Sustainability of Handshake Stewardship: Extending a Hand Is Effective Years Later

Abstract

Background

Children’s Hospital Colorado created a unique method of antimicrobial stewardship, called handshake stewardship, that effectively decreased hospital anti-infective use and costs in its pilot year (2013). Handshake stewardship is distinguished by: (1) the lack of prior authorization; (2) a review of all prescribed anti-infectives; (3) a shared review by the physician and the pharmacist; and (4) a daily, rounding-based, in-person approach to supporting providers. We sought to reevaluate the outcomes of the program after 5 years of experience, totaling 8 years of data.

Methods

We retrospectively measured anti-infective (antibiotic, antiviral, antifungal) use hospital-wide by unit and by drug for an 8-year period spanning October 2010 to October 2018. Aggregated monthly use was measured in days of therapy per thousand patient days (DOT/1000 PD). The percentage of children admitted ever receiving an anti-infective was also measured, as well as severity-adjusted mortality, readmissions, and lengths of stay.

Results

Hospital-wide mean anti-infective use significantly decreased, from 891 (95% confidence interval [CI] 859–923) in the pre-implementation phase to 655 (95% CI 637–694) DOT/1000 PD in post-implementation Year 5; in a segmented regression time series analysis, this was a rate of -2.6 DOT/1000 PD (95% CI -4.8 to -0.4). This is largely attributable to decreased antibacterial use, from 704 (95% CI 686–722) to 544 (95% CI 525 –562) DOT/1000 PD. The percentage of children ever receiving an anti-infective during admission likewise declined, from 65% to 52% (95% CI 49–54). There were no detrimental effects on severity adjusted mortality, readmissions, or lengths of stay.

Conclusions

The handshake method is an effective and sustainable approach to stewardship.

handshake stewardship, antimicrobial stewardship, antimicrobials, anti-infectives, pediatrics

(See the Editorial Commentary by Goff and Kullar on pages 2333–5.)

The rise of antimicrobial resistance is a major public health concern, leading to the creation of antimicrobial stewardship programs (ASPs). The goal of ASPs is judicious use of antimicrobials, in order to decrease the untoward effects anti-infectives can engender [1]. These include adverse drug events, superinfections such as Clostridioides difficile, poor clinical outcomes, and excessive costs. Methods of stewardship anti-infective review include prior authorization and prospective audit and feedback (PAF) [2, 3]. Typically, PAF includes a review of selected, target anti-infectives; the review is primarily done by a PharmD, with an MD only for backup, and feedback is communicated to the provider electronically or by pager/phone. In 2013, Children’s Hospital Colorado (CHCO) created a unique, expanded PAF model and subsequently reported the first year’s results, demonstrating significant decreases in anti-infective use [4] and costs [5]. To distinguish it from a typical PAF, we termed it handshake stewardship, intending to convey the in-person methodology used to reach agreement. Handshake stewardship is now considered a leading practice by the Joint Commission [6]. In contrast to traditional stewardship approaches, handshake stewardship is distinguished by: (1) the lack of prior authorization; (2) a review of all prescribed anti-infectives (antibacterials, antifungals, antivirals) by all intravenous, enteral, and inhaled routes; (3) an anti-infective review shared by the physician and the pharmacist (each perform half of the review per day) at 2 time points (24- and 48–72-hour order duration); and (4) a daily, rounding-based, in-person approach to feedback by the pharmacist-physician team. This study replicates the initial 2016 analysis with the subsequent 4 years of handshake stewardship data, to assess the durability of this approach.

