In Conversation With… Sharon K. Inouye, MD, MPH
Editor's note: Sharon K. Inouye, MD, MPH, one of the world's leaders in geriatrics research and innovation, is a Professor of Medicine and Milton and Shirley F. Levy Family Chair at Harvard Medical School, and Director of the Aging Brain Center in the Institute for Aging Research at Hebrew SeniorLife. She has developed and validated a widely used tool to identify delirium called the Confusion Assessment Method (CAM), and she founded the Hospital Elder Life Program (HELP) to prevent delirium in hospitalized patients.
Dr. Robert Wachter, Editor, AHRQ WebM&M: Tell us what got you interested in delirium.
Dr. Sharon Inouye: I originally got interested in delirium when I did my first stint as a general medicine attending, before I was a geriatrician. That month, I remember so vividly six patients who became really confused while in the hospital. They were all elderly and admitted for different things, like acute myocardial infarction (MI), congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), pneumonia, and cancer. I talked to various colleagues and the Chief of Medicine and asked, "What's this thing about these older people who become confused in the hospital?" Every single person said, "Oh, that just happens, don't worry about it. We see that all the time. They'll get better." And I said, "Well, actually they all ended up doing really poorly." So I spent hours abstracting their charts, making timelines on huge graph paper of when they got confused, everything we did to them, and I concluded that the care we gave them contributed: the medications, the procedures, the immobilization, the Foley catheters. Then I took my flow sheets back to the Chief of Medicine. I showed him this, and he said, "You need to give a lecture to the house staff on prudent use of meds in the elderly." That was the conclusion, and I think he thought that was it—that would take care of the problem. But I said, "No, this is a bigger issue." Then I just never stopped.
RW: When you went to the literature at the time to see what was known about delirium, what did you find?
SI: There was very, very little in the literature. In fact, for a talk I just did, I charted out how many publications per year there were in delirium or acute confusion back in the 1980s, and it was about 20 to 30 per year. Now there are about 400 per year and it's an exponential curve. But, back then, there was no way to measure delirium. No one even understood what it was, how you recognized it at the bedside, what patients were vulnerable, if it had bad outcomes. That is the early work that I did.
RW: You developed a measurement method that still is widely used. Why was it so important to develop a simple tool to make the diagnosis?
SI: I discovered that an entity doesn't exist until you can measure it—and I have to tell you honestly I didn't realize how important an instrument was when I was starting out in this field. Before creating the Confusion Assessment Method (CAM), I tried to find an existing instrument—it was kind of a great adventure. I approached everyone who was interested or had started working in the field and tried to get them to share what they were using with me because I really viewed that as a means to an end. But after being forced to create the instrument, I realized that creating it helped to define what delirium was. Now the definition that I created in that little tiny study when I was a fellow is used everywhere and the instrument has been translated into every language. It was used in 1400 articles as of a couple of years ago.
RW: I've noticed a couple of times you've said delirium or acute confusion or an acute confusional state. Is the name delirium helpful or would it be better if we had a more descriptive name?
SI: Well, it's the term that people use. It's difficult right now because there is no diagnosis code for delirium. If you look at the ICD-9, there are 13 different codes that could be delirium, some of them say delirium, some of them say things like delirium following alcohol withdrawal, or delirium following renal failure, some of them say toxic metabolic encephalopathy, and some of them say clouded mental status or something like that. To me, if there are 13 different ways to code something, then that means no one is in agreement about what it really is. So if delirium could be accepted as a term and if people will recognize that and use a set definition, I think that would really help. If it's acute confusional state that people can understand better then that is okay too.
RW: When you came up with a standard way of measuring delirium and then began looking at the misdiagnoses and the ways people didn't understand this entity, what are the most common diagnostic pitfalls that you've seen?
