Intended for healthcare professionals

Quality Improvement Reports

Using rapid-cycle quality improvement methodology to reduce feeding tubes in patients with advanced dementia: before and after study

BMJ 2004; 329 doi: (Published 26 August 2004) Cite this as: BMJ 2004;329:491
  1. Carol Monteleoni, coordinator of speech-language pathology (cmonteleoni{at},
  2. Elizabeth Clark, director of division of geriatric medicine1
  1. 1 Lenox Hill Hospital, 100 East 77th Street, New York, New York 10021, USA
  1. Correspondence to: C Monteleoni
  • Accepted 3 June 2004


Problem Despite lack of evidence that enteral feeding tubes benefit patients with dementia, and often contrary to the wishes of patient and family, patients with dementia who have difficulty swallowing or reduced food intake often receive feeding tubes when hospitalised for an acute illness.

Design We conducted a retrospective chart review of all patients receiving percutaneous endoscopic gastrostomy or jejunostomy tubes between March and September 2002. QI interventions including a palliative care consulting service and educational programmes were instituted. We conducted a second chart review for all patients receiving feeding tubes between March and September 2003.

Setting 652 bed urban acute care hospital.

Key measures for improvement We measured the number of feeding tubes placed in patients with dementia, the number of feeding tubes placed in patients with dementia capable of taking food by mouth, and the number of feeding tubes placed in patients with dementia with an advance directive stating the wish to forgo artificial nutrition and hydration.

Strategies for change Medical and allied health staff received educational programmes on end of life care and on feeding management of patients with dementia. A palliative care consulting team was established.

Effects of change After the interventions, the number of feeding tubes placed in all patients and in patients with dementia was greatly reduced.

Lessons learnt Multidisciplinary involvement, including participation by the administration, was essential to effect change in practice. The intensive focus on a particular issue and rapid change led to “culture shift” within the hospital community. The need to establish unified goals of care for each patient was highlighted.

Background A growing body of research over the past decade has questioned the utility of placing feeding tubes (percutaneous endoscopic gastrostomy (PEG) or jejunostomy) in patients with advanced dementia.1 Studies have found no evidence that feeding tubes in this population prevent aspiration,2 3 prolong life,46 improve overall function,7 or reduce pressure sores.8 Additionally, the quality of life of a patient with advanced dementia can be adversely affected when a feeding tube is inserted. The patient may require wrist restraints to prevent pulling on the tube1 3 or may develop cellulitis at the gastrostomy site, develop decubitus ulcers,1 be deprived of the social interaction and pleasure surrounding meals,9 10 and require placement in a nursing home. Unfortunately, many doctors are unfamiliar with this literature or face barriers—attitudinal, institutional, or imposed by the healthcare industry—to applying its findings to their practice.11 Thus feeding tubes are placed in patients who will not benefit from this intervention and whose quality of life in the terminal stage of their illness will be adversely affected. With the expected increase of elderly people with dementia,12 a great change in doctors' knowledge, attitudes, and practice is necessary to prevent even greater numbers of patients receiving this futile treatment.


At Lenox Hill Hospital, a 652 bed acute care facility in New York City, a multidisciplinary group of professionals concerned about improving medical care and quality of life for patients with dementia undertook a quality improvement project to address the issue of feeding tube placement in these patients. The project was conducted under the auspices of the New York Palliative Care Quality Improvement Collaborative (PCQuIC), a multisite initiative sponsored by the United Hospital Fund, RAND Corporation, and the Washington Home Center for Palliative Care Studies, with the aim of improving palliative care services in the 23 participating healthcare institutions.13 Faculty at the collaborative supplied training in the “plan-do-studyact” model for improvement14 15 and mentoring for this project over a 10 month period from September 2002 until June 2003. The project's core team included a geriatrician, a speech pathologist, a nurse, and an administrator. The hospital's vice president for medical affairs served as senior leader, acting as an advocate for the project.

Outline of the problem

Despite lack of evidence that feeding tube placement benefits patients with late stage dementia, despite medical risks and compromised quality of life, and sometimes contrary to the patient's advance directive, placement of feeding tubes in this population continues. We aimed to identify the extent and nature of this problem in our hospital and to intervene to change medical practice so that patients with dementia would receive medically appropriate treatment, consistent with their wishes. A retrospective chart review showed that within a six month period most feeding tubes were placed in patients with dementia. We found minimal documentation of patients' wishes regarding artificial nutrition and hydration, and few instances of formal assessment of the patients' capacity for decision making. Reversible causes for failure to eat were rarely addressed. The doctor's rationale for tube placement was documented infrequently. In cases of documented “failure to thrive,” no calorie count or other assessment had been made to determine actual food intake or to assess whether it was consistent with the nutritional needs of a patient with late stage dementia. A small but important number of patients received feeding tubes despite their explicit advance directives stating the wish to forgo artificial nutrition and hydration if they were irreversibly ill and unable to make their own decisions.

Key measures for improvement

We established three aims to measure the success of our interventions:

  • Reduce by 50% the number of feeding tubes placed in patients with dementia

  • Reduce to zero the number of feeding tubes placed in patients with dementia who are capable of taking food by mouth

  • Reduce to zero the number of feeding tubes placed in patients with dementia who have advance directives stating the wish to forgo artificial nutrition and hydration.

We hoped to achieve these aims within nine months of initiating quality improvement interventions.

Methods used to identify problems

Names of all inpatients receiving percutaneous endoscopic gastrostomy or jejunostomy in the hospital from March to September 2002 were obtained from the hospital's computer database. The core team conducted an in-depth review of the 58 available charts of the 71 patients who received tubes. Data collected included patients' characteristics, medical diagnoses, presence of dementia (either reported as a diagnosis or inferred from notes in the chart), advance directives or other evidence of the patient's wishes regarding artificial feeding, doctor's rationale for tube placement, and a detailed chronology of the patient's hospital course. We conducted a second chart review, after initiation of quality improvement interventions, on all patients who received feeding tubes for the six month period March-September 2003.

