Study design
YESS consisted of a 32-week yoga program with 2 phases: a 16-week beginning phase (Series I) and a 16-week intermediate phase (Series II). The study design and poses that were used in each phase have been detailed [6]. The primary biomechanical outcome variables were the net JMOFs during the performance of the individual yoga poses (asanas). Muscle activation patterns associated with the asanas, and adherence to the yoga program, were also assessed. Biomechanical data was collected at the Musculoskeletal Biomechanics Research Laboratory (MBRL) at the University of Southern California (USC). Subject recruitment and the yoga classes were conducted at the University of California Los Angeles (UCLA) and TruYoga studio (Santa Monica, CA), respectively. The USC and UCLA Institutional Review Boards approved the study protocol and all participants provided informed, written consent.
Subjects
YESS was designed to design and test senior-adapted Hatha yoga poses intended to be suitable for ambulatory older adults. The study sample size was determined by power analysis (β = 0.8; p < 0.01) using the JMOF findings from a previous study [7]. Safety exclusions were adopted in order to decrease potential cardiovascular, musculoskeletal, and neurological risks to the participants; these included: active angina; uncontrolled hypertension (SBP > 160 or/and DBP > 90); high resting heart rate (greater than 90) or respiratory rate (greater than 24); unstable asthma or exacerbated chronic obstructive pulmonary disease; cervical spine instability or other significant neck injury; rheumatoid arthritis; unstable ankle, knee, hip, shoulder, elbow, or wrist joints; hemiparesis or paraparesis; movement disorders; peripheral neuropathy; stroke with residual deficit; severe vision or hearing problems; walker or wheelchair use; not able to attend in-person classes; has not had check-up by regular provider within 12 months (if not taking any prescription medications) or in the past 6 months (if any regular medicines taken). Participants also had to execute the following safety tests stably and independently: transition from standing to recumbent on the floor and reverse; lift both arms to shoulder level; stand with feet side-by-side for 30 seconds; and stand with feet hip-width apart for 60 seconds.
Yoga program
Participants attended 2 60-minute yoga sessions per week, for 32 weeks. The yoga program was developed by the research team, which included an experienced yoga therapist (RYT-500), a geriatrician, an exercise physiologist/biomechanist, and a physical therapist. In general, the program was an adapted form of Hatha yoga [8]. Two sets of poses (Series I and Series II) were taught. We report herein the biomechanical findings assessed after the completion of the second series, because the second series was more homogenously performed than that was the first series. This increasing homogeneity in pose form over time is inherent in working with senior participants, who initially exhibit a broad range of yoga-performance capabilities, related to each subject’s strength, flexibility, balance, overall fitness and group-exercise experience. This heterogeneity in capability necessitates greater pose modification to avoid harm. However, the second series poses build on the training achieved in the first series. Thus, they require fewer modifications from the standard forms of the asanas. By the end of the second series training period (32-weeks), all participants could perform all second series poses. Thus, the present analysis examines 7 standing poses, performed at the completion of the series two training period. These are: Chair, Wall Plank, Tree, Warrior II, Side Stretch, Crescent, and One-Legged Balance. A detailed description of the poses, including photographs, can be found in the report by Greendale et al. [6].
Kinematics and kinetics
Reflective markers were placed on a head band and over the following anatomical landmarks of the lower- and upper-extremities bilaterally: first and fifth metatarsal heads, malleoli, femoral epicondyles, greater trochanters, acromions, greater tubercles, humeral epicondyles, radial and ulnar styloid processes, and third metacarpal heads. Markers were also attached to the spinous process of the 7th cervical vertebrae (C7), jugular notch, L5/S1, bilateral iliac crests, and bilateral posterior superior iliac spines, in order to define the trunk and pelvis. Based on these markers, a total of 15 body segments were modeled: the head, trunk, pelvis, the upper arms, forearms, hands, thighs, shanks, and feet. Non-collinear tracking marker plates were placed on each of these segments to track segmental position during the poses, using previously documented procedures [9, 10].
