| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
a Department of Critical Care Medicine and Surgery, Section of Gerontology and Geriatrics, University of Florence, Italy
b Azienda Ospedaliera Careggi, Florence, Italy
Riccardo Pini, Department of Critical Care Medicine and Surgery, Section of Gerontology and Geriatrics, University of Florence, Via delle Oblate 4, 50141, Florence, Italy E-mail: rpini{at}unifi.it.
Decision Editor: Jay Roberts, PhD
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
MEASUREMENT of stature is of primary importance in the clinical examination of the individual patient, as well as in epidemiological surveys, since it is an essential parameter in assessing body composition. Moreover, stature, alone or in combination with body weight, is used to correct several structural and functional cardiovascular parameters. Therefore, the accuracy of its estimate may influence, in turn, the accurate estimate of these parameters. Although obtaining an accurate measurement of stature is substantially simple in the young adult population, this is not always possible in older people. In fact, older individuals are often unable to stand, since they may be chair- or bed-bound, or may present with spinal deformity and loss in vertebral height (1) that determine a disproportionate loss of height, resulting in an incorrect stature measurement. In this population, it may be more appropriate to estimate stature from measurements of other skeletal segments. In actuality, knee height shows the strongest correlation with stature (2)(3)(4). It should be taken into account that equations predicting height may substantially differ according to the race of the study population (5), although it is unknown whether differences in such equations arise from measurements of knee height in populations of a primarily identical racial group that live in different geographic areas and present only minor ethnic differences.
An age-associated decrement in body height has been reported by both cross-sectional studies, which may be influenced by the birth cohort effect (6), and longitudinal studies (7)(8)(9). The latter confirmed that this phenomenon is the result of an actual decrement in stature along with aging, but the entity of this change was not homogeneous across published series.
The first objective of the present analysis was to determine the applicability and accuracy of formulas previously proposed to derive stature from knee height in an Italian elderly population and verify whether kyphosis of the spine may influence the accuracy of such predictive formulas. The second aim of the analysis was to assess longitudinally, in the same population, the change in stature over a 6-year interval. To address these issues, we carried out a secondary analysis of a database from a two-wave, population-based, epidemiological survey of an older Italian cohort.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
65 years) population living in Dicomano, a small town near Florence (10). In 1995, a second epidemiological survey (Insufficienza Cardiaca negli Anziani Residenti a Dicomano, ICARe Dicomano), whose general design has been detailed elsewhere (11), was carried out in the same population to identify the clinical characteristics of heart failure in elderly individuals. All 864 (373 men and 491 women) home-dwelling residents aged 65 years and older, out of 4749 total residents, were eligible for the study. Of these, 614 subjects (71.1% of the eligible study population) underwent the clinical examination and cardiopulmonary assessment of the study protocol and have been included in the present analysis. All participants provided written informed consent. The reasons for nonparticipation were as follows: 229 (26.5%) subjects refused to participate, 18 (2.1%) had died, two (0.3%) had moved to a different town, and one (0.1%) had been admitted to a nursing home before enrollment. Because two subjects (of 614, or 0.3%) were unable to stand and knee height data were missing in six subjects (of 614, or 1.0%), the present analysis comprises 606 subjects (253 men and 353 women, 70.1% of the eligible population).
Physical examinations, including history and physical findings, were carried out by five physicians, three of whom were fellows, trained in geriatric medicine, specifically the Division of Gerontology and Geriatrics of the University of Florence.
To assess the longitudinal changes in body height, the stature of 258 subjects (114 men and 144 women) enrolled in both the first (1989) and second (1995) phase of the study was analyzed. For all these subjects, the age reported in the present analysis was calculated in 1995.
Anthropometric Measurements
The stature of barefooted subjects standing upright against a wall was obtained using a vertical steel measurement bar, with the subject's head positioned according to the Frankfurt plane (i.e., with the lower margins of the orbital openings and the upper margins of the auditory meatus lying in the same horizontal plane). The distance between the occiput and the wall was also measured with subjects standing in the same position. In this phase, presence of clinically evident kyphosis was also recorded.
