Filetype PDF | Posted on 20 Aug 2022 | 4 years ago
Authentication
digital resources
281x Filetype PDF File size 0.98 MB Source: academic.oup.com
Journal of Gerontology: BIOLOGICAL SCIENCES Copyright 1998 by The Gerontological Society of America
1998. Vol. 53A, No. 5, B362-B368
Comparison of Cross-Sectional and Longitudinal
Designs in the Study of Aging
of Upper Extremity Performance
12 12 12 1
Johanne Desrosiers, Rejean Hebert, Gina Bravo, and Annie Rochette
Downloaded from https://academic.oup.com/biomedgerontology/article/53A/5/B362/588231 by guest on 19 August 2022
'Centre de Recherche en G6rontologie et Geriatrie, Sherbrooke Geriatric University Institute, Sherbrooke, Qu6bec, Canada.
Taculte" de Me*decine, University de Sherbrooke, Sherbrooke, Quebec, Canada.
The purpose of the study was to compare two research designs, namely the cross-sectional design and the longitudi-
nal design, in the context of upper extremity performance and age-related changes. Upper extremity performance of
360 randomly recruited, healthy, community-dwelling elderly persons was evaluated with reliable and valid sensori-
motor tests. Three years later, survivors (n = 264) were reevaluated with the same tests. In many tests, cross-sec-
tional and longitudinal designs were comparable for estimating the changes in upper extremity performance with
age. However, in some tests, the decline with age using a cross-sectional design was underestimated. The upper
extremity performance decline observed with the longitudinal design was larger than the decline predicted with the
cross-sectional design. The withdrawal and survivor biases related to the longitudinal design and the cohort bias
associated with the cross-sectional design may, in part, explain these results.
N the study of age-related changes, two major research used by clinicians to compare the performance of their
I designs may be used: the cross-sectional design and the patients to that of a normal population. These data are
longitudinal design. Both designs have particular advan- important in gerontology because it is important to differ-
tages and disadvantages. Even though the cross-sectional entiate between the difficulties attributable to normal aging
design is less expensive and frequently used in the devel- and those attributable to pathologic aging.
opment of normative data, this design may have an impor- The objective of the present study was to compare cross-
tant limitation in the study of aging. This limitation is the sectional and longitudinal designs in the context of upper
cohort bias in that people born at the beginning of the cen- extremity performance changes with age. The method-
tury have not experienced the same events nor under the ologic relevance of the study is based on the appropriate-
same conditions as younger people, which may influence ness of the cross-sectional design for studying age-related
their performance. changes. The clinical relevance of the study is based on the
In addition to being longer and more expensive, the durability of the normative data developed with a cross-
study of aging with a longitudinal design presents other sectional design. If the performance decline predicted by
possible biases related to withdrawal (refusals) and to sur- the cross-sectional design is equivalent to the decline ob-
vival (deaths or ineligibility factors). The bias associated served with the longitudinal design, normative data may be
with withdrawal was studied by Hubert et al. (1) with 500 applied to successive cohorts. If not, they should be regu-
subjects aged 75 years and older, followed on a 3-year larly revised.
period. Men had more tendency to drop out, but no health-
related variables were associated with withdrawal from the METHODS
study. Mihelic and Crimmins (2) found that withdrawal
(nonresponse) was associated, among other variables, with Subjects
age (older people), living situation (alone), and functional During 1992-1993, a random sample of 360 subjects (179
impairments (more), but not with gender. The survival bias women and 181 men), aged 60 and older (mean, 73.9; SD,
was more serious because people who died during a study 8.0), and living at home was drawn from the electoral list of
were those whose health was more affected (3-5). the city of Sherbrooke, Quebec, Canada. The purpose of this
Upper extremity performance is an important prerequi- cross-sectional study was to develop reference values for
site to functional independence in older people (6-9). With many upper sensorimotor parameters (10). The eligibility
the aging of the population, more and more people may criteria were lucidity (clinical judgment), independence in
have a decline in upper extremity performance secondary activities of daily living (eating, washing, dressing, groom-
to age-related sensorimotor deficits. Therefore, sensorimo- ing, and toiletting), and absence of upper extremity impair-
tor parameters related to good upper extremity perfor- ments (neurologic or orthopedic). The participation rate in
mance, such as gross and fine dexterity, motor coordina- this study was 78% and statistical analyses revealed no dif-
tion, global performance, grip strength and sensibilities, are ference between those who refused to participate and those
often measured in clinical settings to monitor their evolu- who accepted in terms of age, gender, height, weight, domi-
tion. Reference values or normative data are frequently nance, self-perceived health, and current activity level (11).
