Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports
Percutaneous Endoscopic Gastrostomy
(PEG) feeding in the enteral nutrition of
dysphagic stroke patients
A West Midlands Development and Evaluation Service Report
Authors: Carole Cummins, Tom Marshall & Amanda Burls
Department of Public Health & Epidemiology
University of Birmingham
Edgbaston
Birmingham
B15 2TT
Correspondence to: Carole Cummins
Institute of Child Health
Children’s Hospital, 4
th
Floor
Steelhouse Lane
Birmingham
B4 6NH
ISBN No. 0704421178
© Copyright, Development and Evaluation Service
Department Of Public Health And Epidemiology
University of Birmingham 2000
Percutaneous Endoscopic Gastrostomy
West Midlands DES reports December 1999
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports
QUESTION ADDRESSED BY THIS REVIEW:
Is percutaneous endoscopy feeding more effective than nasogastric tube feeding in stroke
patients requiring enteral feeding with regard to mortality, morbidity and health related quality of
life?
CONCLUSION:
PEG feeding of dysphagic stroke patients is associated with small increases in patient wellbeing
and small differences in resource use compared with NGT feeding. The impact of enteral
feeding of stroke patients on survival without severe disability is unknown. Trials in progress
may provide further information.
EXPIRY DATE: 2002
This report was completed December 1999. The searches were conducted in September
1998. The expiry date of this report is provisionally December 2002, but is dependent upon
the reporting of the Pegasus and FOOD trials which should provide new evidence on this
topic.
Percutaneous Endoscopic Gastrostomy
West Midlands DES reports December 1999
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports
West Midlands Development & Evaluation Service
The West Midlands Development and Evaluation Service (DES) produce rapid systematic
reviews about the effectiveness of healthcare interventions and technologies, in response to
requests from West Midlands Health Authorities or the HTA programme. Reviews usually
take 3-6 months and aim to give a timely and accurate analysis of the quality, strength and
direction of the available evidence, generating an economic analysis (where possible a cost-
utility analysis) of the intervention.
About Intertasc
West Midlands DES is a member of InterTASC which is a national collaboration with three
other units who do rapid reviews: the Trent Working Group on Acute Purchasing; the Wessex
Institute for Health Research and Development; York Centre for Reviews and Dissemination.
The aim of InterTASC is to share the work on reviewing the effectiveness and cost-
effectiveness of health care interventions in order to avoid unnecessary duplication and improve
the peer reviewing and quality control of reports.
Contribution of Authors
Carole Cummins undertook the collection and collation of evidence and wrote this review.
Amanda Burls provided duplicate data extraction and commented on the report. Tom
Marshall provided extra material on health economics.
Percutaneous Endoscopic Gastrostomy
West Midlands DES reports December 1999
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports
West Midlands Development and Evaluation Committee
Recommendation:
Supported
Anticipated expiry date
This report was completed in December 1999
The searches were completed in October 1998
The expiry date of this report is provisionally December 2002, but is dependent upon
the reporting of the Pegasus and FOOD trials which should provide new evidence on
this topic.
Percutaneous Endoscopic Gastrostomy
West Midlands DES reports December 1999
Contents
1 Introduction.................................................................................................................2
2 Background .................................................................................................................2
2.1 Incidence of dysphagia in stroke patients...................................................................2
2.2 Natural history, health related quality of life and functional status ................................3
2.3 Numbers of patients treated......................................................................................3
2.4 Outline of typical current alternative service...............................................................5
3 Questions addressed by this review ..............................................................................5
4 Methods......................................................................................................................5
4.1 Search strategy ........................................................................................................6
4.2 Inclusion criteria for the evidence..............................................................................6
4.2.1 Study design ....................................................................................................6
4.2.2 Study population..............................................................................................7
4.2.3 Intervention......................................................................................................7
4.2.4 Outcome measures...........................................................................................7
4.3 Criteria for the evaluation of the evidence..................................................................7
4.4 Data extraction.........................................................................................................7
4.5 Economic analysis ....................................................................................................7
5 Quality, direction and strength of the evidence...............................................................7
5.1 Randomised control trials .........................................................................................8
5.1.1 Quality, design and execution............................................................................8
5.1.2 Patients............................................................................................................8
5.1.3 Outcomes........................................................................................................9
5.2 Evidence from case series.......................................................................................13
5.3 Trials in progress....................................................................................................15
5.3.1 The FOOD multicentre RCT..........................................................................15
5.3.2 The PEGASUS multicentre RCT....................................................................15
5.4 Summary: quality and direction of the evidence.......................................................15
6 Economic analysis ......................................................................................................16
6.1 Economic literature.................................................................................................16
6.2 Economic analysis ..................................................................................................16
6.2.1 Health consequences of NGT and PEG feeding ..............................................16
6.2.2 Health consequences: summary.......................................................................16
6.2.3 Resource consequences of NGT and PEG feeding..........................................17
6.2.4 Identifying resource consequences..................................................................17
6.2.5 Measuring resource consequences..................................................................18
6.2.6 Prices of PEG insertion...................................................................................19
6.2.7 Resource consequences - summary................................................................19
6.3 Summary of economic analysis ...............................................................................19
6.4 Implications for other parties...................................................................................19
7 Conclusions ...............................................................................................................19
7.1 Areas of uncertainty...............................................................................................20
7.2 Time limit for this report..........................................................................................20
8 References.................................................................................................................21
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 1
Summary
Technology
Percutaneous endoscopic gastostomy (PEG) is a method of enteral feeding through a
gastrostomy tube placed endoscopically under local anaesthesia and sedation. It is an
alternative to nasogastric tube (NGT) feeding.
Condition
The main use of PEG is for the enteral feeding of dysphagic stroke patients.
Evidence of effectiveness
Three small RCTs have compared PEG and NGT feeding. The evidence indicates that
PEG feeding is technically superior to NGT feeding and is more acceptable to patients.
One trial found that early PEG feeding was associated with a reduction in mortality, but
this result needs to be confirmed.
Economic analysis
Compared to NGT feeding, PEG feeding is associated health gains including improved
patient wellbeing and possibly earlier discharge. There are small differences in the
resource implications of PEG compared to NGT, principally that PEG requires the use
of more senior staff.
Conclusion
PEG feeding of dysphagic stroke patients is associated with small increases in patient
wellbeing and small differences in resource use compared with NGT feeding. The
impact of enteral feeding of stroke patients on survival without severe disability is
unknown. Trials in progress may provide further information.
