The case, explained simply
Hospital warning system · external evaluation
A widely used system still needs to be tested where it will be used.
What actually happened
A University of Michigan external study evaluated the Epic Sepsis Model retrospectively. At the studied threshold, it missed 67% of sepsis cases while identifying 18% of hospitalisations for potential alerts. The study raised concerns about missed cases and the burden of checking alerts. It evaluated a particular historical model and setting, not every current Epic product.
What the evidence does not tell us
This was a retrospective evaluation at one academic medical centre. It did not establish that deploying the alerts caused patient harm. The findings cannot be used as a current clinical recommendation or as an assessment of later versions.
What a Greek organisation can learn
- For a hospital considering a system, involve clinical, data and safety leads before a purchasing decision. Ask which version and patient population were tested.
- Have qualified teams examine local performance, missed cases and workload before any use that affects care. A promising accuracy figure alone is insufficient.
- Set formal oversight, change review and stopping arrangements. Keep patient-care decisions within the appropriate clinical and regulatory process.
Before spending money
The wider business lesson is simple: check missed problems and the work created by false alarms, not just correct predictions. In healthcare, specialist evaluation is essential.
Read the original evidence
Wong et al. · JAMA Internal Medicine ↗The practical steps above are AIStrategy’s editorial suggestions. They are not reported actions or promised results from the organisation featured. Photography illustrates the subject unless its caption identifies the actual facility.