ANGELA KENISTON, MSPH
Director, Data and Analytics/Instructor, University of Colorado

Commonly used discharge communication workflows hinder timely and efficient discharge. Studies exploring the use of the EHR for discharge planning have been limited to electronic reports constructed from EHR data elements, including barriers to discharge documented at admission, care management data, and discharge criteria or other targeted interventions such as improving discharge summaries for patients or medication reconciliation at discharge. To address these deficits, we developed an innovative EHR tool to facilitate communication in real-time between hospitalists and other clinicians about discharge readiness and barriers to discharge.

All clinicians who were scheduled to be on an inpatient Hospital Medicine service were trained and asked to use the Discharge Today tool with all patients assigned to their team.

This study is a prospective, single center, pragmatic, interrupted time series study. Clinicians were asked to update patient discharge readiness (Definite, Possible, Tomorrow, In 24-48 hours, > 48 hours) and any barriers to discharge, every morning and anytime patient status changed. Primary outcomes were time of day the clinician enters the discharge order, time of day the patient leaves the hospital, and hospital length of stay. Secondary outcomes were proportion of patients with a discharge order before 11 am and proportion of patients discharged before 11 am. We used linear mixed modeling and generalized linear mixed modeling with team and discharging provider included in all models to account for patients cared for by the same team and the same provider.

We found that, after adjusting for pre-specified confounders and effect modifiers, for every one patient increase in the morning census, there was a statistically significant reduction in the time of day the discharge order was entered into the EHR by the discharging physician (3 minutes per patient (95% CI: 42 seconds, 6 minutes), p=0.0245) for the pilot implementation period compared to the pre-implementation period, though not for the post-implementation period (p=0.4526). We also found a statistically significant reduction in hospital length of stay for the pilot implementation period compared to the pre-implementation period (56 minutes (95% CI: 52 minutes, 1 hour), p=0.0047), though not for the post-implementation period (p=0.4342).

Our Discharge Today tool is a real-time communication tool, created by hospitalists and other healthcare professionals who participate in discharge planning to not only document and communicate what tasks need to be completed before a patient can be discharged but also to help clinicians, nursing, and other staff to prioritize their work in real-time. Our analysis suggests this tool is useful for improving discharge timing, particularly as the number of patients being cared for by a team increases.

BRAD MORSE PhD, MA on behalf of MICHAEL HO, PhD
Research Instructor, University of Colorado Anschutz Medical Campus – D2V

The National Academy of Medicine defines a Learning Health System as a health system that assembles, analyzes, and interprets data. Findings are leveraged to adapt and improve delivery of patient-centered care. A challenge for a Learning Health System is responsive learning and adapting, i.e., thinking differently to address what seem like problems that can be managed with the application of traditional methods. To assist UCHealth and Children’s Hospital of Colorado with this process, the Navigation Lab (NavLab) performs interdisciplinary evaluation of health system clinical programs and initiatives.

The University of Colorado Anschutz Medical Campus/UCHealth Learning Health System. The NavLab works with physicians and departments serving the many patients that utilize the network of hospitals associated with the Learning Health System. Due to the unique setting in which the NavLab works, our projects engage a diverse spectrum of communities within the general population.

The NavLab’s multidisciplinary team includes a health economist, systems engineer, biostatisticians, qualitative analysts, clinicians, analytics developers, user experience (UX) designer, and a program manager. The NavLab utilizes an interdisciplinary approach for program evaluation including comparative effectiveness analysis, economic evaluation, workflow and staffing assessment, and user-center design. The team engages health system partners to identify opportunities for Quality Improvement (QI). Stakeholder engagement is critical for all evaluations in terms of defining the scope of the QI and how change, effectiveness, or efficiency will be measured.

NavLab evaluations include interdisciplinary outcomes related to care quality, efficiency, and cost savings from multiple perspectives to improve the healthcare system.

The NavLab’s next steps include expanding use of economic modelling, workflow evaluations and simulations, user-center design, and predictive analytics with operational and clinical partners.

ERIN LEISTER CHAUSSEE
PhD Candidate, Colorado School of Public Health

In stepped wedge (SW) designs, differing cluster-level characteristics or individual-level covariate distributions that differ by cluster can lead to imbalance by treatment arm and potential confounding of the treatment effect.

SW cluster randomized trials

Adapting a method used in cluster-randomized trials, we propose a covariate-constrained randomization method to be used in SW designs. First, we define a balance metric to be calculated for all possible randomizations of cluster order for a given SW design (denoted BSW). The resulting distribution of this balance metric across all possible randomizations is used to select a candidate set of randomizations with acceptable covariate balance, for example, the best tenth percentile (P10) of the BSW distribution. One cluster order is selected at random from this candidate set to be used as the cluster order for treatment implementation. In a simulation study, we implement the covariate-constrained randomization procedure and computed treatment effect estimates and average absolute bias, and estimates of type I error and power. We used these outcomes to evaluate differing cutoffs of the BSW distribution used to define the candidate set and various analysis methods, under varying SW design and confounding settings.

We observed optimal statistical properties when the balance metric was used to exclude a small set of potential randomizations with the highest level of imbalance, and when analysis methods were adjusted for the potential confounders (see Figure for average absolute bias by BSW cutoff results). The covariate-constrained randomization was most beneficial in settings with a small number of clusters, low intra-class correlation, a low number of participants per cluster, and in the presence of cluster-level confounding variables.

We recommend researchers using the SW design carefully consider potential confounders, both cluster-level and individual-level, prior to cluster order randomization and consider a covariate-constrained randomization if appropriate. Treatment effect estimation should be adjusted for these potential confounders, and other covariates associated with the outcome of interest.