METHODS

Setting

CHCO is a free-standing, quaternary-care pediatric hospital located in Aurora, Colorado, with 444 total licensed beds in the main hospital, including 82 in the Neonatal Intensive Care Unit (NICU, no well newborn unit), 32 in the Pediatric Intensive Care Unit (PICU), 16 in the Cardiac Intensive Care Unit (CICU), and 48 in Hematology and Oncology. There are approximately 97 000 patient days and 15 000 admissions every year; the hospital provides bone marrow transplants, immunotherapy, and heart, liver, and kidney transplants. CHCO has had an infection control program since 1973; notably, there have not been any large outbreaks during the study period, apart from enterovirus D68 in the summer of 2014, which did not significantly impact antibiotic use [7]. The CHCO ASP team consists of 1 pediatric infectious disease physician and 2 pediatric infectious disease pharmacists, all of whom have completed quality improvement training [8]. The 3–4 total hours per day spent on handshake stewardship is broken down as follows. Each day, 1 physician and 1 pharmacist (the 2 stewards for the day) review all anti-infective orders at the 24- and 48-72-hour time points (each steward reviews half of the orders). This is ~18 000 orders per year, generating ~1800 interventions per year [9], and takes 1 hour per steward per day. A daily meeting with the Microbiology, Infection Control, and Infectious Diseases teams (1/2 hour per steward per day) follows. The stewards then round in-person throughout the hospital to address any patient-specific concerns and outstanding queries generated by the ASP drug review or generated by the primary teams (21 total teams in 2018, 1.5–2 hours per steward per day). The overall acceptance rate for interventions is high [9]. Education is built into these interactions, providing an opportunity for teams to ask questions of the ASP team concerning appropriate anti-infective strategies. The stewards also call positive results for rapid testing on blood cultures and spinal fluids during business hours [10]. Handshake stewardship rounds were done 3 days per week October 2013 to June 2014 and 5 days per week July 2014 forward; this effort was financially supported by CHCO leadership and has incrementally increased with the success of the program, from a 0.2 to a 1.7 combined full-time equivalent.

Study Design

We retrospectively measured anti-infective use hospital-wide for 8 consecutive years, encompassing the pre-implementation, planning, and post-implementation phases of handshake stewardship. To mitigate the effects of seasonal variation in our data, we divided each phase into a full year; thus, each phase began in October and ended in September. The pre-implementation phase spanned October 2010 to September 2011; the planning phase (consisting in the design of handshake stewardship and the development of clinical pathways) [11] lasted 2 years (October 2011 to September 2013); and the post-implementation phases each spanned October to September for the subsequent 5 years (2013 through 2018). We extracted all data for anti-infectives (antibacterials, antifungals, and antivirals; intravenous, enteral, and airway routes) prescribed on the inpatient services from the electronic medical record (Epic, Epic Systems Corporation, Verona, WI). We excluded patients on the Psych/Behavioral Science or Maternal Fetal Medicine/Obstetrics services. These data were aggregated for each month of each phase. Anti-infective use was measured in days of therapy per 1000 patient days (DOT/1000 PD), based on a midnight census. A DOT is defined as the receipt of at least 1 dose of an anti-infective on a single calendar day; for example, if a child received 2 antibiotics for 5 days, this was counted as 10 DOT (5 for antibiotic 1 plus 5 for antibiotic 2).

We also determined the percentage of patients admitted (October 2010 to September 2018) who received any anti-infective at any time during their admission. This was done by calculating the number of patients who received an anti-infective, divided by the total admissions for the same time period, aggregated as described above.

To assess for unintended consequences of decreased use [12], we used deidentified data, including number of discharges, number of fatalities, number readmitted within 30 days, lengths of stay, and the All Patient Refined—Diagnosis Related Group severity scores [13], as reported by CHCO to the Pediatric Health Information Systems database. This database was queried for all CHCO inpatients discharged between 1 October 2010 and 30 September 2018, and excluded patients on the Psych/Behavioral Science or Maternal Fetal Medicine/Obstetrics service. This review was approved by the local Organizational Research Risk and Quality Improvement Review Panel.

