A 93-Year-Old Man With Advanced Dementia and Eating Problems

DR BURNS: Mr P is a 93-year-old retired engineer from Russia with a history of dementia, hypertension, type 2 diabetes mellitus, benign prostatic hypertrophy, chronic renal insufficiency, and presumed renal cell cancer. He has Medicare and Medicaid insurance.

Mr P's dementia was diagnosed 10 years ago, manifesting as gradually progressive short-term memory loss, with a sharper decline during the past 2 years. He moved to the United States in 1994 and has lived with his daughter since his wife died 20 years ago. Prior to October 2006, he could be left alone for several hours, ambulated with a walker, and fed and toileted himself independently but needed assistance with his other activities of daily living.

In October 2006, Mr P was hospitalized for a witnessed episode of “unresponsiveness.” Magnetic resonance imaging/magnetic resonance angiography revealed age-related atrophy of his brain and an electroencephalogram result was normal. A methicillin-resistant Staphylococcus aureus urinary tract infection was diagnosed. An abdominal computed tomography scan showed a right kidney lesion suggestive of renal cell carcinoma, but Mr P's daughter declined further evaluation. His unresponsiveness was attributed to dementia. He was discharged to his daughter's home but was readmitted a few weeks later with another episode of unresponsiveness. An electroencephalogram demonstrated encephalopathy without focal abnormalities. At this point, his daughter felt unable to continue caring for him. He was discharged to a nursing home.

At the time of nursing home admission, Mr P had profound memory deficits, lacked the cognitive skills to participate in daily decision making, was dependent in all his activities of daily living, and was incontinent of urine and stool. He could recognize his daughter. His dementia was complicated by agitation and aggressive behavior. He ambulated with a walker but had recurrent falls. His oral intake was poor and he occasionally refused to eat. A bedside swallowing evaluation revealed mild oral dysphagia. A mechanically altered diet, nutritional supplements, and dental referral for ill-fitting dentures were recommended. Nonetheless, his weight declined from 138 lb to 128 lb over the next 2 months. His body mass index was 19.5.

In January 2007, Mr P was transferred to a special dementia care unit because of behavior problems. After he became accustomed to the staff and was treated for constipation, his oral intake improved. In March 2007, Mr P fell and fractured his right femur. He was hospitalized and underwent surgical repair of his hip. On returning to the nursing home, Mr P was at his baseline cognitive status but refused to eat or drink. Intravenous hydration was provided for several days, although his intake did not improve and he ultimately pulled out the intravenous line. After consultation with Mr P's physician, his daughter decided not to restart intravenous fluids or initiate tube feeding.

In September, all of a sudden he had seizures. He was in the hospital for 4 weeks and he went home, then in a short time he was back. And they couldn't find anything. He didn't want to eat, he was in terrible condition. I was offered to have a feeding tube at that time because he didn't really eat. But he was yelling at me, “Leave me alone!” He was pushing the table, he didn't want to eat, and then when he did it was just a little. Then he was moved over to the nursing home and it was like—I don't know how to say, but little by little, he was eating better and he was drinking better. But if he doesn't want it, he just pushes the food. I can't force him, I can’t. Because some people you can do it. But with him, he will tell you how much he loves you, but he wouldn't do it, because he doesn't want it.

More recently, he was taking just 1 or 2 spoons or some juice, and then he was pushing it away. He was asking for a drink, and they gave it to him, and he was pushing the drink away and nobody could understand what he was doing. Also he has his IV, he pulls it out. “Father, you're going to die,” I told him. “I know, but I don't want any more.” He was saying, “What a terrible life, I don't want.” I talked to my relatives in Russia and they are doctors, and they said don't do it. Because you would never do it for yourself, like everybody. He's 93, and oh, the life he would have. For me it would be just so nice to come to see him, and to feel still that I have somebody who could somehow protect me.

Mr P's medical history represents a lot of older people who get to the end of a long, dementing illness. He first came to this center in October 2006 and was on one of the more functional floors. In January he was transferred to this unit, which is more a palliative care unit. He very much settled into the floor, and everything was going progressively better until he fell in March and fractured his right hip. As we often see, you can sort of fix the individual pieces but you can't get the jigsaw back together. And his hip was fixed and it was orthopedically a success, but he refused to eat and drink.

