Abstract
Seventy-seven prognostic factors were considered in an evaluation of more than 5,000 patients with inoperable bronchogenic carcinoma of the lung; these patients were entered on the Veterans Administration Lung Group protocols 9–15 (1968–78). Fifty prognostic factors for survival were identified, and their relative contributions to patient survival were considered. The three most important prognostic factors affecting survival were the Karnofsky initial performance status score, extent of disease, and weight loss in the previous 6 months. These factors denoted three general prognostic components: current physical status, current disease status, and prior physical status. Initial performance status was the dominant prognostic factor. Characterization of patient's function status with the use of the Karnofsky scale was preferred to a summarized scale or a dichotomization into ambulatory versus nonambulatory. The current practice of dichotomizing factors resulted in the loss of much prognostic information. Other factors, such as tumor size, histologic type, and institution, appeared to be important when considered alone. However, their contribution was relatively minor after a correction was made for the effects of initial performance status, extent of disease, and prior weight loss. Depending on these three prognostic factors alone, median survival varied between 6 weeks and over a year. Reporting of results without the consideration of such strong prognostic factors severely hampers any comparisons that may be made between investigations.