Table 3.

Factors associated with the reduction in carbapenem use from random-effects logistic regression model

Variable	Univariable analysis		Multivariable analysis
	OR	p	OR	p a	95% CI
Type of ward (reference: Medical)
ICU	0.94	0.83	0.07	0.005	0.01–0.46
Surgical	0.95	0.81	0.66	0.19	0.35–1.23
Ward size in 2013, per increase of 50 beds	0.90	0.14	0.79	0.02	0.64–0.97
Type of HCF (reference: General hospital)
University hospital	0.42	0.35	0.28	0.25	0.03–2.39
Private hospital	1.06	0.85	0.66	0.28	0.31–1.40
Cancer centre	0.37	0.29	0.09	0.04	0.01–0.88
Local hospital	2.33	0.18	1.19	0.83	0.25–5.54
Rehabilitation and long-term care facility	0.38	0.50	0.24	0.40	0.01–6.59
Military hospital	0.43	0.34	0.18	0.09	0.02–1.28
Geographical area (reference: South-east)
North	1.42	0.42	2.33	0.09	0.88–6.16
West	1.15	0.80	1.09	0.90	0.30–3.94
East	1.93	0.15	3.11	0.03	1.10–8.77
South-west	1.78	0.17	1.57	0.36	0.60–4.12
Antibiotic use, DDDs per 1000 patients-days
Initial CBP useb	1.01	0.01	0.97	0.08	0.95–1.00
Initial prescribing profile of CBPsc	1.22	0.002	3.16	< 0.001	1.88–5.34
Initial 3GC useb	1.00	0.79
Initial prescribing profile of 3GCsc	0.97	0.16	0.89	0.002	0.83–0.96
Reduction in 3GC used	1.34	0.23
Initial FQ useb	1.00	0.25
Initial prescribing profile of FQsc	1.02	0.29
Reduction in FQ used	2.21	0.004	2.3	0.02	1.17–4.53
Initial AG useb	1.00	0.50
Initial prescribing profile of AGsc	0.97	0.44
Reduction in AG used	1.56	0.06	1.13	0.67	0.65–1.96
Initial AP ± BLI useb	1.00	0.83
Initial prescribing profile of AP±BLIsc	1.00	0.99
Reduction in AP ± BLI used	1.10	0.68
Initial Imidazole useb	1.00	0.98
Initial prescribing profile of imidazolesc	0.97	0.45
Reduction in imidazole used	1.65	0.03	1.50	0.15	0.87–2.59
Initial GP useb	1.00	0.95
Initial prescribing profile of GPsc	0.97	0.61
Reduction in GP used	1.83	0.01	1.98	0.02	1.13–3.45
Initial TZP useb	1.00	0.99
Initial prescribing profile of TZPc	0.95	0.33
Reduction in TZP used	4.24	< 0.001	4.02	< 0.001	1.99–8.14
Initial use of other alternatives to CBPsb	1.00	0.88
Initial prescribing profile of other alternatives to CBPsc	1.31	0.25
Reduction in the use of other alternatives to CBPsd	1.07	0.79
Initial overall antibiotic useb	1.00	0.72	1.00	0.97	0.99–1.01
Reduction in the overall antibiotic used	2.02	0.003	1.38	0.29	0.76–2.51
Region with high ESBL-PE incidence rate in acute care settingse	0.65	0.12	0.40	0.01	0.19–0.82

OR odds ratio, CI confidence interval, ICU intensive care unit, HCF healthcare facility, DDD defined daily dose, CBP carbapenem, 3GC third-generation cephalosporin, FQ fluoroquinolone, AG aminoglycoside, AP ± BLI aminopenicillin ± β-lactamase inhibitor, GP glycopeptide, TZP piperacillin/tazobactam. ESBL-PE extended-spectrum β-lactamase-producing Enterobacteriaceae

^aSignificant associations (p < 0.05) from multivariable analysis are highlighted in bold italic

^bConsumption of a given antibiotic in 2009

^cProportion (%) of a given antibiotic in the total antibiotic consumption in 2009

^dReduction in the consumption of a given antibiotic between 2009 and 2013

^eData from the French multidrug-resistant bacteria surveillance network; BMR-RAISIN, 2011