Gait Variability Detects Women in Early Postmenopause With Low Bone Mineral Density

Study Design

This study was an observational cohort study. Participants were assigned to groups on the basis of BMD status. Participants were assigned to the low BMD group if their BMD was ≥2.0 standard deviations below the mean for adults who are healthy at one or more skeletal sites in accordance with the DXA scan results.⁸ The results of each DXA scan were confirmed by the radiologist's written DXA scan report. Participants were assigned to the normal BMD group if their BMD was within 1.0 standard deviation of the mean for adults who are healthy. Participants were excluded from the latter group when any skeletal site had a bone density of 1.1 to 1.9 standard deviations below the mean for adults who are healthy. This exclusion criterion recognizes that although severe osteopenia is defined as 2.0 standard deviations below the mean for adults who are healthy,²⁵ the range of 1.1 to 1.9 is still classified as osteopenia.

All testing occurred at Arcadia University. Pilot testing was conducted to estimate sample size. Eight women, 4 with low BMD and 4 with normal BMD, completed the pilot testing. These women had a mean (SD) age of 59.1 (3.1) years; had been postmenopausal for, on average, 6.3 (SD=2.2) years; had a mean (SD) body mass index (BMI) of 24.3 (1.5) kg/m²; and had a mean (SD) physical activity level of 2,214.1 (189.8) kcal/d, as determined with the Stanford 7-Day Physical Activity Recall Questionnaire. Results from the pilot study indicated that women in early postmenopause and with low BMD exhibited differences in backward tandem walk time and quadriceps femoris muscle strength compared with their peers who were healthy. Effect sizes were large (r=−.75) for backward tandem walk time and medium for the strength of both quadriceps femoris muscles (r=.43). The study was powered with a medium effect size because it represented the lowest effect size for the specific aims. Given a power of at least 0.80, a medium effect size of r=.43, and an alpha level of .05, a target sample of 28 women per group, for a total of 56 participants, was assumed.²⁶ Data collection for the main study took place from August 2006 to April 2007. Performance data were collected in one session, which lasted approximately 1 hour. Participants were contacted by telephone 1 year after performance testing.

Participants

Participants were recruited through informational flyers placed in Arcadia University–area community buildings for people aged 55 years and older, churches, gymnasiums, synagogues, physical therapy clinics, and libraries and advertisements placed in church bulletins and local newspapers. A total of 82 women responded to recruitment flyers and advertisements.

After potential participants made contact with the principal investigator (K.M.P.), they were telephoned and screened with inclusion and exclusion criteria. To be considered for inclusion in the study, participants had to be community-dwelling women who were 50 to 65 years of age and 3 to 10 years after menopause, who had DXA test results from within the preceding 2 years, and who were able to ambulate independently. Years after menopause was assessed by reading the definition of menopause, which is “cessation of the menstrual period for a span of time of at least 12 months,”²⁷ and then asking, “At what age did your period cease?”

Participants were excluded if they were organ transplant recipients, had active, severe liver or kidney disease, were undergoing chemotherapy or renal dialysis, had chemical menopause or menopause because of oophorectomy, had chronic, severe cardiac or pulmonary disease, had a history of bone cancer or bone metastases, used corticosteroids for a long period of time, or had medical issues that affected the lower extremities and that, in turn, could affect ambulation. Examples of such medical issues were joint replacement; joint inflammation causing swelling, tenderness, or both; lower-extremity surgery within the preceding 6 months; or neurological diseases that could affect ambulation, such as Parkinson disease, multiple sclerosis, or residual weakness from a cerebrovascular accident. Twenty-four women did not meet all of the inclusion criteria (age, n=1; years after menopause, n=12; DXA scan test date, n=2; chronic lung disease or use of asthma inhalers, n=2; menopause because of oophorectomy, n=4; lower-extremity joint replacement, n=2; and neurological disease affecting the lower extremities, n=1), and 1 woman declined to participate without financial compensation.

