British Journal of Anaesthesia 86 (3): 366±71 (2001)

Post-operative recovery after inguinal herniotomy in

ex-premature infants: comparison between sevo¯urane and
spinal anaesthesia
J. M. Williams*, P. A. Stoddart, S. A. R. Williams and A. R. Wolf

Department of Anaesthesia, The Royal Hospital for Sick Children, Bristol BS2 8ET, UK
*Corresponding author: Sir Humphry Davy Department of Anaesthesia, Bristol Royal In®rmary, Marlborough Street,
Bristol BS2 8HW, UK

We prospectively studied the post-operative recovery pro®le of 28 ex-premature infants

undergoing inguinal herniotomy. All infants had a post-conceptual age of less than 46 weeks at
the time of surgery and were randomized to receive either sevo¯urane (group 1, 14 patients)
or spinal anaesthesia (group 2, 14 patients). All patients received supplemental caudal analgesia
before skin incision. Cardiorespiratory function was continuously recorded in all patients
before and after surgery. A blinded observer analysed each paired recording for prede®ned
episodes of apnoea, hypoxaemia or bradycardia and the reports were used to compare the
two groups. Spinal anaesthesia was attempted unsuccessfully in four patients in group 2. Five
patients in group 1 demonstrated an `excess' number of episodes (median 4, range 3±12) of
clinically silent post-operative cardiorespiratory complications. (`Excess' in our study was
de®ned as a 3-fold or greater increase in the number of post-operative episodes of bradycardia
or apnoea relative to pre-operative occurrence). Three of these patients had pre-existing
abnormal respiratory function and accounted for 80% of the episodes (26/32) of post-operative
bradycardia and all ®ve episodes of post-operative apnoea identi®ed. All episodes of bradycar-
dia and apnoea were temporally unrelated. None of the remaining patients in group 2 demon-
strated an unacceptable number of post-operative cardiorespiratory complications. Our
limited study suggests that general anaesthesia with an inhalational agent such as sevo¯urane
may induce or unmask abnormalities of cardiopulmonary function in predisposed infants. Spinal
anaesthesia may be preferable but it is potentially stressful for the infant and associated with a
clinically signi®cant failure rate.
Br J Anaesth 2001; 86: 366±71
Keywords: anaesthesia, paediatric; anaesthetics volatile, sevo¯urane; anaesthetic techniques,
subarachnoid
Accepted for publication: October 16, 2000

Inguinal hernias are common in former pre-term infants and
require early repair to avoid the risks of incarceration.1 2
However, the risk of life-threatening apnoea after surgery
is signi®cant in this group regardless of the anaesthetic
technique used.2±4 The reported incidence of post-operative
respiratory dysfunction ranges from 20 to 50% and re¯ects
the variation in both anaesthetic technique and study
design.2 5±9 Spinal anaesthesia appears to reduce the risk
of respiratory dysfunction but does not abolish it.2 8
Moreover, the technique requires considerable expertise,
is contraindicated in some infants, has a reported failure rate
of 10±20%, and can be of insuf®cient duration for surgery.
As a result, the technique may need to be abandoned or
supplemented with sedatives or anaesthetic agents, negating
any potential bene®ts.2 7 Therefore, despite the perceived
advantages of this approach, awake spinal anaesthesia has
not gained universal acceptance amongst paediatric anaes-
thetists and many institutions have rejected it.2 3 10
The introduction into paediatric practice in the UK of
sevo¯urane and des¯urane with their properties of enhanced
recovery characteristics may offer a more practical solution.
In neonates undergoing pyloromyotomy, it has been
observed that recovery after des¯urane is twice as fast as
iso¯urane with less post-operative ventilatory disturbance.11
Sevo¯urane has similar recovery characteristics but unlike
des¯urane provides excellent conditions for inhalation
induction.12 13
We therefore designed a prospective, randomized pilot
study that would re¯ect both clinical practice at our
institution and explore the hypothesis that the `recovery

Ó The Board of Management and Trustees of the British Journal of Anaesthesia 2001
 Comparison between sevo¯urane and spinal anaesthesia

pro®le' after sevo¯urane anaesthesia is comparable with
that after spinal anaesthesia. With neither group experien-
cing an increase in the number of episodes of apnoea,
hypoxaemia or bradycardia relative to pre-operative levels.
Table 1 Patient characteristics and intra-operative data. A comparison of
post-conceptual age (PCA), gestational age (GA), weight (Wt), pre-operative
haemoglobin (Hb) and anaesthetic time (induction±skin closure) for the two
groups (median [range]). There was no signi®cant difference between the
groups with regard to and variable. n=number of patients