Percutaneous Endoscopic Gastrostomy
2 West Midlands DES reports December
1999
1 Introduction
Percutaneous endoscopic gastostomy (PEG) is a method of enteral feeding through a
gastrostomy tube placed endoscopically. The procedure can be performed with
benzodiazapine sedation and local anaesthetic in patients who are high anaesthetic risks. The
method has low morbidity and a procedure related mortality of 1 to 2%.
1
2
. Percutaneous
gastrostomy tubes can also be inserted as a radiological procedure
3 4
This method of enteral feeding is increasingly used in long term feeding of dysphagic patients
(patients with swallowing difficulties), the majority of whom have had strokes, with a minority
suffering from other neurological indications, principally motor neurone disease, or with a need
for nutritional support for other reasons including malignancy
5 6 7
.
The selection of patients for PEG and the timing of PEG insertion are important. Some patients
are likely to die quickly as a result of their strokes whether or not enteral feeding is initiated and
will therefore derive little benefit from this invasive procedure. Other patients may survive with
maximal dependency and minimal quality of life
1
. The impact of enteral feeding on survival and
quality of life is currently unknown.
Ethical guidelines for PEG placement
8
have suggested that PEG is not ethically justified where
the patient will derive no physiological benefit (as in permanent cachexic states) and where the
patient will not experience any improvement of quality of life (as in permanent vegetative
states), with PEG only enabling the maintenance of physiological function. Where dysphagia
exists without complications, and the patient will unequivocally benefit, then PEG feeding is
ethically justified. Where PEG is of uncertain clinical benefit, because of deficits in quality of
life or progressive underlying disease, the authors state that the decision to have a PEG should
be made by patients or substitute carers after receiving full information from the clinician. Most
stroke patients considered for PEG will fall into the latter category.
2 Background
2.1 Incidence of dysphagia in stroke patients
Dysphagia, that is difficulty in swallowing, is a common, serious consequence of stroke and
results from damage to the upper motor neurone of the lower cranial nerves. In an unselected
hospitalized group of stroke patients, 45% had difficulty in swallowing when admitted to
hospital, 7% had dysphagia for nine or more days, but only 3% were dysphagic after 40 days,
with a 6 week mortality rate of initially dysphagic patients of 46%
9
. In a series of hemispheric
stroke patients, although nearly 30% of stroke patients who were conscious within 48 hours of
their stroke initially had difficulty swallowing, by 1 month, only 2 percent still had dysphagia
10
.
It has been estimated, using data from the above studies, that a typical district general hospital
serving a population of 280,000 will have approximately 8-10 patients out of 450 new stroke
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 3
patients per year with dysphagia persisting 14 days
1
. Approximately 30 patients, however, will
have dysphagia persisting more than a week and will be candidates for early nutritional support.
2.2 Natural history, health related quality of life and functional status
The health related quality of life of dysphagic stroke survivors will vary with functional status
following stroke. Dysphagia is associated with severe stroke. Although most dysphagic
patients either recover their swallowing ability or die
9 10
in the first few weeks following the
stroke, some patients regain the ability to eat at a later stage. Little information is available
about the functional status of dysphagic stroke patients in the longer term.
Although dysphagia has a poor prognosis and is associated with early death following stroke
9
,
medium term survivors with dysphagia are probably not at greater risk of death than patients
with otherwise comparable functional status without dysphagia. Dysphagia, however, is
associated with poorer functional outcome
10
.
Patients with dysphagia are at a high risk of aspirating their food and consequently of aspiration
pneumonia. Enteral feeding does not remove this risk
11
but does contribute to an improved
nutritional status. Malnutrition has been associated with poor outcomes, including higher rates
of urinary and respiratory infections and a higher prevalence of bedsores, but it is not clear
whether early enteral nutrition of any type prevents increasing malnutrition following stroke
12
.
A small observational study has found that pneumonia prior to PEG, oesophagitis on
endoscopy and age of 70 or over were risk factors for pneumonia after PEG
13
.
Even in community series, the reported proportion of stroke patients achieving independence
(variable definitions) at one year varies from 60 to 69%, with best estimates of 45% achieving
independence at 6 months and 60% by one year. Around 20% of survivors are in institutions
one year after the stoke
14
. One community based study found that at 6 months 4% were very
severely disabled, 5% were severely disabled, 12% were moderately disabled, 32% were
mildly disabled and 47% were independent in activities of daily life on the Barthel Index of
Daily Living. By definition, survivors with dysphagia are at least mildly disabled (Barthel Index
15-18). That is, no patients are able to perform all self-care activities.
Raha and Woodhouse
6
comment “the target population is, by definition, markedly disabled,
usually with severe underlying disease”. They believed that, in these circumstances, PEG
improved their patients’ quality of life in their final weeks. The reasons put forward for this
were that PEG feeding allows relief from hunger to be provided where the patient was able to
communicate that they were hungry and patients preferred PEG over NGT on grounds of
comfort and cosmetic acceptability
6 15
.
2.3 Numbers of patients treated
Dysphagia following stroke is likely to be the indication for at least half of all PEG insertions in
a district general hospital
15
. Hospital Episode Statistics for the West Midlands from April 1996
to March 1997 recorded 411 gastrostomies, of which 57% were specified as endoscopic
(Table 1). Most gastrostomies where the approach was unspecified would have been PEGs,
rather than open operations. Stroke patients accounted for 270 gastrostomies, 66% of the
total. Some of the ninety patients described only as dysphagic would also have suffered
strokes.
Percutaneous Endoscopic Gastrostomy
4 West Midlands DES reports December
1999
Calculations based on the incidence of persistent dysphagia given above (see 3.1) suggest that,
in 1995, the West Midlands (population 5,315,000) would have had around 600 stroke
patients with dysphagia persisting for at least a week and between 150 and 190 new stroke
patients with dysphagia persisting for 14 days
16
. The number of gastrostomy procedures
carried out in stoke patients suggests that nutritional support has often been provided at an
early stage. This is supported by audit figures from West Midlands hospitals (personal
communication).
A high proportion of these patients were seriously ill, as 29% of the total and 35% of the
stroke patients died during the admission in which the gastrostomy was carried out.
Gastrostomies were usually carried out as part of an inpatient stay, although there were a few
day cases (19 (5%) of the total, 7 (3%) of stroke patients). Inpatient length of stay by
diagnosis is given in Table 2. Some gastrostomies were carried out during the course of very
long admissions, and the median length of admission for stroke inpatients who had a
gastrostomy was 46 days.
