Iatrogenic puncture of the laryngeal mask airway cuff

Correspondence

Dorsal column function afterspinal/epidural anaesthesia

I take issue with Parry et al.’s suggestionthat walking ‘remains a safe option in themajority of parturients after low-doseepidural and combined spinal/epidural(CSE) labour analgesia’ (Anaesthesia1998; 53: 382–7). In their study ofdorsal column function in labouringpatients receiving either intrathecalbupivacaine 2.5 mg/fentanyl 25 mg (in1 ml), epidural bupivacaine 15 mg/fen-tanyl 30 mg (in 15 ml) or dense spinalanaesthesia for Caesarean section (bupi-vacaine 10 mg/fentanyl 25 mg), theyfound the incidence of dorsal columnimpairment merely 7%, compared withup to 66% in our earlier work [1]. Theirmethodology does not justify this con-clusion for the following reasons,although the unknown effect of pro-longed or repeated doses of local anaes-thetics on dorsal column sensation isacknowledged.

Firstly, assessment of proprioreceptionbegan 20–30 min following instigationof analgesia, in contrast to our studies,which assessed dorsal column signsbefore, as well as after, onset of labouranalgesia [1, 2]. This had the effect offamiliarising patients with the postinter-vention tests, rendering them morelikely to be accurately reported andvalid [3]. Furthermore, Parry et al. donot detail how they actually conductedtheir test of distal joint proprioreception,except to state there was ‘manual dis-placement of the large toe and anklejoint’. We offered each patient a choiceof three responses to distal interphalan-geal movement, i.e. ‘up’, ‘down’ or

‘don’t’ or ‘don’t know’ and took sixtrials, both before and after the analgesicintervention [2]. Three or more incor-rect responses from six attempts wererequired after the analgesia in our studybefore a patient was classified as having‘abnormal’ proprioreception.

Similarly, the pilot study quoted asshowing no dorsal column impairmentafter standard intrathecal analgesia doesnot report its methods in detail [4],which is crucial where there are noobjective measures of impairment andsigns may vary subjectively. Clinicalstudies in this area highlight the diffi-culty of objectively measuring neurolo-gical signs in general and dorsal columnsensory modalities in particular. Soma-tosensory evoked potentials have beenmooted as a putative objective means ofdorsal column function assessment, butcurrent evidence indicates that while itmay be a useful barometer of globalspinal cord function, dorsal columnmodalities cannot currently be distin-guished from other modalities travellingin the spinocerebellar tracts, for example[5]. In these circumstances, it is impor-tant to minimise interobserver variabil-ity by having a single investigatorperform all neurological assessments.While Parry et al. were unaware of thestudy group to which the patient theywere examining belonged, it is unclearwhether more than one co-authorconducted these clinical evaluations.

While Parry et al. are to be com-mended on their ingenuity in designinga purpose-built 128-Hz device fordelivering a vibratory stimulus, suchlengths were hardly necessary, giventhat the traditional clinical method of

using the tuning fork is perfectly ade-quate to evaluate vibration sense if usedcorrectly [3]. Moreover, they reporttesting the lateral malleolus and head offibula, despite the teaching that this testought to commence distally at the term-inal interphalangeal joint of the greattoe, where a vibration abnormality maybe detected with greater sensitivity thanat the more proximal sites [3]. Impor-tantly, Parry et al. did not report anyassessment of Romberg’s sign, which issurprising in view of their emphasis onthe concept that safe walking dependson maintenance of balance. A recenteditorial suggested testing Romberg’ssign would be a minimum test require-ment before allowing ambulation [6]. Iwould also disagree with their interpre-tation of the effect of increasing dose oflocal anaesthetic on different types ofnerve fibres. Scurlock et al. showedthat sensitivity to local anaesthetics wasdependent on the rate of basal dischargeof nerve fibres and because propriore-ceptive fibres have a greater rate ofimpulse transmission, they are moresensitive to a given dose of local anaes-thetic than similarly sized motor fibres[7]. Finally, as the authors state, it isunlikely that compensatory adjustmentsin balance and posture could occur asrapidly as onset of dorsal column impair-ment after regional analgesia, especiallysince the physiological postural changesof pregnancy shift the normal centre ofgravity of the woman’s body anteriorly,rendering balance more difficult [8].

In summary, I welcome Parry et al.’sconcluding call for dorsal column func-tion to be assessed and found normalbefore permitting ambulation after

Anaesthesia, 1998, 53, pages 925–933................................................................................................................................................................................................................................................

925Q 1998 Blackwell Science Ltd

All correspondence should be addressed to Dr M. Morgan, Editor of Anaesthesia, Department of Anaesthetics, Royal Postgraduate Medical School,Hammersmith Hospital, London W12 0HS, UK.

Letters (two copies) must be typewritten on one side of the paper only and double spaced with wide margins. Copy should be prepared in the usualstyle and format of the Correspondence section. Authors must follow the advice about references and other matters contained in the Notice toContributors to Anaesthesia printed at the back of each issue. The degree and diplomas of each author must be given in a covering letter personally signedby all the authors.

Correspondence presented in any other style or format may be the subject of considerable delay and may be returned to the author for revision. If theletter comments on a published article in Anaesthesia, please send three copies; otherwise two copies of your letter will suffice.

regional analgesia for labour, as the safetyof this practice has not been clearlyestablished.

D. BuggyLeicester General Hospital,Leicester LES 4PW, UK

References1 Buggy DJ, Hughes N, Gardiner J.

Posterior column sensory impairmentduring mobile extradural analgesia inlabour. British Journal of Anaesthesia1994; 73: 540–2.

2 Buggy DJ, MacDowell C. Extraduralanalgesia with clonidine and fentanylcompared with 0.25% bupivacaine inthe first stage of labour. British Journalof Anaesthesia 1996; 76: 319–21.

