Lead/Presenter: Boback Ziaeian, COIN - Los Angeles
All Authors: Ziaeian B (1. VA HSR&D Center for the Study of Healthcare Innovation, Implementation & Policy (CSHIIP), Los Angeles), Steers, WN (VA HSR&D Center for the Study of Healthcare Innovation, Implementation & Policy (CSHIIP), , Los Angeles CA) Hoggatt, KJ (VA HSR&D Center for the Study of Healthcare Innovation, Implementation & Policy (CSHIIP), Los Angeles CA) Wong MS (VA HSR&D Center for the Study of Healthcare Innovation, Implementation & Policy (CSHIIP), Los Angeles CA) Washington DL (VA HSR&D Center for the Study of Healthcare Innovation, Implementation & Policy (CSHIIP), Los Angeles CA)
National guidelines recommend using the 2000 U.S. Population Standard for reporting standardized mortality or disease rates. However, the decision to use a specific population standard may influence reporting of relative disparities, which has not been examined among Veterans. Our objective was to estimate the effect of utilizing alternative standard populations for estimating disparities.
We report rates and disparities for all-cause mortality using four alternative population standards for Veterans receiving care from 10/2008-9/2009 who had linked National Death Index data (n = 5,032,009). We utilized four standard populations to compare disparities of standardized mortality rates (SMR) by race/ethnicity. Ages were collapsed into seven CDC-defined categories. For the age-standardization, we first used the same VA white population cohort to generate age-adjustment weights for SMR. The second standard was the 2000 U.S. standard population from the U.S. Census. The World Health Organization global population standard was the third standard. The fourth standard was derived from the Global Burden of Disease project to weight ideal life expectancy as a function of age. All-cause SMR/100,000 person-years and SMR ratios were reported.
There is large variability in the reported SMR based on the standard population used. For whites, SMR was 5,123 for the VA white population, 2,007 using the 2000 U.S. Standard population, 1,524 using the WHO standard population, and 2,043 using the life expectancy standard. Depending on the standard used, there were shifts in the reporting of relative disparities related to the selection of a population standard. For African-Americans, an all-cause mortality disparity (1.040, 95% CI 1.033-1.047) reversed when using all three alternative population standards (e.g., 0.980, 0.973-0.987). In general, minority Veterans with a younger age distribution had lower relative disparities using alternative age-adjustment methods when compared to white Veterans.
There is marked variation in how disparities are presented statistically. The decision to use a certain standard population may influence the reporting and trend in reported disparities.
Guidelines should be developed on reporting disparities and age-adjustment of disease and mortality rates for Veterans, that provide consistent reporting using age distributions reflecting VA users, to avoid under detection of disparities.