I am deeply saddened today to learn of the passing of my valued colleague Basil Proimos of Crete. I was privileged to know him and to call him friend. My thoughts are with his widow and their two young daughters. Such a tragic loss.
Look, I take it all back. Varian has demonstrated once again, and conclusively, that the organization does not understand safe design. I tried to give them the benefit of the doubt, but I was just wrong. Continue reading
I’ve just returned from the annual Symposium and Meeting of the Southeast Chapter of the AAPM. I was a member of the SEAAPM for about 25 years before moving out of the region. I was thrilled to be invited back down this year to speak at the Symposium on the topic of software system QA and safety.
I was struck by a consistent theme in the questions from the floor, and in fact from some of the presenters (not me). Continue reading
One very interesting chapter of my recent conversation with the folks at Varian regarding software safety had to do with the interaction of hospital networking infrastructure with the heavily network-reliant computer systems in radiation oncology. This theme had also emerged recently in a conference call of the AAPM’s Professional Council in the context of our review of ASTRO’s draft white paper on IMRT safety. And it has been a constant conversation in my workplace since I arrived a year ago.
I’m here to tell you that inadequate computer and network infrastructure is a very significant and widely underappreciated risk in a modern radiotherapy department.
The issue is on the face of it simple enough. Continue reading
I am (for the second time in two weeks) reminded of the Monty Python “Summarizing Proust Competition.” I am not at all sure how to capture the texture of my day at Varian in the space of a blog entry. But I’ll give it a go.
Frank, sometimes sharp, differences of perspective were expressed strongly but respectfully on both sides. I have gotten a much better sense of what they are doing, and of their frustrations. And I am optimistic that they have a better sense of my part in the safety conversation, and of my frustration. I count that as a successful meeting and do appreciate their initiating it. Continue reading
Here are some incidents involving Varian software and accidental loss of MLC shielding (or in one case a whole field) as reported to the MAUDE database. Not an exhaustive list, but interesting. Continue reading
So here’s the story so far.
Back in 2005 there was a serious misadministration at St. Vincent’s Hospital in NYC in which the patient was treated to most of his head and neck for 3 fractions with IMRT monitor units but no MLC in the field. The patient received an estimated 13 Gy per fraction and died two years later of complications from that significant radiation overdose. Shortly after the accident I received a copy of an internal forensic analysis by the equipment manufacturer indicating that while the staff at the facility were clearly culpable for having not followed their own safety process, the triggering event was that the treatment planning database was able to get into an inconsistent state (intended fluence, no MLC) that was not detected until too late. The manufacturer concluded that the cause was user error and that the software had performed as designed. My reading was that the design was not safe. Still, I honored the confidentiality of the information for many years as I had promised I would. Continue reading
… and I was not very reliable about updating it. But I’m thinking I might try again. From time to time I write something that folks seem to want to share with others and maybe this is a good way to facilitate that. Maybe.