3: Goals of Care & Consideration of Prognosis

Goals of care discussions provide a broad framework for decision making, helping align patients’ underlying values and hopes with the realistic and achievable options for care given the current medical circumstances. This is no easy task, however, as patients and their family members may simultaneously express multiple goals for their health care, which may include maintenance of independence, prevention of illness, prolongation of life, relief of suffering, and maximization of time with family and friends. The relative importance placed on each goal may change over time as new information is shared with the patient or family, such as new diagnosis or a worsening prognosis. These goals should serve as a guide from which patients and their physicians can develop specific plans for treatment when dealing with acute or chronic illness.

Goals of care can provide a guide for various decisions, including immediate decisions regarding life-sustaining treatments, decisions regarding preferences for preventive therapies such as cancer screening, and for the completion of advance directives. There is no one right way of having these discussions; however, the following outlines 7 practical steps for having a discussion (see Table 3–1 for words to use, and Table 3–2 for words to avoid).

1. Prepare: Clinicians should establish an appropriate setting, one that is quiet with enough space for all participants to sit down. The clinician should identify appropriate participants, including extended family, other consultants, or team members, such as social work or chaplaincy. A facilitator should be identified in advance if more than one clinician or team member will be present. Also, ensure adequate time is set aside for the meeting and that interpreters are used if needed.

2. Create structure: At the start of the meeting, all participants should introduce themselves. The purpose of the meeting should be made explicit. Clinicians should also ask about patient and family preferences for information sharing and decision making.

3. Explore understanding of medical situation and underlying values: Effective decision making depends on both health care providers and patients having an understanding of the patient’s illness and prognosis. Clinicians should determine what the patient and family members understand about the patient’s illness and its expected natural course. Information should be given in small, easy-to-understand statements with frequent checks to assess for comprehension. This is also a time to explore what outcomes patients and families are hoping for and which ones they would want to avoid, as well as what is most important in their lives and what they would like most to accomplish.

4. Define overarching goals: Based on what was learned about the patient’s and family’s hopes and expectations, providers can explore or suggest overarching goals. This should also be a time to address hopes and goals that may be unreasonable or unrealistic given the current health state or future prognosis.

5. Assist in making a decision based on the patient’s beliefs and values: Discuss how goals can be achieved by discussing treatment options consistent with the patient’s goals of care. This should include the potential benefits, harms, and burdens associated with various therapies, and the likelihood that the proposed intervention will accomplish the goals that have been specified.

6. Plan for follow-up: Goals and preferences may change over time, so these discussions should be considered part of an ongoing process.

7. Document goals and decisions: This may include documentation in the chart, in advance directives, or if preferences for potentially life-prolonging therapies are clear, in state authorized portable orders such as the physician orders for life-sustaining treatments (POLST).

One out of 4 older adults may require surrogates to make or help make medical treatment decisions before death. Physicians have a responsibility to help these surrogates make decisions consistent with the preferences, values, and goals for care of the patient. However, because of the often uncertain and unanticipated nature of medical illness, even if specific preferences have been laid out in advance directives, these directives may not address the decision at hand and may still require interpretation by the surrogate. Complicating matters further, older adults may desire that future decisions be made on wishes and interests of family members, not just their own stated preferences for care.

Involving surrogates in advance care planning discussions with the patient prior to incapacitation may help increase the chances that the wishes of a patient are known to the surrogate and may help lessen the burden of surrogate decision making. These discussions should focus on preparing surrogates for future decisions, including appointing a health care proxy to serve as a surrogate in the event of incapacity, clarifying and articulating patients values and preferences, and addressing how much leeway surrogates have in decision-making.

Prognostication can be divided into 2 parts. The first is the estimation of the patient’s prognosis by the clinician. The second is communicating the prognosis to the patient and/or family. Prognostication involves more than predictions of survival or mortality. Older adults care about their prognosis for remaining independent, functional, and free from dementia. However, life and death predictions are often implied when individuals ask about “prognosis.” Clinicians should ask patients to clarify the outcome they are concerned about.

