the n of 1

We’d like to think that medicine is entirely evidence-based, but it’s not. Some of what we do has a great amount of evidence behind it, but sometimes the evidence is a little more shaky. Sometimes there’s practically no evidence. As a paramedic put it to me in an ACLS class one day, aside from this short list of drugs, as far as the evidence goes, we could put mayonnaise in that IV and there’d be a similar amount of research to support it.

I mean, we try our best to operate under what we think to be the best. We read consensus statements from working groups, we go by clinical practice guidelines published by a committee of experts, and when we aren’t sure, we go by conjecture that we base on previous clinical experience and whatever tangentially associated evidence we happen to have packed away in our brains. We usually get pretty close, but the truth is that we could get a lot better. A lot of the time, we basically run on nothing.

In truth, we don’t really run on “nothing” — clinical experience isn’t irrelevant, and recommendations from those with more experience than us, while not exactly evidence, isn’t exactly nothing. But there’s lots of situations, especially with the sort-of new-frontier type medicine, where the answer to the question of what the best thing to do is honestly that we don’t know.

So imagine my joy when I find a document on a subject that I know has gotten little research. Somebody (in this case the Canadian Thoracic Society) has compiled all of the best available information into one document and made recommendations based on it. I skipped off to the printer (sorry, trees) and pulled out my highlighter and began swiping away at passages I found most relevant. I got two swipes and three paragraphs into the actual recommendations before I found this gem:

“Unfortunately, each of these techniques suffers from the lack of well-designed prospective trials. As such, recommendations were informed by observational studies and professional consensus.”

Professional consensus and observational studies. So clinical experience (times a lot of clinicians) plus tangentially related evidence (with a small sample size and no controlled conditions) are literally the best evidence we have. Like I said, it’s not exactly nothing, but when you consider the way things tend to fall apart under close scrutiny in this field, it’s about as close to nothing as you can get while still having a half-assed idea what you’re doing.

In school they teach us to operate on this version of ‘nothing’. They teach us models and give us context and try and assist us in developing the skills necessary to work outside the textbook. Very few patients are cookie-cutter. We operate like that a little bit when we’re brand new, but as we gain experience we learn things that are unteachable. We learn how much wiggle room we really have — that it’s not necessarily the end of the world if we try something and it doesn’t work. We learn that the limits we were given are margins of safety, and that there’s a lot of space between the margins.

Enter the patient. I’m lucky, I say — so many of my patients are heavily sedated and won’t remember what I did to them — I have a sort of list of things I can try in order to achieve the result that I want. It’s a common refrain in health care that patients don’t read textbooks, and it’s true. It’s exactly because no two patients are exactly alike that no two treatments are exactly alike. It’s the nature of what I do that I intervene and look for a particular patient response; when I don’t get the response I want, I change my intervention. In this way each patient is its own isolated experimental model: a kind of so-called n-of-1 trial.

I think if most people knew how much of my job (especially with regards to ventilating people) is “well, let’s try it and see what happens,” they’d be a little concerned. The truth is that that’s the essence of a lot of medicine. The beauty of ventilating someone with a piece of equipment that retails for more than a small condominium is that I get the benefit of immediate information about how my experiment is working. I don’t have to wait for days for antibiotics to work or steroids to kick in. I don’t even have to wait the minutes it can take for sedation to kick in. I will usually know in under 5 minutes if what I want to do is going to work or not, and because things respond so fast, unless I do something exceedingly stupid it’s actually very difficult for me to harm somebody with an experiment of this kind.

Sometimes I get another kind of immediate data: sometimes my patients are awake and talking. The home ventilator stuff I linked up there is so interesting precisely because of that. 99% of the time when I ventilate a patient, they’re out cold and I’m left to do the guesswork based on some animations and a few fluctuating numbers on an LCD screen. When the patient’s awake, they can tell me what they want and how they feel, and if they’re articulate about it and it’s a problem I can solve, in a way this gives me far more fine-grained control over what I end up doing.