Table 1: Gastrostomies, type and diagnosis, West Midlands Hospital Episode Statistics
1996-1997
Gastrostomy type Total
permanent temporary unspecified permanent,
endoscopic
temporary,
endoscopic
Other,
endoscopic
n %
malignant neoplasms 4
2
3 9 2.2
subarachnoid
haemorrhage
1 5
2
1 9 2.2
stroke, total 16 76 1 46 129 2 270 65.7
stroke, haemorrhagic 2 2
2
6 12 2.9
stroke, infarction 6 7
7
14 34 8.3
stroke, unspecified 8 67 1 37 109 2 224 54.5
other cerebrovascular
disease
1 3
1
4 12 2.9
motor neurone disease 1 11
1
6 19 4.6
dysphagia,
unspecified
1 57 1
5
26 90 21.9
other
1
4 2 0.5
Total 20 156 2 58 173 2 411
4.9% 38.0% 0.5% 14.1% 42.1% 0.5%
78% of gastrostomies were carried out in patients aged 65 and over. 84% of stoke patients
were aged 65 and over, and the median age of stroke gastrostomy patients was 76 (Table 3).
Hospital episode statistics do not record enteral feeding via nasogastric tubes, so the number of
patients receiving nasogastric tube feeding is unknown.
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 5
Table 2: Gastrostomies, in-patient length of admission, West Midlands Hospital Episode
Statistics 1996-1997
Mean Std.
Deviation
Median N Minimum Maximum
malignant neoplasms 43.9 29.3 39 9 5 107
Subarachnoid haemorrhage 62.8 40.2 61 8 0 117
stroke, total 56.9 47.4 46 261 0 382
motor neurone disease 45.7 22.2 42.5 18 5 99
dysphagia, unspecified 39.3 50.7 23.5 76 0 282
other 52.7 42.2 65.5 12 0 106
Total 52.5 46.9 43 384 0 382
Table 3: Gastrostomies, patients’ age and diagnosis, West Midlands Hospital Episode
Statistics 1996-1997
Mean Std.
Deviation
Median N Minimum Maximum
Malignant neoplasms 70.4 11.5 71
9
52 88
Subarachnoid haemorrhage 61.9 23.9 67
9
2 83
stroke, total
74.8 10.7 76 270 39 95
motor neurone disease 76.8 7.2 75 19 63 89
dysphagia, unspecified 64.6 21.0 71 90 4 92
other 68.6 20.2 69.5 14 14 93
Total
72.1 14.9 75 411 2 95
2.4 Outline of typical current alternative service
Nasogastric tube feeding has traditionally been used in stroke patients with a prospect of
extended survival (>4 weeks) who are unable to maintain adequate oral nutrition and fluid
balance. PEG feeding is increasingly used in this situation, and can be carried out in a home or
nursing home setting by carers and staff.
Nasogastric tubes may become dislodged or be
removed by patients, so that replacement tubes are required. An audit of an acute geriatric
assessment unit found that 139 out of 2332 patients (6%) required nasogastric feeding at some
stage, and that an average of 12 tubes per intubated patient were used
17
.
Although home nutritional support is becoming more common
18
, most GPs will not have a
patient on home enteral tube feeding using either method. The support available to the patient
on discharge on hospital will vary. GPs can prescribe feeds, but not other items. Local
financial and organisational arrangements are therefore critical to successful use of enteral
feeding in community settings.
3 Questions addressed by this review
Is percutaneous endoscopy feeding more effective than nasogastric tube feeding in stroke
patients requiring enteral feeding with regard to mortality, morbidity and health related quality of
life?
4 Methods
Percutaneous Endoscopic Gastrostomy
6 West Midlands DES reports December
1999
4.1 Search strategy
Reviews and primary studies were identified in the following databases:
Medline, Science Citation Index, Embase, DARE, Cochrane Database of Systematic Reviews,
Cochrane Controlled Trials Register, ISI Conference Proceedings and Transcripts. Searches
were conducted in July 1998.
The following searches (adapted for each platform) were used:
1. (percutaneous adjacent endoscopic adjacent gastrostomy as text) or
(peg as text) or (gastrostomy as MESH term or text) or enteral nutrition as MESH term or
text) or (surgery endoscopic as MESH term or endoscopic surgery as text) or (intubation,
gastrointestinal as MESH term or gastrointestinal intubation as text);
and
(cerebral hemorrhage as MESH term or text) or (cerebrovascular disorders as MESH term
or text) or (cerebral infarction as MESH term or text)
2. (cerebrovascular disorders as MESH term or stroke as text) or (cerebr$ as text));
and
(feeding methods as MESH term or feeding as text) or (intubation, gastrointestinal as
MESH term or enteral nutrition as MESH term or nasogastric as text) or (gastrostomy as
MESH term or text) or (endoscopy, esophagoscopy, gastroscopy as MESH terms or
endoscop$ as text).
The searches were inclusive, rather than restrictive, and reviews and primary studies with
relevant subject matter were identified by inspection of titles and abstracts, obtaining papers
where necessary.
Searches of the NEED and HTA databases and the National Research Register were made
using each of the following text words, stroke, feeding, gastrostomy, enteral, endoscop$ and
PEG. The results were scanned for relevant studies.
Other sources included hand search of Drugs and Therapeutic Bulletin (1996 and 1997),
follow-up of citations from reference lists and personal contacts.
4.2 Inclusion criteria for the evidence
4.2.1 Study design
A search was made for randomised control trials to compare the outcomes of PEG versus
NGT.
Additionally a search was made for case studies of PEG to provide further information on
procedure related mortality rates.
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 7
4.2.2 Study population
The study population was preferably stroke patients with persistent dysphagia requiring enteral
feeding, but studies which included patients with dysphagia resulting from conditions other than
stroke were included.
4.2.3 Intervention
PEG feeding compared with nasogastric tube feeding.
4.2.4 Outcome measures
It has been suggested that consideration of whether PEG should be used in preference to NGT
depends upon its relative efficiency in terms of achieving adequate intake, nutritional indices,
safety, patient tolerability, flexibility and ease of use, effects on other rehabilitation activities,
duration of hospitalisation and costs
1
. Other possible outcomes include health related quality of
life and mortality.
The outcomes considered here were: mortality and survival, procedure related mortality and
complications, patient health related quality of life, patient and carer preferences, volume of
food delivered.
RCTs that measured any of these outcomes were to be included.
4.3 Criteria for the evaluation of the evidence
Trials were to be evaluated in accordance with suggested guidelines
19
and important features of
trial design, including blinding, randomisation, concealment of randomisation and reporting of
drop-outs, crossovers and losses to follow-up
20
, were recorded.
Caution was exercised where there were small effect sizes or conclusions based on small
numbers of cases or sub group analyses, as such results generally need to be confirmed in
further trials.
4.4 Data extraction
Data extraction of RCTs was carried out by CC and AB and differences were discussed and
reconciled. Data extraction of case series was carried out by CC alone.
