Seventy consecutive patients (May 2012 – May 2014) undergoing primary standardised cemented unilateral TKR (single surgeon; 15-years’ experience of knee replacement; 50 knee replacements per annum) were invited to participate in the study. The inclusion criteria for participants were: a) Ambulatory at the time of surgery patients with OA (clinical and radiological findings of advanced osteoarthritis, 6–12 months length of wait for surgery) undergoing primary standardised cemented TKR by the same surgeon; b) Aged 65–80 years old. Patients were excluded from the study if they had: a) Infection, or complications after TKR; b) Neurological/neuromuscular conditions; c) Vestibular disorders that might affect balance; d) Other lower extremity orthopedic problems that limited function; e) Cardiovascular diseases, high blood pressure not controlled with medication and f) Unable to communicate or follow instructions or complete objective assessments.
A clinical trial was undertaken at a primary care university hospital in Greece (International Standard Randomised Control Trial Registration: ISRCTN12101643). All patients provided written informed consent, following a verbal and written explanation of the study procedures. The study had been ethically approved by two Institutional Committees (University Hospital of Patras, Greece and Queen Margaret University Edinburgh, UK [7052/4-7-2011]) and adhered to the Consort guidelines.
Ten blocks of 5 patients were randomly assigned to two groups (SMT [experimental]; Control [usual practice]) using a computer-generated number sequence overseen by an independent statistician. The confidential coded listing, maintained until after data analyses), assured allocation concealment from participants. This study involved a single-blind design as the principal investigator undertook assessment and training sessions. However, every effort was made to preclude bias (i.e. an undergraduate student recorded data during assessment sessions, and the principal investigator analysed data at the end of the rehabilitation period using the coded listing). A further 2 patients were included and assigned in the original block-allocation order, contributing to the study’s 52 patients.
Time-matched rehabilitative procedure performed by both groups
All participants underwent a standardised post-surgery care-protocol, involving bedside physiotherapy and gait-retraining up to hospital-discharge (4–5 days after TKR) with hospital-based physiotherapists. After discharge, they were encouraged to continue the same exercise protocol and gait practice at home. A 12-week programme of self-managed, home-based exercises designed to enhance functional capabilities (modified from Piva et al ) was initiated at ~ 2 weeks after surgery (range 15–20 days). Additional file 1: Appendix 1 reproduced by an associated parent RCT study  details the delivery of SMT and control programmes. At the programme’s inception, an experienced physiotherapist (principal investigator) conducted an educational training session with patients in order to teach the key features and characteristics of safe delivery of the exercise programme that they would follow at home. Patients’ training programmes were further prescribed using a standardised illustrated guidebook of 14 exercises to regulate exercise-specific dosages. From week 3 to week 8, patients undertook 5 exercise sessions per week. Sessions increased progressively in duration from 35 to 45 min, incorporating progressively longer durations of walking from 10 to 20 min. Weeks 9 to 14 required patients to complete 45-min sessions of exercise, 3 times per week. The level of difficulty was progressed by adjusting exercise intensity to calibrate with weekly changes in each patient’s strength capability. Clinical oversight involved patients freely reporting effusion or discomfort and clarifying the delivery (accuracy, dose or safety) of the home-based exercises by telephone and by voluntary attendance ad libitum, for patients within both groups, within weekly scheduled clinical practical sessions. Patients’ compliance with the prescribed intensity, duration and frequency of exercise was verified by 7-day recall activity diaries. Experimental and control groups were prescribed identical procedures, number of exercises and total programme’ duration.
Experimental group: sensori-motor training (SMT)
The experimental group undertook exercises that focused predominantly on enhancing sensori-motor functioning of patients. The SMT exercises included novel formulations of agility and perturbation training techniques [15, 16, 18, 19] which substituted for a proportion of training (50% – 7/14 exercises) within usual practice. Since the sensori-motor exercises were instructed to be delivered within a home-based environment, no specialised equipment was required. Exercise challenges and progression was achieved by using regular pillows to substitute for unstable surfaces, plastic cups for overcoming obstacles, and strategies such as bipedal to monopedal stance and eyes open to eyes closed in order to increase difficulty in maintaining or achieving balance.