Statistical Analysis

The mean use (in DOT/1000 PD) for each year of study was calculated. Results were then assessed by 3 statistical methods. First, an analysis of variance (ANOVA) was used for the overall significant change in mean use across the time periods. Second, a t-test was used to compare mean uses between 2 time periods only: Year 5 post-implementation compared to pre-implementation (to represent overall decreases during the program) and Year 5 post-implementation compared to Year 1 post-implementation (to represent ongoing changes/sustainability since the results first published [4]). In a third statistical method, an interrupted time series study separated data into the pre-implementation, planning, and post-implementation periods, and a slope of DOT/1000 PD was created for each. The rates of change of the slopes were then calculated within and between time periods for all anti-infectives, antibacterials, antivirals, and antifungals, using segmented regression. Seasonal effect (October-March) was considered as a covariate in our models but not included in the final models due to lack of statistical significance. Autocorrelation was adjusted for using an autoregressive (lag 7) model for antimicrobial DOT and autoregressive (lag 1) model for the percentage of patients who received an anti-infective at any point during admission. The appropriate autocorrelation lag was determined by evaluating residuals and comparing Akike information criterion between candidate models. For the statistical analysis on mortality, readmissions, and average length of stay from the CHCO Pediatric Health Information Systems data, linear regression was used to test the changes in rates after adjusting for the proportion of patients in severity categories 3 and 4 [14]. Hypothesis tests were assumed to be 2-sided, with a significance level of 0.05. SAS version 9.4 was used for all analyses.

RESULTS

Hospital-wide, all anti-infective use declined over the post-implementation time period (Tables 1 and 2; Figure 1). The mean DOT/1000 PD in post-implementation Year 5 was 665 (95% confidence interval [CI] 637–694), a decrease of 226 (95% CI 184–267) from 891 DOT/1000 PD (95% CI 859–923) during pre-implementation (P < .01). When post-implementation Year 5 is compared to post-implementation Year 1, the mean total anti-infective DOT decrease was 121 DOT/1000 PD (95% CI 116–125; P < .01). By segmented regression, the estimated rate of change was -2.6 DOT/1000 PD (95% CI -4.8 to -0.4) each month in the post-implementation time period (P = .021).

Table 1.

Hospital-Wide Rates of Change in the Days of Therapy per 1000 Patient Days in the 3 Study Phases

Therapy	Pre-Implementation, 2010–2011	Planning, 2011–2013	Post-Implementation, 2013–2018	Pre-Post Difference
	Q4 2010–Q3 2011	Q4 2011–Q3 2013	Q4 2013–Q3 2018
Anti-infectives	8.5 (-5.5 to 22.5)	-5 (-11.2 to 1.2)	-2.6 (-4.8 to -.4)a	-11.1 (-25.3 to 3)
Antifungal	2.8 (-2.6 to 8.2)	-0.8 (-2.8 to 1.3)	-0.4 (-1 to .3)	-3.2 (-8.6 to 2.2)
Antiviral	0.3 (-3.4 to 4.1)	-0.5 (-2.3 to 1.3)	0 (-.7 to .7)	-0.3 (-4.2 to 3.5)
Antibiotics	3.2 (-4.9 to 11.2)	-3.4 (-6.8 to 0)	-1.9 (-3 to -.8)a	-5.1 (-13.2 to 3)
Patients ever receiving anti-infective during admission, %	-0.1 (-.6 to .4)	-0.2 (-.4 to -.1)a	-0.1 (-.2 to -.1)a	0 (-.5 to .5)

Data were calculated by segmented regression and are shown as slopes (95% confidence intervals).

^aStatistically significant change in slope over indicated time period (P ≤ .05) by segmented regression.

Table 2.

Hospital-Wide Mean Days of Therapy per 1000 Patient Days and Percent of Patients Ever Receiving Anti-Infectives by Study Year

