Vital Statistics Consulting Vital Statistics Consulting

Case Study: Optimization of Emergency Department Metrics

The Project: One of our current projects evaluates an intervention meant to optimize the time that Emergency Department (ED) physicians may provide high-quality, patient-centered care. A group of our consultants, with assistance from two outstanding medical student/interns at VSC, are studying data from a community hospital in New England that implemented a 6-month trial whose goal was to use medical scribes to ease the clerical/administrative burden of its ED doctors. Our first objective is to determine whether employing scribes shortened the average time for a patient to be treated, discharged, admitted to the hospital, or transferred to another hospital. These outcomes, called process and performance metrics, are essential to hospitals, because they are used in reimbursement calculations by the Centers for Medicare & Medicaid Services (CMS), which view such non-clinical metrics as vital to capturing the totality of the ED experience. We also have a secondary objective, and that is to evaluate whether better process outcomes enhance patient experience and quality outcomes. In other words, if scribes allow clinicians to focus more of their time on patients (and less on data entry and monitoring of diagnostic testing), is this additional time reflected in patients’ subjective (i.e. satisfaction) and objective (e.g., likelihood of returning to the ED) experiences?

How We Address Key Knowledge Gaps: Clinical and organizational changes that optimize process and performance metrics are critical to reimbursement from public insurance programs. Even so, predictors of such metrics, as well as the influence of process outcomes as mediating factors in healthcare quality and patient safety, have received insufficient scientific attention, and therefore remain largely unknown.

How Our Findings May Improve Outcomes & Inform Decision-making: While it is common practice for hospitals to collect extensive metrics for reporting purposes, these data are vastly underutilized. The overarching purpose of this project is to explore the rich data collected in the ED, so as to better understand factors that are associated with process and performance outcomes, and to link those outcomes to patient safety and experience. While special emphasis is paid to the influence of scribes, we are also focusing on other “modifiable” predictors, which may help target organizational interventions. In other words, we are transforming data to actionable evidence, which, in turn, supports to organizational decision-making that may support timeliness, delivery, and quality of care.

What Are the Design, Setting & Patient Population: This project is a non-experimental investigation that involves time-dependent analyses of EMR data. The units of analysis are the clinical encounter and the patient. The setting is a non-profit, community hospital located in New England with an ED volume of over 30,000. The population is all ED patient encounters that took place over a 12-month period, 6 months prior to the employment of scribes and during the 6-month scribe trial period.

What Are the Key Measures:Primary outcomes include AHRQ process and performance measures for discharged patients (i.e., door to doctor time, doctor to discharge time, ED length of stay (LOS)), for admitted patients (i.e., door to doctor time, doctor to decision to admit, ED LOS, daily boarding hours), and transferred patients (i.e., door to doctor time, doctor to decision to transfer time, decision to transfer time to transfer accepted time, transfer accepted to left ED time, transfer accepted to left ED time, ED LOS). Secondary outcomes include patient mortality, readmission, length of stay, and satisfaction (where possible). Explanatory variables (predictors) include the presence of a medical scribe associated with a given encounter, patient-level variables (e.g., age, reason for visit, HCC score/ICD codes/DRGs/diagnoses), temporal factors (e.g., day of week/time of day/shift type), organizational factors (e.g., provider mix, shift change/patient hand-off, case management/care coordination), and related laboratory/diagnostic variables (test utilization, pharmacy/lab times, radiology times, ancillary turnaround times).

How We Are Analyzing the Data: ED data have been extracted from the hospital’s EMR, and are currently being analyzed with multivariable (risk-adjusted) regression models. Mediation models will then be estimated to determine whether process and performance variables are in the causal pathway between predictors and patient outcomes.