Table 1. Reasons for discrepancies in primary outcome measures.

				Change
Reasons for discrepancy between primary outcome(s) specified in the protocol and the review	Inclusion	Exclusion	Inclusion and Exclusion	Upgrade	Downgrade	Upgrade and Downgrade	Number of reviews
Recommendation by editors/peer reviewers	-	1	1	3	2	-	7
Recognition of the importance of the outcome before reading the results for the included trials	-	-	-	5	2	-	7
†Recognition of the importance of the outcome after reading the results for the included trials	3	-	-	2	2	-	7
†Outcome reflects the same domain as another outcome specified. Decision made after reading the results for the included trials	-	1	-	-	-	-	1
No results reported in the literature	-	1	-	-	1	-	2
Change in author from protocol/review – change reflects opinion of the importance of the outcome from another expert	1	1	-	2	-	-	* 4
Reviewer responded but could not recall reason for discrepancy	-	-	-	6	-	-	6
No response from authors	2	1	1	5	4	1	14
Total	6	5	2	23	11	1	48

^†Reasons where potential bias was suspected.

^*Delay between publication of the review and protocol for these four reviews: 27 months, 66 months, 75 months and 99 months (median for all 288 reviews was 24 months).