3 MacLeod J (ed.). Clinical examination,6th edn. Edinburgh: Churchill-Livingstone, 1983; 267–95.

4 Plaat F, Singh R, Alsoad SM,Crowhurst JA. Selective sensoryblockade with low-dose combinedspinal-epidural allows safe ambulation inlabour: a pilot study. International Journalof Obstetric Anaesthesia 1996; 5: 220.

5 Loughnan BA, Fennelly ME, HenleyM, Hall GM. The effects of differingconcentrations of bupivacaine on theepidural somatosensory evokedpotentials after posterior tibial nervestimulation. Anesthesia and Analgesia1995; 81: 147–51.

6 Elton CD, Ali P, Mushambi MC.Walking epidurals in labour: a stepforward? British Journal of Anaesthesia1997; 79: 551–4.

7 Scurlock JE, Meymaris E, Gregus J.The clinical character of localanaesthetics: a function of frequency-dependent conduction block. ActaAnaesthesiologica Scandinavica 1978; 22:601–8.

8 MacEvilly M, Buggy D. Back pain andpregnancy. Pain 1996; 64: 405–14.

A replyThank you for the opportunity torespond to Dr Buggy’s letter comment-ing on our recent article (Anaesthesia1998; 53: 382–7).

We disagree that our methodology isflawed. We considered it unethical towithhold epidural analgesia in labouruntil detailed dorsal column testing was

performed as Dr Buggy did [1] in hishighly criticised paper [2]. Although notstated in our article, distal joint proprio-ception was recorded in a similar way tothat of Buggy et al. [1]. Plaat et al.’s study,reported in abstract form, found no evi-dence of dorsal column impairment inany of their patients, which is far more inkeeping with our own data than that ofBuggy et al. [1].

Using somatosensory evoked poten-tials (SEPs) to assess dorsal column func-tion after low-dose combined spinal–epidural (CSE) labour analgesia, wefound no difference in P40 latenciescompared to a control group [3]. Stimu-lation of the posterior tibial nerve at theankle, as performed in this study, isknown to evoke positive scalp potentialswith initial latencies at 40 ms [P40] andcompromise of this pathway at the cordor nerve root level will reduce both theamplitude and the latency of SEPs [4].

Although ideally a single observershould perform and score every neuro-logical test, all observations in our studywere carried out by one of three authorswith each patient being tested to severalclearly defined end points. Interobserverbias was further reduced by using fourmechanical buzzers to independentlydeliver a 128-Hz vibratory stimulus tojoints both within (lower limbs) andoutside (upper limbs) the field of blockto accurately assess abnormalities of vibra-tion sense. Unfortunately the physicalsize of the buzzer prevented testing ofthe big toe.

Romberg’s test is a highly subjectiveand inaccurate test of dorsal columnfunction abnormality. In our recentlycompleted study using computeriseddynamic posturography [5] only minorand clinically insignificant changes instability scores were noted between anambulating labour CSE analgesia groupand a control group of nonlabouringmothers. Indeed three patients withabnormal Romberg signs were shownto have normal proprioceptive scores ondynamic posturography testing.

We disagree that proprioceptive fibresare more sensitive to a given dose oflocal anaesthetic than a similarly sizedmotor fibre. The actual response in vivois less clear. Both fibres may actually beblocked at similar rates. This is because

impulse transmission depends both onthe axon diameter and the degree ofmyelination [6]. Since proprioceptiveand somatic motor fibres are both of asimilar diameter surrounded by similaramounts of myelin, approximately thesame rate of local anaesthetic blockadewould be expected. Therefore with amotor block of the lower limbs afterregional blockade, one should alsoexpect a reduction in proprioception.Our high-dose Caesarean section groupdemonstrating a high incidence of bothmotor block and dorsal column abnorm-ality in the lower limbs, in sharp contrastto the low-dose labour epidural/CSEgroup, supports this theory.

Currently all that we can recommendis that if lower limb motor power isnormal after ambulatory epidural/CSEanalgesia, it is probably safe to allowmobilisation providing that the patientfeels confident enough to do so.

R. FernandoM. G. ParryRoyal Free Hospital,London NW3 2QG, UK

References1 Buggy D, Hughes N, Gardiner J.

Posterior column sensory impairmentduring ambulatory epidural analgesia inlabour. British Journal of Anaesthesia1994; 73: 540–2.

2 Fernando R, Price CM. Posteriorcolumn sensory impairment duringambulatory epidural analgesia inlabour. British Journal of Anaesthesia1995; 74: 349–50.

3 Bell R, Parry MG, Fernando R, et al.Assessment of dorsal column functionusing somatosensory evoked potentials(SEPs) after ambulatory combinedspinal epidural analgesia (CSE) forlabour. International Journal of ObstetricAnesthesia 1997; 6: 199–200.

4 Beric A, Prevec TS. The early negativepotential evoked by stimulation of thetibial nerve in man. Journal ofNeurological Sciences 1981; 50: 299–306.

5 Pickering AE, Parry MG, Ousta B,Fernando R. Use of computeriseddynamic posturography to assessmaternal balance function after lowdose combined spinal-epidural analgesia[abstract]. Anesthesiology 1998 (in press).

Correspondence Anaesthesia, 1998, 53, pages 925–933................................................................................................................................................................................................................................................

926 Q 1998 Blackwell Science Ltd

6 Wildsmith JAW. Peripheral nerve andlocal anaesthetic drugs. British Journal ofAnaesthesia 1986; 58: 692–700.

Are you all right, Mrs Tilt?

