Differential Outcome of an Antimicrobial Stewardship Audit and Feedback Program in Two Intensive Care Units

A Controlled Interrupted Time Series Study

Linda R. Taggart; Elizabeth Leung; Matthew P. Muller; Larissa M. Matukas; Nick Daneman

BMC Infect Dis. 2015;15(480)

Abstract and Introduction

Abstract

Background: Antimicrobial decision making in intensive care units (ICUs) is challenging. Unnecessary antimicrobials contribute to the development of resistant pathogens, Clostridium difficile infection and drug related adverse events. However, inadequate antimicrobial therapy is associated with mortality in critically ill patients. Antimicrobial stewardship programs are increasingly being implemented to improve antimicrobial prescribing, but the optimal approach in the ICU setting is unknown. We assessed the impact of an audit and feedback antimicrobial stewardship intervention on antimicrobial use, antimicrobial costs, clinical outcomes and microbiologic outcomes in two ICUs with different patient populations.

Methods: The audit and feedback intervention was implemented in a trauma and neurosurgery ICU (TNICU) and a medical surgical ICU (MSICU) at a 465-bed teaching hospital in Toronto, Canada. ICU patients were reviewed Monday to Friday by a physician and pharmacist with infectious diseases training. Recommendations related to appropriate antimicrobial use were presented to ICU teams during a dedicated daily meeting. A controlled interrupted time series analysis was used to compare outcomes in the 12 months before and after the intervention. Cardiovascular and coronary care ICUs served as control units.

Results: Mean total monthly antimicrobial use in defined daily doses (DDD) per 1000 patient days was reduced 28 % in the TNICU (1433 vs. 1037) but increased 14 % in the MSICU (1705 vs. 1936). In the time series analysis, total monthly antimicrobial use in the TNICU decreased by 375 DDD per 1000 patient days (p < 0.0009) immediately following the intervention, followed by a non-significant downward trend in use of −9 DDD per 1000 patient days (p = 0.56). No significant changes in antimicrobial use were identified in the MSICU. Antimicrobial use temporarily increased in one control unit and remained unchanged in the other. There were no changes in mortality, length of stay, readmission rate, incidence of C. difficile infection or resistance patterns of E. coli and P. aeruginosa in either intervention unit.

Conclusions: Audit and feedback antimicrobial stewardship programs can lead to significant reductions in total antimicrobial use in the ICU setting. However, this effect may be context-dependent and further work is needed to determine the ingredients necessary for success.

1 2 3 4 5 6

Next Section

Abstract and Introduction
Background
Methods
Results
Discussion
Conclusions
References

Table 1. Patient characteristics for those in the trauma and neurosurgery intensive care unit and medical surgical intensive care unit during the pre- and post-intervention periods.
Unit	Characteristic	Pre-intervention period	Post-intervention period	p-value
TNICU
	Admissions	n = 1330	n = 1387
	Age, mean (SD)	55 (18) ^b	56 (18)	0.12
	Male	757 (57)	801 (58)	0.66
	Admitting diagnosis			0.0086
	Neurological	879 (66)	937 (68)
	Trauma	327 (25)	364 (26)
	Other	124 (9)	86 (6)
	Ventilator utilization ratio	0.59	0.60	0.80
	Multiple organ dysfunction score, mean (SD)	2.39 (0.34)	2.32 (0.24)	0.56 ^a
	Febrile respiratory illness rate, cases per 1000 patient days	0.5	0.3	0.66
	Influenza rate, cases per 1000 patient days	0	0	1.00
	Patient days attributable to cystic fibrosis	0 (0)	0 (0)	–
MSICU
	Admissions	n = 1305	n = 1369
	Age, mean (SD)	61 (17)	59 (18)	0.016
	Male	784 (60)	833 (61)	0.68
	Admitting diagnosis			0.0082
	Respiratory	192 (15)	265 (19)
	Gastrointestinal	142 (11)	123 (9)
	Neurological	88 (7)	95 (7)
	Other	883 (68)	886 (65)
	Ventilator utilization ratio	0.69	0.69	0.77
	Multiple organ dysfunction score, mean (SD)	4.26 (0.38)	3.83 (0.33)	0.0061 ^a
	Febrile respiratory illness rate, cases per 1000 patient days	21.4	17.6	0.099
	Influenza rate, cases per 1000 patient days	1.7	2.1	0.52
	Patient days attributable to cystic fibrosis	68 (1)	277 (4)	<0.0001

