To determine the efficacy of an intensive upper limb exercise program for women soon after breast cancer surgery, we will conduct a multi-center, single-blind randomised controlled trial. Women will be recruited from 3 large hospitals with breast cancer oncology services in Sydney, Australia. Breast Cancer Nurses will inform potential participants about the study when they visit the patient after surgery. To provide sufficient time for the wound to heal, women will enter the study 4 weeks after surgery.
A record will be kept of the number of invitations to participate, the number of potential participants who volunteer to participate, and the number of screened patients who are ineligible and the reason for their ineligibility.
Ethical approval has been received from the participating hospitals as well as from The University of Sydney. All participants will provide written informed consent.
Women will be included in the study if they: have undergone a breast cancer intervention that included surgery to the axilla; can attend for treatment and follow-up; understand English and consent to participate and be randomised to either treatment group. Women scheduled for adjunct treatments including radiotherapy, chemotherapy and breast reconstruction will be included.
Women will be excluded if they have not had surgery to the axilla or have undergone bilateral operations, present with infection at the baseline measurement or have received previous treatment for breast cancer or have metastatic disease. In addition, women will be excluded if they have sustained a fracture, undergone surgery in the upper limbs, or suffered any neurological deficit or other injury to either upper limb that may interfere with the test procedures. Lastly, women will be excluded if they undergo breast reconstruction and as a consequence, their specialist has indicated that they are not to participate.
Women will receive the same care as is now provided in the three participating hospitals. Women are typically admitted on the day of surgery and are discharged 2 to 7 days after surgery. While in hospital, they are seen by a physiotherapist who reviews gentle shoulder exercises which are outlined in a pamphlet. The active assisted shoulder exercises are for shoulder protraction, elevation and abduction. In addition, women are seen by an occupational therapist who discusses issues related to prevention of lymphedema. Women are not followed by either the physiotherapist or the occupational therapist after discharge from the hospital.
Women in the Usual Care group will not be given a home program of exercises, apart from those described in the pamphlet. To control for attention from the research assistant, they will meet with the research assistant fortnightly at which time their arm will be assessed with multifrequency bioimpedance analysis  for the development of lymphedema, but no specific advice will be given about exercise and management.
Commencing Week 4 following surgery, women will attend 8 weekly exercise sessions comprising stretching exercises and resistance training, and will commence a daily home program of similar exercises. We have used these exercises in our pilot study , and they are well tolerated.
Each week, there will be a supervised weekly session at which women will perform the shoulder exercises targeting the muscles and shoulder movements likely to be at risk. Stretches will consist of prolonged holds of at least 5 minutes duration. The stretches are performed in supine and are undertaken for the shoulder extensors, and for pectoralis major and minor muscles. For women unable to achieve >90 deg forward flexion, shoulder extension and abduction stretches will be performed whilst seated. For both stretches, the arm will be elevated in the appropriate direction and supported on pillows; the aim of these stretches is to provide a comfortable stretch in the region of the participant's axilla. Each of these stretches will also be held initially for 5 minutes and progressed up to 20 min. Once shoulder flexion >90 deg is achieved, the participant will undertake the stretches in supine. The muscles targeted for resistance training include the shoulder flexors, abductors, external rotators and horizontal flexors. For strength training, the resistance will be determined by the Borg Effort Scale  and provided by free weights. Each week, the exercises will be monitored, and progressed as required.
In addition to the weekly supervised sessions, women will perform a daily home program. The program will consist of i) a daily stretching program and ii) a strengthening program undertaken on alternate days. The stretching program will be the same as that undertaken at the weekly supervised session. For the strengthening program, the same muscles will be trained as in the weekly supervised sessions but rather than free weights, the resistance will be provided by theraband. Participants will be instructed to perform two sets of 8 – 12 repetitions for each exercise. Participants will be instructed to work at a target of 15 ('Hard') on the Borg Effort Scale 
At the conclusion of the program, women will be encouraged to continue the home program. A compliance diary will be provided to record the number of times each exercise was performed each day. In addition, women in both groups will be asked to record any other treatments that they have received the dose, and the effect.