I think for families, there's often the concern of “Will they suffer more by not receiving intravenous fluids or sufficient nutrition?” And I think part of the job of geriatricians and palliative care physicians is to help them understand that it doesn't automatically mean that there will be more suffering. This is such a gift that you can give to a family because it's amazing how many decades later people carry the imagery of their loved ones dying in their head.

What is the epidemiology of advanced dementia? What are the typical clinical course and common problems experienced by patients with advanced dementia? What tools are available to prognosticate life expectancy in advanced dementia? What is the role of hospice in caring for patients with advanced dementia? What is the role of pharmacotherapy? Why do patients with advanced dementia get eating problems? What are the treatment options and the pros/cons and common misconceptions for each? How can clinicians help guide feeding decisions? What do you recommend for Mr P and his daughter? What does the future hold?

In the year 2000, more than 4.5 million individuals in the United States were estimated to have Alzheimer disease. By 2050, this number is expected to exceed 13 million.¹ In 2003, Alzheimer disease was the fifth leading cause of death in US individuals older than 65 years.² The annual mortality rate from Alzheimer disease increased by 24% from 2001 to 2004, while the mortality rate for heart disease, cancer, and stroke declined during the same period.^{3 - 4} The cost of caring for individuals with Alzheimer disease is more than $150 billion annually in the United States alone.⁵ These statistics likely underestimate the clinical and societal burden of dementia because they do not consider other causes of dementia (eg, vascular) and are derived from death certificate data, which typically underrepresent dementia as a cause of death.⁶

The best current estimates indicate that median survival after the onset of symptoms of dementia ranges from 3 to 6 years, shorter than previously estimated.^{7 - 8} Approximately 70% of persons with dementia die in nursing homes.⁹ Therefore, nursing homes are key providers of terminal care to these patients.

Although advanced dementia clearly meets accepted standards of an end-of-life condition,¹⁰ unlike metastatic cancer, it is not widely recognized as a terminal illness.^{11 - 16} To date, little research has focused on the experience of persons dying with dementia. In a 1993 survey, 70% of physicians and 60% of family members thought that care in advanced dementia should focus on palliation.¹⁷ Nonetheless, data suggest that these patients do not receive adequate palliative care.^{11 ,15 - 16 ,18 - 19} Compared with other terminally ill patients, those with dementia receive less optimal pain control,^{18 - 19} have fewer advance directives,¹¹ undergo more burdensome interventions,^{11 - 12 ,14} and receive less hospice care.^{15 - 16 ,20 - 22}

Unless another fatal illness intervenes, patients with dementia progressively lose cognitive and physical function and reach the advanced stage of their disease. The Global Deterioration Scale is a widely used dementia staging system ranging from 1 (least severe) to 7 (most severe).²³ Stage 7 is defined as having profound memory deficits, total functional dependence, no knowledge of recent or past events, little to no verbal communication, and limited ability to ambulate. The duration of this stage ranges from months, as was the case with Mr P, to years.

The clinical course of advanced dementia is characterized by “sentinel events” that require decision making to direct care toward either palliation or more aggressive measures. These events include complications that are expected in end-stage dementia, particularly eating difficulties and recurrent infections,²⁴ such as Mr P's urinary tract infections. Pneumonia is the immediate precipitating cause of death in more than 50% of cases.^{25 - 26} Sentinel events in advanced dementia also include acute illnesses common among frail elderly persons; for example, the hip fracture experienced by Mr P.

