End-of-Life Planning in Primary Care Practice

Norman Weinberg, MD, is a primary care internist and researcher focusing on the root causes of physician errors and system issues that affect the quality of health care.

William B Stason, MD, MSci, is an internist, cardiologist, and health policy researcher who has focused his research at the Harvard School of Public Health and Brandeis University on the cost-effectiveness and quality of health care.

ABSTRACT: Primary care physicians (PCPs) perform important tasks by helping patients understand their illnesses and develop advance directives (ADs) that aim to optimize their quality of life and make sensible use of health care resources. However, these ADs are not always followed for several reasons, including failures in communication between patients, specialists, and PCPs and failures of hospital-based staff or emergency departments to take electronic health records or ADs into consideration when making clinical decisions. Increased attention to establishing ADs and using them to guide clinical decisions near the patient’s end of life holds promise to improve his or her quality of life and to reduce health care costs. 

KEYWORDS: End of life, advance directives, quality of life


Recent emphasis on patient safety and medical errors has increased the medical community’s awareness of untoward events in medical care that impair an individual’s quality or length of life.1,2 This emphasis has also prompted questions related to the care that patients with severe chronic illnesses receive during the end of their lives. Some argue that disregard for an individual’s care preferences atthe end of life constitutes a “medical error” that is as significant as wrong limb surgery or a life-threatening medication error.3 The American College of Physicians (ACP) highlights the importance of the compassionate and competent care of dying patients, effective communication with patients and their families, advance care planning, and the legal and ethical rights of informed and alert adults to refuse recommended life-sustaining medical treatments in its Ethics Manual.4 The ACP also underscores the ethical considerations of withholding or withdrawing treatment to clarify the patient’s prognosis. Further, The Conversation Project (http://theconversationproject.org) emphasizes the value of individuals and their families engaging in open, value-directed, end-of-life conversations and expressing personal preferences long before medical crises occur.5 These discussions will provide the building blocks for the development of advance directives (ADs), living wills, or health care proxies in collaboration with primary care physicians (PCPs), specialists, hospitals, andhealth care organizations. 

The goals of ADs are to increase the likelihood that a patient will receive the medical care he or she wishes at the end of life, to enhance his or her quality of life, and to provide important support to his or her family members. Despite their usefulness, however, physicians often face barriers in obtaining an AD because of the patient’s inherent reluctance to discuss end-of-life issues,6 the time pressures of a busy practice, and realistic uncertainties over the patient’s prognosis.7 When ADs are acquired, effective coordination of patient care between PCPs and involved specialists, hospitals, or nursing homes is essential if they are to be successfully implemented. 

Here are 3 case vignettes that illustrate different aspects of clinical decision- making occurring among such patients and a discussion with information on developing best practices for AD development and implementation follow. 

Case 1: Establish ADs as Soon as the Life-Limiting Condition Is Recognized

An 86-year-old female was admitted to a hospital for the surgical repair of a fractured hip. The patient has had multiple falls and sustained a fracture of the tibia and hip 9 months before her death. Her underlying conditions included severe aortic stenosis, moderate pulmonary hypertension (blood pressure of 53 mm Hg),rapidly declining cognition during the last year of her life, recurrent aspiration following surgery for an adenocarcinoma of the tongue, and a previous case of pneumonia for which she was admitted to the hospital 6 months before her death. The patient was placed on hospice care after being discharged from her hip repair surgery; she first established her ADs at this time. She died 3 months later. 

Comments. The consequences of delaying the institution of ADs underscore the need to have concrete criteria for establishing the goals of care. Several of the patient’s conditions carried prognoses that would have given her a life expectancy of under 1 year. During the last 9 months of her life, she had 1 emergency department visit and 3 hospitalizations, including surgery for a fractured hip. It was never determined whether these treatments were consistent with her wishes.