Usual care exercise sessions involved strengthening, stretching, and task-oriented functional exercises of the lower-extremity as described in other studies [15, 16, 34]. The content of the usual care programme was pragmatically adjusted to match the current trends of TKR rehabilitation .
The selected indices included measures of neuromuscular performance capability (muscle force and activation), muscle size and knee ROM. Randomly-ordered assessments of outcome data were collected by the principal investigator at pre-surgery, at 8 weeks post-surgery and at the study’s primary endpoint, 14 weeks post-surgery.
Neuromuscular performance capability
The knee extensors’ peak force (PF), measured in Newtons, was the study’s primary outcome and tested on an isokinetic dynamometer (Primus RS BTE, The Technology of Human Performance, USA). Muscle peak force (PF) was assessed during a maximum voluntary isometric contraction (MVIC) using a protocol adapted from Gleeson et al, . The latter was recorded as the mean peak response from three intra-session muscle contractions. The reliability and reproducibility of assessing peak force has been verified [36,37,38].
Neuromuscular performance capability was assessed indirectly by surface EMG [38, 39] during MVICs (50 ms epoch, rectus femoris; Spike 2, version 5.16, Cambridge Electronics Design Ltd., UK). The EMG activity from the rectus femoris (RF) was recorded concomitantly with participants’ performance of static PF, using bipolar rectangular surface electrodes (self-adhesive, Ag/AgCl; 10 mm diameter; Unilect, UK). The root mean square (RMS) and peak amplitude, were used to describe the time-domain information of the EMG signal [40, 41], using commercially available software (Spike 2, version 5.16, Cambridge Electronics Design Ltd., UK). Normalisation of the EMG signal’s peak amplitude and RMS [42, 43] to the baseline MVIC (100%) facilitated inter-group comparisons over time. Reliability and validity of assessing EMG during MVICs has been verified by McKenzie et al. .
Active range of flexion and extension movement (ROM) of the operated and non-operated knees was assessed by goniometry [43, 44] using the best of three attempts. The ICCs for flexion ROM has been found as 0.96, whilst for extension as 0.81 in a supine position [45, 46].
Muscle size alterations of the RF were examined throughout the study. Real time ultrasound image was captured at 7.1 MHz with a 55 mm linear probe (BK, mini focus, USA). Imaging was conducted in a seated position with the knee in 60° of flexion. Measurement of CSARF was undertaken using the method described by Bruin et al, . Two images were captured with the muscle in maximum relaxation, and subsequently, another two images with the muscle during maximum voluntary isometric contraction (at the end of a 5 s contraction). Analysis of images was performed with ‘Image J’ software (https://imagej.nih.gov/ij/). Intra-rater reliability ICCs of 0.87 to 0.99 have been reported in studies measuring CSARF with a corresponding coefficient of variation (%) within the range from 3.5 to 8.9% [48, 49].
The effects of the SMT were assessed per protocol for each outcome measure using separate factorial ANOVAs involving group (experimental; control) by leg (non-operated; operated) and by test occasion (pre-surgery; 8 weeks post-surgery; 14 weeks post-surgery) comparisons, with repeated measures on the latter two factors. Assumptions underpinning the use of ANOVA were assessed and corrections used Greenhouse-Geisser (GG), where appropriate. For outcomes that had focused on bilateral limb capabilities, group (experimental; control) by test occasion (pre-surgery; 8 weeks post-surgery; 14 weeks post-surgery) interactions were assessed, with repeated measures on the latter factor.
Effect size (ES; Cohen’s d) was calculated using pooled standard deviations . A sample size of 30 participants per group had been computed a priori to discriminate moderate inter-group effects  at the study’s primary endpoint (14 weeks post-surgery) for its primary outcome (PF). Statistical significance was accepted at p < 0.05. Analyses used the Statistical Package for Social Sciences (SPSS; v. 16.0).