Quick COVID-19 Severity Index (qCSI)

Why did you develop the qCSI? Was there a particular clinical experience or patient encounter that inspired you to create this tool for clinicians?

COVID-19 is a novel disease and healthcare providers are in need of new clinical tools to help us provide safe and high-quality patient care. As emergency department providers, we saw early reports of risk factors and predictive models for eventual critical illness, but none addressed the uniquely emergency department perspective - “how will this patient do over the next 24 hours”. After all, it does not necessarily  matter to the ED provider whether a patient would decompensate as late as a week into hospitalization. We also saw that many patients who were being admitted to general floors were becoming much more ill within hours. Taken together, we sought to develop a clinical decision aid to help acute care providers anticipate how patients would fare over their first 24 hours of hospitalization.

What pearls, pitfalls and/or tips do you have for users of the qCSI? Do you know of cases when it has been applied, interpreted, or used inappropriately?

As with any clinical decision tool, the qCSI needs to be used with local practices and standards in mind. For example, early in the COVID-19 pandemic, our institutional standards included ICU admission based on nasal cannula oxygen requirement. A few months into the outbreak, however, patients on non-rebreather masks were routinely cared for outside of ICUs. We suggest that this scoring system be used to guide interdisciplinary conversations between emergency department practitioners, general floor, and critical care teams.

What recommendations do you have for doctors once they have applied the qCSI? Are there any adjustments or updates you would make to the score based on new data or practice changes?

The qCSI is not intended to replace clinician judgement. Instead, it is meant to present acute care providers with data to augment their assessments, recognizing the challenges inherent to a new disease process like COVID-19. It is our hope that the simplicity of the scoring system will contribute to both easy application and robust generalizability.

How do you use the qCSI in your own clinical practice? Can you give an example of a scenario in which you use it?

The qCSI is most useful when evaluating patients who we are admitting with some new oxygen requirement, work of breathing, and hypoxia. The dispositions of intubated or high-flow patients are obvious, but there is an opportunity to improve care of a large population of admitted, but non-critical patients.

Any other research in the pipeline that you’re particularly excited about?

The qCSI has strong performance within our initial study, but requires external validation. Similarly, it does not tell us who needs to be hospitalized and who can be discharged.

Anything else you'd like to add?

This is a tool created by healthcare providers with patient care in mind. Please do not hesitate to reach out to us with ideas.

Adrian Haimovich, MD, PhD, is a resident in the department of emergency medicine at Yale University in Connecticut. He is also a member of the Yale Emergency Scholars (YES) program. Dr. Haimovich's primary research is focused on machine learning and big data in the context of emergency medicine.

Andrew Taylor, MD, MHS, is an associate professor of emergency medicine at Yale University in Connecticut. He is also the Director of Clinical Informatics and Analytics at Yale New Haven Hospital. Dr. Taylor’s primary research is focused on machine learning and big data in the context of emergency medicine.