Resuscitation of the obstetric patientdemands immediate manoeuvres torelieve aortocaval compression, i.e. uter-ine displacement either by the use of awedge or lateral tilt of the bed or table[1]. Failure to perform such a man-oeuvre at the outset of basic life supportwill make further resuscitative measuresmuch less effective and is likely toaffect outcome adversely [2]. We haveobserved during instructing basic obste-tric life support that many candidates,while performing the task exactly toEuropean Resuscitation Council guide-lines, omit this vital manoeuvre duringsimulated resuscitation situations. Theseguidelines suggest, when faced with acollapsed patient, that responsivenessshould be checked by shaking thepatient and asking ‘Are you all right?’.In an obstetric situation it is at this pointthat uterine displacement, either by useof a wedge or lateral tilt of the table,must be established.

It appears to us that candidates,obviously very well prepared in EuropeanResuscitation Council guidelines, do notautomatically change their mindset in thesetting of an obstetric resuscitation. Byreplacing the ‘Are you all right?’ of theguidelines with ‘Are you all right, MrsTilt?’ (the ‘Mrs Tilt’ would obviously notbe verbalised in real situations, it being inthe mind of the rescuer as a prompt to tiltthe patient) in obstetric resuscitationteaching, the mindset of the rescuer canbe changed to ‘obstetric’ mode. Thisshould lead to the rescuer carrying outthis vital manoeuvre right at the start ofobstetric resuscitation. Used in this way asan ‘aide memoir’, it may be a useful adjunctto obstetric resuscitation teaching.

A. DarkT. EngelhardtAberdeen Royal Hospitals,Aberdeen AB25 2ZN, UK

References1 Lee RV, Rodgers BD, White LM,

Harvey RC. Cardiopulmonaryresuscitation of pregnant women.

American Journal of Medicine 1986; 81:311–8.

2 Rees GAD, Willis BA. Resuscitationin late pregnancy. Anaesthesia 1988; 43:347–9.

Caudal epidurals: the ‘whooshtest’

Eastwood and colleagues (Anaesthesia1998; 53: 305–7) state that they havedemonstrated that the specificity of the‘whoosh test’ is better than that of clini-cal impression. There are some pointsthat I would like to make.

Firstly, I find the use of the terms‘sensitivity’ and ‘specificity’ confusingin this context, where there is noobvious ‘abnormal’ or ‘positive’ result,but merely two different results. Indeed,I would have considered a result indicat-ing an incorrectly sited needle to be amore intuitively ‘positive’ result ratherthan ‘negative’ as in the paper. Secondly, itis possible that the observed differenceswere due to chance and there is no realdifference between the tests. Typically,McNemar’s test would be used to lookfor correlation between observations oftwo binary variates. Moreover, in thiscase the availability of a ‘good standard’allows a different approach. Consideringonly those 121 cases in which fluoro-scopy demonstrated that the needle wascorrectly located, clinical impressionidentified 114 correctly and the ‘whooshtest’ identified 97 correctly. We canconstruct a contingency table thus

Correct Incorrect

Clinical impression 114 7‘Whoosh test’ 97 24

and use a Chi-squared test with onedegree of freedom to give a p value of0.0011, or of 0.0021 if Yates’ continuity

correction is used. The ‘whoosh test’ issignificantly worse at identifying acorrectly placed needle.

Now, considering instead the 10 casesin which fluoroscopy demonstrated thatthe needle was not correctly located,clinical impression identified only twocorrectly, but the ‘whoosh test’ identi-fied six correctly. The numbers in thiscase are far too small too perform a Chi-squared test, so an exact test is used togive a p value of 0.170 and a mid p valueof 0.0948. Neither of these two resultsapproach significance, so Eastwood et al.have failed to demonstrate that the‘whoosh test’ is significantly better atidentifying an incorrectly placed needle.

Finally, the paper does not give all theinformation that the study provided. Weknow the results of the two tests indivi-dually, but not the results of the tests incombination. When actually perform-ing a caudal injection a clinical impres-sion of the position of the needle isobtained during insertion, whether ornot we go on to perform a ‘whoosh test’.These data could conveniently havebeen provided in the format shown atthe foot of this page, which is essentiallya ‘2 ×2 ×2’ table encapsulating all theavailable information.

What I suspect I would have found,had this information been provided, isthat the best strategy is to only perform a‘whoosh test’ if the clinical impression isthat the needle is incorrectly placed. Ofcourse it may be that the number ofincorrectly sited needles is too small toprove this – perhaps a less skilled opera-tor is needed to provide the data on thissubgroup. The investigation does con-firm the validity of the ‘whoosh test’, butthe conclusions that may be drawn arenot those of the original paper.

A. M. J. AthertonFazakerley Hospital,Liverpool L9 7AL, UK

Anaesthesia, 1998, 53, pages 925–933 Correspondence................................................................................................................................................................................................................................................


Fluoroscopy

Correct IncorrectClinical impression Clinical impression

‘Whoosh test’ Correct Incorrect Correct IncorrectCorrect n1 n2 n3 n4

Incorrect n5 n6 n7 n8

A replyDr Atherton criticises the use of thediagnostic test statistics of sensitivity andspecificity in our prospective evaluation ofthe whoosh test. I do not agree with hiscriticisms and I would like to providefurther statistical evidence.

Fluoroscopic imaging provides a goldstandard against which to measure othertests for caudal epidural needle place-ment; statistically, the treatment ofneedle placement status is no differentto that of ‘disease’ status more com-monly associated with diagnostic tests.Here is a fuller analysis of the diagnostictest results concerned:

For the study as a whole: prevalence(pre-test likelihood of correct needleplacement)� 92%.