SI: The common ones are in someone with underlying dementia; delirium is often missed, overlooked, or attributed to the dementia. The other very common scenario is that people don't know the baseline status of the patient, so it's impossible to say if it's an acute change or not. If someone you know was found on the street and brought in to the emergency room confused, it's very hard to know if that person is delirious or not, because the acute onset is one of the major diagnostic criteria. In terms of recognizing it at the bedside, you really do need to do some formal cognitive testing even if it's very short. You at least have to check orientation. It's very nice if you could do some kind of attention task to make sure that someone can keep something in their mind even for a short time. They can repeat after you or say the days of the week or the months of the year backwards. Just a very quick screening test, because I've found when people have good social skills they can completely hoodwink you and cover up all of their cognitive deficits, and you won't know it in routine conversation. You have to probe and check cognitive status to see if people have cognitive deficits to rule out delirium.
RW: When we see people in the hospital with delirium, what are the major errors that you see in workup? Do we get too many or not enough CT scans or LPs? When you instruct your house staff or your students about how to approach such patients, what do you tell them?
SI: A few key points. I agree that we get neuroimaging too often. My criteria for neuroimaging are that if someone has a history of head trauma or evidence of bruising anywhere on the upper body then they definitely need a non-contrast head CT scan to rule out a bleed. You do need to do a careful neurologic exam. In the absence of any neurologic changes, head trauma, or fever where you think there could be herpes encephalitis, then you don't need the neuroimaging or the LP. In the other cases, you should get it. In all cases, you do need to look very carefully for factors that may be contributing in terms of psychoactive meds, any evidence of infection, signs of DVT, acute abdomen, things that could be occult in older people. They may be having an MI and just present with the mental status change. You have to have your antennae very high. The other common pitfall in the elderly—I say this over and over—it's usually not caused by just one thing. It's usually multifactorial. They're a little hyponatremic, going into renal failure, and on medications that maybe the doctor has been titrating upwards or maybe they weren't taking correctly, so maybe because they got a little dehydrated or dig-toxic. You have to very thoughtfully look through everything and it's probably not going to be solved by just addressing one thing.
RW: You've found that the prognosis for patients with delirium is significantly worse than for those without it. Is your sense that the delirium does something to make the prognosis worse or is the delirium a marker for something else that is already there?
SI: It's both. Delirium is a marker of a more vulnerable patient, but delirium contributes excess morbidity and mortality in and of itself. We see this all the time in the hospital: someone becomes delirious with maybe a waxing and waning mental status, and they're much more likely to develop aspiration pneumonia, decubiti, UTI, skin breakdown, or poor nutrition, and when they're agitated there may be psychosis and hallucinations associated with the delirium—then they get treated with medications associated with a whole host of complications themselves. Delirium often sets off a spiral where someone is very confused and a fall risk, so they get assigned to bed rest, get immobilized, and cannot walk—you just see these cascades. Delirium starts this downward spiral, and if you cannot turn it around quickly then the person just spirals downhill.
RW: Can you talk a little bit about what you've learned about prevention strategies?
SI: If you can act upon the very common risk factors for delirium, you can make a big difference in preventing it. Basically if you can address keeping older people off of psychoactive meds, the Beers criteria medications published by the American Geriatrics Society (AGS). The AGS has done a good job of examining medications that are not good for the elderly. If you have to prescribe something in a class of meds, like a certain antidepressant, choose one that is of the lowest toxicity, the shortest acting or the least anticholinergic effects, and the least associated with confusion.
You want to keep people in the hospital well hydrated, mobile, and oriented. You want to allow them to get a good night's sleep. I know that sounds very simple, but it's incredibly hard—patients really cannot get any sleep in the hospital. It's ironic because we think of the hospital as a place for healing, but patients are woken up all night long. If an older person who is very, very sick and is also getting tremendously sleep-deprived, it's not a good setup for good mental functioning and physical functioning. I try to work with the nursing staff to coordinate so that we wake up the patient only once, and we give the breathing treatment, the IV change, the lab, blood draws, the medications, the vitals, the O2 sat—everything can be done at that one time.
RW: How successful are you in those interventions?