Strategy for change

Beginning in January 2003, the team formed a palliative care consulting team and instituted educational programmes.

The palliative care consulting team was led by the director of geriatrics. Working with medical residents on geriatrics rotation and the existing ethics and pain consult services, the team provided support and guidance for attending physicians (figure). A letter from the hospital's vice president for medical affairs requested attending physicians to call a member of the team whenever a feeding tube was being considered for any patient, demented or otherwise. Primary care physicians, gastroenterologists, nutritionists, nurse managers, and case managers attended in-service training sessions that explained the project and enlisted participation (boxes 1 and 2) The project was publicised at the quarterly medical staff conference and at grand rounds.


Role of palliative care consulting team

Four educational programmes were instituted. Firstly, a nationally recognised geriatrician with expertise in feeding issues in dementia gave a medical grand rounds presentation addressing the issue of feeding tube placement. The other three programmes involved the 24 medical residents in postgraduate year 2: they have a rotation of 12 sessions of pain management consultation; participate in presenting modules of the EPEC (Education for Physicians in End of Life Care) curriculum to the remainder of the medical house staff (attendance is mandatory); and spend one month on geriatrics rotation, which includes answering all palliative care consulting requests.

Box 1: Specialties targeted

  • Speech pathology

  • Nutrition

  • Gastroenterology

  • Nurse managers

  • Case managers

  • Primary care physicians

  • Hospitalists (physicians specialising in inpatient care)

  • Ethics committee

  • Administration

Box 2: Components of in-service training

  • Review of medical literature on burdens and benefits of tube feeding in patients with dementia

  • Discussion of advance directives

  • Linking advanced dementia to palliative care

  • Explanation of quality improvement project

  • Assignment of role in project

Effects of change

The table shows the preliminary results. The total number of tubes inserted and the number of tubes placed in patients with dementia both decreased significantly, exceeding our first aim, but the proportion of tubes placed in patients with an advance directive refusing artificial nutrition was not significantly reduced. This finding highlights a resistance by some doctors and surrogate decision makers to withholding treatment even when the patient's wishes to withhold treatment were clear. Retrospective data on ability to take food by mouth by patients in the 2002 group was not obtainable because few had received speech pathology or geriatrics evaluations to assess their ability to eat. Of the 2003 group of eight patients with dementia who received feeding tubes, two had been determined by the speech pathologist to be able to take some nutrition by mouth, four had been seen and determined to be unable to eat, and two had not received speech pathology assessments.

Feeding tube placement before and after quality improvement interventions

View this table:

Lessons learnt

Use of rapid cycle quality improvement methodology13 14 to change medical practice in this case was successful on many levels. The intensive educational initiative, coupled with implementation of a palliative care consulting service, brought the issue of tube placement in patients with dementia into focus for the hospital community. Medical grand rounds and in-service training sessions were well attended. From January to September 2003, the consult team intervened on 12 feeding tube cases. It is unlikely, however, that the sharp reduction in tube placement is attributable only to the organised quality improvement interventions. During the study period, physicians, nurses, nutritionists, speech pathologists, case managers, social workers, and other clinical staff seemed to talk more, formally and informally, about feeding tubes and the wider issue of medically futile treatment, suggesting a “cultural shift” within the hospital. Staff from many disciplines took pride in the rapid, dramatic results of the project, and this increased morale and belief in the possibility of positive change within the institution. Doctors' response to the project was generally positive, although a few practitioners resented “interference” with their care of patients. More commonly, doctors were grateful for the help the consulting team offered in dealing with difficult cases. Education of house staff, in both didactic sessions and clinical rotations, was a key component in the success of the project.

Key learning points

Despite literature questioning value of this intervention, physicians insert feeding tubes in patients with dementia

Interdisciplinary teamwork and focused educational effort can rapidly produce change in practice

Nutritionists, speech pathologists, gastroenterologists, case managers, social workers, nurses, and house staff can be targeted for education

The palliative care team works with attending physician to address prognosis, the patient's decision making capacity, the patient's wishes, and treatable causes of poor food intake

We wondered if the sharp decline in feeding tube placement in patients with dementia in our hospital had been accompanied by an increase in outpatient tube placements in outpatient settings or at nearby facilities. We have joined a consortium of hospitals in New York City to address the issue on a regional basis.

It became clear from chart review and palliative care consultations that absence of clear goals of care often led to unnecessarily long, complicated stays in hospital, contradictions among recommendations made by various clinicians on a case, and care inconsistent with patients' wishes. The palliative care consultant became a facilitator of communication among medical staff and between doctor and patient or surrogate, to establish goals of care. We have started a new quality improvement activity for early establishment and documentation of goals of care.


  • Contributors CM and EC participated in project design, data collection, data analysis, palliative care case consultations, in-service instruction, and other staff education in palliative care, and writing of this report. Julie Wityk participated in data collection and served as liaison with the PCQuIC faculty. Nancy Mooney, Tamar Kotz, and Barrie Guise participated in data collection. Walter Ettinger assisted with project design and served as the project's senior advocate within the institution. John Rapoport and Arthur Blank assisted with data analysis. CM is guarantor.

  • Funding Participation in the PCQuIC collaborative was supported by grants from Lenox Hill Hospital and the United Hospital Fund. These grants subsidised attendance at learning sessions and access to PCQuIC faculty for guidance throughout the course of the activity.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
View Abstract