Group 1 Group 2
(n=14) sevo¯urane (n=10) spinal

Patients and methods PCA (weeks) 38 [32±46] 40 [36±44]

GA (weeks) 30 [23±35] 8 [26±33]
After obtaining Ethical Committee approval and informed Wt (kg) 2.6 [1.2±3.5] 2.8 [1.7±3.6]
parental consent we studied 28 patients undergoing inguinal Hb (g dl±1) 10.2 [9.0±13.4] 10.9 [9.6±12.7]
Bilateral repairs (n) 7 5
herniotomy. All infants had been born at less than 36 weeks
Induction±incision (min) 23 [16±29] 19 [11±28]
gestation and had a post-conceptual age (PCA) of less than Incision±closure (min) 26 [10±45] 28 [12±48]
46 weeks at the time of surgery. Infants with pre-existing
cardiac, neuromuscular or metabolic diseases were ex-
cluded from participation in the study. Pre-operative
haemoglobin concentration and a history of `pre-existing
Table 2 Number of episodes of pre- and post-operative apnoea and
abnormal respiratory function', with or without an on-going bradycardia recorded in each subject. a, apnoea lasting 15 s or longer;
requirement for supplemental oxygen therapy, were noted b, bradycardia (less than 100 beat min±1) for at least 5 s. Y, yes; N, no.
(Tables 1 and 2). We de®ned `pre-existing abnormal *Subjects identi®ed as exhibiting an `excess' of post-operative
cardiorespiratory complications
respiratory function' as the persistence of clinical features
of respiratory distress in association with an abnormal chest Subject Group Abnormal respiratory Number of episodes
no. function/oxygen
radiograph beyond the ®rst 28 days of life, either occurring therapy
alone or in combination with a `current' history of apnoea Pre-op 12 h Post-op 12 h
(i.e. episodes witnessed within the 2 weeks before surgery).
Patients were allocated randomly, using random number a b a b
1* 1 Y/Y ± ± ± 10
tables, to receive either sevo¯urane anaesthesia with caudal 2 1 N ± ± ± ±
analgesia (group 1) or awake spinal anaesthesia with caudal 4 1 N ± ± ± ±
7 1 Y ± 5 ± ±
analgesia (group 2). No sedative pre-medication was
8 1 Y/Y ± ± ± ±
administered but all patients received paracetamol 15 mg 11 1 N ± 1 ± ±
kg±1 per rectum 1 h before surgical incision. 14* 1 Y/Y ± 3 4 12
16 1 N ± ± ± 1
In group 1, general anaesthesia was induced with a 2
17* 1 N ± 1 ± 3
MAC equivalent value for age of sevo¯urane in 100% 18* 1 N ± 1 ± 3
oxygen (1 MAC=3.2% in neonates).14 15 Atracurium 0.5 mg 22* 1 Y/Y ± ± 1 4
24 1 N ± ± ± ±
kg±1 was used to provide muscle relaxation for tracheal
26 1 N ± ± ± ±
intubation and the lungs were ventilated to normocapnia 28 1 N ± 1 ± 1
with an end-tidal sevo¯urane concentration of 0.5±1.0 MAC 6 2 N ± ± ± ±
10 2 Y/Y ± 1 ± 1
in an air±oxygen mixture until completion of skin closure.
12 2 N 1 2 ± 1
In group 2, spinal anaesthesia was established with 0.5% 13 2 Y ± 1 ± 1
bupivacaine 1 mg kg±1. Patients were placed in the left 15 2 N ± ± ± ±
20 2 N ± 1 1 2
lateral position and a 25-gauge short bevelled spinal needle
21 2 Y/Y ± 2 ± ±
inserted in the midline through either the fourth or ®fth 23 2 Y/Y ± 1 ± 1
lumbar interspace until free ¯owing cerebrospinal ¯uid was 25 2 N ± ± 1 2
27 2 N ± ± ± ±
obtained. To ensure that the groups were comparable to one
another in terms of the duration and intensity of analgesia
provided in the initial post-operative period, both patient
groups received before skin incision a single injection of
0.25% bupivacaine 2 mg kg±1 through a 22-gauge cannula All patients were monitored for a 12 h period pre- and
inserted into the caudal epidural space.16 17 `Induction to post-operatively using a portable three channel continuous
incision' (group 1, start of gaseous induction to surgical cardiorespiratory monitor (Nellcor Puritan-Bennett
incision; group 2, skin penetration with spinal needle to EdenTrace II Plus) with internal memory multi-channel
surgical incision) and `incision to closure' times were recorder interfaced with an EdenTec 3310 Assurance
recorded for both patient groups (Table 1). monitor. This integrated monitoring system provided con-
Upon completion of skin closure, sevo¯urane adminis- tinuous contemporaneous recordings of respiratory effort
tration in group 1 was discontinued and residual neuromus- and pattern (impedance pneumography/air¯ow thermistor),
cular block antagonized with neostigmine 50 mg kg±1 and heart rate (ECG) and haemoglobin oxygen saturation (pulse
glycopyrrolate 10 mg kg±1. oximeter) for each patient. In order to allow each patient to