Why Prognosis in Older Adults Is Important

Prognostication is a key component in clinical decision making. Prognostication provides patients and families with information to determine realistic, achievable goals of care. It targets interventions to those likely to live long enough to realize the beneficial outcomes. It establishes patients’ eligibility for care programs such as hospice or advance illness management programs. It also impacts decisions outside of the health care setting, including how individuals decide to spend time and their money.

A key part of decision making based on goals of care is the need for explicit consideration of the likely outcomes of possible medical interventions. Simply asking a patient’s preferences for an intervention such as cardiopulmonary resuscitation (CPR) is rather meaningless unless there is consideration of likelihood that the intervention will produce a desirable outcome consistent with the individual’s goals. Furthermore, if outcomes are not explicitly discussed, patients may hold on to erroneous ideas about the likelihood of particular outcomes. However, if misconceptions are corrected and outcomes are clearly discussed, patients may change their preferences for certain interventions to those more consistent with the underlying values.

There are 3 important concepts to remember when considering prognosis in the older adult. The first is that estimating prognosis in older adults is made more complicated in that they are more likely to have more than one chronic progressive illness that impacts life expectancy. In these individuals, it would be inadequate to focus on only 1 problem when estimating prognosis, as it would not take account of the interaction of their medical problems. The second is that most prognostic algorithms in younger patients are based on specific diseases; in the oldest old, however, functional limitations are greater predictors of mortality than chronic conditions. Most disease-specific prognostic algorithms do not adequately account for functional status. The third is that clinical decision making must take into account the likelihood that a patient will live long enough to survive to benefit from a proposed intervention. For example, preventative therapies, such as cancer screening, blood pressure management, and glycemic control, have all been shown to be effective in healthier, highly functional cohorts of older adults. As the benefits of these treatments all require many years to accrue, frail older adults may not realize the benefit in the time they have left to live. They are though exposed to the harms of the intervention, which often occur much earlier than the delayed benefits.

Estimating Prognosis

The most common type of prognostication is simply using clinician judgment and experience. Prognostication based on clinician judgment is correlated with actual survival, however, it is subject to various shortcomings that limit prognostic accuracy. Clinicians are more likely to be optimistic and tend to overestimate patient survival by a factor of between 3 and 5. Clinical predictions also tend to be more accurate for short-term prognosis than long-term prognosis. The length of doctor–patient relationships also appears to increase the physician’s odds of making an erroneous prognostic prediction. Accuracy of clinician predictions may be improved by integrating clinical predictions with some other form of estimating prognosis such as life tables or prognostic indices.

Life tables estimate remaining life by comparing to national averages for individuals of similar age, sex, and race. These estimates give information on median life expectancy, although the heterogeneity in health states and prognosis among older adults of the same age significantly decreases its value. Using clinical characteristics such as comorbidities and functional status to estimate whether a patient will live shorter or longer than the median life expectancy may help individualize prognostic estimates in the clinical setting.

Prognostic indices are a useful adjunctive in prognostication. Clinicians should select indices that predict mortality over a time frame equal to that time to benefit for the intervention. Clinicians should also select indices that have been tested in settings that resemble the patient’s clinical situation, that have reasonable accuracy in predicting risk, and that use readily available data as its variables. A helpful repository of published geriatric prognostic indices can be found at www.ePrognosis.org. Prognostic indices are intended to supplement rather than replace the clinical judgment of clinicians based upon their assessment of the patient’s condition. When using any of these methods to estimate prognosis, it is important to know that it is not a one-time event. Rather, it is a process that involves periodic reassessment.

Non–Disease-Specific Prognosis

Many older adults do not die from a single disease; instead, they die from the interacting effects of multiple chronic conditions, functional impairment, and cognitive decline. Several non–disease-specific prognostic indices have been created in recognition of this fact. These indices were the subject of a systematic review. Here we list some of the highest-quality indices, commenting on their practical application in clinical settings.

