UCLA School of Public Health, Los Angeles, California (firstname.lastname@example.org)
To the Editor: In the study by Ms Kucirka and colleagues1 on an effect modification by age for mortality in black compared with white dialysis patients, the overall relative risk (RR) of mortality for black vs white patients was reported as 0.90, with an adjusted hazard ratio of 0.84 after adjusting for demographics (including age) and comorbidities and censoring for transplantation. In contrast, the unadjusted RRs for the age subgroups were all higher, ranging from 0.95 for the 71- to 80-year-old group to 1.94 in the 18- to 30-year-old group (Figure 1 in the article). Could the authors please explain why the RR in the overall group was lower than in any of its component age subgroups, since ordinary adjustments are roughly equivalent to averaging over strata of the adjustment factors?2 Without an explanation for this apparent conflict, it is unclear if the results of the age subgroup analyses can explain the lower mortality rate seen in black dialysis patients.
Ricks J, Streja L, Greenland S. Race, Age, and Mortality Among Patients Undergoing Dialysis. JAMA. 2011;306(20):2215-2216. doi:10.1001/jama.2011.1715