UBC Faculty Research and Publications

Using a single question to assess physical activity in older adults: a reliability and validity study Gill, Dawn P; Jones, Gareth R; Zou, Guangyong; Speechley, Mark Feb 28, 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12874_2011_Article_702.pdf [ 236.9kB ]
JSON: 52383-1.0223945.json
JSON-LD: 52383-1.0223945-ld.json
RDF/XML (Pretty): 52383-1.0223945-rdf.xml
RDF/JSON: 52383-1.0223945-rdf.json
Turtle: 52383-1.0223945-turtle.txt
N-Triples: 52383-1.0223945-rdf-ntriples.txt
Original Record: 52383-1.0223945-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessUsing a single question to assess physical activityin older adults: a reliability and validity studyDawn P Gill1*, Gareth R Jones2, Guangyong Zou3,4 and Mark Speechley3AbstractBackground: Single-item physical activity questions provide a quick approximation of physical activity levels. Whilerecall questionnaires provide a more detailed picture of an individual’s level of physical activity, single-itemquestions may be more appropriate in certain situations. The aim of this study was to evaluate two single-itemphysical activity questions (one absolute question and one relative question) for test-retest reliability, convergentvalidity, and discriminant validity, in a sample of older adults.Methods: Data was obtained from the Project to Prevent Falls in Veterans, a fall risk-factor screening andmodification trial. One question measured absolute physical activity (seldom, moderately, vigorously active) andone measured relative physical activity (more, about as, less active than peers). Test-retest reliability was examinedusing weighted Kappa statistics () in a sample of 43 subjects. Validity was assessed using correlation coefficients(r) in participants who received clinical assessments (n = 159).Results: The absolute physical activity question was more reliable than the relative physical activity question ( =0.75 vs.  = 0.56). Convergent validity, however, was stronger for the relative physical activity question (r = 0.28 to0.57 vs. r = 0.10 to 0.33). Discriminant validity was similar for both questions. For the relative physical activityquestion, there was moderate agreement when this question was re-administered seven days later, fair tomoderate/good associations when compared with indicators of physical function, and little to no associationswhen compared with measures hypothesized to be theoretically not related to physical activity.Conclusions: The relative physical activity question had the best combination of test-retest reliability, convergentvalidity and discriminant validity. In studies requiring a measure of physical activity, where physical activity is notthe primary focus and more detailed measures are not feasible, a single question may be an acceptable alternative.Keywords: Physical activity, self-report, single-item measure, assessment, validity, reliability, older adultsBackgroundWhen selecting measures for a study, investigators usuallyneed to strike a balance among several factors such asrequired sample size, the level of detail needed, theresources available, and the burden posed by their mea-surement protocol on research participants. In particular,for a given research budget, there is typically a trade-offbetween measurement detail and sample size [1].Each of these issues is present in the study of physicalactivity (PA) in older adults. Regular PA assists withmaintaining independence and preventing disabilityamong older adults and it is associated with a decreasedrisk of morbidity and all-cause mortality [2]. PA isdefined as “any bodily movement produced by skeletalmuscles that results in energy expenditure” [3]. PA is acomplex behavioral construct that can be categorizedand quantified in many ways. For example, PA can bebroken down into routine activities such as housework,and those done for exercise, such as swimming. Eachspecific type of PA can be quantified in terms of fre-quency, intensity, and duration [4].These complexities are reflected in the many methodsused to measure PA or related energy expenditure inolder adults [4-7]. Measurement of PA can be categor-ized into direct and indirect methods. Direct methodsare defined as those that measure movement as it* Correspondence: dpgill2@uw.edu1National Alzheimer’s Coordinating Center, Department of Epidemiology,University of Washington, Seattle, WA, USAFull list of author information is available at the end of the articleGill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20© 2012 Gill et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.occurs and indirect methods provide indicators of PAand energy expenditure [5]. Examples of direct methodsinclude motion sensors, such as pedometers, acceler-ometers, and Global Positioning Systems, whereas indir-ect methods include daily PA records or log books andself-report questionnaires.There are a number of recall questionnaires that havebeen used in older adult populations, with varyingdegrees of evidence for reliability and validity [8]. Whilethese questionnaires provide a more detailed picture ofan individual’s PA, global questionnaires or single-itemquestions may be favored in certain situations. Wheninvestigators have a choice of questions, they mightcompare the evidence of validity and reliability in mak-ing their selection. Reliability and validity results fromexisting single-item PA questions [9-12] indicate a needto evaluate additional single-item questions as possiblemeasures of PA under certain conditions (i.e., when PAis not the primary focus of a study but a quick approxi-mation of activity levels