 shows that those with low MWS were olderand heavier and had lower LEP and poorer standingbalance than those with high MWS. In addition, lowMWS correlated significantly with physical inactivity,smoking, use of alcohol, musculoskeletal and pulmonarydiseases and depressive symptoms. Cardiovascular dis-eases did not correlate with MWS.

In the multivariate regression model, age, gender andBMI explained 11% of the variation in MWS (Table 2,model 1). Adding LEP and standing balance to the mod-el increased the coefficient of determination (R2) to 38%(model 2). Further adjustments for physical activity, smok-ing status and use of alcohol increased the R2 to 45%(model 3). The final model, which included musculoskeletaland pulmonary diseases and depressive symptoms, ex-plained 47% of the variation in MWS (model 4).

In the final model, leg extensor power was the singlemost important factor explaining MWS with the higheststandardized regression coefficient (β) (model 4). Thenext most important factors associated with MWS werestanding balance and physical activity. In addition, age,gender, BMI, smoking, use of alcohol, musculoskeletal andpulmonary diseases and depressive symptoms were in-dependently associated with MWS. Body mass indexand gender did not have significant interactions on the as-sociation between leg extensor power, standing balanceor physical activity on MWS.

DISCUSSIONIn our sample of community-living ambulatory older

people, the strongest explanatory factors of MWS wereLEP, standing balance and physical activity. Chronicconditions, including musculoskeletal and pulmonary dis-eases and depressive symptoms, explained only a few ad-ditional percent of the variability in MWS.

This study is one of the few reporting the relative con-tribution of multiple domains on walking speed amongcommunity-living older adults (6-8). Depressive symp-toms, impaired cognition, dizziness, fear of falling andcardiovascular disease contributed 21-27% of the pop-ulation’s attributable risk for walking disability in a walk-ing test in the Leiden 85-plus Study (7). In the Cardio-vascular Health Study, subjective health and symptomsexplained 12-13%, physical function and activity 2-10%, and age and anthropometry 3-6% of the vari-ability in walking speed (6). However, very old personsvolunteering for the Leiden 85-plus Study (7) may rep-resent a special group of healthiest survivors, reducingthe representativeness of the sample. Second, the preva-lence of overweight and obesity was presumably lower inthe earlier samples than these days (20). Third, physicalfunction was judged based on grip strength, although im-paired knee extensor strength and poor balance areknown to be crucial determinants of walking disability(21). This study expands our understanding of central ex-

planatory factors of walking in present-day older peopleusing a multidimensional approach including measuredLEP and standing balance.

Our results on the association of standing balanceand walking speed are in line with previous research (8,10, 21), which suggests that maintaining balance in an up-right position is an immediate prerequisite for being ableto walk. The positive correlation of leg extension poweror strength has already been reported (9, 11, 22, 23). Forexample, leg extension power explained about a third ofthe variability in walking speed in the InCHIANTI study(11). However, the effect of muscle weakness on walkingability depends upon the presence of other physiologicalimpairments and the psychological reserves of the indi-vidual. For example, good knee extensor strength re-duces the risk of walking disability among individualswith poor balance while, among those with good balance,strength is not a limiting factor for walking ability (21),showing that good lower limb strength may compen-sate for other impairments the person may have. In ourstudy, we observed that standing balance and leg exten-sion power both had an independent effect on walkingspeed.

In our study, physical activity was an important ex-planatory factor of walking speed. Unfortunately, in across-sectional setting, it is difficult to know which is thecause and which the consequence. We assume that ahigher level of physical activity leads to better mobility,but equally well good mobility enables a person to bemore physically active. It is possible that there are cir-cular processes, in which a high level of physical activ-ity leads to better mobility, which makes it easier to bephysically active. Earlier findings have shown that thephysical activity level may have an important influenceon the relationship between strength and body weight.In the EPIDOS study, among physically active women,the knee extensor strength /body weight ratio, some-times termed functional strength, did not differ withBMI (24). Among inactive women, functional strengthdecreased with increasing BMI. In addition, in the activewomen, the likelihood of mobility difficulties did not in-crease with increasing BMI (24). Correspondingly, theHealth ABC Study revealed a progressive negativetrend in walking endurance (timed 400-m walk), knee ex-tensor strength and lower extremity function (EPESE bat-tery) when physical activity decreased from exerciseactivity to lifestyle activity and further to physical inac-tivity (25). Low physical activity was also the onlylifestyle factor associated with mobility loss in bothnon-obese and obese older persons in the Health ABCStudy (5).