• Schonberg 5- and 9-year index for community-dwelling older adults: This index was developed from a nationally representative survey of older adults. Included risk measures are generally aspects of clinical care that most geriatric providers would have access to, including history of diabetes, cancer, independence in instrumental activities of daily living (IADLs), and mobility. The only exception is self-rated health. The 9-year time frame may be particularly useful for making long-term screening decisions.

• Lee 4-year index for community-dwelling older adults: Similar to the Schonberg index, this index was also developed from a national representative survey of older adults. Included risk measures are clinically accessible.

• Walter 1-year index for hospitalized older adults: This index was developed from the Acute Care for Elders dataset from 2 hospitals in Cleveland, OH. All risk measures would be easy to locate in the patient’s medical record, including admission creatinine and albumin, and activities of daily living (ADL) disability at the time of discharge. For decisions about hospice eligibility at hospital discharge, the risk of death at 6-months crosses the 50% threshold in the highest-risk group.

• Porock 6-month index for nursing home residents: All risk measures are derived from the minimum dataset, and should be readily accessible to the clinician.

Advanced Dementia

The long clinical course of advanced dementia makes estimating an accurate short-term prognosis difficult. Individuals with advanced disease may survive for long periods of time with severe functional and cognitive impairments. They are also at risk of sudden, life-threatening complications of advanced dementia, such as pneumonias and urinary tract infections. These complications can serve as a marker of a very poor short-term survival. In one prospective study of advanced dementia residing in a nursing home, the 6-month mortality rates after the development of pneumonia, a febrile episode, or eating problems, were 47%, 45%, and 39%, respectively. Short-term survival rates are similar for individuals with advanced dementia who are admitted to the hospital with either pneumonia or a hip fracture, with 6-month mortality rates exceeding 50%.

Several validated indices have been developed to predict survival in advanced dementia; however, their ability to predict the risk of death within 6 months is poor. An example of a mortality index that can be used in nursing home residents with advanced dementia is the Advanced Dementia Prognostic Tool (ADEPT). The ADEPT can help identify nursing home residents with advanced dementia who are at high risk of death within 6 months, although only marginally better than current hospice eligibility guidelines.

Congestive Heart Failure

The majority of deaths from advanced heart failure are preceded by a period of worsening symptoms, functional decline, and repeated hospitalizations as a result of progressive pump failure. Despite significant advances in the treatment of heart failure, the prognosis in patients who have been hospitalized for heart failure remains poor, with a 1-year mortality rates ranging from 20% to 47% after discharge. The prognosis only worsens for those with multiple hospitalizations. In one prospective study, the median survival after the first, second, third, and fourth hospitalization was 2.4, 1.4, 1.0, and 0.6 years, respectively. Advanced age also worsens prognosis as the median survival decreases to 1 year for 85-year olds after 1 hospitalization and approximately 6 months after 2 hospitalizations.

Other indicators of a poor prognosis in heart failure include patient demographic factors, heart failure severity, comorbid diseases, physical examination findings, and laboratory values. Heart-failure-specific prognostic indices often combine many of these factors to help identify patients who are have a high short-term mortality. The Seattle Heart Failure Model is a well-validated index composed of 14 continuous and 10 categorical variables that provides accurate estimates on 1-, 2-, and 5-year mortality, as well as mean life expectancy both pre- and postintervention. An online calculator is available at http://depts.washington.edu/shfm/.

Chronic Obstructive Pulmonary Disease

Severity of disease, comorbidities, and, to a lesser degree, acute exacerbations influence prognosis in chronic obstructive pulmonary disease (COPD). The most widely studied mortality index in COPD is the BODE index (Table 3–3). It includes 4 variables known to influence mortality in COPD: weight (body mass index [BMI]), airway obstruction (forced expiratory volume at 1 second [FEV₁]), dyspnea (Medical Research Council dyspnea score), and exercise capacity (6-minute walk distance). The BODE index has been shown to be more accurate than mortality predications based solely on FEV₁. However, the BODE index is not useful in predicting short-term life expectancy (in weeks to months).