Case 2: PCP Involvement to Ensure That Care Delivered Is Consistent With the Patient’s Wishes 

An 89-year-old man was admitted to the coronary care unit (CCU) of a hospital with recurrent chest pain and an acute coronary syndrome. The patient had known coronary artery disease (CAD) and had undergone a percutaneous angioplasty with placement of drug-eluting stents 1 year earlier. Six months prior to his death, he had been hospitalized for a non–ST-segment elevation myocardial infarction (NSTEMI); he recovered uneventfully. At that time, the patient and his wife agreed that an AD that included a do not resuscitate and a do not intubate order was important. 

At a follow-up medical evaluation, the patient’s PCP noticed that his cognitive status was beginning to decline and that his cardiac status had deteriorated further with the development of severe aortic stenosis and peripheral vascular disease. The patient, his wife, and his physician agreed that the goals of treatment at this point should be focused on comfort measures. Two months later, he developed chest pain that mimicked prior episodes. His wife became alarmed and called 911 when his pain was not relieved after taking 5 doses of nitroglycerin. 

At the emergency department, the patient was diagnosed with NSTEMI; he was subsequently admitted to the CCU and given intravenous heparin, analgesics, aspirin, and clopidogrel. His wife refused to grant permission for a cardiac catheterization, and he was placed onpositive air pressureto avoid intubation. The patient deteriorated and died from cardiogenic shock on the third hospital day.

Comments. This final hospitalization could have been avoided had the patient’s wife been able to contact his PCP or a call-in service with access to the patient’s ADs and the primary goal of keeping him comfortable at home. 

Case 3: Patient’s Prognosis Should Be Defined Using the Best Available Evidence 

An 89-year-old man with advanced vascular and Alzheimer disease died shortly after being admitted to a hospital for aspiration pneumonia. The patient had a score of 19/29 on the Mini-Mental State Examination and could name only 2 of his 3 children and none of his 6 grandchildren upon admission. He had not been ambulatory and had been rapidly losing cognitive skills throughout the year prior to his death. 

His medical history included CAD and a stroke while on warfarin. The patient never established an AD at any point, despite having had multiple hospital admissions during the last year of his life for pneumonia, incarcerated hernia, chest pain, abdominal pain, lysis of adhesions for an intestinal obstruction, and aspiration pneumonia. 

Comments. Comfort care at home should have been the principal goal of care in this case. The multiple hospitalizations and interventions could have been avoided if the patient had established an AD.


ADs can provide valuable guidance for making health care decisions for patients with life-threatening, chronic illnesses. In our study (see An Observational Study), the challenges to developing and implementing ADs in PCP practices and their affiliated hospitals was emphasized: 78% of patients with advanced diseases had ADs, indicating that they did not want life-sustaining care, but 81.4% of those with ADs received treatment in an emergency department or a hospital during the last 6 months of their lives. 

The reasons for hospital or emergency department referrals or admissions often include: 

• Acute changes in the patient’s status that caused the patient or his or her family distress and led directly to 911 calls or to direct transport to an emergency department or hospital.

• Specialists admitted the patient for tertiary treatments without input from his or her PCP.

• Nursing homes sent the patient to a hospital when his or her symptoms changed.

• The patient or his or her family changed what was previously decided.

• The patient suffered an acute trauma, the comfort care provided at the patient’s home was inadequate.

• Difficulties in contacting PCPs when new or increased symptoms developed. Note: In our study, although difficulty in contacting the PCP or the covering physician at the time of an acute change of status i a theoretical barrier, there were no cases where this was a cause for the transition of care. In a number of cases, the patient or family contacted the specialist directly, acording to the type of symptom that developed. The specialist would typically have the patient transferred to the emergency department.

• The patient’s life-limiting diagnosis was not known at the time of his or her emergency department or hospital admission.

• Failures of emergency department or hospital-based medical staffs to examine patients electronic health records (EHRs) and ADs prior to making treatment decisions. 

