Statistical analysis of publicly funded cluster randomised controlled trials: a review of the National Institute for Health Research Journals Library

Table 4 Cluster and sample size characteristics of the trials included in the review

Characteristics	n	%	Mean (SD)	Median	Range	IQR
No. of clusters randomised (N = 86)
4–10	2	2	77 (121)	44	7–922	25–74
11–20	11	13
21–50	40	47
51–100	21	24
101–200	5	6
> 200	7	8
No. of clusters analysed (N = 86)
0–10	2	2	76 (118)	43	7–864	25–69
11–20	12	14
21–50	40	47
51–100	21	24
101–200	4	5
> 200	7	8
No. of subjects recruited (N = 84^b)
≤ 300	7	8	15,348 (48315)	1184	141–265,434	597–3653
301–600	14	17
601–900	11	13
901–1200	9	11
1201–1500	9	11
1501–1800	3	4
> 1800	29	34
Not reported	2	2
No. of subject analysed (N = 84^b)
≤ 300	15	18	14,367 (48419)	870	42–264,325	441–2356
301–600	15	18
601–900	13	15
901–1200	5	6
1201–1500	5	6
1501–1800	2	2
> 1800	25	30
Not reported	4	5
Planned ICC for sample size (N = 86)
0.00–0.02	18	21	0.065 (0.082)	0.05	0.0002–0.5	0.0258–0.0700
> 0.02–0.05	33	38
> 0.05–0.08	9	11
> 0.08–0.11	8	9
> 0.11–0.14	2	2
> 0.14	6	7
Not reported	10	12
Planned design effect (N = 86)
0.00–2.99	47	55	4.5 (8.90)	1.96	1.03–70.5	1.384–4.600
3.00–5.99	12	14
6.00–8.99	10	12
9.00–11.99	1	1
≥ 12	3	3
Not reported	13	15
Observed ICC of analysed primary outcome (N = 100)
− 0.02 to 0.02	35	35	0.06 (0.12)	0.02	-0.02 to 0.63	0.0010–0.0600
> 0.02–0.07	9	9
> 0.07–0.12	3	3
> 0.12–0.17	6	6
> 0.17–0.22	2	2
> 0.22	3	3
Not reported	42	42

^bTwo trials were excluded because the analysed subjects were measured in person-years. N = total number of trials and/or primary outcomes; n = counts observed in each level of a category; SD = standard deviation. Not reported means that the information of interest was not considered and/or provided in the trial

ISSN: 1745-6215