4.5 Economic analysis
A search was made for economic analyses of different methods of enteral feeding in stroke
patients (see 5.1).
An assessment of the cost effectiveness of PEG compared to NGT feeding was carried out
based on the evidence provided by the review.
5 Quality, direction and strength of the evidence
Percutaneous Endoscopic Gastrostomy
8 West Midlands DES reports December
1999
5.1 Randomised control trials
There have been three randomised trials of PEG versus nasogastric feeding
21,22,23
(Table 4).
One trial (Baeten) is described as "not yet completed".
5.1.1 Quality, design and execution
Table 4 describes relevant aspects of trial design.
All of the trials were small.
In only one trial (Park)
21
was it clear what the primary hypothesis of the trial had been, and
what the trial had been powered to detect.
The nature of the intervention (different methods of enteral feeding) made blinding of clinicians,
other staff and patients and the blind assessment of outcomes impossible.
Randomisation was via sealed or “closed” envelopes.
Reporting of results was by intention to treat. In one trial (Park)
21
, however, patient crossovers
meant that there was only sparse and censored information on NGT feeding. Dropouts,
crossovers and losses to follow-up were adequately reported.
In one study (Baeten)
23
, it was unclear whether standard deviations or reference ranges are
reported, so only mean values are quoted here. The duration of this trial was not stated, and
survival data were not presented, making interpretation of the death data in the trial impossible.
5.1.2 Patients
Patients in the Park trial
21
had dysphagia of at least four weeks duration resulting from
neurological conditions, were likely to survive for at least the trial duration and were able to
communicate (Table 5).
Patients in the Baeten trial
22
had neurological, surgical or ENT indications for PEG and no
preference for PEG or NGT.
Patients in the Norton trial
23
had dysphagia of at least eight days duration 14 days after an
acute stroke and had been unconscious on hospital admission and were described as being in a
stable condition. They were likely to have been sicker than the patients in the other two trials
and at higher risk of death.
It was considered that the patient populations in the trials were so different that the trial results
could not be pooled to produce summary estimates of the results.
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 9
5.1.3 Outcomes
5.1.3.1 Treatment success
In all three trials, treatment failure occurred more often with NGT than with PEG feeding
(Table 6).
Patients were more likely to have the entire amount of their prescribed foods with PEG rather
than NGT feeding
21,23
. PEG feeding was also associated with an increase in nutritional
indices
21,23
(Table 6)
.
Table 4: Randomised control trials of NGT and PEG feeding, design, quality and
execution
Park RHR et al 1992
21
Baeten C & Hoefnagels
J 1992
23
Norton B et al 1996
23
Power calculation for
study reported
Yes No No
Study was powered to: Detect difference in
treatment success of
40% with NGT and 90%
with PEG, with power of
.9 and statistical
significance at p<.05
Proportion of eligible
patients randomised
Not given, some
patients opted for PEG
outside the study.
Of 200 eligible patients,
90 (45%) had no
preference for NGT or
PEG and no
contraindication.
First 30 eligible patients
recruited.
Method of
randomisation
Numbered sealed
envelopes.
Sealed envelopes,
stratification by
indication (neurologic,
ENT, surgical)
Closed envelopes.
Blinding None None None
Reporting of drop-outs Yes. 1 patient in each
arm died after
randomisation but
before receiving
intervention. Treatment
failure in 18 (95%) NGT
group and 0 in PEG
group.
Yes. Failure to introduce
NGT 4 patients, later
failure 8 patients, total
failure 26%. Failure to
introduce PEG 3 patients
(7%)
Yes. 3 NGT treatment
failures.
Reporting of cross-
overs
18 (95%) of NGT group
successfully switched to
PEG.
6 NGT patients had a
PEG on treatment failure
None reported.
Reporting of losses to
follow-up
Yes Yes Follow-up lab data not
available for 1 PEG and 4
NGT patients (deaths
and drop-outs).
Percutaneous Endoscopic Gastrostomy
10 West Midlands DES reports December
1999
5.1.3.2 Mortality
23
found a significant difference in mortality (Table 6) at 6 weeks in stroke
patients with persistent dysphagia at 14 days post stroke who had PEG as opposed to NGT
feeding, with a relative risk for PEG of 0.29 (p=.02, Fisher’s exact test). The 6 week mortality
of dysphagic stroke patients, however, is high
9
, and the results cannot be applied to patients
who remain dysphagic, do not die, and are therefore candidates for PEG feeding in a later
stage of their disease. The patients had a mean Barthel Activites of Daily Living Index of <3
(maximum 100) at recruitment, indicating the patients had poor functional status and a poor
prognosis. This is a small trial (n=30), and this finding needs confirmation in further studies
The Baeten trial
22
found a greater number of deaths in PEG patients than in NGT patients, but
could offer no explanation for this finding. It was impossible to interpret this finding, as no
indication was given of length of follow-up. The number of stroke patients included in the trial
was not given, but must have been less than the 42 (47%) who had neurological conditions.
The PEG procedure appeared to have contributed to one surgical patient’s death: the needle
for the introduction of the PEG had perforated the liver with subsequent intraperitoneal
bleeding. The patient died of sepsis following anastomatic leakage follow rectal cancer
resection.
In the Park trial
21
, the only two deaths occurred post randomisation but prior to tube insertion.
Patients in this study had neurological dysphagia of four or more weeks duration, were
considered likely to survive at least for the duration of the trial. The trial therefore offers no
evidence on the impact of PEG feeding compared to NGT feeding on mortality in patients with
longstanding dysphagia.
5.1.3.3 Treatment acceptability
Two trials (Park, Baeten)
21 22
found that PEG was more acceptable to patients than NGT
feeding (Table 6).
In one trial (Baeten), PEG was preferred by nursing staff and reduced patient fixation was
required to prevent removal of tubes by patients
22
.
5.1.3.4 Complications
Apart from the death in the Baeten
22
trial described above, which may have been related to the
PEG procedure, there were only minor complications (Table 6). Two cases of aspiration
pneumonia in the Park trial
21
were probably not related to the PEG feeding method.
5.1.3.5 Other outcomes
Two trials reported a small increase in weight with PEG feeding
21 23
and one reported an
increase in serum albumin
23
in PEG patients. In two trials, more PEG than NGT patients were
discharged
22 23
(Table 6). In one trial, PEG resulted in earlier discharge, probably because
PEG was more acceptable to nursing homes than NGT feeding
23
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 11
Table 5: Randomised trials of NGT and PEG feeding, patients
Park RHR et al 1992
21
Baeten C & Hoefnagels
J 1992
22
Norton B et al 1996
23
Entry criteria Dysphagia of duration
>=4 weeks, neurological
diseases, likely to
survive >=1 month, able
to communicate, normal
gastrointestinal tract
Neurological, ENT and
surgical indications, no
patient preference for
PEG or NGT
Persistent dysphagia
(duration >=8 days) 14
days after acute stroke.