The knee height of barefooted subjects was measured in the supine position with ankle, knee, and tibia-tarsus joints flexed at 90°, using a SECA model 209 caliper (Vogel and Halke, Germany), with a fixed bar placed along the lateral tibial condyle and the lateral malleolus, and a moving component placed on the upper face of the thigh. All measurements, rounded to the nearest centimeter, were taken by two physicians only, who were unaware of the 1989 measurements. For each subject, measurements were made by the same physician and they were repeated until two identical results were obtained in a row.
To derive the stature from knee height, the following equations developed by Chumlea and Guo (12) for Caucasian men and women were used:
Men: stature (cm)=59.01+
Women: stature 
=75.00+ 
x knee height 
-
.
Statistical Analysis
Data were analyzed using the SPSS (SPSS, Inc., Chicago, IL) (13) statistical package and are reported as mean ± standard deviation (SD). Mean values were compared using Student's t test or analysis of variance followed by a post hoc test (Tukey's multiple comparison test). Age-associated changes in stature and knee height, as well as the correlation between predicted and measured stature, were analyzed by simple linear regression models. The independence of associations was assessed by stepwise multiple regression models. Comparisons between data collected in the two study phases were performed using the t test for paired observations. A p value <.05 was considered statistically significant.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
85 years), men were taller than women within each group (Table 1 ). Stature was different among the three age groups for both men (F = 18.97, p < .001) and women (F = 23.41, p < .001) and was greatest in the youngest age group. Conversely, in both sexes, stature did not differ between the intermediate and the oldest age group. In general, age- and sex-associated changes in knee height paralleled changes observed for stature, with women exhibiting a less marked age-associated decrease in knee height (Table 1 ).
|
|
|
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
As expected, the linear correlation between stature and age and the comparison of the three age groups showed a progressive decrease in body height with increasing age. However, differences in body height were not significant when the oldest and the intermediate age groups were compared, a result that might derive simply by the fewer number of subjects in the oldest age group. Alternatively, this observation, suggesting a progressive attenuation of height loss during the aging process, may be explained by a "secular effect," consisting of a progressive increase of the average stature of the population across subsequent generations (16). We may hypothesize a stronger influence of secular trend on cohorts born after the First World War (65–74 year-old age group). This hypothesis is supported by published analyses of knee height measurements. Some authors demonstrated that age-associated changes in body height are principally due to structural changes in the spine. As a result, the aging process affects mainly the upper body height, whereas the secular trends in skeletal changes can be detected in lower body segments even at an advanced age (6)(9)(17). In our study, only the youngest group showed a statistically significant difference in mean knee height when compared with the other two groups. Therefore, the wider difference in stature between the youngest group and the other two groups might be ascribed to the secular effect.
Because traditional measures of stature are sometimes difficult to be applied to older adults, several estimates of stature derived from measurement of different body segments have been proposed. Many authors have reported that knee height measurement provides the best estimate of stature (3)(4)(5)(18). Since predictive formulas specifically tested in Italian population samples are not available at present, this study verified the extent of applicability of predictive measurements of stature by the Chumlea and Guo (12) equations in an elderly population. We chose this method for its wide use in the clinical setting and in anthropometric studies, which demonstrated a strong correlation between standing stature and knee height in the recumbent position (2)(19). Comparing measured and derived stature, our study confirmed the applicability of the predicting equation in this older, rural population of central Italy. Indeed, when subjects with clinically evident dorsal kyphosis were excluded from analysis, directly measured and derived statures were comparable. This suggests that the Chumlea and Guo (12) method does provide a more accurate estimate of stature in subjects with spine deformities, in whom measured stature may be erroneously less because of the trunk position. In fact, in our series, the difference between measured and derived stature was correlated with the occiput-to-wall distance, which represents an indirect index of dorsal kyphosis. Accordingly, the difference between measured and estimated stature in our series was, although not significantly, slightly less in women than in men, who exhibited a greater occiput-to-wall distance. Moreover, the use of this predictive method may be considered mandatory in subjects unable to stand or in bedridden subjects.
