fluid resuscitation traumatic haemorrhage

REVIEW ARTICLE

Fluid resuscitation in traumatic haemorrhage

R. CUTRESS

Nuffield Department of Surgery, John Radcliffe Hospital, Headington, Oxford, UK

It is ancient knowledge that stopping haemorrhageis beneficial, and that blood is necessary to sustainlife. The world's oldest medical record, a Sumerianclay tablet of about 2150 BC, describes the use ofbandages and poultices to stem bleeding. Incontrast the Greeks, over a thousand years later,believed that illness resulted from an imbalance ofone of four humours: blood, phlegm, yellow bileand black bile. If patients were bleeding, physiciansmight even have bled them further, in order toremove some of the 'excess bad humour', blood.'These beliefs and practices, popularized by

Galen, persisted for another 2000 years until in1628 William Harvey published his work on 'themovement of the heart and blood', which hesupported by a long series of experiments. He thusfinally overthrew Galen's theories and put an endto blood letting. After this breakthrough, treatmentfor haemorrhage became more like that which isnow familiar; battlefield blood transfusions werefirst performed during the American Civil War, andprehospital infusion with 5% dextrose was firstroutinely used by the civilian ambulances in Belfastin 1967. Even since these times, however,controversy has abounded with regard to volumeresuscitation in trauma.Trauma is the commonest cause of death in

those aged under 35 years, and the 500 000 Britishhospital admissions after injury result in 13000hospital beds being occupied each day. Since1988, Advanced Trauma Life Support (ATLS)training has been instituted in the UK. This laysdown specific guidelines on how trauma is to bemanaged, and has greatly improved training, butit has not resolved the controversy. As trauma isso common, and a high proportion of those treatedfor trauma receive fluid resuscitation, it is clearlyan important subject that deserves dueconsideration. This review aims to assess theefficacy and the scientific basis of fluid

resuscitation in modern treatment (as exemplifiedby ATLS) of traumatic haemorrhage.Keywords: haemorrhage, hypovolaemia,

resuscitation, shock, trauma

HAEMORRHAGIC SHOCK

Haemorrhagic shock is widely understood to be aclinical syndrome caused by inadequate perfusionof nutrient substrates to the tissues due tohaemorrhage.2 It was not until the 1940s thatWiggers and co-workers separated shock fromhaemorrhagic hypotension.3 In a series of classicalexperiments they withdrew set volumes of bloodfrom dogs and then reinfused the withdrawn(heparinized) blood. They showed that if the volumewithdrawn was too great, or if the delay beforereinfusion was too long, the blood pressure wouldnot then be maintained by reinfusion of the blood,or by subsequent further infusions. They concludedthat, as a result of the withdrawal of the blood, anirreversible process was occurring in the dogs,which they termed 'shock'. They alsodemonstrated that there was a very wide variationbetween individual dogs in the insult required toproduce shock, and that other variables such astrauma increased the susceptibility of the dogs todevelopment of shock. As well as having obviousclinical implications, this finding demonstrated theneed to produce an experimental model thatproduced shock with reasonable reliability. Wiggersand co-workers therefore developed a model theycalled the 'Western Reserve Method', whereby thedog was bled to a blood pressure of 50mmHg,which was maintained for 90 min, and then furtherbled to a pressure of 30 mmHg, maintained for45 min. This reliably led to a mortality of82% within6 h of reinfusion. All the classic experimental workon haemorrhagic shock has used this model, or amodification of it.Modern initial fluid replacement therapy in

trauma is based on concepts developed by Shires

1995 Blackwell Science Ltd

Journal ofAccident andEmergencyMedicine 199512, 165-172

INTRODUCTION

Correspondence:Dr RamseyCutress,1 Mill CloseHemingford Grey,CambridgesuirePE1 8 9DJ,UK

on July 10, 2022 by guest. Protected by copyright.