B362
COMPARISON OF CROSS-SECTIONAL AND LONGITUDINAL DESIGNS B363
Three years later (1995-1996), the same subjects were Statistical Analyses
recontacted to replicate the study. The same eligibility crite- To simplify the results presentation, the terms time 1 and
ria were applied and subjects who no longer satisfied these time 2 will be used respectively for the first evaluation in
criteria were excluded. As in the cross-sectional study, peo- 1992-1993 and the second in 1995-1996. t test and chi-
ple who refused to participate, although eligible, were square analyses were used to compare, at time 1, personal
asked to reply to a short telephone questionnaire in order to characteristics and upper extremity performance of subjects
estimate the refusal bias. who did not participate in the second measurement with
those who participated in both measurements. The same sta-
Procedure tistical tests were used to verify if those who refused to par-
The same evaluation procedure was followed for the two ticipate were comparable in personal characteristics, at time Downloaded from https://academic.oup.com/biomedgerontology/article/53A/5/B362/588231 by guest on 19 August 2022
measurement periods. To avoid an information bias in the 2, to those who agreed. Paired t tests were used to verify if
second measurement, data from the first assessment were the changes between time 1 and time 2 were significant.
not available to the examiner. For both measurements, each The main analysis of this study focused on comparing
subject was evaluated once at the Upper Limb Function upper extremity changes predicted with the cross-sectional
Measurement Laboratory at the Centre de Recherche en design (predicted difference: PREDDIF) to changes ob-
Ge'rontologie et G6riatrie or at their home under the same served by the longitudinal design (observed difference:
conditions, but not by the same examiner who took the first OBSDIF). The observed difference is defined by the ob-
measurement. The duration of each measurement varied served score at time 1 minus the observed score at time 2.
between VA and 2 hours, depending on the subject, includ- The predicted difference is calculated by subtracting the
ing a rest. This study was submitted to and accepted by the predicted score at time 2 from the score at time 1. The pre-
Research Ethics Committee of the Sherbrooke Geriatric dicted score at time 2 was estimated using simple linear
University Institute. regression analyses developed with time 1 data (cross-sec-
Anthropometric data were first collected, followed by a tional design), where age was the independent variable and
structured interview in order to quantify personal characteris- the upper extremity test score was the dependent variable.
tics potentially related to upper extremity performance: age, In some tests, a high score indicates a low performance
living situation (living alone, living with somebody else, liv- whereas in others, the scoring system is reversed. Conse-
ing in a senior's residence), self-perceived health status, and quently, in order to facilitate the data interpretation, the pre-
activity level. Regarding self-perceived health status, the dicted and observed differences were standardized so that
subject was asked: "Compared to other people your age, how positive numbers represent a decline.
would you describe your present health? Excellent, good, For each upper extremity test, a t score was attributed to
fair, or poor." Activity level was evaluated using questions each subject. These t scores were defined by
regarding the frequency of physical activities. Based on this (OBSDIF-PREDDIF)
information, the evaluator estimated the level of activity on a
global scale: very active, active, slightly active, or sedentary.
Subsequently, upper extremity tests were administered. where Spred is the standard deviation of the predicted score
model. A t test was then used to check if, on average, these
Measurement Instruments t scores were null. If the longitudinal and cross-sectional
Many tests were chosen in order to reflect upper extrem- designs described the aging effect in the same way, then the
ity function. These tests are all reliable and valid. Gross observed difference from the longitudinal design and the
manual dexterity was measured with the Box and Block predicted difference from the cross-sectional design should
Test (12-14) whereas fine manual dexterity was measured be equivalent. In order to confirm the null hypothesis that
with the Purdue Pegboard (15-17). Global upper extremity no difference was observed between the observed upper
performance was estimated with the TEMPA (Test Evaluant extremity performance decline (longitudinal design) and
la performance des Membres superieurs des Personnes the predicted decline (cross-sectional design), the differ-
Ag6es) (18-20). ences should be near 0. To take into account the high num-
Upper extremity motor coordination was estimated with ber of statistical analyses done, the level of p < .01 was
the Finger-Nose Test (21-23). Grip strength was measured retained.
