Defining certainty of net benefit: a GRADE concept paper
https://bmjopen.bmj.com/content/9/6/e027445.long
診療ガイドライン作成における「判断の閾値・範囲」のGRADE概念の必要性が理解されない。。。。（Classification of precision of net effect estimate）
●Classification of precision of net effect estimate
正味の効果の閾値として「ゼロ効果」を設定したというシナリオでの、各パターンの意味が以下です。
| Pattern of net effect estimate | Classification | Precision of net effect estimate is consistent with … |
| Entire CI is beneficial | Net benefit | High certainty of net benefit |
| Point estimate is beneficial, lower bound of CI is harmful and point estimate has larger absolute value than lower bound of CI | Likely net benefit | Moderate certainty of net benefit |
| Point estimate is beneficial, lower bound of CI is harmful and point estimate has smaller absolute value than lower bound of CI | Possible net benefit | Low certainty of net benefit |
| Point estimate is close to zero, wide CI* | Possibly no net benefit or harm | Very low certainty of net benefit or harm |
| Point estimate is close to zero, narrow CI* | Net benefit or harm likely near zero | Moderate certainty of little net benefit or harm |
| Point estimate is harmful, upper bound of CI is beneficial and point estimate has smaller absolute value than upper bound of CI | Possible net harm | Low certainty of net harm |
| Point estimate is harmful, upper bound of CI is beneficial and point estimate has larger absolute value than upper bound of CI | Likely net harm | Moderate certainty of net harm |
| Entire CI is harmful | Net harm | High certainty of net harm |
*Differentiation of wide versus narrow CIs could be based on a threshold of minimally important differences.
