Multivariate analysis of lifestyle, constitutive and body composition factors influencing bone health in community-dwelling older adults from Madeira, Portugal

Multivariate

analysis

of

lifestyle,

constitutive

and

body

composition

factors

influencing

bone

health

in

community-dwelling

older

adults

from

Madeira,

Portugal

E´lvio

Ru´bio

Gouveia

a,

*

,

Cameron

Joseph

Blimkie

b

,

Jose´ Anto´nio

Maia

c

,

Carla

Lopes

d

,

Bruna

Raquel

Gouveia

e

,

Duarte

Luı´s

Freitas

a

a

DepartmentofPhysicalEducationandSports,UniversityofMadeira,Funchal,Portugal

b

McMasterUniversity,DepartmentofKinesiology,FacultyofScience,Hamilton,Ontario,Canada

c_{CIFI2DandFacultyofSport,UniversityofPorto,Porto,Portugal} d

DepartmentofClinicalEpidemiology,PredictiveMedicineandPublicHealth,FacultyofMedicineandInstituteofPublicHealth,UniversityofPorto,Porto, Portugal

e

HealthTechnologies,CompetenceCentre,UniversityofMadeira,Funchal,Portugal

1. Introduction

Bone mass, strength and quality in older adults have been associated with genetic factors (Eisman, 1999) and several environmentalinfluences,oneofthemostimportantofwhichis habitual PA (Nguyen, Center,& Eisman, 2000). Bone-formation

declines with advancing age (Stenderup, Justesen, Clausen, & Kassem,2003)asdoesthelevelsofhabitualPA(Dalyetal.,2008); therefore,inbothgenders,butparticularlyinwomen,thereisan increasedlossofbonemasswithaging(Jones,Nguyen,Sambrook, Kelly,&Eisman1994).Theenhancedbonefragilityandconsequent increaseinfractureriskamongtheolderadultshasbeenlinkedto changes in intrinsic material properties (i.e. mass, density, stiffness, and strength) and its gross geometric characteristics (size,shape,corticalthickness,cross-sectionalarea,andtrabecular architecture)(Kanisetal.,2008;Khanetal.,2001).

InEurope,thenumberofallosteoporoticfracturesin2000was estimatedat3.79million,ofwhich0.89millionwerehipfractures

ARTICLE INFO Articlehistory:

Received7September2013

Receivedinrevisedform28February2014 Accepted7March2014

Availableonline15March2014 Keywords:

Aging

Bodycomposition Bonemineraldensity Physicalactivity

ABSTRACT

Thisstudydescribestheassociationbetweenhabitualphysicalactivity(PA),otherlifestyle/constitutive factors,bodycomposition,andbonehealth/strengthinalargesampleofolderadultsfromMadeira, Portugal.

Thiscross-sectional studyincluded401 malesand401 femalesaged 60–79yearsold. Femoral strengthindex(FSI)andbonemineraldensity(BMD)ofthewholebody,lumbarspine(LS),femoralneck (FN),andtotalleantissuemass(TLTM)andtotalfatmass(TFM)weredeterminedbydual-energyX-ray absorptiometry-DXA.PAwasassessedduringface-to-faceinterviewsusingtheBaeckequestionnaire andforasub-sample byTritracaccelerometer.Demographic andhealthhistory informationwere obtainedbytelephoneinterviewthroughquestionnaire.

TherelationshipbetweenhabitualPAvariablesandbonehealth/strengthindicators(wholebody BMD,FNBMD,LSBMD,andFSI)investigatedusingPearsonproduct-momentcorrelationcoefficientwas similarforfemales(0.098r0.189)andmales(0.104r0.105).Resultsfromstandardmultiple regressionanalysisindicatedthattheprimaryandmostsignificantpredictorsforFNBMDinbothsexes wereage,TLTM,andTFM.ForLSBMD,themostsignificantpredictorwasTFMinmenandTFM,age,and TLTMinfemales.Ourregressionmodelexplained8.3–14.2%and14.8–29.6%ofthetotalvariancein LSBMDandFNBMDformalesandfemales,respectively.

ThisstudysuggeststhathabitualPAisminimallybutpositivelyassociatedwithBMDandFSIamong olderadultmalesandfemalesandthatbodycompositionfactorslikeTLTMandTFMarethestrongest determinantsofBMDandFSIinthispopulation.

ß2014ElsevierIrelandLtd.Allrightsreserved.

* Corresponding author at: Departamento de Educac¸a˜o Fı´sica e Desporto, UniversidadedaMadeira,CampusUniversita´riodaPenteada,9000-390Funchal, Portugal.Tel.:+351291705313;fax:+351291705249.

E-mailaddresses:erubiog@uma.pt,rubiogouveia@gmail.com(&.R.Gouveia).

ContentslistsavailableatScienceDirect

Archives

of

Gerontology

and

Geriatrics

j ou rna l h om e pa ge : w w w. e l s e v i e r. co m/ l oc a t e / a rch ge r

http://dx.doi.org/10.1016/j.archger.2014.03.001

(179000hipfracturesin menand711000inwomen)(Kanis& Johnell,2005).Theincidenceofhipfractures/yearinwomenand menover65yearswas67.9/10000and26.1/10000,respectively (International Osteoporosis Foundation [IOF], 2001). The total direct cost of this disease wasestimated at 31.7 billionEuros, whichisexpectedtoincreaseto76.7billionEurosin2050,based onprojecteddemographicsinEurope(Kanis&Johnell,2005).In Portugal,thenumberofhipfracturesincreasedfrom8500in2000 (rate:8.24per10000)to9821in2007(rate:9.26per10000);the directhospitalcostswere51321300Eurosand53433131Euros, respectivelyin2000and2007(IOF,2008).

PortugalisconsideredoneofthemostagedcountriesofEurope with17.1%ofitstotalpopulationolderthan65yearsofagein2008 (StatisticsPortugal,2009).Withitsrelativelylowfertilityrateand rapidlyexpandingpopulationofolderadults,Portugalhasbeen recognizedas a high risk countryfor the development of hip fracture (Kanis &Johnell, 2005). Particularly in thePortuguese AutonomousRegionofMadeira,olderadults(65yearsandolder) are projected to comprise approximately 57.4% of the total population by 2050. The main occupations in this region are farmingandconstructionwork.Therearealmostasmanyfemale (47%)asmalefarmers(53%)inMadeira,withanaverageageof64 years(Census2001)(StatisticsPortugal,2002),thehighestofall regionsofPortugal.

Numerous lifestyle, constitutive (age, heightand body mass (BM)) and body composition factors have been implicated as determinantsofbonehealthintheelderly(Dishaman,Heath,& Lee,2013;Khanetal.,2001).HabitualPA,oneofthekeyputative environmentaldeterminantsofbonehealthhasbeenassociated withincreasesinbonemass(Felson,Zhang,Hannan,&Anderson, 1993; Pluijm et al., 2001) and improvements in muscle mass, musclestrength,balanceandbonestrength,allofwhichmitigate falls and fractures among the older adults (United States

Department of Health and Human Services [USDHHS], 2004).

Theevidencein supportofthebenefitsofhabitual PAforbone healthpromotionissocompellingthatmanyphysiciansandpublic healthofficialsnowrecommendincreasedhabitualPAandregular exerciseprogramsregardlessofage(USDHHS,2004).