For clinical impression: � ve pre-dictive (post-test likelihood of incorrectneedle placement)� 93%; ÿ ve predic-tive value A (post-test likelihood ofcorrect needle placement)� 78%; ÿ vepredictive value B (post-test likelihoodof correct needle placement despiteÿ vetest)� 22% sensitivity (true positiverate)� 94%; specificity (true negativerate)� 20%; likelihood ratio (and 95%confidence interval) for � ve test� 1.18(0.98–1.92); likelihood ratio (and 95%confidence interval) for ÿ ve test� 0.29(0.08–1.18).

For whoosh test: � ve predictive(post-test likelihood of incorrect needleplacement)� 96%;ÿ ve predictive valueA (post-test likelihood of correct needleplacement)� 80%;ÿ ve predictive valueB (post-test likelihood of correct needleplacement despite ÿ ve test)� 20%sensitivity (true positive rate)� 80%;specificity (true negative rate)� 60%;likelihood ratio (and 95% confidenceinterval) for � ve test� 2.00 (1.15 to4.78); likelihood ratio (and 95% confi-dence interval) forÿ vs test� 0.33 (0.19to 0.68).

It is perhaps more meaningful, in aclinical sense, to think in terms of oddsin this situation. The post-test odds ofcorrect needle placement equals the pre-test odds of correct needle placementmultiplied by the likelihood ratio. Youcan consider the likelihood ratio as indi-cating the value of the test for increasingcertainty about correct needle place-ment. The likelihood ratios for whoosh

test suggest its superiority (in statisticalterms) to the clinical impression. Theconfidence intervals for the likelihoodratios for clinical impression includeunity and therefore one cannot makerobust inference from them; ideallylarger numbers would be sought. Thegreater variation here is explained by thegreater subjectivity of clinical impressioncompared with the whoosh test.

Analysis of the value of the whooshtest when indicated by clinical impres-sion compared with routine use of thewhoosh test would, as Dr Athertonsuggests, be useful. This was not part ofthe current study design and therefore itwould not have been appropriate toperform a retrospective subgroup analy-sis. Any future larger study of thewhoosh test should aim to eliminate asmuch bias and confounding as possiblein the study design if the real effect ofwhoosh test directed by clinical impres-sion is to be resolved. For example, theintention to use a whoosh test could biasthe clinical impression; therefore ran-domisation to whoosh, no whoosh orclinically indicated whoosh shouldhappen after the clinical impression isrecorded.

Dr Atherton suggests working on anull hypothesis of no associationbetween test and fluoroscopy resultseither in discrete paired comparisonsof binomial proportions or in a strati-fied analysis of the contingency tables.One would hope that the ethics com-mittee would not have allowed thestudy to proceed had the answer tothis test of hypothesis not been deter-mined in advance. The most challeng-ing part of the diagnostic testevaluation is to weigh the clinical con-sequences of a false positive resultagainst those of a false negative result.To consider these elements in separatestatistical analyses with independentprobabilities would be specious. Thediagnostic test approach facilitates clinicalinterpretation and maintains statisticalrobustness.

I. E. BuchanCentre for Clinical Informatics,Institute of Public Health,University of Cambridge, CB2 2SR,UK

Iatrogenic puncture of thelaryngeal mask airway cuff

A 30-year-old male, intravenous druguser presented for drainge of a forearmabscess. Due to problems with veinsanaesthesia was induced with sevoflur-ane in oxygen; a size 4 laryngeal maskairway (LMA) was easily inserted. Thecuff was inflated with 25 ml of air andcorrect placement confirmed. Peri-pheral venous access was initially unsuc-cessful and cannulation of the leftinternal jugular vein was attemptedusing an 18G abbocath cannula with a10-ml syringe attached; the right inter-nal jugular vein was sclerosed. Advanc-ing the cannula, 5 ml of air underpressure entered the syringe unexpect-edly and the cannula was withdrawn.The patient was breathing spontan-eously and there was no suggestion orevidence of a tension pneumothorax. Itwas obvious the LMA cuff had beenpunctured and a further 10 ml of airwas introduced into the cuff. Despitethe puncture of the LMA cuff, the pilotballoon remained inflated and noobvious air leak was present. Thecapnograph trace and reservoir bagmovements remained unchanged, withnormal breath sounds heard on ausculta-tion of the lungs and side of the neck.The pressure in the cuff was measuredthrough the pilot balloon and found tobe 22 cmH2O. Throughout the 25-minprocedure no signs of cuff deflationappeared. The pressure in the cuffremained at 22 cmH2O. Intravenousaccess in a dorsal foot vein was estab-lished and anaesthesia maintained withisoflurane in a N2O/O2 mixture. Theremaining procedure was uneventful.

Examination of the LMA cuff in abeaker of water and inflated with 45 mlof air resulted in a stream of bubblesfrom the puncture site (figure top ofnext page). Whatever volume of airwas introduced, the leak always stoppedwith a residual volume of 24 ml in thecuff and a cuff pressure of 22 cmH2O.Interestingly, the deflation test with themask cuff to a high vacuum did notresult in gradual re-inflation as mightbe expected.

The silicone cuff of the LMA ispermeable to a variety of gases, with



CO2 and N2O rapidly diffusing throughthe membrane [1], causing a rise inpressure within the air-filled cuff [2].The diffusion of gas into the LMA cuffmay have compensated for any leak, butmore likely, as gas diffused into the cuffand the pressure increased, the hole inthe silicone rubber opened, acting as apressure limiting leak at 22 cmH2O.Dental damage to the LMA cuff hasbeen reported [3, 4]. In the first reportthe LMA was used in a ‘can’t intubate,can’t ventilate’ scenario and a 0.5-cmtear noted. The second report duringelective use described the replacementof the LMA on three occasions due to aleak. We are not aware of any reports ofiatrogenic puncture of the LMA cuff andits subsequent use.

