Lower Extremity Strength and Power Are Associated With 400-Meter Walk Time in Older Adults: The InCHIANTI Study

Design

We used baseline data from the InCHIANTI study (“Invecchiare in Chianti,” i.e., Aging in the Chianti Area) (25), which were collected during three separate testing sessions: a home-based interview, a medical exam, and a functional performance evaluation. The data collection for this study began in September 1998 and concluded in March 2000. The study protocol was approved by the Italian National Institute of Research and Care on Aging ethics committee.

Participants

In August 1998, 1299 individuals aged 65 years and older were randomly selected from the population registry from the rural town of Greve in Chianti and the suburban town of Bagno a Ripoli, located in the Chianti region of Tuscany, Italy. Individuals were contacted by mail and given a description of the study. Thirty-nine individuals were not eligible for the study due to death or because they moved away from the area. In those who were eligible, the response rate was excellent (91.7%, 1155/1260). Of the initial 1155 participants, 435 participants were excluded from our analyses, resulting in 720 participants with data available for analyses. Participants were excluded for one or more of the following reasons: they (a) were blind (n = 24) or deaf (n = 28); (b) had a cognitive impairment defined as a Mini-Mental State Exam score lower than 21 (n = 172); (c) had Parkinson’s disease (n = 15), a stroke (n = 57), lower extremity amputation (n = 2), paralysis (n = 1), or peripheral neuropathy (n = 2) as determined from both medical documentation and a baseline medical exam; (d) had an interview classified as unreliable, by proxy, or did not provide consent to participate (n = 97); and (e) were missing data for lower extremity strength or power (n = 194). Reasons for missing strength or power data included: (a) severe hip, knee, or back pain (n = 45); (b) range of motion limitation (n = 44); (c) home visit (n = 18); (d) unsafe, dangerous (n = 39); (e) high blood pressure (n = 18); (f) cognitive impairment (n = 13); (g) paralyses/amputation (n = 6); (h) dyspnea (n = 3); (i) angina (n = 1); (j) low heart rate (n = 1); and (k) refusal (n = 6).

In addition to these exclusions, some participants failed to provide complete data for the 400-meter walk (n = 65, see criteria below). The final analytical sample included 655 participants.

Sociodemographic Variables

Participant demographic information included age, gender, height, body mass, body mass index (BMI), education (years completed), depressive symptoms (Center for Epidemiologic Studies Depression scale, CES-D) (26), and cognitive function (Mini-Mental State Exam, MMSE) (27). Comorbid conditions were confirmed for participants who had both past medical records and a current medical exam suggestive of the condition. Comorbidities assessed in this study included cancer, respiratory disorder (diagnosed asthma, chronic bronchitis, and lung emphysema), dyspnea with light or mild exertion, diabetes, cardiovascular disease (diagnosed congestive heart failure, myocardial infarction, and angina), hypertension, peripheral arterial disease, and hip fracture.

400-Meter Walk

Participants were instructed to walk 20 laps of a 20-meter course around 2 cones “at a steady and, if possible, constant pace.” Standardized verbal encouragement was given on each lap, directing participants to maintain their pace, and indicating the number of laps remaining. Exclusion criteria for the test were: heart rate,< 40 bpm or.> 135 bpm; within the previous 3 months, anterior myocardial infarction, cardiac surgical intervention, angina, severe dyspnea or dyspnea at rest, loss of consciousness; systolic blood pressure (BP).> 180 mmHg or diastolic BP.> 100 mmHg, pathologic changes on electrocardiogram; difficulty keeping feet together for 10 seconds; and difficulty in walking 8 meters. Criteria for test termination were: palpitations; chest pain, constriction, feeling of oppression; respiratory difficulty or dyspnea; sensation of fainting, empty head, or postural instability; pain in the lower limbs; vertigo; and muscle fatigue. The 400-meter walk time was measured using an optoelectronic system with two photo-cells connected to a digital chronometer.

Lower Extremity Isometric Strength (LEIS)

Maximum voluntary isometric strength of eight lower extremity muscle groups of the right and left lower extremity (hip flexors, hip extensors, hip abductors, hip adductors, knee flexors, knee extensors, plantarflexors, and dorsiflexors) was measured in kilograms (kg) using a portable dynamometer (Penny and Giles Instrumentation Ltd, Christchurch, Dorset, England), following a standard protocol (28).

