Participants
In this cross-sectional study of a sample of women aged 60 years or older, who were physically active and independent from assistance or assistive tools, were consecutively recruited from the Open University of the Third Age of the XXXXXXXX. Inclusion criteria comprised good cognitive condition (MiniMental State Examination (MMSE) score > 24 [15], Body Mass Index below 30 kg/m2, and robustness [16]. Exclusion criteria comprised the reporting of falls in the previous year, complaints of back pain or impaired gait, ≥ 7 points in the Clinical-Functional Vulnerability Index (IVCF-20) [16], scoliosis Cobb angle > 20°; skeletal abnormalities, neurological or musculoskeletal disabilities that interfered with gait; uncontrolled diabetes mellitus; alcoholism; history of spine surgery, and fractures or prosthesis implants in the lower limbs.
The research was approved by the Ethical Committee of XXXX (#824.155), and written informed consent was obtained from each participant after the explanation of the objective and methodology of the study. During the first visit, participants answered questionnaires regarding cognitive condition, vulnerability, falls, and self-reported exercise adherence. The answers were evaluated by a rheumatologist (XX) and in case of eligibility for the study, the participants were sent to radiological evaluation of the sagittal balance, carried out by a trained radiologist (XX). In a second visit, performance tests and accelerometry recordings were conducted by a trained physical therapist (XXX).
Questionnaires
The IVCF-20 was used to assess vulnerability. The IVCF-20 is a 20 question multidimensional inventory that covers aspects of the elderly’s health conditions, which includes age, health self-perception, functional disabilities, cognition, mood, mobility, communication, and multiple comorbidities. The maximum score is 40 points and the higher the value obtained, higher is the elderly’s risk for clinical-functional vulnerability. In our study, robustness was defined for scores < 7 as reported for the Brazilian elderly [16].
Cross-cultural adaptation of the Falls Efficacy Scale-International (FES-I) for Brazilians was used to assess the fear of falling during the previous year. This 16-question inventory inquiring “how concerned are you about the possibility of falling” is rated as not at all, somewhat, fairly, and very concerned. The total score ranges from 16 to 64 points, being the higher values indicative of higher concern about falling [17].
Self-reported exercise adherence was assessed by the weekly frequency of reported aerobic or resistance training exercises in the previous month. According to the answers, subjects were considered as highly active (physical exercises at least 3 days a week), moderate (at least once a week) and inactive (less than once a week).
Performance tests
Participants were subjected to two sessions of a 10-m walk at a comfortable speed, which was measured in meters/second. The tests were performed on a level and clear 30-m corridor, with 1.5-m space for acceleration and deceleration, and without encouraging words. Gait speed between 0.90–1.42 m/s has been considered adequate for the walking economy during the ageing process [18, 19]. Berg Balance Scale validated to Brazilian Portuguese [20] was used to assess functional balance performance; and was based on 14 questions with answers ranging from 0 to 4 points. Total scores of 0 to 20 represent balance impairment, of 21 to 40 indicate acceptable balance, and of 41 to 56, good balance [21].
Timed Up-and-Go (TUG) was also carried out for functional performance. Participants, wearing their usual footwear, were seated on a standard chair with solid seat, with their backs supported by a flat backrest, and their arms supported by armrests. At the evaluator’s instruction, they had to stand up, walk a 3 m-distance in a straight line, in a comfortable speed, cross a mark on the floor, then turn around 180°, walk back, and sit down again. There was one practice session immediately before the test session was performed. The evaluator initiated the chronometer at the moment the participants stood up, and stopped it only when they were seating on the chair with their arms and back supported again. TUG values over 12.47 s are considered as predictive of falls for the Brazilian older adults [22]..
Radiological evaluation
For Pelvic Incidence, complete overlap between the femoral heads was required. Sagittal images were obtained with subjects in the orthostatic position, with upper limbs flexed, hands on the shoulders in 30° flexion, knees extended, and looking straight ahead. The image scope was from the cervical spine to the femoral heads, and the distance from source-to-film was 2.3 m [23].
Thoracic Cobb angle was computed as the angle formed between the two straight lines that were orthogonal to the upper endplate of T4 to the endplate of T12. For Lumbar Cobb the reference points were the upper endplate of L1 to the lower endplate of L5 [24]. Thoracic Cobb mean values for older Brazilian women are 43.1° ± 13.6°, and for the Lumbar Cobb, − 41.7° ± 11.9° [25].
Pelvic Incidence angle was defined by a line drawn from the center of the femoral heads to the midpoint of the sacral endplate, and a line perpendicular to the center of the sacral endplate. Pelvic Tilt was defined as the angle subtended by a vertical reference line originating from the center of the femoral head and the midpoint of the sacral endplate. Sacral Slope was defined as the angle subtended by a horizontal reference line and the sacral endplate line. The reported mean reference values for Brazilians, older than 60 years were: Pelvic Incidence angle of 50.9° ± 6.6°, Sacral Slope of 37.4° ± 5.1°, and Pelvic Tilt of 13.8° ± 5° [26].
Accelerometry
We used the MiniMod DynaPort™ (McRoberts, Netherlands) with a signal acquisition frequency of 100 Hz, in a range of variation of ±2 g in a 30-m walking trial [27]. The device was attached to a belt that was placed firmly at the point behind the spinal process of L3. This point is easily located and it is convenient for global measure of walking parameters, due to its proximity to the CM in the upright stance and during walking [28, 29]. Gait was assessed in a 30-m walk distance held at comfortable speed, with a 1.5-m space for acceleration and deceleration, and without encouraging words.
Data was collected by MiniMod Acquire (MIRA™; McRoberts, Netherlands) and processed using iGait, which is an open-source software from MatLab™ environment [30]. After alignment correction, gait data extracted in the temporal domain included: Stride Regularity, Gait Symmetry, Root Mean Square (RMS) value in the anteroposterior (AP), vertical (V), and medio-lateral (ML) planes. The square root values of the three planes were normalized to the average value of square root/RMS. Step and Stride Regularities were defined as the correlation between the original acceleration signal and the acceleration signal phase shifted to the average step and stride time, respectively [28,29,30]. Gait Symmetry was defined as the difference between the consistency of strides and steps in the accelerometer waveforms. Step Length was calculated dividing the 30-m walk distance by the number of peaks of acceleration detected by the accelerometer during the foot contact with the ground.
Statistical analysis
A sample of 38 elderly women allowed detection of a correlation of 0.44 or more between radiographic and accelerometer-derived data with an alpha of 0.05 and power of 80% [31].
Mean and standard deviation (SD) were used for data with parametric distribution whereas median and interquartile range (IQR) were used for the non-parametric data. The Spearman’s rank correlation coefficient was used to calculate correlations between the angles of the spine and pelvis with clinical, functional and accelerometer-derived parameters.
Multivariate regression was used to evaluate the existing association of age with radiological and accelerometer-derived data of normal distribution. Independent variables tested were age, height, weight, Thoracic Cobb, Lumbar Cobb, Sacral Slope, Pelvic Tilt, whereas the dependent variables tested were, RMS, Step Length, Stride Regularity and Gait Symmetry in all planes. Data were tabulated and analyzed using SPSS Statistics for Windows, Version 22.0, IBM Corp and the significance was set at 5%.