Three points arise from this incident.First, neck vein cannulation may bedifficult due to distortion of the oro-pharynx by the LMA. Second, theimportance of performing all pre-usetests is highlighted [5]. In this case thepuncture of the LMA was obvious, but ifthe device had been sterilised for re-use,the leak would have been detected bythe inflation test only (50% overinflationwith the device under water). Finally,

inadvertent puncture of the LMA cuffmay not result in complete cuff deflationand the device may continue to functionnormally.

A. PatelA. PearceGuy’s Hospital,London SE1 9RT, UK

References1 Lumb AB, Wrigley MW. The effect of

nitrous oxide on laryngeal mask cuffpressure. In vitro and in vivo studies.Anaesthesia 1992; 47: 320–3.

2 Brimacombe J, Berry A. Laryngealmask cuff pressure and position duringanaesthesia lasting 1-2 hours. CanadianJournal of Anaesthesia 1994; 41:589–93.

3 Pennant JH, Gajraj NM, Griffith K.Puncture of the laryngeal mask airwaycuff. Anaesthesia 1994; 49: 448.

4 McLure HA. Dental damage to thelaryngeal mask. Anaesthesia 1996; 51:1078–9.

5 Brimacombe JR, Brain AIJ. TheLaryngeal Mask Airway: A Review andPractical Guide. London: W. B.Saunders Company Ltd, 1997.

The intubating laryngeal maskairway (ILMA) in failed anddifficult intubation

We were interested to read about thethree cases of difficult intubation over-come by the use of the ILMA (Anaes-thesia 1998; 53: 343–8). We havesimilarly noted success rates comparableto those of the original investigators [1, 2]in a series of 100 patients with normalairways. However, we wish to report onecase in which tracheal intubation via theILMA proved unsuccessful.

A 58-year-old Chinese man withmyasthenia gravis was admitted to theintensive care unit because of ventilatoryfailure 2 days after an emergencyappendicectomy. At the time of sur-gery, trachael intubation had provedimpossible following rapid sequenceinduction and was eventually achievedusing a fibreoptic bronchoscope, afterthe patient had safely been woken up.Postoperatively, he had been sent to thehigh dependency unit. On admission toICU, he was responsive to commandsand was able to maintain an oxygensaturation of 98%, but breathing waslaboured with shallow tidal volumes, itwas possible to manually assist ventila-tion using a facemask. Following propo-fol 20 mg, direct laryngoscopy revealed agrade 4 view [3]. Tracheal intubationwas attempted, using a gum elasticbougie and stylet, by three specialistanaesthetists, with no success. Initial useof the fibreoptic bronchoscope alsoproved unsuccessful. A size 4 ILMAwas then inserted and adequate ventila-tion achieved. However, it provedimpossible to pass a size 7.0 mm silicontracheal tube via the ILMA. A fibre-optic bronchoscope was passed downthe metal stem of the ILMA, but theview was obscured because of bleeding.The ILMA was removed and afterfurther tries the trachea was eventuallyintubated using a size 7.0 mm trachealtube railroaded over a fibreopticbronchoscope passed nasally.

It is important to report both successand failure especially since the ILMA isa new device which has been devel-oped based on MRI studies in oneparticular population. In addition, allthree cases described by Parr et al. had



received neuromuscular blockers to aidintubation.

C. L. LimL. HawthorneP. C. Ip-YamSingapore General HospitalSingapore

References1 Kapila A, Addy EV, Verghese C, Brain

AIJ. The intubating laryngeal maskairway: an initial assessment ofperformance. British Journal ofAnaesthesia 1997; 79: 710–3.

2 Brain AIJ, Verghese C, Addy EV, KapilaA, Brimacombe J. The intubatinglaryngeal mask. II: a preliminary clinicalreport of a new means of intubatingthe trachea. British Journal ofAnaesthesia 1997; 79: 704–9.

3 Cormack RS, Lehane J. Difficultintubation in obstetrics. Anaesthesia1984; 39: 1105–11.

Ethics and research inanaesthesia

In his recent editorial on Ethics andResearch in Anaesthesia (Anaesthesia1998; 53: 422–3), Dr Sneyd made anerror which I wish to correct. In thesection on Financial interests, he indi-cated that the British Journal of Anaesthe-sia does not require authors to make anydisclosure of relevant financial interests.However, in the Guide to Contributors,published in every issue of British Journalof Anaesthesia, there is a statement whichhas appeared since March 1996: ‘thereshould be a clear declaration of anyfinancial or commercial interest whichany author may have in the material’.

The British Journal of Anaesthesia hasheld a long-standing interest in promot-ing high ethical standards in anaestheticresearch. We drafted the editorial pub-lished in the four leading anaestheticjournals on duplicate publication [1]and collaborate in order to try to reducethe extent of this problem. During mytime as Editor of the British Journal ofAnaesthesia from 1987 to 1997, I refrainedfrom publishing some material passed bylocal ethics committees, on the groundsthat it did not meet our higher ownethical standards. More recently, the

BJA was invited to join a committeeon publication ethics, convened by DrRichard Smith, Editor of the BritishMedical Journal.

Detection of fraud by the Editor of aJournal is particularly difficult. This pro-blem can only be obviated by close localinstitutional monitoring of research;however, the Guide to Contributors inthe BJA also indicates that the Editormight wish to inspect all raw data if anysuspicion of fraud arises.

Obviously, there are many aspects ofDr Sneyd’s editorial on which I couldcomment at length. However, I wouldmerely make a plea that all authors,either novice or seasoned, read theguide to Contributors for a specificjournal before submitting a manuscriptto that journal; clearly it would beunreasonable not to anticipate thatthese guidelines are updated at regularintervals.