Besides habitual PA, constitutive factors may also influence bonehealthstatus(Dalyetal.,2008;Hannanetal.,2000;Lane, 2006). Age-related changes in body composition have been considered potential determinants of bone health/density with aging. Although, their relative importance remains equivocal, changesinBM,TLTMandTFMhaveeachbeenidentifiedashaving important (Dargent-Molina, Poitiers, Bre´art, & EPIDOS Group, 2000;Dytfeld,Ignaszak-Szczepaniak,Gowin,Michalak,& Horst-Sikorska, 2011; Felson et al., 1993), independent roles in determiningskeletalintegrityinolderadults(Ho-Pham,Nguyen, Lai,&Nguyen,2010).Leanmassispostulatedtobeadeterminant onbonemassbecauseofthestructuralandfunctional relation-ships between muscle and bone (Looker, Melton, Borrud, Shepherd, & McGowan, 2009). BM and fat mass could exert protectiveeffectsonbonebyincreasingthemechanical loading forcesactingontheskeletonduringweight-bearing,withfatmass havinganadditionalpotentialinfluencethroughtheconversionof steroids to estrogen (Reid, Plank, & Evans, 1992). Age-related changes in general health and socio-economic status (Booth, Owen,Bauman,Clavisi,&Leslie,2000;Lim&Taylor,2005),andin smoking,calcium intake, alcohol consumption, nutrition status andprescriptivemedications(Felsonetal.,1993; Hannanetal., 2000) have also been identified as equivocal yet potential determinantsofbonehealthinolderadults.

Whilefragilityfracturesmaybeaninevitableconsequenceof aging,themorbidity,mortalityandfinancialburdenof osteoporo-sismaybemitigatedbymanagementofknownriskfactors.No attempthas been made in either Portugal or the Autonomous

RegionofMadeira(ARM),toidentifyorcharacterizerelationships between potential determinants of bone health/strength in Portugueseolderadults.Thepurposeofthisstudywastodescribe theassociationbetweenlevelofhabitual PA,otherlifestyleand constitutivefactors(stature,BM,leantissuemass,andfatmass) andbonehealth/strength(assessedasmulti-siteBMDandFSI)ina large Portuguese sample of active community dwelling older adultsmenandwomen(60–79years),controllingformanyofthe known important confounding and covariateinfluences among potential determinants. The Autonomous Region of Madeira representsarelativelyuniqueandisolatedgeographicandcultural regioninEurope.Itisuncertainwhethertherelationshipsamong putative determinants of bone health evident in other more heterogeneousregionsofEuropeapplyalsotoMadeira. Anticipat-inghighoccupationalparticipationratesfromtheCensusdata,we hypothesized that constitutive factors rather than habitual PA levels, would be stronger determinants of bone density and strengthinthispopulation.

2. Methods

2.1. Studydesignandparticipants

Thiscross-sectionalstudyincluded802participants(401males and401females)distributedsimilarlyoverfourage-cohorts(60– 64,65–69,70–74,and75–79years).UsingGPower3.1software (Erdfelder,Faul,&Buchner,1996),samplesizeestimationswere doneforcorrelationsandregressionanalysisusingCohen’seffect sizeof0.25,alpha=0.05andapowerof0.80.Thesuggestedsample sizevaried between 100 and269 subjects.In total,the sample comprised 2.1% of the older adults from ARM (60–79 years). Participantswereabletowalkindependentlyandavailabletovisit our laboratory at the University of Madeira on their own. Proportionalregional(geographic)representationwasdetermined by stratified sampling based on Census 2001 data from the PortugueseStatisticsNationalInstitute(StatisticsPortugal,2002) withthenumberof subjectsperagecohort andsexserving as stratificationfactors.Thisisarelativelyuniquegeographicallyand culturally isolated group of volunteers recruited via advertise-mentsfor a large study on bone health and PA distributed via newspapers and through churches, senior groups and senior centersthroughouttheARMin2008–09.

The study was approved by the University of Madeira, the Regional Secretary of Education and Culture, and theRegional SecretaryofSocialAffairs.Allparticipantswereinformedaboutthe natureandpurposesofthestudyandwritteninformedconsent wasobtainedfromeachsubject.

2.2. Anthropometryandbonedensitometry

BM(kg)wasmeasuredwithabalancescaleaccurateto0.1kg (Seca alpha digital scales model 770, Germany) and standing height(cm)witha Holtain stadiometer (HoltainLtd., Crymych, UnitedKingdom)accurateto0.1cm.Subjectsworelight,indoor clothingwithoutshoesduringthemeasurements.

BMD(g/cm2)wasdeterminedbydual-energyX-ray absorpti-ometry-DXA(LunarProdigyPrimo,withtechnologicfanbeam–GE Healthcare, Encore 2007 software version 11.40.004). After removing all objects suspected or known to contain metal, participantswerepositioned bythetechnicianaccordingtothe manufacturer’srecommendedprotocol.Subjectswereinasupine positionandthefollowingsiteswereinvestigated:wholebody,LS (anterior–posterior),andhip(FN,trochanter,Ward’striangleand totalhip). Furthermore, the scansyielded information on body composition,includingTLTMandTFM.

Inadditiontotheconventionaldensitometrymeasurements, structuralvariableswerealsodeterminedusingtheHipStrength Analysisprogram,including hipaxislengthand cross-sectional momentofinertia(CSMI).Thesebonegeometryvariables were usedtocalculatetheFSI,theratioofestimatedcompressiveyield strengthoftheFNtotheexpectedcompressivestressofafallon thegreatertrochanteradjustedforeachsubject’sage,heightand BM(Yoshikawaetal.,1994).FSIprovidesanestimateoffunctional bone strength of the hip region based on both its level of mineralizationandthegeometricdistributionofmineralaround centralaxisoftheFN(Beck,Ruff,Warden,Scott,&Rao,1990).

Scanswerestandardizeddailyagainstacalibrationphantom; theprecisionerrorexpressedasthecoefficientofvariation(CV%) was 0.31%. Scans were taken alternately by four different technicians over the course of data collection. All technicians receivedanidentical5daysDXAtrainingcoursebeforethestart the study using the manufacturer’s recommended protocol. ReliabilityofourDXAmeasurementswasdeterminedona sub-sampleof17malesandfemalesaged69.35years.Technicians werepairedandmembersofeachpairperformedseparateLSandhip scansonhalfthesubjectseach(9and8subjects,respectively,per pair).Subjectswererepositionedaftereveryscan.Resultsfromboth pairs of assessors were pooled and the technical error of the measurements(TEM)wasdetermined.TEMwasusedtodetermine inter-observererror,asoccurswhentwotechniciansindependently measurethesamething.Thevaluesrangedfrom0.19%fortotalhipto 0.50%fortheLS.Inter-observerreliabilitywasalsodeterminedusing theCV.TheCV%was1.72%forLS,2.10%fortheFN,2.53%forWard’s triangleand0.88%forthetotalhip.

2.3. PAmeasures

HabitualPAwasassessedduringface-to-faceinterviewsusing theBaeckequestionnairedevelopedintheNetherlands(Baecke, Burema,&Frijters,1982),withareferencetimeperiodoflastyear. Thisquestionnaireincludesatotalof16questionsclassifiedinto threespecificdomains:PAatwork/housework,sportandleisure time,thelatterexcludingsports.Ifthesubjectswerenotemployed oriftheywereretired,theiroccupationwascodedashomemaker. ThequestionnairealsoprovidesameasureofhabitualPAwhichis thesumofthesethreespecificdomains.Numericalcodingformost responsecategoriesvariedfrom1to5(Likertscale)rangingfrom nevertoalwaysorveryoften.Adetaileddescriptionofthescoring proceduresfor calculationofhabitual PAanditssubcomponent categories (PA at work, sport and leisure time) is provided by

Baecke et al. (1982). The questionnaire took about 20min to complete.

