Eleven cohorts were available in which to replicate any associations for NEO personality measures. These cohorts were of European descent and are described in de Moor et al [29]. In brief, they comprised samples of varying age range (15 – 87 years) from Italy (Cilento, SardiNIA), The Netherlands (NTR/NESDA, ERF), United States (BLSA, SAGE), Finland (HBCS), Australia (QIMR, NAG/IRPG), Estonia (EGPUT) and Germany, totalling 17 106 participants with mean age ranging from 19.4 ±3 (Australia: QIMR) to 78.9 ±5.4 (Italy: Cilento) years. Personality scores for the five factors were based on the 60 items of the NEO-FFI (12 items per factor)[6]. SNP data were available from varying sized genome-wide scans which had all been imputed to ~2.4M SNPs using HAPMAP II data. Association analysis of each SNP (under an additive model) had been performed using either PLINK, SNPTEST or MERLIN[29]. The results for selected longevity candidate gene SNPs were meta-analysed using METAL [30]. Two separate meta-analyses of older cohorts (HBCS, BLSA, Cilento, Germany: mean age > 63.4 years; N = 2555) and younger cohorts (SardiNIA, NTR/NESDA, ERF, SAGE, NAG/IRPG, QIMR, EGPUT: mean age < 49.3 years; N = 14 551), then a combined analysis, were performed.
