Study design
This study was designed as an assessor-blinded pragmatic randomised controlled trial (RCT). The participants were randomly allocated to either a 12-week mainly self-guided resistance-training program or a control group that did not change their usual lifestyle habits (Fig. 1). The trial was prospectively registered as ISRCTN90224545 (www.isrctn.com, BMC, Springer Nature) and ethical approval was obtained by the Ethics Committee of Canton Berne, Switzerland (No. 222/12). All participants gave written informed consent and the study was conducted according to the Declaration of Helsinki. This paper follows the CONSORT statement [27] and the intervention is described according to the TIDieR checklist [28].
Participants
Women aged between 20 and 40 years with GJH were eligible for the study if they scored at least 6/9 points on the Beighton score, and right knee hyperextension was mandatory. The higher cut-off was chosen based on more recent publications [29, 30] and knee hyperextension was required because training focused on the lower limb and assessments were performed mainly on the right side. As further inclusion criteria, participants needed to have a body mass index between 18 and 30 kg/m2 and be able to understand German questionnaires.
Excluded were women who had had surgery of the lower extremities or lumbar spine in the last two years, because this might affect their current condition and the ability to perform strength training. In addition, women with acute pain in the back or lower extremities were excluded. Women who regularly undertook more than four hours per week of sport activities were excluded to ensure better homogeneity of the groups in terms of muscle strength and training experience. Pregnant women and those less than one year after delivery were excluded, since changes in the hormonal state may affect the outcome of strength training [31]. Finally, women with known inherited diseases of the connective tissue, mainly Marfan syndrome and Ehlers-Danlos syndromes except hypermobility type and Osteogenesis imperfecta, were excluded. A formal diagnosis of Ehler-Danlos syndrome, hypermobility type, was not a reason for exclusion. Note that the criteria for this study were defined in 2012 and thus not based on the new 2017 nosology for EDS and HSD [7, 10].
Recruitment, inclusion and allocation
Participants were mainly recruited from an existing database of previous studies [17, 32] and via the staff of Bern University Hospital and students of the Bern University of Applied Sciences, Department of Health, Switzerland. Furthermore, announcements in the local newspapers were published to recruit participants. The recruitment period was between August 2013 and November 2015 and the recruitment, as well as all the measurements and training sessions took place at Bern University Hospital, in Bern, Switzerland.
Interested participants were informed by phone before their first appointment and received information sheets by mail. After signing the informed consent, inclusion and exclusion criteria were confirmed face-to-face by one physiotherapist (CM), with more than 12 years of clinical experience. The participants performed a standard pregnancy test themselves using a urine sample. For the Beighton score the test movements were a.) Hyperextension of elbow more than 10°, b.) Hyperextension of knee more than 10°, c.) Ability to touch the floor with the palms of the hands, keeping the knees fully extended, d.) At least 90° dorsiflexion of 5th metacarpophalangeal joint, and e.) Ability to touch the inner side of the forearm with the thumb [1]. All items, except c.), were tested bilaterally, resulting in a possible total score of 9 points.
The range of motion of the right knee in flexion and extension was measured with a standard inclinometer while lying supine. Additional measures included body weight, body height, arm span, and arm and leg length on both sides. Finally, anamnestic checking of the Brighton criteria [8] was done by semi-structured interview by the same experienced physiotherapist (CM). The Brighton criteria were recorded for a clearer description of the study group and to allow for potential analysis of the effects or the feasibility of the training for women with and without JHS.
After inclusion, the participants were randomised based on an independently computer-generated randomisation list either to the resistance training or to the control group. After recording of the personal and anamnestic data in the database and confirming inclusion, the physiotherapist responsible for the inclusion accessed the allocation electronically to ensure concealment. The randomisation list was kept secret from the assessor and statistician until all analyses had been performed.
Intervention
The intervention for the training group was a mainly self-guided 12-week resistance training program to address hypertrophy, focusing on the muscles of the lower extremities and the trunk. Two training sessions of about 50 min were performed each week in the medical training centre of the Berne University Hospital, resulting in 24 training sessions.
The strength training program was developed based on recommendations of the American College of Sports Medicine [21, 33]. The details of the training program are provided in a supporting information file (Luder-G_S1-file_training-program-intervention.pdf). Resistance was mainly set at 80% of the one repetition maximum and three series with 12 repetitions for each side were performed. Four experienced physiotherapists gave the instructions on a 1:1 basis for the training program. All of them regularly instructed patients and healthy persons in the medical fitness and were specifically instructed for this project. In the first week, a one-hour session was dedicated to basic instructions and determining the one repetition maximum. In week three a half-hour session aimed to reassess exercise performance and adapt the resistance. Finally, in week six an additional half-hour session was spent monitoring the proper practice. All other training sessions were performed individually and not directly supervised; however, a responsible physiotherapist was always available in the training room for questions and support. Participants were encouraged to increase the resistance gradually whenever more than 12 repetitions were possible. If pain or discomfort occurred because of the exercise, the women could always refer to the physiotherapist in charge. During the instruction sessions possible adaptations to pain or muscular problems were discussed and suggested, e.g. reduction of resistance, increased rest time between series or a reduction to one or two training series instead of three.
The participants in the control group were advised not to change their lifestyle habits for the next 12 weeks. After the post-measurement, all participants of the control group were offered to participate in the same structured training program as the intervention group.
Adherence and problems triggered by training
A secondary aim of the project was to assess the feasibility of the resistance training for women with GJH. Thus, the participants recorded the number of training sessions and the exercises performed with all details in a diary. Additionally, personal notes and experiences were documented, e.g. pain, discomfort, or reasons for reduced performance. Performance of more than 80% of the training sessions was deemed as acceptable adherence.