G. SmithLeicester Royal Infirmary,Leicester LE1 5WW, UK

Reference1 Smith G, Miller R, Saidman LJ,

Morgan M. Ethics in Publishing.British Journal of Anaesthesia 1991; 66:421–422.

A replyProfessor Smith has correctly pointedout an error in my editorial. I apologisefor this inaccuracy.

J. R. SneydDerriford HospitalPlymouth PL6 8DH

Acute arrhythmias on inductionof anaesthesia in a child with ablocked shunt

A 13-year-old child with Dandy–Walkersyndrome and a ventriculo-atrial shuntpresented with a history of occipitalheadache, associated with nausea andvomiting, but normal conscious level.A CT scan showed dilated ventricles anda cyst consistent with the Dandy–Walker malformation. This contained apoorly functioning ventricular catheterand shunt revision was indicated. The

anaesthetist detected a pan-systolicmurmur and ordered an ECG, whichshowed sinus rhythm at 64 beats.minÿ1

with inverted Twaves in leads III, aVR,aVF, V1, V2 and V3. In the absence of aprevious cardiac history, it was decidedto proceed. A modified rapid sequenceinduction was performed, with fentanyl,thiopentone and suxamethonium, fol-lowed by vecuronium. A bradycardiaon intubation resolved, but 5 min latera supraventricular tachycardia at 160beats.minÿ1 developed, with a bloodpressure of 190/110 mmHg and oxygensaturation of 90% whilst receiving 100%oxygen. The child developed frankpulmonary oedema. The neurosurgeonaspirated 20 ml of cerebrospinal fluidfrom the Holter valve and the arrhythmiaceased. Blood pressure settled and oxygensaturation improved. The patient’s lungswere ventilated on the intensive care unitafter insertion of an external ventriculardrain and the trachea extubated 24 h later.The ECG changes reverted and themurmur disappeared. Echocardiographyconfirmed a normal heart. Generalanaesthesia for insertion of a definitiveventriculoperitoneal shunt was subse-quently uneventful.

Ventriculo-atrial shunts are associatedwith pulmonary thrombo-embolismand pulmonary hypertension, but pre-operative findings were not suggestive ofthis [1]. Suxamethonium increases intra-cranial pressure, but the vomiting childposes an anaesthetic dilemma. Modifiedrapid sequence induction was appro-priate but may have precipitated thearrhythmia. Fentanyl and thiopentonedid not offset the effects of suxametho-nium and the hypertensive response tointubation. The conscious level of thechild was a poor indication of his criti-cally raised intracranial pressure, but themurmur and ECG changes were sig-nificant. Pulmonary oedema has beendescribed after induction of anaesthesiafor revision of a blocked ventriculoperi-toneal shunt [2]. The termination of asupraventricular tachycardia on induc-tion of anaesthesia, by emergency aspira-tion of CSF from the Holter valve hasnot previously been described.

E. M. C. AshleySt Peter’s Hospital,Chertsey, Surrey KT16 0PZ, UK



J. O’RiordanSt Georges Hospital,London SW17 0QT,UK

References1 Butler PJ, Wheeler RA, Spargo PM.

Life-threatening complications duringanaesthesia in a patient with aventriculo-atrial shunt and pulmonaryhypertension. Anaesthesia 1990; 45:946–8.

2 Braude N, Ludgrove T. Neurogenicpulmonary oedema precipitated byinduction of anaesthesia. British Journalof Anaesthesia 1989; 62: 101–3.

Ventilation duringpercutaneous tracheostomy

Percutaneous tracheostomy is a tech-nique that has revolutionised airwaymanagement in the intensive care set-ting. One of the potential problems withthe technique is damage to the trachealtube or cuff thereof with the introducerneedle of the percutaneous tracheost-omy set. To prevent this complication,the laryngeal mask airway (LMA) hasbeen used to provide a means of ventila-tion while the procedure is performed[1]. However, this technique is notwithout risk and the requirement tore-intubate the trachea with a cuffedtube may occur at any stage, necessitat-ing removal of the LMA, laryngoscopyand subsequent intubation.

The recently introduced intubatinglaryngeal mask (ILM) may provide auseful alternative technique, enablingventilation of the lungs and the abilityto perform rapid re-intubation shouldthe need arise. Preliminary results wouldindicate a 99% success rate for blindtrachael intubation using the ILM [2]and its successful use in cases of difficultintubation has recently been described[3]. The use of the LMA is, however,contraindicated in cases with decreasedgastric motility and increased pulmonarycompliance. We have used the ILM as aventilatory aid during percutaneous tra-cheostomy in a number of patients inwhom use of LMA was not contra-indicated and have found it to be aneffective technique, freeing the trachea

of the presence of a tracheal tube whileretaining the ability to quickly re-securethe airway should the need arise.

D. A. JonesA. J. BallDorset County Hospital,Dorchester DT1 2JY,UK

References1 Lyons BJ, Flynn CGM. The laryngeal

mask simplifies airway managementduring percutaneous dilationaltracheostomy. Acta AnaesthesiologyScandanavica 1995; 39: 414–5.

2 Brain AIJ, Verghese C, Addy EV,Kapila A, Brimacombe J. Theintubating laryngeal mask. II Apreliminary clinical report of a newmeans of intubating the trachea.British Journal of Anaesthesia 1997; 79:704–9.

3 Parr MJA, Gregory M, Baskett PJF.The intubating laryngeal mask. Use infailed and difficult intubation.Anaesthesia 1998; 53: 343–8.

Percutaneous dilatationaltracheostomy in children

The case report on complications ofpercutaneous dilatational tracheostomy(PDT) in children (Anaesthesia 1998;53: 477–80) emphasizes the risks ofextending an invasive techniquebeyond the original target population,but does not present conclusive evidencethat the reported complications weresolely attributable to the technique,were obligatorily inherent in the tech-nique or occurred with greater fre-quency and to a greater degree inchildren. Without the total numbers oroutcomes for different ages of childrenfor tracheostomy via a formal or percu-taneous approach, or indeed prolongedintubation, it is impossible to qualifythe cases reported, and therefore therecommendations, in any risk analysis.