All in all, our findings confirm that standing balance andleg extension power are important correlates of walkingspeed. In the Disablement Process model, they are alsothe most proximal variables to walking speed. It has

J. Sallinen, M. Mänty, R. Leinonen et al.

277 Aging Clin Exp Res, Vol. 23, No. 4

REFERENCES1. Guralnik JM, Ferrucci L, Pieper CF et al. Lower extremity func-

tion and subsequent disability: Consistency across studies, pre-dictive models, and value of gait speed alone compared with theshort physical performance battery. J Gerontol A Biol Sci Med Sci2000; 55: M221-31.

2. Verbrugge LM, Jette AM. The disablement process. Soc SciMed 1994; 38: 1-14.

3. Shinkai S, Watanabe S, Kumagai S et al. Walking speed as a goodpredictor for the onset of functional dependence in a Japanese ru-ral community population. Age Ageing 2000; 29: 441-6.

4. Onder G, Penninx BW, Ferrucci L, Fried LP, Guralnik JM,Pahor M. Measures of physical performance and risk for pro-gressive and catastrophic disability: Results from the Women'sHealth and Aging Study. J Gerontol A Biol Sci Med Sci 2005;60: 74-9.

5. Koster A, Penninx BW, Newman AB et al. Lifestyle factors andincident mobility limitation in obese and non-obese older adults.Obesity (Silver Spring) 2007; 15: 3122-32.

6. Hirsch CH, Fried LP, Harris T, Fitzpatrick A, Enright P, SchulzR. Correlates of performance-based measures of muscle functionin the elderly: The Cardiovascular Health Study. J Gerontol A BiolSci Med Sci 1997; 52: M192-200.

7. Bootsma-van der Wiel A, Gussekloo J, De Craen AJ, Van Exel E,Bloem BR, Westendorp RG. Common chronic diseases andgeneral impairments as determinants of walking disability in theoldest-old population. J Am Geriatr Soc 2002; 50: 1405-10.

8. Tiedemann A, Sherrington C, Lord SR. Physiological and psy-chological predictors of walking speed in older community-dwelling people. Gerontology 2005; 51: 390-5.

9. Rantanen T, Avela J. Leg extension power and walking speed invery old people living independently. J Gerontol A Biol Sci MedSci 1997; 52: M225-31.

10. Ringsberg K, Gerdhem P, Johansson J, Obrant KJ. Is there a re-lationship between balance, gait performance and muscularstrength in 75-year-old women? Age Ageing 1999; 28: 289-93.

11. Bean JF, Leveille SG, Kiely DK, Bandinelli S, Guralnik JM,Ferrucci L. A comparison of leg power and leg strength within theInCHIANTI study: Which influences mobility more? J Gerontol ABiol Sci Med Sci 2003; 58: 728-33.

12. Leinonen R, Heikkinen E, Hirvensalo M et al. Customer-ori-ented counseling for physical activity in older people: study pro-tocol and selected baseline results of a randomized-controlled tri-al (ISRCTN 07330512). Scand J Med Sci Sports 2007; 17: 156-64.

13. Freedman DM, Ron E, Ballard-Barbash R, Doody MM, Linet MS.Body mass index and all-cause mortality in a nationwide UScohort. Int J Obes (Lond) 2006; 30: 822-9.

14. Pajala S, Era P, Koskenvuo M et al. Contribution of geneticand environmental factors to individual differences in maximalwalking speed with and without second task in older women. JGerontol A Biol Sci Med Sci 2005; 60: 1299-303.

15. Bassey EJ, Short AH. A new method for measuring power out-put in a single leg extension: feasibility, reliability and validity. EurJ Appl Physiol Occup Physiol 1990; 60: 385-90.

16. Portegijs E, Sipila S, Alen M et al. Leg extension power asym-metry and mobility limitation in healthy older women. ArchPhys Med Rehabil 2005; 86: 1838-42.

17. World Health Organization. Obesity: Preventing and managingthe global epidemic. Report of a WHO consultation. WHOtechnical report series 894: Geneva, 2000. Available at:http://whqlibdoc.who.int/trs/WHO_TRS_894.pdf.

been suggested that they capture those consequences ofchronic conditions which influence walking. This is sup-ported by our study, because the variability of walkingspeed explained by variables more distal from walking wassmaller (2). Many chronic conditions common in old agemay not, as such, be cured, but their consequences maybe alleviated e.g., through increasing physical activityand particularly by targeting training to the lower ex-tremities. However, older people have special problemswhich need to be taken into account. In our previous anal-yses of the present data, we observed that poor health,fear and negative experiences and unsuitable environmentwere often perceived as constraints on physical activityamong older people (26, 27).

