Aging Clin. Exp. Res. 7: 392-397, 1995

A physical fitness model of older adults H. Nagasaki, H. Itoh, and T. Furuna Department of Kinesiology, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan

ABSTRACT. Physical fitness for young people is viewed as a multidimensional construct, in that it consists of specific components such as strength, mobility, balance, flexibility, and stamina. This study examined whether this structure underlying physical fitness is also relevant to older adults. A 10-item performance test, which was assumed to assess six components of physical fitness, was ad­ministered to 69 healthy volunteers ranging in age from 61 to 83 years. A covariance structure model was applied to the test data: the second-or­der factor was Physical Fitness, and the first-order factors were Strength, Walking, Balance, Flexi­bility, Stamina, and Manual Speed which were assumed to be measured based on the ten ob­served variables. Goodness-of-fit index (GFI) of the model was acceptable (GFI=0.93). While four factors relating to basic motor performances (Strength, Walking, Balance, and Manual Speed) had loadings more than 0.62 to Physical Fitness, Flexibility and Stamina had less than 0.35. It was suggested for elderly people that strength, mo­bility, balance, and speed components of physical fitness were highly correlated and explainable by a single factor, while flexibility and cardiorespi­ratory endurance were to be measured by use of specific measures. (Aging Clin. Exp. Res. 7: 392-397, 1995)

INTRODUCTION

The concept of physical fitness has recently tend­ed to be oriented toward health-related physical fit­ness, shifting from traditional performance-related physical fitness (1, 2). Health-related physical fit­ness is particularly relevant to older persons, as the purpose of the assessment is not to find and develop a specific motor ability, but instead to predict the lev­els of physical functioning and health status in their daily life. Thus, physical fitness of the elderly may be defined as a set of attributes of functional capacity

Key words: Elderly, motor ability, physical fitness, performance test.

that are related to the ability to perform physical activity in daily life. A variety of performance-based measures has been used for assessing health-related physical fitness of older persons (3-7). However, several items are usually selected ad hoc from among the existing tests for young people, and adminis­tered to older adults primarily considering their safe­ty and ease of testing. The conceptual basis that underlies these measurement tools is still unclear: what are we measuring by using performance-based measures?

For younger people, Fleishman's work on motor ability (8, 9) might be one of the most systematic studies on the structure of physical performance. The term "motor ability" usually refers to a construct that is assumed to underlie performance in a number of motor tasks or activities. Fleishman (8, 9) admin­istered a great number of motor performance tests to young men mainly from the armed forces, and ex­tracted several factors from the test scores, each of which was defined as a motor ability. In the latest ver­sion of the ability list, Reishman (8) gave a total of 52 motor and intellectual abilities. Along with psy­chomotor and skill-oriented control abilities, 9 fun­damental motor abilities among them are referred to as underlying dimensions of physical fitness or proficiency (9): Strength (static, explosive, dynamiC, and trunk), Flexibility (extent and dynamic), Gross Body Coordination, Gross Body Equilibrium, and Stamina.

Reishman's physical fitness model may be schemat­ically illustrated as in Figure 1. At the top of the hi­erarchical structure is a single common factor (phys­ical Fitness) that may be relevant to all domains of physical fitness. Physical Fitness has a number of component factors underneath it. At the bottom of the hierarchy are various tasks, the performance of which may be affected by related physical fitness compo­nent(s).

Thus, the concept of physical fitness was extend­ed by Fleishman to include not only cardiorespirato-

Correspondence: Hiroshi Nagasaki, Ph.D., 35-2 Sakaecho, Itabashi-ku, Tokyo-173, Japan.

Received March 10,1995; accepted in revised form July 31,1995.

Aging Clin. Exp. Res., Vol. 7, No.5 392

H. Nagasaki, H. Itoh, and T. Furuna

Figure 1 - Structure of physical fitness (after Fleishman, 8).

ry endurance (Stamina), but also fundamental motor abilities involving gross body movements. It is noted that physical performance measures popular in the geriatric literature are also chosen according to the components of physical fitness as defined above; e.g., one-leg standing duration for Gross Body Equi­librium, and walking speed for Gross Body Coordi­nation.