He asks me questions, and the answer I have is an honest one: we don’t know, there’s not a lot of research to support this, we don’t have a lot of good models for what we’re doing, it depends on how you respond. It sounds terrifying to somebody who wants the patriarchal model of medicine to hand down a pronouncement from on high about what their therapy will entail. Sometimes we do that, but we try not to. Care plans shouldn’t be about what I think is best for you. I don’t live in your body for 24 hours a day and once you walk out those doors the life you live is your own. If I’m going to come up with something that you’re going to be able to live with day in and day out, it’s far better if we can come up with something together.

It’s easy 99% of the time with my heavily sedated patients. The tube comes out, they come to, (sometimes not in that order,) and what I’ve done is something that was profoundly uncomfortable and yet saved their life. They don’t have to live with the therapy on a day-in-day-out basis — it was a short term thing and once over, it can be forgotten.

With someone who’s vented at home, it’s an entirely different story. Their life is my therapy and without it their life would be shortened considerably. They can’t ignore what I’m doing if it’s uncomfortable and they can’t forget about it because it’s ever present. I need his feedback to do my job properly: the equipment I use in the home is 1/10th the sophistication of the equipment I use in the ICU. I lose my raw data and get subjective information and I have to glean a course of action from that.

The benefit to this is his subjective response is just as quick oftentimes. He knows his body and I can trust that. He gives me far better data than I can get off of an LCD screen and it allows me to individualize his vent settings in a way that I would never dream of doing with an acutely ill patient. Admittedly it helps that most chronic ventilator patients have healthy lungs, and I’m using settings that are far gentler than anything I’d use on someone really sick, but an experiment is an experiment and it can still go awry.

It depends heavily on the patient too. Some people become very uncomfortable if they think you don’t know. Some people are anxious and when you say “we’ll have to try it and see how it goes” they hear you say that you’re not confident in what you’re doing. (Those are times you have to be part salesman.) But most people actually respond really well to an authentic voice, when you tell them we just don’t know and in a lot of ways we have no way of knowing. I tried hard to be honest without being wishy-washy, and I think they appreciated my lack of fatalism and my willingness to be flexible.

Even inasmuch as the n of 1 is a terrible way to conduct scientific research, it’s a great way to conduct patient care. We are all individuals and what works for one of us will likely not work for the next, and applying cookie-cutter approaches doesn’t always work. At some point a really good clinician will be willing to go beyond the textbook, to look at the data they have, to try new things and to see how they work out. The ability to think critically in this way is what separates out those who really know what they’re doing from those who use the paint-by-numbers or recipe-book method of healthcare, and not just for their critical thinking skills. Some of the most valuable things to come out of such an experiment is the experience of having done it in the first place, of learning those things that are unteachable. We shouldn’t fear experimentation. It’s how we become truly great at what we do.

by the numbers

A part of having a highly variable workload from day to day is that it becomes necessary for me to show my work. I have to keep statistics, in the form of a list of time definers. Checkmarks, quite literally, for exactly what I did as part of saving a life.

I always have a bit of a chuckle as I’m doing my stats. How are these numbers even calculated? How have they figured out that it takes 17 minutes to do a respiratory assessment, 8 minutes to change a cylinder, 18 minutes to insert an arterial line? There are many where I write down actual time spent, and it seems a cold kind of truth to me that at the end of the day, 45 minutes spent counselling a family about the impending death of their loved one gets aggregated with other staff members for my department over the month, and written in a tidy little box. #3740, service recipient support, a nice round number for a bureaucrat to sign off on. The funniest part is that I’ve done stats at other places where this kind of service wasn’t even something they had a time definer for. As far as the bureaucrats could see, emotional support didn’t exist.

