Analysis of Competing Risks with Frequent Tied Event Times

• Activity Number: 72
• Type: Contributed
• Date/Time: Sunday, August 7, 2005 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Robert Glynn*+ and Bernard Rosner and William G. Christen
• Companies: Brigham and Women's Hospital and Harvard Medical School and Brigham and Women's Hospital
• Keywords: competing risks ; survival analysis ; generalized estimating equations

Abstract

For diseases that usually require examination for detection, multiple subtypes of the disease often can be present at the time of first diagnosis. For example, at the time cataract reduces visual acuity to 20/30 or worse in an eye, that eye often has evidence of two or all three of cortical, nuclear sclerotic, and posterior subscapsular cataract. The impact of risk factors on different subtypes of cataract can be elucidated through methods of competing risk survival analysis that account for tied events. We studied the incidence of cataract among 20,599 participants in the Physicians' Health Study who were free of cataract at enrollment and were followed until the end of the trial in 1985. We describe the use of generalized estimating equations to account for tied failure times in comparisons of the effects of factors including age, diabetes, cigarette smoking, body mass index, and multivitamin use on risk of different types of cataract. We compare results to alternative analyses based on polytomous logistic regression.