Interviewsweretakenalternatelybyfourdifferentresearchers overthecourseofdatacollection.Previously,intra-class correla-tion-coefficients were calculated to determine the test–retest reliabilityofthequestionnaireinapilotstudyinvolving32males and 59 females (68.37.6 years). Over an interval of 1 week, correlationsrangedbetween0.83,0.85and0.85forthework,sport andleisure-timeindices,respectively.Ourreliabilityscoresforwork andsportPAweresimilartothoseobtainedbyBaeckeetal.(1982)in asampleofDutchadultmenandwomen(0.88,0.81)andwithamore recentstudybyOnoetal.(2007) of61middleaged(53.3 years) women(0.84,0.83).However,ourcorrelationswerehigherforleisure timeindexthanthosereportedbyeitherBaeckeetal.(1982)orOno etal.(2007),0.74and0.78,respectively.ThevalidityoftheBaecke questionnairehasalsobeenestablishedbyOnoetal.(2007)forthis populationagainstthemoreobjectivemeasureofmovementcounts using digital pedometry and uniaxial accelerometry (Lifecorder, SuzukenCo.,Nagoya,Japan);correlationsrangedfrom0.30to0.49for the 3 dimensions of habitual PA assessed with the Baecke questionnaireinthisstudy.

AsaninternalvalidationoftheBaeckeQuestionnaireapproach, oftherelationshipsbetweenhabituallevelsofPAandbonehealth indicators,inourstudyPAwasalsoassessedusingtheRT3triaxial accelerometerinasub-sampleof173olderadults,89malesand84 females, age 60–79 years. The RT3 accelerometer measures accelerationineachanatomicalaxiswithvertical(x), anteriopos-terior(y),andmediolateral(z)measurements.Todeterminethe totaldaily PA‘‘counts’’,we usedthesquare-rootof thesumof squared accelerations for each axis, which provides the vector magnitude(VM)incountsperminute(counts/min).Participants wereinstructedtoweartheRT3fora7-dayperiodduringwaking hoursandremoveitforsleeping,swimmingandbathingonly.

Forinclusioninthepresentstudy,participantswererequiredto have worn the accelerometer on at least five complete days (includingSaturdayandSunday).Participantswereinstructedto weartheRT3fora7-dayperiodduringwakinghoursandremoveit for sleeping, swimming and bathing only. For inclusion in the present study, participants were required to have worn the accelerometeronatleastfivecompletedays(includingSaturday andSunday).Thecrosscorrelationsweregenerallyweak,whenthe relationshipsbetweenthemoreobjectivemeasureofPAandthe keydependentvariableswereexaminedintheregressionanalysis, PA waspersistentlystill arelatively weakdeterminant of bone sites and FSI in this population (0.002

b

0.098 and 0.042

b

0.084 for men and women, respectively). We also correlatedthetotalscoreofPAfromBaeckequestionnairewiththe total ‘‘counts’’ from RT3 using the Pearson product-moment correlation coefficient. However, the correlations for men and women together were positive, but non-significant (r=0.131; p>0.87).

2.4. Healthquestionnaireanddietaryintake

Demographicinformationandacompletehealthhistorywere obtained by telephone interview. The health questionnaire employed in the FallProof! Program (Rose, 2003) was used to assesssmokinghistoryandmedication.

Dietary intakewasestimatedusinga semi-quantitativefood frequencyquestionnairedevelopedbytheDepartmentofHygiene and Epidemiology of Porto University Medical School, and previouslyvalidatedin2415Portugueseparticipants(926males and 1489 females) aged 18–92 years (Lopes, 2000). The semi-quantitative food frequency questionnaire was done with a referencetime periodof lastyear. Thisquestionnairewasused toquantifycalciumandalcoholusingthesoftwareFoodProcessor Plus1

(ESHAResearch, Salem-Oregon, 1997), adaptedto Portu-guesetraditionalfoodsanddishes.

2.5. Statisticalanalysis

Descriptive characteristics of participants were reported as meansSDs.AlldataweretestedfornormalitybytheKolmogorov– Smirnovstatistic.Whenrequired,non-normaldistributed character-istics wereappropriately transformedusinglog10,squarerootor inversetransformfunctions.

Sex specific bivariate associations between the bone health indicators(BMDandFSI)andputativepredictorsofbonehealth (age, BM, TLTM,totalcalciumintake, alcohol consumption and habitual PA in women and sports related PA in men) were calculatedforallage-cohortscombinedusingPearsoncorrelations. Sex-specificstandardmultipleregressionanalysiswasthenused to identify theindependent contribution of theindividual and combinedpredictorsforBMDatthedifferentskeletalsites.Betas, namelystandardizedregressioncoefficients,wereusedtoassess the relative independent contributions of each predictor or combination of predictors, and the adjusted R2_{s indicated the}

percentageofexplainedvarianceinthebonehealthoutcomes.The level of significance was set at p<0.05. All analyses were performedusingSPSS(version18.0).

3. Results

Table 1 shows that age-related declines were significant

(p<0.05)forheightandTLTMforbothmalesand females.BM declinedsignificantly across age only in women and TFM was unchangedacrossageinmalesandfemales.

Likewise, sports related PA for men, work related PA and habitualPAforwomen,decreasedsignificantlyacrossage-groups by13.6%,10.7%,and8.0%,respectively.Therewerenosignificant agerelateddeclinesforanyoftheotherhabitualPAvariablesfor menandwomen.Onlymendemonstratedasignificantage-related reductioninalcoholconsumptionand therewereno significant age-relateddifferencesincalciumintakeforeithersex.

BMDforallsitesandforbothsexeswas,withoneexception(LS males65–69years),highestintheyoungestage-cohortandforthe most part decreased progressively with advancing age-cohort (Table2)forboth sexes.BMDdeclinedsignificantly(9.6–21.1%) withage(60–79years)atallsitesforwomenwhereassignificant age-reductionswereobservedonlyfortheFN(9.8%)formales.FSI was highest in theyoungest age-cohort in both sexes, didnot change significantly across ages for males, but decreased progressively and significantly with increasing age-cohort in females.

WiththeexceptionoftheLSformales,BMDatmultiplesites weresignificantlynegativelycorrelatedwithageinbothgenders (Table 3). All BMD measures were significantly positively correlated withheight, BM,TLTMand TFM forboth malesand females. Sport related PA (Baecke Questionnaire) was the only activity category to relate significantly with BMD for males, correlatingweakly,butnonethelesspositively,withLSBMD.The

Table1

Ageandsex-specificdescriptivecharacteristics.

