Rethinking the Role of Tube Feeding in Patients with Advanced
Dementia
The New England Journal of Medicine
January 20, 2000 -- Vol. 342, No. 3
A byproduct of the aging of the population has been a dramatic
rise in the rate of Alzheimer's disease and other types of
dementia. A conservative estimate is that there are currently 4
million people in the United States with dementia. In the final
stage of dementia, patients are typically unable to walk or to
feed themselves, they are incontinent and aphasic, and they have
lost the capacity to have relationships with other people. Family
members or other surrogate decision makers must make difficult
and often painful decisions about limiting care. Should they
authorize surgery, hospitalization, intravenous medication? Is a
proposed treatment excessively burdensome to a patient who cannot
understand its purpose and who is nearing the end of life?
Of the many decisions that family members and physicians must
make about medical care for patients with dementia, none is more
heart-wrenching than the decision about artificial nutrition and
hydration. Despite an extensive bioethical literature arguing
that the use of feeding tubes is not mandatory and despite the
opinion by a majority of the Supreme Court justices that
artificial nutrition and hydration constitute a form of medical
care, family members repeatedly state that they cannot let a
relative "starve to death." They often feel they have
no choice but to authorize the placement of a feeding tube.
Furthermore, many physicians are either unfamiliar with or
unpersuaded by the bioethical arguments or the law. A study of
1446 physicians and nurses found that 34 percent of the
respondents who were medical attending physicians and 45 percent
of those who were surgical attending physicians believed that
even if all forms of life support, including mechanical
ventilation and dialysis, are stopped, nutrition and hydration
should always be continued. As a result, when patients with
advanced dementia start to have difficulty swallowing or lose
interest in eating, as happens frequently in such patients, the
decision is often made to insert a feeding tube. Percutaneous
endoscopic gastrostomy tubes are being used with increasing
frequency. In 1995, gastrostomy tubes were inserted in 121,000
elderly patients in the United States; approximately 30 percent
of these patients had dementia.
How can we help physicians and surrogate decision makers make
compassionate, morally sound, and technically reasonable
decisions about the treatment of persons with advanced dementia?
Ideally, such decisions should reflect the preferences and values
of the patient and should arise from a clear determination of the
overall goals of care. Unfortunately, the preferences of the
patient are seldom known, and the goals of care -- such as
prolonging life, promoting independence in activities of daily
living, or maximizing comfort -- cannot readily be translated
into practice. For the specific case of artificial nutrition and
hydration, perhaps the most emotionally troubling of all the
treatment-limitation issues, data accumulated over the past
decade can aid in the decision-making process. Surrogate decision
makers for demented patients who opt for feeding through a
gastrostomy tube usually do so because they hope to extend life
and prevent aspiration pneumonia, because they wish to prevent
suffering, or because their values, particularly their religious
beliefs, dictate that sustenance must never be withheld.
Recent information supports the conclusion that tube feeding
seldom achieves the intended medical aims and that rather than
prevent suffering, it can cause it. Moreover, just as many
secular bioethicists have argued that feeding tubes are not
required in patients with advanced dementia, in recent years
several religious ethicists have come to the same conclusion.
Given the weight of the scientific evidence that gastrostomy
tubes are not effective and given the lack of a compelling
ethical argument for their use, I suggest that physicians,
nursing homes, and hospitals adopt a policy of recommending that
gastrostomy tubes not be used in patients with advanced
dementia.
Do Feeding Tubes Work in Patients with Dementia?
A gastrostomy tube is placed in a patient to provide nutrition
and hydration. Although in theory tube feeding can provide
adequate nutrition in a patient with dementia, in reality this is
often not the case. Because of problems with diarrhea, clogging
of the tubes, and the tendency of patients with dementia to pull
out the tubes, nutritional status often does not improve with the
use of feeding tubes.
Aside from maintaining nutrition, one of the primary reasons
to use a feeding tube is to interrupt the cycle of eating,
aspiration, and then pneumonia that is so common in patients with
advanced dementia. It is now clear from multiple observational
studies that feeding tubes do not prevent aspiration in patients
with dementia. The continued risk of aspiration appears to result
from reflux of gastric contents and aspiration of saliva. I am
unaware of any randomized study that has compared the risk of
aspiration with and without tube feeding in patients with
advanced dementia. Hence, whether feeding tubes reduce the risk
of aspiration is unclear. Some studies have found that
jejunostomy tubes may be slightly less likely to result in
aspiration than gastrostomy tubes, but other studies have not
confirmed this finding. Despite the lack of evidence that tube
feeding reduces the risk of aspiration in persons with dementia
who have difficulty swallowing, the standard of care in many
institutions is to withhold oral intake and to insert a
gastrostomy tube if there is videofluoroscopic evidence of
aspiration.
