Comparative survival curve analysis in sickle cell disease

摘要

In a typical followup study of a chronic disease patients enter and leave the study throughout its course. The usual Kaplan-Meier estimate of survival probability as a function of chronological age assumes that all patients are at equal risk from the beginning of the study irrespective of their age at entry. We proposed a modification of the Kaplan-Meier estimate which takes into account patients entering the study after commencement by treating them as “negative losses”. In this paper the usual Kaplan-Meier estimate is compared with our modification of the Kaplan-Meier estimate by applying each method to a data base of an ongoing study of 525 sickle cell disease patients. The data base was subdivided into groups by sex and hemoglobin type. We found that the usual Kaplan-Meier estimate has a lower resolution between groups than the modified Kaplan-Meier estimate, particularly when the true differences between groups are not great. In this sample male Hb SS patients had a better overall prognosis than female patients; a less acute form of the disorder associated with Hb SC, as expected, had a significantly higher survival than either the female or male SS patients. These subgroup and group differences were more clearly shown by the modified Kaplan-Meier estimate.