These findings provide important insight into the need for the further development and implementation of ADs and related health care practice and policy initiatives by physicians, ambulatory practices, and hospitals. They also confirm the results of published studies that indicate that physicians or the patient’s surrogates are often unaware of the patient’s wishes when he or she is being treated for serious conditions8 and that ADs do not influence clinical decisions to admit the patient to an intensive care unit (ICU).9

The benefits of hospitalization in patients with advanced or chronic diseases need to be balanced against the increased risks of infection in the hospital setting, disorientation of the patient in an unfamiliar setting, diminished muscle strength, bone demineralization, and decreased respiratory function due to inactivity, decubitus ulcers, and diminished nutrition.10 In their 2003 study,11 Covinsky et alreported declines in activities of daily living following hospital discharge in 23% of patients aged 70-74 years and in 63% of those aged 90 years and older; decreases in activities of daily living after hospitalizations are strong predictors of 1-year mortality.12 Hospitalizations in elderly patients with multiple comorbid conditions may therefore shorten rather than prolong their lives and may increase rather than improve their levels of disability.13 Comfort rather than cure may be the more appropriateobjective in these individuals. 

In addition, hospitalization requires an increase in the level of care, which adds additional expenses at one’s end of life. Ten percent of the population in the United States accounts for 70% of all health care expenditures,14 and elderly people with severe chronic medical conditions are at especially high risk of repeated and prolonged hospitalizations and therefore, are more likely to be among the highest spending individuals. Admissions of patients with advanced cognitive impairments vary widely in different parts of the country and create particularly difficult challenges.15 A 2011 study found that 19% of nursing home residents had at least 1 burdensome transition of care.15 Further, nursing home residents in the regions with the highest quintile of burdensome transitions were more likely than those in the regions with the lowest quintile to have a feeding tube, spend time in an ICU in last month of life, have a stage IV decubitus ulcer, or be enrolled in hospice.Higher admission rates of such patients tend to occur in regions of the country that have more hospital beds per capita.16 

Although the number of cases in this study was small, important trends were detected that allowed for specific actions targeted at root causes of unwanted and/or inappropriate care. 

An Observational Study

We conducted a 3-year observational study to examine the practices of 8 board-certified PCPs who admit patients to a suburban hospital in eastern Massachusetts. Electronic medical records for these physicians were reviewed to identify patients who had died within a 3-year period between 2008 and 2010. 

Specific objectives were to determine: (1) whether patients had previously diagnosed life-limiting conditions; (2) whether ADs were in place at the time of death; (3) levels of ambulatory primary or specialty care received; and (4) to identify hospital admissions or emergency room (ER) visits that occurred during the last 6 months of their lives. A principal focus was to examine whether transitions in sites of care were consistent with the ADs. The ultimate goal of the study was to identify opportunities for reducing hospitalizations and their associated costs while improving the quality of life of patients with far-advanced diseases. Physicians were interviewed using a structured format to examine these issues (Table 1).        

The Results

Eighty-six patients from the participating practices died during the 3-year observational period between 2008 and 2010. Of these, 67 (78%) had ADs that focused principally on do not resuscitate/do not intubate guidelines. Some ADs also included specifications for preferred site of care, pain relief, or method of feeding. ADs were not obtained from patients who indicated that they wanted to receive all available medical measures. The most frequent underlying diagnoses of these patients were cancer, cardiac-related causes, severe dementia, chronic obstructive pulmonary disease with respiratory failure, and infectious disease (Table 2). 

Site-of-Care Transitions of Patients With Life-Limiting Conditions and ADs

Fifty-nine patients with documented ADs had life-limiting conditions. Accidental deaths, infectious, and postoperative complications are excluded. Of these, 48 (81.4%) were admitted to an ER or hospital during the last 6 months of their lives. Principal reasons for these transitionsor admissionswere acute changes in patient status (38%), treatment decisions by specialists without inputs from PCPs (31%), referrals from nursing homes (13%), and patient or family decisions (10%) (Table 3).  