Agegroups(years) pvalue Comparison 60–64(1) 65–69(2) 70–74(3) 75–79(4)

MeanSD MeanSD MeanSD MeanSD Men (n=103) (n=99) (n=107) (n=92) Height(cm) 166.95.2 165.96.2 164.56.1 164.66.2 0.011 1,2>3,4 BM(kg) 80.312.1 79.713.1 79.913.3 75.813.0 0.058 – TLTM(kg) 54.35.9 53.35.8 52.85.5 51.26.3 0.003 4<3,2,1 TFM(kg) 22.68.1 23.37.9 23.68.6 21.58.2 0.285 PAwork(1–5units) 2.80.6 2.60.5 2.60.6 2.70.5 0.066 PAleisure(1–5units) 2.60.6 2.50.6 2.50.7 2.50.5 0.520 PAsport(1–5units) 2.20.6 2.20.6 2.00.6 1.90.5 0.015 4<2,1 HabitualPA(3–15units) 7.61.3 7.31.2 7.11.4 7.21.2 0.065 – Diet.calcium(mg/day) 721.7321.7 743.7365.2 723.9350.6 734.5384.9 0.083 – Alcohol(dl/day) 16.720.5 13.117.3 11.618.2 9.215.3 0.028 4<3,2,1 Women (n=102) (n=108) (n=98) (n=93) Height(cm) 154.25.4 153.85.5 152.05.8 150.15.4 <0.001 4<2,1;3<1 BM(kg) 72.211.7 71.212.7 70.610.6 67.811.5 0.063 4<3,2,1 TLTM(kg) 39.94.9 40.05.5 39.33.9 38.24.3 0.033 4<2 TFM(kg) 29.77.8 28.78.0 28.27.3 27.17.8 0.131 – PAwork(1–5) 2.80.6 2.80.4 2.70.4 2.50.4 <0.001 4<2,1 PAleisure(1–5) 2.50.7 2.50.5 2.40.5 2.40.6 0.071 – PAsport(1–5) 2.20.6 2.30.6 2.30.6 2.10.6 0.072 – HabitualPA(3–15) 7.51.3 7.51.1 7.31.1 6.91.2 0.002 4<2,1 Diet.calcium(mg/day) 906.5387.3 817.9381.2 821.8398.6 796.5384.5 0.197 – Alcohol(dl/day) 2.13.9 1.54.1 1.22.7 1.32.9 0.178 –

SD,standarddeviation;BM,bodymass;PA,physicalactivity;TLTM,totalleantissuemass;TFM,totalfatmass;1and2>3,4,60–64and65–69agegroupsarelowerthan70– 74and75–79agegroups;4<3,2,1,75–79agegroupislowerthan70–74,65–69and60–64agegroups;4<2,1,75–79agegroupislowerthan65–69and60–64agegroups; 3<1,70–74agegroupislowerthan60–64agegroup;4<2,75–79agegroupislowerthan65–69agegroup.

Table2

Ageandsex-specificdescriptivecharacteristicsforBMDandFSI.

Agegroups(years) pvalue Comparison 60–64(1) 65–69(2) 70–74(3) 75–79(4)

MeanSD MeanSD MeanSD MeanSD Men TotalbodyBMD(g/cm2 ) 1.230.10 1.220.11 1.220.11 1.190.09 0.077 – LSBMD(g/cm2 ) 1.190.18 1.210.21 1.180.24 1.140.19 0.276 – FNBMD(g/cm2 ) 1.020.15 0.990.14 0.950.14 0.920.13 <0.001 4<2,1;3<1 FSI 1.860.45 1.740.52 1.760.51 1.830.57 0.264 – Women TotalbodyBMD(g/cm2 ) 1.140.09 1.100.10 1.070.09 1.030.08 <0.001 4<3,2,1;3and2<1 LSBMD(g/cm2_{) 1.06} 0.18 1.010.16 0.960.18 0.950.16 <0.001 1>2,3,4 FNBMD(g/cm2_{) 0.92} 0.13 0.880.12 0.830.10 0.780.10 <0.001 4<3,2,1;3<2,1;2<1 FSI 1.610.47 1.530.39 1.480.38 1.440.37 0.027 1>2,3,4

SD,standarddeviation;FN,femoralneck;LS,lumbarspine;FSI,femoralstrengthindex;4<2,1,75–79agegroupislowerthan65–69and60–64agegroups;3<1,70–74age groupislowerthan60–64agegroup;4<3,2,1,75–79agegroupislowerthan70–74,65–69and60–64agegroups;3and2<1,70–74and65–69agegroupsarelowerthan 60–64agegroup;1>2,3,4,60–64agegroupishigherthan65–69,70–74and75–79agegroups;3<2,1,70–74agegroupislowerthan65–69and60–64agegroups;2<1, 65–69agegroupislowerthan60–64agegroup.

relationship between habitual PA and BMD in females was generallymorepositivethanformales,withseveral significant, butsimilarlyweakcorrelationsbetweenhabitualPAvariablesand the various measures of BMD (Table 3). FSI was significantly negatively correlated withage in females but not males, with heightinmalesbutnotfemales,andwithBM,TLTMandTFMin bothsexes.FSIwassignificantlynegativelycorrelatedwithwork relatedPA,leisuretimePAanddietarycalciumintakeinfemales only.

Resultsfrommultipleregressionanalysismodelingcontaining TLTM,TFM,habitualPA,ageandtotalcalciumintakearepresented inTable4fortheFN,LSandFSImeasuresonly.Similarmodeling wasdonefortheremainingbonemeasuresbutthesefindingsare notreportedinthesetables.Age,TLTM,TFMforFNBMDinboth sexes,TFMinmenandTFM,ageandTLTMinfemaleforLSBMD, enteredastheprimaryandmostsignificantcontributors.Habitual PAcontributedsignificantlytovariationinFNBMDinfemales.In men,habitualPAdidnotsignificantlycontributetothevariationin BMD at any bone site. Also, dietary calcium intake did not contributetothevariationinBMDofFNorLSineithersex.Our regressionmodelexplained14.2and29.6%ofthetotalvariancein theFNBMDinmalesand females, respectively.For LSBMD,the totalvarianceexplainedbythemodelwas8.3%inmalesand14.8% infemales.Theregressionmodelexplained13.7and14.6%ofthe totalvariationinmechanicalFSIinmalesandfemales, respective-ly.Inbothsexes,agewasthemostsignificantpredictorforFNBMD, andTFMforLSBMDandforFSI.

4. Discussion

The bivariate correlation and multiple regression analyses revealedonlyminorassociationsbetweenhabitualPAandBMD. Our study reinforces that constitutive and body composition factorsarethestrongestpredictorsofBMDandFSIinoldermales

and females.Moreover,associationstendedtobesex-and site-specific.

Additional information about FSI, a derived measure that providesanestimateofhipfracturerisk,wasalsoassessedinour study.FSIisnotonlydependentonfemoralBMD,butitisalsoa function of the spatial distribution of bone mass and bone’s intrinsicstructuralgeometricpropertiessuchasitsdiameter,area, length,andangleattheFN.FSIhasbeenrecognizedasasignificant independent predictor of hip fracture risk (Beck et al., 1990; Faulkneretal.,2006).InthisstudyhabitualPAwasnotasignificant predictorofFSIinourmultipleregressionanalysis.However,work related PA was significantly correlated with FSI in males and females, whereas the association with leisure time PA was significantonlyinfemales.AswithourfindingsforBMD,these resultsalsosuggestthatPAingeneral,isonlyweakly–moderately associatedwithFSIinolderadultmenandwomenofourstudy.

TheassociationbetweenhabitualPAandBMDinolderadults hasnotbeenextensivelystudiedandcollectivelythefindingsare equivocalatbest.Uncertainlyinthisareastemsfromthefactthat previousactivityassessmentapproacheshavenotfullyconsidered theentirearrayofputativemechanicalloadingparameters(type, intensity, frequency and duration of activities) now known to independently or interactively affect skeletal adaptation ( Maı¨-moun&Sultan,2011).Itiswidelyacknowledgedthatbonetissue (corticalboneandporoustrabecularbone)adaptsitsshape and structure according to its mechanical environment. The bone remodelingmechanismis complexandis determinedalsobya speculativenetworkofinteractionsinfluencedbyfactorssuchas genetics, growth factors, gender, age, soft tissue composition (TLTM,TFM),lifestylechoices(smoking,alcoholintake), medica-tion,hormones,andnutrition(Khanetal.,2001).

