Missleading Article in New York Times On Radiation
February 16th, 2010
|
| Share |
There’s no doubt that radiation saves lives when applied properly in a medical setting. In recent years, narrow beam radiation therapy has made it possible to zap cancerous tumors while keeping damage to nearby tissue to a minimum. CT scans and digital filmless x-ray imaging has improved at an astonishing rate, making it possible to see inside the body with unprecedented detail.
The exposure from imaging procedures tends to be fairly low and no empirical evidence exists for any harm from routine diagnostic imaging. However, it’s no surprise that a lot of scaremongering has come about over this. People seem to fear “radiation” like few other things an reports that indicate patients are exposed to more radiation than ever before (on average) raise alarm even though they never state that there’s actually any harm arising from this.
Of course, the greater danger is that these reports, empty though they may be, could lead some to avoid life saving diagnostic imaging and radiation therapy. Forging radiation therapy for a tumor that cannot be removed completely with surgery can be a death sentence, but some of the sensational reporting makes it seem like radiation could be even worse, emphasizing the dangers without any context.
Via the New York Times:
Radiation Offers New Cures, and Ways to Do Harm
The radiation boom
As Scott Jerome-Parks lay dying, he clung to this wish: that his fatal radiation overdose — which left him deaf, struggling to see, unable to swallow, burned, with his teeth falling out, with ulcers in his mouth and throat, nauseated, in severe pain and finally unable to breathe — be studied and talked about publicly so that others might not have to live his nightmare.
ensing death was near, Mr. Jerome-Parks summoned his family for a final Christmas. His friends sent two buckets of sand from the beach where they had played as children so he could touch it, feel it and remember better days.
Mr. Jerome-Parks died several weeks later in 2007. He was 43.
A New York City hospital treating him for tongue cancer had failed to detect a computer error that directed a linear accelerator to blast his brain stem and neck with errant beams of radiation. Not once, but on three consecutive days.
Soon after the accident, at St. Vincent’s Hospital in Manhattan, state health officials cautioned hospitals to be extra careful with linear accelerators, machines that generate beams of high-energy radiation.
But on the day of the warning, at the State University of New York Downstate Medical Center in Brooklyn, a 32-year-old breast cancer patient named Alexandra Jn-Charles absorbed the first of 27 days of radiation overdoses, each three times the prescribed amount. A linear accelerator with a missing filter would burn a hole in her chest, leaving a gaping wound so painful that this mother of two young children considered suicide.
Ms. Jn-Charles and Mr. Jerome-Parks died a month apart. Both experienced the wonders and the brutality of radiation. It helped diagnose and treat their disease. It also inflicted unspeakable pain.
Yet while Mr. Jerome-Parks had hoped that others might learn from his misfortune, the details of his case — and Ms. Jn-Charles’s — have until now been shielded from public view by the government, doctors and the hospital.
Americans today receive far more medical radiation than ever before. The average lifetime dose of diagnostic radiation has increased sevenfold since 1980, and more than half of all cancer patients receive radiation therapy. Without a doubt, radiation saves countless lives, and serious accidents are rare.
There are several things that I find a bit offensive about this article. The first is not even in the article text. The use of the sub-headline “The radiation boom” would seem to imply that this is a dangerous technology being overused to make money by those evil for-profit companies. The fact of the matter is, however, that while someone might be making a buck in the equation, the primary reason for the expansion of radiation therapy is that it is better than ever.
As usual there is the obligatory “errors are rare” statement, but a read of the entire article leaves one with a number of graphic and gut-wrenching examples of radiation therapy gone wrong. It should be noted first of all that radiation therapy is very different from medical imaging. The doses are much much higher. Even if an x-ray machine malfunctions and gives you ten times the dose of radiation, it won’t kill you, but a machine delivering radiation for cancer treatment just might. After all, the dose levels are intended to destroy tissue.
For people like Scott Jerome-Parks and his family, a miscalculation or incorrect setting on a machine can result in tragedy. What happened to him and others who were overexposed to radiation to the point of fatal doses is not a side effect or a consequence of normal treatment – it’s a fundamental failure of the system. Somewhere, someone didn’t do their job right or a piece of equipment was not tested and calibrated properly. In a medical setting that’s never acceptable and I certainly hope that the hospital that let it happen gets slapped with a lawsuit that will smart enough to make sure they won’t make the same mistake again.
Radiation therapy must have tight controls and safeties. I’d suggest that each patient should have a dosimeter device which is independent of the equipment being used and can be checked to verify that they’re not being over-exposed. Some of the most grotesque examples occurred when patients were given much higher doses than prescribed on multiple occasions, with the error not being caught even after several treatment sessions.
But lets take a step back from the radiation issue:
Every year millions around the world undergo surgery while under general anesthesia. Few would suggest that a person forgo an important operation because of the use of anesthesia and without general anesthesia, most major surgeries would be all but impossible. Yet anesthesia can and does kill. The same substances which produce the effect, whether they be inhaled or given through an IV will kill if the levels are too high. On occasion, an improper setting, a malfunctioning machine or a distracted technician results in an unnecessary death in the operating room. While under some forms anesthesia, a person is unable to breathe for themselves and must be on a ventilator. Again, death is only a technical failure or human error away.