Tracheal stenosis is of multifactorialorigin, circumferential ischaemia being aprincipal cause in the intubated patientand the extent of local trauma being onedeterminant with an invasive procedure.The smaller more mobile structure inchildren, with less rigid cartilage, could

engender such trauma in PDT by alower ‘first-pass’ success rate, a higherincidence of posterior wall injury andthe likelihood of invaginated splinteredcartilage if the dilatation takes placethrough, as opposed to between, therings. Whilst these factors may poten-tiate stenosis in children, there is clearlyan operator-determined element in theextent to which they occur, particularlywith a new technique, this considerationbeing implicitly acknowledged by theauthors’ reference to the complicationsof a formal surgical tracheostomy beingnegligible in ‘skilled’ hands.

On the basis of medical and lay con-cern over such hazards of new pro-cedures, SERNIP [1] was established asthe statutory body ‘policing’ these devel-opments, making it worthy of commentwhen a technique not registered foradult use is introduced into paediatricpractice outwith a formal trial or obser-vational study. This aspect and sub-sequent doubt as to the quality ofinformed consent obtained in this sce-nario appear to be of greater significancethan the two reported complications.

Formal appraisal is warranted, but atrue comparison between the outcomesfrom skilled exponents of both approachesis intrinsically difficult because of a lackof paediatric experience with the percu-taneous technique and its usual practiceby nonsurgical personnel. Given that itwould be hard to justify complicationsoccurring during a ‘learning-curve’ inthis vulnerable population, it is difficultto anticipate ethical approval for sucha study, or SERNIP ratification formore widespread adoption even withfavourable results.

As such it is unclear how this issue canprogress, but it would appear in theinterim that, in view of the above con-siderations, complications of any natureafter a percutaneous approach in child-ren may well prove difficult to defend.

M. D. D. BellThe General Infirmary at LeedsLeeds LS1 3EX,UK

Reference1 Safety and Efficacy Register of New

Interventional Procedures of TheMedical Royal Colleges (SERNIP).



Dangers of using a pneumaticpressure bag to increasetransfusion rate

I was interested to read the letter byKinnear and Stase (Anaesthesia 1998;53: 408) relating a case of extravasationof fluid from a pressurised peripheralline in a patient following cardiopul-monary bypass (CPB). The leakoccurred through an upstream hole inthe vein due to a previous attempt atsiting the cannula in the same arm. Ihave had a similar, though potentiallymore serious, experience.

A large-bore cannula was sited in avein in the patient’s left forearm togetherwith a radial artery cannula prior tocommencement of anaesthesia for CPB.Anaesthesia progressed uneventfully untilafter separation from CPB, when, whilstthe patient was hypovolaemic, a rapidtransfusion of fluid was commenced viathe peripheral line with the aid of apneumatic pressure bag device. Followingtransfusion of 500 ml of gelofusine and aunit of autologous blood, the patient’sarterial pressure, as measured via theradial artery cannula, started to fall rapidly,the central venous pressure remaininglow. Palpation of the aorta suggestedthat the blood pressure was significantlyhigher than that displayed via the radialline. This was confirmed by direct aorticpressure measurement, a discrepancy of> 60 mmHg being found. The peri-pheral line was abandoned as a route offluid administration at this point andanaesthesia progressed uneventfully. Onremoval of the surgical drapes the leftforearm was observed to be very swollenand tense. Luckily the swelling of theaffected limb abated over the ensuing48 h with elevation alone. I concludedthat the peripheral line had ‘tissued’ anda large proportion of the transfusion hadextravasated under pressure leading to a‘compartment syndrome’ externally com-pressing the radial artery and so arte-factually decreasing the systolic bloodpressure displayed on the monitor.

The patient’s arms were secured byusing a drawsheet folded over the armand tucked in, as in Kinnear and Scase’scase; however, no previous attempt atvenous cannulation had occurred in thispatient’s arm. I can only surmise that theuse of the pneumatic pressure bag had

resulted in the bursting of the peripheralvein, perhaps contributed to by thedrawsheet compressing the vein proxi-mally, leading to extravasation of fluidat high pressure with the resultantcompression of the radial artery.

I suggest that pneumatic pressure bagsshould not be used to increase the rate oftransfusion into any peripheral line thatcannot be observed easily by the anaes-thetist. I would also suggest that when afolded drawsheet is used to secure thearms of a pateint prior to cardiopulmon-ary bypass, sufficient slack is left to allowfor the inevitable swelling that mayoccur following CPB due to an increasein extracellular fluid.

D. N. HunterSt Thomas’ Hospital,London SE1 7EH, UK

Pain and venous cannulation

In Dr Brown’s paper regarding pain fromcannula insertion (Anaesthesia 1998; 53:495–6), the results show no statisticaldifference between pain from cannula-tion with a 20G versus a 22G Venflon inhealthy volunteers. The hand used firstand which hand received which size ofcannula did not influence these results. Iwondered whether there was any influ-ence on the results by the order in whichthe cannulae were inserted. Pain path-ways are complex, subject to modulationfrom many sources and existing in aplastic, adaptable system [1]; the stimula-tion arising from the first cannulationmust surely influence the perceivedstimulus of the second. One mightexpect that ‘the first cut is the deepest’.

J. NansonNorth Hampshire Hospital,Basingstoke RG24 9NA, UK

Reference1 Coniam SW, Diamond AW. Practical

Pain Management, 1st edn. OxfordUniversity Press.

