The Role of Polyneuropathy

in Motor Convalescence After

Prolonged Mechanical Ventilation
Frans S. S.

Leijten, MD; Joukje E. Harinck-de Weerd, MD; Dick C. J. Poortvliet, MSc\s=d\;Al

Objective.\p=m-\Totest the hypothesis that prolonged motor recovery after

term ventilation may be due to polyneuropathy and can be foreseen at an

long\x=req-\

early

stage by electromyography (EMG).

Design.\p=m-\Cohortstudy with an entry period of 18 months. Polyneuropathy was

identified by EMG studies in the intensive care unit (ICU). During a 1-year follow\x=req-\
up, amount of time was recorded to reach a rehabilitation end point.
Setting.\p=m-\Thegeneral ICU of a community hospital.
Patients.\p=m-\Fiftypatients younger than 75 years who were receiving mechanical
ventilation for more than 7 days.
Main Outcome Measures.\p=m-\Arehabilitation end point was defined as return of
normal muscle strength and ability to walk 50 m independently.
Results.\p=m-\In29 of 50 patients, an EMG diagnosis of polyneuropathy was made
in the ICU. Patients with polyneuropathy had a higher mortality in the ICU (14 vs 4;
P=.03), probably related to multiple organ failure (22 vs 11; P=.08) or aminoglycoside
treatment of suspected gram-negative sepsis (17 vs 4; P=.05). Rehabilitation was
more prolonged in 12 patients with polyneuropathy than in 12 without polyneuropathy (P=.001). Of nine patients with delays beyond 4 weeks, eight had polyneuropathy, five of whom had persistent motor handicap after 1 year. In particular, axonal
polyneuropathy with conduction slowing on EMG indicated a poor prognosis.
Conclusions.\p=m-\Polyneuropathyin the critically ill is related to multiple organ
failure and gram-negative sepsis, is associated with higher mortality, and causes
important rehabilitation problems. EMG recordings in the ICU can identify patients
at risk.
(JAMA. 1995;274:1221-1225)

TECHNICAL progress and the emer

gence of intensive care units (ICUs) have
led to prolonged survival in the critically
ill during the last 40 years. Artificial ven
tilation has been a key accomplishment,
because the lungs are probably the most
vulnerable organs in critical illness.1
Much attention has been given to the
problem of weaning from the ventila
tion apparatus. Pulmonary complications
often (up to 50% in one study2) do not
explain why prolonged ventilation is nec
essary. In 1983, neuromuscular distur-

From the Departments of Neurology (Drs Leijten and

Weerd), Clinical Neurophysiology (Drs Leijten and
de Weerd and Mr Poortvliet), and Anesthesiology and
Intensive Care (Dr Harinck-de Weerd), Westeinde
Hospital, The Hague, the Netherlands. Dr Leijten is now
with the Department of Clinical Neurophysiology, Meer
en Bosch, Heemstede, the Netherlands.
de

\s=d\Deceased.

Corresponding author: Frans S. S. Leijten, MD, Department of Clinical Neurophysiology, Meer en Bosch,
PO Box 21, 2100 AA Heemstede, the Netherlands.

bances were recognized in some patients

by electromyographic (EMG) and biopsy
methods.3"6 A distinct polyneuropathy
called "critical illness polyneuropathy"
was differentiated from Guillain-Barr
syndrome7 and associated with weaning
difficulties.8 Many reports on the sub
ject have appeared since.9"16 Critical ill
ness polyneuropathy is now defined as
a predominantly motor axonal polyneu
ropathy with acute onset in a setting of
systemic inflammatory response with
multiple organ dysfunction.17 Its features
are flaccid tetraparesis, respiratory pa
resis, hyporeflexia, muscle atrophy, and
distal sensory disturbances. Because
neurological examination of critically ill
patients is difficult and nearly always
incomplete, a diagnosis of critical illness

polyneuropathy requires neurophysiological or histological methods.

There have been few prospective stud
ies on neuromuscular disease arising in
critical illness. One study18 reported a 70%

Downloaded From: http://jama.jamanetwork.com/ by a Monash University Library User on 09/21/2013