367
 Williams et al.

Table 3 Comparison of median [range of medians] pre- and post-operative cardiorespiratory changes within and between the groups. There was no
statistically signi®cant difference either within or between the groups (P>0.05). *Patient 22

Sevo¯urane group 1 (n=14) Spinal group 2 (n=10)

Pre-op Post-op Pre-op Post-op

SpO2 (%) 97 [88*±99] 97 [90*±100] 96 [93±99] 96 [92±99]

Heart rate 150 [128±179] 155 [137±176] 142 [134±150] 150 [138±168]
Per cent time SpO2 <90% 6 [1±63*] 6 [0±48*] 6 [0±17] 5 [2±28]
Number of episodes of SpO2 <90% h±1 9 [3±20] 10 [4±14] 6 [2±11] 7 [3±16]

act as their own control, those infants presenting for surgery

with an on-going oxygen requirement were prescribed a
constant ¯ow rate of supplemental nasal oxygen for the
duration of the perioperative period of cardiorespiratory
monitoring (Table 2).
Using the following analysis criteria, each computer-
generated printout was individually analysed for evidence
of either movement artefact or electrode dislodgement by an
investigator (J.M.W.) blinded to the anaesthetic technique
used. An event was de®ned as being a result of an artefact if
one or both of the following occurred: (1) one or more of the
above waveforms out of range or unrecognizable and (2) a
SpO2 less than 90% for more than 20 s with no associated
bradycardia or apnoea.
Clinical signi®cance was de®ned as a bradycardia of less
than 100 beat min±1 for at least 5 s, a haemoglobin oxygen
saturation (SpO2) of less than 90% for more than 10 s and
Fig 1 Group distribution of `excess' post-operative cardiorespiratory
apnoea as a sustained respiratory pause of 15 s or longer, or complications related to pre-operative respiratory function.
less than 15 s if accompanied by an SpO2 less than 90% or
bradycardia. These manually validated reports, in which
each subject acted as their own control, were then used to
Table 4 Total number of post-operative episodes of apnoea and bradycardia
draw comparisons between the two groups in terms of the recorded in relation to the pre-operative respiratory function of those patients
number of `excess' post-operative cardiorespiratory com- in group 1, subsequently identi®ed as exhibiting an `excess' of post-operative
plications recorded. `Excess' in our study was de®ned as a cardiorespiratory complications

3-fold or more increase in the number of post-operative Abnormal Normal

episodes of bradycardia or apnoea relative to pre-operative respiratory respiratory
function function
occurrence.
In accordance with departmental guidelines regarding the Number of patients 3 2
routine post-operative observation and care of infants Episodes of apnoea 5 0
potentially at risk of respiratory dysfunction, real-time Episodes of bradycardia 26 6

apnoea monitoring (Graseby MR10 respiration monitor) in

combination with nurse observation was undertaken for all
study patients in the post-operative period.
In view of the small sample size, data are reported as
median (range). Comparison of cardiorespiratory changes
within and between groups (Table 3) was performed using
the Mann-Whitney U test. A P<0.05 was considered
statistically signi®cant.
Results
Twenty-eight patients were recruited into the study. Both
groups were comparable in terms of PCA, gestational age
(GA), weight (Wt), pre-operative haemoglobin level (Hb)
and anaesthetic time (induction±skin closure) (Table 1).
Five patients (four receiving supplemental oxygen) in group
1 and four patients (three receiving supplemental oxygen) in
group 2 ful®lled the study de®nition of `pre-existing
abnormal respiratory function' (Table 2). No patients had
a current history of apnoea or were receiving respiratory
stimulants in the 2-week period before surgery.
Spinal anaesthesia was attempted unsuccessfully in four
patients in group 2 (28% failure rate). All had normal
respiratory function and went on to have their surgery
successfully completed under general anaesthesia but were
excluded from further analysis.
Median pre- and post-operative values for heart rate,
SpO2, relative duration of a decrease of SpO2 (per cent time