is of interest as a covariate orpossible confounding factor, when the sample size islarge, when resources are limited, and when more com-plex methods would add to respondent burden).Similar research has been done with general healthmeasures. A previous study found that two single-itemgeneral self-rated health measures showed good mea-surement properties when compared to a multi-iteminstrument, thus providing a less burdensome alternative[13]. In that study, researchers compared “standard” and“comparative” versions of general self-rated health mea-sures, where the comparative version referred to a ques-tion that had respondents compare their general healthto a reference group. Findings indicated that both ques-tions represented reasonably similar assessments ofhealth. Another study in this area, which comparedthree different single-item questions of self-rated health(two “standard” questions and one “comparative” ques-tion) found similar results [14].In a previously completed fall risk factor modificationtrial, two single-item questions of PA were included,both intended to easily classify activity levels of partici-pants. Similar to the self-rated health literature, onequestion was a “standard” measure and one was a “com-parative” measure. Specifically, one question measuredabsolute PA (seldom, moderately, vigorously active) andthe other measured relative PA (more, about as, lessactive than peers). Using the self-rated health literatureas a model, since both PA questions have the sameintent (i.e., to quickly classify PA levels), it is of interestto determine if properties of reliability and validity aresimilar between these questions and whether they couldbe used interchangeably. Thus, the aim of this study wasto evaluate the test-retest reliability, convergent validityand discriminant validity of an absolute PA questionand a relative PA question, in a sample of community-dwelling older adults.ResultsThe characteristics of participants who took part in thereliability sub-study are described in detail elsewhere[15]. Briefly, the mean age was 79 (standard deviation(SD) 2.9) years and approximately one-half were male.For the validity sample, the mean age was 80 (SD 3.9)years, and close to two-thirds were males. Other charac-teristics of participants included in the validity sampleare presented in Table 1. About 20% self-reported fairor poor health, 38% reported one or more falls in thepast 12 months, 49% reported that their memory wasworse than five years ago, 15% reported being seldomactive and 12% reported that they were less active com-pared to their peers. In comparison to women, menwere older (mean age 81 years, SD 3.5), and a slightlyhigher percentage reported fair or poor health, one ormore falls in the past month and worse memory com-pared to five years earlier. Men and women providedsimilar responses in regard to their PA compared totheir peers. The median time between administration ofvalidation measures and PA questions ranged between33 days for the subset of participants who had beenadministered both the earlier version of the interRAICommunity Health Assessment (interRAI) and Veterans’Comprehensive Assessment (VCA), and 37 days whenconsidering all participants in the validation sample(also see Figure 1).Results indicated that the absolute PA question hadbetter test-retest reliability than the relative PA ques-tion. The weighted kappa value for absolute PA was0.75 (95% confidence interval (CI): 0.60 to 0.91) whereasfor relative PA, the weighted kappa value was 0.56 (95%CI: 0.30 to 0.82). For the absolute PA question, theweighted kappa value indicated substantial agreementwhereas for relative PA, the weighted kappa value indi-cated agreement in the moderate range.The validation results for both the absolute and rela-tive PA questions are presented in Table 2. For both PAquestions, correlation coefficients were in the expecteddirections according to the type of validity beingassessed. For the relative PA question, there was greatercontrast between values obtained for convergent validityand discriminant validity. For absolute PA, correlationswith convergent validation measures ranged from 0.10(95% CI: 0.00 to 0.26) to 0.33 (95% CI: 0.19 to 0.49),indicating relationships ranging from little to a fairdegree of association. For relative PA, correlations wereconsistently higher, with most comparisons indicatingfair to moderate or good associations. Specifically, corre-lations ranged from 0.28 (95% CI: 0.15 to 0.44) to 0.57(95% CI: 0.38 to 0.78). The total score on the balanceGill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 2 of 10assessment, limiting PA due to fear of falling, and all ofthe gait measures (unsteady gait, gait-path, gait-trunk,and gait abnormality) had stronger positive correlationswith relative PA, compared to absolute PA.For discriminant validity, the magnitude of correlationcoefficients was similar between absolute PA and rela-tive PA. For absolute PA, correlations with discriminantvalidation measures ranged from 0.04 (95% CI: 0.00 to0.19) to 0.29 (95% CI: 0.15 to 0.50) and for relative PA,correlations ranged from 0.02 (95% CI: 0.00 to 0.12) to0.24 (95% CI: 0.00 to 0.44). In general, comparisons ofdiscriminant validity measures and PA questions led tocorrelations that indicated little to no association.DiscussionThe absolute PA question had better test-retest reliabil-ity than the relative PA question. Paradoxically, evidencefor convergent validity was stronger for relative PAcompared to absolute PA. For both questions, resultsindicated evidence for discriminant validity. The relativePA question