Table 2.

Percentage study weights for the examination of a treatment-age interaction in one- and two-stage IPD random effects meta-analyses of 10 randomised trials of anti-hypertensive treatment versus control on systolic blood pressure.

		Percentage study weights using equation (12)
Trial ID (i)	Mean age (${\overline{z}}_i$)	Model (19): Meta-regression to estimate ${\gamma }_{\text{A}}$ (across-trial interaction)	Model (20): Two-stage approach to estimate ${\gamma }_W$ (within-trial interaction)	Model (21): One-stage approach to estimate ${\gamma }_{\text{WA}}$ (amalgamated interaction)	Model (22): One-stage approach to estimate ${\gamma }_{\text{E}}$ (ecological bias)
1	42.27	21.26	4.52	8.88	17.33
2	69.63	2.91	0.66	1.24	2.36
3	73.34	3.64	0.81	1.56	3.13
4	41.56	24.90	11.64	15.13	21.72
5	45.27	16.68	28.78	25.62	19.63
6	70.42	6.15	1.45	2.70	4.93
7	71.59	8.16	17.13	14.81	10.21
8	75.95	7.56	0.29	2.15	5.87
9	66.58	2.35	8.35	6.79	3.74
10	70.23	6.40	26.36	21.11	11.09
Meta-analysis result (95% CI)		${\hat{\gamma }}_{\text{A}}$ = −0.132 (−0.246, −0.018)	${\hat{\gamma }}_W$ = −0.050 (−0.116, 0.016)	${\hat{\gamma }}_{\text{WA}}$ = −0.072 (−0.128, −0.0149)	${\hat{\gamma }}_{\text{E}}$ = −0.086 (−0.220, 0.048)