W. de Weerd, MD, PhD

incidence of critical illness polyneuropa

thy in sepsis and multiple organ failure.
Another found myopathy and several
kinds of polyneuropathy (critical illness
polyneuropathy as a major category) in
86% of patients with difficult weaning.19
Recently, it was established that 96% of
patients who receive intensive care for
more than 7 days show myopathie and
neurogenic abnormalities in muscle bi
opsy specimens.20 Problems in the ex
trapolation of these prevalence rates are
the choice of less well-defined entry cri
teria, such as sepsis21 or difficult weaning,
the number of people not studied because
of early death, incomplete cohorts, pos
sible sampling errors (eg, in muscle bi
opsy), and unknown differences in care
and practice.
Previous studies of critical illness poly
neuropathy were inspired by the prob
lem of weaning failure. This remains a
complex issue with many factors in
volved.22 Less attention has been paid
to other clinical consequences of poly
neuropathy. such as motor convales
cence. In one retrospective study of 22
patients,23 complete recovery was seen
in 45% of patients with critical illness
polyneuropathy after 6 weeks to 12
months (mean, 4.5 months); recovery was
incomplete in 18%, and 37% died. An
other study24 noted recovery after in
tensive physiotherapy for 5 to 17 months
in two of five patients with severe poly
neuropathy; the other three remained
invalidated. Generally, a 50% chance of
complete recovery from critical illness
polyneuropathy is reported.17
In this prospective study, we deter
mined the incidence of polyneuropathy
in patients mechanically ventilated for
more than 1 week, using repeated E MG
studies in the ICU. We evaluated sev
eral risk factors for polyneuropathy. Pa
tients were followed up for 1 year after
discharge from the ICU with special at
tention given to their motor perfor
mance. We then tested the hypothesis
that the presence of polyneuropathy in
the ICU is associated with relevant de
lays in convalescence.

METHODS

Study Design
From July 1991 to January 1993, all
patients younger than 75 years who were
receiving mechanical ventilation for more
than 7 days were enrolled in the study.
Two patients with Guillain-Barr syn
drome and four with thiopental-induced
coma were excluded. The patient or, most
often, a close relative was questioned
about signs of polyneuropathy before the

onset of critical illness. Nerve conduc

tion studies and needle examinations
(both referred to as EMG) were per
formed on day 7 to 9 after the onset of
mechanical ventilation and repeated af
ter 3 weeks and 2 months if the patient
was still in the ICU. After discharge,
there was a 1-year follow-up with stan
dard neurological examinations and scor
ing of functional status. In nine patients,
further EMG studies were done on a
voluntary basis during follow-up; other
patients refused a repeated study after
explanation of the lack of clinical conse
quences.

Patients

Seventy-two patients fulfilled the en

try criteria of the study. In 50 patients,

at least one complete EMG study could

be obtained. Studies were incomplete in
22 patients because of severe edema
(n=2), referral from other hospital after
mechanical ventilation for more than 21
days (n=l), curarization until death (n=2),
multiple fractures or bilateral leg am
putation (n=4), refusal to participate or
inability to obtain informed consent
(n=5), or practical circumstances in which
the investigators were hindered (n=8).
Indications for admission in the 50 pa
tients who were enrolled were abdomi
nal surgery (n=12), noncardiac thoracic
surgery (n=4), cerebral surgery (n=3),
primary infection and sepsis (n=12),
polytrauma (n=9), cardiac resuscitation
(n=5), status asthmaticus (n=3), and intracranial hemorrhage (n=2). The crite
ria used to assess complications, such as
sepsis and multiple organ failure, were
obtained from the literature.2526

Characteristics of the ICU

All studies were done in the ICU of our
hospital, which is a multidisciplinary unit
for medical, surgical, neurosurgical, and
neurological patients generally older than
12 years. It has 16 beds with mechanical
ventilation facilities for nine, serving a
621-bed general hospital in The Hague,
the Netherlands (about 450 000 inhabit
ants). The hospital also acts as a regional
trauma and neurosurgical center, but no
cardiac surgery is performed. From July
1991 to January 1993, there were 1764
ICU admissions, with mechanical venti-

lation for more than 24 hours in 368. Of

these, 72 patients were younger than 75
years and ventilated for more than 7 days

(six patients excluded as just mentioned).

