Table 2. Changes in incidence of MRSA in three different phases of intervention.

	Immediate intervention impact				Pre- intervention trend		Change in trend
	Phase 2 vs Phase 1		Phase 3 vs Phase 2				Phase 3 vs Phases 1 & 21
	IRR (95% CI)2;p-value	Percent change(95% CI)3	IRR (95% CI)2;p-value	Percent change(95% CI)3	IRR (95% CI)2;p-value	Percent change(95% CI)3	IRR (95% CI)2;p-value	Percent change(95% CI)3
Incidence rate of hospital-acquiredMRSA per 1000 admissions	0.637 (0.560, 0.724);p<0.001	−36.3%(−44.0%, −27.6%)	0.726 (0.625, 0.842);p<0.001	−27.4%(−37.5%, −15.8%)	NS	NS	NS	NS
Incidence rate of hospital-acquiredMRSA per 1000 patient-days	0.696 (0.612, 0.791);p<0.001	−30.4%(−38.8%, −20.9%)	0.759 (0.654, 0.881);p<0.001	−24.1%(−34.6%, −11.9%)	NS	NS	NS	NS
Incidence rate of hospital-acquiredMRSA per 1000 MRSA-positive-days	0.804 (0.652, 0.990);p = 0.040	−19.6%(−34.8%, −1.0%)	0.781 (0.615, 0.990);p = 0.041	−21.9%(−38.5%, −1.0%)	1.006 (1.002, 1.011);p = 0.007	0.6%(0.2%, 1.1%)	0.982 (0.971, 0.992);p = 0.001	−1.8%(−2.9%, −0.8%)

Note. Phase 1: baseline observation period; phase 2: launch of the first intervention - hand hygiene campaign; phase 3: continuation of phase 2 plus launch of the second intervention - contact precautions;

¹Insignificant change in trend in Phase 2 vs Phase 1 was found for all models, implying the same slope within Phases 1 and 2;

²IRR denotes incidence rate ratio obtained from segmented Poisson regression, and CI denotes confidence intervals;

³Percentage change in incidence rate obtained from (IRR−1)×100%; NS, not significant.