All unconscious on
admission, in stable
condition, enteral
feeding appropriate
Dysphagia definition Neurological dysphagia Not given Absence of normal gag
reflex/inability to
swallow 50ml water
easily without choking
Exclusion criteria dementia, mechanical
lesions, intra-abdominal
inflammation, relevant
adverse history, major
systemic disease
NGT or PEG
contraindicated. Patient
preference
Gastrointestinal disease
precluding siting of
gastrostomy tube, unfit
for upper GI endoscopy
or intravenous sedation.
Setting 3 teaching hospitals Academic hospital One university hospital
and one district general
hospital
Number in trial 40 (18 cerebrovascular
disease)
20 NGT, 20 PEG
90 (42 neurologic
indication, 39 ENT
indication, 9 surgical
indication)
NGT 46, PEG 44
Neurological indication:
NGT 23, PEG 19
30 recruited over 1 year
NGT 14, PEG 16
Patient characteristics Mean age; NGT65 (se
2.9), PEG 56 (se 4.8)
NGT 10F/10M PEG
12M/8F
Mean age
72 (neurologic indication
79), 62% male (45%,
neurologic indication)
NGT: 4M, 10F, mean age
79
PEG: 7M, 9F, mean age
76
Mean Barthel Index at
recruitment <3
Trial duration 28 days Not given 6 weeks
Percutaneous Endoscopic Gastrostomy
12 West Midlands DES reports December
1999
Table 6: Randomised control trials of PEG and NGT, outcomes
Park RHR et al 1992
21
(NGT 20, PEG 20)
Baeten C & Hoefnagels J 1992
22
(NGT 46, PEG 44)
Norton B et al 1996
23
(NGT 14, PEG 16)
Treatment
failure
NGT 18 (95%)*
PEG 0
NGT 26% (4 could not insert,
problems:8/42, 4/21 neurological
indications)*
PEG 7% (3 could not insert)
NGT 3 (21%) (unable to
resite 2, recurrent removal
1)
PEG 0
N of tube
insertions
12 patients displaced
tube 3 times, 2 patients
displaced tube twice
NGT mean 2.7 (underestimate)
PEG 1
NGT mean 6 (range 1 to 10)
PEG 1
Insertion time NGT 8.4 minutes
PEG 11.4 minutes
Mortality NGT 5/46 (4/23 neurological
indication*)
PEG 13/44 (11/19 neurological
indication) (1 PEG related death)
NGT 8/14*
PEG 2/16
Duration of
feeding
NGT 5.2 (se 1.5) days*
PEG 28 (se 0)
NGT 16.4 days
PEG 21.6 days (in hospital)
Feed intake NGT had 55% (se 4) of
prescribed food*
PEG had 93% (se 2) of
prescribed food
NGT 10 (71%) missed
feeds waiting for resiting
of tube, mean loss 22%
(95% CI 6-37%)
PEG no omitted feeds
Weight gain After 1
st
week:
NGT 0.6kg (se 0.1)*
PEG 1.4kg (se 0.5)
NGT 1/8 gained weight,
mean change -2.6kg*
PEG 10/13 gained weight,
mean change +2.2kg
Serum albumin
concentration
NGT mean change (N=10) -
9.5g/l*
PEG mean change (N=15)
+2.7g/l
Patient
preference
PEG acceptability
excellent (16), very good
(21), fair (1)
(includes crossovers)
PEG preferred
NGT mean 2.3 (n=21) (1=very
good)
PEG mean 1.78 (n=22)
Nurse
preference
PEG preferred NGT mean 2.6 out
of 5 (1=very convenient)
PEG mean 2.0
Patient fixation NGT 10 (22%)
PEG 3 (7%)
Discharge
rates
NGT 7 (7 neurological)
PEG 11 (4 neurological)
At 6 weeks:
NGT 0*
PEG 6 (38%)
Complications NGT none (but short
duration of treatment)
PEG 2 aspiration
pneumonia, 1 minor
infection
NGT clotting 15%, aspiration
7%, swallowing problem 17%,
nasal decubitus 6%
PEG clotting 16%, aspiration
3%, mild inflammation 27%,
abdominal pain (<=3 days) 11%,
abdominal bleeding 2%
PEG 1 peristomal infection
Natural history
of dysphagia
3 stroke patients
swallowing improved
and PEG was removed
3 patients regained normal
swallowing and had PEG
tube removed.
* NGT vs PEG, P <.05
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 13
5.2 Evidence from case series
Case studies were sought to provide information on any procedure related mortality associated
with PEG insertion. A comprehensive search for case studies was not attempted, as it was
considered that sufficient case series had been obtained to identify the range of procedure
related mortality rates that were likely to be found in current practice. Studies which included
patients who had open as well as endocopic gastrostomies, for example that of Ciocon
11
, and
some smaller studies have been excluded.
Table 7: PEG case series, patients
Study Design N Subjects Days between
stroke and PEG
insertion
Criteria for PEG
insertion
Mamel
24
review,
including local
case series
1.338
(327
local)
Not specified. Local
series mostly
neurological
indications
not specified
Raha et al
6
case series 161 neurological
dysphagia 88%
(stroke 81%),
nutritional support
12%
mean 44 (6-200)
Wanklyn et al
25
Retrospective 41 all stroke patients
who had PG
median 26 days
(12-131)
Panos et al
15
Prospective 76 76% neurological
indications, 51%
stroke
Impaired
swallowing and
nutritional need
for enteral
feeding expected
to exceed 2
weeks.
Inpatients:
intolerant of
NGT.
Outpatients:
opted for PEG
Hull et al
2
Prospective 49 referred dysphagic
patients (33%
stroke)
referred patients
Larson et al
26
Retrospective 314 75% neurological
indications
referred for PEG
Kaw et al
7
Retrospective 46 nursing home
residents, 24%
stroke (2 thirds
coma), 52%
dementia
Finucane et al
17
Retrospective 28 dysphagic patients
(26 stroke)
mean 63 (6-210)
Percutaneous Endoscopic Gastrostomy
14 West Midlands DES reports December
1999
Table 8 PEG case series, outcomes
Study Duration of
follow-up
(range)
Results Comments
Mamel
24
not specified Complications 14%, 30 day mortality
9-15%
Raha et al
6
mean 152 (11-
106)
30 day mortality 20% (80%
nutritional support), 90 day
mortality 39%complications 12%,
mostly minor, 1 PEG related death
(<1%), 16 patients had 18 tube
replacements
19 tubes removed
after mean of 199
days (45-365) as
swallowing
regained. PEG
preferred by
patients
Wanklyn et al
24
median 53 days
(2-528)
Complications 30% (11 cases) 5
chest infections 3 local infections, 2
tubes removed, 1 perforation. 3
deaths from complications (2%).