OurfindingsforBMDareinpartialagreementwithprevious select cross-sectionaland longitudinalstudiesof similarly aged populations.Hannanetal.(2000)andStewartetal.(2005)failedto

Table3

Sex-specificPearsoncorrelationsbetweenBMDindicators,FSIandselecteddescriptivecharacteristics. Characteristics TotalbodyBMD(g/cm2

) LSBMD(g/cm2 ) FNBMD(g/cm2 ) FSI Men Age(years) 0.110z _– 0.259y _– Height(cm) 0.279y _0.158y _0.262y 0.147y BM(kg) 0.425y _0.293y _0.309y _0.343y TLTM(kg) 0.389y 0.179y 0.308y 0.165y TFM(kg) 0.319y 0.292y 0.228y 0.383y PAwork(1–5units) – – – 0.105z PAleisure(1–5units) – – – – PAsport(1–5units) – 0.104z _{– –} HabitualPA(3–15units) – – – –

Diet.calcium(mg/day) – – – –

Smoking(units) 0.147z 0.124z – – Alcohol(dl/day) – – – – Medication(units) – 0.123z – – Women Age(years) 0.394y 0.248y 0.443y 0.164y Height(cm) 0.396y _0.314y _0.426y _– BM(kg) 0.530y _0.395y _0.352y _0.334y TLTM(kg) 0.477y 0.323y 0.336y 0.220y TFM(kg) 0.486y 0.334y 0.314y 0.326y PAwork(1–5units) 0.098z – 0.164y 0.099z PAleisure(1–5units) 0.127z _{– 0.100}z _0.108z PAsport(1–5units) – – 0.140y _– HabitualPA(3–15units) 0.117z _{– 0.189}y _–

Diet.calcium(mg/day) 0.108z _{– – 0.108}z

Smoking(units) 0.150y 0.150y 0.192y – Alcohol(dl/day) – – – – Medication(units) – – – 0.115z

FN,femoralneck;LS,lumbarspine;FSI,femoralstrengthindex;BM,bodymass;PA,physicalactivity;TLTM,totalleantissuemass;TFM,totalfatmass. Onlycorrelationsthatwerestatisticallysignificantwereincluded.

z _{Correlationissignificantatthe0.05level(2-tailed).} y _{Correlationissignificantatthe0.01level(2-tailed).}

findanyassociationbetweenPAscoreandBMDchangeinolder adultsmaleswithintheagerangeofourstudy,whereasKenny, Prestwood,Marcello,andRaisz(2000)andNguyenetal.(2000)

foundasignificantpositiverelationshipbetweenPAandFN,but not LSBMD in similarly aged males. A few additional studies, reportedpositiveassociationsbetweensportrelatedPAwithFN andhipBMDinolderadultsmales(Pluijmetal.,2001;Vuillemin, Guillemin,Jouanny,Denis,&Jeandel,2001),andpositive associa-tions between PA and BMD using simple univariate analyses, which became weaker or non-significant with more stringent multivariateapproachesthatadjustedforputativeBMDcovariates (Lauetal.,2006).

There ismoreabundantcomparativeresearchforolderadult womenthanmen,buttheresultsarenolessequivocal.Previous studies(Nguyen,Sambrook,&Eisman,1998)haveshownamodest favorable effect of PA on the rate of bone loss at the FN and significantpositiveassociationswithFNbutnotLSBMDinolder adultsfemales(Hagbergetal.,2001;Nguyenetal.,2000;Pluijm etal.,2001).OtherkindofanalysisprovidedbyMavroeidi,Stewart, Reid,andMacdonald(2009)havereportedsignificantinteractions between PA tertiles, classified according to metabolic (energy expendedincarryingoutactivitiesexpressedinMETh/week)and mechanicalcomponentsofPA(fromgroundreactionforcesonthe

skeletonexpressedinpeakscores),andleftandrighthipBMDina largepopulationofolderwomenlivinginthecommunity.Others havereportednorelationshipbetweenPAlevelandBMDinolder adultsfemales(Ga´ba,Kapusˇ,Pelclova´,&Riegerova´,2012;Nahas, Kawakami, Nahas-Neto, Buttros Dde, & Cangussu, 2011) or insignificant relationships withproxy measures of habitual PA afteradjustmentforcovariates(Schoffletal.,2008),whereassome havereportedspecificpositiveassociationswithwalking(Pluijm etal.,2001)andsportsrelatedPA(Vuilleminetal.,2001).

There are few publishedreports of the associationbetween geometric or derived biomechanical and functional (e.g. bone strengthindex)measuresofbonestrengthandPAinolderadults populations. However, similar toour findings, Nurzenski et al. (2007)andUusi-Rasi,Sieva¨nen,Pasanen,Beck,andKannus(2008)

reported significant positive associations between aspects of femoralbonegeometryandPAinolderadultsfemales,whereas thereisonlyonereporttoourknowledge(Semanicketal.,2005), which likewise,indicated a weakpositive relationshipbetween walkingandFNsectionmodulusinolderadultsmen.

Inourstudy,theindicesofhabitualPAweremeasuredbythe Baeckequestionnaire,whichincludesthreereliabledimensions: PAatwork(participantswhowereretiredweclassifiedinlowlevel for occupation),sportduringleisure-time, and otherPA during

Table4

StandardMLRbetweenFN,LSandFSIandputativepredictors(TLTM,TFM,habitualPAandtotalcalciumintake). Predictors CrudeBstd.error AdjustedBstd.error Betay

pvalue 95%CIa FNBMD(g/cm2 ) Men(R2 adj=0.142) TLTM(kg) 0.0070.001 0.0050.001 0.210 <0.001 0.003;0.008 TFM(kg) 0.0040.001 0.0020.001 0.130 0.016 0.000;0.004 HabitualPA(5–15units) 0.0040.006 0.0030.005 0.028 0.551 0.007;0.014 Age(years) 0.0070.001 0.0050.001 0.211 <0.001 0.008;0.003 Diet.calcium(mg/day) 2.3865

0.000 2.3265 0.000 0.057 0.220 0.000;0.000 Women(R2 adj=0.296) TLTM(kg) 0.0090.001 0.0040.001 0.162 0.002 0.002;0.007 TFM(kg) 0.0050.001 0.0030.001 0.182 <0.001 0.001;0.005 HabitualPA(units) 0.0200.005 0.0130.005 0.124 0.004 0.004;0.022 Age(years) 0.0100.001 0.0080.001 0.369 <0.001 0.010;0.006 Diet.calcium(mg/day) 4.0375

0.001 1.5855 0.000 0.049 0.253 0.000;0.000 LSBMD(g/cm2 ) Men(R2 adj=0.083) TLTM(kg) 0.0060.002 0.0010.002 0.041 0.460 0.002;0.005 TFM(kg) 0.0080.001 0.0070.001 0.282 <0.001 0.005;0.010 HabitualPA(5–15units) 0.0010.008 0.0080.008 0.049 0.319 0.008;0.025 Age(years) 0.0030.002 0.0020.002 0.044 0.374 0.005;0.002 Diet.calcium(mg/day) 1.8805