In our preliminary analyses, we considered several ways to quantify lower extremity strength. First, the lower extremity isometric strength (LEIS) measurements of the eight muscle groups were reduced to a single measure using a principle component analysis. Separate analyses for the right and left side showed that a single factor captured approximately 81% of the variance for the measures on each side. Because of similar factor loadings, we summed the scores of the eight muscle groups on the right side and left side to create an isometric strength sum score. Since the sum scores of the right and left sides were highly correlated (r = .97), we used the sum score of the right side as a measure of maximal lower extremity isometric strength. The sum scores were normalized by dividing by the participant’s body mass and multiplying by the body mass sample mean. Second, we examined single isometric strength measures at the ankle (ankle plantarflexion) and knee (knee extension) since both these muscle groups are important contributors to gait performance and can be assessed with little difficulty in a clinical environment. Of these three measures of strength, we elected to use the ankle plantarflexion isometric strength as the predictor variable for lower extremity isometric strength (LEIS), and our rationale is explained below.

Lower Extremity Muscle Power (LEMP)

Lower extremity muscle power in watts (W) for the right and left sides was measured using the Nottingham leg extensor power rig, according to the method described by Bassey and Short (29). The maximal LEMP measurements of the right and left lower extremities were highly correlated (r = .94). Therefore the LEMP of the right lower extremity was used to represent a participant’s LEMP. We also conducted analyses using the normalization scheme of Lauretani and colleagues (30) whereby the LEMP score was divided by the participant’s body mass and multiplied by the body mass sample mean. The conclusions reached were no different from those using the nonnormalized measure of LEMP, and we used the nonnormalized measure of LEMP in the article data.

Statistical Analyses

All analyses were performed using the SAS statistical package (SAS Institute Inc., Cary, NC). Univariate descriptive statistics were calculated for the covariates, the 400-meter walk time, and the predictor variables (LEIS, LEMP). Because of a highly skewed distribution, log-transformed values of 400-meter walk time were used in regression analyses. However, the fit of models to untransformed data was also investigated for an apparent threshold.

Examination of the regression analyses using the three different measures of isometric strength showed that the sum score resulted in the highest R² (32.3%), followed by ankle plantarflexion strength (28.6%), and the knee extension strength (23.9%). Regardless of the measure of isometric strength, the conclusions reached about the nature of the relationship between strength and 400-meter walk time remained the same. We used the ankle plantarflexion isometric strength measure because its predictive ability fell between the other two measures and, pragmatically, measuring one muscle group is easier than measuring eight.

Linear multiple regression analyses and piecewise regression were used to characterize the relationships between 400-meter walk time and the predictor variables (LEIS, LEMP), and to try to estimate a threshold in the relationship between the predictors and 400-meter walk time. The knot and confidence intervals (CIs) for the piecewise regression models were estimated using PROC NLIN in SAS and Gauss-Newton estimation. An alternative statistical method used for fitting the relationships of interest was locally weighted regression smoothers (loess technique) (31). This technique resulted in plotted lines for the LEIS or LEMP versus 400-meter walk time relationships that were almost identical to those obtained using multiple linear regression with higher order terms included in the model. We also converted the 400-meter walk time into gait velocity in meters per second and fit these data using the loess technique. This analysis revealed even less evidence for a threshold, that is, a plateau in the relationship between gait speed and LEMP. Since the interpretation and conclusions remained the same in all instances, the loess results are not presented but are available from the author. Body mass index was not entered into the regression analyses as a covariate because both strength and power analyses were adjusted for body mass and height.

We examined interactions between LEIS and LEMP and several covariates that we believed may influence the relationship between strength or power and 400-meter walk time, and that also had a sufficient number of participants with the covariate (approximately > 50). These included gender, respiratory disorder, dyspnea, diabetes, cardiovascular disease, hypertension, and peripheral arterial disease.

Analysis of variance was used to explore for differences in age, LEIS, and LEMP for three groups defined by whether the participant completed, did not complete, or did not attempt the 400-meter walk. Fisher’s protected least significant difference procedure was used to examine for differences between the group means (32). For all aforementioned analyses, p values less than .05 were used to determine statistical significance.

Sociodemographic Variables

Baseline characteristics of the study participants are shown in Table 1. The sample consisted of 720 participants with a mean (± standard deviation [SD]) age of 73.0 ± 6.1 years. Women represented approximately 53% of the sample and were, on average, slightly older than the men. Participants were slightly overweight with a mean BMI of 27.4 ± 3.9 kg/m² and, on average, had fewer than 6 years of formal education. Approximately 28% of the sample was free from chronic medical conditions. The most common comorbidity was hypertension, followed by dyspnea with mild exertion, diabetes, and cardiovascular disease.

The mean 400-meter walk time was comparable to a previous report (33), although the range we observed was greater, probably reflecting the larger age range in our study (65–91 years vs 70–78 years). The muscle power observed in this sample was consistent with data in previous studies that have used the power rig in older adults (19,29). The plots of bivariate relationships between muscle strength and power with log-transformed 400-meter walk time showed evidence of a curvilinear relationship (Figure 1).