6 survivors assessed : 1 good
functional outcome, mean Barthel
index 7.
8 patients recovered
safe swallowing at
median of 30 days
post PG.
Panos et al
15
median 93 (3-
785) days
Nursing time same for PEG (8
patients)as for NGT (12 patients)
Weight, BMI arm circumference
improved, PEG related mortality 4%
(1 PEG related death, 1 possibly PEG
related, 1 pneumonia death (PEG
5/12 duration) PEG related, 1 major
complication, 25 minor
complications.
Swallowing
recovered in 16%,
26% mortality at
one month
Hull et al
2
mean 175 (30-
560)
1 death from peritonitis 11/12 after PEG
(2%),
Long term complications 22% 51%
patients had no complication 47%
complications required hospital visit
8% 30 day mortality
Larson et al
26
at least 45
patients >1 year
Placement 95%, mortality 1%, major
complications 3%.minor
complications 13%, 14% regained
ability to eat
Kaw et al
7
mean 321 (2-520) No improvements in functional
status, no improvements in mean
serum albumin, complications 35%
(tube obstruction 30%, tube
migration 17%), aspiration
pneumonia in 20%, survival 40% at
18 months
Finucane et al
17
median 98 in
survivors, mean
interval PEG and
death 92 (6-200)
Successful in all patients, 1 PEG
related death (4%), 30 day mortality,
7%.
mean age 82 (range
66-99)
The case series found
2 6 7 15 17 24 25 26
confirm that there is some procedure related mortality
attached to PEG, ranging from <1% to 4% in groups of patients that experience high mortality
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 15
rates from underlying condition. The 30 day mortality rates reported in these series vary from
7 to 26% and reflect the heterogeneous case mix of the PEG patients. The proportions of
patients regaining swallowing function also reflects case mix and, where quoted, ranged from
12% to 20%.
5.3 Trials in progress
Two relevant randomised control trials are in progress and listed in the National Research
Register, Issue 1 (http://www.doh.gov.uk/research/nrr.htm).
5.3.1 The FOOD multicentre RCT
27
Patients: patients admitted with a stroke (excluding subarachnoid haemorrhage) within seven
days of onset in whom the clinician is substantially uncertain about the best feeding policy.
Primary hypotheses: To assess whether nutritional supplementation increases the proportion
surviving without disability.
To evaluate whether, in patients unable to take an adequate diet orally, the early initiation of
tube feeding (NG or PEG) increases the proportion surviving without severe disability.
To assess whether a PEG tube instead of a NG tube, is associated with improved outcomes.
5.3.2 The PEGASUS multicentre RCT
Patients: inpatients with significant dysphagia unable to take adequate oral diet five days post
stroke.
Primary hypothesis: To compare a policy of early PEG feeding versus initial conservative
management (nasogastric or restricted oral feeding). The primary outcome measure is a
modified Rankin scale incorporating death, and secondary outcomes are incidence of chest
infections, length of stay, survival, discharge destination and Barthel ADL score at discharge.
These trials when completed should provide important evidence on the impact of PEG feeding
in the early post stroke period on survival, functional status and other outcomes.
5.4 Summary: quality and direction of the evidence
• Three small non-blinded RCTs have compared NGT and PEG feeding.
• Two trials included some stroke patients, the third included solely stroke patients.
• The evidence indicates that PEG feeding is more efficient than NGT feeding and is more
acceptable to patients who are able to express a preference.
• In one small trial
23
, PEG feeding was associated with reduced risk of death of 0.3 at 6
weeks after insertion in patients who had persistent dysphagia at 14 days post stroke. This
result requires confirmation in further RCTs.
• There is no useful evidence from randomised trials of the impact of PEG compared to
NGT feeding on mortality in stroke patients with more longstanding dysphagia.
• The impact on longer term mortality and morbidity has not been adequately evaluated in
RCTs.
• Some case series have reported procedure related mortality rates for PEG of up to 4%.
One death in one of the trials
22
may have been related to the PEG procedure.
• Trials in progress will address some of the uncertainties around the benefits of enteral
feeding in stroke patients and the use of PEG versus NGT in these patients.
Percutaneous Endoscopic Gastrostomy
16 West Midlands DES reports December
1999
6 Economic analysis
6.1 Economic literature
No economic analyses comparing PEG feeding with nasogastric feeding in stroke patients were
found.
6.2 Economic analysis
6.2.1 Health consequences of NGT and PEG feeding
6.2.1.1 Process measures: food consumption
There is evidence that more food is consumed with PEG than with NGT
21 23
. It is possible that
this may aid rehabilitation. However there is no evidence that the increase in food consumption
leads to any significant improvement in rehabilitation.
6.2.1.2 Outcome of treatment: mortality
One small trial found that stroke patients fed with PEG had a reduction in mortality compared
to NGT. However, PEG has not been shown to reduce mortality in dysphagic patients after the
first 30 days. Evidence from case series suggests that there may be a risk of procedure related
mortality.
6.2.1.3 Outcome of treatment: quality of life
There have been no formal evaluations of the effect on quality of life of PEG feeding in
comparison to NGT feeding. Most patients who receive enteral feeding following stroke have
poor functional status and a low quality of life. Any improvements in quality of life are therefore
likely to be sufficiently small as to be undetectable on scales for the measurement of generic
quality of life such as the EQ5D. There have been case reports illustrating the positive role
which PEG feeding can play in the rehabilitation of a some stroke patients. In particular this has
been when neurological damage apart from dysphagia is relatively limited. In these specific
cases, quality of life gains may be more substantial
28
29
.
Two trials
21 22
indicate a patient preference for PEG over NGT feeding: one in reported patient
preferences, the other by patient crossover to PEG. This preference for PEG suggests that for
the majority of patients subjective quality of life was better with PEG than NGT. As quality of
life was not specifically measured, it is difficult to quantify this preference.
In two trials
22 23
, more patients fed by PEG than by NGT were discharged from hospital. If
these additional discharges could be attributed to PEG feeding, this suggests that patients fed
by PEG may have a better functional outcome than those fed by NGT.
6.2.2 Health consequences: summary
Patients fed by PEG seem to have a slight improvement in their quality of life. It is not possible
to quantify this in the context of their overall quality of life. There is no clear evidence that PEG
offers advantages in terms of mortality, although more patients may be discharged from
hospital. This must be offset against a small risk of procedure related mortality.