had the best combination of test-retestreliability, convergent validity and discriminant validity.Specifically, there was moderate agreement when thisquestion was re-administered seven days later, fair tomoderate or good associations when compared withindicators of physical function, and little to no associa-tions when compared with measures hypothesized to betheoretically not related to PA. Although we wereunable to evaluate the five-level form of the relative PAquestion, a previous study examining the validity of asimilar question from the National Health InterviewSurvey (NHIS) found that very little was gained with the5-level question compared to the 3-level question [12].Indicators of physical function, often referred to asindirect measures of PA, have not been commonly usedto evaluate the convergent validity of single-item PAquestions in older adults, despite recommendations fortheir use [16,17]. One study, evaluating two different PAquestions in older adults, examined convergent validityagainst indicators of health (i.e., health conditions suchTable 1 Participant characteristics (validity sample)aTotal (N = 159) Women (n = 58) Men (n = 101)Characteristic n (%) n (%) n (%)Activity levelSeldom active 24 (15.1) 7 (12.1) 17 (16.8)Moderately active 96 (60.4) 41 (70.7) 55 (54.5)Vigorously active 38 (23.9) 10 (17.2) 28 (27.7)Activity level compared to peersLess active 19 (11.9) 7 (12.1) 12 (11.9)About as active 61 (38.4) 21 (36.2) 40 (39.6)More active 78 (49.1) 30 (51.7) 48 (47.5)Participant StatusVeteran 103 (64.8) 3 (5.2) 100 (99.0)Caregiver 56 (35.2) 55 (94.8) 1 (1.0)Finances at end of monthJust enough money 35 (22.0) 11 (19.0) 24 (23.8)Money left over 123 (77.4) 47 (81.0) 76 (75.2)1+ falls in past 12 months 61 (38.4) 17 (29.3) 44 (43.6)1+ injurious falls in past 12 months 20 (12.6) 8 (13.8) 12 (11.9)Memory compared to 5 years earlierWorse 78 (49.1) 25 (43.1) 53 (52.5)About the same 79 (49.7) 32 (55.2) 47 (46.5)Better 2 (1.3) 1 (1.7) 1 (1.0)Self-rated healthPoor 4 (2.5) 2 (3.5) 2 (2.0)Fair 28 (17.6) 7 (12.1) 21 (20.8)Good 62 (39.0) 24 (41.4) 38 (37.6)Very good 48 (30.2) 15 (25.9) 33 (32.7)Excellent 17 (10.7) 10 (17.2) 7 (6.9)aAll participants included had a second clinical assessment (CA2) and were evaluated with the earlier version of the interRAI Community Health Assessment(interRAI). A subset (n = 94) receiving a CA2 were also evaluated with the Veterans’ Comprehensive Assessment (VCA)Percentages were calculated excluding those with missing valuesGill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 3 of 10                                     Lost to follow-up n = 15 Refusals   n = 4 Deceased   n = 3  Moved away  n = 2  Too sick   n = 2  Long-term care  n = 3 Completed CA1  (interRAI and VCA)  n = 129 Refusals   n = 4 Lost to follow-up  n = 1 Completed CA1 (interRAI only) n = 80 Median number of days between: 33 to 37 daysc   Validity Sample  n = 159 All participants have CA2 interRAI data  and a subset (n = 94) also have CA2 VCA data  Proxy interviews n = 3 Refusals   n = 3 Lost to follow-up  n = 1 Deceased   n = 1 Completed CA2b  (interRAI and VCA) n = 102 Completed telephone interview with physical activity questions n = 94 Proxy interviews n = 3 Refusals  n = 3  Lost to follow-up  n = 3 Too sick  n = 1  Completed CA2b  (interRAI only) n = 75 Completed telephone interview with physical activity questions n = 65 Zero-modifiable Fall Risk Factors Group n = 91 Refusals for clinical assessmentsa  n = 11 Specialized Geriatric  Services Group n = 188 Refusals for clinical assessmentsa  n = 59 Figure 1 Formation of validity sample from the Project to Prevent Falls in Veterans. The dashed line indicates the point in the largerstudy where measures of interest for the present study begin. a Clinical assessments were not required to be part of the larger study. bValidation measures used in the present study were taken from CA2. c CA2 and telephone interviews were completed as close together in timeas possible. The median number of days was 37 days for all participants (n = 159) and 33 days for the subset who also had VCA data (n = 94).Abbreviations: CA1 = first clinical assessment; interRAI = interRAI Community Health Assessment (earlier version); VCA = Veterans’ ComprehensiveAssessment; CA2 = second clinical assessment.Gill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 4 of 10as heart attack, stroke, and diabetes). This study did notreport any type of validity coefficients, making compari-sons with our findings difficult [9].Two other studies that evaluated an additional fourPA questions in populations of older adults, examinedvalidity by comparing questions with summary measuresfrom PA recall questionnaires. In the first study, a PAquestion designed to be used as a screening question inprimary care was evaluated in a population of olderwomen [11]. This question, “As a rule, do you do atleast half an hour of moderate or vigorous exercise(such as walking or sport) on five or more days of theweek?”, was compared to two summary scores from theNew Zealand Physical Activity Questionnaire - LongForm. Results indicated moderate agreement ( = 0.46to 0.56). In the second study, three PA questions fromthe NHIS (job-related activity, main daily activity, andactivity compared to peers) were compared with sum-mary measures from a detailed PA question set [12].The main daily activity question asked, “How muchhard physical work is required in your main daily activ-ity? Would you say a great deal, a moderate amount, alittle, or none?” The activity compared to peers ques-tion, “Would you say that you are physically moreactive, less