All measurements will be undertaken by a research assistant blinded to the group to which women were allocated. Variables will be measured before randomization, at 8 weeks (the end of the intervention phase) and then at 6 months to determine the effect of the intervention on physical and psychosocial outcomes. In addition, information about the breast cancer treatment regimen, pathology report, age, weight, and previous shoulder pathology will be obtained from the patient's medical record at the baseline assessment. Occurrence of side-effects and adverse events will be evaluated at each measurement occasion.
The primary outcome measure for treatment efficacy is arm symptoms on the operated side, derived from 3 items of the Breast Module (BR23). The BR23 is a 23-item survey specific to women with breast cancer. The 3 selected items ask whether in the past week the woman has had pain in her arm or shoulder, a swollen arm or hand, and whether it was difficult to raise or move her arm sideways. For each question, there are 4 possible responses, ranging from 'not at all' to 'very much'. The raw scores for these questions are summed and then subjected to linear transformation to provide an overall score, with a maximum of 100. A high score indicates a high level of symptoms. The primary endpoint for the trial is change in primary outcome from baseline to 6 months.
Secondary outcomes include other aspects of quality of life and measures of physical impairment.
Quality of life will be measured using the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Version 3 (QLQ-C30) and the Breast Module (BR23). These surveys are specific to cancer patients and comprise a global health status/quality of life scale, as well as functional and symptom scales. These questionnaires are reliable, valid and sensitive to change and able to distinguish patients in different disease stages and with different performance status [17–19].
Shoulder range of motion will be assessed using an inclinometer and a standardised protocol will be followed to minimise compensatory movements. Passive forward flexion, external rotation and horizontal abduction at the shoulder will be assessed in a supine position and with gravity as the standardised force. Active abduction will be assessed in the sitting position. Each arm will be measured separately. Pain at the end of the passive range of each movement will be measured using a standardised 10 cm visual analogue scale.
Maximal isometric shoulder strength will be measured objectively with a dynamometer using a standardised protocol. The arm will be positioned at 90 degrees elevation to assess forward flexion, abduction, and horizontal abduction and adduction. Movement in these planes has been selected as it has been shown to be weak following breast cancer surgery [20, 21]. For each direction, women will perform 3 maximum voluntary contractions, and the highest of the three attempts will be recorded. Each arm will be measured separately.
The presence of lymphedema will be measured non-invasively using multiple frequency bio-impedance analysis. This analysis identifies early changes in tissue fluid accumulation  and is substantially more sensitive than total limb volume measurements in detecting changes in extracellular fluid volume. It has an extremely low (<1 %) false-positive rate for detecting lymphedema . In addition, circumference measures at multiple sites on each arm will be obtained using a standardised measurement technique.
All women will be requested to record use of resources (number and type of health care contacts, medication use, adjunct treatments) during the trial period and the effect of these treatments.
The occurrence of lymphedema, regardless of treatment group allocation, is considered an 'adverse' event. If lymphedema develops, women will be referred to the Breast Cancer Nurse for referral to an appropriate specialist.
Sample size calculations
We will recruit 180 women, 90 to each of the two treatment groups. This will give 80% power to detect as significant at the 2-sided 5% significance level a difference of 12 points (out of 100) for arm symptoms derived from the BR23, based on our pilot data standard deviation of 25. In addition, we have allowed for 20% mortality and loss to follow-up. Based on our pilot data, we will also be able to detect at least 10° difference in range of motion at the shoulder and 5 N difference in shoulder strength.
Treatment group will be coded to enable blinded analysis, which will be by "intention-to-treat". For the primary outcome measure, the change from baseline score for arm symptoms on the operated side derived from the BR23 will be determined. The mean change will be compared between the two groups using Student's two-sample t-test.
Our secondary analyses will examine the effect of adjusting for the stratifying factor (type of surgery) or any important prognostic factors that show a clinically important imbalance between groups. This will be done using multiple linear regression, with the baseline arm symptoms score as a covariate and the final arm symptoms score as the outcome variable.
For measures of strength, range of motion, and lymphedema, the difference between the operated and non-operated side will be determined. These differences at 6 months will be compared using a two-sample t-test or distribution-free equivalent, as appropriate. For other aspects of the quality of life score, the mean changes from baseline will be compared between the two groups using Student's two-sample t-test. As for the primary outcome, multiple regression analyses will be used for all secondary variables to adjust for covariates, including baseline differences.