Sample representativeness and influence of attrition on longitudinal data collected as part of a national medical career tracking project