Several studies demonstrated a decrease in stature with aging. Longitudinal studies, such as the present one, provide the most accurate information on the real impact of aging on statural changes, while results from cross-sectional studies may be influenced by the secular effect, consisting of an increase in stature in newly born generations. However, the rate of the age-associated decline in stature differs with study subject selection (7)(8)(17)(20)(21). In our study, between 1989 and 1995, Dicomano's elderly population exhibited a statural decrease rate of approximately 1 cm every 6 years for men and 2 cm every 6 years for women, corresponding to an annual decline rate of 0.17 and 0.33 cm per year for men and women, respectively. In spite of this observation, inclusion of age as a correction term did not improve the accuracy of the Chumlea and Guo (12) equation, either in their original work or in our study. Indeed, the R2 of the regression predicting stature from knee height in our series was not improved when age was introduced in the model. However, overall, our data are comparable to those obtained in the European SENECA study (21) and in the study conducted by Cline and colleagues (7), although Chumlea and colleagues (8) reported a higher decrease rate without significant sex-related differences.
Some relevant limitations of the present study have to be acknowledged. In the cross-sectional analysis, we could analyze the data from only 606 subjects, who represented 70.1% of the entire eligible population. Subjects who did not participate in the study (29.9%) may pose a selection bias to data analysis, since these may include a substantial proportion of persons unable to reach the ambulatory setting because of physical limitations deriving from musculoskeletal disorders and, therefore, with particular difficulties in measurement of stature. However, the mean age and the overall health status, as assessed by information collected from family physicians, were similar for participants and nonparticipants (11). The average difference in stature from 1989 to 1995 for the entire population was -1.7 ± 3.0 cm. The large standard deviation observed, which is substantially greater than the mean value for both the entire population and the two sexes analyzed separately, implies a large variability in differences in body height between 1989 and 1995. Such a large variability originates, in part, by the fact that, over the 6-year interval, 30 of 258 (11.6%) subjects exhibited an increase, which was limited to 1 cm or less in 16 of 30 subjects, rather than a decrease in their stature. Beyond the obvious possibility of errors in measurements that may occur in an epidemiological survey collecting a great quantity of data, we may hypothesize that increases in stature originated from the inclusion of subjects who, during the 1989 study, were suffering from transient musculoskeletal disorders, such as backache. These disorders would imply some degree of dorsal kyphosis and their transient nature might have caused an increase in stature measured at the 1995 follow-up. In contrast, acute events, such as vertebral fractures leading to sudden and abnormally large reductions in body height, and the development of dorsal kyphosis cannot be excluded as further reasons for the large variability in the statural change that was observed longitudinally in our study population. Unfortunately, in 1989, we did not record the presence of dorsal kyphosis, nor did we measure the occiput-to-wall distance or the knee height. Thus, we were unable to detect the origin of the statural changes or to analyze longitudinally the possible age-associated changes in the relationship between standing stature and knee height. Moreover, the measurements were rounded to the nearest centimeter, an approximation that may be inadequate to assess changes in stature over a 6-year interval. This method was chosen essentially because our study was primarily aimed at determining the prevalence of congestive heart failure in the general older population rather than at carrying out extremely accurate anthropometric measurements. A further study limitation was the lack of a test of the reliability of anthropometric measurements that were made, since repetition of measurements by the same physician—although well trained—does not guarantee their precision.
In spite of these limitations, we believe that our study demonstrated that methods to predict stature from measurement of knee height (12) can be correctly and usefully applied to an elderly population of a rural area of central Italy. Indeed, measurement of occiput-to-wall distance and of knee height should be routinely included among the other classic anthropometric measurements, since this might represent a method to analyze the origin of otherwise unexplainable statural increases that may be encountered in a longitudinal perspective. This is particularly true since, as we demonstrated in our study, derived measurements of body height may be more accurate in subjects with a clinically evident kyphosis. Finally, in keeping with previous reports, the present study confirmed, both cross-sectionally and longitudinally, that a significant decrease in stature occurs with aging and that such statural reduction appears slightly more pronounced in women than in men.
|
Acknowledgments |
|---|
Received November 30, 1999
Accepted May 9, 2000
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
