| pubmed-article:1820119 | pubmed:abstractText | 1. Most multivariate analyses of first cadaver-donor kidney grafts have assumed additivity of covariate effects, ignoring possible interaction terms. Using 21,895 primary cadaveric renal transplants reported to the UNOS Kidney Registry, we attempted to assess whether or not strong interaction effects on graft survival existed among 11 important factors. Due to the small number of cases for many multiway factor combinations, our interaction analyses were limited to studying how stratification by 1 factor affected the graft failure risks of a second factor. Two stratification factors were chosen: recipient race and age. 2. Although there was little evidence of strong factor interaction across 5 racial groupings, differences in risk for several factors (recipient age, donor race, donor age, HLA-B,DR mismatching, and cold ischemia time) between Blacks and non-Blacks appeared to be nonadditive (2 examples listed in Table 3). 3. In a more restricted population of White recipients older than age 16, there was only evidence of interacting effects on 3-month graft survival between donor sex and recipient age. 4. Despite many pairwise factor associations (Figs. 1, 3), it appears that the majority of the transplant covariates are additive with respect to the risk of early graft loss. In future analyses, we suggest inclusion of appropriate interaction terms relating the above-mentioned factors with recipient race and age. | lld:pubmed |