Three patients had more than 1 reason for the transition. Transitions in sites of care were not consistent with ADs in 51 patients (71.8%). Of these, 47 patients were sent to an ER or admitted to a hospital. The most frequent reasons for referral were an acute change in the patient’s status (37.2%), management decisions by specialists (29.4%), referrals from nursing homes (13.7%), patient or family decisions (11.8%), and trauma in 5.9% (Table 3).

Development of ADs

A thoughtfully conceived AD requires the skills, knowledge, and time of physicians as well as active involvement of the patient and key family members, sensitivity to highly charged end-of-life issues, and ongoing commitments to reviewing and updating ADs as the patient’s physical and mental conditions and prognosis evolve. Explicit training of physicians and other health care providers is important. Multiphysician practice groups affirmthe importance of consistency in methods for obtaining and implementing ADs. Some key issues identified in our study include:

The PCP should discuss ADs sensitively and thoroughly with his or her patient. These discussions may occur earlier or later in the patient’s clinical course but, preferably, should happen when the patient is physically stable and mentally able. 

• The development of ADs should be explicitly scheduled for 1 or more office visits. Preferably, both the patient and a family member or health care proxy should be present. Advanced planning and allowing sufficient time for the meetings are important.

• The physician should use the best available, most recent scientific evidence from the literature or experts to estimate the patient’s prognosis from principal diagnoses and important comorbidities.

• ADs should be reviewed and updated at regular intervals and following any hospitalization or major clinical event, including any changes in symptoms, treatments, or prognosis. 

• When a medical specialist is involved in the patient’s care, consensus with the PCP is needed on treatment objectives and approaches to ensure that they are consistent with the patient’s disease(s) and prognosis—and the AD.


There is currently no consensus on the ideal time for initiating discussions on planning ADs. Some believe that these discussions should begin around a patient’s midlife and evolve before he or she develops serious illness or disability.5 ADs are actually completed more commonly, however, by individuals with advanced diseases or disabilities who are in stable physician–patient relationships.17 The patient’s age, type, and severity of clinical diagnosis, level of physical or psychological function, and family support are all important considerations. Further, certain patient characteristics, such as functional decline, race or ethnicity, chronic disease, and the location of family members, are important determinants of end-of-life expenditures.16 It has also been shown that a physician’s race, age, and gender influence his or her attitudes and preferences for end-of-life treatment.18 For example, the authors of a 1999 study reported that white physicians were more likely than black physicians to agree that tube feeding in terminally ill patients is “heroic,” to agree that physician-assisted suicide is an acceptable treatment alternative, and to want physician-assisted suicide for themselves whereas black physicians were more likely than white physicians to request aggressive treatments for themselves in a persistent vegetative state scenario and to request aggressive treatment in a state of brain damage with no terminal illness.18

Considerations of both the patient and health care professional are essential when developing ADs. Patients may be reluctant to face issues resulting from the progression of diseases or disabilities, and health care proxies may have to intervene. For physicians, time limitations and discomfort engaging in end-of-life care discussions may be factors in executing appropriate AD planning. At a minimum, impediments must be identified and explored. An approach that worked well in the primary care practices that we examined in our study was for the physician to introduce the topic and provide key background material at 1 office visit, and then schedule a discussion on the topic for a future visit. 

All individuals who are closely involved in the patient’s treatment should be present for AD discussions and should be familiar with his or her medical problems, related prognoses, and treatment objectives. The patient’s goals for treatment, symptom relief, and physical and psychological comfort as well as his or her attitudes toward cardiac resuscitation, life support, pain relief, and feeding are all important. Preferences should be made explicit before the AD is finalized and signed by the patient, his or her health care proxy, and the responsible physician or other health care professional. As stated previously, it is important that ADs be reviewed at regular intervals, in the face of worsening symptoms or levels of disability, and following medical events or hospitalizations. 