The cross-sectional design of the current study didnot allow cause-effect conclusions. The participants wereambulatory older adults, and the most vigorous and mostfrail were excluded from the study, which may have leadto underestimation of associations. Third, dietary infor-mation was not available and, thus, the effect of diet onwalking speed could not be determined. However, po-tentially undernourished persons with BMI less than 20were excluded from all analyses, due to their potentialhealth problems and higher risk of mobility difficulties (13).Lastly, the validity of BMI as an approximation of body fat-ness may be compromised, since changes in body com-position and fat distribution with aging are not well cap-tured by standard anthropometry (28). However, mea-suring of BMI is easy and quick to perform and it is as-sociated with walking limitations, like other obesity indi-cators (29).

The major strengths of our study were a large popu-lation-based sample of older people and measurement ofwalking speed and related physical characteristics instandardized laboratory conditions. Self-reports maybetter reflect day-to-day reality, but are prone to mem-ory bias, socially desirable responses, and the influenceof contextual factors. For example, older adults mayoverestimate their height and underestimate their weight,causing underestimation of the BMI (30). We did notneed to use, for example, grip strength as a surrogatemeasure of lower extremity strength, as we had data onleg extensor power, which is probably the most impor-tant strength measure to be studied when predictingwalking speed (11). Standardized measurement of MWSas an outcome is highly reliable and minimizes potentialreporting bias (1).

CONCLUSIONSLeg extensor power, standing balance and physical ac-

tivity were the strongest explanatory factors of maximalwalking speed among older people. The results emphasizethat promotion of lower extremity fitness, includingstrength and balance training, may be a good startingpoint for interventions to prevent disability.

Factors associated with walking speed in older adults

Aging Clin Exp Res, Vol. 23, No. 4 278

18. Grimby G. Physical activity and muscle training in the elderly. ActaMed Scand Suppl 1986; 711: 233-7.

19. Radloff LS. The CES-D scale: a self-reported depression scale forresearch in the general population. Appl Psychol Meas 1977; 1:385-401.

20. Arterburn DE, Crane PK, Sullivan SD. The coming epidemic ofobesity in elderly Americans. J Am Geriatr Soc 2004; 52: 1907-12.

21. Rantanen T, Guralnik JM, Ferrucci L et al. Coimpairments as pre-dictors of severe walking disability in older women. J Am GeriatrSoc 2001; 49: 21-7.

22. Ostchega Y, Harris TB, Hirsch R, Parsons VL, Kington R. Theprevalence of functional limitations and disability in older personsin the US: data from the National Health and NutritionExamination Survey III. J Am Geriatr Soc 2000; 48: 1132-5.

23. Ostchega Y, Dillon CF, Lindle R, Carroll M, Hurley BF. Isokineticleg muscle strength in older Americans and its relationship to astandardized walk test: data from the National Health andNutrition Examination Survey 1999-2000. J Am Geriatr Soc2004; 52: 977-82.

24. Rolland Y, Lauwers-Cances V, Cristini C et al. Disability inobese elderly women: lower limb strength and recreational activity.Obes Res Clin Practice 2007; 1: 39-51.

25. Brach JS, Simonsick EM, Kritchevsky S, Yaffe K, Newman AB,Health, Aging and Body Composition Study Research Group. Theassociation between physical function and lifestyle activity and ex-ercise in the Health, Aging and Body composition study. J AmGeriatr Soc 2004; 52: 502-9.

26. Sallinen J, Leinonen R, Hirvensalo M, Lyyra TM, Heikkinen E,Rantanen T. Perceived constraints on physical exercise amongobese and non-obese older people. Prev Med 2009; 49: 506-10.

27. Rasinaho M, Hirvensalo M, Leinonen R, Lintunen T, RantanenT. Motives for and barriers to physical activity among olderadults with mobility limitations. J Aging Phys Act 2007; 15:90-102.

28. Seidell JC, Visscher TLS. Body weight and weight change andtheir health implications for the elderly. Eur J Clin Nutr 2000; 54:S33-9.

29. Ramsay SE, Whincup PH, Shaper AG, Wannamethee SG. Therelations of body composition and adiposity measures to illhealth and physical disability in elderly men. Am J Epidemiol2006; 164: 459-69.

30. Kuczmarski MF, Kuczmarski RJ, Najjar M. Effects of age onvalidity of self-reported height, weight, and body mass index: find-ings from the third National Health and Nutrition ExaminationSurvey, 1988-1994. J Am Diet Assoc 2001; 101: 28-34.