This motivates a proxy case–control association study where 'proxy cases' are the relatives of affected individuals and controls are the relatives of unaffected individuals.