One of the most difficult challenges for the physician is to understand and convey the patient’s prognosis using the best, most recently available scientific evidence related to the principal medical diagnosis and associated comorbidities.Reliable prognoses are essential when making future decisions on the types and intensities of treatment. Most patients in our study had prognoses indicated in their medical records only during the final months of their lives. Some examples of evidence on the relationships between patient prognosis and treatment preferences relate to decisions on cardiopulmonary resuscitation,19 treatment preferences for cancer, and the care of seriously ill hospitalized patients.20 

Clinical decision-making becomes even more complex when patients are cared for by both PCPs and medical or surgical specialists. In these cases, ADs need to be endorsed by all involved physicians involved in the patient’s care and timely communication must occur when changes in treatment or decisions to hospitalize the patient are considered. However, the  physician’s understanding of the prognosis of patients and discussions about the patients’ and family wishes has been found in 1 study21 to be ineffective in improved care once the patient is hospitalized. Note: The patients in this study all had advanced illnesses and limited life expectancies. 

Implementation of ADs

Once developed, the AD becomes an intrinsic part of the patient’s health record and needs to be referred to when making treatment decisions. A critical element of successfully implementing ADs is effective communication between the patient and his family and the involved physicians.22 As previously stated, the frequent barriers to AD implementation in our study included family members calling 911 or taking the patient directly to the emergency department or hospital in the face of anxiety caused by new or more severe symptoms, communication problems with the PCP offices, and failures of emergency departments or hospital clinicians to access patients’ EHRs, which included ADs, before implementing treatments. A reliable means of 24 hour/7 days-per-week access to communication, easy access of complete and up-to-date EHRs and ADs for covering physicians, and steps to increase clinician motivation to access EHRs and consider ADs while making health care decisions are some solutions to these problems. 

Frequent challenges to AD implementation in our study were patients who were receiving care from both a PCP and a specialist and patients who were referred from skilled nursing facilities (SNFs) to the emergency department or hospital without first considering ADs. Specialists frequently referred patients with far-advanced disease to hospitals for tertiary treatments, such as chemotherapy or cardiac procedures, without considering the AD or discussing this course of action with the PCP. Improved communication is important to achieve the best possible balance between the benefits of tertiary treatments and patient wishes expressed in ADs. In our study, referrals to hospitals from SNFs often reflected a lack of knowledge of patient wishes expressed in ADs coupled with legitimate medical and medicolegal concerns. Steps to facilitate AD implementation by emergency department staff, hospitalists, or PCPs are included in Table 4.Elements that promote successful end-of-life care based on the results of our study are listed in Table 5

Hospital Collaborative Working Group on Adherence to ADs 

The Care Transitions Collaborative (a working group comprised of hospital staff and numerous community partners) at the community hospital in Massachusetts stimulated the formation of a physician subgroup, an information systems subgroup, a community education subgroup, and an emergency department subgroup charged with addressing the major concerns relative to the use of ADs. Each of these groups is addressing 1 or more of the following major issues that impact the use and effectiveness of ADs: (1) disease-specific guidelines for recognizing a patient’s end of life; (2) increase in the use of ADs in guiding clinical decisions; (3) improved information technology to ensure that all health care providers have access to the most recently executed ADs; (4) emergency department protocols that enhance the use of ADs; (5) reduced avoidable admissions; and (6) education of community groups in the value and use of ADs. 

Efforts are being made to establish quality measures that reflect the number of patients with life-limiting expectancies who have ADs and the compliance of treatment with the ADs. Pay-for-performance measures will be considered.

Establishing ADs is not, by itself, enough to ensure that the patient’s preferences will be respected. System wide changes in education of the family as to what to expect at the end of life is critical in avoiding panic with acute changes in the patient’s status. In addition, specialists need to become familiar with ADs determined by the PCPs to avoid unnecessary and unwanted hospital admissions.  


The authors would like to thank Dr. Henry Vaillant, chair of the department of medicine, Emerson Hospital, Concord, MA, for his guidance and support of this study. We would also thank the following physicians for their cooperation in the study: Drs.Charles Keevil, Lynn Weigel, Peter Hoenig, Marguerite Roach, Pamela DeLuca, Betty Pomerleau, Magda Mikhil, and Sunita Hunjura.