0.000 2.6455 0.000 0.043 0.369 0.000;0.000 Women(R2 adj=0.148) TLTM(kg) 0.0110.002 0.0050.002 0.118 0.042 0.000;0.009 TFM(kg) 0.0080.001 0.0060.001 0.244 <0.001 0.003;0.008 HabitualPA(5–15units) 0.0060.008 0.0030.007 0.018 0.704 0.012;0.017 Age(years) 0.0070.002 0.0060.002 0.176 <0.001 0.009;0.003 Diet.calcium(mg/day) 3.2675

0.000 1.3045 0.000 0.028 0.558 0.000;0.000 FSI Men(R2 adj=0.137) TLTM(kg) 0.0140.004 0.0000.005 0.005 0.931 0.009;0.009 TFM(kg) 0.0240.003 0.0240.003 0.383 <0.001 0.031;0.018 HabitualPA(5–15units) 0.0300.020 0.0050.019 0.012 0.803 0.033;0.043 Age(years) 0.0020.005 0.0020.004 0.025 0.600 0.011;0.006 Diet.calcium(mg/day) 5.7255_{0.000 3.820}5_{0.000 0.027 0.570 0.000;0.000}

Women(R2 adj=0.146) TLTM(kg) 0.0190.004 0.0070.005 0.085 0.141 0.017;0.002 TFM(kg) 0.0170.003 0.0160.003 0.301 <0.001 0.022;0.010 HabitualPA(5–15units) 0.0330.017 0.0140.016 0.041 0.392 0.018;0.046 Age(years) 0.0120.004 0.0150.003 0.203 <0.001 0.021;0.008 Diet.calcium(mg/day) 6.1065

0.000 4.9595

0.000 0.047 0.315 0.000;0.000 TLTM,totalleantissuemass;TFM,totalfatmass;PA,physicalactivity.

a

95.0%confidenceintervalforB-values.

y

leisure-time(Baeckeetal.,1982).Thisinstrumentwasalsousedin slightly younger healthy 57 year old post-menopausal women reportedinthestudybyWalsh,Hunter, andLivingstone(2006)

andyounger healthyadultBelgians30–40 yearsof age( Philip-paerts&Lefevre,1998).Althoughoursampleisolderthanthosein thesereferencestudies,wetrustthattheBaeckequestionnaireis stillausefulinstrumentsinceourolderpopulationwasveryactive fortheirage,inrelationtoothersregionsfromPortugalasreported inCensus2001(StatisticsPortugal,2002).

Notwithstandingthisinformation,discrepanciesamong find-ingsregardingPAmaybeattributedtoavarietyoffactors.Studies oftenincludedifferentmeasuresofPAanditssub-classifications, different age ranges, differing cut points defining age-specific cohorts,variablehealthstatusandlevelsofsocialindependence andco-morbidities,differentdegreesofstatisticalsophisticationin dataanalysis, differing levelsof historical and habitual PA and variable measures of bone health. Apart these limitations, our findings and the cumulative literature to date suggest that ‘‘current’’ levels of PA contribute only minimally in explaining variationinBMDandFSIamongolderadultsmales,withperhapsa slightlystronger influence among olderadults females and for certaingeometricmeasuresreflectingbonestrengthinbothsexes. OurfindingsregardingPAwerenotaltogetherunexpected,aswe anticipated,basedontheNationalSurvey(Census2001)(Statistics Portugal,2002),arelativelyhighlevelofoccupationalengagement among our subjects, which we hypothesized would reduce variabilityinPAlevels,therebyaccentuatingbroaderdifferences inconstitutivefactors.Inourstudy,however,only35.2% ofthe menand11.5%ofthewomenwereactiveinfarming.Thisdiffers markedly from the percentage found in the National Census, suggesting that subjects in ourstudy were less occupationally engagedthantheregionalpopulation atlarge.Notwithstanding this differencein occupationalengagement, ourfindings never-thelesssuggestreducedvariabilityingenerallevelsofPAamong oursubjectscomparedtoconstitutivefactors.

Inthepresentstudy,ashypothesized,constitutivefactorslike age,heightandBM,andbodycompositionfactorslikeTLTMand TFMwereconsistentlymorestronglycorrelatedwithBMDandFSI thananyofourmeasuresofPA.Themultipleregressionanalyses, revealed that these constitutive and body composition factors make the strongest contribution to explained the dependents variables,whenthevarianceexplainedbyallothervariablesinthe model was controlled for. Specifically, age,TLTM and TFM for FNBMDin both genders and TFM in menand TFM and agein womenfor LSBMDand FSI,Otherstudieshaveshown that age (Hannanetal.,2000),BM(Dargent-Molinaetal.,2000;Felsonetal., 1993),TLTM(Dytfeldetal.,2011;Ho-Phametal.,2010;Travison, Araujo,Esche,Beck,&McKinlay2008)andTFM(Ho-Phametal., 2010)areimportantpredictorsofBMDinolderadultsmalesand females,corroboratingourfindings.Resultsfrommultiple regres-sionanalysisinourstudy,showedthatage,TLTMandTFMentered astheprimaryandmostsignificantcontributorsforFNBMDinour study,accountingfor between 13–21.1% and16.2–36.9% of the explained variation in these measures in males and females, respectively.For LSBMD,TFM enteredastheprimaryand most significantpredictoraccountingfor28.2%ofthevariationinmales and24.4%infemales.

Ourstudyis uniqueinseveral aspects;it includedrelatively largepopulationsofbotholderadultmenandwomen,theseolder peoplewerelivingfreelyandindependentlyamongthegeneral community,andthepopulationhadahighdegreeofoccupational engagement.Alsothehomogeneityofoursamplereducesethnic, socioeconomic, nutritionaland lifestyles variability, which may accentuate the more important determinants. There were, however, several limitations associated with this study. The cross-sectional design does not allow conclusions about the

cause-and-effect relationship between PA and BMD. Although the Baecke questionnaire has been shown to have acceptable reproducibility(r=0.71–0.82),thelimited abilityofsome parti-cipantstoaccuratelyrecallpastsportandleisureactivitiescould introduce bias and lead to misclassification. In addition, PA questionnairesareunlikelytobesufficientlysensitivetodetectall the mechanical loading effects in all bone types and all bone regions.Inourstudy,althoughweusedtheTritracaccelerometer in a small sub-sample with similar results to the Baecke questionnaire, this analysis was also probably insufficiently sensitive to detect subtle effects of differing types of loading parameters.Thisissueneedstobeaddressedinfuturestudies.The datawereobtainedfromindependentlylivingolderadultmenand womenfromMadeira,Portugal,ageographicallyisolatedregion wheretheculturalbackgrounds,livingandworkingconditionsand environmentalinfluencesaregenerallyhomogeneous.The homo-geneityoftheseenvironmentalinfluences,especiallywithregards toworkinghistory,wouldminimizetheapparentimportanceofPA asadeterminantofBMDandbonestrengthinthisstudy.Further, theparticipantswereessentiallyvolunteers,whocouldhavebeen generally healthier than those who did not participate, and survivor bias,especially among malesin the olderage-cohorts cannotberuledoutasapotentialconfoundingfactorparticularly forbetweensexcomparisonsinourstudy.Lastly,ourswasavery unique older population as witnessed by their lower rates of retirement,highprevalenceofgainfulemploymentinfarmingand reduceddependencyonsocialassistance.