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 17
6.2.3 Resource consequences of NGT and PEG feeding
Because NGT feeding is the alternative to PEG, the overall resource consequences of PEG
should be compared to the overall resource consequences of NGT feeding. The cost
perspective adopted is that of the NHS. Because a formal cost analysis of either PEG or NGT
feeding has not been carried out, a number of estimates are presented below. The first estimate
is based on the estimated additional resource use which results from carrying out NGT feeding
or PEG feeding. Since some of the resources (endoscopy unit) used are available anyway, the
marginal costs of using them are assumed to be zero. The costs of food have been ignored as
these are likely to be the same for both NGT and PEG. This cost estimation therefore focuses
on the additional financial cost and to the NHS and the additional staff time needed.
The second estimate is based on the notional costs of NGT insertion or PEG insertion. This
uses the prices charged for a private procedure as an estimate of the total resource costs.
While it is clearly better to include all the costs of a procedure, there is no guarantee that the
price charged is an accurate reflection of the cost of the procedure.
Table 9: Resource consequences of NGT and PEG feeding.*
NGT PEG
Number of
procedures per
patient
NGT typically reinserted a further
1 or 2 times (possibly more)
Rarely needs to be inserted more
than once
Staff time per
procedure
20 minutes of staff time per NGT
insertion (nurse or junior medical
staff)
30 minutes each of consultant
medical staff time and senior nurse
time per PEG
Additional facilities
per procedure
None (carried out in ward) Use of endoscopy unit
Additional
investigations per
procedure
Chest x-ray
Disposables per
procedure
NGT and associated disposables PEG catheter
Effect on other
costs
Possibly reduced length of stay in
hospital
* Cost estimates supplied by the Royal Wolverhampton Hospitals NHS Trust.
6.2.4 Identifying resource consequences
NGT insertion requires the use of 20 minutes of staff time (typically nurse or junior medical staff
time). The procedure is carried out on the ward. Some disposables and the NGT itself are also
needed. NGTs typically need to be reinserted. In one trial, a mean of 2.7 NGTs were used per
patient, compared to a single one PEG catheter. The figure of 2.7 insertions is therefore used in
this cost estimation although the use of as many as 12 nasogastric tubes has been reported
elsewhere
17
. To carry out PEG requires the use of the endoscopy unit, consultant medical staff
and nursing staff. About 30 minutes of staff and endoscopy unit time is needed. The PEG
Percutaneous Endoscopic Gastrostomy
18 West Midlands DES reports December
1999
catheter itself costs £41. The main resource implications to the health service of NGT and PEG
feeding are listed in
Table 10: Evaluation of resource consequences of NGT and PEG feeding.*
NGT PEG
Staff time Per NGT
insertion:
20 minutes
Per patient (x
2.7):
54 minutes
Per patient:
60 minutes
Additional
investigations
Chest x-ray:
£9.20 to £11
Per patient (x
2.7):
£25 to £30
Nasogastric
tube:
£3.20 to
£8.80
Per patient (x
2.7):
£9 to £24
Disposables
Disposables:
£1 to £2
(est.)
Per patient (x
2.7):
£3 to £5
Cost of a PEG catheter:
£41
Total Financial cost:
£36 to £59
Staff time (nursing or junior
medical staff):
54 minutes
Financial cost:
£41
Staff time (nursing & senior
medical staff):
60 minutes
* Cost estimates supplied by the Royal Wolverhampton Hospitals NHS Trust.
6.2.5 Measuring resource consequences
In terms of staff time and financial costs to the NHS, there is little difference between the cost
of NGT feeding and PEG feeding. (Table 10) NGT feeding is likely to be slightly less costly
than PEG feeding because it can be inserted by less senior staff and because the PEG catheter
is more costly than disposable nasogastric tubes.
This approach ignores the marginal costs associated with the use of the endoscopy unit and
possible marginal savings associated with reduced length of stay in hospital. If the endoscopy
unit is not in continuous use, the opportunity cost of additional use for PEG is likely to be low.
Similarly, cost savings associated with reduced length of stay are only likely to be realised if
hospital beds and hospital staff are redeployed.
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 19
Table 11: Prices charged for PEG insertion (West Midlands NHS Trusts)
Day
case
In-
patient
Trust A £582 £904
Trust B £401 £753
Trust C
(physician)
£170 £1295
Trust C
(surgeon)
£276 £858
6.2.6 Prices of PEG insertion
Table 11 shows the prices charged by trusts for PEG insertion. These vary widely from one
trust to another and range from £170 to £1295. Unfortunately, equivalent prices for NGT
insertion are not available. If junior medical and nursing staff time costs £15 an hour, we would
expect NGT insertion costs to be in the region of £49 to £72. The resource costs of increased
lengths of stay in patients fed by NGT are difficult to estimate. They are likely to be measured
in hundreds rather than tens or thousands of pounds. By this method, it appears that the
resource costs of PEG insertion feeding are much higher than those of NGT insertion but are
likely to be offset by shorter lengths of stay.
6.2.7 Resource consequences - summary
The full economic costs of PEG feeding and NGT feeding are difficult to estimate. Prices
charged are unlikely to be a true reflection of the economic costs.
In more practical terms, the financial and staffing implications of PEG feeding and NGT feeding
are similar. The principal differences are that PEG insertion is carried out in an endoscopy unit,
by consultant staff with senior nurses in attendance, whereas NGT insertions are carried out on
the ward by ward nursing staff or junior medical staff.
6.3 Summary of economic analysis
The balance of evidence suggests that, compared to NGT feeding, PEG feeding is associated
health gains. These include improved patient wellbeing and possibly earlier discharge. In
practical terms, there are small differences in the resource implications of PEG compared to
NGT. These are principally that PEG requires the use of more senior staff.
6.4 Implications for other parties
PEG is preferred to NGT feeding by nursing staff
6 15 22 23
.
7 Conclusions
A small proportion of stroke patients will have persistent dysphagia. If enteral feeding is
initiated, there is a choice of feeding via a PEG or a NG tube. Approximately 300 West
Midlands stroke patients per year have a gastrostomy. The number having NGT feeding is
unknown.
Percutaneous Endoscopic Gastrostomy
20 West Midlands DES reports December
1999
Although one small trial found that the early use of PEG feeding as opposed to NGT feeding
was associated with a reduction in mortality in stroke patients, this result needs to be confirmed
and cannot be generalised to patients where enteral feeding is initiated at a later stage.
Some case series have recorded a procedure related mortality of up to 4% for PEG in patient
populations whose condition results in a high risk of death from other causes. One death in a
patient who had not had a stroke in one trial
22
may have been related to the PEG procedure.