active, or about as active compared to otherpersons you age?”, was also expanded to a 5-levelquestion with the following response options: a lotmore, a little more, about the same, a little less, a lotless. For participants 65 years of age or older, correlationcoefficients ranged from 0.17 to 0.21 for the main dailyactivity question and from 0.24 to 0.28 for the activitycompared to peers question. The validity results fromthe present study, in particular for the relative PA ques-tion, have been similar or better than previous studiesof single-item PA questions in older adults.At least two studies have evaluated test-retest reliabil-ity of single-item PA questions in older populations. Inthe first study, researchers found intraclass correlationcoefficients (ICCs) ranging from 0.75 to 0.80 for two PAquestions that asked regular exercisers about their fre-quency and intensity of activity [9]. Another study eval-uated the test-retest reliability of three different PAquestions (work PA, strenuous PA, and moderate PA) ina sample of participants from the Canadian MulitcentreOsteoporosis Study [10]. The kappa statistic was 0.57(0.47 to 0.68) for the strenuous PA question and 0.30(0.23 to 0.37) for the moderate PA question.Reliability results achieved in the present study for therelative PA question were similar or better than thosereported by Nadalin et al. [10] but worse than thosereported by Davis et al. [9]. Comparing the results inthe present study to those reported by Davis et al. [9] isTable 2 Convergent and discriminant validity results for PA questionsAbsolute Questiona Relative QuestionbValidation Measure r (95% CI)c r (95% CI)cConvergent validityTotal score on Berg Balance Scale (n = 158) 0.31 (0.17, 0.45) 0.48 (0.35, 0.59)Unsteady gait (n = 152) 0.24 (0.12, 0.40) 0.40 (0.26, 0.56)Gait-path (n = 93) 0.32 (0.17, 0.53) 0.57 (0.38, 0.78)Gait-trunk (n = 94) 0.29 (0.15, 0.50) 0.42 (0.25, 0.63)Gait-abnormality (n = 93) 0.26 (0.00, 0.47) 0.36 (0.20, 0.57)Lower extremity weakness (n = 92) 0.21 (0.00, 0.43) 0.37 (0.21, 0.58)Postural stability (n = 94) 0.17 (0.00, 0.38) 0.33 (0.17, 0.54)Walking ability (n = 158) 0.33 (0.19, 0.49) 0.42 (0.28, 0.58)Housework difficulty (n = 158) 0.28 (0.16, 0.45) 0.34 (0.20, 0.50)Limits activity-fear of falling (n = 155) 0.10 (0.00, 0.26) 0.28 (0.15, 0.44)Pain affects mobility (n = 90) 0.28 (0.00, 0.50) 0.34 (0.18, 0.55)Discriminant validityHome hazards (n = 152) 0.04 (0.00, 0.19) 0.02 (0.00, 0.12)Vision (n = 157) 0.05 (0.00, 0.21) 0.07 (0.00, 0.23)Skin problems (n = 157) 0.13 (0.00, 0.29) 0.15 (0.00, 0.32)Hearing deficit (n = 94) 0.29 (0.15, 0.50) 0.24 (0.00, 0.44)a Response categories: i) seldom active; ii) moderately active; or iii) vigorously active.b Response categories: i) less active; ii) about as active; or iii) more active.c Specific type of correlation coefficient (r): Spearman’s rho (total score on Berg Balance Scale) and Cramer’s V (all other validation measures)Validation measures and PA questions were ordered so that higher scores indicated better functioning or higher PA levels.Abbreviations: CI = Confidence Interval; PA = Physical Activity.Gill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 5 of 10also problematic, since the PA questions evaluated inthat study were only posed to participants who hadalready reported engaging in regular exercise.Indicators of physical function have been used to eval-uate the convergent validity of many PA recall question-naires designed for older adults. For a number of themost well-known questionnaires, evidence for conver-gent validity is not substantially stronger than thatobtained in this study; in fact, in some instances, therelative PA question evaluated in this study, performedbetter. For example, correlations between summaryscores from the Community Healthy Activities ModelProgram for Seniors (CHAMPS) Physical Activity Ques-tionnaire and various measures of physical functioningranged between 0.10 and 0.54 [16,18-20]. For theCHAMPS Physical Activity Questionnaire and the YalePhysical Activity Survey, test-retest reliability was evalu-ated over a similar interval to this study (one to twoweeks), and ICCs ranged from 0.55 to 0.79 [18,19,21].The intent of both the absolute and the relative PAquestions was to quickly and easily classify older adults bytheir activity level. Since specific details related to fre-quency, duration and intensity are not referenced withinthe relative PA question, this question will remain accu-rate for assessment even when PA recommendations forolder adults are revised, such as was done in the UnitedStates in 2007 [22] and in Canada in 2011[23]. The relativePA question may also be less prone to recall errors, com-pared to the absolute PA question, since participants donot need to remember the duration or frequency of theirtypically performed activities.It is known that in general, people tend to over-reportPA levels [24]. In the self-reported health literature, it wasnoted that with increasing age, people tended to overesti-mate their health when comparing themselves to others oralternatively, they underestimated the health of others[14]. Thus, it is plausible that the participants in this studymay have overestimated their PA, and perhaps to a greaterextent when responding to the relative PA question. Thisshould be kept in mind when interpreting the results ofthis study and when considering the merits of measuringPA using an absolute or a relative question.Other limitations exist for the present study. Partici-pants included in this study were Canadian veterans ofWorld War II or the Korean War and their caregivers, ahighly