Data are number (%) unless otherwise indicated. All p-values calculated using Chi-square unless otherwise noted. SD, standard deviation; TNICU, trauma and neurosurgery intensive care unit; MSICU, medical surgical intensive care unit. ^at-test; ^b2 data points missing

Table 2. Autoregressive model for total antimicrobial use per month measured in defined daily doses per 1000 patient days for intervention and control intensive care units.
Unit	Baseline level	Baseline trend	Change in level (standard error)	p-value	Change in trend (standard error)	p-value
TNICU	1427	1	−375 (94)	0.0009	−9 (15)	0.56
MSICU	1626	10	297(249)	0.25	−30 (35)	0.40
CVICU	1009	−3	169(177)	0.35	−12 (25)	0.63
CICU	591	−7	454 (128)	0.0023	−50 (19)	0.017

TNICU, trauma and neurosurgery intensive care unit; MSICU, medical surgical intensive care unit; CVICU, cardiovascular intensive care unit; CICU, cardiac intensive care unit.

Table 3. Comparison of the use of specific antimicrobial classes and agents measured in defined daily doses per 1000 patient days in the pre- and post-intervention periods. P-values meeting the pre-specified Bonferroni corrected significance threshold are bolded.
Unit	Class or Agent	Pre-intervention mean (SD)	Post-intervention mean (SD)	p-value
TNICU
	antibacterials	1409 (203)	1001 (232)	0.0001
	antibacterials with antipseudomonal activity	349 (89)	195 (62)	<0.0001
	antifungals	24(24)	36 (25)	0.33
	penicillin	15 (21)	19 (24)	0.63
	ampicillin	86 (67)	52 (39)	0.24
	cloxacillin	337(127)	235(128)	0.068
	piperacillin-tazobactam	72 (15)	71 (30)	0.71
	cefazolin	189 (55)	138 (43)	0.028
	ceftriaxone	78 (35)	67 (41)	0.29
	ceftazidime	52 (40)	34 (21)	0.41
	ertapenem	4 (7)	3 (7)	0.47
	meropenem	11 (18)	22 (32)	0.23
	imipenem	9 (10)	16 (14)	0.20
	fluoroquinolones	210 (81)	43 (32)	<0.0001
	TMP-SMX	76 (73)	83 (57)	0.67
	azithromycin	12 (13)	21 (16)	0.20
	vancomycin	203 (110)	134 (52)	0.068
	aminoglycosides	4 (8)	8 (10)	0.18
MSICU
	antibacterials	1547 (274)	1715 (263)	0.16
	antibacterials with antipseudomonal activity	445 (136)	588 (196)	0.028
	antifungals	177 (195)	202 (124)	0.14
	penicillin	55 (48)	44 (68)	0.29
	ampicillin	81 (58)	94 (89)	0.98
	cloxacillin	166 (119)	136(66)	0.93
	piperacillin-tazobactam	197 (53)	200 (34)	0.98
	cefazolin	68 (29)	83 (30)	0.35
	ceftriaxone	97 (27)	113 (40)	0.48
	ceftazidime	28 (21)	32 (29)	0.98
	ertapenem	19 (24)	15 (15)	0.77
	meropenem	28 (48)	55 (51)	0.033
	imipenem	49 (35)	47 (28)	0.84
	fluoroquinolones	138 (49)	180 (60)	0.078
	TMP-SMX	118 (93)	85 (41)	0.63
	azithromycin	101 (42)	132 (53)	0.11
	vancomycin	153 (38)	175 (43)	0.18
	aminoglycosides	29 (34)	55 (43)	0.088

All p-values calculated using Wilcoxon rank sum test. SD, standard deviation; TNICU, trauma and neurosurgery intensive care unit; MSICU, medical surgical intensive care unit; TMP-SMX, trimethoprim-sulfamethoxazole

Table 4. Autoregressive model for use of tracer medications per month in each intervention intensive care unit measured in defined daily doses per 1000 patient days.
Unit	Variable	Baseline level	Baseline trend	Change in level (standard error)	p-value	Change in trend (standard error)	p-value
TNICU
	H2 blockers	611	−8	48 (107)	0.66	10 (15)	0.52
	Proton pump inhibitors	194	10	−113 (114)	0.33	−2 (17)	0.91
MSICU
	H2 blockers	329	1	−82 (53)	0.14	3 (8)	0.75
	Proton pump inhibitors	924	−8	301 (151)	0.061	4 (21)	0.86

TNICU, trauma and neurosurgery intensive care unit; MSICU, medical surgical intensive care unit.

