No on-site doctor: Dental student died in ICU overseen by remote tele-health MD

This note in the article caught my eye, and I think I have that explanation:

> ”It also appears from the sparse ICU records that the ICU RN was only contacting the tele-ICU service for sedation orders as [their] condition deteriorated in the ICU”

It reads as though the on-site staff weren’t trained to detect and report conditions that are normally witnessed by the prescribing doctor. But. If this follows the same trends as our industry, this hospital’s innovative idea was to shift the workload of monitoring and reporting fully away from expensive doctors and onto cheaper nurses with oncall-remote escalation. The IT prodops equivalent here would be, perhaps, “the lower-cost engineer only reported a couple of sporadic latency spikes to the higher-cost oncall tech lead, who suggested checking swap and transaction logs and went back to other work, and a few hours the raid array died after losing a third drive”, and the post-mortem would look rather like what this article contains. We tolerate that sort of thing in tech when lives aren’t at stake — it’s just a raid array, after all! — but I think this hospital took it too far by cutting the oncall out of the ‘hands on’ loop and likely pressured the staff onsite to minimize costly escalations by making specific requests, probably through an app, bypassing the entire “a certified medical doctor occasionally puts eyeballs on the patient and their records as part of their rounds”. When the raid array fails, the junior eng is typically first to be blamed, for not reporting the symptoms with the context that a senior eng would have had, resulting in the senior eng not surveying the server’s condition.

So, then, if the remote-oncall medical senior ‘eng’ was not engaged to survey the patient’s full condition, and the ‘junior’ eng team was allowed to continue treating a patient without having done so, then the failure is the process, not the personnel: we would lose the raid array in this scenario, and they did lose the patient in this scenario. This process that I’m theorizing existed at the facility would guarantee an increase in deaths in exchange for a decrease in expenses. It’s easy to blame the nurses or the doctors, but as many ignored tech post-mortems would say:

“The data loss event ultimately stemmed from employment and process structures imposed by executives for cost savings, and we expect further incidents in the future. Specific mitigations were identified that should prevent this precise scenario, but ultimately the fault lies with department’s design, rather than with the engineers and their imperfect efforts. Additional headcount supporting a revised departmental five-year strategy would be necessary to cure the underlying process defects that allow incidents of this type to occur.”