selected group of older adults. In addition, someof the validation measures were only available on parti-cipants in the study who had reported at least one mod-ifiable fall risk factor. The Project to Prevent Falls inVeterans (PPFV) began as a randomly selected sample;however, only 13% of the original participants wereincluded as part of the risk factor modification trial anda smaller percentage completed the second clinicalassessment and the final telephone interview. As aresult, it is likely that the participants included in thisstudy are different than the general population of olderCanadian veterans and their caregivers. Caution shouldbe taken in generalizing results from our study to popu-lations that may differ clinically and demographically.The present analyses were done because we had datathat allowed us to do these comparisons, but were notpart of a validation study planned a priori. It is thereforepossible that the modest validity correlations achievedmay be partially due to the measures selected for valida-tion. Since there is no widely accepted criterion of PA[24], we chose to evaluate the convergent validity of twosingle-item PA questions, by comparing them with indi-cators of physical functioning. We recognize, however,that capacity to perform PA does not equal actual per-formance. As a result, correlation coefficients indicatingmore than a moderate association may not be possiblewhen using indicators of physical functioning as valida-tion measures. A related limitation is that while some ofthe indicators of physical function were objectively mea-sured performance-based outcomes, others were mea-sures of self-reported functional ability. Self-reportedmeasures can be affected by factors such as cognitiveimpairment and guessing among older populations [25].Additionally, it would have been preferable if the indica-tors of physical functioning were measured at the sametime as the PA questions. Even so, we hypothesize thatany resulting bias is likely toward the null, indicatingthat correlations may have been stronger if these mea-sures had been conducted closer in time.ConclusionsIn large sample research, there is a trade-off betweenthe intensity of measurement of a single variable andthe comprehensiveness of all variables. In this study, arelative PA question had the best combination of test-retest reliability, convergent validity and discriminantvalidity. The magnitude of the reliability and validitycoefficients achieved for this question are similar, and insome cases better, than those previously reported forother single-item PA questions evaluated in older adults.Reliability and validity results of many recall question-naires for older adults have also not substantiallyexceeded the results obtained in this study. This simplePA question may be useful in studies of older adultswhere PA is not the primary focus, but a brief classifica-tion of activity levels is needed.In this study, we have taken an initial step in evaluat-ing convergent validity of a relative PA question usingindicators of physical functioning as validation measures.Future research should evaluate convergent validityusing other validation measures such as accelerometersand more detailed recall questionnaires. This questionor other single-item questions cannot replace recallGill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 6 of 10questionnaires or other direct measures of PA whenresources are available or when study objectives requiremore comprehensive measures. In summary, this simplePA question may provide an alternative to researcherswhen lengthy PA measures, which increase both costand participant burden, are not possible or necessary.MethodsParticipantsWe used data from the PPFV, a fall risk-factor screeningand modification trial. The PPFV was approved by theResearch Ethics Board for Health Sciences ResearchInvolving Human Subjects at the University of WesternOntario. Written informed consent was obtained fromall participants.In 2002, the PPFV began with screening question-naires mailed to 3,000 addresses of older adults living incentral or southwest Ontario, sampled randomly fromthe client list of Veterans Affairs Canada. To be eligible,persons had to be: i) a Canadian veteran of World WarII or the Korean War or someone providing care forthis individual; ii) living independently; and iii) able tounderstand and provide responses to a screening ques-tionnaire. Caregivers were not proxy respondents forveterans but were recruited as full participants in thestudy. Questionnaires were received from 1,913 veteransand 1,398 caregivers, which corresponded to a 70%response rate for veterans. The response rate for care-givers could not be calculated since the number ofveterans who had a caregiver was unknown.Participants from the London and Windsor regionswere eligible to enroll in a one-year risk factor modifica-tion effectiveness trial. In total, there were 348 partici-pants who consented to be re-contacted and who hadself-reported at least one modifiable risk factor for fall-ing. These participants were randomized to either theSpecialized Geriatric Services (SGS) group (n = 188) orthe Family Physician group (n = 160). Participants inthe SGS group made two in-person visits where theyreceived comprehensive clinical assessments; the firstclinical assessment was conducted at the start of thetrial (CA1) and the second clinical assessment was con-ducted at the conclusion of the trial (CA2). The SGSgroup was evaluated with the interRAI [26] and anassessment tool developed specifically for the PPFV, theVCA (see Figure 1). Because participants in