Potential participants who met the inclusion criteria were scheduled for assessment. Participants were informed of the study objectives and provided written informed consent before data collection. Fifty-seven women, 33 with low BMD and 24 with normal BMD, participated in the testing for this study. Two participants were excluded from the low BMD group because their physical activity levels were greater than 2 standard deviations above the participant mean (4,130.23 and 4,166.20 kcal/d, respectively), and 1 participant was excluded from the normal BMD group because both her physical activity level and BMI were greater than 2 standard deviations above the group mean (3,953.73 kcal/d and 42.2, respectively) (Fig. 1). Four participants (3 in the normal BMD group and 1 in the low BMD group) were lost to follow-up for the 1-year evaluation of falls and fractures despite numerous attempts to contact them.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Flow diagram of study recruitment. BMD=bone mineral density, BMI=body mass index.

Demographic Variables

The participants completed a demographic and health status questionnaire that included questions about age, number of years after menopause, date and results of the last DXA scan, medical history, current medications, and use of calcium and vitamin D supplements. With the exception of the DXA scan results, these variables are associated with BMD,^28–31 physical performance,^32–34 or both. Participants’ height (in meters) and weight (in kilograms) were measured on a calibrated physician scale,* and BMI (kg/m²) was calculated from these values. The BMI was calculated because it is associated with sarcopenia³⁵ and BMD.²⁸

Next, the participants were administered the Stanford 7-Day Physical Activity Recall Questionnaire.³⁶ Physical activity level was measured as a potential covariate because activity level could affect both physical performance^37–39 and BMD.⁴⁰ The Stanford 7-Day Physical Activity Recall Questionnaire was selected because it has been used with both older women⁴¹ and women who are postmenopausal and have low BMD.^42–45 This questionnaire asks people to recall the number of hours slept each night as well as the number of hours spent in moderate, hard, and very hard activities during the preceding 7 days. Examples of moderate, hard, and very hard physical, social, and occupational activities are provided to each individual. A description of exertion levels also is provided. Energy expenditure is calculated from the numbers of hours spent in sleep and in activity and is reported as kilocalories per day. The reliability³⁶ and validity⁴⁶ of Stanford 7-Day Physical Activity Recall Questionnaire scores have been reported.

Of the 57 women who participated in the testing for this study, 55 were white and 2 were African American. On average, the women were 58.8 (SD=3.3) years of age and had been postmenopausal for 6.1 (SD=2.5) years. Analysis of the results indicated significant between-group differences for BMI and level of physical activity. Participants with low BMD had a mean BMI that fell within the healthy category, and those with normal BMD had a mean BMI that fell in the overweight category. Participants with low BMD had a mean (SD) physical activity level of 2,290.61 (376.32) kcal/d, or 13 h/wk, and those with normal BMD had a mean (SD) physical activity level of 2,683.79 (573.05) kcal/d, or 15 h/wk. All participants walked at speeds that exceeded reported thresholds for successful community ambulation, such as crossing the street within the timing of a traffic light (1.0–1.2 m/s).⁴⁷

The most common types of medications included over-the-counter supplements, daily aspirin, statins, and osteoporosis medications for participants in the low BMD group. The most common comorbidities were high cholesterol levels and hypothyroidism (Tab. 1).

Physical Performance Variables

Physical performance measures included assessment of dynamic balance, quadriceps femoris muscle strength, and gait. The order of testing was the same for all participants. Dynamic balance was assessed with a timed, backward tandem walk test.^48,49 Participants were instructed to walk backward as quickly as possible in a heel-to-toe fashion on a 2.44-m line. This test is timed with a stopwatch.³⁷ Participants were allowed 1 practice trial to become accustomed to the testing procedure. They then performed 2 trials of the test. The time for the 2 trials was recorded in seconds, and the mean time for the 2 trials was calculated. This test was selected to minimize the ceiling effect that can occur when subjects under age 65 are tested; other balance measures that use ordinal scales, such as the Berg Balance Scale, have a documented ceiling effect for such subjects.⁵⁰ The backward tandem walk test has been used for older adults⁴⁹ and people with osteoporosis.⁴⁸ The validity of the backward tandem walk test for isokinetic knee extension strength has been reported.⁴⁹

