Clostridioides difficile infection (CDI) is the leading cause of hospital-acquired infections (HAI) in the United States.Reference Magill, Edwards and Bamberg1 Although multiple interventions have been shown to reduce CDI, the adoption of these evidence-based practices remains suboptimal, and the burden of CDI remains high.Reference Yanke, Moriarty, Carayon and Safdar2 There is a pressing need to develop strategies that bridge the gap between the available evidence and clinical practice to reduce harm from CDI.
The ‘Agile Implementation’ (AI) framework was used to reduce central-line–associated bloodstream infections (CLABSIs) at our institution.Reference Azar, Kelley and Dunscomb3 In study described here, we used the AI model to achieve reductions in CDI.
Methods
Setting
The study was conducted in 2 large academic hospitals in the Midwest between October 2016 and December 2018.
AI model
The AI model was developed at the Indiana University School of Medicine Center for Innovation and Implementation Science. It recognizes the healthcare system as a complex adaptive system (CAS) and acknowledges the importance of individual barriers and enablers in influencing implementation outcomes. It therefore incorporates the CAS theory and the social cognitive theories of behavioral economics to facilitate implementation.
The first step is to confirm opportunities and engage stakeholders. The second step is to identify evidence-based solutions. The third step is to develop an evaluation strategy in which process and outcome measures are defined and a termination plan is agreed upon if these measures are not met. An interdisciplinary team then converts the evidence-based solution into a minimum viable service, the minimum specifications required to implement the intervention effectively, while taking into consideration the local environment in which it is applied. Implementation occurs in “sprints” (repeatable tests of change) with modification through continuous feedback. Impact on the organization is also measured. If the process results in the desired outcome, then the team develops a minimally standardized operating procedure, an implementation blueprint that can be utilized by other departments.
Implementation
An interdisciplinary team met in October 2016 to create a CDI reduction strategy. A gap analysis was conducted and opportunities were identified in environmental services (EVS), antibiotic stewardship, hand hygiene, and CDI testing practices. Four subteams were created to address each identified opportunity using evidence-based guidelines.Reference McDonald, Gerding and Johnson4
The subteams met monthly to strategize and modify interventions based on feedback from the implementation sprints. Localization of the implementation of the minimal viable solutions was left to individual units where appropriate.
Measurements
Outcome measures. The primary outcome was the CDI rate, as defined by the National Healthcare Safety Network. A standardized incidence ratio (SIR) was calculated based on the Centers for Disease Control definition.
Implementation outcomes. Data regarding the implementation were obtained from direct feedback received during team meetings. Changes in practice habits were assessed by tracking antibiotic days of therapy (DOT), acceptance rate for antibiotic stewardship recommendations, black-light data for monitoring of environmental cleaning, number of CDI tests ordered, and hand hygiene compliance data.
System outcomes. Non-CDI hospital-acquired infections were monitored.
Statistical analysis
We used Poisson regression to test whether the outcomes (incidence rates of infections, SIR, DOT) differed between the baseline and the implementation period. Time period was included as a fixed effect in all models and an offset equal to the log(exposure) to account for the differential exposure time across periods. Relative risks (RRs) and 95% confidence intervals (CIs) were reported. Logistic regression was used to determine whether hand hygiene and EVS compliance differed between the baseline period and the implementation period. The odds ratios (OR) and 95% CIs were reported. Linear regression was used to determine change in CDI testing. All analyses were conducted using SAS version 9.4 statistical software (IBM, Armonk, NY).
Results
The CDI rate decreased from 12.47 to 7.23 cases per 10,000 patient days during the implementation period (RR, 0.58; 95% CI, 0.51–0.66; P < 0.001). Data on implementation and system outcomes are summarized in Table 1.
Table 1. Calculated Rates (95% confidence intervals) of Hospital-Acquired Harms and Regression Results
Note. CLABSI, central line-associated bloodstream infection; CAUTI, catheter-associated urinary tract infection; CDI, Clostridioides difficile infections; DOT, days of therapy; SIR, standardized incidence ratio (observed/expected infections); RR, relative risk; OR, odds ratio. Poisson and logistic regression, respectively, were used to compare implementation to baseline periods, for infections and compliance procedures.
a Per 1,000 central-line days.
b Per 1,000 Foley catheter days.
c Per 10,000 patient days.
The stewardship recommendations acceptance rate started at 45% in January 2017 and increased to an average of 84% in 2018. CDI testing orders were static in 2017 then decreased by 38.5% in 2018, coinciding with the implementation period of the intervention.
Discussion
We utilized the AI model for implementing evidence-based CDI guidelines, and we observed a 42% reduction in CDI. This model focuses on the characteristics of the users and adopters of implementation as well as the organizational context in implementing these processes. This process results in the engagement of key stakeholders and localization of measures to adapt to the local environment within the organization, leading to increased and rapid adoption of practices and sustainability in processes.Reference Azar, Kelley and Dunscomb3 By viewing the healthcare system through the lens of complexity science, taking into account nonlinear and unpredictable interactions among its individuals, monitoring for unintended consequences, and using minimal viable solutions to allow for localization,Reference Pype, Mertens, Helewaut and Krystallidou5 the AI model successfully facilitated the implementation of evidence into practice in the real-world care setting, thereby leading to a significant reduction in CDI. In this model of implementation, a termination plan is developed initially, which eases stakeholders’ loss aversion during change and reduces waste. Additionally, the model focuses on local adaptation of solutions, and on monitoring of consequences of the implementation on other system outcomes.
This is a single-center study and our results may not be generalizable to other institutions. We did not perform a cost–benefit analysis. We successfully utilized the AI framework to impact clinician behavior and to implement evidence-based practices to prevent CDI. We hope that others can learn from our journey.