the FamilyPhysician group did not receive geriatric assessments,they were not included in the present analyses.The main analysis of the PPFV revealed no significantdifferences between randomized groups in regard tofalls or injurious falls. Accordingly, data from the PPFVwere analyzed as a prospective cohort study. Participantswith no reported modifiable risk factors for falling(Zero-Mod group) formed an open study arm (n = 91).Most participants in this group also received CA1 andCA2; however, only the interRAI was administered tothese participants. At the end of the trial, a telephoneinterview including two PA questions was administeredto all study groups. This telephone interview was com-pleted as soon as possible after CA2. Participants in theSGS and Zero-Mod groups, who completed both a CA2and the telephone interview, made up the validity sam-ple (n = 159). Additional details related to the formationof the validity sample are outlined in Figure 1.A reliability sub-study of items in the telephone inter-view was conducted in a convenience sample from thePPFV. This sub-study evaluated test-retest reliability,which refers to agreement among measurements on thesame participants at different time points [27]. Partici-pants who completed the telephone interview wereasked if they would be willing to be re-interviewedseven days later, by the same interviewer. This processcontinued until the target sample size for the reliabilitysub-study was achieved (n = 43). Additional details onthe PFFV have been presented elsewhere [28,29].Self-report PA measuresThe absolute PA question was developed for the PPFV.This question, “What best describes your activity level?”had three response options: vigorously active for at least30 min, 3 times per week; moderately active at least 3times per week; or seldom active, preferring sedentary activ-ities. Participants were asked to select the response optionthat best described their typical activity level. The relativePA question is similar to two questions included in the1985 NHIS in the United States [12,30]. This question,“Compared to other people your own age, do you thinkyou are . . . “ had five response options: much more active,more active, about as active, less active, or much lessactive. Due to small numbers in the two most extremecategories, this question was re-coded as follows: muchmore active and more active were collapsed to more active,about as active remained unchanged, and less active andmuch less active were collapsed into less active.Validation measuresValidation measures were taken from CA2 since thesemeasurements were completed closest in time to the PAquestions. The earlier version of the interRAI Commu-nity Health Assessment is a standardized assessmenttool that is a subset of the Minimum Data Set for HomeCare (MDS-HC) version 2.0 [26]. Reliability and validityof the MDS-HC has been previously reported in com-munity settings [31,32]. The version of the interRAIused in the present study provided detailed assessmentin the following domains: cognition, communication/hearing, mood/behaviour, social and physical function-ing, continence, disease diagnoses, health conditions,Gill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 7 of 10preventive health measures, nutrition/hydration, skincondition, environmental/home safety, service utiliza-tion, and medications. An additional section providedassessment of risk factors for falling.The VCA tool was developed under the guidance of ageriatrician and a physical therapist specializing in geria-tric assessments. This instrument was designed to beadministered by trained geriatric health care profes-sionals to capture information related to fall risk, includ-ing: home environment risk, chronic disease, healthstatus indicators, sensory function, mobility, continence,cognition, pain, footwear, blood pressure, balance,strength, range of motion, gait and medications.Convergent validity is present if two measures believedto reflect the same underlying phenomenon correlatestrongly [33]. Eleven indicators of physical functionfrom the interRAI and VCA were selected as convergentvalidity measures. Seven of these indicators are objectiveperformance-based measures of physical function andfour are self-report measures of functional ability. PAquestions and indicators of physical function wereordered such that higher scores indicated higher PAlevels or better functioning. We hypothesized a priorithat evidence for convergent validity would exist if PAquestions positively correlated with indicators of physi-cal function.Discriminant validity indicates that two measuresbelieved to assess different characteristics will have littleor no relationship [33]. Measures hypothesized to betheoretically not related to PA were selected as discrimi-nant validity measures, and we hypothesized correlationsclose to zero. Three objectively measured items and oneself-report item from the geriatric assessments wereselected for evaluation. See Table 3 for an overview ofTable 3 Convergent and discriminant validation measuresValidation Measure Response Options Assessment ToolaObjective Measures (Convergent Validity)1. Total score on Berg Balance Scale Score range from 0 to 56 interRAI2. Unsteady gait ▪ No▪ YesinterRAI3. Gait (path) ▪ Straight without walking aid VCA4. Gait (trunk) ▪ No sway, flexion, use of arms, or walking aid▪ Sway/Uses aid/Flexion/SpreadsVCA5. Gait (abnormality) ▪ No▪ YesVCA6. Evidence of lower extremity weakness ▪ No▪ YesVCA7. Postural stability test (nudged)b ▪ Steady▪ Staggers, grabs, catches self/Beings to fallVCASelf-Report Measures (Convergent