with two apparatuses: the Jamar dynamometer (11,24) and
the Martin vigorimeter (25,26). The Jamar dynamometer RESULTS
was set at the second position and the large bulb of the Of the original 360 subjects in the cross-sectional study,
Martin vigorimeter was selected for all subjects. 264 (128 women and 136 men) with a mean age of 75.0
Tactile recognition was estimated with the Pick-Up Test (SD = 7.4) participated in the second measurement. There-
(27), modified by Dellon (28,29). Static and moving two- fore, 96 subjects were not reevaluated: 26 had died, 15
point discriminations were measured on the palmar face of were not located, 29 refused, and 26 were no longer eligible
the distal phalanx of the index and little finger of both because of the development of impairment.
hands, using the Mackinnon-Dellon disk-criminator (30). When characteristics at time 1 of the 96 drop-out subjects
Touch/pressure thresholds (31) were estimated at the distal were compared to those who participated, drop-outs were
phalanx of the index of the dominant hand using the older (p < .001), had a lower body mass index (p = .006), per-
Semmes-Weinstein monofilaments (32). Finally, the same ceived themselves in poorer health (p = .001), were less
filaments were used for the tactile localization test (33). active (p = .001), and were more likely to live in a senior's
B364 DESROSIERS ETAL.
residence (p = .04). From the phone questionnaire at time 2, Table 1 presents the results related to the objective of the
the subjects who refused to participate in the second mea- study. For each test, when significant differences were found
surement (n = 29) were older (p = .01), perceived themselves between the upper extremity performance of the women and
in poorer health (p < .01), and were less active {p < .001) than the men, the results are presented by gender. If not, they
those who participated. In addition, the subjects who partici- were combined. The first two columns of the table report
pated in both measurements showed, at time 1, significantly upper extremity scores obtained at the time 1 measurement
higher performance on all upper extremity tests (p < .01) than (1992-1993) and at the time 2 measurement (1995-1996).
those who participated in only one measurement, with the The third column shows the result of the difference between
exception of the tactile localization of the left index (p = . 14). the two measurements (time 1 minus time 2), which corre-
Downloaded from https://academic.oup.com/biomedgerontology/article/53A/5/B362/588231 by guest on 19 August 2022
Table 1. Comparison of the Observed and Predicted Differences
OBSDIF T PREDDIF OBSDIF -
1 2 1 PRFnniF n
92-93 (T ) 95-96 (T ) (T score Predicted (T' score r
1 2 2 2 p>rd
Score T Score T T score) score T pred. score) S value
Box and Block Test (No. of blocks)
Men and women
Right hand 68.9 (8.3)* 59.6 (7.6)* 9.3 (6.8)* 66.2 (4.7)* 2.7 (7.3)* 0.85 (0.9)* <.001
Left hand 68.2 (8.7) 59.2 (7.5) 9.0 (6.5) 65.6 (4.8) 2.6 (7.6) 0.80 (0.9) <.001
Purdue Pegboard (No. of pins)
Women
Right hand 13.3(1.8) 12.4 (2.3) 0.9(1.9) 12.7(1.1) 0.5(1.5) 0.24(1.2) .03
Left hand 12.5(1.9) 11.7(2.3) 0.9(1.6) 12.0(1.2) 0.6(1.6) 0.18(1.0) .05
Both hands 10.1 (1.9) 9.3(1.9) 0.8(1.5) 9.6(1.1) 0.5(1.6) 0.17(0.9) .04
Total 35.9 (5.0) 33.2 (6.0) 2.6 (3.6) 34.3 (3.4) 1.6(3.9) 0.24(1.0) .01
Assembly 27.5 (6.3) 25.5 (6.5) 2.0 (5.0) 25.5 (3.5) 2.0 (5.5) -0.007 (0.9) .93
Men
Right hand 11.8(1.9) 10.8(2.1) 0.9(1.7) 11.1(1.1) 0.7(1.8) 0.16(1.0) .07
Left hand 11.5(2.0) 10.5 (2.2) 1.0(1.7) 10.8(1.3) 0.7(1.7) 0.17(1.1) .08
Both hands 9.0(1.9) 8.2 (2.0) 0.8(1.4) 8.4(1.1) 0.7(1.7) 0.11 (1.1) .27
Total 32.3 (5.3) 29.5 (5.7) 2.8(3.1) 30.3 (3.5) 2.0 (4.5) 0.17(1.1) .07
Assembly 24.4 (6.0) 22.0 (6.0) 2.3 (4.0) 22.2 (3.8) 2.1(5.1) 0.03 (0.9) .73
TEMPA (log. sec.)
