How Can We Improve What We Don’t Measure? Selecting Key Medical Metrics and Wrangling Them Out of the Database [Abstract]

Abstract

Introduction

Shelters and high-volume community clinics are among the closest veterinary parallels to human hospitals. Large numbers of patients move through these systems, and systems-level approaches to quality control, population health  and efficient and judicious use of resources are needed. Yet, unlike human medicine, veterinary medicine has few tools to objectively monitor the quality or outcomes of care at a population level—whether in shelters, community clinics, or private practice. The metrics we can readily extract from our software systems—intakes, outcomes, live release rates, and length of stay—do not capture key medical measures such as infectious disease incidence, surgical outcomes, mortality in vulnerable populations, and wait times for care. In human medicine, a system without such measures would be unthinkable.

Description

Several years ago, our group began developing automated medical metrics for our shelter. Our goals were to monitor quality of care, identify problems, celebrate successes, measure the impact of protocol changes, and create benchmarks for improvement. Prior to developing automated reports, medical concerns were identified and investigated on an ad hoc basis. These investigations were time-consuming, and were difficult to replicate and interpret without knowledge of baseline data.

Creating tailored reports has required careful decision-making about metric selection, definitions of terms, and technology selection to ensure the reports are accurate, versatile, and actionable. Each chosen metric reflects outcomes that are influenced by multiple functional areas, allowing for broad monitoring of the system as a whole, or large segments of it. For example, the incidence of feline URI is influenced by animal source, population density, stress, cleaning protocols, vaccination, housing, and husbandry. When problems are identified, the investigation determines which functional areas are involved.

Outcomes

We have developed five automated monthly metrics from our shelter’s medical databases (PetPoint and Ezyvet). These are: cumulative incidence of URI, cumulative incidence of diarrhea, kitten mortality, surgical wait times and surgical complication rates. The detailed reports are automatically emailed monthly to the medical leadership team, and are also simplified as an online dashboard for staff.  The reports can be expanded or modified as needed. For example, a cumulative incidence report can be run for any medical condition that is routinely recorded in the database.

The primary limitation of this project is that it was not possible to create a “cookie cutter” solution that could be shared with other shelters, as we had originally hoped. However, we have been able to develop a toolkit detailing our methodology, that other shelters can build from. It is important to note that data analytics skills are essential for any project like this.

Conclusion

Measurement of non-medical metrics is routine for well-run shelters, and their value is clearly understood. Our hope is that medical metrics will become a standard addition to existing metrics.