It seems funny to me that I can distill a really fraught encounter down to numbers. A code on the floors, intubated, sent to ICU, set-up on the vent and handed off. It rounds up to around 4 hours of work, spread across a multitude of time definers. What isn’t in the stats is the looks I exchanged with the ICU nurse, the frustration at the physician who was content to sit on his hands, using my ass to hold open an elevator door, rearranging a barely-set-up-freshly-clean ICU room, and the heartbreak of prolonging the inevitable indefinitely. It doesn’t include the bone-weariness that comes with five flights of stairs times five or six trips up and down. It doesn’t include an entire team of people content to place their anxiety at not knowing what to do squarely on your shoulders, because now you’re here, and they don’t have to worry. But no pressure.

No pressure. I mean, I don’t stat mistakes as mistakes, they just get lumped in with an actual time definer. (#4420, arrest attendance.) It’s funny to me that things which are truly chaotic, which can truly not be distilled down to a series of single actions are lumped in together. The time definer for arrest attendance may as well be #4420: unmitigated chaos.

How does one stat “agonize over a decision”? How do you stat “sat in boss’s office venting”? It’s a rhetorical question — I could find a way to stat either one — but the point is, I can’t put a numeric representation on how hard I’ve worked when sometimes the hardest things I’ve done aren’t things with statistics attached to them.

a perfunctory passing

If we ever had a fly on the wall when we talk to each other in ICU, in the RT department, they’d think we’re heartless.

We have a pretty intimate relationship with death. Death isn’t unexpected here, and it’s rarely sudden. Most deaths are well anticipated, planned for, welcomed even. The family may not anticipate it the way we do, but we have seen that shadow on the horizon many times, and we know it’s coming. We see small signs, we know the diagnoses and have a solid understanding for when it’s a lost cause. Usually, there’s nothing in writing regarding how the dying wanted their death to look. Few people plan their own deaths; usually these people are cancer patients, and not everybody dies of cancer. Those who have chosen to use and use up their bodies are especially prone to poor planning, which means when their end-organ failure catches up to them and assures their death is inevitable and probably soon, we find ourselves lost in the grey area between patient wishes and family wishes.

The dying is usually unable to give us much input at this point; especially in the elderly, especially with organ failure, confusion is common and not readily reversible. People also don’t talk about their deaths with their family. They might talk about it with their next-of-kin but they’re usually vague. “I don’t want to be put on life support” is a really vague statement, how do you respond when the doctor says “we want to put a tube down his throat”? Without specialized knowledge, how do you know they’re talking about life support? All that’s obvious is that the doctor wants to do something to stop his inevitable decline into the grave. What’s not obvious is that the doctor isn’t telling, the doctor is asking, and in that grey area, the answer they want is a clear end-of-life decision.

These are heavy decisions to make on the spot, and people need time. There is no small amount of grief in circling around this decision, the discussion between family members about whether they want to throw in the towel and accept that this really is the end. We can see it: the liver failure, the kidney failure, the heart failure, the respiratory failure. We can compensate for some of these things somewhat for a small amount of time, and we often do, in order to let people come to terms with that decision. But we, as the seemingly prescient-ones, know how close he is to the edge. We know that our various treatments and therapies and interventions and invasive procedures are only buying time. We whisper it to each other: “He needs to die.”

It’s said glibly, and we understand each other. We are not saying that this person’s life is insignificant, we are saying that this person’s natural death is being held off by medicine. We know which organs have failed and how completely. We know there’s no feasible way for this person to live a life unencumbered by pain, immobility, incontinence, isolation, and pathology.

Similarly, we know that the family is afraid, grieving, dealing with that process first before death even comes. We know the situation is emotionally very fraught. We know that what’s best for the patient is also what’s best for the family: we need to find a middle ground where it’s not about freeing up an ICU bed, where it’s not about medicine giving up on somebody and putting them on an ice floe out to sea. The middle ground is when we can present reasonably that what we want is to give the patient a kinder, gentler passing, one eased by painkillers and anti-nauseants, anti-anxiety drugs, and peace and calm. We don’t need to be convinced of these things, so we communicate in short-hand with each other.