This is, of course, not limited to anesthesia. Chemotherapy drugs are quite toxic and will kill if the dose is too high. Indeed, nearly any drug can kill if the dose is improper or if it is administered without concern for potential interactions with other drugs being taken. Medical technologies and automated systems of all kinds also do, from time to time, fail and cause death or injury. The failure of a ventilator or malfunction of a dialysis machine can turn an otherwise routine medical procedure deadly. Some have even been killed when an implanted defibrillator malfunctioned and delivered a fatal electrical shock.
Yet the article seems to focus on radiation therapy as if risks of this type are exclusively the domain of nuclear medicine. The accounts of Mr. Jerome-Parks, tragic and sad though it may be, is no more painful nor tragic than others who were maimed or injured because of an error having nothing to do with radiation.
There are several lessons from Mr. Jerome-Parks tragic death that can be learned, and they don’t have as much to do with radiation as they do with medical technology in general. In Mr. Jerome-Parks case, it appears that the problem stemmed from a computer error. After a computer software program (which based on the accounts seems to have been running on a PC) crashed, it appears to have sent some corrupted commands to the control unit on the linear accelerator that was providing the radiation treatment. The result was the a collimator, which was supposed to focus the radiation beam on only the tumor, instead opened up completely and allowed the patient’s entire upper body to be exposed to the intense radiation.
In addition to a complete lack of proper oversight and confirmation of the machine’s proper operation and dosimetry to the patient, there are clearly some major issues with the design of the system and the software. Issues which, unfortunately, are all to common. The system did not have proper input validation and was setup in such a way that it interpreted null input on the collimator setting to mean that the collimator should be fully open. This is a textbook case of bad programming. In addition to this, the interface between the machine and the control software commands only operated in one direction. The software sent commands to the machine, but the machine did not return those commands to the computer in order to confirm that they had been received properly. (These kind of glaring oversights are unfortunately much more common than one might think in life-critical systems. It’s also not the first time we’ve covered this issue here.)
I’m left stumped and appalled that this could happen. Why was an automated, software-driven machine allowed to provide treatment without direct oversight of its actions? Why was there no independent verification of the radiation dose the patient was receiving? Why would such an important piece of equipment be designed with no fail-safe to prevent it from operating if it did not receive all the treatment perimeters?
It appears to be an all too common case of operators not being entirely familiar with their equipment, equipment not providing good situational awareness to operators, lack of hard breaks to stop unsafe situations and over-reliance on unseen software code without verification that it’s functioning correctly. By no means is this problem exclusive to nuclear medicine or even to medicine in general. From airplane autopilots to automated dam gates, designers should never forget the importance of fail-safety and the dangers of lack of good, simple and concise interfaces between operators and equipment which provide for instant situational awareness.
Whether this error had occurred on a linear accelerator or a chemotherapy pump, it’s unacceptable and should not be allowed to happen again. It’s just not a radiation issue.
Less I be called an armchair quarterback for telling system designers and programmers how to do their job, I should point out that despite these considerations being fairly basic and something that they really should already know about, there have been numerous reports of incidents that have happened because of a failure to use good fail-safe design philosophy. Airplanes have crashed because of it, anesthesia machines have malfunctioned because of it, chemical spills have happened because of it and lives have been lost because of it. In an extreme example, even an automated air defense cannon started randomly shooting and could not be shut down due to a software error. Clearly these considerations are not being taken into account to the degree they should in all situations.
This entry was posted on Tuesday, February 16th, 2010 at 3:01 am and is filed under Bad Science, Good Science, Misc, media. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.
View blog reactions
February 16th, 2010 at 3:22 pm
Well of course if RADIATION is involved in any accident regardless of the cause, that’s the story. A steam leak at a cyanide plant wouldn’t make the news, even if a worker or two got parboiled badly. If it were to happen at a nuclear power plant on the generator side, it would be treated like the end of the world was narrowly averted by shear good fortune, and editorials would be written wondering why everybody in a 100Km radius wasn’t evacuated as a precaution.
Quote Comment
February 16th, 2010 at 4:00 pm
Okay, two points here:
You’re right, because these were not really ‘radiation’ incidents, in so far as radiation being the root reason for the problem. The problem was a mistake made someone get an overdose. People die because they get way too much radiation, anesthesia, chemotherapy or whatever. Radiation is just incidental. Any therapy can be deadly if you give way too much of it because of an error.
I read the article and when it talked about how the dying man wished his case be studied to prevent it from happening again, well, it’s not the radiation effects that need to be studied. What needs to be looked at is the breakdown in the human-machine-patient interface.
Point number two:
Armchair quarterbacking on the safety of things like medical devices? Maybe, but your point is totally valid because this kind of design flaw does happen even though it shouldn’t.
