Table 1.

‘best’ and ‘worst’ case scenarios used to estimate the optimum sample size and duration of the preintervention and postintervention periods in the eight ICUs

Participants per ICU	Power (%)	Reference process rate (%)	Mean process rate change detectable (%)	Base sample size per group	ICC	Inflation factor	Inflated sample size (4 ICUs per analysis group): participants per group
(A) ‘Best case scenario’ sample size justification
250	80	70	19.7	74	0.05	13.45	1000
250	80	70	26.2	38	0.10	25.90	1000
250	80	70	29.3	29	0.13	33.37	1000
(B) ‘Worst case scenario’ sample size justification
66	80	70	19.7	74	0.040	3.57	264
66	80	70	26.2	38	0.092	6.95	264
66	80	70	29.3	29	0.125	9.10	264

Inflation factor is 1+(m−1)×ICC, where m=cluster size.

ICU, intensive care unit.