“He needs to die,” we say, and the rest of us nod solemnly. What we mean is that his body wants to die, that without some form of life support he will surely die soon. What we mean is that he is living on borrowed time, at the expense of his comfort and health. Life support is not a benign thing; it is almost always inherently invasive, and with invasiveness comes risks.

We don’t usually go that far. Usually, all it takes is being honest: bring family into the room, show them what we see, show them his organs are failing and that he has a limited amount of time. They are encouraged to spend their time while they have it, and we stress to them that anything more we can do will only cause more pain with no true healing. There’s true understanding then, at the end, where they see what we see: that death will be soon and inevitable, and it’s only a matter of how easy we can make it on the dying.

And then after, when the family has finished streaming in and out, when the patient has been whisked away to the morgue and the bed has been cleaned and left as prim and pristine as it was when they arrived, when all trace of their presence has been erased, the next shift comes on, and asks where the patient went. “Oh, he died,” we say, and the next shift nods with relief.

“Oh, that’s good,” they say, a quote which when taken out of context sounds callous, from such a caring profession. Out of context it’s hard to understand why we’d be happy to hear somebody died;  he got what he needed, and what was best for him.

the man and the machine

Gear comes up so often as an RT that a day where I don’t do at least some troubleshooting is a boring day indeed. Where most of the other equipment other team members deal with are either ways of gathering patient data (lab analyzers, patient monitors) or ways of giving drugs to the patient (IV pumps), there are relatively few situations where the patient is directly interfacing with the machine in question. It is, after all, somewhat frowned upon for patients to adjust their monitor alarms or change their IV rate.

For many RTs, though, it’s a daily thing. I can think of two examples in particular, and they’re polar opposites: the spirometer, and the ventilator.

Spirometry is a difficult test to do sometimes, as it’s almost 100% effort dependent on the part of the patient. This is in contrast with blood analysis by the lab, where the major difficulty regarding the patient is whether or not you can get blood out of the person. (Sometimes it takes the lab tech who can get blood from a stone. Sometimes it takes four strong staff and five point restraints.) The most accurate results will only happen when a patient can understand my instructions, will cooperate and follow my directions, and when they try as hard as they possibly can.

They emphasize to us in school how important it is for us to coach the patient enthusiastically, since a major portion of getting the best effort out of someone is coaching them to try their hardest. It’s not uncommon for me to look at a patient’s results on the screen or watch a patient blow and go, “you can do better than that.”

Then there’s the interface itself: I’m trying to make a person do a very specific thing to a machine, and if I can’t make that person do what I need them to do, the test is nearly useless. This is a classic example of troubleshooting the patient. I made sure at the beginning of the testing day that my machine is working correctly, so unless something catastrophically fails, the problem is almost universally one with the patient. It’s my job to look at the output the machine displays, and try and explain to the patient I’m testing what I need them to do differently in order to get the output I want. This includes noticing things like air escaping the system at the mouthpiece (and therefore not being measured,) a lack of effort on the part of the patient (meaning the results appear as if the patient has terrible lung funtion to the untrained eye,) or some things the patient has no control over, like whether their dentures are loose, whether that stroke they had five years ago means they can’t make a mouth seal at the mouthpiece, whether they can’t understand what I’m saying but are smiling and nodding anyway, or whether they’re just not going to put the effort in and I’m wasting my time trying to teach a pig to sing.

I can’t stare only at the machine and ignore the patient: the forced exhalation maneuver can cause a transient decrease in blood pressure and a transient slowing of the heart rate, especially towards the end of exhalation, where the patient is forcing against airways that are empty of air and for the most part, closed. In this way it resembles a valsalva maneuver and I need to pay attention to my patient, because if they faint and fall off the chair and bounce their skull off the concrete floor, well, I have a bigger problem than poor test results. If I don’t notice that air is leaking out the side of the patient’s mouth, I’m gonna have a really hard time figuring out why I can’t get decent results. At the same time, I can’t only stare at the patient and ignore the machine. The whole trick to the success of the test is ensuring that the patient is interfacing properly with the machine. Ignoring the machine is to ignore half of the system.