	Pre, 2010–2011	Planning, 2011–2013	Post 1, 2013–2014	Post 2, 2014–2015	Post 3, 2015–2016	Post 4, 2016–2017	Post 5, 2017–2018	P	PPre-Post 5	PPost 1-Post 5
Hospital-wide, all units
All anti-infectives	891 (859–923)	885 (856–913)	786 (752–819)	798 (756–839)	846 (811–881)	727 (677–777)	665 (637–694)	<.01	<.01	<.01
All antibacterials	704 (686–722)	689 (673–706)	624 (596–653)	610 (588–632)	632 (617–647)	570 (541–599)	544 (525–562)	<.01	<.01	<.01
All antifungals	107 (90–125)	109 (100–119)	91 (83–100)	91 (78–104)	115 (105–126)	82 (69–94)	62 (51–73)	<.01	<.01	<.01
All antivirals	73 (64–82)	76 (70–82)	57 (52–63)	81 (69–94)	95 (84–106)	73 (62–84)	56 (51–61)	<.01	<.01	.71
Hospital-wide, all units, specific antibacterials
Vancomycin	100 (91–109)	83 (80–87)	72 (67–78)	67 (61–72)	74 (68–81)	60 (56–64)	47 (44–50)	<.01	<.01	<.01
Meropenem	43 (36–50)	42 (38–46)	33 (28–39)	32 (24–40)	34 (24–45)	24 (15–33)	15 (13–18)	<.01	<.01	<.01
Ertapenem	15 (12–17)	5 (3–7)	1 (1–2)	0 (0–1)	0 (0–0)	0 (0–0)	0 (0–0)	<.01	<.01	<.01
Fluoroquinolonesa	34 (30–37)	28 (25–32)	28 (23–33)	32 (30–34)	25 (21–28)	20 (15–25)	19 (17–22)	<.01	<.01	<.01
Ceftriaxoneb	30 (25–34)	35 (32–38)	40 (36–44)	41 (36–46)	48 (44–51)	48 (43–53)	46 (41–52)	<.01	<.01	.05
Cefepimeb	12 (7–17)	44 (41–41)	46 (43–50)	48 (43–53)	47 (30–65)	57 (42–73)	61 (55–67)	<.01	<.01	<.01
Ceftazidime	38 (31–45)	10 (8–11)	6 (4–8)	7 (5–10)	16 (3–28)	15 (8–22)	12 (10–15)	<.01	<.01	<.01
Piperacillin-tazobactam	5 (4–7)	10 (8–11)	7 (6–9)	8 (6–11)	9 (6–12)	5 (3–6)	3 (2–5)	<.01	.04	<.01
Unit specific, all anti-infectivesc
Pediatric intensive care unit	1578 (1478–1679)	1531 (1426–1636)	1352 (1235–1469)	1329 (1243–1415)	1314 (1226–1401)	1328 (1254–1403)	1179 (1135– 1222)	<.01	<.01	<.01
Neonatal intensive care unit	521 (470–572)	478 (444–513)	450 (403–497)	460 (397–523)	533 (501–564)	407 (341–474)	395 (365–425)	<.01	<.01	.04
Cardiac intensive care unitb	795 (716–873)	796 (738–854)	808 (729–887)	881 (813–949)	946 (812–1079)	816 (693–939)	983 (896–1069)	.004	<.01	<.01
Surgery, 6th Floor	590 (559–621)	563 (546–580)	511 (474–547)	526 (493–558)	529 (496–562)	558 (521–595)	494 (459–528)	<.01	<.01	.46
Hematology/oncology/BMT	2882 (2507–3257)	3021 (2602–3440)	1803 (1673–1934)	1810 (1655–1964)	2035 (1939–2132)	1565 (1420–1709)	1379 (1245–1512)	<.01	<.01	<.01
General medicine	653 (613–694)	599 (568–630)	534 (506–562)	535 (496–573)	522 (487–557)	455 (426–484)	444 (398–490)	<.01	<.01	<.01
Pulmonary/cystic fibrosis	829 (700–959)	947 (884–1011)	880 (778–981)	788 (729–847)	778 (719–837)	669 (578–761)	647 (579–716)	<.01	.01	<.01
Patients ever receiving anti-infective during admission, %	65 (64–66)	61 (60–62)	57 (55–59)	54 (53–55)	55 (53–56)	52 (50–53)	52 (49–54)	<.01	<.01	<.01

Data are shown as mean (95% confidence interval). Means were calculated by ANOVA, with P value for overall trend (all phases) and t-test analysis P_{Pre-Post 5} and P_{Post 1-Post 5} comparisons.Abbreviations: ANOVA, analysis of variance; BMT, bone marrow transplant.

^aIncludes ciprofloxacin and levofloxacin (moxifloxacin is not on formulary and is rarely used at our institution).

^bDenotes that the results were significant due to increased use overtime (overall trend).

^cOverflow unit was excluded from subanalysis due to inability to attribute to a particular service line.