Cancer

Prognosis for earlier stage cancer is primarily based on tumor type, disease burden, and aggressiveness suggested by clinical, imaging, laboratory, pathologic, and molecular characteristics. Tumor-specific factors tend to lose prognostic significance for patients with very advanced cancer. For these advanced cancers, patient-related factors, such as performance status and clinical symptoms, have increasing significance in regards to short-term mortality. Performance status has consistently been found to be a strong predictor of survival in cancer patients. Several different measures of performance status have been developed, including the Eastern Cooperative Oncology Group (ECOG) (Table 3–4) and the Karnofsky Performance Status Score (KPS) (Table 3–5). High performance status score does not necessarily predict long survival, although low or decreasing prognostic scores have been shown to be reliable in predicting a poor short-term prognosis. Symptoms that are associated with a poor short-term prognosis in advanced cancer include dyspnea, dysphagia, weight loss, xerostomia, anorexia, and cognitive impairment. The Palliative Prognostic Index (PPI) is an example of a tool that predicts short-term survival of advanced cancer patients in the palliative care setting by combining functional status with presence of symptoms of edema, delirium, dyspnea at rest, and oral intake.

Communicating bad news, such as a poor prognosis, to a patient or a patient’s family is one of the most difficult tasks in medicine. Most physicians are not trained in how to communicate about prognosis, most believe their training in prognostication is deficient, and the prognosis clinicians communicate to family tends to be overly optimistic. Yet, the majority of patients and families prefer to discuss prognosis with physicians, even in the face of uncertainty. The consequences of failing to communicate prognosis with patients and their surrogates are great. For instance, patients are more likely to receive aggressive end-of-life care and less likely to receive symptom-directed care when they have a poor understanding of their prognosis.

The SPIKES mnemonic is one way to help remember key steps in deliver bad news such as a poor prognosis (Table 3–6). Prognosis should be framed in the context of ones illness, and be framed in both the positive and negative (eg, “If there were 100 patients in your father’s current condition, in 5 years roughly 80 would die and 20 would survive. I am basing this on his advanced heart failure and his worsening functional status.”). Technical language should be avoided. For example, most individuals do not understand the term “median” survival when used by their physicians. Similarly, vague language such as “good” or “poor” chance of survival may also lead to misinterpretations. Combining both qualitative and numeric language may improve comprehension of prognostic statements.

Exploring patient and surrogate understanding and personal beliefs about prognosis is imperative in these discussions, as there may exist poor concordance in what information the professional perceived was given and information the patient or surrogate understood from a conversation. In addition, few surrogates report basing their view of their loved one’s prognosis solely on the physician’s prognostic estimate. Rather, most attempt to balance the physicians judgment of prognosis with other factors, including (a) their own knowledge of the patient’s intrinsic qualities and will to live; (b) their observations of the patient; (c) their belief in the power of their support and presence; and (d) optimism, intuition, and faith. Furthermore, even in the face of poor prognostic information, patients and surrogates remain optimistic and overestimate survival.

Accurate prognostication allows for clinicians to provide patients and families with realistic options for care given current medical circumstances, and aids in determining which interventions offer little chance of benefit because of competing risks of morbidity and mortality. The use of structured approaches, such as SPIKES, is one way to ensure that this information is delivered in an effective and empathic manner. Prognostic information should be used along with consideration of other health priorities, such as maintaining independence, as part of shared decision making with older adults and their family members.

ePrognosis: www.eprognosis.org (a repository of geriatric prognostic indices)

Seattle Heart Failure Index: http://depts.washington.edu/shfm/ EPERC: http://www.eperc.mcw.edu/EPERC (accessible and clinically relevant monographs on palliative care topics)