Inconclusion,ourfindingspointtotheimportanceofage,TLTM and TFM in both sexes, as predominant determinants of bone healthinolderPortuguesemenandwomen.HabitualPAshowed significantpositivecorrelationswithBMDforthetotalbodyand FNBMDinwomen,butwaseithernotimportantorrelativelyless importantthanconstitutivefactorsinexplainingvariationinBMD or FSI.Neither dietarycalciumintake noralcohol consumption appearedasimportantdeterminantsofbonehealthinthisstudy. Ourfindingssuggestthatbonehealthpromotionandpreservation mightbeenhancedamongtheolderadultsbyencouragingdietary practicesandPAbehaviorsaimedatbodycomposition stabiliza-tionwithadvancingage.

Funding

This work was supported by doctoral degree grants from Fundac¸a˜o para a Cieˆncia e a Tecnologia reference: SFRH/BD/ 29300/2006. Technical assistance in collection of the data was supportedbyMadeiraRegionalGovernmentandRegionalSecretary ofEducationandCulture.Thedual-energyX-ray absorptiometry-DXAscansweregenerouslysponsoredbyIberdataEquipment’sS.A. Conflictofintereststatement

Theauthor(s)declarethattheyhavenopersonalorfinancial conflictofinterest.

Acknowledgments

We would like to greatly acknowledge the supervision and assistanceofProfessorEmeritusGastonBeunenoftheFacultyof KinesiologyandRehabilitationSciences,DepartmentofBiomedical Kinesiology,KatholiekeUniversiteit Leuven,who assistedinthe dataanalysis andcontributedsignificantly tothequality ofthe manuscriptinitsearlystages.Gastonunexpectedlypassedawayin Augustof2011.TheauthorsarealsogratefultoErcı´liaFena,Joana CastroandLetı´ciaSousafortechnicalassistanceinthecollectionof thesampleanddatamanagement.

References

Baecke,J.A.,Burema,J.,&Frijters,J.E.(1982).Ashortquestionnaireforthe measure-mentofhabitualphysicalactivityinepidemiologicalstudies.TheAmericanJournal ofClinicalNutrition,36,936–942.

Beck,T.J.,Ruff,C.B.,Warden,K.E.,Scott,W.W.,Jr.,&Rao,G.U.(1990).Predicting femoralneckstrengthfrombonemineraldata:Astructuralapproach.Investigative Radiology,25,6–18.

Booth,M.L.,Owen,N.,Bauman,A.,Clavisi,O.,&Leslie,E.(2000).Social-cognitiveand perceivedenvironmentinfluencesassociatedwithphysicalactivityinolder Aus-tralians.PreventiveMedicine,31,15–22.

Daly,R.M.,Ahlborg,H.G.,Ringsberg,K.,Gardsell,P.,Sernbo,I.,&Karlsson,M.K.(2008).

Associationbetweenchangesinhabitualphysicalactivityandchangesinbone density,musclestrength,andfunctionalperformanceinelderlymenandwomen. JournaloftheAmericanGeriatricsSociety,56,2252–2260.

Dargent-Molina,P.,Poitiers,F.,Bre´art,G.,&EPIDOS,Group.,(2000).Inelderlywomen weightisthebestpredictorofaverylowbonemineraldensity:Evidencefromthe EPIDOSstudy.OsteoporosisInternational,11,881–888.

Dishaman,R.K.,Heath,G.W.,&Lee,I.(2013).Physicalactivityepidemiology(2nded.). Champaign,IL:HumanKinetics.

Dytfeld,J.,Ignaszak-Szczepaniak,M.,Gowin,E.,Michalak,M.,&Horst-Sikorska,W. (2011).Influenceofleanandfatmassonbonemineraldensity(BMD)in postmeno-pausalwomenwithosteoporosis.ArchivesofGerontologyandGeriatrics,53,237–242.

Eisman,J.A.(1999).Geneticsofosteoporosis.EndocrineReviews,20,788–804.

Erdfelder,E., Faul,F., &Buchner,A.(1996).GPOWER:A generalpower analysis program.BehaviorResearchMethods,28,1–11.

Faulkner,K.G.,Wacker,W.K.,Barden,H.S.,Simonelli,C.,Burke,P.K., Ragi,S.,etal. (2006).Femurstrengthindexpredictshipfractureindependentofbonedensity andhipaxislength.OsteoporosisInternational,17,593–599.

Felson,D.T.,Zhang,Y.,Hannan,M.T.,&Anderson,J.J.(1993).Effectsofweightand bodymassindexonbonemineraldensityinmenandwomen:TheFramingham study.JournalofBoneandMineralResearch,8,567–573.

Ga´ba,A.,Kapusˇ,O.,Pelclova´,J., &Riegerova´,J. (2012).Therelationshipbetween accelerometer-determinedphysicalactivity(PA)andbodycompositionandbone mineraldensity(BMD)inpostmenopausalwomen.ArchivesofGerontologyand Geriatrics,54,315–321.

Hagberg,J.M.,Zmuda,J.M.,McCole,S.D.,Rodgers,K.S.,Ferrell,R.E.,Wilund,K.R.,etal. (2001).Moderatephysicalactivityisassociatedwithhigherbonemineraldensityin postmenopausalwomen.JournaloftheAmericanGeriatricsSociety,49,1411–1417.

Hannan,M.T.,Felson,D.T.,Dawson-Hughes,B.,Tucker,K.L.,Cupples,L.A.,Wilson,P.W., etal.(2000).Riskfactorsforlongitudinalbonelossinelderlymenandwomen:The FraminghamOsteoporosisStudy.JournalofBoneandMineralResearch,15,710–720.

Ho-Pham,L.T.,Nguyen,N.D.,Lai,T.Q.,&Nguyen,T.V.(2010).Contributionsoflean massandfatmasstobonemineraldensity:Astudyinpostmenopausalwomen. BMCMusculoskeletalDisorders,11,59.

InternationalOsteoporosisFoundation (IOF).(2001). Osteoporosisin theEuropean Community:acalltoaction.Anauditofpolicydevelopmentssince1998Available from http://www.iofbonehealth.org/publications/eu-policy-report-of-2001.html

Accessed15.04.11.

InternationalOsteoporosisFoundation(IOF).(2008).OsteoporosisintheEuropeanUnion in2008:TenyearsofprogressandongoingchallengesAvailablefrom http://www.iof-bonehealth.org/publications/eu-policy-report-of-2008.html.

Jones,G.,Nguyen,T.,Sambrook,P.,Kelly,P.J.,&Eisman,J.A.(1994).Progressivelossof boneinthefemoralneckinelderlypeople:LongitudinalfindingsfromtheDubbo osteoporosisepidemiologystudy.BritishMedicalJournal,309,691–695.

Kanis,J.A.,Burlet,N.,Cooper,C.,Delmas,P.D.,Reginster,J.Y., Borgstrom,F.,etal. (2008).Europeanguidanceforthediagnosisandmanagementofosteoporosisin postmenopausalwomen.OsteoporosisInternational,19,399–428.

Kanis,J.A.,&Johnell,O. (2005).Requirements for DXAfor themanagement of osteoporosisinEurope.OsteoporosisInternational,16,229–238.

Kenny,A.M.,Prestwood,K.M.,Marcello,K.M.,&Raisz,L.G.(2000).Determinantsof bonedensityinhealthyoldermenwithlowtestosteronelevels.TheJournalsof Gerontology.SeriesA,BiologicalSciencesandMedicalSciences,55,492–497.

Khan,K.,McKay,H.,Kannus,P.,Bailey,D.,Wark,J.,&Bennell,K.(2001).Physicalactivity andbonehealth.Champaign,IL:HumanKinetics.

