Here’s a story from the UK about an x-ray incident at a hospital:
http://www.dailyrecord.co.uk/news/scottish-news/patient-given-13-times-normal-1515154
From the story, does it sound like they found the root causes?
What lessons learned might they have missed?
Category: Accidents, Current Events, Investigations, Pictures
4 Comments »
RSS feed for comments on this post.
The X-ray machine should of course have been tested after the software update. That is the most important lesson to remember.
Comment by Jan Waehrens — January 22, 2013 @ 4:45 am
My first though was that the testing protocol is inadequate. Do they do scenario simulations after upgrades are installed? Answering the question around quality control may get closer to the rot cause.
Comment by Graham Dobson — January 22, 2013 @ 3:23 pm
It appears from the minimal amount of information in the article that the hospital is not conducting any investigation into the incident therefore missing the opportunity to find root causes and apply corrective actions. An investigation into the incident would need to be concerned with, among other issues, the machine itself, the maintenance procedure (if any) and personnel executing the upgrade, requirements for testing the unit (if any), the machine operator and interface, operating procedures, software design, and other instances of similar failures.
If software design is determined to be a cause, running a root cause analysis on the software design process itself might yield significant improvements.
I would also recommend that the hospital lock the machine out of service if possible to prevent any chance of inadvertent use; signs can be accidentally removed. They should also utilize a return-to-service procedure to ensure that the problem is really fixed.
Food for thought… Since Software is essentially a procedure that a machine follows, might there not be some application of the Procedures Basic Cause category to software failures?
Comment by Christopher W — January 23, 2013 @ 1:24 pm
Software can be thought of as an automated process. Therefore, RCFA for software failures should be performed as it would for any other process. However, software rarely functions alone. It usually relies upon data, hardware, operating systems (which is more software), human interaction, sensors (providing data), consistent voltage and amperage of electricity, etc. So, if a proper RCFA was performed on the original problem, and if that RCFA can legitimately point to the software failure alone as a “cause,” then I would say they have so far only identified a symptom (bad software) needing further RCFA, and have not yet discovered the actionable root cause(s) of the software failing.
Comment by Eric Nielsen — January 23, 2013 @ 3:20 pm