Pick up and move ajar
Women
Right hand 0.5 (0.2) 0.5 (0.2) 0.03 (0.2) 0.5(0.1) 0.03 (0.2) 0.03 (0.9) .75
Left hand 0.5 (0.2) 0.5 (0.2) 0.02 (0.2) 0.5(0.1) 0.03 (0.2) -0.02 (0.9) .79
Men
Right hand 0.4 (0.2) 0.4 (0.2) 0.05 (0.3) 0.4(0.1) 0.04 (0.2) 0.05 (0.8) .49
Left hand 0.4 (0.2) 0.4 (0.2) 0.04 (0.3) 0.4 (0.4) 0.03 (0.2) 0.06 (0.8) .38
Open ajar and take a spoonful of coffee
Women and men 2.2 (0.2) 2.3 (0.2) 0.11(0.2) 2.2(0.1) 0.06 (0.2) 0.20(1.1) .004
Pick up a pitcher and pour water into a glass
Women and men
Right hand 2.1 (0.2) 2.1 (0.2) 0.04 (0.2) 2.1 (0.1) 0.05 (0.2) -0.008(1.0) .89
Left hand 2.1 (0.2) 2.1 (0.2) 0.03 (0.2) 2.1 (0.1) 0.05 (0.2) -0.13(0.9) .02
Unlock a lock and open a pill container
Women and men 2.4 (0.2) 2.6 (0.2) 0.17(0.2) 2.4 (0.2) 0.05 (0.2) 0.56 (0.9) <.001
Tie a scarf around one's neck
Women and men 2.1 (0.3) 2.3 (0.3) 0.13(0.3) 2.2 (0.2) 0.11(0.2) 0.09(1.1) .34
Handle coins
Women
Right hand 2.0 (0.2) 2.2 (0.2) 0.15(0.2) 2.1 (0.1) 0.04 (0.2) 0.60(1.0) <.01
Left hand 2.1(0.2) 2.2 (0.2) 0.09 (0.2) 2.2(0.1) 0.04 (0.2) 0.30(1.1) .002
Men
Right hand 2.1 (0.2) 2.3 (0.2) 0.14(0.2) 2.1 (0.1) 0.05 (0.2) 0.50(1.3) <.001
Left hand 2.2 (0.2) 2.3 (0.2) 0.11(0.2) 2.2(0.1) 0.04 (0.2) 0.37(1.3) <.001
Pick up and move small objects
Men
Right hand 2.1(0.2) 2.2 (0.2) 0.06 (0.2) 2.1 (0.1) 0.05 (0.2) 0.04 (0.9) .60
Left hand 2.1(0.2) 2.2 (0.2) 0.05 (0.2) 2.1 (0.1) 0.05 (0.2) 0.01 (0.9) .89
Continued on next page
COMPARISON OF CROSS-SECTIONAL AND LONGITUDINAL DESIGNS B365
Table 1. Comparison of the Observed and Predicted Differences (Continued)
OBSDIF T2 PREDDIF OBSDIF -
2 1 PREDDIF
92-93 (T) 95-96 (T ) (T score - Predicted (T score - P
1 2 2 2 gpncd
Score T Score T T score) score T pred. score) value
Finger-Nose Test (No. of movements)
Women
Right upper limb 22.3(4.1) 20.2 (3.8) 2.0(4.1) 21.2(2.1) 1.0(3.5) 0.27(1.1) .004
Left upper limb 21.6(4.1) 19.4(3.7) 2.2 (3.6) 20.6(2.1) 1.0(3.6) 0.33(1.0) <.001
Men Downloaded from https://academic.oup.com/biomedgerontology/article/53A/5/B362/588231 by guest on 19 August 2022
Right upper limb 23.0 (3.9) 22.5 (4.0) 0.5 (3.6) 21.8(2.0) 1.2(3.5) -0.18(1.0) .03
Left upper limb 22.9 (4.2) 21.6(3.8) 1.3(3.3) 21.5(2.2) 1.3(3.7) -0.01 (0.9) .88
Jamar dynamometer (kg)
Women
Right hand 24.2 (4.9) 23.2 (4.9) 1.0(2.8) 22.8 (2.3) 1.4(4.6) -0.08(1.0) .35
Left hand 22.5 (4.7) 22.1 (5.0) 0.4 (2.8) 21.2(2.3) 1.4(4.4) -0.21 (1.0) .02
Men
Right hand 41.9(9.3) 38.9 (9.8) 3.0(4.1) 39.8 (4.9) 2.0 (8.2) 0.12(1.0) .17
Left hand 40.2 (9.4) 38.1 (9.3) 2.1 (4.0) 37.7(5.1) 2.5(8.1) -0.05(1.0) .60
Martin vigorimeter (Kpa)
Women
Right hand 52.2(11.2) 51.0(11.9) 1.2(7.1) 49.7 (4.2) 2.5(10.8) -0.12(1.1) .18
Left hand 50.4(10.9) 48.3(11.3) 2.0 (7.7) 48.1 (4.1) 2.3(10.6) -0.02(1.0) .80
Men
Right hand 81.2(18.1) 75.5(17.6) 5.7 (9.7) 76.9(11.0) 4.4(15.7) 0.09 (0.9) .24
Left hand 80.0(17.9) 74.0(17.7) 6.2(10.8) 75.3(10.3) 4.7(15.5) 0.09 (0.9) .24
Pick-Up Test (log. sec.)