Automation is great, software is great and it doesn’t make the same kind of mistakes as people do, but sometimes it makes even worse ones. Simple stuff like, if a command is sent to a machine, verify that it actually is going to preform that command and did not just get lost because of a bad connection. Monitor it in real time – always important for something like a medical device. VERIFICATION – Make sure it gives the dose it is supposed to. No matter how good the system is, parts get jammed, stuff stops working mid-process. Any potentially dangerous piece of equipment needs a kill switch.
Okay, this is all stuff everyone who builds or operates a machine like this should know. Most of the time, they do know. Most of the time machines like this are properly designed and operated and everything goes smoothly. Sometimes though they don’t have the features they should. Not just these machines, automated guns, auto-pilots etc. That should NEVER happen, but it does anyway. All I can say is that the somehow not everyone is getting the message who should.
Quote Comment
February 16th, 2010 at 10:31 pm
Wow. Agreed, this is not an issue inherent to radiation, but this part really got me:
Stunning. How on earth could it be designed so poorly? No input validation?
The way I read this it sounds like the shield thing that controls the radiation is open “By default” so that if the control file is null or corrupt or something then the shield is open and he software tells it which ones need to be closed. That’s horrible! It needs to be the other way. It should be closed by default and the shields are selectively opened by the program input. There should be no way that anything other than a verified open command will make them open. Not a closed command! Closed should be the default state!
How could they miss that? How could something so important not be verified? When someone goes to delete a file on a windows computer it asks “Are you sure” but this didn’t show the status of the machine and ask “ARE YOU SURE THIS IS HOW IT IS SUPPOSED TO BE SET” and make sure that it was checked? Also, how could they miss this because of not watching the screen? Was this like off in a little corner of the screen?
Okay, I’m starting to suspect this thing runs Windows or something, but that’s horrible again. The computer froze and it still created a usable file? WTF???? Important files should not be written in a way that allows an interrupted operation to result in a file that can be mistakenly read.
Even buisiness software and database applications are written to prevent this. It’s a common error to not have this in a program and a reason for data read failures, but why didn’t they bother here.
It should be:
write temp file
verify temp file
copy temp file to working file
verify copy
flag working file as validated.
If it were done properly like that, a crash would never result in radiation error.
How could anyone get a job at a company like that and suck so badly as a programmer??????
Quote Comment
February 17th, 2010 at 12:39 am
Interesting. It looks like Walt Bogdanich is a busy man. I blogged about another article he wrote in the January 26 NYT. I checked and he has written 8 articles recently on the topic.
http://topics.nytimes.com/top/reference/timestopics/people/b/walt_bogdanich/index.html?inline=nyt-per
My blog entry is here:
http://www.anupchurchchrestomathy.com/2010/01/new-york-times-article-on-radiation.html
My point was that these problems aren’t really new and went into the history of the Therac-25. I summed it by saying, “Most of the issues came down to poorly trained people using new machines with confusing software.”.
I’d also like to mention a book I read recently by Dr. Atul Gawande called, “The Checklist Manifesto”. He talks about how a simple checklist could reduce surgical complications by up to 1/3. My first thought and reading the article was that these guys could use a checklist. BTW, Dr. Gawande book can be used for almost complex tasks and many that aren’t that complex. Much of his inspiration was drawn from commercial aviation. When you think how few scheduled flightss crash in the United States, that is a pretty good role model.
Quote Comment
February 17th, 2010 at 12:48 am
Because said programmer is being pushed by the company to get the product out there, to maximize the return.
I’m not surprised there are these kind of shortcomings, the way our medical system is set up, many doctors are focused on quantity instead of quality, as that is where the greatest money is to be made. Service the maximum number of patients as quickly as possible. The incentives need to be fixed, so there is money in preventing illness, instead of these more profitable radiation treatments. Or as Obama pointed out, the focus needs to be treating diabetes, instead of the high reimbursement procedures such as amputations.
Quote Comment
February 17th, 2010 at 12:52 am
I was also thought they would have mentioned the Therac-25 fiasco. I would have thought to, that the lesson would have sunk in to the whole medical radiation industry from that event. God knows something like that would go through the aviation industry like a dose of salts.
Another thing that would happen in aviation with a repeat event like this, is that the regulator would be looking to lay criminal negligence charges, and massive punitive fines, to show their displeasure at seeing this again.
Quote Comment
February 17th, 2010 at 2:07 am
Bruce said:
Having done a lot of programing myself, I’d contest that many of these issues are not really issues of getting it done quickly. It doesn’t necessarily take any longer to do things properly (like not using case control statements on a positive condition for terminating from a loop in a life critical application.)
It’s more just bad design practice. Not to mention the side of it of operators. How they would not check the collimator is beyond me.
Bruce said:
Well, there’s a limited number of doctors to treat a nearly unlimited number of patients. So the quality/quantity thing is always an issue. If ER doctors spent all day on every person the line would be even worse.
There’s a legitimate argument that they should try to treat as many as possible, while still maintaining quality to what degree is possible, but sometimes there are hard choices.
Bruce said:
Er… not sure you can blame that one so much on the healthcare system. Patients need to manage their own diabetes and can’t expect a doctor to be with them 24 hours a day always giving them their insulin on time and manage their diet.
