Wednesday, October 15, 2014

Assuming Your Conclusion

Health officials have said there was a breach in protocol that led to the infections, but they don't know where the breakdown occurred.

Officials have said they also don't know how the first health worker, a nurse, became infected.
 Both quotes are from the same article.

I was struck earlier by headlines asserting that a breech of the protocol, the rules for preventing contagion, had occurred. The basis for that claim was and is the fact that a nurse got Ebola, not any evidence of how it happened. The obvious alternative is that the existing protocols are in one way or another inadequate, possibly because the beliefs about the disease on which they are based are in part mistaken. It is only if you assume that the protocol is correct that you can conclude it was not followed.


At 10:32 AM, October 15, 2014, Anonymous Anonymous said...

I don't know that is entirely fair. If I write a program that crashes unexpectedly, I can say that there is a bug in my program.

That does not mean I necessarily know what the bug is, or what failure in the software development process occurred to allow the bug through.

At 11:18 AM, October 15, 2014, Anonymous Anonymous said...


At the highest level - the level where the CDC head operates - the protocol is "don't come in contact with the bodily fluids of the patient".

The rest is implementation detail.

At 11:22 AM, October 15, 2014, Anonymous Daublin said...

Anonymous #1: in your analogy, it would be like observing a bug, and then saying someone must have not followed the development process. In software, this would sound crazy; bugs just happen, no matter how good you are.

Anonymous #2: what if it can be transmitted over the air after all?

At 12:28 PM, October 15, 2014, Anonymous Anonymous Hebrew said...

Yes, that's the great fear and concern, if it's possible for it be transmitted over the air.

At 1:37 PM, October 15, 2014, Blogger William Newman said...

I don't have any great respect for health officials, and I do agree that the quote does sound like circular reasoning. Nonetheless I still think the sloppiness conclusion is probably the way to bet, just based on the information I had before. It is not at all easy to keep several percent of people from having a careless moment at work and/or from having misunderstood something relevant in training. Since as I understand it well over a dozen people worked with that patient for days, even a fraction of a percent of slippage accumulates fast. And even though theoretically our medical system is serious about controlling contagion, the chronic difficulties with hospital-transmitted infections and the occasional report of the effectiveness of some new incentive to reliable cleanliness make me think the actual level of quality is not all that impressive. (Though not too laughably bad, either: people mostly surviving after surgery and/or trauma is evidence of some level of competence.)

I did a summer of undergraduate research in a virology lab, and then some in-the-school-year time too. I worked with tame viruses, with long strands of bare RNA (horribly vulnerable to being nicked by even the barest trace of the RNAse enzymes that are present in our skin, in part as a first line of defense against viruses), radiolabeling (making it easy to detect in my urine when I screwed up once), and avian cell culture (horribly vulnerable to contamination by various microorganisms), so I have some practical appreciation of how hard it is to control contamination. I haven't spent all that much time in contact with the US medical system, or studied it systematically, but what I've seen and read makes me think it's not nearly as serious about cleanliness as even ordinary threats like MRSA would justify in a system with fewer principal-agent problems than the US medical mixed economy. At least we have a lot of cheap disposable stuff, which can help a lot. But even in my limited direct experience with the medical system, I've seen medical people wearing disposable gloves in a protocol that's clearly intended to stop the spread of infection between patients ... and typing on a computer keyboard with them. Patient, keyboard, patient, keyboard. Then when the next patient comes along, change the gloves, and go to the new patient, to the keyboard, patient, keyboard. (me: "Argh!")