  1. Kohn LT, Corrigan JM, Donaldson MS, eds. To Err is Human: Building a Safer Health System. Washington, DC: National Academy PR; 2000. 
  2. Graber ML. The incidence of diagnostic error in medicine. BMJ QualSaf. 2013;22:ii21-ii27.
  3. Allison TA, Sudore RL. Disregard of patients’ preferences is a medical error: comment on “Failure to engage hospitalized elderly patients and their families in advance care planning.” JAMA Intern Med. 2013;173(9):787.
  4. American College of Physicians. ACP Ethics Manual. 6th ed. Philadelphia, PA: ACP; 2012.
  5. Bisognano M, Goodman E. Engaging patients and their loved ones in the ultimate conversation. Health Affairs. 2013;32:203-206. 
  6. Christakis NA, Lamont EB. Extent and determinants of error in doctors’ prognoses in terminally ill patients. BMJ. 2000;320(7233):469-473.
  7. Gillick MR. A broader role for advanced medical planning. Ann Intern Med. 1995;123(8):621-624.
  8. Covinsky KE, Fuller JD, Yaffe K, et al. Communication and decision-making in seriously ill patients: findings of the SUPPORT project. The Study to Understand Prognoses and Preferences for Outcomes and risks of Treatments. J Am Geriatr Soc. 2000;48(5 suppl):S187-S193.
  9. Kish Wallace S, Martin CG, Shaw A, Price KJ. Influence of an advance directive on the initiation of life support technology in critically ill cancer patients. Crit Care Med. 2001;29(12):2294-2298. 
  10. Creditor CC. Hazards of hospitalization of the elderly. Ann Intern Med. 1993;118(3):219-223. 
  11. Covinsky KE, Palmer RM, Fortinsky RH, et al. Loss of independence in activities of daily living in older adults hospitalized with medical illnessess: increased vulnerability with age. J Am Geriatr Soc. 2003;51(4):451-458.
  12. Walter LC, Brand RJ, Counsell SR, et al. Development and validation of a prognostic index for a 1-year mortality in older adults after hospitalization. JAMA. 2001;285(23):2987-2994.
  13. Inouye SK, Peduzzi PN, Robison JT, et al. Importance of functional measures in predicting mortlity among older hospitalized patients. JAMA. 1998;279(15):1187-1193.
  14. Bodenheimer T, Fernandez A. High and rising health care costs. Part 4: can costs be controlled while preserving quality? Ann Intern Med. 2005;143(1):26-31.
  15. Gozalo P, Teno JM, Mitchell SL, et al. End of life transitions among nursing home residents with cognitive issues. N Engl J Med. 2011;365(13):1212-1221.
  16. Kelley AS, Ettner SL, Morrison RS, et al. Determinants of medical expenditures in the last 6 months of life. Ann Intern Med. 2011;154(4):
  17. 235-242.
  18. Duffield P, Pozamsky JE. The completion of advance directives in primary care. J Fam Pract. 1996;42(4):378-384.
  19. Mebane EW, Oman RF, Kroonen LT, Goldstein MK. The influence of physician race, age and gender on physician attitudes toward advance care directive and preferences for end-of-life decision making. J Am Geriatr Soc. 1999;47(5):579-591.
  20. Murphy DJ, Burrows D, Santilli S, et al. The influence of the probability of survival on patient’s preferences regarding cardio-pulmonary resuscitation. N Engl J Med. 1994;330(8):
  21. 545-549.
  22. Weeks JC, Cook F, O’Day SJ, et al. Relationship between cancer patients predictions  of prognosis and their treatment preferences. JAMA. 1998;279(21):1709-1714. 
  23. The SUPPORT principal investigators. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). JAMA. 1995;274(20):1591-1598.
  24. Gordon NP, Shade SB. Advance directives are more likely among seniors asked about end-of-life care preferences. Arc Intern Med. 1999;
  25. 159(7):701-704.