368
 Comparison between sevo¯urane and spinal anaesthesia
SpO2 less than 90%) and number of episodes of desaturation
(SpO2 less than 90% h±1) in group 1 were comparable to
values recorded in group 2 (Table 3).
In group 1, three out of the ®ve patients with pre-existing
abnormal respiratory function and two out of the nine
patients with normal respiratory function experienced an
excess of post-operative cardiorespiratory complications
(Table 2). In contrast, none of the 10 patients in group 2
exhibited an excess of post-operative complications when
compared with their respective pre-operative recordings
(Fig. 1).
Table 4 relates the total number of post-operative
episodes of apnoea and bradycardia recorded to the pre-
operative respiratory function of those patients in group 1
identi®ed as exhibiting an excess of post-operative cardior-
espiratory complications (Table 2). The three patients with
abnormal respiratory function accounted for 80% of the
episodes (26/32) of post-operative bradycardia and all ®ve
episodes of post-operative apnoea identi®ed. All recorded
episodes of apnoea (®ve in group 1 and two in group 2) were
temporally unrelated to the recorded episodes of bradycar-
dia and haemoglobin oxygen desaturation (SpO2 less than
90%) (Tables 2 and 3).
One patient in group 1 (patient no. 22), presented for
surgery with a median SpO2 less than 90%. This level of
hypoxaemia persisted throughout the post-operative period
and accounted for the wide range in the duration of
hypoxaemia (median per cent time SpO2 less than 90%; 63
versus 48%) recorded for each study period (Table 3).
The concurrent use post-operatively, as per hospital
policy, of real-time apnoea monitoring in combination with
nurse observation failed to detect any of the above recorded
episodes of cardiorespiratory disturbance. All resolved
spontaneously without intervention.
Discussion
We have endeavoured to design a study that re¯ects as
closely as possible clinical practice at The Royal Hospital
for Sick Children in Bristol. After the work conducted at our
institution by Peutrell and Hughes, a combined approach of
separate spinal and caudal epidural injections has been the
preferred method of anaesthesia for those infants presenting
for lower abdominal surgery who are thought to be at risk of
post-operative cardiorespiratory complications.17 We have
found this approach provides anaesthesia of a suf®cient
quality and duration to permit the completion of bilateral
inguinal herniotomies without the need for additional
incremental boluses of local anaesthetic. The dosing regime
(bupivacaine 3 mg kg±1) in the awake-spinal±caudal group
has been used without untoward sequelae at our institution
for at least the last decade. It is less than the 3.8 mg kg±1
dose used by Gunter and colleagues18 to establish single
injection caudal epidural anaesthesia and compares favour-
ably with the mean total dose of bupivacaine 2.8 mg kg±1
369
(range 2.5±3.7 mg kg±1) used by Peutrell and Hughes in
their study of a combined spinal±epidural technique.17
Several studies to date have determined that the incidence
of clinically signi®cant post-operative apnoea falls expo-
nentially with time from the completion of surgery.2 6 19
Proposed aetiological factors for this include the hang-over
effect of residual anaesthetic agents combined with an
elevated plasma level of circulating endorphins. The
contribution of the latter is in turn related to the intensity
and duration of the surgical insult.5 19 By attempting to
ensure that both groups were comparable in terms of the
duration and intensity of analgesia provided, we hoped to be
able to detect in the initial post-operative period any excess
cardiorespiratory complications attributable to residual
anaesthesia alone.
The results of our pilot study suggest that despite the
enhanced recovery characteristics of sevo¯urane, the recov-
ery pro®le in this vulnerable age group may not be
comparable to that of `awake' spinal anaesthesia in all
infants. However, balanced against this, was our inability to
establish spinal anaesthesia in four subjects (28%), despite
the technique being performed by experienced paediatric
anaesthetists. This highlights the considerable expertise and
practice required and is one of the primary reasons why the
seemingly advantageous procedure of spinal anaesthesia has
not gained universal acceptance in many paediatric institu-
tions.3 7 These ®ndings need to be interpreted within the
