The Hartung-Knapp-Sidik-Jonkman method for random effects meta-analysis is straightforward and considerably outperforms the standard DerSimonian-Laird method

Table 2 Number (%) of statistically significant Cochrane meta-analyses according to the DerSimonian-Laird and Hartung-Knapp-Sidik-Jonkman methods

Outcome	Selected meta-analyses	N	DL test significant	HKSJ test significant	HKSJ test not significant, positive DL test
Continuous	All	255	130 (51.0)	102 (40.0)	31/130 (23.8)
	Ratio > 5, < = 5 studies	46	22 (47.8)	13 (28.3)	13/22 (59.1)
Dichotomous	All	434	185 (42.6)	147 (33.9)	48/185 (25.9)
	Ratio > 5, < = 5 studies	76	28 (36.8)	15 (19.7)	14/28 (50.0)

All: all meta-analyses with a continuous or dichotomous outcome that fulfilled the following criteria: the first meta-analysis in a review in the Cochrane Database for Systematic Reviews Issues of 2012, based on at least three studies. Ratio >5, < = 5 studies: a selection of these meta-analyses based on at most five studies, where the ratio of the largest vs. the smallest trial size was > 5. DL: DerSimonian & Laird meta-analysis method. HKSJ: Hartung-Knapp-Sidik-Jonkman meta-analysis method. DL test significant: DL p-value <0.05; HKSJ test significant: HKSJ p-value < 0.05. Note that in a few cases the HKSJ test was significant when the DL test was not.

ISSN: 1471-2288