The nondepolarizing neuromuscular

drug vecuronium bromide (Norcuron, Or
ganon Teknika

BV, Boxtel, the Nether

lands) was used for intubation at a dose

of 0.1 mg/kg, if necessary, followed by a

maintenance dose of 0.03 mg/kg per hour
or more until the patient stopped breath
ing spontaneously. The administration
was interrupted daily to determine
whether the patient could be ventilated
without paralysis. Intravenous midazolam (Dormicum, Koche Nederland BV,
Mijdrecht, the Netherlands) was admin
istered in a dose of 0.25 to 1 mg/kg per
day until the desired level of sedation
was achieved. All patients on mechanical
ventilation received subcutaneous heparin calcium, sucralfate (via a nasogastric
tube), airway nebulization, and selective
gut decontamination. Systemic antibiot
ics included cefotaxime and, less fre
quently, gentamicin, metronidazole, floxacillin, erythromycin, a combination ofimipenem and cilastatin, and vancomycin.
Fourteen patients received theophylline,
seven of whom also received low-dose
prednisolone. Three patients were given
high-dose prednisolone, two for treatment
of end-stage respiratory distress syn
drome and one for intracerebral disease.
Common ICU practice was not altered
because ofthe study protocol, except that
serum levels of vitamin B and B12, mag
nesium, and phosphate, and thyroid func
tion were checked and corrected.
EMG Studies
The EMG protocol included nerve con
duction studies of the median, tibial, per
oneal, and sural nerves, preferably on the
right side, with assessment of nerve con
duction velocity and amplitudes of the
compound muscle and sensory nerve ac
tion potentials at distal and proximal su
pramaximal stimulation. All motor re
sponses were obtained with surface elec
trodes. After stimulation of the median
nerve at wrist and elbow, motor responses
were deducted from musculus abductor
pollicis brevis, and sensory responses
were deducted from the second digit us
ing ring electrodes. After stimulation of
the tibial nerve at medial malleolus and
knee, responses were obtained from mus
culus flexor hallucis. The peroneal nerve
was stimulated at the anterior distal tibia
and below and above the head ofthe fibula
(the latter to exclude a compression palsy);
responses were recorded from musculus
extensor digitorum brevis. Sural nerve
potentials were elicited using the retro
grade technique with averaged responses.
Concentric needle studies were done in
the musculus abductor pollicis brevis,

Downloaded From: http://jama.jamanetwork.com/ by a Monash University Library User on 09/21/2013

musculus interosseus I, musculus flexor

hallucis brevis, and musculus tibialis an
terior. If abnormal spontaneous muscle
activity (ie, fibrillations and sharp posi
tive waves) was found in any of these, the
contralateral side and other muscles were
studied. Decrement studies were done
by 3-Hz stimulation of the median nerve
in patients who had received vecuronium
for more than 5 days, before a standard
EMG was done. All studies were done at
the ICU with a transportable Mystro Medelec MS20 apparatus (Vickers Health

Co, Cambridge, England).

Criteria for Polyneuropathy
In our experience, a reliable clinical di
agnosis of polyneuropathy in the critically
ill is difficult, especially in the early weeks
ofmechanical ventilation. Therefore, EMG
care

results were taken as a criterion for poly

neuropathy. The EMG findings were clas
sified into three groups according to the
presence of abnormalities of needle stud
ies, conduction studies, or both. Polyneu
ropathy was defined as definitely axonal
when abnormal spontaneous muscle ac
tivity was present in more than one muscle
group, in more than one extremity, and
action potential amplitudes in at least two
nerves ranged below normal on supramaximal stimulation. Polyneuropathy was
classified as conduction-slowing type when
spontaneous muscle activity was absent
but minimal conduction velocities were
not reached and distal latencies were pro
longed in more than two nerves. If both
criteria were met (axonopathy and con
duction slowing), this was classified as
mixed polyneuropathy. Lower limits of
normal (peak-to-peak amplitude and ve
locity) and maximal limits (distal latency)
in the conduction studies, respectively,
were 5.0 mV, 48 meters per second (m/s),
and 4.2 milliseconds (ms) for the motor
median nerve; 16 , 53 m/s, and 3.5 ms
for the sensory median nerve; 5.2 mV, 41
m/s, and 6.0 ms for the tibial nerve; 4.5
mV, 40 m/s, and 5.5 ms for the motor
peroneal nerve; and 5 and 39 m/s for
the sural nerve (reference values at 35C).
Because skin skin temperatures were
greater than 31C in all patients, no cor
rection factors were used. None of the
patients studied showed purely demyelin-

ating polyneuropathy according to pub

lished criteria.27

Clinical Studies During

Convalescence
All patients received passive and ac
tive physiotherapy as part of intensive
care and entered a training program as
soon as

possible. Complete independence

from ventilatory support (except for na

sal oxygen) was considered the first day
of rehabilitation in 32 patients surviving
the intensive care period. Neurological

examinations were done at day 4, after 2

and 4 weeks, and after 3,6, and 12 months.
These focused on detection of paresis,
ataxia, and sensory disturbances, and as
sessment of overall functional motor per
formance. Paresis in all muscle groups of
the extremities was scored using the Medi
cal Research Council (MRC) scale.28 Sen
sory evaluation was done qualitatively and
included vibration, proprioception, and
sharp-blunt discrimination tests. Overall
performance was measured by asking pa
tients to rise from a chair and counting
the number of meters they could walk
without aid.
Criteria for clinical polyneuropathy
were predominantly distal paresis in legs
more than arms, impaired sensory test
ing in the feet with hyporeflexia, signs
of ataxia, and a feeling of "walking on
rubber."
End Point
End point was strength greater than
MRC grade 4/5 in all muscles with ability
to walk for more than 50 m without aid or
ataxia. After analyzing the data, patients