Validity)1. Ability to walk 3 city blocks in last 3 days ▪ No difficulty▪ Difficulty/UnableinterRAI2. Difficulty with ordinary housework ▪ No difficulty▪ Some/Great difficultyinterRAI3. Limits going outdoors (fear of falling) ▪ No▪ YesinterRAI4. Pain affects mobility ▪ No▪ YesVCAObjective Measures (Discriminant Validity)1. Vision ▪ Adequate▪ Impaired/Moderately, highly or severely impairedinterRAI2. Skin problems ▪ No▪ YesinterRAI3. Evidence of hearing deficit ▪ No▪ YesVCASelf-Report Measures (Discriminant Validity)1. Home environment hazardous ▪ No▪ YesinterRAIaRefers to the measurement tool that contained the validation measurebItem from the Tinetti Balance Assessment ToolAbbreviations: interRAI = interRAI Community Health Assessment (earlier version); VCA = Veterans’ Comprehensive AssessmentGill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 8 of 10validation measures and corresponding responseoptions.Statistical analysesTest-retest reliability was assessed using the weightedkappa statistic along with 95% CIs [34]. Guidelinesadopted for interpreting the strength of agreement forkappa values were as follows: less than 0.41 representspoor to fair agreement, 0.41 to 0.6 represents moderateagreement, 0.61 to 0.8 represents substantial agreement,and 0.81 to 1 represents almost perfect agreement [35].Convergent and discriminant validity was assessed bycorrelation coefficients. Spearman’s rho was used whenthe PA questions were compared with continuous vali-dation measures whereas Cramer’s v was used for vali-dation measures that were categorical. For theSpearman’s rho correlations, the Fisher z transformationwas used to obtain 95% CIs [36]. Qualitative descriptorsadopted for interpreting correlation coefficients were asfollows: 0 to 0.25 represents little or no association, 0.26to 0.5 represents a fair association, 0.51 to 0.75 repre-sents moderate to good association, and greater than0.75 represents good to excellent association [33]. Allstatistical analyses were performed using SAS v. 9.1.3(SAS Institute Inc., Cary, NC, 2003).AbbreviationsCA1: First clinical assessment; CA2: Second clinical assessment; CI:Confidence interval; ICC: Intraclass correlation coefficient; InterRAI: interRAICommunity Health Assessment (earlier version); MDS-HC: Minimum Data Setfor Home Care; NHIS: National Health Interview Survey; PA: Physical activity;PPFV: Project to Prevent Falls in Veterans; SD: Standard deviation; SGS:Specialized Geriatric Services; VCA: Veterans’ Comprehensive Assessment;Zero-Mod: Zero modifiable risk factors for falling.AcknowledgementsFunding provided by the Ontario Neurotrauma Foundation (Gill &Speechley) and Veterans Affairs Canada-Health Canada Falls PreventionInitiative (Speechley, Project # 6793-06 2000/16009). Guangyong Zou is arecipient of the Early Researcher Award, Ontario Ministry of Research andInnovation, Canada. The authors would like to sincerely thank the CanadianCentre for Activity and Aging, Shannon Belfry, Krista Bray Jenkyn and SueMuir for their contributions to the Project to Prevent Falls in Veterans. Theauthors would also like to thank the referees of this manuscript whoprovided extremely helpful comments. The authors gratefully acknowledgethe individuals who participated in this study.Author details1National Alzheimer’s Coordinating Center, Department of Epidemiology,University of Washington, Seattle, WA, USA. 2School of Health and ExerciseSciences, University of British Columbia, Kelowna, BC, Canada. 3Departmentof Epidemiology and Biostatistics, Schulich School of Medicine and Dentistry,University of Western Ontario, London, ON, Canada. 4Robarts Clinical Trials ofRobarts Research Institute, Schulich School of Medicine and Dentistry,University of Western Ontario, London, ON, Canada.Authors’ contributionsDPG conceived of the study, participated in the design of the study,performed the statistical analysis and drafted the manuscript. GRJparticipated in the design of the study and helped to draft the manuscript.GZ helped with the statistical analysis and helped to draft the manuscript.MS conceived the study, participated in the design and coordination of thestudy and helped to draft the manuscript. All authors read and approvedthe final manuscript.Competing interestsThe authors (DPG, GRJ, GZ, MS) declare that they have no competinginterests.Received: 12 September 2011 Accepted: 28 February 2012Published: 28 February 2012References1. Shi L: Health Services Research Methods. 2 edition. Clifton Park, NY:Thomson/Delmar Learning; 2008.2. Paterson DH, Jones GR, Rice CL: Ageing and physical activity: evidence todevelop exercise recommendations for older adults. Appl Physiol NutrMetab 2007, 32(Suppl 2E):S69-S108.3. Caspersen CJ, Powell KE, Christenson GM: Physical activity, exercise, andphysical fitness: definitions and distinctions for health-related research.Public Health Rep 1985, 100:126-131.4. Warren JM, Ekelund U, Besson H, Mezzani A, Geladas N, Vanhees L:Assessment of physical activity-a review of methodologies withreference to epidemiological research: a report of the exercisephysiology section of the European Association of CardiovascularPrevention and Rehabilitation. Eur J Cardiovasc Prev Rehabil 2010,17:127-139.5. Ainsworth BE: How do I measure physical activity in my patients?Questionnaires and objective methods. Br J Sports Med 2009, 43:6-9.6. Lagerros YT, Lagiou P: Assessment of physical activity and energyexpenditure in epidemiological research of chronic diseases. Eur JEpidemiol 2007, 22:353-362.7. Westerterp KR: Assessment of physical activity: a critical appraisal. Euro JAppl Physiol 2009, 