Subjects consenting to participate in this study were randomly allocated to receive either yoga (intervention) or supportive therapy plus exercise therapy prior to their surgery using random numbers generated by a random number table. Randomization was performed using opaque envelopes with group assignments, which were opened sequentially in the order of assignment during recruitment with names and registration numbers written on their covers. Yoga being a popular intervention, it was not possible to mask the yoga intervention from the subjects although they were initially told that they would be participating in a postoperative rehabilitation program. However, the investigators (treating surgical oncologists) were blind to the intervention.

Stress was assessed using standard self-report questionnaires such as the State Trait Anxiety Inventory (STAI)[42] for anxiety and Beck’s Depression Inventory (BDI)[43] for depression.
STAI consists of separate self-report scales for measuring two distinct anxiety concepts: state anxiety and trait anxiety, each having twenty statements. The respondents are required to rate themselves on a four point scale: ‘not at all to very much so’ on various anxiety-related symptoms which they experience. This has been used widely in earlier studies on cancer populations and with a concurrent validity ranging from 0.75 to 0.80 with other tests.
Beck’s Depression Inventory is a self-report measure used to assess behavioral manifestations of depression. The inventory is composed of 21 categories of symptoms and attitudes, each with a graded series of 4–5 evaluation statements ranked to indicate the range of severity of symptoms from neutral to maximal severity. This instrument has a reliability of 0.48–0.86 and validity of 0.67 with the Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnostic criteria for depression.

Quality of life of study participants was ascertained using the Functional Living Index of Cancer (FLIC).[44] This scale is a self-administered measure of the global quality of life for cancer patients having a high correlation (0.44–0.75) with other scales. A subjective symptom checklist was developed during the pilot phase to assess stress, treatment-related side effects, problems with sexuality and image and relevant psychological and somatic symptoms related to breast cancer. The checklist consisted of 31 such items, each evaluated on two dimensions, the severity graded from “none to very severe (0–4)” and distress from “not at all to very much (0–4)”. This scale measured the total number of symptoms experienced, total and mean severity and distress score and was evaluated previously in a similar breast cancer population.[45]

Blood samples were collected in three separate vaccutainers: 7 ml of each blood sample was collected in a heparinized vacuette for the separation of peripheral
blood lymphocytes, 2 ml in a sodium citrate vacuette for plasma and 2 ml in a plain vacuette for serum. Blood cell separation: Peripheral Blood Lymphocytes
were isolated from 7 ml of heparinized blood using ficoll gradients (Histopaque 1077 R, Sigma Inc.). The isolated lymphocytes were then washed in phosphate-buffered saline (PBS), treated with 4% glacial acetic acid to lyse the red blood cells (RBCs) and again washed with PBS. The lymphocytes were then counted on a hemocytometer/ coulter counter and diluted at concentration of 50,000 cells/ml in Tris-buffered saline (TBS). Thereafter, the cell suspension was centrifuged and the cells fixed on the slides using an acetone-methanol medium. The NK Cell count was determined by immunohistochemistry using the standard Alkaline Phosphatase Anti-Alkaline Phosphatase technique (APAAP). Briefly, the cells fixed on the slides were treated with an anti-CD56 antibody
(DAKO Cytomation) and then with a secondary antibody conjugated with APAAP. Cells with the CD56 surface antigen bound to these antibodies and took up the fast red stain giving a red glow over the periphery. The cells were counterstained with hematoxylin. Only those cells which took up the fast red stain and had pink to red stained edges were counted as CD56-positive cells as against others, which took up only hematoxylin and appeared blue. The cells were counted in two hundred fields and the mean percentage of CD56-positive cells per hundred fields extrapolated. Serum Immunoglobulins (IgG, IgM
and IgA) were assessed using an Immunoturbidometry assay with an autoanalyzer.

The intervention group received an “integrated yoga program” and the control group received “supportive counseling and exercise rehabilitation.” While the
goals of the yoga intervention were stress reduction and improvement in shoulder mobility, the goals of the control intervention were to reinforce social support
and prevent shoulder restriction. The yoga intervention consisted of a set of breathing exercises, pranayama (voluntarily regulated nostril breathing) and yogic
relaxation techniques. These practices were based on the principles of attention diversion and relaxation to cope with day-to-day stressful experiences. These sessions were administered by an instructor at the subjects’ bedside prior to surgery and during their postoperative recuperation in the hospital. Following their discharge, subjects were asked to practise at home for the next four weeks. Subjects were also provided audiotapes of an instructor’s voice to help them practise at home so that a familiar voice could be heard on the cassette. Their practice was monitored on a day-to-day basis by their instructor through telephone calls once a week. Subjects were encouraged to maintain a daily log listing the yoga practices done, use of audio- visual aids for practice, duration of practice, experience of distressing symptoms, intake of medication and diet history.
Supportive counseling sessions as control intervention included two important components: “i) education and reinforcing social support and ii) shoulder exercise for postoperative rehabilitation.” We chose to have this as a control intervention to prevent shoulder restriction and to control for any nonspecific effects of the yoga program that may be associated with adjustment such as attention, support and a sense of control. Moreover, these interventions have been shown to hasten recovery from surgery in earlier studies.[22,24] Subjects and their caretakers were invited to participate in an introductory session lasting 60 minutes before surgery where they were given information about surgery and management of its related side effects, taught shoulder exercises and mobilization by the physiotherapist and provided information about a variety of common questions. The interventions were imparted at the patient’s bedside and subjects were asked to perform the shoulder exercises at their home following their discharge until they receive adjuvant therapy. Both groups received four such in-person sessions of intervention during their hospital stay.

Data was analyzed using SPSS 10.0 for windows. Data was tested for normality and homogeneity. An analysis of covariance was done on all assessments 3–4 weeks postsurgery using their respective baseline (presurgery) measures as covariates. The within-groups effects was analyzed using a paired t test.