Intra and interobserver variability in cancer patients' performance status assessed according to Karnofsky and ECOG scales.

The Karnofsky (K) and ECOG (E) performance status (PS) scales are widely used to evaluate the functional status of cancer patients to determine their eligibility for clinical trials and their prognosis, but knowledge of inter and intraobserver variability of these scales is scarse. We therefore planned a prospective study on 209 consecutive cancer patients to evaluate this critical point. Two independent observers evaluated the KPS and EPS of each patient by interviewing them on the same day. After their interviews the patients were asked to fill in, again on the same day, a self-evaluation scale concerning their ability to perform the routine activities of daily life. The 209 patient self-evaluation scales were presented twice, randomly and blinded, to the two observers who had participated in the evaluation of PS as well as to one other observer who had not. The interobserver correlation for both scales was very high (K = 0.921 for KPS and K = 0.914 for EPS) as was the intraobserver correlation (for KPS: K = 0.993, K = 0.960, and K = 0.959 and, respectively, for EPS: K = 0.982, K = 0.970, and K = 0.920). On the basis of these results, it appears that evaluation of PS made by a clinical oncologist using K or E scales can be very reliable and is a guarantee of optimal selection of cancer patients for inclusion in clinical trials.