The other major example I have, the ventilator, is somewhat the opposite sort of system. In this case, the patient is often, but not always, passive (rather than with the spirometer, where the machine is passive.) With the ventilator, I’m trying to make the machine do this very specific thing, and if I can’t get the machine to do what I need it to do, then I’ve got some serious problems. The confounding extra factor, for bonus fun, is that sometimes the patient isn’t passive. Sometimes, I’ve got this machine that’s supposed to be breathing for a patient that’s trying to breathe, and a major limitation to this is that they don’t share a brain (though they’re working on that.) Sometimes the patient’s trying to breathe and the machine is trying to breathe and they end up fighting each other, or what some people call “the patient fighting the ventilator.” The term we prefer to use is ‘patient-ventilator dyssynchrony’ or, more colloquially, “hypo-sedation-emia.”

In an ideal world, the patient would initiate a breath and the ventilator would detect this and deliver a breath in synchrony with the patient’s efforts. This is called ‘triggering’ the breath. Then, when the patient is done inhaling, the ventilator can also detect this and, depending on the mode of ventilation, can stop delivering the breath near the same time the patient stops inhaling. This is called ‘cycling’ the breath.

Being as, from a gas flow physics standpoint, the sensors that make these detections are a considerable distance from the patient, sometimes the ventilator isn’t able to detect what the patient wants the ventilator to do. Accordingly, the ventilator is more likely to have a harder time the sicker you are and the worse your lungs are. The ventilator is, after all, a machine, and one of the major downfalls of machines is they don’t think, they only do what we program them to do. Sometimes we can program them with complex algorithms designed to eliminate some of the thinking, but that does not obviate the need to think.

This is where I come in. I observe both the machine and the patient and try to tweak what variables I can tweak in order to get the machine and the patient to agree with each other. I try and see what the patient is trying to do and try and manipulate the variables I have to try and make the vent do what the patient is trying to get it to do. It’s a magical sort of alchemy, and sometimes I can’t make it work. Most of the time, though, I can manage to find the sweet spot between the totally passive patient who “rides” the ventilator and the dyspneic, agitated, desaturating, magical self-extubating patient.

The opposite is also true: sometimes the patient is so sick and so short of breath that they consume so much oxygen trying to breathe (and doing so only ineffectively) that they’re better off anaesthetized so I can take over and make the patient’s lungs do what I want them to do so they get some actual gas exchange going and they can get better.

The trend with new ventilator modes is to try and make it so the machine can adjust itself continually according to what the patient appears to ‘want’ according to a software algorithm. This seems like a good idea in theory, and it can work pretty good for some patients, but us RTs tend to hate these newfangled modes for one reason and one reason only: we can’t tweak them. “I can’t make the stupid thing do what I want it to do” is what we think, trying to find that sweet spot. I can see what the patient is trying to get the vent to do and the software algorithm can’t.

It’s alchemy, it’s guesswork, it’s “let’s try this and see how that works”, it’s tweaking and a process of elimination. Half the time the process is a series of judgement calls and failed experiments until something works.

I’m the respiratory therapist, and when I troubleshoot a system, it’s got both a machine and a person in it, and I have to troubleshoot them both at the same time. It leads to a unique set of skills and challenges, and I don’t even think these systems are unique to my profession. I think these systems are visible in many disciplines where people interact with machines. I think most of us are just not used to seeing the person as a part of the system.

the ounce of prevention

A code that came in this weekend brought me back in time about a decade.

He came in dead, and he left that way too, as these things are sometimes wont to go. Later, after the post-mortem, the infection control officer did her customary chasing down of the staff members who were at the code. Seems the patient had meningococcal meningitis, a severe infection of the brain linings caused by a bacterium which spreads in the droplets of respiratory secretions. The infection control officer was there with a stack of prescriptions for antibiotics, a dose of which we all had to take to ensure that we who were exposed would not succumb to this infection.