classified into four categories for

reaching the end point within 4 days, 4
weeks, after 4 weeks, or after more than
were

eight patients, motor perfor

impeded by factors unrelated
to polyneuropathy, such as fractures, am
putations, or hemiparesis; these patients
were excluded from the analysis.
Ethical Aspects
The protocol was approved by our hos
pital ethics committee. Informed con
sent was obtained from the patient when
possible, but due to the severity of ill
1 year. In

mance was

ness

involved, often

tained from

consent

was

ob

close relative.

Statistical Analysis
Patients were divided into a nonpolyneuropathy group and a polyneuropathy
group according to the EMG criteria just
described. Differences between groups in
mean age, number of days on medication
and ventilation, and Acute Physiology and
Chronic Health Evaluation II (APACHE
II) scores were tested using a two-sided
Student's t test (corrected for unequal vari
ances). We used 2 analysis to study pro
portional differences in sex, comorbidi-

ties, mortality, occurrence of potential

preexistingpolyneuropathy, admission di
agnoses, and complications. The nonparametric Mann-Whitney Utest was used to
study differences in duration of convales
values less than
cence delay. Two-sided
.05 were considered significant.

RESULTS
EMG Studies
The EMG results during mechanical
ventilation were considered criteria for

Diagnostic Electromyographic (EMG)

Table 1.Results of

Velocity, m/s
I

Studies

Latency,
I

During

Mechanical Ventilation*

Amplitude, mV, |iV

ms

Nerve_No PNP_PNP_No PNP_PNP_No PNP_PNP

3.60.5
Median, motor_52.94.4
52.64.8t
3.70.6t
8.64.4_8.75.6t
Median, sensory

60.37.7

56.910.8

2.60.4

3.00.9

24.9i10.5

Tibial_45.46.0
Peroneal_47.34.6

40.86.6
41.06.1
45.95.8H

5.30.7

6.11.0
7.42.1

10.1

Sural

48.58.2||

5.01.5

22.813.5

5.4_3.32.8
7.14.8_1.91.4

17.010.9||
...

14.610.5H

...

*Values are meanSD. PNP indicates polyneuropathy.

tNo response in one patient. tNo response in three patients. No response in nine patients. ||No response in seven
patients. IJNo response in 20 patients.

polyneuropathy. Neurological examina

tion of critically ill patients on the res
pirator is difficult, always incomplete,

and cannot be relied on in the detection

of polyneuropathy. We will present our
clinical ICU data and discuss these con
clusions in a separate publication. In the
current study, EMG diagnosis was cor
related only to clinical signs after re
covery when an extensive examination
of the patient was possible.
The EMG studies that allowed clas
sification were obtained in 50 patients.
Postmortem microscopic examination of
sural nerves and gastrocnemius muscle
confirmed the EMG classification in eight
of nine cases. In one, segmental demyelinization was found, whereas the EMG
result was normal. Concomitant myopathy not detected by the EMG was found
in two biopsy specimens.
A total of 78 EMG studies were per
formed in 50 patients. There was no cut
off date for classifying patients as having
polyneuropathy. Conclusions from the
second EMG (done after 3 weeks of ven-

tilatory support) were never changed by

later EMG results. The EMG studies per

formed at day 7 to 9 were already ab
normal in four cases. Mean values of con
duction parameters ofthe diagnostic EMG
studies are given in Table 1. The total
incidence of polyneuropathy was 58% (29/
50). Seventy-six percent (22/29) of these
polyneuropathies showed predominantly
axonopathic features of fibrillations and
sharp positive waves on concentric needle
studies as well as diminished amplitudes
of muscle or nerve action potentials after
stimulation. In half of these axonal poly
neuropathies (11/22), conduction veloci
ties were slowed and distal latency times
were increased (mixed polyneuropathy).
In the other half, these parameters were
within normal limits (purely axonal poly
neuropathy). In the remaining 24% (7/29)
of cases, generalized conduction slowing
without signs of abnormal spontaneous
muscle activity was found.
In the group without evidence of poly
neuropathy (21/50 [42%]), the EMG re
sult was normal in 12 patients. In the
other nine patients, four had isolated
neuropathies in the peroneal nerve, two
in the sural nerve, one in the median

