Although prognostication is most often associated with survival prediction, it is inextricably linked in a triad with diagnosis and treatment. Indeed, there is always the question of the likelihood (prognosis) that a test being ordered will accurately identify the problem. That is, should I put the patient through the investigations, what are the associated burden and costs, and will this likely lead to a diagnosis? Prognosis is also tied to treatment: Will a proposed treatment be successful, when, what would the likely adverse effects be, and how effective will it be in alleviating illness? Both evidence and judgment are required in deciding whether a particular patient will likely benefit from treatment.
So how are we when it comes to prognosis regarding survival prediction? Unfortunately, many patients and family members readily recall how inaccurately and insensitively physicians have given the “blunt” truth, or have avoided the subject altogether, whereas others have found comfort and support from the doctor. Both patients and their bereaved family may have long-lasting anger and difficulty when bad news has been broken badly and incorrectly. In addition, prognostic discussion affects timely access to services, decisions about further treatment, funding allocation, and client decision making. At the same time, it is ludicrous to expect that clinicians will be all knowing and will be precise for each individual patient when considering all diseases, all stages, and various treatment responses. So what are the issues, and can anything improve this?
Most physicians tend to overestimate survival by a factor of 1.2 to 5 times, some are fairly accurate, and a few overly pessimistic. Such optimism can be due to limited clinical experience with end-of-life illness trajectories, a lack of skill or reliance on wrong factors to determine prognosis, dependence on selective recall (remembering only the significant and outlier cases from one’s experience), the belief in a self-fulfilling prophecy that “positive attitudes” affect outcomes including survival, or even purposeful exaggeration to avoid removing hope.
The concern for hope in the presence of a terminal illness is quite legitimate in that one does not want to inflict harm such as depression or despair by disclosing a poor prognosis, but let’s look further. Christakis discusses the value of being optimistic in that it preserves hope, provides encouragement, fosters treatment, and engenders confidence even in the face a terminal illness. He then notes that the “ritualization of optimism” goes beyond this; it is the favorable outlook held by physicians in spite of, or even as a result of, evidence suggesting an unfavorable outcome. Although some physicians have made comments such as, “When in doubt, suspect recovery, and act accordingly,” this may in fact be optimism that is out of proportion to the objective reality. This results in false hope and subsequent decision making by the patient and family that likely would have been different if they had known otherwise.
The flip side is “ritualization of pessimism,” wherein purposeful disclosure is toward an unfavorable outcome. When both the stakes and degree of uncertainty are high, it can be regarded as a protective no-lose strategy. Christakis notes that some physicians have tried to rationalize this with statements such as, “It may often be better to be pessimistic because if the outcome proves poor, it was not unexpected; and if the outcome proves good, you are a hero.” However, neither of these approaches, if intended as so, are helpful to a patient or family who are given information the physician feels should be stretched. These behaviors tend to occur where actual uncertainty of the outcome is higher, because if it is actually known, most will share accordingly. Thus, the skill and accuracy of prognostication come into play as well as the sharing of such information.
10 Steps to Improve Prognostication
As shown in Table 5-1 , prognostication involves two overarching concepts: foreseeing and foretelling. The clinician’s internal formulation of a prognosis can be improved by linking with emerging evidence of relevant factors and prognostic tools to assist this, and external sharing of bad news by attention to other factors. By paying attention to each of the 10 steps, the prognosis for better prognostication is quite favorable.
| Concept | 10 Steps to Better Prognostication | Action Steps | |||
|---|---|---|---|---|---|
| FORESEE | SCIENCE | Disease | 1. | Start with an anchor point. |
|
| Function | 2. | Assess changes in performance status (amount; rate of change). |
| ||
| Tests | 3. | Know physical signs and laboratory markers related to prognosis. |
| ||
| Tools | 4. | Utilize palliative or end-stage prognostic tools. |
| ||
| SKILL | Judgment | 5. | Clinician prediction of survival: Would I be surprised? |
| |
| FORETELL | ART | Center | 6. | What is important to my patient? To the family? |
|
| Frame it | 7. | Use probabilistic planning and discussion. |
| ||
| Cautions | 8. | Share limitations of your prognosis. |
| ||
| Changes | 9. | Review and reassess periodically. |
| ||
| Follow-up | 10. | Stay connected. |
| ||
1
Start with An Anchor Point
It is best to start with what is known and what Quill and Epstein called an anchor point. Most diseases do have general survival statistics that can be found in textbooks, journal articles, or via the web and are usually measured in 1-, 5-, and 10-year survival statistics. They are often based on a staging system, such as tumor-node-metastasis (TNM) staging in cancer, and response rates to first- and second-line treatments. For example, the prognosis for stage I melanoma is very different than stage III, and the diagnosis of metastatic brain cancer alters prognosis significantly. Chronic kidney disease and heart failure (New York Heart Association [NYHA] classification) are other examples where staging has prognostic value. They provide a realistic anchor or ballpark from which to begin, as well as a quick review of the illness trajectory and potential future complications. Discussion of one’s patient with the relevant specialist is invaluable. Relying solely on personal prior cases is problematic and likely to lead to optimistic or pessimistic bias.
