For shoulder abduction, the starting position was sitting (as for flexion) with the arm at the side, the shoulder in external rotation and the elbow extended. The participant was asked to abduct the arm while maintaining elbow extension. For shoulder external rotation, the starting position was supine JQ1 mouse with the arms at the side and supported by the bed, the affected elbow flexed to 90°, and the hand in a loose

fist. The participant was asked to externally rotate the arm, keeping the elbow on the bed and leading with the dorsum of the hand. Anatomical surface markings were made to guide placement of the inclinometer. After a practice movement, each range of motion was repeated twice and the higher measure recorded. Shoulder muscle strength was measured using a handheld dynamometerb. Strength measurements were taken for flexion, abduction, extension, and internal rotation as these are some of the actions of the muscles divided during open thoracotomy. All measurements were taken with the

participant sitting (as above) with the affected arm one gripped fist’s width (at the lower end of the humerus) from the side of the body, the elbow flexed to 90° and the forearm in neutral rotation. Anatomical surface markings were again used to guide dynamometer placement. Resistance was applied against the direction of shoulder movement for 3–5 sec using the ‘make’ rather than ‘break’ technique (Stratford and Balsor 1994). Standard instructions

and verbal Capmatinib encouragement were given. After one practice contraction, each movement was measured 3 times with 1 min between measurements and the highest value was recorded. Shoulder function was measured using the Shoulder, crotamiton Pain and Disability Index (Roach et al 1991), which is a selfrated questionnaire designed to measure shoulder pain and disability. Although this questionnaire has not been used previously in a post-thoracotomy population, its validity, reliability, responsiveness, and ease of completion have been demonstrated in patients with primary shoulder disorders (Bot et al 2004, Paul et al 2004). It has 13 items divided into two subscales (pain and disability). All items were rated on a visual analogue scale anchored with ‘No pain’ and ‘Worst pain imaginable’ for pain, and ‘No difficulty’ and ‘So difficult it requires help’ for disability. Scores for each subscale range 0–100, with higher scores indicating greater pain or disability. A total score (0–100) was calculated by averaging the two subscale scores. If more than two items of a subscale were not answered, no subscale or total score could be calculated. Health-related quality of life was self-rated using the Medical Outcomes Study Short Form 36-item version 2 (New Zealand) survey.