The use of feeding tubes in persons with dementia is thought
to prolong life. In fact, however, it has been remarkably
difficult to demonstrate any difference in longevity between
patients with feeding tubes and those without tubes. A carefully
performed study of the outcomes of nursing home patients with
advanced dementia and eating disorders found that the patients
who were fed through a gastrostomy tube and those who continued
to be fed by hand had the same survival rates. Other studies have
confirmed that swallowing or eating problems in patients with
dementia are independently associated with mortality, regardless
of whether a gastrostomy tube is in place. These were
observational studies, however, not randomized trials. Despite
adjustment for coexisting conditions, the patients who were fed
through a gastrostomy tube may have been sicker than the patients
who were fed by hand, and the equivalent survival rates may
indicate that the patients with feeding tubes lived longer than
they otherwise would have. It is also possible that there is a
subgroup of persons with dementia and feeding problems whose
lives are prolonged by the use of feeding tubes -- for example,
persons with vascular dementia who have difficulty swallowing
because of a small brain-stem stroke.
There is a more plausible explanation for the failure to find
a survival advantage with the use of gastrostomy tubes -- or more
generally, the failure to find that such tubes are effective.
Difficulty with eating is a marker of severe dementia, which is a
uniformly fatal disorder. Since eating is typically among the
last activities of daily living to become impaired in persons
with dementia, difficulty with eating unfortunately signals that
the person has entered the final phase of the illness.
Do Feeding Tubes Promote the Comfort of Patients with
Advanced Dementia?
Families and physicians often believe that the use of a
feeding tube will promote the comfort of a person with advanced
dementia. Someone with inadequate nutrition to maintain his or
her weight is widely assumed to suffer from the adverse
consequences of dehydration or malnutrition. A gastrostomy tube
is therefore used as a means of preventing a protracted and
painful death. Is a person with advanced dementia who has
difficulty eating actually uncomfortable in the absence of
artificial nutrition and hydration?
It is difficult, if not impossible, to obtain data on the
subjective experience of patients with advanced dementia who stop
eating and drinking. We can, however, try to extrapolate from the
experience of patients dying of cancer or stroke who have
anorexia or profound dysphagia and who are sufficiently lucid to
describe their sensations. A growing literature from the hospice
movement indicates that such patients do not experience more than
transient hunger and that any thirst they experience can be
assuaged with the use of ice chips and mouth swabs. Moreover,
many elderly patients do not feel distress from dehydration
because they have an impaired thirst mechanism. Whereas in other
circumstances the inability to experience thirst is maladaptive,
in the terminal phase of Alzheimer's disease, it minimizes
discomfort. A few observers have suggested that maintaining
hydration promotes comfort by preventing delirium, but most
clinicians find that, on balance, hydration without nutrition
causes discomfort, because it prolongs the process of dying and
increases the production of urine and sputum.
Not only do gastrostomy tubes fail to prevent suffering, they
may actually cause it. The percutaneous insertion of a
gastrostomy tube, the most common method used, requires upper
endoscopy, an incision in the abdominal wall, and intravenous
sedation. Once the endoscope is in place, the permanent
gastrostomy tube is threaded into the stomach, anchored in place,
and pulled through the abdominal wall. Although it is a
relatively simple procedure, the long-term rate of complications
has been reported to range from 32 percent to 70 percent. The
problems, such as superficial skin infections and dislodgment or
clogging of the tube, are usually relatively minor.
When tube feeding is used as a permanent alternative to oral
feeding, patients are deprived of the enjoyment derived from
eating. Patients with dementia who are fed through a tube no
longer have the opportunity to taste food or experience the
social satisfaction associated with mealtimes. Feeding by hand is
an act of nurturing that cannot be accomplished by hanging a bag
of nutrients on a pole for delivery through a tube. These
problems might be overcome -- but only in some instances and to a
limited extent -- by supplementing tube feeding with the
provision of small amounts of food to persons with dementia who
have difficulty swallowing. The most serious potential adverse
consequence of tube feeding is the need to restrain the patient.