I know a guy with diabetes who seems to talk all the time about how he’s going to start trying to manage it soon. He’s morbidly obese and he’s constantly “going to start the new diet next week.” His blood sugar levels are always very high.
Quite honestly, I don’t think he’s going to live more than a few more years and it’s not any doctor’s fault.
Joel Upchurch said:
Actually, I was going to write something about this. There was a study a while ago where they had several hospitals adopt an adapted version of the aviation checklist for surgery and it was found it reduced errors more than almost any other procedure tested.
The checklist used in aviation is very flexible and extremely effective. For all it’s simplicity, having a hard list of things to make sure are in order for a procedure seems to be one of the best ways of keeping up safety.
There may be a kind of culture block to it though. The idea of a surgical checklist is that the head surgon will sign off on everything as a pilot would – it doesn’t actually increase the time spent in surgery, but the issue is that it’s sometimes seen as being insulting or stupid because of the things on it like “Check patients name: is this the correct patient” and “check, all personnel are scrubbed”
The response is something like “What you really think we’d operate on the wrong patient???” of course, occasionally they do.
There was similar resistance when the aviation checklist was standardized during world war II. Pilots thought it was some kind of a joke, half the stuff on it being so mundane “Wheel blocks removed? what the hell kind of an idiot would forget to remove the wheel blocks. All engines running? What am I, retarded” There’s a kind of macho attitude to cut corners on those. However, it has become accepted.
DV82XL said:
Yeah, I probably should have brought that up. I don’t know what it is, but these fundamental problems of design and operator issues are not going away. It’s really a design philosophy issue, I think. Too much trust in soft and logical control. Big things keep being left out: Always fail to safe, always inoperative by default, provide continuous situational awareness of basic systems, independent sensor confirmation, direct acknowledgment and approval.
Really this could be attacked on either end: system design or operational check lists and simple step by step procedures. Ideally it should be on both.
Quote Comment
February 17th, 2010 at 2:31 am
drbuzz0 said:
That is the whole point though. Checklists are supposed to be “idiotproof” and if you don’t think you can be an idiot, then you’re going to really feel like one when you start to take off with one of the doors still open and the cowling off one of the engines. That kind of mistake (and more complex ones too) is exactly what a checklist prevents.
It amazes me they seem to wing it when they do radiation treatments, There’s no step-by-step SOP or anything. Checks are not mandatory? They were not watching the screen?
Obviously radiation therapy is a complex technical task and that’s what check lists and itemized procedures are for. They are designed to make complex tasks managable and fault proof by breaking them down and catching mistakes.
I wonder how much of this might be due to the medical culture. Not that I don’t think modern medicine is great, but it seems like it’s too separated from other worlds. The medical world should really consider looking at other sectors for procedures to apply. Aviation safety was mentioned, but also industrial quality control, construction management, food quality control, the military and so on. These sectors all have developed very good methodologies for managing complex things and catching mistakes. Even top surgeons should not bee too proud to consider learning a lesson from them. I think checklists are a brilliant place to start.
Quote Comment
February 17th, 2010 at 4:27 am
As a critical software development engineer, I totally agree.
A few years ago, I was working on drivers for a cardiovascular X-Ray imager manufactured by a large corporation founded by the inventor of the lightbulb (the kind of device that allowed your 42th president to undergo a vasular intervention a few weeks ago without having to stop all activities for mounths or face a significat risk of dying on the opperation table).
The thing is that this system was designed in a way that every module was constantly checking the sanity of the other components, and turning on the generator required quite of lot of checks. Also, these systems had to be FAA approved, with the auditors starting with the assumption that we deliberately tried to kill the patient and the manufacturer QA team spending mounths proving the contrary.
BTW, that manufacturer pulled out of radiotherapy about ten years ago because the finantial risk to benefit was too high: the market is very small and a large proportion of patients die anyway, so even without system malfunction, the risk of trial was high.
Quote Comment
February 17th, 2010 at 9:38 am
The thing is that the Therac-25 caused a complete overhaul of how the FDA validates software for medical devices. I actually read the draft back in 1997 and I estimated it would take several times as much effort to do the validation as it did to write the software. I think my estimate was pretty good, since I had gone through a similar effort getting the operating system a C2 security approval at a previous company I worked for.
http://www.fda.gov/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm085281.htm#_Toc517237966
Yet all this effort lets simple software bugs through.
That reminds me of a story about when my boss called and he asked me how a bug in my code ended up getting released in the operating system and I told him there was a technical term for what happened. He asked me what the term was. I said that we call it “screwing up”. He laughed and said that he was familiar with the term and that ended the discussion.
Quote Comment
February 17th, 2010 at 10:55 am
No amount of effort or intelligence can prevent a bug to come in production once in a while, but that’s not the point here.
In every serious place I worked, we always spent days in meetings listing every piece of a design and any way it could fail and asked the following questions:
-What are the most terrible possible consequences?
-How can it be prevented, or at least detected?
-What does the design do to mitigate the consequences?
Any less than good answer to any of those questions caused a rejection of the design, that’s why we always ended with things like power-on self test, range check, communication acknolegements, redundancy and/or degraded mode.
