1. The document discusses identifying patients in the acute health setting who have a life-limiting illness and high risk of death in the next year, so person-centered end-of-life care can be provided.
2. A study was conducted before and after implementing shared decision making discussions for these high-risk patients. It found an increase in such discussions from 50% to 69% and a decrease in 90-day mortality from 43% to 34%.
3. The goal is to properly identify these patients to engage in palliative care rather than see death as a failure of care.
14. The question
1. Where ?
2. Poor prognosis?
3. Time period?
4. Why?
“Can we identify patients with a poor prognosis?”
15. The question
1. Where ?
2. Poor prognosis?
3. Time period?
4. Why?
“Can we identify patients with a poor prognosis?”
“Can we identify patients in the acute health setting with a high-risk of
dying in the next year… due to a life-limiting illness…why?..."
16. The question
1) No LLI with acute disease…………………………Cure…Shared decision-making
2) LLI with non-reversible acute disease……….Comfort….Gate keeper?
3) LLI with reversible acute disease……………..Cure or Comfort….Gate keeper or SDM
“Can we identify patients in the acute health setting with a high-risk of
dying in the next year… due to a life-limiting illness…so we can..."
18. Is SDM feasible in ICU?
Before (n=119) After (n=103) P-value
Age 72.6 (+13.6) 73.9 (+12.4) 0.47
LLI Criteria
Cancer 24% 22% 0.83
CCF 29% 12% 0.16
COPD 23% 21% 0.68
Renal failure 11% 6% 0.18
Frailty / dementia / stroke 45% 48% 0.74
Nursing home 13% 13% 1.00
Neurological disease 3 % 5% 0.35
Patient-centred discussion
All 50% 69% 0.004
Deceased by day-90 43% 94% <0.0001
19. Changes in outcomes?
Before (n=119) After (n=103) P-value
Admission ICU/HDU 21% 29% 0.16
MET call incidence 88% 73% 0.009
Palliative Care referral 24% 24% 0.90
Hosp LOS 9 [4,19] 9 [5,15] 0.80
Hospital mortality 29% 24% 0.47
90-day readmit 39% 31% 0.24
90-day mortality 47% 34% 0.05
Survival for cancer before and after
20. Frailty
Before (n=48) After (n=43) P-value
ICU/HDU admission 15% 21% 0.4
MET incidence 94% 79% 0.04
Palliative care referral 13% 21% 0.3
90-day readmission 48% 19% 0.003
90-day mortality 35% 44% 0.4
21. The question
1. Where ?
2. Poor prognosis?
3. Time period?
4. Why?
“Can we identify patients with a poor prognosis?”
“Can we identify patients in the acute health-setting with a life-
limiting illness and high risk of death and disability in the next
year, so we can engage in person-centred end-of-life care?”
What about Keith?
22. “90% of adults in the US have no or limited knowledge of palliative care, but after
reading a definition, more than 90% would want it for them or their family” Amy
Kelley, NEJM 2015
“Everyone dies. Death is not an inherent failure. Neglect, however, is.”
(Atul Gawande, JAMA 2016)
http://barwonhealthicu.com
Editor's Notes
Thankyou for inviting me to speak today. Can I start by saying I believe I am here for two reasons;
The first is I am Director of a tertiary ICU in Victoria that has worked to improve patient-centred end-of-life care. Of course like all ICUs, we relentlessly, with insatiable effort and curiosity, try and save lives. This is appropriate, intensive care is designed to take our sickest patients, stop them dying, and return them to their lives. Like all of us we have come a long way in the last 20 years, we are better and better at saving lives.
The second is I very publically shared the difficulties I had accessing good end-of-life care for my father 2-years ago, through an opinion piece that ended up in one of the major Australian newspapers. The difficulties my family experienced highlighted everything we get wrong with end-of-life care – identification, communication, interventions, comfort vs cure.
So the second reason I am here is that I am a person.
It is worth noting that I am not alone in this. We are seeing a strong socio-political interest in end-of-life care. Our institutions are including patient-centred end-of-life care in their strategic plans, our governments are discussing end-of-life care in their legislative councils. In Australia we are actively debating assisted dying legislation.
Our reading and movie material that involves end-of-life care is hugely successful. Books written by doctors about dying resonate as best-sellers. Movies about dying are blockbusters, particularly for young people. The point is our society is interested in dying, and wants us to better.
The title for this talk “identifying patients with a poor prognosis”, is important, as perhaps the first step in improving care is defining a cohort of patients. But I think we need to refine the topic.
To illustrate this I would like to refer to a patient I met last week. I was clinical for my ICU, and arrived one morning to find a 75 year old man who had been admitted overnight. He had metastatic bowel cancer with secondaries in liver, peritoneum, and extra-abdominally. He was receiving palliative chemotherapy, and had been told he was unlikely to survive to Christmas. Over the last 48-hours he had developed what was thought to be a bowel pseudo-obstruction, and had gone to theatre for a decompressive colonoscopy. This had happened, but he became hypotensive, and ended up in recovery, on a vasopressor infusion via his port-a-cath. The oncology registrar had written on his Goals of Management form that he was for full ICU care, so he was referred to us. There had been no discussions with him about his future, so in the end he was admitted to ICU, and the following morning I met him.’
The question, “does this patient have a poor prognosis”, seems simple when we apply it here. But there are important issues to address.
Where
What do we mean by poor prognosis- quality of survival or death
over what time period
due to the acute critical illness or long-term disease or both
and why do we want to know – is it to refuse admission to ICU, to engage in goals of care, or something else.