Table 5. Comparison of clinical outcomes in each intervention unit in the pre- and post-intervention periods.
Unit	Variable	Pre-intervention	Post-intervention	p-value
TNICU
	Discharges	n = 1302	n = 1358
	ICU mortality	86 (7)	115 (8)	0.069
	ICU length of stay in days, mean (SD)	4.7 (0.6)	4.6 (0.6)	0.38^a
	Readmission to unit within 48 hours	17 (2)	19 (2)	0.81
MSICU
	Discharges	n = 1247	n = 1307
	ICU mortality	140 (11)	147 (11)	0.99
	ICU length of stay in days, mean (SD)	5.5 (0.8)	5.4 (1.0)	0.76^a
	Readmission to unit within 48 hours	28 (3)	33 (3)	0.62

Data are number (%) unless otherwise indicated. All p-values calculated using Chi-square unless otherwise noted. SD = standard deviation; TNICU, trauma and neurosurgery intensive care unit; MSICU, medical surgical intensive care unit.
^aWilcoxon rank sum test

Table 6. Susceptibility of *E. coli* and *P. aeruginosa* isolates to commonly used antibiotics.
Unit	Organism and antibiotic	Pre-intervention	Post-intervention	p-value
TNICU
	E. coli
	ampicillin	42/83 (51)	40/76 (53)	0.80
	cefazolin	70/83 (84)	61/76 (80)	0.50
	cefotaxime	72/83 (87)	64/76 (84)	0.65
	ciprofloxacin	68/83 (82)	60/76 (79)	0.64
	TMP-SMX	66/83 (80)	61/76 (80)	0.91
	piperacillin-tazobactam	70/83 (84)	63/76 (83)	0.81
	imipenem	85/85 (100)	76/77 (99)	0.48^a
	gentamicin	77/83 (93)	70/76 (92)	0.87
	tobramycin	76/83 (92)	70/76 (92)	0.90
	P. aeruginosa
	ceftazidime	27/30 (90)	22/23 (96)	0.62^a
	ciprofloxacin	27/30 (90)	18/23 (78)	0.27^a
	piperacillin-tazobactam	27/30 (90)	22/23 (96)	0.62^a
	imipenem	30/30 (100)	20/23 (87)	0.076^a
	gentamicin	28/30 (93)	23/23 (100)	0.50^a
	tobramycin	30/30 (100)	23/23 (100)	-
MSICU
	E. coli
	ampicillin	37/95 (39)	30/81 (37)	0.79
	cefazolin	65/95 (68)	59/81 (73)	0.52
	cefotaxime	73/95 (77)	64/81 (79)	0.73
	ciprofloxacin	46/95 (48)	49/81 (60)	0.11
	TMP-SMX	58/95 (61)	43/81 (53)	0.29
	piperacillin-tazobactam	64/95 (67)	59/80 (74)	0.36
	imipenem	93/95 (98)	81/82 (99)	1.00^a
	gentamicin	76/95 (80)	73/81 (90)	0.063
	tobramycin	72/95 (76)	72/81 (89)	0.025
	P. aeruginosa
	ceftazidime	50/64 (78)	42/58 (72)	0.46
	ciprofloxacin	58/64 (91)	41/58 (71)	0.0049
	piperacillin-tazobactam	50/64 (78)	40/58 (69)	0.25
	imipenem	51/64 (80)	41/58 (71)	0.25
	gentamicin	58/64 (91)	48/58 (83)	0.20
	tobramycin	62/64 (97)	56/58 (97)	1.00^a

Data are number of isolates susceptible/total number of isolates tested (%). All p-values calculated using Chi-square unless otherwise noted. A Bonferroni corrected significance threshold of 0.0033 was used. TMP-SMX, trimethoprim-sulfamethoxazole.
^aFisher's Exact test