It happens to “screw up”, and that’s a shame that an error goes through the various review or verification steps, but that’s absolutely nothing compared to not having proper review or verification plan.
Quote Comment
February 17th, 2010 at 12:35 pm
Franck said:
That is true, but this was more than just an issue of a bug in the design. The bug would not have caused any fatal concequences if the system had been monitored properly or if it there was a mandatory check to make sure that the system did configure itself as it was supposed to.
Yes, technology won’t always work the way it should. That’s why there’s an off switch.
I tend to have a fair amount of trust for technology, but I still count my money when it comes out of the ATM, because even though I’ve never been shortchanged, it only takes a second and I know the currency counting thing can skip or get jammed. Also, when I print something important, I don’t just hand it in or mail it without glancing at all the pages, because I know that the printer sometimes skips or improperly loads the paper and ends up with an error on it.
The other thing that gets me is how they could not have been aware of the way the machine was operating. Something fundamental like the collimator setting and the total dosimetery should be displayed on a big central display and not just one of a page of variables. It should be something that you’ll notice even at a glance. The most basic and fundamental information should be displayed like that.
In a car, the center of the dashboard is where the speedometer is. It’s big and prominent. It’s so obvious and in the field of view that its all but impossible to not see it when driving. In an airplane, the artificial horizon is front and center, lit and brightly colored – again, all but impossible not to see. Then you have other important indicators like the airspeed and altitude next to it on easily read dials.
There are smaller dials that are important, but not as basic and not necessary to be continuously read off to the side and smaller. The gauges for things like oil pressure, transmission fluid pressure, voltage, coolant level and such are small and not in the center. On an airplane, the indicator of bleed air flow is not going to be as big as the artificial horizon, because you don’t need to keep an eye on the continuously.
That’s part of situational awareness and the human-machine interface issue. The operator should be given an obvious, impossible to ignore, unambiguous readout of the most important data as to what is happening.
This issue seems to keep coming up, despite the fact that it has been cited a number of times as one of the critical reasons for a mishap. The investigation of Three Mile Island cited the fact that there was no easily readable and unambiguous display of the reactor fluid level in the control room. Fluid level was determined by inferring it based on three pressure gauges. That made it easy to not notice if it was unusual.
Quote Comment
February 17th, 2010 at 1:23 pm
Franck said:
As much as ISO 9001: xxx is a bureaucratic pain in the behind, it is very good standard for quality assurance documentation and quality control system design. It is however misunderstood, and there is resistance to using these tools even in those companies that have adopted the standard.
The power of ISO 9001 is that it creates a common framework that allows those outside a particular area of expertise to fairly judge if inspection standards and quality control procedures have been effectively carried out downstream from their position in the supply chain. Standards also exist for training, such that there is broad input from every part of system design into making sure that the training regimen covers all of the important aspects of operation.
What is very much lacking is the proper attitude, and getting through to some companies that these standards are not some sort of elaborate fraud with the objective to pick their pockets every few months, is not easy. I was astonished the first time I accepted a contract outside of the aviation supply chain at the resistance I met to this whole idea. Apropos to this topic, the company was supplying chassis assemblies to a maker of X-ray real-time imaging targets for mammography, who properly ( in my opinion) wanted their whole supply chain fully traceable.
Quality is nether cheap or easy to implement in any activity, but the cost might be your business, to say nothing of someone’s life.
Quote Comment
February 17th, 2010 at 3:24 pm
I don’t see this as just an issue of standards for quality or manufacture control and working the bugs out of the design. It’s just as much an issue of planning and operating with the assumption that there may be a failure and how you can prevent it from being catastrophic.
Even the best built, well designed, proven and reviewed ships still should always have lifeboats.
No matter how reliable jet engines are these days, pilots still train for a glide-in landing and airplanes still have an APU, ramair turbine or both.
All dams should have a spillway.
All reactors should have a scram button.
All reberather divers should carry a small tank of compressed air, or at least have a bailout procedure for open-circuit operation
All helicopters, regardless of the quality of the engines, should be capable of auto-rotation
All parachutes should have reserve
All pressurized aircraft should have breathable oxygen avaliable
All large rockets should have a destructive abort capability and a range safety officer constantly monitoring them
All space stations should have enough escape capsules for all crew at all times
All submarines should be able to surface using only a compressed air ballast blow
All nuclear weapon cores should fail to reach criticality with anything other than a fully symmetrical implosion
All grid-powered life support equipment should have a UPS, a generator or both
Never go diving without a buddy
Never go into the wilderness without telling someone when to expect you back
Never skip items on a safety check list
Never weld without a fire extinguisher
Never bench press without a spotter
Never go on a space walk without tether, a maneuvering unit or both
Never smoke in a coal mine
Never take any medication without reading the label
If you’re not sure if you’ve had too much to drive, you’ve had too much to drive.
If you can’t identify a snake, assume it’s venomous.
if it might be contaminated, it is contaminated.
