Table 4

Recommendations to improve adverse event analysis and reporting in clinical trial report publications

	Recommendation
Analysis	Incorporate objective statistical methods to assist the evaluation of adverse event information.
	Consider avoiding dichotomising continuous data.
	When count outcomes are available (such as repeated events within participants) use appropriate statistical methods.
	Clearly define exposure and specify a suitable safety analysis population.
	Use graphical approaches to help summarise large amounts of data.
Reporting	Report adverse event data according to the CONSORT harm checklist.
	Increase the uptake of mandatory submission of CONSORT harm by journals.
	Include a relevant summary of the adverse event profile in the main article. Resist depositing all adverse event data into appendices without summarising.