Performance status assessment in cancer patients. An inter-observer variability study

Br J Cancer. 1993 Apr;67(4):773-5. doi: 10.1038/bjc.1993.140.


The ECOG Scale of Performance Status (PS) is widely used to quantify the functional status of cancer patients, and is an important factor determining prognosis in a number of malignant conditions. The PS describes the status of symptoms and functions with respect to ambulatory status and need for care. PS 0 means normal activity, PS 1 means some symptoms, but still near fully ambulatory, PS 2 means less than 50%, and PS 3 means more than 50% of daytime in bed, while PS 4 means completely bedridden. An inter-observer variability study of PS assessment has been carried out to evaluate the non-chance agreement among three oncologists rating 100 consecutive cancer patients. Total unanimity was observed in 40 cases, unanimity between two observers in 53 cases, and total disagreement in seven cases. Kappa statistics reveal the ability of the observers compared to change alone and were used to evaluate non-chance agreement. Overall Kappa was 0.44, (95% confidence limits 0.38-0.51). The Kappa for PS 0 was 0.55 (0.44-0.67), while those for PS 1, 2, 3 and four were 0.48 (0.37-0.60), 0.31 (0.19-0.42), 0.43 (0.32-0.55), and 0.33 (0.33-0.45), respectively. If one observer allocated patients to PS 0-2, then another randomly selected observed placed the patients in the same category with a probability of 0.92. For patients with PS 3-4 the probability that the same category would be chosen was 0.82. Overall, the non-chance agreement between observers was only moderate, when all ECOG Performance Status groups were considered. However, agreement with regard to allocation of patients to PS 0-2 versus 3-4 was high. This is of interest because this cut-off is often used in clinical studies.

Publication types

  • Clinical Trial

MeSH terms

  • Activities of Daily Living
  • Female
  • Humans
  • Male
  • Medical Oncology
  • Neoplasms / physiopathology*
  • Neoplasms / psychology*
  • Observer Variation
  • Quality of Life
  • Reproducibility of Results
  • Self Care
  • Statistics as Topic / methods