As indicated by Fleishman's method for defining motor ability, i.e., factor analysis, the physical fitness components (Fig. 1) are assumed to be specific. For example, Gross Body Equilibrium, which may underlie groups of motor tasks such as standing on a ladder or walking on ice, is hardly dependent upon Static Strength. The concept of physical fitness is thus a multidimensional construct including a range of spe­cific motor abilities.

Reishman's motor ability model, in other words, as­sumed that "general motor ability" contributes very lit­tle to each specific component. If there were a unifying ability that could serve as a powerful determiner of all motor tasks, then general motor ability as such would demand that the correlations between motor tasks be quite substantial. However, this was clearly not the case for younger adults (10). In the physical fitness model illustrated in Figure 1, the effect of the hy­pothesized general factor ("Physical Fitness") on its components is also considered to be virtually in­significant.

Is the concept of physical fitness developed for youth also relevant to older adults? If so, a compre­hensive test battery including at least a measure for each specific component would be essential to the assessment of fitness in the elderly (6). Even given that such a test were developed, a test consisting of many items would have practical difficulties in its ap­plication to the community-based elderly popula­tion. It has frequently been observed, however, that

393 Aging Clin. Exp. Res., Vol. 7, No.5

considerable correlations exist among performance measures for Strength, Gross Body Coordination, and Equilibrium in older adults (6, 7). Our test ad­ministered to community-dwelling elderly people showed that the correlations among performances for strength, speeded movements, mobility, and balance were highly significant (11). These results suggest that some of the fitness components de­fined for young adults are not specific in older per­sons, but instead are determined by a common fac­tor.

The aim of this study was to examine the specificity hypothesis of the physical fitness components in old­er adults. A short physical performance battery was designed so that each component of Fleishman's physical fitness (9) could be assessed at least by one measure, and administered to healthy elderly volun­teers. The result of the test was analyzed by applying a covariance structure model constructed according to Fleishman's model of physical fitness as illustrated in Figure 1. The specificity of each component was determined by examining its factor loading on the hy­pothesized general factor, Physical Fitness. The validity of the model and its practical consequences suggest­ed by the present study will be described in our sub­sequent paper for a representative study population of the elderly (11).

SUBJECTS AND METHODS Participants

The subjects of the present study were 69 healthy volunteers (35 males and 34 females) from a local community, ranging in age from 61 to 83 years. The subjects were engaged in temporary jobs in the community, and paid for participating in this study.

Test items

A physical performance battery (12) consisted of items that have been frequently used in the geriatric literature for assessing physical functioning of older adults: grip and leg strength; trunk flexion; walking at preferred and maximum speed; single-leg bal­ance with eyes open and closed; a step test; finger tapping; and a pegboad test. Height and weight were also measured. The grip strength (GRIP) of the preferred hand was measured using a hand-held Smedley-type dynamometer. Leg strength (LEG) was isometric force produced by knee extension at 90 degrees joint position in sitting posture. Maximum trunk flexion (TFLEX) was assessed in an upright posture. In the walking test, the subject first walked a straight walkway of 11 meters on a flat floor at

preferred speed (P-WALK), and then as fast as pos­sible (M-WALK). Towards the middle of the walkway, the speed and steps were measured for 5 meters. The walking test and its reliability was described elsewhere (13). Only speed data were used in this study. Duration of standing while using the pre­ferred leg was measured for a maximum of 60 sec­onds with eyes open (O-BALANCE), and for 30 seconds with eyes closed (C-BALANCE). Physical work capacity (PWC, watts/kg) was estimated by ad­ministering a step test (14). In the finger-tapping test (13, 15), the maximum tapping rate, and constant and variable errors for tapping in time to a 4-Hz metronome were measured. The maximum tapping rate (TAPPING) was used in the present study. The pegboad test (PEG) measured the number of small pegs (3 mm diameter) picked up one at a time and placed into pegboad within a time limit of 30 sec­onds.