105:823-828.8. Forsen L, Loland NW, Vuillemin A, Chinapaw MJ, van Poppel MN,Mokkink LB, van Mechelen W, Terwee CB: Self-administered physicalactivity questionnaires for the elderly: a systematic review ofmeasurement properties. Sports Med 2010, 40:601-623.9. Davis HS, MacPherson K, Merry HR, Wentzel C, Rockwood K: Reliability andvalidity of questions about exercise in the Canadian Study of Health andAging. Int Psychogeriatr 2001, 13(Suppl 1):177-182.10. Nadalin V, Bentvelsen K, Kreiger N: Reliability of self-reports: data from theCanadian Multi-Centre Osteoporosis Study (CaMos). Chronic Dis Can 2004,25:28-31.11. Rose SB, Elley CR, Lawton BA, Dowell AC: A single question reliablyidentifies physically inactive women in primary care. N Z Med J 2008,121:1-7.12. Weiss TW, Slater CH, Green LW, Kennedy VC, Albright DL, Wun CC: Thevalidity of single-item, self-assessment questions as measures of adultphysical activity. J Clin Epidemiol 1990, 43:1123-1129.13. DeSalvo KB, Fisher WP, Tran K, Bloser N, Merrill W, Peabody J: Assessingmeasurement properties of two single-item general health measures.Qual Life Res 2006, 15:191-201.14. Eriksson I, Unden AL, Elofsson S: Self-rated health. Comparisons betweenthree different measures. Results from a population study. Int J Epidemiol2001, 30:326-333.15. Gill DP, Jones GR, Zou GY, Speechley M: The Phone-FITT: a brief physicalactivity interview for older adults. J Aging Phys Act 2008, 16:292-315.16. Moore DS, Ellis R, Allen PD, Cherry KE, Monroe PA, O’Neil CE, Wood RH:Construct validation of physical activity surveys in culturally diverseolder adults: a comparison of four commonly used questionnaires. Res QExerc Sport 2008, 79:42-50.17. Rikli RE: Reliability, validity, and methodological issues in assessingphysical activity in older adults. Res Q Exerc Sport 2000, 71:S89-S96.18. Cyarto EV, Marshall AL, Dickinson RK, Brown WJ: Measurement propertiesof the CHAMPS physical activity questionnaire in a sample of olderAustralians. J Sci Med Sport 2006, 9:319-326.19. Harada ND, Chiu V, King AC, Stewart AL: An evaluation of three self-reportphysical activity instruments for older adults. Med Sci Sports Exerc 2001,33:962-970.20. Stewart AL, Mills KM, King AC, Haskell WL, Gillis D, Ritter PL: CHAMPSphysical activity questionnaire for older adults: outcomes forinterventions. Med Sci Sports Exerc 2001, 33:1126-1141.Gill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 9 of 1021. Schuler PB, Richardson MT, Ochoa P, Wang MQ: Accuracy andrepeatability of the Yale Physical Activity Survey in assessing physicalactivity of older adults. Percept Mot Skills 2001, 93:163-177.22. Nelson ME, Rejeski WJ, Blair SN, Duncan PW, Judge JO, King AC, Macera CA,Castaneda-Sceppa C: Physical activity and public health in older adults:recommendation from the American College of Sports Medicine and theAmerican Heart Association. Med Sci Sports Exerc 2007, 39:1435-1445.23. Tremblay MS, Warburton DE, Janssen I, Paterson DH, Latimer AE, Rhodes RE,Kho ME, Hicks A, Leblanc AG, Zehr L, Murumets K, Duggan M: NewCanadian physical activity guidelines. Appl Physiol Nutr Metab 2011,36:36-58.24. Shephard RJ: Limits to the measurement of habitual physical activity byquestionnaires. Br J Sports Med 2003, 37:197-206.25. DiPietro L: The epidemiology of physical activity and physical function inolder people. Med Sci Sports Exerc 1996, 28:596-600.26. interRAI Community Health Assessment. [http://www.interrai.org].27. Koepsell TD, Weiss NS: Epidemiologic Methods: Studying the Occurrence ofIllness New York: Oxford University Press; 2003.28. Gill DP, Zou GY, Jones GR, Speechley M: Injurious falls are associated withlower household but higher recreational physical activities incommunity-dwelling older male veterans. Gerontology 2008, 54:106-115.29. Speechley M, Belfry S, Borrie MJ, Bray Jenkyn K, Crilly R, Gill DP, McLean S,Stolee P, Vandervoort AA, Jones GR: Risk factors for falling amongcommunity-dwelling veterans and their caregivers. Can J Aging 2005,24:261-274.30. National Health Interview Survey. [http://www.cdc.gov/nchs/nhis.htm].31. Landi F, Tua E, Onder G, Carrara B, Sgadari A, Rinaldi C, Gambassi G,Lattanzio F, Bernabei R: Minimum data set for home care: a validinstrument to assess frail older people living in the community. MedCare 2000, 38:1184-1190.32. Morris JN, Fries BE, Steel K, Ikegami N, Bernabei R, Carpenter GI, Gilgen R,Hirdes JP, Topinkova E: Comprehensive clinical assessment in communitysetting: applicability of the MDS-HC. J Am Geriatr Soc 1997, 45:1017-1024.33. Portney LG, Watkins MP: Foundations of Clinical Research: Applications toPractice Upper Saddle River, NJ: Prentice Hall, Inc; 2000.34. Fleiss JL, Cohen J: The equivalence of weighted kappa and the intraclasscorrelation coefficient as measures of reliability. Educ Psychol Meas 1973,33:613-619.35. Landis JR, Koch GG: The measurement of observer agreement forcategorical data. Biometrics 1977, 33:159-174.36. Fisher RA: Frequency distribution of the values of the correlationcoefficient in samples of an infinitely large population. Biometrika 1915,10:507-521.Pre-publication historyThe pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2288/12/20/prepubdoi:10.1186/1471-2288-12-20Cite this article as: Gill et al.: Using a single question to assess physicalactivity in older adults: a reliability and validity study. BMC MedicalResearch Methodology 2012 12:20.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitGill et al. BMC Medical Research Methodology 2012, 12:20http://www.biomedcentral.com/1471-2288/12/20Page 10 of 10


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items