Characteristics of a randomised sample of 842 question–answer pairs in the RELIS public database 1995–2006
| Period | 1995–2000 | 2001–2006 | p Value |
| Questions, n | 312 | 530 | |
| Drugs involved, category: % | One: 49 | One: 45 | 0.434§ |
| Two: 19 | Two: 22 | ||
| Three or more: 32 | Three or more: 34* | ||
| Judgemental questions, n (%) | 207 (66) | 399 (75) | 0.005§ |
| Patient-related questions, n (%) | 170 (54) | 379 (72) | <0.001§ |
| Answers, n | 312 | 530 | |
| Literature search, type: %† | None: 3 | None: 2 | 0.315§ |
| Simple: 43 | Simple: 40 | ||
| Advanced: 54 | Advanced: 58 | ||
| Reference to RELIS database, n (%) | 39 (13) | 164 (31) | <0.001§ |
| References, median (average) | 3 (4) | 4 (5) | 0.100¶ |
| Days, median (average)‡ | 2 (8.4) | 1 (5.0) | 0.001¶ |
↵* The total number exceeds 100% because of rounding off to the nearest whole number.
↵† Simple literature search included monographs (eg, Martindale, Summary of Product Characteristics), advanced literature search included articles (eg, databases like Medline and Embase).
↵‡ Answer time is influenced by some extreme outliers (>100 days) and choice of deadline for answer (‘not important’ is an option).
↵§ χ2 Test for contingency tables.
↵¶ Mann–Whitney U test for two independent samples.