These test items were selected so that five di­mensions of Fleishman's physical fitness (Fig. 1), and one manual ability could be assessed at least by one measure: static Muscle Strength by GRIP and LEG, extent Flexibility by TFLEX, Gross Body Co­ordination by P- and M-WALK, Gross Body Equi­librium by 0- and C-BALANCE, Stamina by PWC, and a manual ability ( manual speed) by TAPPING and PEG.

Figure 2 - Covariance structure model for physical fitness of older adults.

A physical fitness model of older adults

Model specification

According to Fleishman's motor ability model (Fig. 1), a structural model of physical fitness was con­structed as illustrated in Figure 2, and applied to the result of the performance test. The model was a sec­ond-order covariance structure model in which Phys­ical Fitness was the second-order latent variable (~). Five fitness components suggested by Fleishman were assumed as first-order latent variables (l1s) with mea­surement errors (1;s): Strength, Walking, Balance, Flexibility, and Stamina. Manual Speed, which was not included in the original model, was also considered in Figure 2 as a component of Physical Fitness, be­cause TAPPING and PEG correlated with muscle strength and others (Table 1). Observed variables consisted of ten test items (ys) with measurement er­rors (es). As illustrated, each first-order factor was assumed to have a loading (11 on the second-order fac­tor, Physical Fitness. Each observed variable (item) was assumed to have loadings (As) on one of the first-order factors (fitness components): grip and leg strength (Yl:GR1P, Y2:LEG) had loadings on Strength; pre­ferred and maximum walking speed (Y3:P-WALK, Y4:M-WALK) on Walking; one-leg standing with eyes opened and closed (Ys:O-BALANCE, Y6:C-BAL­ANCE) on Balance; and tapping and peg speed (Y9:TAPPING, YlO:PEG) on Manual Speed. Flexibility and Stamina factors were assessed by only one mea­sure, i.e., trunk flexion (y7:TFLEX) and physical work capacity (ys:PWCj, respectively. In order not to take the morphological factor (subject's height and weight) into consideration, muscle strengths and walking speeds were corrected by diViding them by each sub­ject's weight and square root of body height (16), respectively.

The factorial structure of physical fitness for older adults was analyzed by use of a CALIS procedure in the SAS system. At first, the physical fitness model in Figure 2 was tested using a correlational matrix of the observed variables. The model was then revised if necessary. Goodness-of-fit index (GFI), adjusted good­ness-of-fit index (AGFI) , root mean square residuals (RMR) , and x2 were used as indices of the model's fit.

RESULTS Parson's simple correlations and partial correlations

with age and gender as control variables are given in Table 1, which shows that the inter-item correlations were remarkable. When age and gender were con­trolled, the correlations were significant at the p<0.05 level, not only within each of the assumed components of physical fitness (e.g., between Yl:GRIP and Y2:LEG), but also between those from different components.

Aging Clin. Exp. Res., Vol. 7, No.5 394

H. Nagasaki, H. Itoh, and T. Furuna

Table 1 - Pearson correlation coefficients between test scores. Upper right: simple correlation: lower left: partial correlation controlling age and gender.

GRIP LEG P-WALK M-WALK O-BALANCE C-BALANCE TFLEX PWC TAPPING PEG ._-_. --_._---- - ----------- ~

GRIP 0.56"" 0.10 0.30*

LEG 0.40"* 0.27* 0.43*'"

P-WALK 0.14 0.30* 0.30'"

M-WALK 0.17 0.29" 0.30*

O-BALANCE 0.37*" 0.18 -0.04 0.16

C-BALANCE 0.02 -0.10 -0.10 0.20

TFLEX 0.08 0.26" 0.27* 0.28*

PWC -0.09 0.03 -0.06 0.08

TAPPING 0.24 0.15 0.19 0.29"

PEG 0.02 0.10 0.10 0.18

p < 0.05. ' p<O.Ol.

For example, walking speed (Y3, Y4), balance (Ys:O­BALANCE), and trunk flexibility (Y7) were all correlated with muscle strength (Yl' Y2)' For young adults, on the contrary, available data indicate that the speed of preferred walking is independent from either one's maximum walking speed, or muscle strength, bal­ance, and trunk flexibility (12), as suggested by the mo­tor ability model. These results suggested that the specificity hypothesis may not necessarily account for the physical fitness components in our data for old­er people. This suggestion was tested by analyzing the physical fitness model that follows.