context of the limited size of the study and, therefore,
require further comment.
Viewed as a whole, the recovery pro®le of both groups
appeared comparable with neither group experiencing a
signi®cant number of episodes of hypoxaemia (SpO2 less
than 90% h±1) relative to pre-operative levels (Table 3). It
must, however, be appreciated that these results were based
on computer generated recordings and are therefore sus-
ceptible to artefact.20 The addition of real-time video
surveillance in a sleep study unit would have helped to limit
the impact of the unde®ned noise-to-signal ratio of the
recording system, but because of the unpredictable sched-
uling of surgery this was not a feasible option. An observer
blinded to the anaesthetic technique used, therefore checked
the validity of each recording manually.
Five out of the 14 patients (36%) in group 1 compared
with none of the 10 patients in group 2 exhibited an excess
of post-operative cardiorespiratory complications (Table 2).
In patients with pre-existing abnormal respiratory function,
three out of ®ve patients in group 1 demonstrated an excess
of post-operative cardiorespiratory complications compared
with none of four in group 2 (Fig. 1). We propose a number
of possible explanations to account for this difference
between the groups.
First, with reference to a standard premature infant
growth chart, we established that those patients in group 2
with pre-existing abnormal respiratory function were born
`small-for-gestational-age' (less than 10th percentile) as
opposed to `appropriate-for-gestational-age' (10th±90th
 Williams et al.
percentile) when compared with the birth weights of
corresponding patients in group 1.21 Cote and colleagues
have determined that when controlled for age, haematocrit
(greater than 30%) and institution, small-for-gestational-age
infants seem to have a somewhat lower risk of post-
operative apnoea.19
Second, it is possible that the observed differences
between the two subgroups are purely a function of the
level of monitoring. In our study, the concurrent use of
standard real-time post-operative apnoea monitoring in
combination with nurse observation, as per hospital policy,
failed to detect any of the subsequently computer identi®ed
adverse episodes (Tables 2 and 4). This raises the interesting
question whether cardiorespiratory complications only
detectable by sophisticated techniques are clinically import-
ant; particularly when they appear to resolve spontaneously
without intervention.
Third, the study of relatively rare events requires a large
patient population that is impossible to acquire at any one
institution. Using a statistical approach recommended by
Hanley and Lippman-Hand, Cote and colleagues deter-
mined that a sample size of at least 300 patients would be
necessary to detect a 1% incidence of apnoea with 95%
con®dence in a particular age group.19 22 Therefore,
considering the limited size of this pilot study, it is possible
that we have rejected our hypothesis when it is, in fact, true.
In other words, the observed difference between the two
groups is not signi®cant and our results are, at best,
inconclusive.
Impedance pneumography is the most widely used form
of respiratory monitoring but has limitations for detecting
obstructive airway problems and a high incidence of false
alarms.20 The importance of upper airway obstruction in the
pathophysiology of apnoea and bradycardia is well
described in infants with periodic breathing and low
PCA.23 24 It is, therefore, reasonable to conclude that
general anaesthesia, which can decrease upper airway
muscle tone, may contribute to the development of
cardiorespiratory complications in susceptible infants.9
Given the limitations of impedance pneumography for
detecting obstructive airway problems, this may account for
why the incidence of apnoea was comparable between the
groups and may also, in part, explain the excess of
temporally unrelated episodes of bradycardia in group 1
(Tables 2 and 4).
It is clear even from our limited study that post-operative
cardiorespiratory complications occur frequently in former
pre-term infants after relatively minor surgical procedures.
Allowing for the high detection rate with sophisticated
computer-assisted continuous recording devices, it is also
apparent that general anaesthesia even with modern evan-
escent agents such as sevo¯urane can induce or unmask
abnormalities of cardiopulmonary function in predisposed
infants. Recent work by O'Brien and colleagues has been
