A Statistical Algorithm for Assessing Homogeneity Among Dried-Blood-Spot (DBS) Pools

• Activity Number: 254
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Quality and Productivity
• Author(s): Maya Sternberg*+ and Victor De Jesus and Joanne Mei
• Companies: Centers for Disease Control and Prevention and Centers for Disease Control and Prevention and Centers for Disease Control and Prevention
• Keywords: quality assurance ; homogeneity ; ANOVA ; equivalence

Abstract

Effective screening of newborns using dried-blood-spot (DBS) specimens collected at birth helps prevent mental retardation and premature death through early detection. Each year, millions of babies in the U.S. are routinely screened for certain genetic, endocrine, and metabolic disorders. CDC prepares and distributes more than 500,000 DBS per year to US domestic and international laboratories as part of a quality assurance program to ensure high levels of technical proficiency. Prior to distribution, the homogeneity of the DBS materials is assessed by randomly sampling DBS cards from across the entire pool, and punching disks from spots across each card. The natural statistical test to assess homogeneity is a one way random effects analysis of variance. However, instead of basing the homogeneity assessment on the standard F-test, Fearn and Thompson (2001) suggest a test called "sufficient homogeneity," which bases the decision on defining a permissible target variance and comparing that value to a one sided 95% confidence interval on the between-sample variance, (L, infinity). This abstract proposes a new homogeneity test based on principles of equivalence hypothesis tests.