Prognostication in advanced dementia is challenging and a barrier to providing end-of-life care.^{15 ,27 - 32} Prognostication often helps guide end-of-life decision making^{33 - 34} and is used to determine hospice eligibility, which requires an estimated life expectancy of less than 6 months.³⁵ Nonetheless, many experts believe that patients with end-stage dementia deserve high-quality palliative care regardless of life expectancy.^{30 - 31 ,36}

Hospice eligibility for dementia is largely based on the Functional Assessment Staging (FAST) scale ( Article ). FAST consists of 7 major stages.³⁷ Stage 7 consists of 6 substages: 7a = speech is limited to 5 words or fewer; 7b = all intelligible speech is lost; 7c = nonambulatory; 7d = unable to sit independently; 7e = unable to smile; and 7f = unable to hold head up. Stage 7c is currently used as the cutoff for hospice enrollment.³⁵ This application of FAST has been criticized because the tool does not accurately predict 6-month survival^{30 - 31} and because scoring requires that patients advance through the stages in a sequential fashion, which often does not occur.^{28 ,36} In one study, the FAST criteria could not be applied to 41% of patients because their disease had not progressed in the ordinal sequence of the scale.²⁸ Mr P, who remained ambulatory until very late in his course, would not have qualified for hospice based on a primary diagnosis of dementia until after his hip fracture.

To be eligible for hospice, patients must meet both of the following criteria:

   Functional Assessment Staging (FAST)³⁷ : Patient must be at or beyond stage 7c and show all of the features of stages 6a-7c.
   Medical conditions: Patients must have had at least 1 of the listed medical conditions over the prior year.

Few validated models have been developed to predict survival in advanced dementia.^{27 ,30 - 31 ,38 - 39} In one study that developed and validated a model, the Minimum Data Set (MDS) was used to create a 6-month mortality risk score in newly admitted nursing home residents with advanced dementia.³⁰ The MDS is a standardized resident assessment instrument mandated for use in all licensed US nursing homes.⁴⁰ The MDS-derived risk score predicted 6-month mortality with reasonable accuracy (area under the receiver operating characteristic curve [AUROC], 0.71) and had greater discrimination than FAST stage 7c simulated with MDS variables (AUROC, 0.51).³⁰

Hospice care improves the end-of-life experience for patients, including those with dementia.^{18 ,21 ,41 - 42} Some demonstrated benefits of hospice among nursing home residents with dementia include a lower probability of hospitalization during the last 30 days of life (hospice, 19%; nonhospice, 39%)²¹ and a higher probability of regular treatment for daily pain (hospice, 44%; nonhospice, 27%).¹⁸ A survey of 77 123 bereaved family members of patients who received care from 796 hospices nationwide found that 73% of respondents whose loved one died with dementia (N = 8686) evaluated hospice services as excellent, similar to that reported for cancer (78%) and other chronic terminal conditions (76%).⁴³ Thus, hospice appears to manage the specific problems encountered in end-stage dementia (eg, symptom control in cognitively impaired persons) with reasonable success.

Even though hospice benefits persons with advanced dementia, most do not receive these services. Sachs et al¹⁵ estimated that 1 in every 10 persons dying with dementia receives hospice care. In 2005, only 10% of all hospice enrollees nationwide had a primary diagnosis of dementia compared with 46% with cancer.⁴⁴ Barriers to hospice enrollment include accurate prognostication,^{15 ,27 - 32} lack of recognition of dementia as a terminal condition,^{11 - 16} and accessibility of hospice services in nursing homes.^{18 ,21 - 22} Hospice providers cite prognostication as the most difficult of these challenges.^{29 ,32} Despite improved models to estimate survival in dementia,³⁰ these tools are imperfect and their practical application to determine hospice eligibility remains problematic.^{30 ,45} Thus, as long as hospice requires a very high likelihood of death within 6 months, this criterion will continue to hinder the delivery of hospice services to dementia patients.

The use of pharmaceutical agents in advanced dementia needs to be guided by the goals of care. Cholinesterase inhibitors (eg, donepezil, rivastigmine) may provide modest reductions in the rate of symptomatic and functional decline in some patients with mild to moderate Alzheimer disease.⁴⁶ Similar benefits have been demonstrated for memantine (an N-methyl-D-aspartate receptor antagonist) in moderate to severe Alzheimer disease (Global Deterioration Scale stage ≤6).^{46 - 47} These drugs have not been tested in very advanced disease (Global Deterioration Scale stage 7) and have adverse effects (eg, nausea, anorexia). Thus, there is little rationale for administering these pharmacologic agents to patients in the final stage of dementia who have profound cognitive and functional deficits and are near the end of life.

