Table 2.

Application of summary approaches to paroxetine vs placebo for major depression in adults

Outcomes	Estimated risk with Placebo	Absolute reduction in risk with Paroxetine	Relative effect (95% CI)	Number of participants (studies)	Confidence in effect estimate 1	Comments
(A) Standard deviation units	The depression score in the paroxetine groups was on average 0.31 SDs (0.24 to 0.38 lower than in the placebo groups)		---	5736 (34)	⊕⊕OO2,3 low	As a rule of thumb, 0.2 SD represents a small difference, 0.5 moderate, and 0.8 large (Cohen, 1988)
(B) Natural units
Major depression measured on Hamilton Rating Scale for Depression, generally scored from 0 to 50, higher scores indicate more severe depression	The mean depression scores with placebo ranged from 3.1 to 11.3	The mean depression score in the intervention groups was on average 2.47 (1.91 to 3.03) lower		5736 (34)	⊕⊕OO2,3 low	Scores estimated based on an SMD of 0.31 (95% CI 0.24 to 0.38)The minimal important difference on the 0 to 50 depression scale is 7 points. Although the depression score was on average only 2.47 lower, the corresponding NNT is 11
(C) Risk difference	50 per 100 patients	39 per 100 patients	OR=1.64 (95% CI 1.47 to 1.84)	5736 (34)	⊕⊕OO2,3 low	This approach uses binomial and equal variance assumptions and baseline risks, and demonstrates that for every 100 patients treated with paroxetine, 11 will achieve important improvement
		Differences in proportion achieving important improvement
		0.11 (95% CI 0.07 to 0.16) in favor of paroxetine
(D) Ratio of means	---	---	Ratio of means	5736 (34)	⊕⊕OO2,3 low	Weighted average of the mean depression score in paroxetine group divided by mean depression score in placebo. RoM method provides similar effect estimates compared with the traditionally used standard deviation unit, with SMDs of 0.2, 0.5, and 0.8, corresponding to increases in RoM of approximately 8%, 22%, and 37%, respectively (Friedrich 2011).
			1.27 (1.18 to 1.36)
(E) Minimal important difference units	The depression score in the paroxetine groups was on average 0.38 (95% CI 0.30 to 0.47) minimal important difference units less than the control group		---	5736 (34)	⊕⊕OO2,3 low	An effect less than half the minimal important difference suggests a small effect

Note: Investigators measured depression using different instruments, higher scores indicate more severe depression. ¹Quality rating from 1 (very low quality) to 4 (high quality); ²Evidence limited by heterogeneity between studies; ³Evidence limited by risk of bias (i.e. missing participant data and potential for selective reporting bias).