All firearms are loaded until you have personally verified that both the clip and chamber are empty
All unexploded ordnance is considered live and dangerous
All injured persons are to be treated as if they have spinal injuries and immobilized until it is confirmed otherwise
All power lines are live until directly verified and/or grounded
Measure twice, cut once
A radiation therapy machine’s settings should be verified. It’s actions should be monitored. It’s dose should be measured and verified. The calculations used to set the dose should be double checked.
NO amount of quality control or manufacturing precision will ever replace the above – EVER.
Remember, not long ago a school nurse injected several with insulin instead of vaccine. If you think that this can be avoided by strict quality control in insulin or vaccine production, then you’ve missed the root of the problem.
Quote Comment
February 17th, 2010 at 4:46 pm
Well, my main take away from this article, besides the point about doctor’s re-imbusrement (which is another subject, but something Obama has spoken on, pointing out much more money is to be made amputating then treating), is the safety systems for radiation are just not there yet.
From reading everything people say here, it seems like there is quite a lot of thought put into safety. The problem is, there’s always some safety shortcoming people don’t think of. In the case of this medical radiation, they were meeting all of these ISO 9000 FDA safety standards or whatever, and this still happened. After the fact, you always say the problem “was obvious” and they were so stupid to have made the mistake – but I bet a lot of the people working on these systems and desiging them are pretty smart. The problem is, complexity! That’s why you go to alternatives that are not so high risk, if they are available. In the case of nuclear medicine, there isn’t an alternative, so we put up with the risk. In other cases, there is no need to put up with the risks of radiation, so it is best to avoid them. That is what I think the moral of the NYT story is.
Quote Comment
February 17th, 2010 at 8:05 pm
Bruce said:
I wouldn’t say that. “Not there yet” is implying that there’s some more development or refining of technology. I don’t think that’s the problem. The problem is not a radiation problem. We can put as much money and development into radiation studies as we want and not solve this, because it’s just an overdose or improper dose. It could be a dose of anything (and many die from overdoses of drugs or anesthesia or something)
There is a whole branch of design and study that deals with this problem. Human-machine interactions are the cause of a lot of issues not relating to radiation but potentially just as deadly.
There are stratagies which can be used to prevent this from happening. One is called “Poke Yoke” which is Japanese for something I forget, but it means that something physically stops you from doing something wrong, like you can’t stick a floppy disk in the wrong way, because one of the corners is cut and it won’t fit in that way or you can’t run something with the safety turned off.
Bruce said:
They are smart, and that’s not the problem. Anyone, no matter how smart, can mess up complex systems. Yes, it is complexity. That’s why you want a simple fail safe or a simple escape sequence. That’s also why check lists are good, becuase you don’t miss anything.
I think both the hospital and manufacturer are at fault here. The manufacturer should have made it impossible to not notice an error (like Steve said, big central displays with confirmation “are you sure this is how you want it set”) and dosimetery alarms.
The hospital should have had a set-in-stone procedure with step by step fault checks.
Bruce said:
Unless, of course, you replace that risk with something much much greater, like the risk of everyone getting sick from coal fumes or something. I mean, I’d rather have lived next to Three Mile Island when it suffered its partial meltdown than I would have TVA Kingston when the ash pond spilled and left homes under several feet of gritty, heavy, toxic, dirty ash. Of course, in that case, everyone managed to get out alive (all their property, their home and possibly health ruined, but at least alive). That’s a lot better than the implications of a sudden natural gas plant explosion, like the one last week that killed five. The one problem with natural gas is that if you get a leak and get it mixing with the atmosphere, it had a tendency to go off like a pile of dynamite.
That’s not all that bad though, not compared to what a dam could do if it suddenly failed. I mean, natural gas explosions kill dozens, at worst, maybe hundreds, but a dam failure can easily kill thousands. There was one in China that killed more than 50 thousand and some estimate 250,000 if you count all the casualties due to famine and disease. If the Three Gorges Dam were to fail or one of the other mega-dams, it would make the 2005 tsunami look small.
Thanks, I’ll take the radiation risk versus all the much greater ones.
Quote Comment
February 18th, 2010 at 11:59 am
While this is clearly a quality control and process control issue, not something that justifies a “OMG SCARY RADIATION!” kinda thing, I do think it is utterly unacceptable that, all these years after the THERAC scandal, medical device manufacturers are still making basic mistakes and not treating their products like the safety-critical systems they are. I’m sure it’s not limited to imaging and radiotherapy devices either; what problems does Da Vinci (for instance) have, that we don’t know about?
That said, I do think we have too many people getting CT scans. (And MRIs, though MRIs don’t involve ionizing radiation.) The cumulative radiation exposure isn’t something we should entirely dismiss, and there are other problems as well.
* Cost: CT scans are not free, and if the scan isn’t going to greatly increase the diagnostic sensitivity, is it worth the cost and nuisance to the patient?
* Availability: there is a finite number of CT scanners; perhaps we should be triaging their use better. Though I haven’t heard of any problems with people in critical care situations being denied a CT scan because it was busy scanning someone who maybe didn’t need it, I think that if the current screening trend continues, there is a very good chance we’ll start to have problems like that.
