Can inaccuracy of reported parental history of diabetes explain the maternal transmission hypothesis for diabetes?