Downloaded From: http://jama.jamanetwork.com/ by a Monash University Library User on 09/21/2013

nerve, and two had widespread fibril

lations without other signs (lowered com
pound muscle action potential amplitude
or conduction block).
Decrement studies of neuromuscular
transmission showed no persistent ab
normalities, except that in three patients
monitored after discontinuation of vecuronium bromide, recovery of the neu
romuscular synapse was prolonged be
yond 12 hours.
In six patients with polyneuropathy,
follow-up EMG studies during rehabili
tation confirmed the diagnosis and
showed abnormal recruitment patterns
on voluntary contraction of leg muscles
with polyphasic motor units in reduced
numbers. No myopathie patterns were
found in these patients, as well as three
patients without polyneuropathy who
were studied during rehabilitation.

General Characteristics
General characteristics were com
pared between patients with and with
out polyneuropathy detected by EMG.
There were no major differences in rea
son for ICU admission. For both groups,
38% were admitted to the ICU because
of acute surgery or complicated elective
surgery (eight patients without poly
neuropathy and 11 patients with poly
neuropathy), 24% because of infection
(five vs seven), 18% because of trauma
(four vs five) and 20% for other reasons
(four vs six; all values, .87 to .99). In
addition, comorbidities were evenly dis
tributed: chronic obstructive pulmonary
disease in 18% (four vs five) and history
of cardiovascular disease in 28% (six vs

eight;

values,

.87 to

.94).

Other variables are shown in Table 2.

Estimates of preexisting polyneuropathy
were made from the medical histories.
One patient had vincristine polyneuropa
thy; he died of central nervous system
complications shortly after weaning from
mechanical ventilation. Three patients had
diabetes mellitus for more than 10 years,
and three consumed more than five al
coholic beverages a day for more than 10
years. Two of these patients had normal
EMG results during intensive care. Only
one patient of the seven at risk for pre-

Table 2.Characteristics of 50 Patients

Undergoing

Mechanical Ventilation for More Than 7

No PNP

rehabilitation

Days*

period. All patients with

polyneuropathy detected by EMG
showed paresis in the lower limbs dur
ing convalescence, and 10 (83%) of 12

PNP

_(n=21)_(n=29)_P__
Men, No. (%)_14 (67)_21 (72)_.66
Age, y (meanSD)_54.316.9_59.713.9_.22
Risk factors for preexisting PNP, No. (%)_2(10)_5(17)_.44
No. of days on ventilator, median (range)
No. of days receiving medication, median
Vecuronium

20(8-49)

(range)

Midazolam

1
7

(0-16)
(0-29)

25(8-109)
1
8

(0-21)
(0-36)

had sensory symptoms in the feet. The

latter tended to persist longer than mo
tor deficits. Sensory symptoms also were
found in two of 12 patients without poly
neuropathy; one of these had paresis as
well and would have been considered
polyneuropathic if clinical parameters
were used (she reached her rehabilita
tion end point within 4 weeks).
Clinical signs of polyneuropathy (mo
tor, sensory, and functional) improved
at least partially in all patients during
follow-up. This finding also held for the
two patients suspected of preexisting
polyneuropathy who reached their re
habilitation end point. After 1 year, se
vere residual functional handicap due to
polyneuropathy was noted in five pa
tients (22% of all 1-year survivors).

.03

.47
9

Aminoglycosides_0(0-11)_4 (0-9)_.05

APACHE II

score

in first 24 hours of

ICU,

meanSD_23.1 6.6_22.39.0_.74

Complications, No. (%)

Sepsis
Multiple organ failure
Deaths, No. (%)
In ICU

Total

(<1y)

9(43)
11(52)

14(48)
22(76)

.70

4(19)
9(43)

14(48)
18(62)

.03

.08

.18

*PNP indicates polyneuropathy; APACHE II, Acute Physiology and Chronic Heath Evaluation II; and ICU, Intensive

care

unit.

3.Delays in Reaching the Rehabilitation End Point in 24 Patients Who Had No Other Impairments
Interfering With Motor Function*

Table

_<3d_3d-4 wk_4 wk-1 y_>1 y_Total

No PNP
PNP

*PNP indicates

1
3

0
5

12
12

polyneuropathy.