Figure 1.

Segmented regression comparing the rate of change (slope) pre-implementation (1 year, Q4 2010–Q3 2011), during planning (2 years, Q4 2011–Q3 2013), and post-implementation (5 years, Q4 2013–Q3 2018). Data show results from (A) all anti-infective, (B) antibacterial, (C) antifungal, and (D) antiviral medications, and (E) the percentage of children ever receiving anti-infectives during admission. There was a decrease of 2.6 DOT per 1000 PD each month for all anti-infectives and a decrease of 1.9 for antibiotics in the post-implementation period. Abbreviations: Abx, anti-infective; DOT, days of therapy; PD, patient days; Q, quarter.

The decrease in all anti-infective use was largely driven by a significant decrease in antibiotic use, which went from a mean of 704 DOT/1000 PD (95% CI 686–722) to a mean of 544 DOT/1000 PD (95% CI 525–562) in post-implementation Year 5, a decrease of 161 (95% CI 132–189) DOT/1000 PD (P < .01). When post-implementation Year 5 is compared to post-implementation Year 1, the mean decrease was 80 DOT/1000 PD (95% CI 71–90 DOT/1000 PD; P < .01). The rate of antibacterial use change during the post-implementation period was -1.9 DOT/1000 PD (95% CI -3 to -0.8; P < .01) per month. Of these 544 antibiotic DOT/1000 PD for post-implementation Year 5, 411 DOT/1000 PD were intravenous; this trended from 583 pre-implementation, 557 in the planning period, and 504, 486, 530, 458, and 411 DOT/1000 PD in post implementation Years 1–5, respectively (P < .01 for ANOVA and t-tests). Changes in antifungal and antiviral use were non-significant using a segmented regression analysis, but were significant using ANOVA and t-test comparisons. Year-by-year results are summarized in Tables 1 and 2 and Figure 1.

When assessing individual anti-infective agents, vancomycin, meropenem, fluoroquinolones (eg, ciprofloxacin and levofloxacin), ceftazidime, and piperacillin-tazobactam mean use significantly (P < .01) decreased from pre-implementation to post-implementation Year 5 by 53 DOT/1000 PD, 28 DOT/1000 PD, 15 DOT/1000 PD, 26 DOT/1000 PD, and 2 DOT/1000 PD, respectively (Table 2). Conversely, ceftriaxone and cefepime mean use increased by 16 DOT/1000 PD (P < .01) and 49 DOT/1000 PD (P < .01), respectively (Table 2).

By unit, comparing pre-implementation to post-implementation Year 5, the PICU and the NICU both experienced significant decreases in overall mean anti-infective use. The PICU use decreased by 396 DOT/1000 PD (P < .01) and the NICU by 126 DOT/1000 PD (P < .01). The CICU increased anti-infective use by 188 DOT/1000 PD (P < .01; Table 2). All other units in the hospital experienced statistically significant declines in mean use over time (Table 2).

The percentage of children ever receiving anti-infectives at any point during admission (including medical and surgical prophylaxis) decreased by 14% (95% CI 12 –16; P < .01) in the 8-year period (Figure 1E). There was a 6% decrease from Year 1 to Year 5 of handshake stewardship, which was significant by ANOVA (95% CI 2–8; P < .01), but not by segmented regression.

After adjusting for patient severity (which increased over time), changes in rates of mortality, rates of readmission, and average lengths of stay over time were evaluated by linear regression (Table 3). No significant changes were present.

Table 3.

Pediatric Health Information Systems Hospital Outcomes Over Time

Outcome	Estimate (95% CI)	P Valuea
Readmissions	0.29 (-.43 to 1)	.3521
Fatalities	0 (-.07 to .08)	.9074
Average length of stay	0.06 (-.23 to .35)	.6221

Data were adjusted for proportion of patients with severity levels 3 and 4.Abbreviation: CI, confidence interval.

^aLinear regression was used to test the change in rates after adjusting for the proportion of patients in the highest severity category (severity 4).