http://emj.bm

j.com/

J Accid E

merg M

ed: first published as 10.1136/emj.12.3.165 on 1 S

eptember 1995. D

ownloaded from

http://emj.bmj.com/

et al.4 working on the Wiggers model. Theymeasured total-body red cell mass using 51Cr-labelled red blood cells, total body plasma volumeusing 131 I-labelled human serum albumin, and 35S-tagged sodium sulfate to measure total bodyextracellular fluid. Using the 'Western ReserveMethod' they showed that after the return of theshed blood, both the red cell volume and theplasma volume returned to normal; however, thereremained a deficit in extracellular fluid volume.They then determined whether or not replacementof this deficit would improve survival, and observedthat it did. If 10mL kg-1 plasma was added withthe shed blood, mortality decreased from 80 to70%, and if lactated Ringer's solution was addedwith the shed blood, mortality dropped to 30%.As there was no external source for the loss of

extracellular fluid, the question arose as to whetherthe fluid moved isotonically into the cells. Toanswer this question, Shires et al.,4 measured thetransmembrane potentials of skeletal muscle invivo, using modified microelectrodes, beforehaemorrhage, whilst 'shock' was occurring, andduring resuscitation, in the animal model. Theyfound that the transmembrane potential droppedfrom -90 to -65mV during acute profoundhaemorrhagic shock, and that this was reversedduring resuscitation with the shed blood andRinger's solution. Muscle biopsies were also takenat all three stages and, using the chloride spacemethods to measure intracellular sodium,potassium, chloride and water, it was shown thatthe cells swell, gaining sodium, chloride and waterwhilst losing potassium. These ionic changes arebelieved to be due to the depletion of adenosinetriphosphate (ATP) and hence the loss of activityof transmembrane pumps, in particular the sodiumpotassium ATPase. This is one of the key piecesof supportive evidence for the initial use ofasanguinous resuscitation fluids as opposed to,for example, 0-negative blood in the managementof traumatic haemorrhage.

THE ATLS ALGORITHM

Over the past 20 years, arguments have raged asto whether crystalloid is better than colloid in earlytraumatic shock. It is now generally accepted inNorth America and some centres in the UK thatcrystalloid is preferable, largely on the basis of theresults of a meta-analysis of the 17 major clinicalstudies dealing with the issue,5 and on theexperimental work on the animal model outlined

above. Cyrstalloid also has the advantage of beingcheaper than colloid and non-antigenic, and so ithas now been advocated for initial resuscitationof trauma both in England6 and in the USA.7The meta-analysis is quoted to show a relative

difference in mortality rates of 12.3% in favour ofcrystalloid in trauma patients. Interestingly, thisstudy is also believed to show a 7.8% mortalitydifference in favour of colloids in patients with non-traumatic haemorrhage. It is postulated that, in thetrauma patient, the stress of trauma leads to agreater increase in the permeability of thepulmonary capillary membrane than is caused bysuch stress in the non-trauma patient. Leakage ofcolloids into the interstitium will cause an increasein the incidence of Adult Respiratory DistressSyndrome (ARDS) in the trauma patient givencolloids, but not in the non-trauma patient whosebasement membrane remains less permeable.Unfortunately, however, even on pooling all thetrials together in the meta-analysis, the numbersare still small (a total of 826 patients including thenon-trauma groups) for a study that uses mortalityas the endpoint. This is reflected by the fact thatthe parameters defining the 95% confidenceintervals for both trauma and non-trauma casesinclude the value zero. Hence, in this study, for bothtrauma and non-trauma cases there may be nodifference in mortality for either crystalloid orcolloid, so in neither case can it be stated whethercrystalloid or colloid is preferable, although theresults suggest an advantage for crystalloid in thetrauma case and for colloid in the non-trauma case.Larger trials providing narrower confidenceintervals are needed in order to demonstratewhether such an advantage exists.ATLS doctrine is that crystalloid resuscitation

should begin either in hospital or, if necessary,before the patient reaches hospital, as soon as fluidloss is recognized, or is suspected to haveoccurred. Signs of fluid loss can be minimal. Forexample, with 15% (750mL) blood loss slightanxiety occurs, whilst cardiovascular parametersremain almost normal. As the volume of crystalloidgiven needs to be three times that of the bloodloss for adequate replacement (empirical evidence7and the fact that the total extracellular volume isthree times the intravascular volume), fluidresuscitation tends to be aggressive.8 Two litres ofRinger's lactate are often given immediatelythrough wide-bore cannulae, when bleeding issuspected to have occurred in trauma.Subsequent treatment according to the ATLS

