SHM Converge Scientific Abstract ePoster Gallery-Do Customized Electronic Health Records-Based Sepsis Alerts Help Clinicians Treat Suspected Sepsis More Quickly?

Pdf Summary

A customized Epic-based sepsis alert identified most suspected sepsis patients in the ED, but it did <strong>not</strong> shorten time to antibiotic ordering.

<strong>Essential results</strong>
- <strong>Study population:</strong> 7,084 ED patients with suspected sepsis.
- <strong>Alert timing:</strong>
  - <strong>78%</strong> (5,487) had the alert fire <strong>before</strong> the first antibiotic order.
  - <strong>22%</strong> (1,597) had the alert fire <strong>after</strong> antibiotics were already ordered.
- <strong>Timing impact:</strong> Patients with the alert firing <strong>before</strong> antibiotics actually received antibiotics <strong>later overall</strong> than those whose alert fired after antibiotics:
  - Median time to antibiotics was <strong>0.9 hours later</strong> in the “alert before antibiotics” group.
  - The alert itself fired much earlier in this group (**0.2 hr vs 3.7 hr**).

<strong>Key interpretation</strong>
- Differences in demographics and clinical characteristics did <strong>not</strong> explain why antibiotics were ordered later in the group with earlier alerts.
- The alert seemed more likely to fire <strong>after treatment had already started</strong> when initial vitals did <strong>not</strong> meet SIRS or qSOFA criteria.
- Overall, the customized EHR sepsis alert may be <strong>more timely and accurate</strong> than the proprietary Epic Sepsis Model, but it <strong>did not expedite antibiotic ordering</strong>.

<strong>Selected comparison table</strong>

| Characteristic | Alert before antibiotics | Alert after antibiotics | p-value |
|---|---:|---:|---:|
| Patients | 5,487 | 1,597 | — |
| Median age | 64 | 64 | 0.679 |
| Female | 45.2% | 45.7% | 0.803 |
| Median Elixhauser index | 11 | 11 | 0.004 |
| Emergent acuity | 59.7% | 47.1% | 0.001 |
| Alert triggered in ED | 96.9% | 82.3% | 0.001 |
| SIRS2 on triage | 48.0% | 9.8% | 0.001 |
| qSOFA2 on triage | 12.2% | 6.5% | 0.001 |
| In-hospital mortality | 8.8% | 7.1% | 0.032 |

Asset Subtitle

Priya A. Prasad

Meta Tag

Author List David Shimabukuro, Kirsten N. Kangelaris, Margaret C. Fang, Mary Sullivan, Priya A. Prasad

Category Research

Concept EHR Alert

Concept Sepsis

Concept Antibiotic Ordering

Concept Antibiotic administration

Concept Emergency Department

Distinguished Non-Finalist

Presenter Organization University of California San Francisco

Presenting Author Priya A. Prasad

Track Outcomes Research

Keywords

sepsis alert

epic EHR

emergency department

antibiotic ordering

suspected sepsis

clinical decision support

SIRS criteria

qSOFA

timing analysis

electronic health record

EHR Alert

Sepsis

Antibiotic Ordering

Antibiotic administration

Emergency Department