SI: If the nurses will buy in and if the nursing leaders will take it on as their mission, then we have been very successful. If we cannot get that to line up, then it doesn't work. You really have to get buy in on the front line. It takes a lot of education, continuous encouragement, continuously working with them. It's positively reinforcing if they see that it makes a difference. But there are so many forces within the hospital setting working against it because of the economic climate and the forces of getting patients out quickly. Ironically, routine or low priority MRIs are now all done at many hospitals between midnight and 6:00 AM. If you need something as an emergency you can get it done right away, but if you want a routine MRI, your patient is going to be up all night some night that week, and I think doctors don't know that.
Certain ways that our health care system is structured are not amenable to good patient health. I also discovered the floors are stripped and waxed between midnight and 6:00 AM—an incredibly noisy procedure that is disruptive. They move patients out into the hallways. Why does that have to be? They told me it's because they don't want people to slip when there is a lot of traffic. But they're keeping all the patients up all night. All these little things happen on the night shift because it's when things are otherwise quieter. When they have to do major room switches, if someone comes up VRE or MRSA positive, they often do them on night shift as well. Then they are moving five people from their rooms. I only find out because I ask all my patients, "Did anything happen last night that disturbed your sleep?" The answer is always yes, and then I ask, okay, what? And they tell me. Then I try to figure out why does that have to be? It's hard; it's a constant battle.
RW: If a patient gets really agitated at 2:00 in the morning, should you be giving them meds? Obviously there are concerns about restraints and concerns about sitters. We want people to move around and be active. We try to avoid psychoactive meds. How would you manage that differently than the run-of-the-mill doctor?
SI: Some locations have delirium units where they have staff trained to work with delirium patients, and they use a lot of non-pharmacologic approaches for relaxation. They have rocking chairs, relaxing music, a bunch of distraction strategies, special beds so the spouses or children can stay as well. They try to make it a kind of homelike environment to make it more comfortable for the person and try to get them reoriented. It often works surprisingly well when you've caught the agitation early on, because what happens when someone becomes acutely agitated, often there is a component of being very frightened or confused about what's going on, or a component of pain or discomfort that is not being adequately addressed.
These specially trained staff, volunteers, and aides are there to help deal with that kind of situation and give attention to that. So you can often get by without any meds at all. But in reality even in those ideal units there are patients you cannot control. If they're at risk in terms of their own safety they might injure themselves, they might injure others, or they might pull out needed tubing like IVs, intubation, dialysis catheters, or central lines, then they will have to be medicated and/or restrained. We try to keep restraints as a last resort. If you need to protect an IV, there are ways you can wrap the IV so it's indestructible and inaccessible and then you can do the other strategies. But if you do have to medicate, we still recommend haloperidol, as long as the person can tolerate that, since it has been demonstrated to be effective in controlled clinical trials.
The newer atypical antipsychotic agents, the only ones that have really been tested in clinical trials, are olanzapine and risperidone, and they are maybe slightly less or equally effective as haloperidol and have less extrapyramidal effects—but they honestly are not well tested yet. There is still the concern of increased mortality with those agents in confused older adults, so we try not to use them as first line. We try to request that the clinicians not use benzodiazepines as the first line drugs because they have been shown in randomized trials to prolong delirium. But they are indicated for alcohol or sedative withdrawal, so those are the only cases where they would be first line. In the ICU where you have uncontrollable agitation and someone might extubate themselves—to sedate someone quickly, they will use faster acting agents like a benzodiazepine or propofol in a supervised setting.
RW: Let's turn to delirium as a patient safety issue. It seems to me it's entered the patient safety world over the last 3 or 4 years. It was really not on the safety radar screen in the early days. Do you agree with that, and do you think that is the right place for it?