1995 Blackwell Science Ltd, Journal of Accident and Emergency Medicine 12, 165-172

R. Cutress

166


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

algorithm depends on the response to the initialinfusion (see Fig. 1). A rapid and sustainedresponse (assumed to indicate satisfactoryreplacement of loss of less than 20% of the bloodvolume) is defined as a return to and maintenanceof normal cardiovascular parameters such as bloodpressure, pulse and pulse pressure, a urine outputof 50 mL-1 and an improvement in capillary returnand 'CNS status'. Patients who respond transientlyare assumed to have lost 20-40% of their bloodvolume or to have ongoing haemorrhage. Infusionof blood is then commenced and the response

observed in order to discriminate between the two.Those patients who do not respond to thistreatment, orwho did not respond at all to the initial2 L of Ringer's solution are assumed to haveongoing haemorrhage, and thus the ATLSalgorithm states that surgical intervention isindicated in order to stop this.The normal response to blood loss is an increase

in heart rate stimulated by a decrease in venousreturn (preload). This initially maintains the cardiacoutput. The blood pressure is maintained by arterialvasoconstriction, leading to increased systemicvascular resistance. If too much blood is lost, thesemeans of compensation fail and blood pressure

falls precipitously. The heart and brain, initiallyprotected by autoregulation, are no longeradequately perfused and tissue death begins tooccur. If blood volume is maintained above a

critical point venous return and hence cardiac

output and blood pressure should be adequate.9Expansion of the vascular compartment is not theonly consideration, however, as the blood thatreaches the tissues must be adequatelyoxygenated. Experiments on dogs have shown thatacute haemodilution per se, without volume loss,is only tolerated to levels of 7 g dL-1 before cardiaccompensation is overwhelmed.10 This is why theATLS algorithm specifies the need for blood afterthe initial 2 L of Ringer's solution. Blood clearlyalso contains many other substances, such as

clotting factors, that may be of particular benefitafter trauma. It appears that, if haemoglobin canbe maintained above 10gdL-1, with packed redcells if necessary, then oxygen delivery will not becompromised.9

IS THE ATLS ALGORITHMSCIENTIFICALLY VINDICATED?

The belief has become almost axiomatic that whenfluid is lost it should be replaced. This logic is asinescapable as that championed by Galen. Wemust be extremely clear as to the intended effectsof our treatments, follow this up with experimentalevidence from appropriate animal models as wellas adequate clinical trials, and hence establishbeyond doubt that the algorithm used producesoptimal results. This is especially important now

that audit has started in the UK to ensure that theATLS algorithm is being followed."

167 Fig. 1. The ATLS algorithm for shock.

1995 Blackwell Science Ltd, Journal ofAccident and Emergency Medicine 12, 165-172

Fluid resuscitationin traumatichaemorrhage


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

Significant criticisms have been levelled at theWiggers model of haemorrhagic shock, namelythat, although the model is extremely effective indemonstrating the fluid shifts that accompanyshock and subsequent resuscitation, it does 'notreproduce the primary pathophysiological eventleading to haemorrhagic shock, namely a vascularinjury'. It will therefore 'not demonstrate howvarious resuscitation regimens affect hemostasisor how alterations in hemostasis affect outcome'.8Blood loss in trauma does not occur in givenvolumes through a catheter to a set blood pressurefor a certain period of time, but occurs throughinjury to blood vessels. To a degree, the losssustained depends at least in part on the normalphysiological mechanisms of stoppinghaemorrhage. This should not be ignored, andunfortunately the Wiggers model does not take thisinto account. Once blood has been withdrawn viathe catheter, further bleeding will not occur throughthe catheter site on resuscitation.As long ago as World War I it was recognized

that 'the injection of a fluid that will increase bloodpressure has dangers in itself. Haemorrhage maynot have occurred to a marked degree becausethe blood pressure has been too low to overcomethe obstacle offered by a clot.12 The article goeson to desctibe how resuscitation should only occurwith the surgeon present to stop any bleeding.During World War II, the American Surgeon Generaladvocated this advice for treating patients withbleeding that was not readily accessible to externalcontrol.13The physiological control of arterial haemorrhage