* False positives: any test can have a false positive, and the risk of that is that you may end up treating something that doesn’t need to be treated. No treatment is without risks, and some have quite a lot of risks, so this also should be considered carefully.
* Detection of things which don’t actually need treatment: this isn’t really a false positive, but the sensitivity of these scans is now good enough that doctors are finding lumps so small it’s hard to judge whether they are liable to kill the person or not. (Many cancers are quite indolent.) Is it worth going through the pain and risk of a lumpectomy if the tumor would never cause problems?
Given that there really does appear to be a risk of *death* with these devices, I think that needs to be considered when deciding whether or not to refer a patient for a scan like these. I do think we are probably screening too many patients with these more sophisticated techniques. Yeah, they can all benefit, but do they benefit enough to justify the scans? Especially since we now know that quality control is a bit iffy at some places?
There’s been a lot of discussion of this at Respectful Insolence (the author, Orac, is a surgical oncologist and a cancer researcher). It’s not really clear-cut where the line ought to be drawn, but it’s something that the medical community probably should be thinking more about.
Quote Comment
February 18th, 2010 at 12:27 pm
drbuzz0 said:
No, of course not, however the role of quality management should not start or stop at manufacturing. Proper design evaluation, establishing training requirements, and follow up from the field also part of the process. This is the Total Quality Management philosophy, that again is looked at as a useless added expense that has been inflicted on a company by a crazy major customer. The degree of resistance to these even in big companies is depressing.
However it has been proven in transportation maintenance, that there is very few ‘human error’ incidents, in those areas when the person responsible for the job has to sign for it and assume unlimited criminal liability for what they have done.
It has been proven by that sector, that good traceable and complete records, can get to the bottom of a systemic failure quickly and accurately, and the findings and corrections can be distributed throughout an industry in a matter of hours.
I can add too that regular recurrent training, and in-service evaluation of those preforming critical tasks, is a part of a quality/safety culture that is missing, particularly in medicine and several other activities where it should be par for the course.
I believe that establishing responsibility, right from the outset is something sorely lacking in most human activities, and that civil actions after the fact is a poor substitution. But maybe that’s just me.
Quote Comment
February 18th, 2010 at 1:21 pm
Calli Arcale said:
Well, to give an anecdote, which may still be worth consideration: very recently (as in last month) my father had some bad lower back pains and was diagnosed with an infection. He took antibiotics for a week and the pain did not go away. His doctor told him that the likely reason was that it was a persistent infection that would take a good two weeks or more of antibiotics to clear up and that it might take even longer for the swelling to go down and the symptoms to be completely gone.
However, there was the possibility it was cancer and not an infection. This was the less likely reason, but couldn’t be ruled out.
So obviously we could have waited a couple weeks and if the antibiotics did their job then it would pretty much confirm that it was an infection. However, even though the probability of cancer was “low” it had a very very big effect on my father and everyone else in my family. I don’t know whether I was stressed because I thought it actually might be cancer or because I saw my father so stressed out.
In any case, this didn’t last long becasue the morning after his second appointment he went in for a CT scan and it was confirmed not to be a tumor nor were there any tumors seen in his lower abdomen in general.
Fast, non-invasive, fairly conclusive peace of mind. In my opinion, more than worth a couple millirems.
Calli Arcale said:
I’m not sure whether that’s really a concern. The place my father went was actually not a hospital at all. It was in his town (a fairly small community) and was a diagnostic radiology office. I don’t think there’s any real concern there that the machine would be needed for anything critical. It’s not an ER, it’s for diagnostic stuff. They also have CT scanners at big ER’s and hospitals in general.
A lot of these places like that, nobody is really coming in on a stretcher with hours to live if they don’t get a proper diagnosis.
Calli Arcale said:
There’s an interesting side to this. Given that cancers do sometimes go into spontaneous remission, does this mean that they should not be treated, and the bigger question: Is it ever better to be ignorant than to be aware of something in your body? Should a doctor ever know less and not more?
Clearly, there are some for whom it is unnecessary, but it’s hard to know who.
So a hypothetical question: Lets say a scan finds something like a breast tumor (that being the classic example.) The tumor looks like it could be cancer and so a biopsy is done with a local anesthetic, and indeed the lump is cancerous.
Based on the best information, there’s a probability of 25% possibly up to 33% that if you do nothing, the tumor will just sit there for a year or so and then shrink and go away. There’s a 66-75% chance that if left alone it will not go away and will grow and start to spread. You could keep monitoring it, but it’s not clear whether it will show any signs of growing before it spreads. It may turn out that before you can get an idea of whether or not it’s going to grow, it’s already spread to your brain, liver and lungs.
The alternative: remove it with a small incision. Possibly a small amount of unwanted change in profile that could be fixed with a little plastic surgery. After that, a couple seasons of radiation beam therapy. The chance of death due to malfunction? Well under .1% However, it would also have some unpleasantness. Granted, it would not be bloody, torturous painful, but any incision is going to be a little uncomfortable for a while.
Now: What do you do? Or… would you rather just not know? What if your decision is wrong? Very hard to answer questions.