2
Assess Performance Status Changes
Continued evidence points toward performance status as one key factor in prognosis for advanced and endstage disease. A common pathway in the last 6 months before death in cancer patients is a steady, almost linear, functional decline that is frequently accompanied by cachexia, anorexia, tiredness, and other symptoms. In the frail elderly with advanced dementia, the decline is very gradual, but final symptoms of dysphagia, development of decubitus skin breakdown, and aspiration pneumonia accompany such frailty and may act as sentinel event flags.
For patients with chronic diseases like congestive heart failure, chronic obstructive pulmonary disease, or end-stage renal disease, it is more difficult to judge, because with acute-on-chronic decompensation and complications rapid decline is often followed by improvement from treatment. Patients may fully regain their previous functional status with initial bouts, but over time, exacerbations requiring hospital admission occur more often and with only partial improvements. The simple question, “Would I be surprised if the patient died within the next 6 months?” can act as a good trigger to reassess the patient and discuss advanced care planning.
One way to visualize the amount and rate of change in functional status is using time blocks, along with a performance status scale such as the Palliative Performance Scale, Eastern Cooperative Oncology Group (ECOG) performance status, the Karnofsky Performance Status in cancer, or other measures such as activities of daily living (ADL). For example, one may assess any functional change in several prior 4-month blocks going back 1 year or possibly more to gauge any differences. When things are changing, move to shorter time blocks such as 4 weeks or 4 days, and vice versa if stable or improving. Sudden change can occur at any time and should be addressed. Using time blocks to show the patient how much and how rapidly change is occurring can assist the clinician to identify and discuss with the patient a number of “what if’s” to gauge how much treatment the patient may want if such were to occur.
3
Review Recent Biological and Laboratory Markers
Other than accidental or sudden death, a declining trajectory in advanced illness inevitably involves physical and laboratory alterations. Physical decline with delirium, dyspnea, dysphagia, weight loss, persistent tiredness, or development of skin breakdown commonly foreshadows the terminal stage. Laboratory findings that have prognostic significance include leukocytosis in association with lymphocytopenia; elevated C-reactive protein (CRP); low albumin; elevated lactate dehydrogenase (LDH); low sodium; and elevated B-type natriuretic peptide (BNP) in heart disease.
Some of these abnormalities may not in and of themselves be prognostic, but they often occur with several diseases and complications. It is perhaps the combined relationship of declining functional status along with changing physical and laboratory findings that should alert the physician.
4
Utilize Palliative or End-Stage Prediction Tools
A systematic review by Lau found several validated prediction scales or tools that could be used to compare with the physician’s view of prognosis. Because they use population-based data, they are not exact and so an individual case could fit anywhere in these projections. They do help in figuratively moving from which “ballpark” the patient is in to whether he or she is now in the outfield or infield. Such probability issues are discussed later. Some provide the likelihood of 30-day survival, such as the Palliative Prognostic (PaP) Score, which incorporates performance status (Karonfsky), clinician prediction, laboratory values, and several symptoms; the Palliative Performance Scale (PPS), using performance, intake, and consciousness level; and the Palliative Performance Index (PPI).
Other examples of tools included in Lau’s review are for heart failure (Heart Failure Risk Scoring System); dementia (Dementia Prognostic Index); cancer (Intrahospital Cancer Mortality Risk Model, Cancer Prognostic Score); and, for the elderly, Prognostic Index for 1-Year Mortality in Older Adults (PIMOA) and Changes in End-Stage Signs and Symptoms (CHESS).
