Practical Considerations in Meta- and Pooled Analyses of Type II Diabetes Data in Normal-Weight Adults

• Activity Number: 190
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Peter de Chavez*+ and Juned Siddique and Alan R. Dyer and Mercedes R. Carnethon
• Companies: Northwestern University and Northwestern University and Northwestern University and Northwestern University
• Keywords: meta-analysis ; pooled analysis ; diabetes ; mortality ; obesity ; normal weight

Abstract

Type 2 diabetes (T2DM) in normal weight adults is an intriguing representation of the metabolically obese normal weight phenotype. Since its prevalence is low (5-15% of all cases), no single study has had an adequate sample size to assess the health consequences of normal weight T2DM. However, several cardiovascular studies that include common measurements of variables relevant to T2DM have been conducted. Consequently, it is possible to carry out pooled and meta-analyses to test the hypothesis that mortality is higher among overweight persons with T2DM than persons with normal weight T2DM. Conducting pooled and meta-analyses presents a number of challenges such as non-overlapping demographic characteristics of cohorts (e.g., studies restricted to a given race or age); and the use of different measurement units, assays, and survey methodologies. This poster presents several methods used to combine results from 5 cardiovascular studies of 2,619 participants in order to estimate the effect of weight on mortality among adults with T2DM. We also describe the results of our analyses and discuss the limitations of our methods.