Isometric quadriceps femoris muscle force production for each leg was measured with a MicroFET 2 handheld dynamometer.^† Participants sat upright on a plinth of sufficient height to prevent the feet from touching the floor. The upper thighs were strapped to the plinth to prevent compensatory movements, and the knees were flexed to 90 degrees.⁵¹ The dynamometer was placed on the anterior tibia just proximal to the malleolus.⁵¹ Participants were asked to push as hard as possible for 5 seconds while the investigator matched the resistance (make test). The left lower extremity was tested first. Participants performed 2 practice trials and then a minimum of 2 maximal-effort trials with a 1-minute rest period between the trials. The peak force for each trial was recorded, and the mean for the 2 trials was calculated.⁵² In the event that the 2 trials were not within 10% of each other (∼50% of the participants), additional trials were performed. The quadriceps femoris muscle complex was chosen because quadriceps femoris muscle strength has been found to predict gait speed¹¹ and is associated with static and dynamic balance in women with osteoporosis.⁴² Handheld dynamometry has been reported to have reliability⁴² and validity.⁵³ Handheld dynamometry⁵² and similar testing procedures^42–44 have been used to assess lower-extremity strength in women with osteoporosis.

Free gait speed data were collected with the GaitMat II^‡ to determine temporal and spatial gait characteristics. The GaitMat II is a 3.87-m walkway containing pressure-sensitive switches⁵⁴ (Fig. 2). Free gait speed was chosen as a variable because walking speed is one of the criteria included in the definition of frailty.⁵ The participants completed 2 trials of free speed with the instruction to “walk across the mat at your normal speed.” Individually determined rest periods were given between the trials. Gait variability was calculated from the walking trials as the coefficient of variation of free speed step length, stride length, step width, step time, and stance time as described by Brach et al¹³ (Fig. 1). The coefficient of variation, calculated as (standard deviation/mean) × 100,¹³ represents the variability within a distribution of numbers. The reliability and validity of data obtained with the GaitMat II have been reported.⁵⁴

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

(A) GaitMAT II System. (B) Step length, defined as the distance between contacts of the contralateral extremities. Step length is defined with respect to the advancing limb.⁷⁷ (C) Step width, defined as the lateral distance between contralateral footfalls. Thus, right step width is defined with respect to the left trailing limb and left step width is defined with respect to the right trailing limb.⁷⁷ (D) Stride length, defined as the distance from the contact of one foot to the subsequent contact of the same foot. Initial contact typically is used as the measurement reference. One stride represents the completion of one cycle of gait.⁷⁷ (E) Step time, defined as the time necessary to complete a right or left step length.⁷⁷ (F) Stance time, defined as the time a limb is in contact with the ground.⁷⁷ Photographs used with permission of E.Q. Inc.

Data on Falls and Fractures

Data on falls and fractures were collected for the year after initial testing. Participants were contacted 1 year from their initial test date and were provided with the definition of a fall: “A fall is defined as unintentionally coming to rest on the ground, floor, or other lower level.”^55(p1619)

Participants then were asked whether they had fallen in the preceding year and, if so, how many falls they had sustained. Participants also were asked whether they had sustained any fractures or broken bones in the preceding year and, if so, which bones they had broken. They were not questioned about the circumstances of the fall or whether a hospitalization resulted from the fall.

Data Analysis

All data were analyzed with SPSS software, version 14.0.^§ Descriptive statistics were determined for all demographic and health-related variables. Data were excluded from the analyses when a participant's physical activity level or BMI exceeded 2 standard deviations of the group mean to prevent outliers from skewing the results. Independent sample t tests were performed to assess between-group differences for all demographic variables and to assess potential between-group differences between the pilot sample and the full-study sample for both the low BMD group and the normal BMD group.

Analyses of variance were performed to test differences in physical performance variables (primary aim) and differences in gait variability (secondary aim). Chi-square analyses were performed to test differences in rates of falls and fractures (secondary aim). We treated falls and fractures as dichotomous variables instead of using the numbers of falls and fractures as metrics. Statistical significance was set at P<.05.

Role of the Funding Source

This study was funded, in part, by grants from the Pennsylvania Physical Therapy Association and the Section on Women's Health of the American Physical Therapy Association. The funding received for this study allowed for office supply costs, such as postage and purchases of paper and statistical software; photocopying costs; advertising costs for recruitment purposes; and partial salary support for the principal investigator (K.M.P.). The principal investigator was required to submit a poster presentation of preliminary results to the Pennsylvania Physical Therapy Association and one article related to her dissertation to the Journal of Women's Health Physical Therapy as part of the funding stipulations. This study was conducted in partial fulfillment of Dr Palombaro's doctoral degree.