The standardized maximum likelihood solution of the initial model (Fig. 2) showed GFI=0.90 and AGFI=0.83. The goodness-of-fit was improved by revision, in which two pairs of measurement error co­variance were added to the initial model, i.e., be­tween P-WALK (Y3) and TFLEX (Y7), and between PWC (Ys) and TAPPING (Y9)' The goodness-of-fit of the revised model reached acceptable levels (Fig. 3): GFI=0.93, AGFI=0.86, RMR=0.075, X2=28.7 (df=29). Among these indices, goodness-of-fit index (GFI) and/or GFI adjusted by degree of freedom (AGFI) are most popularly used for the evaluation of covariance structure models. GFI=0.93 in the re­vised model (Fig. 3) indicated that the model ex­plained 93% of the correlation matrix of the sample used.

The factor loadings of first -order factors on the sec­ond-order factor (Physical Fitness) were greater than 0.62 for Strength, Walking, Balance and Manual

395 Aging Clin. Exp. Res., Vol. 7, No.5

---- ------------

0.35* 0.06 -0.15 0.15 0.25* 0.06

0.28* -0.02 om 0.27' 0.18 0.19

0.02 -0.09 0.26* -0.03 0.19 0.14

0.27'" 0.24 0.12 0.24 0.30" 0.26

0.33** 0.10 0.16 0.20 0.12

0.31" 0.06 0.11 0.23 0.22

0.19 0.10 -0.03 0.12 0.03

0.05 0.06 0.15 -0.17 0.03

0.19 0.23 0.17 -0.24 0.36"*

0.02 0.19 -0.03 -0.08 0.36"'"

Speed, but negligible for Flexibility (0.10) and Stami­na (0.35): only 2 and 12% of variance were ex­plained by the hypothesized Physical Fitness for Flex-

Phys;cal Fitness

(-~-\

A 0.71 0.78 0.62 0.100.35 0.65

/;// \\~

0.68 0.83 0.33 0.89 0.73 0.46 1.00 1.00 0.68 0.49

I ~ J \ ~ \ \ ~ t \ [!J ~ IV!] ~4~ ! Y 511 Y 6! [Ij ~ ~J 1 Y10J

t t t t t t t t t t 0.74 0.56 0.94 0.46 0.68 0.89 0

"----- 0.22 ---'

o 0.74 \, )

-0.30

0.87

Figure 3 - Standardized maximum likelihood solution of the physical fitness model in Figure 2. N=69. GFI=O.93, AGFI=O.86, RMR=O.075, X2=28.7 (df=29).

Basic Yotor Abi I ity

~ 0.69 0.59 0 .64

/ \ ~ Strength Walking Balance Manual Speed

o,~~t;f o~ of"K 0.68 0.83 0.36 0 .84 0 .70 0.48 0.70 0.50

I~' ~ J \ ~\ ~~~IT3[!J~~[EJ t t t t t t t t

0.74 0.56 0.93 0.54 0.72 0 .88 0.71 0.87

Rgure 4 - Standardized maximum likelihood solution of the "Ba­s ic Moto r Ability" model. N=69, GFl=O.94, AGFl=O.87, RMR=O.068, X2=17. 98 (df=17).

ibility and Stamina, respectively. The results suggest that these four factors (Strength , Walking, Balance, and Manual Speed) are mainly affected by a single common factor, but that Flexibility and Stamina are specific components of physical fitness.

The model then was simplified in that the four factors (Strength , Walking , Balance , and Manual Speed) were first-order variables, and to be explained by a single common factor at a higher level , i.e. , "Basic Motor Ability". The solution of this model is il­lustrated in Figure 4 . The goodness-of-fit of this mod­el was acceptable: GFI=0 .94, AGFI=0.87 , RMR==0.068, X2=17.98 (df=17). The factor load­ings of first-order factors were greater than 0.59, indicating that more than 35% of variance for these four latent variables were explained by a single ability factor, Basic Motor Ability.