Eyes close-eyes open
Women
Right hand 2.6 (0.5) 2.7 (0.5) 0.13(0.6) 2.7(0.1) 0.06 (0.5) 0.13(0.9)
Left hand 2.6 (0.4) 2.6 (0.5) 0.03 (0.6) 2.6(0.1) 0.06 (0.4) -0.07(1.0) .42
Men
Right hand 2.8 (0.5) 3.1 (0.5) 0.20 (0.5) 3.0 (0.2) 0.11 (0.4) 0.21 (1.0) .02
Left hand 2.8 (0.5) 3.0 (0.5) 0.15(0.5) 2.9 (0.3) 0.11 (0.5) 0.10(1.0) .24
Two-point discrimination (mm)
Static
Women and men
Right index 4.9 (0.9) 4.7 (0.9) -0.12(1.1) 5.0 (0.3) 0.18(0.9) -0.30 (0.9) <.001
Left index 4.9 (0.9) 4.8 (0.9) -0.06(1.1) 5.0 (0.2) 0.14(0.9) -0.21 (1.0) <.001
Women
Right little finger 5.1(1.1) 5.9(1.1) 0.76(1.4) 5.3 (0.3) 0.14(1.0) 0.59(1.0) <.001
Left little finger 5.1 (0.8) 6.3(1.5) 1.23(1.5) 5.3 (0.3) 0.15(0.8) 1.21 (1.7) <.001
Men
Right little finger 5.4(1.1) 6.1 (1.4) 0.59(1.3) 5.6 (0.3) 0.16(0.9) 0.45(1.2) <.001
Left little finger 5.5(1.0) 6.3(1.3) 0.77(1.4) 5.7 (0.2) 0.14(1.0) 0.64(1.3) <.001
Moving
Women and men .02
Right index 4.4(1.0) 4.6(1.0) 0.27(1.1) 4.5 (0.2) 0.13(1.0) 0.13(0.9)
Women
Left index 4.1 (1.0) 4.6(1.0) 0.48(1.1) 4.3 (0.3) 0.18(0.9) 0.31 (1.0) .001
Right little finger 4.5(1.1) 5.3(1.0) 0.86(1.2) 4.6 (0.3) 0.15(1.0) 0.67(1.0) <.001
Left little finger 4.5(1.0) 5.3(1.1) 0.81(1.1) 4.6 (0.4) 0.16(0.9) 0.70(1.1) <.001
Men
Left little finger 4.3(1.0) 4.5(1.1) 0.20(1.2) 4.5 (0.2) 0.19(1.0) 0.01 (1.0) .88
Right little finger 4.8(1.0) 5.4(1.2) 0.57(1.3) 5.0 (0.3) 0.19(1.0) 0.37(1.1) <.001
Left little finger 4.9(1.0) 5.3(1.3) 0.47(1.5) 5.0 (0.2) 0.12(1.0) 0.33(1.2) .002
Touch/pressure threshold
(SW filaments number)
Women 3.42 (0.4) 3.50 (0.3) 0.08 (0.4) 3.46(0.1) 0.04 (0.4) 0.12(0.9) .13
Men 3.56 (0.3) 3.61 (0.3) 0.05 (0.4) 3.62(0.1) 0.06 (0.3) -0.05(1.0) .58
Tactile localization (/12)
Women and men 9.2 (2.5) 8.3(2.1) 0.95 (2.7) 8.8 (0.4) 0.34 (2.5) 0.22 (0.8) <.001
Note, p < .01, cross-sectional design differs significantly from the longitudinal design in estimating the age-associated decline. If the difference is posi-
tive, the cross-sectional design underestimates the decline. If negative, the cross-sectional design overestimates the decline,
•mean (SD).
no reviews yet
Please Login to review.
