Taxonomy of EBM-based inferences
| Evidence classification | Inference | Relevant to advisements in text? |
| No direct evidence—but BHE favourable | Probably indicated until better evidence emerges. | No |
| No direct evidence—but BHE unfavourable | Probably not indicated until better evidence emerges. | Possibly 1 and 2 |
| Direct evidence favourable but uncertain (not tested with sufficient statistical power to rule-out a clinically significant difference and/or unexplained differences in direction of effect). | If BHE is favourable, indicated until better evidence emerges. If BHE is unfavourable, then there may be some indications for its use but caution is necessary. | No |
| Direct evidence unfavourable but uncertain (not tested with sufficient statistical power to rule-out a clinically significant difference and/or unexplained differences in direction of effect). | If BHE is favourable, there may be some indications for its use but caution is necessary until better evidence emerges. If BHE is unfavourable, it is probably not indicated until better evidence emerges. | Possibly 1 and 2 |
| Direct evidence favourable (tested with sufficient statistical power to rule-out a clinically significant difference). | BHE not relevant unless encompasses reasons unconsidered during evidence collection. Indicated except for those rationales. | No |
| Direct evidence unfavourable (tested with sufficient statistical power to rule-out a clinically significant difference). | BHE not relevant unless encompasses reasons unconsidered during evidence collection. Not indicated except for those rationales. | Possibly 1 and 2 |
See text for description.
BHE, benefit to harm expectancy; EBM, evidence-based medicine.