Regression Analyses

The summaries from the unadjusted and adjusted regression analyses are shown in Table 2. In the unadjusted models for LEIS, the addition of the quadratic term resulted in an increase of 3.9% in total explained variability; whereas the cubic term increased the R² by less than 0.1%. For unadjusted power analyses, the significant quadratic term explained an additional 3.3% of variability, and the cubic term an additional 0.6%. However, overall, the model containing linear, quadratic, and cubic power terms explained 35.0% of the variability in 400-meter walk time, in contrast to 28.7% for strength. When all linear, quadratic, and cubic terms for strength and power were entered into the model, the total explained variability was 38.3%, indicating that the addition of the linear and quadratic strength terms only increased the explained variability beyond power alone by 3.3%. Inspection of Figure 1 shows that the major difference between the quadratic and cubic models is how the models fit those individuals with the greatest power or strength: the cubic model for LEMP permitting a reduction in the predicted walk time for these individuals. However, these cubic terms explain very little variability in models containing power or strength.

In the adjusted regression analyses, a model containing the covariates (age, gender, body mass, height, years of education, MMSE score, CES-D score) and comorbid conditions accounted for 48.6% of the total variability in log-transformed 400-meter walk time. The addition of the linear strength term resulted in an R² equal to 52.3%, with addition of the quadratic term increasing R² by 1.1%, to 53.4%. The addition of the cubic strength term explained <0.1% of the total variability. The addition of the linear power term to the covariates resulted in an R² equal to 55.6%, with the quadratic term increasing the R² by an additional 1.1% and the cubic term increasing the R² by an additional 0.4%. The fact that power and strength terms are close in total explanatory power after controlling for all covariates, even though power is the stronger predictor in the uncontrolled analyses, indicates that some of the covariates are more highly correlated with power than strength. In particular, gender and height are more strongly associated with power than strength. If these covariates are not included, the total explained variability of covariates alone is 42.7%, increasing by approximately 8% (to 50.9%) if linear and quadratic strength terms are added, and by approximately 14% (to 56.9%) if similar terms for power are added. The magnitude of the difference in explained variability (6%, 56.9 vs 50.9%) after inclusion of linear and quadratic power versus strength terms (adjusting for covariates other than gender and height) is larger than observed for the adjusted analyses that included gender and height (4.3%, 57.7 vs 53.4%), indicating more overlap in the 400-meter walk variability explained by these two covariates and power compared to strength. An alternative perspective is that the contribution of gender and height in explaining the variability of 400-meter walk time is less after entering power (0.8%, 57.7 vs 56.9%) in the model compared to entering strength (2.5%, 53.4 vs 50.9%).

In Figure 1, the relationship between 400-meter walk time and power and strength is displayed relative to the observed data. On each figure, we identify a dashed line that represents the fitted model from a piecewise regression that estimated a threshold at which the quadratic equation becomes a flat line. For LEIS, introduction of the cubic term did not add to the explanatory ability of the quadratic term, with the quadratic term increasing the explained variability of the unadjusted analyses by approximately 4%. For the piecewise regression models contained in Figure 1, the threshold value was estimated to be 215.4 W (95% CI 178.1–242.7) for LEMP and 49.6 kg (95% CI 43.1–56.3) for LEIS.

In analyses where we examined interaction terms between strength or power and covariates, we found that the curvilinear form of the relationship between strength or power and 400-meter walk time did not change. While those participants with dyspnea, diabetes, hypertension, and peripheral arterial disease completed the 400-meter walk in a significantly slower time, the curvilinear relationship and lack of strong evidence of a threshold value remained consistent for all of these analyses.

Means of strength and power measurements were calculated for participants that completed the 400-meter walk and those that either did not complete the walk or did not attempt the walk. Participants who completed the task had significantly greater strength and power than the other two groups (Power means [W]: completed 400-meter walk = 115.1; did not complete 400-meter walk = 58.3; did not attempt 400-meter walk = 67.5; Strength means [kg]: completed 400-meter walk = 33.1; did not complete 400-meter walk = 25.9; did not attempt 400-meter walk = 24.3).

Conclusion

Lower extremity muscle strength and power are important predictors of the time to walk 400 meters at a steady pace. It remains questionable due to the small changes in explained variance as to whether LEMP is a better predictor of the time to walk 400 meters than lower extremity strength. Finally, curvilinear relationships exist between strength and power with the time to walk 400 meters, and the curvilinear form of the relationship is very robust to gender and comorbidities. However, the data do not appear to be suggestive of a threshold value for strength or power.