Survival tables and Kaplan-Meier graphs provide general population sampling, and nomograms and algorithms break down data further by age, gender, disease, and so on and are preferable if available. Examples of the latter in heart failure patients are the Seattle Heart Failure Model and the Canadian Cardiovascular Outcomes Research Team ( www.CCORT.ca ). Several nomograms for patients with breast, colon, and bladder cancers can be found on the internet.
5
Clinician Prediction of Survival: Would I Be Surprised?
The final part of “foreseeing” is clinical skill and judgment. It cannot be expected that medical students or clinicians early in practice will have had the experience of seeing many patients die and noting the manner and timing of these deaths. On the other hand, they are also not clouded by recall bias, and all physicians, including those specializing in palliative medicine, need to be aware of their personal tendency to optimism or pessimism in forecasting survival. It is likely that those who are generally good at prognostication are utilizing many of the factors discussed earlier and have distilled these into skilled judgment. Yet the sudden occurrence of complications, even if not unforeseen, make this more difficult, such as if and when a patient may hemorrhage or develop sepsis, a bowel obstruction, or a fall with resulting hip fracture.
The “surprise” question is not so much about whether one thinks the person will in fact die as it is about the risk of dying and whether it would be perplexing or startling that it occurred. If the answer is no, then one is becoming more attuned to possible demise, and this may alert clinicians to observe the patient more closely for signs such as those noted earlier and also provide opportunity to raise the topic of advanced care planning. A simple but valuable practice point is to jot down in patients’ charts your clinician prediction of survival (CPS), or several CPSs over a period of time, and then compare these with the actual survival time. Doing so may help to refine your skill and identify a tendency to overestimate or underestimate survival.
6
What Is Important to My Patient? To the family?
A detailed discussion of communication skills and breaking bad news is beyond the scope of this chapter. The “foretelling” aspect of prognostication is, however, the most important, because one can just as easily harm as help. The approach to discussion and disclosure should always be individualized. Although the physician must be guided by the patients’ wants at this point, be aware that patients commonly expect the following of their physician :
- •
To be realistic
- •
To listen well
- •
To communicate in a straightforward and clear manner
- •
To be sensitive and empathic, especially when delivering a poor prognosis
- •
To allow time for questions
- •
To maintain realistic hopes
- •
To not abandon them
Although it is advocated and indeed helpful to hold a family conference to discuss prognosis, especially if the news is not good, caution is needed in first understanding divergence toward death. This means that the needs and desires for specific information often become different between the patient, who may want less information, and the family, who may need more information. It can sometimes be too much for the patient to hear details of prognosis and expected decline, but this can be quite valuable for the family. Some patients, of course, will want as much information as possible and want the family in attendance, whereas other patients, if asked beforehand by the physician, may decline to attend or leave some discussion with the family until the patient has left.
Culture and religion may also affect what information regarding prognosis is expected and with whom it can be shared. It is imperative that the clinician check with the family and patient regarding how information is to be shared. One should not make assumptions about what is appropriate to share; taking an individualistic approach is imperative.
Before discussing survival information, ascertain the patient’s and family members’ hopes and goals, because this may help in deciding how to discuss prognosis. When patients ask how long they have to live, it may be preferable to begin with a qualitative prognosis (“You are likely to die from your cancer”) rather than quantitative (“You are likely to die in 6 months”). That doesn’t mean that is all the information you are prepared to share, but it is a reasonable starting point. In a survey by Kaplowitz, 80% wanted a qualitative prognosis, whereas only 53% wanted a quantitative prognosis. The question of dashing all hope arises here, and a few points can be shared.
7
Use Probabilistic Planning and Discussion to “Foretell”
Never give a specific time projection to patients because you will always be wrong, with the patient dying either before or after a stated prognosis. Many patients (and their families) hold anger when they are told they have 3 months or 1 year to live and make life plans and changes based on this information, only to find out such a “temporal” prediction was wrong. At the same time, saying, “No one knows,” “I don’t play god,” or “Could be anytime” is equally unhelpful.
So what can be said and how? Having obtained some insight from the “foresee” section, you can then frame the discussion around the likelihood of death occurring, using facts such as the average survival, the median, or percentages. Patients and families usually prefer that the chances of living be given in lay terms, although some patients (e.g., academics, accountants, or scientists) may want actual numbers. There are exceptions and outliers that may help to frame the conversation. Here are some examples of what might be said:
“We do not know for sure how long [John] has to live, but given his situation, the average time is about 3 months. That is, about one half of patients live longer than this and one half less. Most people (10%–90%) live somewhere between 1 to 6 months. However, [John] has now developed [complication]; even if we treat it, I feel this may place him on the shorter side of the 3-month time frame.”