DISCUSSION

In our free-standing, quaternary-care pediatric hospital, handshake stewardship reduced anti-infective use by 25% over a 5-year period, endorsing the methodology’s efficacy and sustainability. These ongoing data build upon previous work demonstrating efficacy in the first year of implementation [4, 9] in association with decreased anti-infective costs [5] and increased infectious diseases consultations [10, 15]. This method differs from most published methods of PAF, in that it is an in-person approach with daily face-to-face interactions between the stewards (pharmacist-physician team) and providers (21 teams) with reviews of all anti-infectives (not just targeted) given by all major routes of administration (not just intravenous) at 2 time points (24 hours and 48–72 hours), and also differs in that it lacks prior authorization. To assure the initial year’s results did not wane or reverse due to theoretical steward burnout or provider intervention fatigue, an assessment of the sustained impact of this approach was undertaken. The results solidify this approach as an alternative to traditional prior-authorization or PAF.

Perhaps the most important measure to monitor as a reflection of the health of the program and ongoing areas for improvement are trends in anti-infective use (Tables 1 and 2; Figure 1). The 25% decrease in anti-infective use was largely driven by a 23% decrease in mean antibacterial use. Antibacterial use likely decreased because we intervened most often on these medications [9] and because of a 14% decrease in the percentage of admitted children ever receiving an anti-infective. The latter phenomenon was likely a combination of clinical pathways (developed with the ASP) [16]; medical trainees and providers who have now “grown up” in an environment of handshake stewardship; and the Hawthorne effect [17], in which a provider may consciously or subconsciously consider the ASP program when deciding to prescribe.

Antifungal use decreased over time for a total decrease of 43%; this was not significant in the segmented regression model, which measures the rate of change of the slope in Figure 1. The regression model has comparatively few points in the pre-implementation period, limiting confidence in the pre-implementation period slope. Thus, to aid in the interpretation of these data, we also analyzed by ANOVA/t-test (Table 2), which were statistically significant. Antiviral use was also decreased pre-implementation to post-implementation, but this was not consistent over time. We postulate that antiviral and antifungal use are more difficult to impact for 2 reasons. First, these agents are most commonly prescribed as medical prophylaxis or preemptive therapy in our transplant and oncology patients [18, 19], which are largely dictated by protocols (national and local). Second, it is difficult for a steward to assess whether a patient meets a certain transplant or oncology protocol criteria for antifungals and antivirals; thus, these medications accounted for only 4 and 5% of our interventions, respectively [9], and discontinuation was an uncommon intervention type (internal data), so DOT were not affected. With this analysis, however, there is renewed local interest in reconsidering our intervention strategies regarding antifungals and antivirals.

Regarding particular drugs, some trends are notable and consistent with local initiatives. For example, the use of broad-spectrum gram-negative drugs, such as meropenem, piperacillin-tazobactam, and ceftazidime, decreased, while the use of cefepime increased; this is consistent with a change in the institutional protocol for the treatment of oncology, bone marrow transplant, and liver disease patients, along with the lack of cefotaxime availability since 2016. Ceftriaxone use increased, reflecting both changes in our local appendicitis guideline (ceftriaxone and metronidazole replaced ertapenem [20]) and the implementation of replacement therapy with ceftriaxone in our PICU, in an effort to decrease vancomycin use for pneumonia (currently a target for ongoing improvement). By unit, decreases were observed hospital-wide, except in the CICU; we are currently analyzing this change in conjunction with our local CICU liaisons and our data on mandatory, provider-selected clinical indications for use, to understand the increase and examine it for appropriateness [18, 19].

It is important to place our results in the context of other stewardship literature. This, however, is difficult, as most published studies have not used hospital-wide anti-infective or antibacterial DOT/1000 PD as a primary outcome; rather, they have reported only intervention data, use of only certain drugs/routes of administration, or stewardship only in certain units [21–29]. While such studies contribute enormously to our understanding of anti-infective use management, they do not represent the overall global impact of a stewardship program [30]. Reporting hospital-wide anti-infective or antibacterial results is an important addition to understanding which stewardship methodologies may be effective in the real-world setting [31–38]. In this context, handshake stewardship performs very well, particularly given the complexity of patients at our institution (42% of patients had a severity score of 3 or 4 in 2018). In fact, ours is the largest percentage decrease for hospital-wide antibiotic use described (23%). In comparing the final year DOT/1000 PD (in contrast to percentage decreases), Jenkins et al [34] described a lower nadir, at 457 DOT/1000 PD (compared to our 554 DOT/1000 PD), though their practice setting was quite different (largely adult safety net hospital without cardiac surgeries or transplant services). Willis et al [33] described a nadir of 527 DOT/1000 PD, but this included only intravenous antibiotics at their pediatric institution; ours, for comparison, is 411 intravenous antibiotic DOT/1000 PD.