Lane,N.E.(2006).Epidemiology,etiology,anddiagnosisofosteoporosis.American JournalofObstetricsandGynecology,194,S3–S11.

Lau,E.M.,Leung,P.C.,Kwok,T.,Woo,J.,Lynn,H., Orwoll,E.,etal.(2006).The determinantsofbonemineraldensityinChinesemen-resultsfromMr.Os(Hong Kong),thefirstcohortstudyonosteoporosisinAsianmen.Osteoporosis Interna-tional,17,97–303.

Lim,K.,&Taylor,L.(2005).Factorsassociatedwithphysicalactivityamongolder peopleapopulation-basedstudy.PreventiveMedicine,40,33–40.

Looker,A.C.,Melton,L.J.,3rd,Harris,T.,Borrud,L.,Shepherd,J.,&McGowan,J.(2009).

Age,gender,and race/ethnic differencesintotal bodyand subregionalbone density.OsteoporosisInternational,20,1141–1149.

Lopes, C.(2000). Reproducibilityand validityof asemi-quantitativefood-frequency questionnaire(dissertation)Porto,Portugal:UniversityofPorto(inPortuguese).

Maı¨moun, L.,&Sultan,C.(2011).Effectsofphysicalactivityonboneremodeling. Metabolism:ClinicalandExperimental,3,373–388.

Mavroeidi,A.,Stewart,A.D.,Reid,D.M.,&Macdonald,H.M.(2009).Physicalactivity anddietarycalciuminteractionsinbonemassinScottishpostmenopausalwomen. OsteoporosisInternational,20,409–416.

Nahas,E.A.,Kawakami,M.S.,Nahas-Neto,J.,Buttros,Dde.A.,Cangussu,L.,&Rodrigues, A.B.(2011).AssessmentofriskfactorsforlowbonemineraldensityinBrazilian postmenopausalwomen.Climacteric:TheJournaloftheInternationalMenopause Society,14,220–227.

Nguyen,T.V.,Center,J.R.,&Eisman,J.A.(2000).Osteoporosisinelderlymenand women:effectsofdietarycalcium,physicalactivity,andbodymassindex.Journal ofBoneandMineralResearch,15,322–331.

Nguyen,T.V.,Sambrook,P.N.,&Eisman,J.A.(1998).Boneloss,physicalactivity,and weightchangeinelderlywomen:theDubboOsteoporosisEpidemiologyStudy. JournalofBoneandMineralResearch,13,1458–1467.

Nurzenski,M.K.,Briffa,N.K.,Price,R.I.,Khoo,B.C.,Devine,A.,Beck,T.J.,etal.(2007).

Geometricindicesofbonestrengthareassociatedwithphysicalactivityand dietary calciumintake inhealthyolder women.Journal ofBoneandMineral Research,22,416–424.

Ono,R.,Hirata,S.,Yamada,M.,Nishiyama,T.,Kurosaka,M.,&Tamura,Y.(2007).

ReliabilityandvalidityoftheBaeckephysical activityquestionnaireinadult womenwithhipdisorders.BMCMusculoskeletalDisorders,8,61.

Philippaerts,R.M.,&Lefevre,J.(1998).Reliabilityandvalidityofthreephysicalactivity questionnairesinFlemishmales.AmericanJournalofEpidemiology,147,982–990.

Pluijm,S.M.,Visser,M.,Smit,J.H.,Popp-Snijders,C.,Roos,J.C.,&Lips,P.(2001).

Determinantsofbonemineraldensityinoldermenandwomen:Bodycomposition asmediator.JournalofBoneandMineralResearch,16,2142–2151.

Reid,I.R.,Plank,L.D.,&Evans,M.C.(1992).Fatmassisanimportantdeterminantof wholebodybonedensityinpremenopausalwomenbutnotinmen.TheJournalof ClinicalEndocrinologyandMetabolism,75,779–782.

Rose,D.J.(2003).Fallproof:Acomprehensivebalanceandmobilitytrainingprogram. Champaign,IL:HumanKinetics.

Schoffl,I.,Kemmler,W.,Kladny,B.,Vonstengel,S.,Kalender,W.A.,&Engelke,K.(2008).

InhealthyelderlypostmenopausalwomenvariationsinBMDandBMCatvarious skeletalsitesareassociatedwithdifferencesinweightandleanbodymassrather thanbyvariationsinhabitualphysicalactivity,strengthorVO2max.Journalof Musculoskeletal&NeuronalInteractions,8,363–374.

Semanick,L.M.,Beck,T.J.,Cauley,J.A.,Wheeler,V.W.,Patrick,A.L.,Bunker,C.H.,etal. (2005).AssociationofbodycompositionandPAwithproximalfemurgeometryin middle-agedandelderlyAfro-Caribbeanmen:theTobagobonehealthstudy. CalcifiedTissueInternational,77,160–166.

StatisticsPortugal.(2002).Census2001.Finalresults:XIVPopulationcensus.IVHousing census.Lisbon:StatisticsPortugal(inPortuguese).

StatisticsPortugal.(2009). ResidentPopulationProjections–Portugal–2008–2060. Lisbon:StatisticsPortugal.DepartmentofSocialandDemographicStatistics(in Portuguese).

Stenderup,K.,Justesen,J.,Clausen,C.,&Kassem,M.(2003).Agingisassociatedwith decreasedmaximallifespanandacceleratedsenescenceofbonemarrowstromal cells.Bone,33,919–926.

Stewart,K.J.,Bacher,A.C.,Hees,P.S.,Tayback,M.,Ouyang,P.,&JandeBeur,S.(2005).

Exerciseeffectsonbonemineraldensityrelationshipstochangesinfitnessand fatness.AmericanJournalofPreventiveMedicine,28,453–460.

Travison,T.G.,Araujo,A.B.,Esche,G.R.,Beck,T.J.,&McKinlay,J.B.(2008).Leanmass andnotfatmassisassociatedwithmaleproximalfemurstrength.JournalofBone andMineralResearch,23,189–198.

U.S.DepartmentofHealthandHumanServices.(2004).Bonehealthandosteoporosis: Areportofthesurgeongeneral.Rockville,MD:U.S.DepartmentofHealthand HumanServices,OfficeoftheSurgeonGeneralRetrievedfrom http://www.sur-geongeneral.gov/library.

Uusi-Rasi,K.,Sieva¨nen,H.,Pasanen,M.,Beck,T.J.,&Kannus,P.(2008).Influenceof calciumintakeandPAonproximalfemurbonemassandstructureamongpre-and postmenopausalwomen.A10-yearprospectivestudy.CalcifiedTissue Internation-al,82,171–181.

Vuillemin,A.,Guillemin,F.,Jouanny,P.,Denis,G.,&Jeandel,C.(2001).Differential influenceofPAonlumbarspineandfemoralneckbonemineraldensityinthe elderlypopulation.TheJournalsofGerontology.SeriesA,BiologicalSciencesand MedicalSciences,56,248–253.

Walsh,M.C.,Hunter,G.R.,&Livingstone,M.B.(2006).Sarcopeniainpremenopausal andpostmenopausalwomenwith osteopenia,osteoporosis andnormalbone mineraldensity.OsteoporosisInternational,17,61–67.

Yoshikawa,T.,Turner,C.H.,Peacock,M.,Slemenda,C.W.,Weaver,C.M.,Teegarden, D.,etal.(1994).Geometricstructureofthefemoralneckmeasuredusing dual-energyX-rayabsorptiometry.JournalofBoneandMineralResearch,9,1053– 1064.