Once instrumented, the subjects performed their poses while guided by the yoga instructor (Figure 1
a). Props, including foam blocks (One-legged Balance) and a chair (Side Stretch and Crescent) were used in the same manner as they were used during the participant’s regular practice sessions in the yoga studio. A firm but portable clear plexiglas wall, which permitted capture of the markers, was positioned for the Wall Plank pose. For the single-limb poses, measurements were taken on the dominant limb. For poses requiring bilateral limb support, each foot was positioned on an independent force platform. For each pose, the participant was instructed to begin in a starting position, move smoothly into the pose, hold the pose while taking a full breath, then return back to the original position. The instructor performed each pose along with the participant in order to provide visual cueing. Two trials of each pose were collected and averaged; data during the middle 3 seconds of each pose was used for analysis (Figure 2
a). Data during the middle 3 seconds, while holding the pose, was used for analysis. For poses that involved asymmetric positioning of the 2 support limbs (e.g. Side Stretch, Crescent, and Warrior II asanas), measurements were obtained by repeating the poses, initially with the dominant limb in the lead (front) position and subsequently in the trailing (back) position. Because the JMOFs varied considerably between the leading and trailing limbs, the limbs were considered separately. Thus, Side Stretch, Crescent, and Warrior II asanas were subdivided into Leading- and Trailing-limb poses (e.g. Crescent Leading and Crescent Trailing). The subjects also completed 2 successful walking trials at their self-selected “comfortable speed”. The walking trials provided a well-studied, stereotypical activity for comparisons of the JMOFs and muscle activation patterns with the respective poses.
Three-dimensional coordinates of the body segments were recorded by an 11-camera system at 60 Hz (Qualisys, Gothenburg, Sweden). Ground reaction forces were measured from separate force platforms at 1560 Hz (AMTI, Watertown, MA). Data processing software (Visual 3D, C-Motion, Inc. Germantown, MD) was used to process the raw coordinate data and compute the segmental kinematics. The principle moments of inertia were determined from the subject’s total body weight, segment geometry, and anthropometric data. Using standard inverse dynamics techniques consistent with the International Society of Biomechanics recommended coordinate systems, the JMOFs in the sagittal and frontal planes, for the ankle, knee and hip, were calculated from the inertial properties, segmental kinematics, and ground reaction forces [11, 12]. JMOFs were normalized to each subject’s bodyweight in kilograms (kg). Additionally, a support moment, calculated as the sum of the ankle, knee, and hip sagittal plane JMOFs, was determined for each pose [13, 14]. These instrumentation and data-processing techniques have previously been used in our laboratory to assess exercise activities in older adults with high reliability (Cronbach’s alpha = 0.98) [15].
Electromyography (EMG)
Surface EMG signals of the gluteus medius (GMED), hamstrings (HAMS), vastus lateralis (VL), and gastrocnemius (GAS) muscles were collected using active surface electrodes (Motion Lab Systems, Baton Rouge, LA). Data from the dominant limb were recorded at 1560 Hz. Standard procedures for older-adult participants including preparation of the skin and electrode placement were employed [16]. The EMG signals were filtered according to ISEK standards, including, notch filtering at 60 Hz, and band-pass filtering between 20 and 500 Hz. A root mean square smoothing algorithm, with a 75-millisecond constant window, was used to smooth the EMG data [17]. EMG smoothing, processing and normalization were performed using a custom written MATLAB program (MathWorks, Natick, MA).
Data analysis
Visual inspection was used to select the top 4–5 ranked poses for statistical comparison, across each joint, plane of motion, and direction. Parametric distribution of the JMOFs was confirmed by analyzing the skewness and kurtosis of the data. Repeated-measure ANOVA omnibus tests were used to identify significant differences in the JMOFs within each cluster of the top 4–5 ranked poses. When the results were significant, Tukey’s post-hoc tests were used to examine the pairwise comparisons. Additionally, Cohen’s d effect sizes (small d = 0.2; medium d = 0.5; large d = 0.8) are reported for all statistically significant post-hoc comparisons [18]. Statistical analysis was conducted via PASW Statistics 18 (IBM SPSS Statistics, Armonk, NY) and significance level was set at α = 0.05. The EMG data was used to support the primary JMOF findings; formal statistical analyses were not conducted on the EMG data.