Pharmacotherapy is sometimes considered to try to control behavior disturbances in advanced dementia. However, a growing body of evidence suggests that antipsychotic medications are not efficacious in this situation⁴⁸ and may be associated with a higher mortality.⁴⁹ Thus, nonpharmacological interventions are the preferred approach to behavior management in advanced dementia.

A primary role of pharmacotherapy in advanced dementia is to help promote comfort. To that end, it is reasonable to prescribe drugs that can be administered safely and comfortably to aim to relieve pain (ie, analgesics), dyspnea, and other burdensome symptoms.²⁴

Eating problems are a hallmark of end-stage dementia²⁴ because the ability to eat independently is generally the last activity of daily living to be lost prior to death.⁵⁰ However, patients with advanced dementia can live for a relatively long time despite poor oral intake.⁵¹ One theory posits that patients with advanced dementia have an altered state of homeostasis, characterized by a reduced metabolic rate and lower caloric requirements.⁵²

There are several causes of eating problems in advanced dementia.^{24 ,53} Many patients develop oral dysphagia, manifesting as absent or continuous chewing with a tendency to pocket or spit food. Pharyngeal dysphagia is also common, presenting with delayed swallowing initiation, multiple swallows, and aspiration that often leads to pneumonia. Some patients with advanced dementia lose the ability to perform the task of eating or to interpret the sensation of hunger, while others, like Mr P, may simply refuse to eat. Depression may also present as disinterest in food, and intake may improve with antidepressant treatment.⁵⁴

At the onset of eating problems, acute medical problems (eg, infection, stroke, medication adverse effects) need to be excluded. Easily reversible causes should be addressed in keeping with the patient’s goals of care, as was done for Mr P by fixing his dentures, managing his constipation, and altering the texture of his food. Conservative measures may improve oral intake, such as offering finger foods, smaller portions, favorite foods, or nutritional supplements.²⁴ A bedside swallowing evaluation may help to identify food consistencies that are easier to swallow and positions that reduce the risk of aspiration.⁵⁵

Despite the aforementioned efforts, most patients with end-stage dementia will continue to have eating problems, leaving essentially 2 options: continuing to feed by mouth or placement of a long-term feeding tube (ie, percutaneous endoscopic gastrostomy tube).

The main goal of continued oral feeding is to provide food and drink to the extent that it is enjoyable for the patient. This option focuses on comfort rather than life prolongation. Conscientious hand feeding is labor intensive. Approximately 45 to 90 minutes per day are needed to hand feed and deliver oral medications to patients with advanced dementia⁵⁶ ; thus, insufficient staff resources are a barrier to hand feeding.⁵⁷ Even with the best efforts, food consumption is often inconsistent and continues to decline. However, in the decision to continue oral feeding, the goal of providing a prescribed daily caloric intake is inherently abandoned in favor of palliation.

There are several benefits to hand feeding. This approach allows patients to maintain the pleasure of tasting food, insofar as it remains comfortable for them to do so. Participation in hand feeding also provides an opportunity for family members to care for their loved ones and for formal caregivers to interact with their patients. Family members must be reassured that the decision to hand feed does not imply the discontinuation of medical care. Palliation includes a host of treatments that should be outlined to the family. Moreover, families may opt for hand feeding while at the same time choosing potentially curative treatments for other problems if consistent with goals of care, such as antibiotics for infections or surgical repair of a hip fracture, as in the case of Mr P.

Placement of a feeding tube is another approach to eating problems in advanced dementia. Approximately one-third of US nursing home residents with advanced dementia are tube fed,⁵⁸ with a 10-fold variation in prevalence across the 50 states.⁵⁹ Patient characteristics consistently associated with a higher likelihood of being tube fed include younger age,^{58 ,60 - 62} nonwhite race,^{13 ,58 ,61 - 62} and lack of advance directives.^{13 ,58 ,60 - 62}

The most commonly cited reasons for tube feeding in advanced dementia include prolonging life, improving nutrition, preventing aspiration, and providing comfort.^{63 - 64} Two sentinel reviews detail the paucity of evidence to support these purported benefits.^{65 - 66} However, a randomized controlled trial of tube feeding in advanced dementia has not been conducted and would be difficult to justify ethically. Therefore, it is impossible to definitively know the outcomes of this intervention. The highest-quality data are derived from observational studies that are limited by selection bias.^{13 ,60 ,67 - 71} Acknowledging these limitations, important findings have emerged from this earlier work.