existing polyneuropathy survived the

1-year follow-up time. Another died at 10

months but had reached the rehabilita

tion end point 5 weeks after the intensive
care period. Both patients were included
in the rehabilitation follow-up study.
The number of days on mechanical
ventilation varied substantially: patients
with polyneuropathy tended to be on the
respirator for a longer period. Adminis
tration of vecuronium was kept to a mini
mum. Only three patients received vec
uronium for more than 10 days (one of
whom did not have polyneuropathy). Six
teen patients (55%) with polyneuropathy
had been given vecuronium for less than
24 hours; 10 received vecuronium only for
intubation or duringoperation. Midazolam
was used more extensively (in seven pa
tients, five with polyneuropathy, for more
than 20 days). Overall, 45% of patients
received vecuronium for at least 24 hours,
and 78% received midazolam. The only
significant difference between the groups
with respect to medication was adminis
tration of aminoglycoside antibiotics
(mainly gentamicin) for suspected gramnegative sepsis. This medication was given
more often to patients with polyneuropa
thy (59% vs 19%). Patients with mixed
polyneuropathy received aminoglycosides
twice as often as those with purely axonal

polyneuropathy.

The overall mortality rate in the

ICU was 36%. This is within the range
predicted by the APACHE II scores at
ICU admission.29 However, although
APACHE II scores had about the same
distribution in both groups, mortality in

the ICU was 3.5 times higher for those

with polyneuropathy. Patients with poly
neuropathy also had a higher mean age
and more multiple organ failures. Mul
tiple organ failure was present in 15 (83%)
of 18 patients who died in ICU, 12 (80%)
of whom had polyneuropathy. The differ
ence in mortality from multiple organ fail
ure with and without polyneuropathy was
not statistically significant (P=.14,

2 test).

Rehabilitation

Twenty-four of 32 patients who sur

vived the intensive care period could be
evaluated for polyneuropathy-related
impairment (Table 3). The remaining
eight patients had other disorders pre
venting evaluation of polyneuropathic
signs and correlated functional impair
ment. Seven had central nervous sys
tem dysfunction, and in one patient, a
leg had been amputated.
Delays in motor convalescence were
found to be different in patients with and
without polyneuropathy (P=.001, MannWhitney U test). Rehabilitation was se
verely slowed in nine patients who did
not reach the end point at 4 weeks after
discharge from the ICU. Eight of these
nine patients had polyneuropathy. All five
patients who still had severe deficits af
ter 1 year had polyneuropathies with con
duction slowing, four of them in the group
with mixed axonal polyneuropathy. Only
one of eight patients with conduction slow
ing reached the rehabilitation end point
within 4 weeks.
Twenty-four patients were examined
for paresis or sensory symptoms in the

COMMENT
In this study, polyneuropathy had an
incidence of 58% among patients younger
than 75 years who were mechanically
ventilated for more than 7 days. Preex
isting peripheral nerve disease probably
does not account for this high rate, al
though it may have been more frequent
than judged by our estimates.
In more than 75% of patients with
polyneuropathy, widespread abnormal
spontaneous muscle activity was found
during needle EMG examination, sug
gestive of acute and active muscle denervation. These cases were classified
as classical (axonal) critical illness poly
neuropathy. The overall incidence in this
study was 44% (22/50), in accordance
with others.19 The exact incidence may
be somewhat lower or higher because
22 patients could not be evaluated ac

cording to protocol.

We cannot substantiate claims by oth

ers10 that the administration of nonde-

polarizing

neuromuscular agents (ve

curonium in our study) is associated with
polyneuropathy in general. Fewer than
half of the patients with polyneuropa
thy were exposed to vecuronium for
more than 24 hours. Even in the few
cases of prolonged vecuronium admin
istration, abnormal EMG results were
not always present. In our cases of
polyneuropathy, amplitude reductions
of muscle action potentials were too
marked to be attributed to the myo
pathie changes described in vecuroniumtreated patients.30 However, we did not
exclude the possibility that vecuronium
enhances denervation in individual cases.
Aminoglycosides were administered
more often to patients with polyneu
ropathy, especially in the mixed-poly-

Downloaded From: http://jama.jamanetwork.com/ by a Monash University Library User on 09/21/2013

neuropathy subgroup. This does not im

ply an etiologic role. It is equally likely
that gram-negative sepsis (the indica
tion for aminoglycoside use) represents
a special threat to neurons, especially in
an autoimmune theory of critical illness
polyneuropathy.17 In the original reports
on critical illness polyneuropathy, pres
ence of gram-negative sepsis was espe
cially noticed.8 We do not think that this
polyneuropathy is drug induced.
Mortality is higher and mechanical ven
tilation is more prolonged in patients with
polyneuropathy. This is not due to
premorbid conditions or immediate risk
factors on ICU admission as represented
by APACHE II scores. Apparently, de
velopment of complications, such as mul
tiple organ failure, are involved. It may
be that a prolonged and more compli

