2.1 Study design and participants
The study design was cross-sectional with a convenience sample (specifically, modal instance sampling) and adhered to the STROBE guidelines (http://www.strobe-statement.org/). Eighty-three participants integrated the sample, of which 40 people with Parkinson's disease composed the PD group (PDG) and 43 participants formed the group of healthy counterparts (HG) matched by age, gender, and height. PDG participants were recruited from two centers: a Parkinson’s disease Association and the Neurology Service of a local public hospital. HG individuals were from a Municipal Senior Social Activity Center. Inclusion criteria for PDG were: diagnosis of idiopathic PD and to present Hoehn & Yahr stadium I, II or III. In addition, both groups should be able to walk by themselves, have a normal cognitive state (i.e. score >25) according to the Minimental test adapted for PD [10], symmetry in lower limb length (<1cm), and at least two months of sedentary life and without having attended physiotherapysessions Exclusion criteria for both groups were the presence of another neurological or symptomatic musculoskeletal disease (e.g. musculoskeletal pain), history of trauma or surgery on the lower limbs, balance disorders due to other diseases and uncontrolled chronic diseases (e.g. hypertension or diabetes).
2.2 Procedures
The assessments were carried out at the Medicine Department of the University of Valencia. PDG participants were evaluated in the on-medication state [4]. The assessment session included first, a clinical interview to record the main personal data of the participants and to verify their cognitive state; second, an anthropometric evaluation to record weight, height, and length of lower limbs [11] used in the standardization of biomechanical variables; and third, biomechanical assessment of gait at a self-selected comfortable speed. All the participants walked barefoot under all conditions tested to avoid the damping provided by the different footwear of the participants and therefore, prevent a confusing variable in obtaining ground reaction forces. Five conditions were randomly evaluated: i. single task (ST): walking without secondary tasks with the attention focused only on walking performance, ii. visual DT: walking while checking the time on an analog clock projected at the end of the walkway, iii. verbal DT: walking while telling the evaluator the activities they had performed the previous day in chronological order, iv. auditory DT: walking while listening and recognizing different daily noises and, v. motor DT: walking while carrying one glass in each hand and repeatedly transferring their contents from one to the other (beans). These tasks were intended to simulate the action of looking at the time, talking with another person, listening to the sounds of the environment and manipulating objects with hands while walking. All these tasks are representatives of daily life activities, which provide external validity to the study and concur with skills previously studied [4,12]. During DT gait, participants were urged to focus attention on the secondary task through all the way walking.
2.3 Biomechanical gait assessment
Gait assessment was carried out in a corridor 10m long and data were registered using 3D photogrammetry with 12 smart cams (Kinescan/IBV software, Biomechanical Institute of Valencia, Valencia, Spain, version 5.3.0.1) and two force platforms (Dinascan/IBV Biomechanical Institute of Valencia, Valencia, Spain). A valid repetition is one in which the patient performs at least five complete strides along the corridor and where one complete and isolated footprint (rigth or left) from the third stride coincide with the dynamometric platform located in the center of the corridor. The previous and subsequent strides were discarded to avoid acceleration and deceleration phases, at the beginning and at the end of the gait, respectively. For each of the five evaluated walking conditions, 10 repetitions were performed, five with each foot, to later use the average of these. Participants were allowed to rest between repetitions, sitting on a stool for less than three minutes, when fatigue was reported.
The biomechanical model was composed of 35 landmarks located in specific anatomical points on both sides of the body: i. the spinous process of the seventh cervical vertebra, ii. the acromioclavicular joint, iii. the posterosuperior iliac spine, iv. the anterior-superior iliac spine, v. the greater trochanter of the femur, vi. the anterior vertex triangle forming the thigh segment, vi. the medial condyle of the knee, vii. the lateral condyle of the knee, viii. the posterior vertex triangle of the leg segment, ix. the medial malleolus of the ankle, x. lateral malleolus of the ankle, xi. the posterior surface of calcaneus, xii. the tuberosity of fifth metatarsal and xiii. cuboids. For suitable biomechanical modeling, the patient's clothing included shorts, sleeveless shirt and slip-on paddings (with the top of the foot uncovered). Before recording gait, participants were allowed to walk in the corridor (ST condition) to familiarize themselves with the test.
2.4 Outcomes
For calculation of biomechanical variables based on the data exported from the photogrammetry system and the dynamometric platform, the software was programmed with Matlab (MathWorks, MA, USA, version R2016b). The average of these dependent variables was calculated from the 10 repetitions for each condition: i. Velocity: distance travelled by the body per unit of time (m/s), ii. Stride length: distance measured between two consecutive heel strikes of the same foot (i.e. two steps) (m), iii. Cadence: number of steps taken in a minute (steps/min), iv. Double support time: the sum of the amount of time in which there is double-limb support in a gait cycle (%), v. Ankle range: range of motion (ROM) that corresponds to the sum of the maximum angle of plantar flexion and maximum angle of dorsiflexion of the foot (°), vi. Hip flexion: maximum flexion angle reached by the hip joint during the swing phase of the gait cycle (°), vii. Hip extension: maximum extension angle reached by the hip joint during the stance phase of the gait cycle (°), viii. Weight-acceptance force: first peak of the vertical component curve of reaction forces corresponding to the maximal weight-acceptance (N), ix. Midstance force: minimum value between the two force peaks of the vertical component curve of reaction forces corresponding to the midstance of the gait cycle (N), x. Braking force: first negative peak of anteroposterior component curve of reaction force that corresponds to the braking (N).
In addition, to inform about the degree of interference of dual tasks during gait, the Dual-task cost was calculated as the percentage measure of performance decline observed under DT conditions (Equation 1) [4,13]. Negative values mean a decrease in the value measured during the dual-task compared with the single-task condition. Likewise, the same equation was used to describe the performance of PDG with respect to the HG participants, which was defined as the Performance of parkinsonian gait (PPgait). When the PPgait percentage (Equation 2) was negative, it meant that the PDG had registered lower values in the variable analyzed than the HG, while positive PPgait indicated that the PDG has recorded higher values than the HG. For the calculation of both equations, the mean values obtained in each outcome and group were used.
Equation 1: Dual-task cost. Interference rate of dual tasks during gait.
(see Equation 1 in the Supplementary Files)
Equation 2: Performance of parkinsonian gait. Impairment of the group with Parkinson's disease compared to the healthy group.
(see Equation 2 in the Supplementary Files)
2.5 Data analyses
Statistical analyses were performed using IBM SPSS v.24 (SPSS Inc., Chicago, IL, USA). Standard statistical methods were used to obtain the mean and standard deviation (SD). Also, as descriptive results, the confidence interval, PPgait percentage, and DT cost were calculated. A two-factor mixed Multivariate analysis of variance was conducted to analyze the effect of within-subject factor conditions with five categories (ST, visual DT, verbal DT, auditory DT, and motor DT) and between-subject factor group with two categories (i.e. PDG and HG) on the dependent biomechanics variables. Some of these variables were standardized according to anthropometric data, concretely, ‘stride length’, which was standardized by the lower limb length, and ground reaction force variables were standardized according to the weight of the participants.
When significant factor effects were found, the Bonferroni correction, provided by the statistics package, was used for pairwise comparisons. Differences were declared statistically significant if p < 0.05 and the 95% confidence interval (CI) of the pairwise mean differences was reported.
To check for differences between the demographic outcomes between groups, multivariate analysis with one-way between-subject factor group was conducted. Further, to demonstrate differences in gender between groups, a chi-square test was used.