A patient with advanced dementia does not have the cognitive
capacity to understand why a tube is protruding from the
abdominal wall and often pulls it out. To prevent the patient
from removing the tube, the physician frequently orders the use
of restraints. In one study, 71 percent of patients with dementia
who had feeding tubes were restrained, regardless of the type of
tube used. Even with educational programs designed to reduce the
use of restraints in nursing homes, the presence of a
"treatment device" such as a gastrostomy tube was
associated with the use of restraints. The experience of being
tied down is distressing, even to persons with severe dementia,
and it often results in agitation, which in turn may lead to the
use of pharmacologic sedation. In summary, data collected over
the past decade suggest that gastrostomy tubes are not necessary
to prevent suffering and may actually cause suffering.
Is Withholding Artificial Nutrition Morally Wrong?
Some believe that failing to provide nutrition and hydration
for patients with dementia is morally wrong. Geriatricians and
ethicists have argued that artificial nutrition and hydration are
forms of medical therapy that can legitimately be withheld if
their risks, judged according to the patient's values,
outweigh their benefits. Patients or their surrogates who choose
tube feeding are most often influenced by the inestimable value
of life itself, and this value is frequently based on a set of
religious beliefs.
Although religious values may prompt some persons to favor any
intervention that offers a chance, however small, of prolonging
life, even these values are not absolute. The Roman Catholic
position on the use of artificial nutrition and hydration near
the end of life is that although "there should be a
presumption in favor of providing nutrition and hydration to all
patients, including patients who require medically assisted
nutrition and hydration," this approach is warranted only as
long as it "is of sufficient benefit to outweigh the burdens
involved to the patient." The Orthodox Jewish tradition,
also commonly assumed to advocate any intervention that might
prolong life, in fact rejects interventions that cause or prolong
suffering. Equally important, many Orthodox Jewish thinkers
regard the dying person in a special light and argue against
"impediments to dying" in the final year of life. For
surrogate decision makers who evaluate the decision on religious
grounds, the ethical question is to determine whether the use of
a feeding tube may have disadvantages and, if so, whether they
outweigh the possible benefits.
A New Standard of Care
Making decisions about artificial feeding in patients with
advanced dementia has been difficult because it has hinged on
assessing the quality of life. Advocates for withholding
gastrostomy tubes have based their argument principally on their
beliefs about patients' comfort, not on data about outcomes.
I have argued, in contrast, that recommendations about tube
feeding should be based on its effectiveness. Gastrostomy tubes
have not been shown to prolong life, ensure adequate nutrition,
or prevent aspiration, and there is neither a secular nor a
religious ethical imperative to use them. In addition, they are
not necessary to prevent suffering. Since there are few if any
benefits and there is considerable potential for harm, the
routine use of gastrostomy tubes in patients with severe dementia
is not warranted. Physicians, professional organizations,
hospitals, and nursing homes should recommend to patients and
their families that nutrition be provided orally, not through a
feeding tube, during the final stage of dementia. This approach
is distinct from the broader recommendation that patients with
advanced dementia receive exclusively palliative care. Decisions
about hospice care as opposed to curative care are typically
based on an assessment of the quality of life. Decisions about
hand feeding versus tube feeding can be made by weighing the pros
and cons of gastrostomy tubes.
There is just enough uncertainty associated with tube feeding,
just enough chance that a tube might, in some unanticipated
situations, prolong life or provide comfort in patients with
dementia, that family members should be able to request a feeding
tube if they believe it is truly what the patient would have
wanted. However, if family members are unable to make a decision
and if there are no extenuating circumstances, the physician
should assume that a person with advanced dementia would not want
a gastrostomy tube. An analogous default position forms the basis
of the do-not-resuscitate policy of some nursing homes. These
facilities advise that cardiopulmonary resuscitation not be
attempted, and in cases in which the resident or a surrogate
cannot decide, a do-not-resuscitate order is written. The
assumption that a patient would not want a gastrostomy tube
reflects the empirical data. A study of 421 randomly selected,
competent persons living in 49 nursing homes found that only one
third would favor a feeding tube if they were unable to eat
because of permanent brain damage. Moreover, 25 percent of the
respondents who initially favored tube feeding changed their
minds about its acceptability when they learned that they might
have to be physically restrained to facilitate use of the tube.
