To be eligible for enrolment subjects must be adults admitted to Auckland City Hospital, New Zealand to undergo elective lung resection via open thoracotomy. They must understand written and spoken English and give informed written consent. Subjects will be excluded if they are unable or unwilling to comply with treatment, have tumour invasion into the chest wall or brachial plexus, present with a PPC prior to randomisation on postoperative day one, remain ventilated for longer than 24 hours following surgery or have received physiotherapy for shoulder problems or respiratory problems within two weeks prior to admission for surgery. Subjects with postoperative neurological or mobility complications requiring comprehensive physiotherapy input to progress towards discharge (defined as being more than two physiotherapy interventions), will be provided with the physiotherapy as required, remain in the group to which they were allocated and analysed using intention to treat principles. Subjects, who live within a 60 km radius of Auckland Hospital and can attend outpatient appointments, will form a subgroup that, in addition to all other measures, will attend for follow up shoulder ROM and muscle strength measures.

Sample size was calculated for the primary outcome measure of PPC with a type I error rate of .05 and a type II error rate of 0.20 (80% power). Estimates of the incidence of PPCs vary throughout the literature dependent upon criteria used for diagnosis but best estimates suggest a PPC rate (after thoracotomy and amenable to physiotherapy interventions) of between 10–25% occurs in this population with routine physiotherapy care 414, Assuming a PPC rate of 20% in the Control Group and 5% in the Treatment Group (thus a 15% difference in PPCs), 84 subjects in each group are required. Allowing for an estimated loss to follow up rate of 9% (based on similar studies) a sample size of 184 subjects (92 per group) is necessary to demonstrate sufficient power for the primary outcome. Because there are no randomised trials that include a control group on which to accurately base this sample size calculation, an interim analysis will be undertaken when total n = 80.

Subjects eligible to take part are screened from the thoracic surgical operating list by the study investigators (JR & KN) on admission to hospital. If eligible to take part the study investigators approach the subject to explain the study and gain informed consent. It is explained that they may withdraw from the study at any time without jeopardising their future care. On day one postoperatively, prior to any physiotherapy interventions and following extubation, subjects are given a consecutive study number and randomly allocated to the Treatment Group or Control Group by an independent physiotherapist. The ward physiotherapist is then told of the group allocation which is recorded and the envelope stored. Group allocation is assigned using a computer generated random allocation (from http://www.randomization.com) and individual group assignment is kept in sealed opaque envelopes.

In the event of more than one thoracic surgical procedure per day randomisation is performed according to order of admission onto the postoperative intensive care/surgical ward. To avoid influencing postoperative regimens, subjects taking part in the research trial are, where possible, allocated to separate rooms during their hospital stay.

In the subgroup who meet the inclusion criteria for shoulder ROM and muscle testing, a computer generated randomised order for performing the ROM and muscle strength measures is determined before preoperative measures and adhered to throughout the study. The randomisation order is generated from http://www.randomization.com.

After consent all subjects have preoperative demographic data recorded in order to compare baseline measures between groups (see Table 1), and are given a physiotherapy written information sheet by a study investigator. This broadly explains the need for postoperative breathing and coughing exercises, early ambulation, and shoulder mobility exercises. It outlines breathing exercises, huffing/coughing and the importance and progression of ambulation postoperatively. It gives general advice about shoulder and arm use in the postoperative period. No other form of preoperative physiotherapy education and treatment is provided to subjects in either group.

Both Treatment and Control Group subjects will receive standard nursing and medical care according to the clinical pathway in use at the unit where the research is being undertaken.

Subjects in the Treatment Group will receive postoperative physiotherapy from the attending ward physiotherapist as per a written protocol on a daily basis until discharge. This includes a minimum predetermined amount of: deep breathing and coughing exercises, assistance with ambulation, and a progressive shoulder and thoracic cage mobility exercise programme (see Figure 2). Subjects will be encouraged by the attending physiotherapist to practice the exercises outside physiotherapy treatment times but this practice will not be quantitatively measured. Upon discharge from hospital Treatment Group subjects will receive a discharge exercise and advice sheet with an exercise diary explained by the attending ward physiotherapist.

Control Group subjects will receive no postoperative physiotherapy. They may be verbally encouraged by attending nursing staff to practice the exercises outlined on their preoperative physiotherapy information sheet throughout their postoperative stay.

Will be diagnosed on a daily basis until hospital discharge using the diagnostic tool shown in Table 3. The data will be collected by an attending physiotherapist and phoned to a blinded assessor who determines the diagnosis of a PPC.

Length (number of days) of postoperative stay will be measured from data held on the hospital database.