SI: In about half the cases, there is an iatrogenic contributor. I wrote a lot back in the 1990s about delirium as a quality marker and a way that we can improve hospital care. Delirium has been proposed to be included on the Medicare no pay list, that is, they were proposing that delirium become a no pay condition. I was very concerned about that because delirium is not well recognized in the hospital right now. By my estimations and those of the American Geriatrics Society, only about 3% to 5% of delirium present in the hospital setting is recognized while that patient is in the hospital. We thought having it become a no pay condition would be the absolute demise of delirium ever being recognized in the hospital. It would create this perverse incentive never to recognize it. So we felt that rather than go that route we would be very much in favor of putting in some process measures looking at things like appropriate use of delirium prevention measures (mobility, non-pharmacologic sleep, and relaxation protocols) and appropriate use of medications. Some of those markers could be adapted to be very helpful for appropriate process measures for delirium prevention.
RW: It's a conundrum. If you believe that many cases of delirium are caused by things that we do to patients in hospitals and you made the case that substantial amounts of delirium can be prevented if things were handled perfectly, the argument would be that we need a standard way of assessing outcomes, adjusted appropriately for the risk factors, and should use that for public reporting or for differential payment. But your argument is that that would create a perverse incentive because people would be incented not to look very hard.
SI: Right now we don't have an accepted standardized way that is practical for hospitals to use because there is no way from administrative databases or just from medical records that we can capture delirium well. There is no accepted coding even if you look at those 13 codes that exist now, only about 3% to 5% are coded that way. I'm trying to see if we can identify key words from electronic medical records that, if they're present, those people have a very high risk of delirium. But right now the only way to reliably identify the delirium is from someone going to the bedside. And that is not necessarily practical for these large-scale patient safety initiatives. So this would be the first, critical step in order to put into place an adjusted method for public reporting.
RW: You've spent most of your career answering the question you confronted as a young attending and demonstrating that delirium is not inevitable, that it's measurable, and that steps can be taken to prevent or ameliorate it. Now that it is on the patient safety radar screen, one would want to take advantage of that light and say I want the hospital that has taken this seriously to be advantaged and for patients or their advocates to know that this is a good hospital as it comes to their approach to delirium. If you were in charge of the safety world and that was your goal over the next 3 to 5 years, what would you want to see in place?
SI: The highest bar would be the presence of HELP (Hospital Elder Life Program) or some delirium prevention program at the hospital. The next level would be, are there nursing staff who are trained in recognizing delirium? Do they use the CAM? Are delirium guidelines in place at the hospital? Are there delirium pathways? And the next level would be to check if the hospital has geriatricians. Is there a geriatric consult service? What do we do if someone gets in trouble? Who is called in if delirium develops? Those are some of the steps people can take. On our Hospital Elder Life Program Web site, we have links for patients and 10 tips for them to know before they come to the hospital: how to recognize delirium, how to talk to their doctor, what to bring with them to the hospital. We try to provide that kind of education. There are probably 100 and 200 official HELP programs in the United States and Canada—not very many given the thousands of hospitals. If we had a delirium prevention program in every hospital and staff that were trained to recognize delirium, we would probably save the US health care system about $40 billion per year. Our estimates weren't in even every hospital but just the hospitals that would be willing to do it.
RW: That is a powerful argument to push for publicly accessible data about whether or not hospitals have these things in place, which can certainly drive some changes. It does sound like one of the twists here is that using the outcome measure of delirium is still problematic because of the various definitions and the possible perverse incentives.
SI: We were working on coding at last year's American Delirium Society meeting (it's a brand new organization, I was the keynote speaker). I said if there is one thing your organization can do is please create a single code for delirium: have it be linked to the multiple chronic conditions or major complicating conditions (MCC) with Medicare, and have it be a quality indicator. A group is already actively working on that. I've been working with DSM-5 group on the delirium criteria. The CAM is part of the new MDS 3.0, the minimum dataset measured in every nursing home across the country. In nursing homes, at least, we're going to have standardized measures of delirium across the board, and they're doing cognitive assessments. We're going to have national guidelines for delirium within 2 years. So there is a lot of stuff going on. It's just as you know, a very, very slow process. I'm trying to make sure that along the way we don't do things that have unintended consequences.
Previous interviews can be heard by subscribing to the Podcast