In this analysis we were interested in whether any particular diagnostic phenotype definition(s) provided strong association evidence when a set of participants with bipolar disorder meeting this definition was analysed against controls for the full set of SNPs within the genome-wide analysis. We used a genome-wide analysis (i.e. a case–control comparison for each of the 276 122 SNPs) as the basic unit of study and summarised the overall association evidence in that analysis by counting the number of independent SNPs that showed association exceeding a specified significance threshold (chosen as P<10–5, the significance benchmark in the WTCCC study used to designate at least ‘moderately strong’ evidence for association). We refer to these associated SNPs as ‘hits’ (which is a shorthand term used in molecular genetics to indicate an independent association signal that meets a specified level of statistical significance). Our basic aim was to determine if one or more definitions of diagnosis possessed greater utility by virtue of identifying more hits. When comparing the number of hits from different diagnostic phenotype definitions there is an important complicating factor – the