Accuracy of bedsidehaemoglobin measurement

The Hemocue is a popular device formeasuring haemoglobin at the point of

care and its accuracy must be established.It was therefore comforting to read thatLardi and colleagues (Anaesthesia 1998;53: 349–52) were able to confirm ourresults in their evaluation of the Hemo-cue using arterial blood samples [1].However, I would like to alert users tothe poor repeatability of haemoglobinestimates from capillary blood samples.In our evaluation of the Hemocue, wefound a high variability between theestimates from sequential capillary sam-ples. As a result, we noted a significantdifference between the variance of theCoulter JT measurements and Hemocueestimates using capillary samples, but notfor venous and arterial samples. As capil-lary sampling is commonly used at thepoint of care especially by nonmedicalstaff, Hemocue’s haemoglobin estimatesfrom these samples must be interpretedwith caution.

P. P. ChenAlice Ho Miu Ling NethersoleHospital,China

Reference1 Chen PP, Short TG, Leung DHY, Oh

TE. A clinical evaluation of theHemocue haemoglobinometer usingcapillary, venous and arterial samples.Anaesthesia and Intensive Care 1992; 20:497–500.

The influence of smoking onpostoperative nausea andvomiting

In response to a letter from Dr Dolenska(Anaesthesia 1997; 52: 1021), Koivurantaet al. (Anaesthesia 1998; 53: 413) defendthe findings of their survey [1] that post-operative nausea and vomiting (PONV)is less common in both male and femalesmokers compared with nonsmokers.We have made some similar observa-tions in recent unpublished work inves-tigating PONV in day-case arthroscopypatients.

Our patients underwent a standard-ised anaesthetic and were exposed toeither isoflurane or desflurane. Wefound a significant incidence of nauseain the desflurane group. Analysis of these76 patients, utilising a Chi-squared test,



showed nonsmokers to be at a signifi-cantly higher risk of nausea than smok-ing patients (Table 1). Further statisticalanalysis of our results was performedwith a Chi-squared test, with Yates’correction for continuity, for the smallersingle sex samples (54 male and 22female patients). There appeared to bestatistical significance for the group as awhole and male patients (p< 0.05).However, for female patients, the asso-ciation of smoking protecting againstnausea was not significant (p< 0.1).This may be due to both sample sizeand a female predisposition to PONV.

There is other support for the obser-vation that smoking protects fromPONV [2]. Koivuranta et al. speculatethat smokers are more tolerant to anaes-thetic gases and other toxins than non-smokers and consequently have a lowerincidence of PONV. We think it isunlikely that any action of inhaled nico-tine should protect from PONV, butpostulate that one of the many othercomplex component chemicals withincigarette smoke may have an anti-emeticeffect. Further research should clarifythe effect of smoking on PONV in allcategories of patient. The prospect ofidentifying the anti-emetic mechanismof smoking may have exciting implica-tions for our understanding and futuretreatment of postoperative nausea andvomiting.

M. HoughB. SweeneyPoole General Hospital,Poole, Dorset, UK

References1 Koivuranta M, Laara E, Snare L,

Alahuhta S. A survey of postoperativenausea and vomiting. Anaesthesia 1997;52: 443–9.

2 Cohen MM, Duncan PG, DeBoer DP,Tweed WA. The postoperativeinterview: assessing risk factors fornausea and vomiting. Anesthesia andAnalgesia 1994; 78: 7–16.

The Budget: bad news forBritish anaesthetists in the USA

We are among the small number ofBritish anaesthetists who have electedto spend time as visiting faculty atinstitutions in the USA.

Until recently our US earnings hadbeen protected from US tax by virtue ofa tax treaty and from UK tax by virtue ofthe co-called foreign earnings deduction.The latter allowed a 100% deductionfor earnings from employment carriedout wholly or partly abroad during a‘qualifying period’ of 365 days or more.For many years this tax break hasallowed US institutions to offer visitingBritish doctors a salary equivalent tothe net income of junior Americancolleagues. Salaries are just highenough to attract British doctors andyet low enough to make visiting facultyprogrammes economically viable. Theconsiderable relocation and start-upcosts mean that most of us only ‘breakeven’ at the end of 12 months.

In his recent Budget statement, theChancellor of the Exchequer announcedthe cessation of the foreign earnings

deduction. The principal targets of thislegislation appear to be a small numberof high earning individuals who haveexploited the system to avoid paying taxon a significant portion of their income[1]. Doctors in our position now face alarge tax bill upon our return to the UK– a bill that we have neither expectednor made provision for. Those currentlyplanning to travel to the USA facefinancial uncertainty. If US salarieshave to be increased to offset this newtax liability the door to future exchangeswill effectively be closed and a trulyunique and valuable educational resourcewill be lost, perhaps for ever. Inevitably asmall number of doctors will choose toremain in the USA. Our belief is that,either way, the National Health Serviceis the loser.

We ask both the Chancellor and theSecretary of State for Health to examineurgently the far reaching implications ofthis budgetary measure before thedamage is irreparable. We suggest thatBritish doctors planning to work in theUSA contact their host institution andtheir tax office.

J. E. ArrowsmithDuke University Medical Center,Durham NC 27710, USA.

M. E. NicolUniversity of Michigan MedicalCenter,Ann Arbor, MI 48105, USA.

R. C. BaconUniversity of Washington MedicalCenter,Seattle, WA 98195, USA.

Reference1 Budget Office Press Release (IR33). 17

March 1998.



Table 1 The incidence of nausea in smo-kers versus nonsmokers exposed to des-flurane.

Sex Smokers Nonsmokers

Nauseated Female 1 7Male 0 6

Nausea-free Female 5 9Male 17 31

Documents

Iatrogenic puncture of the laryngeal mask airway cuff