Preference-based condition-specific measures of health: what happens to cross programme comparability?