DISCUSSION

The present study evaluated the specificity hy­pothesis of motor ability (8-10) in the physical fitness dimensions of older adults. To this aim, a covari­ance structural analysis was applied to data from a ten­item test of physical performance. The second-order physical fitness model for ten observed variables (Fig. 3) indicated that more than 38% of variance in four factors of physical fitness (Muscle Strength, Walk­ing Speed, Balance, and Manual Speed) was deter-

A physical fitness model of older adults

mined by a single higher-order factor, whereas this fac­tor explained little of the variance in Trunk Rexibility and Stamina components. These results suggest that some of the components of physical fitness defined by Fleishman were not necessarily specific, but rather combined into a common factor. This conclusion is consistent with the fact that Muscle Strength, Walking Speed, Balance, and Manual Speed had consider­able inter-correlations (Table 1). The results may also support the notion of "health-related physical fit­ness" in AAHPERD (1) and Fitness Canada (2) that consists of four speCific components, i.e. , Muscle Strength/Endurance, Flexibility, Cardiorespiratory Endurance , and Body Composition.

Muscle Strength, Walking Speed, Balance, and Manual Speed were then explained by a single factor in the revised model (Fig. 4). This single factor was la­beled as "Basic Motor Ability", because four first-order factors may be related to an ability that is fundamen­tal to physical functioning of older adults. Although Fleishman's factor "Strength" consists of four inde­pendent strength components (static, explosive, dy­namic, and trunk) , correlations may increase with aging between muscle forces produced by different muscles or modes of contraction (17). Manual motor abilities are not included in the Fleishman's model of physical fitness. However, Manual Speed, measured by finger tapping and pegboad tests in the present study, exhibited a considerable loading on the higher­order factor . This is consistent with the results of Greene et al. (6) that showed significant correlations between speed of handskill tasks and Muscle Strength and Walking Speed. The manual tasks in our test were simpler than Fleishman's tasks relating to man­ual control abilities , and as such , the manual speed showed a factor loading on Basic Motor Ability simi­larly to Muscle Strength and Walking Speed.

In a factorial analysis of their performance test for older adults, Greene et a!. (6) found that strength, body mobility/agility, and static balance were separate dimensions. Their method, however, was an ordi­nary exploratory factor analysis which assumed that all factors were uncorrelated, and all observed variables were directly affected by all common factors. The data of Greene et al. (6) , on the other hand, showed Significant correlations among variables for measuring Strength, Mobility, and Balance, thereby suggesting considerable inter-factor correlations. The analysis model, based on a confirmatory factor analysis, should be employed in order to reveal potential structures in multidimensional data (18).

The factor loadings of Strength, Walking, Bal­ance , and Manual Speed on the general factor, i.e., Basic Motor Ability, were not suffiCiently large in the

Aging Clin. Exp. Res ., Vol. 7 , No.5 396

H. Nagasaki, H. Itoh, and T. Furuna

present study to confirm this factor as a common determiner of these components. This may be partly due to the fact that our subjects were at high levels of physical functioning. A model similar to Figure 4 ex­amined in our population-based study of aging (11) showed that the model had a GFI (goodness-of-fit index)=0.996 and loadings more than 0.8 for first-or­der factors. Thus, it is suggested that the specificity hy­pothesis as proposed by Fleishman (9) for physical fit­ness of younger adults may not account for elderly people. A general decline in physical capacity with ag­ing would bring about interrelations between several fitness components, which had been differentiated in individuals through developmental process.

The Basic Motor Ability model (Fig. 4) suggests that a short physical performance battery can be designed with a high internal consistency for assessing physical functioning of older community residents. A scoring method of physical performances of the elderly may be constructed on the basis of the model proposed by the present study. These practical consequences of the Basic Motor Ability model will be described in our sub­sequent paper (11).

In the present study, extent flexibility and cardio­vascular endurance appeared to be specific compo­nents of physical fitness for older adults. Since Trunk Flexion had considerable correlations with Muscle Strength and Walking Speed (Table 1), it may be in­cluded in observed variables for Strength or Walking in the Basic Motor Ability model. The assessment of cardiorespiratory endurance, on the other hand, may be important not only because it was shown to be a specific component of physical fitness of older adults, but also to provide protection against heart dis­ease. Performance measures for evaluating the Stami­na component of physical fitness for older adults in the community should be developed in population-based studies of aging (19).