more encouraging.12 They suggest that des¯urane, in terms
of the quality of recovery that it offers, may be the most
370
appropriate potent inhalational agent with which to maintain
general anaesthesia in high risk infants.
In conclusion, spinal anaesthesia may be preferable to
general anaesthesia for ex-premature infants undergoing
lower abdominal surgery within the ®rst weeks of life. But
until this has been demonstrated in large prospective,
controlled studies in comparison with modern low solubility
potent inhalational agents, we will continue to have
reservations about routinely subjecting our patients to an
awake technique that is potentially stressful for the infant
and associated with a clinically signi®cant failure rate.
References
1 Peevy KJ, Speed FA, Hoff CJ. Epidemiology of inguinal hernias in
preterm neonates. Pediatrics 1986; 77: 246±7
2 Welborn LG, Rice LJ, Hannallah RS, Broadman LM, Ruttimann
UE, Fink R. Postoperative apnea in former preterm infants:
prospective comparison of spinal and general anesthesia.
Anesthesiology 1990; 72: 838±42
3 Kunst G, Linderkamp O, Holle R, Motsch J, Martin E. The
proportion of high-risk preterm infants with postoperative
apnoea and bradycardia is the same after general and spinal
anaesthesia. Can J Anaesth 1999; 46: 94±5
4 Sartorelli KH, Abajian JC, Kreutz JM, Vane DW. Improved
outcome utilizing spinal anaesthesia in high risk infants. J Paediatr
Surg 1992; 27 1022±25
5 Welborn LG, Hannallah RS, Luban NL. Anaemia and post-
operative apnoea in former preterm infants. Anesthesiology 1991;
74: 1003±6
6 Kurth CD, Spitzer AR, Broennle AM, Downes JJ. Post-operative
apnea in preterm infants. Anesthesiology 1987; 66: 483±4
7 Krane EJ, Haberkern CM, Jacobson LE. Postoperative apnea,
bradycardia and oxygen desaturation in formerly premature
infants: prospective comparison of spinal and general
anaesthesia. Anesth Analg 1995; 80: 7±13
8 Harnick E, Gregory RH, Potolicchio S, Stewart DR, Seigelman
RE. Spinal anesthesia in premature infants recovering from
respiratory distress syndrome. Anesthesiology 1986; 64: 95±9
9 Liu LMP, Cote CJ, Goudsouzian NG, et al. Life-threatening apnea
in infants recovering from anaesthesia. Anesthesiology 1983; 59:
506±10
10 Wolf AR, Stoddart PA. Awake spinal anaesthesia in ex-
premature infants. Lancet 1995; 346: S13
11 Wolf AR, Lawson RA, Dryden CM, Davies FW. Recovery after
des¯urane anaesthesia in the infant: comparison with iso¯urane.
Br J Anaesth 1996; 76: 362±4
12 O'Brien K, Robinson DN, Morton NS. Induction and emergence
in infants less than 60 weeks post-conceptual age: comparison of
thiopental, halothane, sevo¯urane and des¯urane. Br J Anaesth
1998; 80: 456±9
13 Zwass MS, Fisher DM, Welborn LG, et al. Induction and
maintenance characteristics of anaesthesia with des¯urane and
nitrous oxide in infants and children. Anesthesiology 1992; 76:
373±8
14 Sarner JB, Levine M, Davis PJ, Lerman J, Cook DR, Motoyama EK.
Clinical characteristics of sevo¯urane in children. Anesthesiology
1995; 82: 38±46
15 Lerman J, Sikich N, Kleinman S, Yentis S. The pharmacology of
sevo¯urane in infants and children. Anesthesiology 1994; 80:
814±24
 Comparison between sevo¯urane and spinal anaesthesia
16 Mahe V, Ecoffey C. Spinal anaesthesia with isobaric bupivacaine in
infants. Anesthesiology 1988; 68: 601±3
17 Peutrell JM, Hughes DG. Combined spinal and epidural
anaesthesia for inguinal hernia repair in babies. Paediatr Anaesth
1994; 4: 221±7
18 Gunter JB, Watcha MF, Forestner JE, et al. Caudal epidural
anaesthesia in conscious premature and high risk infants.
J Paediatr Surg 1991; 26: 9±14
19 Cote C, Zaslavsky A, Downes J, et al. Postoperative apnoea in
former preterm infants after inguinal herniorraphy: a combined
analysis. Anesthesiology 1995; 82: 809±22
20 Miyasaka K, Kondo Y, Suzuki T, Sakai H, Takata M. Toward
371
better home respiratory monitoring: a comparison of inductance
and impedance pneumography. Acta Paediatr Japonic 1994; 36:
307±10
21 Battaglia F, Lubchenco L. A practical classi®cation of newborn
infants by weight and gestational age. J Paediatr 1967; 71: 159±63
22 Hanley J, Lippman-Hand A. If nothing goes wrong, is everything
all right?: interpreting zero numerators. JAMA 1983; 249: 1743±5
23 Ruggins N, Milner A. Site of upper airway obstruction in infants
following an acute life-threatening event. Pediatrics 1993; 91:
595±601
24 Milner A, Saunders R, Hopkins I. Apnoea induced by air¯ow
obstruction in infants. Arch Dis Child 1977; 52: 379±82