Our study presents certain limitations. First, we present limited measures to address unintended consequences from decreased use. Ideally, in addition to decreased anti-infective use, we would also present data showing the lack of harm due to decreased use, increased appropriate use, and better outcomes [12, 39]. We attempted to address this with information on mortality rates, readmissions, and average lengths of stay, and found no differences over time (Table 3). Though commonly presented in stewardship manuscripts, these measures are impacted by innumerable variables. We have also examined outcomes in targeted clinical entities, such as musculoskeletal infections, appendicitis, and the implementation of provider-selected clinical indications for antimicrobials, all demonstrating benefits [18–20, 40]. Infectious diseases consultations were not adversely impacted by handshake stewardship and, in fact, increased [42]. Second, we do not present Clostridioides difficile infection rates, a common stewardship measure, due to changes in testing and reporting over time at CHCO with the adoption of various new testing platforms and protocols over time; notably, in our pilot year, prior to testing changes, rates did decrease [4]. Third, this study was done at a single center, limiting generalizability. A multicenter study would be necessary to assess the method’s efficacy and adoptability in different settings.

Handshake stewardship is an effective and sustainable antimicrobial stewardship strategy. Though it is time intensive, requiring 3–4 hours per day each for a PharmD and an MD for a 444-bed quaternary care hospital with the review of all drugs and discussions with all teams, the face-to-face interactions may be perceived to be more efficient and effective than electronic or pager/phone interactions [6, 42–47]. The stewards have not found the handshake stewardship method to cause ASP fatigue, and instead have appreciated the opportunity to interact daily with their colleagues, find optimal therapy options for patients, and investigate new areas of multi-disciplinary collaborations to improve patient care. Handshake stewardship is also an effectual segue to diagnostic stewardship, cross-disciplinary clinical decisions and collaboration, and improved hospital culture. For programs that are not resourced for hospital-wide implementation, alternatives such as selected drug reviews (rather than all anti-infectives at 2 time points) or targeted units (such as the intensive care unit, which theoretically has the advantage of downstream effects on prescribing hospital-wide) could be considered for this methodology. Every setting, however, will necessitate a careful design and stakeholder recruitment; for us, this process took 2 years (planning period). Though we are a pediatric institution, implementation at adult care facilities or adult-pediatric facilities may also be effective, provided the ratio of stewardship resources per hospital beds is similar. In terms of value (increased quality for less cost), handshake stewardship should be attractive to hospital leadership and is financially justifiable; the first year post-implementation pharmacy drug cost savings were 1 to 2 million dollars, compared to the planning period [5]. Relatively few studies comparing methods of inpatient stewardship have been published [29, 48, 49] and none have included handshake stewardship, though other centers are using this method (Sarah Parker, personal communication with Jason Newland, Scott Weissman, Natasha Nakra, Preeti Jaggi, Ritu Banerjee, and Michelle Mitchell; 30–31 May 2019) [6, 35, 46]. Further study is needed to assess whether this method is superior to others in terms of efficacy, cost savings, and ideally, outcomes; we hope these data will compel such an evaluation on a larger scale.

Notes

Acknowledgments. The authors thank Tim Jenkins, Holly Frost, and Chris Czaja for their careful reading and advice about this manuscript, and the leadership and providers at Children’s Hospital Colorado for their ongoing support.

Financial support. This work was supported by the Clinical Operational Effectiveness and Patient Safety Small Grants Program for Children’s Hospital Colorado and the School of Medicine.

Potential conflicts of interest. The authors report no conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.