The scariest part about meningococcus isn’t antibiotic resistance (MRSA) or disease mutations in animal hosts making the pathogen more dangerous to humans (H1N1), but rather how fast the disease progresses from an asymptomatic incubation stage to causing severe brain damage or death. Vaccination is available against the infection but vaccination campaigns in Canada haven’t been part of the routine immunization series, so herd immunity is nonexistent. The last time a number of teenagers died of this vaccine-preventable disease was when I received my immunizations against it.

About ten years ago, several high-school aged kids where I lived contracted meningococcal infection. Some of them got treated early enough on for meningitis, but the unlucky ones ended up like this kid. The infection overwhelmed their bodies, resulting in a the typical meningitis-type rash growing and growing into purpura, a sign the infection had caused DIC. At that point, the sepsis had progressed so far that the likelihood of a good outcome was extremely poor, and as a result, two very healthy kids died very suddenly.

Meningococcal disease is downright frightening, and is one of the few examples of infections you shouldn’t wait to get antibiotics for. It’s one of relatively few infections that, even in industrialized nations, still kills people fast enough that a matter of hours can make the difference between alive and dead. It’s one infection where the best treatment is that ounce of prevention: vaccination for those who are at risk. It’s one infection that emphasizes how important vaccination can be in avoiding preventable deaths. It’s one infection that really lays bare how little the medical community can do if we don’t catch you in time. And it makes it very starkly clear how dangerous it can be to not pay attention to your body.

Dr. Peter Saul at TED: Let’s talk about dying

The weight of their grief

We approach the humorless with humor; I think it’s the only way to cope. Sometimes there’s nothing to be done. Sometimes it’s sudden. Sometimes it’s a sudden insult but death doesn’t come suddenly. Then, their watch begins.

The rules about how many visitors to an ICU room go out the window. As many people as that room can fit, and outside visiting hours too. Sometimes we break other rules too, to allow children and the burning of smudges. The treatment decisions are guided by family, not the doctors anymore. The doctors have washed their hands.

The neurosurgeon has seen the CT scan (ah, technology) and too much time has passed since the bleeding begun. Your brain, in a box. Increase the pressure too much and it becomes toothpaste squeezed through the hole at the bottom of your skull. The phrase we use for this injury is “not compatible with life.” It’s less politically correct amongst ourselves. There’s just joking: “go towards the light;” when we have to shut off life support sometimes we dryly refer to it as being “the grim reaper.” In his room, though, it’s different. Their vigil makes it a sacred space.

Those who enter do so only out of necessity, with more respect than even the cemetery commands. We are quiet and as discreet as we can be, but somehow it still feels like a violation to go inside and do my job. Moreover, it’s difficult to watch their internal struggle: they have heard our words, they know there is no hope, but they reach for it anyway. They are grieving not only the death that hasn’t happened yet but the lost opportunities for his recovery. They’re grieving their lost hope at the same time as his loss, and all their losses before it as well.

They share these losses with me when I enter, while he actively dies; they’ve lived through this more than once. I try and keep my words to a minimum, aside from condolences and apologies for my intrusion. I am more proper than I usually am, even on a good day. I ask if they want the door closed. I enforce silence in myself. It isn’t my words that they want.

When I leave that room I feel like I walk slower, my breathing and my body heavier. I am moving in slow motion through water, the resistance of the air demanding more of my strength. I am brought back to where I was, when I last experienced such grief. I say a silent prayer for the fact that I had no vigil, in retrospect the finality of these things is a gift. I try to shake off the heaviness of it all, perhaps remarking at it to my manager or a coworker.

It is impossible to be wholly unaffected when they only currency in use is hope, and I have empty pockets. I can’t make this easier. All I can do is try to remember enough that it’s harder to forget, as I will eventually, as I always seem to. I’ll be reminded again of their grief soon enough, except at the same time it won’t be theirs, it will be someone new, and the grief will be fresh, and I will have my task, to be the cup-bearer of the good death.