cated critical illness increases the risks of

developing polyneuropathy. This is logi

cal when polyneuropathy is seen as part
of multiple organ dysfunction syndrome

pathophysiology.17 It cannot be answered

here if polyneuropathy within the syn

drome of multiple organ failure poses an

independent mortality risk or simply re
flects the severity of other complications.
Several neurological factors may influ
ence rehabilitation in the critically ill. They
include myopathy, severe muscle atrophy,
contractures, mononeuropathies, pain, de
pression, and polyneuropathy. Myopathy
is an especially difficult problem because
it concurs with polyneuropathy in critical
illness820,23 and is only reliably indicated
by multisite biopsy. In our patients, we

believe that polyneuropathy rather than

myopathy was a major determinant of
motor convalescence for four reasons: (1)
There was a good correlation between
the EMG classification of polyneuropathy
and clinical criteria; (2) 78% of those with
prolonged recovery beyond 4 weeks
showed sensory symptoms; (3) concentric
needle EMG studies during rehabilitation
showed signs of reinnervation; and (4)
polyneuropathy could be confirmed mi
croscopically in eight patients and by
repeated EMG studies during follow-up
in six patients. However, we cannot
exclude a contribution of muscular fac
tors, and if so, it is probable that they play
a role in delaying motor recovery. Con
ventional EMG studies such as ours are
not sensitive in the diagnosis of myopa
thy, especially if there is concomitant

polyneuropathy.
The presence of polyneuropathy is
intuitively and statistically related to im
paired convalescence. Electromyography
is probably the most sensitive means of
detecting polyneuropathy in the inten
sive care situation where the neurologi
cal examination can only be cursory.
By performing EMG studies in the criti
cally ill, motor recovery delays can be
anticipated. This may direct more intense
physiotherapy programs for certain pa
tients at an earlier stage. In our study,
the mixed-polyneuropathy subgroup
showed the highest risk of motor deficit

after 1 year. This may reflect the fact

that any severe and enduring axonal in
jury will lead to secondary demyeliniza-

tion and thus conduction slowing on EMG.

We conclude that prolonged mechani
cal ventilation is associated with the de
velopment of polyneuropathy. Polyneu
ropathy can be found by EMG studies in
the ICU within 3 weeks after the start
of ventilation. Most cases can be defined
as critical illness polyneuropathy: an
acute axonal and predominantly motor

polyneuropathy. Polyneuropathies are

seen in the more severely ill patients.
They give rise to prolonged motor defi

cits that may take more than a year to

In the ICU, EMG characteris
tics may provide prognostic clues for
motor recovery. In particular, when axo
nal disease is accompanied by conduc
tion slowing, recovery is often slow. Fur
ther confirmation of this association will
be necessary. The contribution of muscle
disorders is still unclear in this respect.
Improvements in technique and avail
ability of supportive care will lead to an
increase in the number of patients with
critical illness, multiple organ failure,
and polyneuropathy. Polyneuropathy ac
counts for extended suffering in these
patients who already have great physi
cal and emotional problems. While ques
tions regarding pathophysiology are still
open, we plead for early recognition.
recover.

We thank Frans G. I. Jennekens, MD, PhD, from

the Department of Neurology at the University of
Utrecht, and Hilbert A. C. Kamphuisen, MD, PhD,
from the Department of Clinical Neurophysiology
at the University Hospital Leiden, for their sup
port and critical review of the manuscript.

References
1. Repine JE. Scientific perspectives on adult respiratory distress syndrome. Lancet. 1992;339:466-469.
2. Kelly BJ, Matthay MA. Prevalence and severity
of neurologic dysfunction in critically ill patients.
Chest. 1993;104:1818-1824.
3. Roelofs RJ, Cerra F, Bielka N, Rosenberg L,
Delaney J. Prolonged respiratory insufficiency due
to acute motor neuropathy: a new syndrome? Neurology. 1983;33(suppl 2):240.
4. Rivner MH, Kim S, Greenberg M, Swift TR.
Reversible generalized paresis following hypotension. Neurology. 1983;33(suppl 2):164.
5. Op de Coul AAW, Lambregts PCLA, Koeman J,
Van Puyenbroek MJE, Palmen FMLHG. Neuromuscular complications after artificial respiration
with Pavulon (pancuronium bromide). Clin Neurol
Neurosurg. 1983;85:197.
6. Bolton CF, Brown JD, Sibbald WJ. The electrophysiologic investigation of respiratory paralysis in
critically ill patients. Neurology. 1983;33(suppl 2):186.
7. Bolton CF, Laverty DA, Brown JD, Witt NJ,
Hahn AF, Sibbald WJ. Critically ill polyneuropathy: electrophysiological studies and differentiation from Guillain\p=m-\Barr\l=e'\syndrome. J Neurol Neurosurg Psychiatry. 1986;49:563-573.
8. Zochodne DW, Bolton CF, Wells GA, et al. Critical illness polyneuropathy: a complication of sepsis
and multiple organ failure. Brain. 1987;110:819-842.
9. Bolton CF, Gilbert JJ, Hahn AF, Sibbald WJ.
Polyneuropathy in critically ill patients. J Neurol
Neurosurg Psychiatry. 1984;47:1223-1231.
10. Op de Coul AAW, Lambregts PCLA, Koeman
J, Van Puyenbroek MJE, Ter Laak HJ, Gabre\l=e"\s\x=req-\
Festen AAWM. Neuromuscular complications in
patients given Pavulon (pancuronium bromide) dur-

