[Factors influencing long-term survival in patients with nonoperable lung cancer: an analysis by Cox model]

This paper reports a prospective survey of 173 patients with nonoperable lung cancer between January. 1, 1983 to March. 1, 1985. The follow-up rate was 97.7% over five years. Fourteen factors including sex, age, course of disease before treatment, clinical stage, performance status, size of mass, metastatic status, hemoglobin before treatment, short-term response to treatment and so on which might influence long term survival were studied by univariate analysis (Kruskal-Wallis test for Kaplan-Meier survival curve) and by multivariate analysis (Cox's proportional hazad model and audio-visual chart test for goodness of fit). Multivariate analysis using Cox's model revealed 6 significant prognostic factors: performance status, short-term response to treatment, clinical stage, hemoglobin before treatment, smoking index and method of treatment. The survival prediction equation was chi 2 = 72.14, nu = 6, P < 0.0001. The results indicate that the performance status and the CR rate of the initial treatment, among other things, is the major factors affecting prognosis.