Will be measured using the Shoulder Pain and Disability Index (SPADI) preoperatively, on discharge from hospital, one month and three months postoperatively. This is a self reported shoulder specific index developed by Roach et al. (1991) 26 which consists of 13 items in two subscales: pain (5 items) and disability (8 items). Each item is measured on an 11 point Likert scale where 0 = "no pain/no difficulty" and 11 = "worst imaginable pain/so difficult it required help" on each subscale respectively. The SPADI is scored by adding then averaging the two subscales to determine a score out of 100. A higher score means greater pain/disability. The SPADI has been demonstrated to detect change in subject's status over time, have good test-retest reliability, internal consistency and good criterion and construct validity although not within the thoracic surgical population 262728.

In a subgroup of patients living within a 60 km radius of Auckland Hospital, active ROM will be measured preoperatively, on discharge from hospital, one month and three months postoperatively by a blinded assessor using digital inclinometry via a Plurimeter-V inclinometer. ROM will be measured in those muscle groups which are divided during surgery thus potentially affecting range of movement. The measures are:

a. Total shoulder flexion (the movement includes scapular motion)

b. Total shoulder elevation through abduction (the movement includes scapular motion)

c. Glenohumeral external rotation

All ROM measures (except external rotation) are taken with the subject fully upright in a standardised chair with the opposite forearm resting on a fixed table to eliminate trunk movement. Sitting has been chosen as the starting position for most measures due to the possibility of subjects with respiratory disorders being unable to tolerate supine lying. The position of the inclinometer is established by measurement from a clear surface marking (e.g. 1 cm above lateral epicondyle).

▪ Shoulder flexion is measured with the arm at the side, elbow extended, shoulder in neutral rotation with palm facing the thigh and thumb facing forwards. The subject actively flexes the arm with the thumb leading throughout, maintaining elbow extension.

▪ Shoulder elevation is measured with the arm at the side, elbow extended, arm in external rotation with the palm facing forward. The subject actively elevates the arm leading with the thumb maintaining elbow extension.

▪ Shoulder external rotation is measured in supine lying (on a standardised bed) with the arm at side, humerus supported by the bed, elbow flexed to 90°, forearm in the mid prone position, hand in fisted position, pointing towards the ceiling. The subject actively externally rotates the arm maintaining the elbow on the bed and leading with the dorsum of the hand.

For all movements care is taken to avoid trunk assisted movements. On each occasion the arm is returned to 0° and each movement is repeated twice after one practice measure. These measures, where possible, follow a standardised protocol that has been shown to have acceptable intra and interrater reliability for use within a randomised controlled trial 29. As well as recording ROM subjects report maximum pain during each movement on a 0 – 10 verbal rating scale. The maximum reading for pain for each movement is recorded.

In a subgroup of patients living within a 60 km radius of Auckland Hospital, muscle strength will be measured preoperatively, one month and three months postoperatively by a blinded assessor using a handheld manual muscle tester (Lafayette instruments) which has been shown to have moderate to high interrater and intrarater reliability 30. Measurements will be made of those of muscle groups that have been divided during surgery and thus may affect muscle strength. The measures are:

a. Shoulder flexion

b. Shoulder abduction

c. Shoulder internal rotation

d. Shoulder extension

All measures are taken in the same position as described for ROM testing with the position of the dynamometer established in the same way using the easily identifiable surface markings of 1 cm above the elbow crease (for flexion), 1 cm above the lateral epicondyle (for abduction), 1 cm above the olecranon (for extension) or the skin crease (usually middle) between the radial styloid process and the head of the ulna (for internal rotation). The starting position is with the arm at the side with one gripped fist width between the distal end of the humerus and chest wall, the elbow flexed to 90° and the forearm in the mid prone position. This position was chosen due to the potential for subjects to be unable to commence muscle strength measures in 90° or 45° glenohumeral abduction as recommended by some authors 3132. Resistance is applied against the direction of shoulder movement for three to five seconds. The "make" (rather than "break") technique is used requiring the examiner to resist a maximal voluntary contraction by the subject which is essentially an isometric contraction and has been shown to have high interrater and intrarater reliability even in inexperienced examiners 33.

Standardised instructions and verbal encouragement is given and, after one practice contraction, each movement is measured three times with one minute between measures. Subjects report maximum pain during each movement on a 0 – 10 verbal rating scale. In keeping with other authors the maximum intensity of pain for each movement will be recorded 333435.

Table 1 gives subject and operative data that will be collected and used to compare characteristics that could influence the outcome of the study. These data will be collected by the study investigators from subjects' medical records. In addition, records of any postoperative radiotherapy, chemotherapy and information regarding any additional shoulder treatments from health care professionals will be sought.