ing artificial ventilation. Clin Neurol Neurosg.

1985;87:17-22.
11. Williams AC, Sturman S, Kelsey S, Curran FT,
Finnegan JA. The neuropathy of the critically ill.
BMJ. 1986;293:790-791.
12. Lycklama \l=a`\Nijholt J, Troost J. Critical illness
polyneuropathy. In: Matthews WB, ed. Handbook
of Clinical Neurology: Neuropathies. Amsterdam,
the Netherlands: Elsevier Science Publishers; 1987;

7:575-585.
13. Barat M, Brochet B, Vital C, Mazaux JM, Arn\l=e'\
L. Polyneuropathies au cours de s\l=e'\joursprolong\l=e'\s
en r\l=e'\animation.Rev Neurol. 1987;143:823-831.
14. Coronel B, Mercatello A, Couturier JC, et al.
Polyneuropathy: potential cause of difficult weaning. Crit Care Med. 1990;18:486-489.
15. Lopez Messa JB, Garc\l=i'\aA. Acute polyneuropathy in critically ill patients. Intensive Care Med.

1990;16:159-162.
16. Gorson KC, Ropper AH. Acute respiratory failure neuropathy: a variant of critical illness polyneuropathy. Crit Care Med. 1993;21:267-271.
17. Leijten FSS, De Weerd AW. Critical illness
polyneuropathy: a review of the literature, definition, and pathophysiology. Clin Neurol Neurosurg.
1994;96:10-19.
18. Witt NJ, Zochodne DW, Bolton CF, et al. Peripheral nerve function in sepsis and multiple organ
failure. Chest. 1991;99:176-184.
19. Spitzer AR, Giancarlo T, Maher L, Awerbuch G,
Bowles A. Neuromuscular causes of prolonged ventilator dependency. Muscle Nerve. 1992;15:682-686.
20. Coakley JH, Nagendran K, Honavar M, Hinds
CJ. Preliminary observations on the neuromuscular abnormalities in patients with organ failure and

Downloaded From: http://jama.jamanetwork.com/ by a Monash University Library User on 09/21/2013

Intensive Care Med. 1993;19:323-328.

21. Sprung CL. Definitions of spesis\p=m-\havewe
reached a consensus? Crit Care Med. 1991;19:849-851.
22. Lemaire F. Difficult weaning. Intensive Care
Med. 1993;19(suppl):S69-S73.
23. Op de Coul AAW, Verheul GAM, Leijten ACM,
Schellens RLLA, Teepen JLJM. Critical illness polyneuromyopathy after artificial respiration. Clin
Neurol Neurosurg. 1991;93:27-33.
24. Seiser A, Schwarz S, Brainin M. Critical illness
polyneuropathie: Klinik und Langzeitergebnisse.
Wien Klin Wochenschr. 1992;104:294-300.
25. Bone RC, Balk RA, Cerra FB, et al. Definitions
for sepsis and organ failure and guidelines for the
use of innovative therapies in sepsis. Chest. 1992;
101:1644-1655.
26. Goris RJA, te Boekhorst TPA, Nuytinck JKS,
Gimbr\l=e`\reJSF. Multiple-organ failure: generalized
autodestructive inflammation? Arch Surg. 1985;120:
1109-1115.
27. Asbury AK, Cornblath DR. Assessment of current diagnostic criteria for Guillain-Barr\l=e'\syndrome.
Ann Neurol. 1990;27(suppl):S21-S24.
28. Medical Research Council. Aids to the Investigation of the Peripheral Nervous System. London, England: Her Majesty's Stationary Office; 1976.
29. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:818-829.
30. Dodson BA, Kelly BJ, Braswell LM, Cohen
NH. Changes in acetylcholine receptor number in
muscle from critically ill patients receiving muscle
relaxants. Crit Care Med. 1995;23:815-821.

sepsis.