Furthermore, during the training pain and disability in the daily life of the participants were monitored with a face-validated questionnaire using 5-point Likert-scales. The first two questions asked for details of disability or pain during and after the training. Three additional questions asked for other pain or impairments during the week. For every question, the location of the problems and additional information could be provided. The questionnaire was developed based on a previous study [34].
Outcome assessments
GJH may affect an individual in several dimensions of life, as defined in the International Classification of Functioning, Disability and Health (ICF) [35]. The outcome assessments in this study aimed to evaluate the effects of resistance training in various dimensions of the ICF: muscle strength and properties as body structures, muscle activity during stair climbing in terms of function; and a set of patient reported questionnaires regarding activities and participation to detect impairments and restrictions in daily life. A detailed description of all assessments and the respective analyses is provided in a supporting information file (Luder-G_S2-file_outcome-measures.pdf).
In brief, muscle strength was measured as maximum isometric contraction and rate of force development of the knee extensors and knee flexors on a custom-built strength table using a strain gauge. For each muscle group three measurements were performed. Maximum strength and rate of force development as the slope of the force curve between 20 and 80% of maximum were calculated, the values normalised to body mass and the best attempt taken for calculations [18]. The muscle properties of the thigh were measured using peripheral quantitative computer tomography (pQCT) and muscle cross sectional area, and muscle mass and density were calculated as previously described [36]. The cross-sectional area parameters were all calculated in relation to body mass.
During stair climbing on a standard six-step stair-case [16, 37] the ground reaction forces were measured by a force plate embedded in the 3rd step. Simultaneously the muscle activity of the vastus medialis, vastus lateralis, semitendinosus and biceps femoris was measured using electromyography (EMG). Electrode placement and measurement procedure were defined according to the recommendations of SENIAM [38]. The participants had to climb up and down the stair ten times at a comfortable, self-selected speed barefoot and without using the handrail. All ground reaction forces, and electromyography data were processed with custom-made software and six trials were selected for the analyses of stair ascent and descent. Dynamic EMG data were normalised to the corresponding 100% maximum voluntary contraction value and peak and mean muscle activation during stance were calculated. The vertical ground reaction force curves were normalised to body mass and standard parameters for force and time were calculated as means of six trials for each condition [37].
To measure general health the widely used Medical Outcomes Study Short Form 36-Item (SF-36) health survey was completed and the scores calculated according to the standard method [39]. As a measure of disability in daily life the Arthritis Impact Measurement Scales 2 (AIMS-2), originally developed for patients with rheumatoid arthritis [40], was used, since there was no specific questionnaire for persons with GJH at the time of the study preparation. All scores were calculated according to the described methods [41]. Additionally, and based on previous studies a face-validated questionnaire for hypermobility (HM-Q) was used, asking for pain at specific sites and disability in selected daily life activities. All items were rated on a five-point Likert scale and the sum score for the whole questionnaire calculated.
All assessments were performed by a single investigator (GL), blinded to group allocation. The first assessment took place before the training or control period and the second within two weeks after the end of training or the 12-week control period of the control group.
Primary and secondary outcomes
The primary outcome for the effect of resistance training was defined as the increase in muscle strength in relation to body mass, measured as maximum voluntary isometric contraction of the knee flexors and extensors. Secondary outcomes included rate of force development of these knee muscles, the cross-sectional area parameters of the thigh, as well as muscle mass and density. All further variables were analysed in an exploratory manner.
Regarding the feasibility of the training intervention, the percentage of completed training sessions was the main parameter. Additionally, pain and disability in daily life as detected by the weekly questionnaire served as further descriptive outcomes.
Power estimation
As this was the first trial to investigate a resistance training program in individuals with GJH there were only approximate data available for the power calculation. In a previous study [32] a 16.2% higher normalised rate of force development was found for hypermobile women compared to women with normal mobility. With a similar change induced by the training, a hypothetical medium effect size of about 0.6 could be expected.
Derived from this data, a power estimation was performed using G*Power 3.1.5 [42]. For an estimated effect size of 0.6 with the significance level (α) set at p ≤ 0.05, a sample size of 21 in each group (total of 42 subjects) was necessary to achieve a power of 0.8. Since some dropouts were expected, the aim was to enrol 50 women in the study.
Statistical analysis
All analyses were performed on an intention-to-treat basis and included all randomised participants. Missing data was processed by means of imputation based on linear regression per group, except by “last carry forward” for the HM-Q. Missing data for EMG measurements due to technical reasons was not imputed. All statistics were performed on a blinded basis, whereby the randomisation code was only broken after completion of the statistical evaluation.
Descriptive statistics for all clinically relevant parameters are presented. Normal distribution of the data was checked by the Shapiro-Wilk test and Q-Q-plotting to decide whether parametric or non-parametric tests were used for significance testing. At baseline, the comparability between the groups in terms of demographic and prognostic factors was assessed using the independent t-test.
For parametric testing the primary outcomes of the two groups were compared by a mixed analysis of variance (ANOVA) with time as the within subjects factor and group as a between subject factor. To account for possible baseline differences all prognostic variables with significant t-test at baseline between the two groups were additionally introduced as co-variates in the model (ANCOVA). The significance level was Bonferroni-corrected to account for multiple testing (two primary variables) and set at p < 0.025 as the accepted significance level.
For the main parameters, mean differences of change for each group as well as 95% confidence intervals (CI) are presented and the respective effect sizes calculated as partial eta square and converted to Cohens d. The additional outcomes of the secondary analyses were not tested for significance but are reported as descriptive data, with mean difference between pre and post and the respective 95% confidence interval (95% CI). A tendency for a change was noted when the 95% CI for the mean difference did not cross the zero line.