The best available cohort studies suggest that tube feeding does not prolong survival in advanced dementia.^{13 ,60} Tube feeding has also not been shown to improve nutritional status or the clinical consequences of malnutrition, such as pressure ulcers.^{70 - 71} While these results may seem counterintuitive, as with other end-stage illnesses, such as terminal cancer or AIDS, persons with advanced dementia may simply be too debilitated to derive a demonstrable survival or nutritional benefit from tube feeding. Alternatively, complications from tube feeding may actually promote decline (eg, electrolyte disturbances, use of physical restraints, diarrhea).^{67 - 68 ,71}

Prevention of aspiration is another commonly perceived goal of tube feeding. However, tube feeding will not prevent aspiration of oral secretions or regurgitated gastric contents. As such, data from nonrandomized studies consistently indicate that tube feeding severely demented patients who aspirate will not prevent further aspiration.^{67 - 69}

Family members often express concern that their relatives with end-stage dementia may suffer from hunger or thirst.^{63 - 64} It is not known to what degree such sensations are experienced in advanced dementia. One study found no measurable increase in discomfort following a decision to withhold artificial nutrition or hydration.⁷² Dying cancer patients with minimal oral intake report that they do not sense hunger or thirst beyond what can be alleviated with swabs or ice chips.⁷³

Tube feeding is associated with risks that warrant consideration.^{67 - 68 ,71} Up to one-third of older tube-fed patients experience transient gastrointestinal adverse effects (ie, vomiting, diarrhea).⁷¹ Serious mechanical problems, such as bowel perforations, are rare (<1%). However, tube dislodgement, blockage, and leakage are more common (4%-11%) and often necessitate transfer to an acute care facility.^{71 ,74} Agitated demented patients may require physical or chemical restraints to prevent tube dislodgment, and tube feeding has been shown to be associated with greater restraint use.⁶⁸ Tube feeding is also independently associated with lower family satisfaction with end-of-life care for nursing home residents with advanced dementia.⁷⁵

In summary, the best available evidence fails to demonstrate any health benefits of tube feeding in advanced dementia. Moreover, there are risks associated with the intervention. Given that feeding problems occur at the final stage of this illness, expert opinion is that hand feeding should be the default approach and that tube feeding should not be offered.^{65 - 66}

Decision making for feeding problems is among the most difficult clinical crossroads in the course of dementia. The decision is challenging because it is emotional, value-laden, reliant on substitute decision making, based on limited outcome data, and influenced by many external factors. Faced with these challenges, practitioners should be prepared to provide effective and compassionate decision support.

Advance care planning plays a critical role in feeding decisions. The lack of advance directives is a consistent risk factor associated with feeding tube insertion in dementia.^{13 ,58 ,60 - 62} Clinicians have an opportunity to prepare patients with early disease and their families about what to expect in the later stages of dementia, including eating problems. Ideally, these preparatory discussions provide the context for broader advanced care planning and a scaffold for future discussions.

Feeding decisions should follow an ethical framework reflecting a balance of benefits and burdens and respect for autonomy.^{65 ,76} The following steps offer a general approach for clinicians to operationalize this framework (Table)⁸⁴ : (1) clarify the clinical situation; (2) establish the primary goal of care; (3) present the treatment options and their risks and benefits; (4) weigh the options against values and preferences; and (5) provide additional and ongoing decision support.

