Adjustment of long-term survival rates for deaths due to intercurrent disease

Abstract

PA= survival is the ultimate criterion for measuring the effectiveness of treatment for a chronic disease, but the interpretation of survival rates is complicated by deaths due to causes other than the disease under study. The risk of death from intercurrent disease is particularly important in comparing the survival of patient groups with different age distributions and in comparing the survival of different age groups. Consider the observed survival curves for women with cancer of the breast who were surgically treated at the Memorial Hospital for Cancer and Allied Diseases in New York City during 1950-1952 (Fig. 1). The data are for 1481 women, divided into 6 age groups. Survival rates were calculated by the life table method [l-3]. It is clear that after ages 45-54 survival decreases as age increases. Is this because treatment is less effective in older women or does the decreasing survival reflect the increasing risk of dying from other causes? The relative survival rate has been used to adjust for ‘normal mortality risk [45]. The relative rate is the ratio of the observed survival rate in a group of patients to the survival rate expected from general population experience. In the United States, Life Tables published by the National Center for Health Statistics are generally used for estimating expected survival. In estimating expected rates the race, sex, and age distribution of the patient group and the calendar period of observation must be taken into account. Thus, the relative rate can be regarded as a measure of the probability of escaping the extra risk of dying from the disease under study. Relative survival rates by age are shown in Fig. 2. The relative survival curves for the 4 age groups under age 65 are close together. The survival curve for patients 75-84 yr of age is at a lower level from the Hth year on, but the rates are subject to considerable chance variation due to the small number of patients involved. A puzzling finding is the apparently more favorable survival of women in the 65-74 yr age group from the ninth year on. If we were to accept this finding at face value, we would conclude that women found to have breast cancer at ages 65-74 and surgically treated have a higher long range probability of not dying from the disease. The use of general population mortality experience as a base for comparison