PEG feeding of dysphagic stroke patients appears to be associated with small improvements in
wellbeing compared with NGT feeding in generally severely ill patients. The evidence for this
improvement in wellbeing lies in patient preference for PEG over NGT in two small RCTs. For
the vast majority of patients PEG feeding will have no impact on their functional status over and
above that of NGT feeding.
Compared to NGT feeding, PEG feeding is associated health gains including improved patient
wellbeing and possibly earlier discharge. There are small differences in the resource
implications of PEG compared to NGT, principally that PEG requires the use of more senior
staff.
Given the small resource difference and patient and nurse preference for PEG, the decision
whether to use NGT or PEG feeding should be made by clinicians, carers and patients who
should be aware that there have been procedure related deaths.
7.1 Areas of uncertainty
The costs of malnutrition in terms of skin breakdown and increased infections are high. Many
stroke physicians believe that some kind of nutrition should be instituted by 7 to 10 days after
stroke to prevent tissue and muscle breakdown. The impact of enteral feeding on malnutrition
and its consequences, is, however, unclear
12
and has not been convincing addressed in the
RCTs
21 22 23
.
There is no evidence that adequately addresses the question of whether enteral feeding of
dysphagic stroke patients has any impact on survival free of severe disability, and if so, whether
PEG or NGT feeding is associated with improved outcomes.
The FOOD trial (which will not report until 2004) and the PEGASUS trial will address these
issues.
7.2 Time limit for this report
The conclusions reported are likely to remain current until the above trials report.
Percutaneous Endoscopic Gastrostomy
December 1999 West Midlands DES reports 21
8 References
1
O’Mahony D, McIntyre A S. Artificial feeding for elderly patients after stroke. Age and Ageing
1995;24:533-535.
2
Hull MA, Rawlings J, Murray FE, et al. Audit of outcome of long term enteral nutrition by percutaneous
endoscopic gastrostomy. Lancet. 1993;341:891-72.
3
Chapman AH Percutaneous endoscopic gastrostomy feeding. BMJ 1992;305:116.
4
Murphy BL, Sidhu PS and Greene RE Radiological placement of the gastrostomy tube should be the
preferred method. BMJ 1996;312:972-3.
5
Raha SK, Woodhouse KW. Who should have a PEG? Age and Ageing 1993;22:313-315.
6
Raha SK and Woodhouse K The use of percutaneous endoscopic gastrostomy (PEG) in 161 consecutive
elderly patients. Age and Ageing 1994;23:162-3.
7
Kaw M and Sekas G Long-term follow-up of consequences of percutaneous endoscopic gastrostomy
(PEG) tubes in nursing home patients. Digestive Diseases and Sciences 1994;39:738-743.
8
Rabeneck L, McCollough L B and Wray NP Ethically justified, clinically comprehensive guidelines for
percutaneous endoscopic gastostomy tube placement Lancet 1997;349:496-498.
9
Gordon C, Langton Hewer R and Wade DT. Dysphagia in acute stroke. BMJ 1987;295:411-414.
10
Barer DH The natural history and functional consequences of dysphagia after hemispheric stroke.
Journal of Neurology, Neurosurgery and Psychiatry 1989;52:236-241.
11
Ciocon JO, Felix A, Silverstone MD, Graver LM and Foley CJ Tube feeding in elderly patients:
indications, benefits and complications Archives of Internal Medicine 1988;148:429-433.
12
Davalos A, Ricart W, Gonzalez-Huix F, Soler S, Marrugat J, Molins A, Suner R and Genis D, Effect of
malnutrition after acute stroke on clinical outcome. Stoke 1996;27:1028-1032.
13
Patel PH and Thomas E Risk factors for pneumonia after percutaneous endoscopic gastrostomy.
Journal of Clinical Gastroenterology 1990;12:389-92.
14
Ebrahim S.Clinical epidemiology of stroke. 1990 Oxford University Press. Oxford.
15
Panos MZ, Reilly H, Moran A, Reilly T, Wallis PPJW, Wears R and Chesner IM. Percutaneous
endoscopic gastrostomy in a general hospital: prospective evaluation of indications, outcome, and
randomised comparison of two tube designs. Gut 1994;35:1551-1556.
16
Department of Health. Public Health Common Data Set. 1996. University of Surrey.
17
Finucane P, Aslan SM and Duncan D Percutaneous endoscopic gastrostomy in elderly patients.
Postgraduate Medical Journal 1991;67:371-373.
18
Elia M (Ed.) Enteral parenteral nutrition in the community. 1994. British Association for Parenteral and
Enteral Nutrition.
19
Guyatt GH, Sackett DL and Cook DJ. Users’ guides to the medical literature: II. How to use an article
about therapy or prevention. JAMA 1994;271:59-63.
20
Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ and McQuay AJ Assessing
the quality of reports of randomised clinical trials: is blinding necessary? Controlled Clinical Trials
1996;17:1-12.
21
Park RHR, Allison MC, Lang J, Spence E, Morris AJ, Danesh BJZ, Russell RI, and Mills PR Randomised
comparison of percutaneous endoscopic gastrostomy and tube feeding in patients with persisting
neurological dysphagia. BMJ 1992;304:1406-9.
22
Baeten C and Hoefnagels J. Feeding via nasogastric tube or percutaneous endoscopic gastrostomy: a
comparison. Scandinavian Journal of Gastroenterology. Suppl. 1992;194:95-8.
23
Norton B, Homer-Ward M, Donnelly MT, Long RG and Holmes GKT. A randomised propospective
comparison of percutaneous endoscopic gastrostomy and nasogastric tube feeding after acute dysphagic
stroke. BMJ 1996:312:13-16.
24
Mamel J J Percutaneous endoscopic gastrostomy. American Journal of Gastroenterology. 1989;84:703-
710.
25
Wanklyn P, Cox N and Belfield P Outcome in patients who require a gastrostomy after stroke Age and
Ageing 1995;24:510-514.
26
Larson DE, Burton DD, Schroeder KW and DiMagno EP Percutaneous endoscopic gastrostomy:
indications, success, complications, and mortality in 314 consecutive patients. Gastroenterology
1987;9348-433.
27
http://www.hta.nhsweb.nhs.uk/projdets/962901.htm.
28
Ghosh S, Eastwood MA Percutaneous gastroscopy feeding BMJ 1992;305:115.
Percutaneous Endoscopic Gastrostomy
22 West Midlands DES reports December
1999
29
Allison MC, Morris AJ, Park RHR and Mills PR Percutaneous endoscopic gastrostomy tube feeding may
improve outcome of late rehabilitation following stroke Journal of the Royal Society of Medicine.
1992;85:147-149.