Clarification of the clinical situation involves an explanation of feeding problems within the context of the natural history of end-stage dementia and any compounding acute conditions, such as Mr P's recent hip fracture. Patients and surrogates are more likely to choose care directed toward comfort if they understand that the end of life may be near.^{33 - 34} A recent study used a video to provide a visual depiction of an actual patient with advanced dementia.⁷⁷ Participants were significantly less likely to opt for aggressive care in the event that they developed advanced dementia after as compared with before they viewed this video. A Dutch ethnographic study underscored that an understanding of the current clinical situation was among the most important factors influencing decisions to withhold artificial nutrition and hydration in patients with advanced dementia.⁷⁸

Counseling for feeding problems in advanced dementia requires the elicitation of the primary goal of care and understanding where the treatment options fit in with this goal. One framework prioritizes treatment goals into the following broad categories: life prolongation, maintenance of function, and maximization of comfort.⁷⁹ Life prolongation prioritizes survival even at the expense of potentially burdensome treatments. Maximizing comfort gives precedence to palliation, with a willingness to forgo uncomfortable interventions even if they may prolong survival. Maintenance of function, lying between these extremes, may include low-risk, minimally invasive but potentially curative treatments (ie, parenteral antibiotics, intravenous fluids).

To promote informed decision making, clinicians should be knowledgeable about the potential risks and benefits of hand feeding vs tube feeding as well as about the limitations of the data. Of concern, in a 2001 survey of 195 primary care physicians (50.8% general internists, 49.2% family physicians), the majority believed tube feeding in advanced dementia reduced aspiration pneumonia (76%), prolonged survival (61%), and improved nutrition (93.7%), revealing a discord with the published literature.⁸⁰ Subsequent work further substantiates that surrogates' high expectations of tube feeding may be based in part on faulty information.⁸¹

Surrogates need to be advised that decisions should be based on their perceptions of the patients' wishes as guided by the ethical principles of substitute decision making: (1) consideration of advance directives; (2) making a substituted judgment; and (3) weighing what is in patients' best interests.⁸² Mr P's daughter made a substituted judgment that her father would not want to be tube fed, as reflected in her repeated phrase, “He does not want it.” In one survey, 48% of surrogates for tube-fed patients with dementia were not confident that the patients would have chosen the intervention for themselves.⁶³

Finally, clinicians must advocate for patients in the face of many external factors influencing feeding decisions in advanced dementia, as evidenced by the varying use of tube feeding across facilities,^{58 ,61} states,^{59 - 62} and countries.^{63 ,83 ,89} External factors driving these differences include regional laws,⁹⁰ facility characteristics,^{58 ,62} fiscal and regulatory policy,^{56 ,58 ,91 - 92} and societal attitudes toward death and dying.⁸³ Approximately 20 US states have more stringent statutory standards to withhold artificial nutrition and hydration.⁹⁰ At the facility level, nursing homes in many states receive higher reimbursement for tube-fed compared with hand-fed patients, while at the same time, hand feeding requires considerable staff resources, presenting a potential fiscal incentive for tube feeding.^{56 ,91} In addition, nursing home quality indicators are now publicly available⁹² that include “weight loss.” Thus, facilities may feel compelled to offer tube feeding to avoid penalization, even when weight loss occurs as part of the dying process.

Additional strategies to improve feeding decisions in advanced dementia are available. A rapid-cycle quality improvement intervention reduced the number of feeding tubes placed in hospitalized dementia patients.⁹³ Decision support tools in the form of audio booklets^{84 - 85} and videos⁷⁷ have been shown to reduce decisional conflict among surrogates. Materials formulated for proxies are available,⁸⁶ and guidelines regarding artificial nutrition and hydration have been developed by professional societies (eg, American Geriatrics Society,⁸⁷ American Academy of Hospice and Palliative Medicine⁸⁸ ). Interdisciplinary team members (eg, nurses, social workers, clergy) are additional resources for families, and team meetings can be helpful to strive for unified decisions. On rare occasions, consultation by an ethics committee may be indicated to resolve conflict between decision makers.

