Table 2.

Principles for the vigiGrade completeness score

Principle	Idea	Example
Measure the amount of information on reports	The score shall reflect the amount of information provided, but not whether it strengthens the suspicion of a causal relation; the magnitude of the score shall reflect the amount of information, and its maximum value shall be 1	Information on time-to-onset is rewarded even if it makes it unlikely that the adverse event is truly caused by the drug
Consider dimensions not data elements	The score shall focus on dimensions, and not on specific data elements. It shall account for the fact that the same information can be provided through different combinations of data elements	For time-to-onset, the reaction start date is only valuable together with the treatment start date, and redundant if time-to-onset is explicitly reported in a separate field
Weigh by importance	The penalties for absence of information on specific dimensions shall reflect the importance of that dimension for causality assessment	Absence of information on time-to-onset receives greater penalty than absence of information on dose
Allow substantial penalties for individual dimensions	Absence even of single dimensions may significantly decrease the scorea	Absence of information on either (1) time-to-onset or (2) both patient age and sex each decrease the score by 50 % or more
Focus on machine-readable information	The score shall account only for information that can be identified by a computer, for scalability and secondary use	Information on time-to-onset in a free-text case narrative is not rewarded

^aThis is the primary motivation for the multiplicative penalties—if missing information would be penalised by subtraction, it would not be possible to penalise individual missing information items to the same extent