Which interactions matter in economic evaluations? A systematic review and simulation study

Table 2 Magnitude of interactions for the 16 studies reporting mean costs and mean health benefits for each cell within the factorial design

	Interaction: effect ratio	Proportion of interactions in different categories
	Interaction: effect ratio	Cost	Health benefit	Net monetary benefit^a
Qualitative	<−1	16.7% (4/24)	33.3% (8/24)	29.2% (7/24)
Sub-additive	Between −1 and 0	41.7% (10/24)	25.0% (6/24)	25.0% (6/24)
Mixed non-qualitative^b	Between −1 and 0	4.2% (1/24)	12.5% (3/24)	12.5% (3/24)
Additive	0	0.0% (0/24)	8.3% (2/24)	0.0% (0/24)
Super-additive	> 0	37.5% (9/24)	20.8% (5/24)	33.3% (8/24)
Interaction:effect ratio > 1 or < −1		29% (7/24)	38% (9/24)	38% (9/24)
Range of interaction:effect ratios		-2, 232	−23, 2	−44, 13

^a A ceiling ratio of £20,000 (or $20,000 or €20,000) was used for quality-adjusted life-years (QALYs) [14] and life-years gained, while a ceiling ratio of £5000 (or $5000 or €5000) per unit of benefit was arbitrarily used for all other health benefits
^b Mixed interaction: one factor increases the outcome of interest, while the other decreases it. The interaction therefore cannot be classified as either sub-additive or super-additive

ISSN: 1471-2288