Let start with where - and here I think we mean the acute hospital setting. Why? we are intensivists. While I happen to believe we could do a lot for our patients and ourselves by engaging at a community level, I have no influence outside the walls of my hospital, and a lot to do inside them. In addition, there are well developed tools used in community palliative care – like the GSF or SPICT, that help identify patients with LLI. So for now, lets restrict ourselves to hospitals. We could further restrict ourselves to deteriorating patients, or patients admitted to ICU. This is certainly the cohort we have been referred, and therefore have some control over. However, it seems logical that identifying patients at high risk of dying prior to ICU contact may have advantages.
What do we mean by poor prognosis - it is a broad umbrella, encompassing ongoing disease, disability, quality of life, and death. There is a great deal that remains unknown about ICU survivorship.
So lets focus on a more defined population, patients at a high–risk of dying, and who will probably experience significant, unwanted, decline in quality of life before death.
This is where I believe there has been an increased understanding over the last decade, as we have started to consider the impact of pre-ICU disease or functional trajectory, when we discuss post-ICU disability and dying.
In the palliative care, geriatric, and primary care setting the recognition of life limiting illness criteria is established. Broadly there are trajectories of LLI, cancer, organ failure, and frailty (dementia, nursing home, frailty).....
What do we mean by poor prognosis - it is a broad umbrella, encompassing ongoing disease, disability, quality of life, and death. There is a great deal that remains unknown about ICU survivorship.
So lets focus on a more defined population, patients at a high–risk of dying, and who will probably experience significant, unwanted, decline in quality of life before death.
This is where I believe there has been an increased understanding over the last decade, as we have started to consider the impact of pre-ICU disease or functional trajectory, when we discuss post-ICU disability and dying.
In the palliative care, geriatric, and primary care setting the recognition of life limiting illness criteria is established. Broadly there are trajectories of LLI, cancer, organ failure, and frailty (dementia, nursing home, frailty).....
Are these criteria applicable in the hospital or ICU setting. We have prospectively looked at them in our ICU, and found clearly different survival curves.
In addition, it is worth noting that frailty is emerging as an important concept in ICU, as not surprisingly, this cohort have very different trajectory of disability and death to others.
Finally there are also studies looking at functional trajectory pre-ICU, and the relationship with both post-ICU functional trajectory and mortality. This study by Ferrante looked at disability in 13 basic areas;
4 basic activities (bathing, dressing, walking across a room, and transferring from a chair)
5 instrumental activities (shopping, housework, meal preparation, taking medications, and managing finances)
3 mobility activities (walk a quarter mile, climb a flight of stairs, and lift or carry 10 lb).
Have you driven a car during the past month?”
Disability was operationalized as the need for personal assistance or an inability to perform the task.
Not only is there a relationship between pre and post ICU disability trajectory, but pre-ICU disability trajectory has a stronger correlation with 1-year survival than other factors. This also raises the question of when, should we look at 1-year rather than hospital or 30-d?
There are also criteria that have been modified for ICU, with Ken Hillmans's group developing the Cristal framework, which incorporates pre-ICU factors, including frailty, and acute factors, to give a prognostic score
It seems we have moved from “can we identify patients with a poor prognosis”, to something else, perhaps to "Can we identify patients in the acute health setting with a high-risk of dying in the next year… due to a life-limiting illness or functional decline….” a very important question remains, why
It seems to me we have identified 3 categories of patients, and the "why” is important;
No LLI with an an acute life threatening disease (ie neutropaenic sepsis in AML, TBI, asthma, sepsis). Unless something unusual happens they will be focused on cure. This group have a good prognosis, and are not the subject of this talk
LLI with minimally or non reversible acute decline
LLI with reversible acute decline
Gatekeeper role is one we are all familiar with, where we decide unilaterally not to offer certain treatments. There are easy examples – ECMO, transplantation. But there are more difficult examples - ventilation, CRRT, NIV. In fact to me this seems more difficult the further I progress.
Only this week we have seen an article in JAMA describing this. Are we able to predict 6-month mortality accurately…Not really. so we should be cautious.
That opens another possibilty, SDM, where we allow patients and families to participate in these decisions, where we give sensible advice, express what we know, what we are concerned about, what we are unsure about prognostically, elicit goals and values, and come up with joint plan.
This creates a very different requirement for our question “can we identify poor prognosis”, because we are not trying to accurately predict death so we can tell patients not to come to ICU, rather we are giving balanced view of risk and benfits and discussing.
Of course the fear is they will all want ICU, and we will be overwhelmed with EOL care. I reject this notion The evdience that exists suggests patients and families are much more sensible, and that the problem is our focus on medicalised dying.
In summary, can we identify patients in the acute health-setting with a high likelihood of death in the next 12-months, determine the presence and influence of life limiting illness, and engage in person-centred end-of-life care?
We can. We have various tools that help us identify patients who are likely to die, particularly due to the trajectory of their underlying functional or disease decline.
The most important question is why do we want to know? Is it to allow us to act as a gatekeeper, or to engage in SDM.
What about my patient, the 75 year old man with metastatic bowel cancer, palliative chemotherapy, 8-months to live, a decompressive colonoscopy, in ICU with vasopressors.
We identifed that he had a LLI, was likely to die in the next year. His oncology doctor had ticked “full ICU care” on his GoM form. However he hadnt talked to him. So we asked him what he wanted. He was clear. His life had been pretty awful for 6 of the last 7 weeks of chemo. He wanted to get through this hiccough, so he could go to the world bowls championships on the Gold Coast with his 2 sons and their families. After that he wanted to stop chemo and die at home, like his wife had 6-year earlier. We all agreed that in this context a short run of vasopressors and ICU suport was appropriate, if it helped him get there.