The present study specified a model to analyze the latent structure of physical fitness measures for old­er adults. Two major limitations of the study must be noted. First, the model was Simplified from the origi­nal model of Fleishman because limited test items were measured. Second, the participants in the pre­sent study were at a higher level of physical fitness than the general elderly population, and the validity of the model should be confirmed in a representative sample from community-dwelling older adults. The lat­ter problem will be considered in our subsequent pa­per (11).

397 Aging Clin. Exp. Res., Vol. 7, No.5

REFERENCES 1. Falls H.B.: Modern concepts of physical fitness. JOPER 51:

25-27, 1980.

2. Canadian standardized test of fitness. Fitness Canada, Ottawa. 1987.

3. Briggs Re.. Gossman M.R. Drews J.E., Shaddeau S.A: Balance performance among noninstitutionalized elderly wom­en. Phys. Therapy 70: 410-415, 1989.

4. Tinetti M.E .. Ginter S.F.: Identifying mobility dysfunctions in elderly patients. Standard neuromuscular examination or di­rect assessment? JAMA 259: 1190-1193, 1988.

5. Carter J.s .. Williams H.G .. Macera e.A: Relationships be­tween physical activity habits and functional neuromuscular ca­pacities in healthy older adults. J. App/. Gerontol. 12: 283-293.1993.

6. Greene L.S .. Williams H.G., Macera e.A: Identifying di­mensions of physical (motor) functional capacity in healthy old­er adults. J. Aging Health 5: 163-178. 1993.

7. Era P.: Sensory. psychomotor and motor functions in men of different ages. Scand. J. Social Med. 39: 9-77. 1990.

8. Fleishman E.A. Quaintance M.K: Taxonomies of human performance. Academic Press. New York, 1984.

9. Fleishman E.A: The structure and measurement of physi­cal fitness. Prentice-Hall. Englewood Cliffs. NJ. 1964.

10. Parker J.F .. Fleishman E.A: Ability factors and component performance measures as predictors of complex tracking be­havior. Psychol. Monograph. 74 (Whole No. 503), 1960.

11. Nagasaki H .. Itoh H .. Furuna T.: The structure underlying physical performance measures for older adults in the com­munity. Aging C/in. Exp. Res. (in press), 1995.

12. Kinugasa T.. Nagasaki H .. ltoh H .. Hashizume K, Furuna T.. Maruyama H.: Aging effect on motor ability for male subjects aged from 18 to 83 years (in Japanese). Jpn. J. Phys. Fitness Sports Med. 43: 343-351.1995.

13. Nagasaki H .• Itoh H., Hashizume K. Furuna T., Maruyama H., Kinugasa T.: Walking patterns and finger festination in the elderly. Exp. Aging Res. (in press), 1995.

14. Furuna T., Nagasaki H., ltoh H., Fujisawa A, Niimi M., Maruyama H .. Kinugasa T.: Effect of aging on the aerobic ca­pacity measured by a step-test (in Japanese). Jpn. J. Geriatr. 31: 40-47, 1994.

15. Nagasaki H., ltoh H., Maruyama H., Hashizume K: Char­acteristic difficulty in rhythmic movement with aging and its re­lation to Parkinson's disease. Exp .. Aging Res. 14: 171-176,1988.

16. Wagenaar Re., Beek W.J.: Hemiplegic gait: A kinematic anal­ysis using walking speed as a basis. J Biomech. 25: 1007-1015,1992.

17. Avlund K, Schroll M., Davidsen M., Lovborg B .. Rantanen T.: Maximal isometric strength and functional ability in daily ac­tivities among 75-year-old men and women. Scand. J. Med. Sci. Sports 4: 32-40, 1994.

18. Long J.S.: Confirmatory factor analysis: A preface to LISREL. Sage Publications, Beverly Hills, 1983.

19. Blair S.N.: Physical activity, physical fitness. and health. Res. Quart. Exer. Sport 64: 365-376, 1993.