Mr P's daughter made the difficult decision not to pursue artificial nutrition and hydration for her father. In my view, this was the appropriate decision based on the clinical circumstances, her father's perceived wishes, and the available evidence. Therefore, at this point, the focus of his care is palliation. For Mr P, this involves standard approaches to alleviate physical discomfort (eg, relief of dyspnea, pain control). Pain measurement is challenging in advanced dementia. However, disease-specific tools to assess pain and discomfort in this condition have been developed and validated.^{94 - 96} If available in Mr P's facility, consultation by a specialized palliative team or hospice referral is worth pursuing. Hospice would not only benefit Mr P but would provide support to his daughter. Family members experience grief both before and after the death of their loved one with dementia.⁹⁷ Few nursing homes offer grief and bereavement support,⁹⁷ whereas these services are standard components of hospice care.^{97 - 98}

The future of advanced dementia care lies in combined clinical, research, educational, and policy initiatives aimed at increasing recognition of dementia as a terminal condition and determining the best way to provide high-quality palliative care to the growing number of individuals dying with this condition. Comprehensive clinical palliative care programs are desperately needed in the nursing homes where the majority of these patients die.⁹⁹ Research is under way to better describe the natural history, the sources of patient burden, and family experience in advanced dementia.¹⁰⁰ Rigorous evaluation of targeted interventions to improve end-of-life outcomes is the next critical step. While the number of specialists with board certification in both geriatrics and palliative care is increasing, educational strategies are also needed to improve the knowledge and counseling skills of primary care physicians, hospitalists, and physicians who insert percutaneous endoscopic gastrostomy tubes (eg, gastroenterologists, surgeons). However, these efforts will be of limited benefit without the implementation of rational health care policies. To that end, new policies should expand hospice eligibility for patients with advanced dementia and promote fiscal (eg, reimbursement for hand feeding) and regulatory (eg, palliative care quality indicators) incentives for nursing homes that better meet their needs caring for patients at the end of life.

QUESTION: In your experience or that of others, are there any examples of tube-fed patients who improve to the extent that tube feeding can be discontinued?

DR MITCHELL: In situations where there is a reversible cause for an eating problem, such as a stroke, older patients may improve to the point that they regain their ability to eat independently and tube feeding can be discontinued.^{71 ,81} However, in advanced dementia, the etiology of the eating problem is usually not reversible, so in my experience, it is very uncommon for tube feeding to be discontinued because the patient regains the ability to eat orally. More often, tube feeding in advanced dementia is discontinued because the patient declines further intervention and the goals of care are redirected toward palliation.

QUESTION: I wonder if you could reflect for a moment on what I see as a big failure in US medicine. One of the things we are supposed to do as physicians is present options. But with everything we know about this situation, why in the world would we be presenting tube feeding as an option to a patient like Mr P?

DR MITCHELL: In my personal practice, I don't introduce the option of tube feeding for a patient with advanced dementia, unless the health care proxy specifically brings it up as a possibility. However, once the option of tube feeding is raised, physicians must be prepared to counsel the family through the decision-making process.

On a broader level, there are several possible reasons why tube feeding remains an option that is commonly presented to patients with end-stage dementia in the US health care system. These reasons include the strong symbolic significance of feeding, an emphasis on individual autonomy in medical decision making, the wide availability of medical technology, discontinuity across health care settings, fear of litigation, and health care policies that may facilitate curative treatments over palliation.¹⁰¹ Of note, tube feeding in advanced dementia is less common in other developed countries (eg, Canada),^{63 ,83 ,89} where different health systems and cultural attitudes toward death and dying may facilitate a greater willingness to forgo aggressive end-of-life interventions.

QUESTION: You use the word “culture” to determine some of the differences in the use of tube feeding, but I wonder where socioeconomics might play a role in decision making? Has that been looked at?

DR MITCHELL: Socioeconomic factors do appear to play a role in tube feeding practice at a facility level.⁵⁸ Nursing homes that care for a comparatively higher proportion of Medicaid beds (ie, that are resource-poor) have higher tube-feeding rates in advanced dementia. At the same time, facilities with more Medicaid beds also care for a relatively larger proportion of African American patients.⁵⁸ African American race is a strong independent risk for tube feeding in advanced dementia at both the individual and facility levels.^{13 ,58 ,61 - 62} Thus, there appears to be some interaction among economic resources, culture, and tube feeding within nursing homes. There are no available data to suggest that socioeconomic factors are determinants of feeding decisions in advanced dementia at the individual level.