The purpose of this project was to develop the M. D. Anderson Symptom Inventory (MDASI), a brief measure of the severity and impact of cancer-related symptoms. A list of symptoms was generated from symptom inventories and by panels of clinicians. Twenty-six symptoms and 6 interference items were rated by a validation sample of 527 outpatients, a sample of 30 inpatients from the blood and bone marrow transplantation service, and a cross-validation sample of 113 outpatients. Clinical judgment and statistical techniques were used to reduce the number of symptoms. Reliability, validity, and sensitivity of the MDASI were examined. Cluster analysis, best subset analysis, and clinical judgment reduced the number of symptoms to a "core" list of 13 that accounted for 64% of the variance in symptom distress. Factor analysis demonstrated a similar pattern in both outpatient samples, and two symptom factors and the interference scale were reliable. Expected differences in symptom pattern and severity were found between patients with "good" versus "poor" performance status and between patients in active therapy and patients who were seen for follow-up. Patients rated fatigue-related symptoms as the most severe. Groups of patients classified by disease or treatment had severe symptoms that were not on the "core" list. The core items of the MDASI accounted for the majority of symptom distress reported by cancer patients in active treatment and those who were followed after treatment. The MDASI should prove useful for symptom surveys, clinical trials, and patient monitoring, and its format should allow Internet or telephone administration.
