Ideal vs. real: a systematic review on handling covariates in randomized controlled trials

Table 2 Covariate Involvement in the Design of Published RCTs ^a

	Covariates Involved in Allocation (N = 298)					Covariate-Adaptive Method (N = 237) ^b
	No		Yes		p-value	No		Yes		p-value
	N	%	N	%		N	%	N	%
Type of Study					0.0140					0.4694
Superiority	43	17	212	83		182	88	26	13
Non-Inferiority	5	22	18	78		14	78	4	22
Other	9	45	11	55		9	82	2	18
Number of Arms					0.0246					0.7461
2	40	17	196	83		167	87	26	13
3	5	21	19	79		16	84	3	16
4	7	27	19	73		15	83	3	17
5 or more	5	42	7	58		7	100	0	0
Multicenter Study					0.0212					0.9843
No	9	38	15	63		13	87	2	13
Yes	48	18	226	82		192	86	30	14
Number of Centers, median (IQR)	12 (2–51)		24 (7–76)		0.8676	23 (7–76)		26 (6–64)		0.3399
Number of Subjects, median (IQR)	546 (200–1032)		584 (250–1442)		0.3601	580 (255–1369)		613 (188–1389)		0.5002
Length of Study, Years, median (IQR)	3 (2–4)		3 (2–5)		0.0158	3 (2–5)		3 (3–7)		0.0403
Months of Subject Follow-up, median (IQR)	6 (3–18)		12 (6–22)		0.1825	12 (6–19)		12 (6–36)		0.5449
Study Initiation Year					0.0889					0.0066
Unable to Determine	1	20	4	80		3	75	1	25
Before 2000	1	6	16	94		11	69	5	31
2000–2004	13	17	62	83		49	82	11	18
2005–2009	23	19	100	81		87	89	11	11
2010 and later	19	24	59	76		55	93	4	7
Primary Outcome					0.0054					0.4861
Continuous	23	26	66	74		55	86	9	14
Binary	19	17	94	83		82	89	10	11
Time-to-Event	9	11	75	89		64	85	11	15
Other	6	50	6	50		4	67	2	33
Number of Covariates (involved in randomization algorithm)										<.0001
1	Not applicable for this outcome.					93	98	2	2
2						78	93	6	7
3						30	77	9	23
4						4	40	6	60
5 or more						0	0	9	100

^aP-values correspond to logistic regression results
^bThis subset of 237 studies includes those studies clearly involving either stratified block method (N = 205) or an adaptive method (N = 32). Refer to Table 1 (205 + 32 studies using stratified block method or covariate-adaptive method, respectively)

ISSN: 1471-2288