Towards better meta-analyses in assisted reproductive technology: Fixed, random or multivariate models?

Table 1 Luteneizing hormone supplementation effect: Comparison between fixed, random and multivariate models

		BPR				CPR				OPR				LBR
Overall	FM	1.17	1.01	1.28	0.03	1.08	0.99	1.17	0.03	1.09	0.93	1.19	0.25	1.28	0.98	1.56	0.07
	RM	1.14	0.95	1.38	0.16	1.07	1	1.19	0.05	1.05	0.93	1.19	0.45	1.23	0.88	1.72	0.23
	MM	1.23	0.94	1.61	0.12	1.09	1	1.2	0.06	1.13	1.03	1.24	0.01	1.13	1.01	1.28	0.04
POR	FM	1.22	0.97	1.53	0.1	1.30	1.03	1.64	0.03	1.29	0.82	2.02	0.28	1.81	0.99	3.31	0.06
	RM	1.19	0.87	1.64	0.27	1.30	0.99	1.64	0.04	1.29	0.82	2.02	0.28	1.70	0.78	3.69	0.27
	MM	1.37	1.00	2.21	0.04	1.27	1.01	1.61	0.04	1.38	1.06	1.81	0.02	1.53	1.09	2.15	0.01

Luteneizing hormone supplementation effect for overall population and POR models. Comparison between fixed (FM), random (RM), and multivariate (MM) models. Values are risk ratio 95%CI, and P-value. BPR: Biochemical pregnancy rate; CPR: Clinical pregnancy rate; OPR: On going pregnancy rate; LBR: Live birth rate; POR: Poor ovarian responder.

Table 2 Between endpoint correlation

	CP	OP	LB	NST	I2
BP	0.95 (0.91, 0.97)	0.91 (0.84, 0.95)	0.87 (0.79, 0.93)	22	41.2
CP	-	0.96 (0.92, 0.97)	0.92 (0.85, 0.95)	39	31.8
OP	-	-	0.96 (0.92, 0.98)	13	46.36
LB	-	-	-	8	40.1

Within-study correlations (95%CI). NST: Number of studies reporting each endpoint; I²: Heterogeneity index; BP: Biochemical pregnancy; CP: Clinical pregnancy; OP: On going pregnancy; LB: Live birth.