I think I can answer those questions for myself, but maybe not, because I’ve never been in that situation. I think I’d want to know as much as possible and talk it over with my doctor. I’d probably go for treatment, unless treatment involved some really major risks or disfigurement. If it was going to be something really major then I might consider watching it for a little bit to try to see what it does.
Calli Arcale said:
I’m not sure that death is a reasonable risk from diagnostic imagery like a CT scan. Radiation therapy does expose patients to very high doses (remember, the whole point is to kill tissue) and if that dose is high outside the treated area and destroys tissue it’s not supposed to – well, that could be very bad.
CT scans don’t produce nearly that much radiation. Worst case for a head to toe scan would probably be less than 2000 mrems. Even if the machine malfunctioned badly and scanned you ten times instead of one, that still wouldn’t cause any acute radiation sickness. It would have to be closer to 40 REM before anyone would start to have any symptoms of acute radiation problems. It takes about 100 REM to make death a possibility. Even at those levels, only 10% fatality in totally untreated persons is expected. With immediate medical treatment (supplements, antibiotics, possibly dialysis) there’d be near 100% survival.
It would take one hell of a CT-scan error to give you that kind of exposure. I mean, the tube in the machine can’t even produce enough radiation to kill you in less than an hour without blowing out. So it would have to be a very intense radiation and you’d have to be exposed repeatedly.
Quote Comment
February 18th, 2010 at 8:22 pm
Calli Arcale said:
I’m pretty sure this was the rationale for the announcement a couple of weeks ago, advising that fewer mammograms should be done on women (under 50 yrs). The probabilities are sometimes such that fewer diagnostic tests will result in a better overall outcome for the population as a whole. This is, of course, no comfort to any given individual who has reason to want the test done. The details of the probabalistic argument are too subtle for our generally innumerate society, but the results surely include a small number of women who suffer needlessly – they are not ill (false positive) and yet they are subject to the (sometimes dire) side effects of the treatment.
Quote Comment
February 21st, 2010 at 6:20 pm
gman said:
Too my understanding, the study did not say that – all things being equal – the risks are greater with treatment then without. Instead, it was that mammograms at under 50 are too expensive to be justifiable, but they do still save lives. Here’s from the article:
http://www.npr.org/templates/story/story.php?storyId=120470428
“If saving lives is the only goal, then doing more screening will save more lives,” says Jeanne Mandelblatt, of Georgetown University, the lead author of the modeling conducted by the cancer centers. A larger societal goal that has to be considered, she says, is whether the cost of saving lives is worth the harm caused.
So, it’s more about the fact that the rising cost of medical care is a problem. One of the best ways of addressing it is limiting the use of these expensive treatments.
DrBuzz: The place my father went was actually not a hospital at all. It was in his town (a fairly small community) and was a diagnostic radiology office. I don’t think there’s any real concern there that the machine would be needed for anything critical. It’s not an ER, it’s for diagnostic stuff. They also have CT scanners at big ER’s and hospitals in general.
That is precisely the kind of place where a lot of money could be saved in our medical system. There are too many such diagnostic tests being performed when we could get by with a lot less. Without healthcare reform, these costs will just keep growing.
Quote Comment
February 22nd, 2010 at 10:11 am
Bruce said:
So, what did she mean when she said “is worth the harm caused” ?
My understanding of the issue was, if you treat 1000 women, and save the life of 1, but meanwhile make say, 5 sick (in a non-fatal way), that’s OK with most people. But if you make 10 women sick, or 100, or 500… when does that become “not OK”? Depends on the number, and ‘how sick is sick’, right?
Quote Comment
July 16th, 2010 at 2:42 am
Noone knows how many cancers have been caused by Cat Scans. At one particular hospital radiation overdoses had been going on for years and the patients were not notified. They would not know that the radiation overdose was causing memory loss or balance problems because there was no hair loss. (hair loss appears at 3 Gry.) We don’t know how many have died. How many women had mammograms every year and developed breast cancer? We have a reason to fear diagnostic CAT Scans. I was overdosed and continue to suffer. It could happen to you or your family member. I was not informed of the large amount of radiation that I would have received if I had had a normal scan. We all have a reason for fear? Now that Bogdanich and others have investigated these radiation overdoses “perhaps” things will change. The public needed to be informed or nothing would change- too many coverups!! Bogdanich should get another Pulitzer Prize!
Quote Comment
July 16th, 2010 at 5:57 am
JC said:
Prove just one.
Hysterical claims like yours based on ignorance are meaningless and have nothing to back them up. The question is why should anyone believe anything you say without supporting evidence other than your assertions, and reject the opinions of the medical community?
Have there been accidents with medical radiation treatments? Yes. There have been accidents with pharmaceutical treatments, and accidents in the operating theatre, and misdiagnosis in the general practitioner’s office. In fact each of these other areas have been the source of more medicine related deaths than medical radiation ever has. Are we then proposing that all of medicine be abandoned in favour of doing nothing?
We accept risk in everything we do, and balance it against the potential gain. So far medical radiation, in all its manifestations has done more good than harm. Unless it can be proven otherwise by numbers, these criticisms are little more than fearmongering.
Quote Comment