My working guess is that in the wisdom of hindsight when we understand the spread of Ebola in much more detail, we will find that the equipment and at least the basic outlines of the protocol were theoretically reasonably adequate: not 100.00000000% perfect because nothing is, but in no way so inadequate that multiple nurses should be getting sick when a single patient is treated. I conjecture that in practice the main problem is not some fundamental technical failure (lack of positive pressure containment suits or some such thing) but that the culture does not currently support anything like the level of attention to detail that is required to go from 99.3% adherence to the letter of the protocol and dirty-shared-keyboards indifference to the spirit of the protocol ("some of you may die, but that's a risk I'm willing to take") to the highly-focused behavior seen in organizations that achieve 99-point-several-nines (in various fields such as airline flight safety and various complex manufacturing processes). That might well start to change, because as y'all like to observe, incentives matter, and now it's not e.g. sick patients unaccountably getting sicker from MRSA (these things happen, y'know) but hospital personnel themselves knocking on death's door bearing an infection they they obviously picked up in the hospital in the previous week or two.

At 3:34 PM, October 15, 2014, Anonymous Anonymous said...

Anonymous #1: Not a fair analogy. It might not be a failure in the process. Could be a freak bug in the CPU as has happened in reality.

At 8:48 PM, October 15, 2014, Blogger Shaddox said...

Regarding first Anonymous' comment:

That's not quite analogous. I think the better programming analogy would be to see a program crash, and conclude that there must be a bug in the compiler or in the hardware running your software.

At 6:51 AM, October 16, 2014, Blogger Josiah Neeley said...

You're point is correct, but it seems like a reasonable conclusion for the CDC to draw nonetheless.

There were somewhere around 70 medical personnel taking care of the Ebola patient. Two got sick. The idea that those two were "transmission by air" of a non-lung based disease is implausible.

At 8:11 AM, October 16, 2014, Blogger William Newman said...

an afterthought/clarification: I've seen elsewhere people dumping on the individual infected nurses, sometimes viciously. If that idea is in the air, then it may sound as though that's my point too. But it's only part of my point, and only as one possibility. Possibly if we had a time machine to check things out we would discover that sloppiness by individual front-line nurses resulted in the infection of those sloppy individual front-line nurses, simple as that. However there are lots of other scenarios, and when you look at the work of people who have successfully gotten serious about quality control (most famously Deming) it doesn't sound like idiosyncratic individual front-line sloppiness is a safe way to bet when something goes wrong. Real organizations end up with an interlocking mess of slightly defective procedures, poorly documented procedures, individual sloppiness, careless training, outright incorrect training, needlessly error-prone procedures, indeed lots of needlessly error-prone procedures, inattention to measuring results, and did I mention error-prone procedures and inattention to measuring results? Because they're really important.

Now, ddfr has remarked wisely that we should doubt an account that is such a delicious story that it would be repeated even if it's untrue. And I recognize that the "oh, quality is not an individual carelessness problem, it's a systems problem" story is one that many people would be motivated to push even if it weren't true, so we should accept it with caution. But looking in from the outside at organizations which have achieved high quality, it sure seems to me as though organizations that succeed at high quality consistently do so only in part by hiring very careful people, and in larger part by fixing systems. That seems like pretty good circumstantial evidence that the story is not purely feelgood BS. And from my experience on the inside of software development and maintenance on systems which are complicated enough that reliability is a huge practical constraint, the story seems basically correct to me: sure, it's very important to rely on careful people and to incentivize them to be careful, but when you look back on years of struggling with quality, changes in procedures (notably to make things less error-prone and to provide more prompt high-quality feedback when something goes wrong) will probably loom even larger than anything humanly achievable in terms of choosing flawless individual paragons.

At 10:14 AM, October 16, 2014, Blogger Richard Ober Hammer said...

A story I see tells there was no protocol in place. Or no protocol for a disease this infectious in this hospital.

At 7:43 AM, October 17, 2014, Blogger Nancy Lebovitz said...

From listening to the BBC-- I have the impression that in Africa, medical professionals have a minder when they're getting out of their protective gear.

This makes excellent sense-- it's unreasonable to expect people to get that much detail right when they're tired.

At 11:50 AM, October 17, 2014, Blogger Nancy Lebovitz said...

More detail about hazmat suits.

At 2:14 PM, October 17, 2014, Anonymous js290 said...

The appropriate use of "begging the question..."


Post a Comment

<< Home