The proportion of respondents who favored feeding tubes might
have been even smaller if they had been informed about the lack
of efficacy of gastrostomy tubes in patients with dementia, just
as in another study many elderly patients who initially said they
would want cardiopulmonary resuscitation if they had a cardiac
arrest changed their minds when they learned about the poor
outcome of attempted resuscitation.
Obstacles to a New Standard of Care
There will undoubtedly be resistance to attempts to alter the
standard approach to difficulties with eating in patients with
dementia. There may also be legal and regulatory barriers.
However, in the one case dealing with artificial nutrition and
hydration that has reached the U.S. Supreme Court, that of Nancy
Cruzan, the majority of justices equated artificial nutrition and
hydration with medical therapy. At the same time, the Court
upheld the right of state legislatures to require that a
surrogate decision maker for a patient provide evidence that a
decision reflects the patient's previously stated
preferences. Two states, Missouri and New York, have adopted such
a policy for decisions about artificial nutrition and hydration.
They have statutes requiring "clear and convincing"
evidence that an incompetent patient would not have wanted tube
feeding in order for a surrogate decision maker to authorize the
withholding of tube feeding. The assumptions underlying such laws
are that tube feeding is a life-sustaining treatment and that the
state has an interest in protecting life.
Such laws are clearly applicable to patients like Cruzan, who
was in a persistent vegetative state, and who could be maintained
by a feeding tube indefinitely. The relevance of the laws to
patients with advanced dementia, a condition in which medical
intervention offers little chance of prolonging life, is highly
questionable. As Lo and Steinbrook have pointed out, "It
would be irrational to interpret the Cruzan ruling as requiring
that futile treatment be administered to incompetent patients
merely because they had not given clear and convincing evidence
of refusal." In subsequent Supreme Court cases, which may be
more germane to patients with advanced dementia (even though the
cases were prompted by a question about the constitutionality of
bans on assisted suicide), the majority of justices stressed that
dying patients have a right to palliative care. By implication,
states should remove existing barriers to palliative care for
patients near the end of life, including restrictions on the
right to withhold artificial nutrition and hydration.
Regulations governing nursing homes are another potential
barrier to the implementation of a new standard of care. Federal
regulations require nursing homes to ensure that their residents
have "acceptable parameters of nutritional status, such as
body weight and protein levels, unless the resident's
clinical condition demonstrates that this is not possible."
A specific example of a clinical condition in which maintenance
of nutrition may be impossible is advanced disease, according to
the regulations. Adherence to federal guidelines, which are
intended to promote high-quality care and to prevent abuse,
should focus on ensuring that nursing home residents with eating
problems are adequately assessed, perhaps by a trained therapist.
Nursing homes will need to supply evidence that the cause of the
feeding problem has been established and that appropriate efforts
are being made to hand-feed the resident. Given the data on the
lack of efficacy of tube feeding and the risk of adverse effects,
routine insertion of a gastrostomy tube as a strategy for
complying with government regulations makes no more sense than
the use of restraints for nursing home residents who fall.
There is a pervasive failure -- by both physicians and the
public -- to view advanced dementia as a terminal illness, and
there is a strong conviction that technology can be used to delay
death. The first step in changing these attitudes is for
physicians to acknowledge that feeding tubes are generally
ineffective in prolonging life, preventing aspiration, and even
providing adequate nourishment in patients with advanced dementia
and that they can have adverse consequences -- principally the
need for restraints. Although the use of feeding tubes is not
unequivocally futile in all cases, balancing the risks and
benefits leads to the conclusion that they are seldom warranted
for patients in the final stage of dementia. Once physicians are
aware of the data on tube feeding, they can move from a position
of discussing the issue in a nondirective manner, as is currently
advocated by some authors, to recommending that artificial
nutrition not be provided. There will always be exceptions and
unusual circumstances. However, patients with dementia who have
feeding difficulties and their families deserve the guidance of
physicians in this highly charged area. Physicians should take
the initiative in promoting hand feeding and allow state
legislatures and nursing home regulators to follow their
lead.
Muriel R. Gillick, M.D.
Hebrew Rehabilitation Center for Aged
Boston, MA 02131
Copyright © 2000 by the Massachusetts Medical Society.
All rights reserved.
|