“It is impossible to predict for any individual with certainty, but the average person with your condition will live[a few weeks to a few months, 3–6 months]. Treatment, if it works, might extend the time (a month or two). It could be longer and we can do everything possible. Unfortunately, it could also be shorter, so we need to be prepared just in case.”
“It is hard to know for any given patient, but the average person with [Kevin’s] illness and the recent changes in his condition will live only a few days or so. It could be longer if we are lucky, but it could also be shorter, so we should talk a little more about what to expect in the next little while, and how we can work to ensure he remains comfortable”.
“[Mary] has changed a lot in the last few weeks, and even more over these past 3 days. It is difficult to know for sure, but if the same amount of change occurs in the next few days, then it is hard for me to see beyond that. She might live longer but I also think it could be quite short, perhaps even by tomorrow, so let’s discuss your goals at this point and adjust our plan for what may occur.”
These examples are not “guesswork” but illustrate how the reality of uncertainty can be balanced with the knowledge of the real facts of survival, current stage of disease, and expected complications for a given patient. They also provide aspects of hope within reasonable possibility.
8
Recognize Limitations of Prognostication
One obvious limitation in prognostication is that these tools and the clinician prediction of survival are based on populations with shared characteristics, not specific individuals. The Kaplan-Meier survival graph, for example, shows the line or curve of survival but does not identify where on that curve any particular patient is. As a palliative nurse noted, the patient at “PPS 30% has many faces”. So, will he or she die in a few days, early on the curve, or months later, in the 10th percentile? Second, treatment response may alter survival by controlling progression, causing improvement, and even resulting in cure; alternatively, the decline of treatment may adversely affect prognosis. Third, there is some evidence that psychosocial factors of the patient and family may influence decision making and survival as well as access to care or financial factors. Finally, the personal experience and skill of the clinician, as discussed earlier, affect understanding and accuracy.
In statistical terms, variations are related to the horizon effect (the farther away death is, the harder it is to predict), outliers, exceptions, odds ratios, correlation, precision, and calibration of prognosis. Thus, one’s best prediction of prognosis should be modulated and informed by such factors. In doing so, the patient and family receive your best sense of prognosis within the context of its limitations; most will appreciate the attempt, the candor, and the compassion.
9
Review and Reassess Periodically
Toward the end of life, things never stay the same for long. Depending on the disease and comorbidities, and where the patient is in his or her illness trajectory, various sudden or slowly progressive changes occur that will affect prognosis. Some diseases come with well-known complications; in other diseases, particularly when affecting the frail elderly, things may be unclear. In the latter, an interesting term, ambiguous dying syndrome , refers to the fact that just a day or so before death, many patients are thought to have several months to live.
Currently, researchers are working to identify sentinel events or triggers that alter prognosis, such as pulmonary embolus, sepsis, and so on. Prognostication is not an event but a process, and therefore clinically one should reassess the patient at regular intervals. Quill and Epstein suggested that the physician review and offer to discuss advanced advance care plans whenever the following occur:
- •
New diagnosis of serious illness occurs
- •
A major medical decision with an uncertain outcome must be made.
- •
The patient or family ask about prognosis.
- •
The patient or family request treatment not consistent with good clinical judgment.
- •
You answer no to the question, Would you be surprised if the patient died in 6 to 12 months?
- •
The patient is actively dying.
10
Stay Connected
There is tremendous value in remaining connected through the final illness trajectory. As discussed earlier, being able to provide some insight on prognosis provides the patient and family opportunity to plan for final visits, share closure issues, and prepare for death. This can only happen with periodic reassessment of prognosis. For those who are at home, there is evidence that the likelihood of a home death is increased when the practitioner remains involved.
Irrespective of location, there is increased safety and reassurance, because nursing and medical visits are times where symptom control can be addressed based on declines in physical status and expected complications. One adage to remember is that toward the end of life, things never stay the same for long, and so plans need to be altered. This leads to valuable discussions and reclarification of goals, hopes, and any change to the advance care plan. Visits do not have to be long; the physician’s presence, rather than the amount of time spent on the visit, is what most patients and families find comforting. Finally, personal satisfaction in one’s medical career can be enhanced and reinforced when clinicians remain connected through death and bereavement. Therefore, we need to foresee and foretell, but not forego.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