was studied in a series of experiments by Shaftanet al.14 They cannulated the carotid artery andjugular veins of dogs in order to measure bloodpressure and give fluid infusions, respectively. Theythen dissected out the saphenous artery andsevered it. They found that haemostasis did notoccur in dogs whose blood pressure wasmaintained at normal levels. This was performedeither by transfusion with autologous blood andthen saline, or by vasopressors. On the other hand,those animals that were used as a control and weregiven no fluid produced a 'soft extraluminal clot'that contained the bleeding. This clot was blownaway and rebleeding started from the site if thecontrol animals were transfused within the firsthour in order to raise the blood pressure back tonormal.Further provocative experiments using a

different animal model were performed by Dronen

et al.8,15 They deliberately chose a model with ahigh 1 -h mortality where they argued no one wouldquestion the need for aggressive fluidresuscitation. They bled pigs through a femoralartery catheter to a mean arterial pressure (MAP)of 30 mmHg. They then produced a 4-mm aorticlaceration by pulling a steel wire that they hadalready sewn in place. When the pulse pressuredropped to 5mmHg they proceeded to resuscitatewith saline at a rate of 6 mL kg-' min-1 up to a totalof 90 mL kg-' when resuscitation was continuedwith shed blood at a rate of 2mL kg-' min.-' Acontrol group received no resuscitation, andexperimental groups were resuscitated to MAPsof 40,60 and 80 mmHg. They measured 1 -h survivaland total haemorrhage volumes, amongst otherparameters. The 1-h survival value for the controlgroup was 12.5%, while 1 -h survival values for the40, 60 and 80mmHg groups were 89, 78 and 22%,respectively. Volumes of shed blood were alsogreater in pigs who were resuscitated to higherMAPs. The conclusion drawn from these results isthat, although moderate fluid infusion to maintainlower than normal blood pressures resulted inimproved survival, more aggressive resuscitationto normal blood pressures led to decreasedsurvival and increased haemorrhage volumes. Inthe words of the authors, 'current standard therapyfor the treatment of haemorrhagic shock includesinitial aggressive colloid resuscitation. Our datasuggest that this is not desirable in the presenceof a noncompressible vascular injury'.The concept of keeping blood pressure down in

order to reduce bleeding is not unique or restrictedto trauma patients. In leaking aortic aneurysms,systolic blood pressure is often maintainedbetween 50 and 70mmHg using only smallvolumes of blood or crystalloid infusion. Using thistreatment it has been found that cardiac and othervital functions can be adequately maintained whilstkeeping blood loss from the aneurysm low.16 In arecent article17, a similar approach has beensupported in which life-threatening cardiac ormajor vascular injury is excluded before theinfusion of 2 L of clear fluid.

In another experiment using the same swinemodel, Dronen et al. compared resuscitation withsaline alone, with resuscitation using saline followedby blood, and with resuscitation using bloodfollowed by saline.18 The mortality was found to be100, 37.5 and 25%, respectively. These results,derived from a more appropriate experimentalmodel than the 'Western Reserve Method' of

1995 Blackwell Science Ltd, Journal ofAccident and Emergency Medicine 12, 165-172

R. Cutress

168


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

Fluid resuscitation Wiggers (as used by Shaftan et a/.), challenge thein traumatic experimental basis of the use of crystalloid infusionhaemorrhage as initial therapy for haemorrhagic shock. It must

be noted, however, that although 0-negative bloodcan theoretically be made available as fast ascrystalloid, crossmatched blood obviously cannot.A clinical study is needed in order to assess andcompare the efficacy of 0-negative blood as initialtherapy with that of other fluids; this has not beenundertaken in a prospective study to date.19Although the swine model of Dronen et a/. is

more appropriate in cases of ongoing haemorrhagethan that of Wiggers, some criticisms still remain.In the initial two experiments up to 90 mL kg-1 wasinfused. Weight for weight this corresponds to 7-8 L in a subject weighing 80 kg. ATLS uses only 2 Lof crystalloid before blood. In the animalsresuscitated to higher blood pressures, a muchgreater amount of haemodilution would thereforeoccur than during an ATLS resuscitation, whereblood would be added earlier. Secondly, by designthese studies assess short-term survival and donot consider renal, cerebral or other end organdamage or survival. Thirdly, in all these casesimmediate surgery would be indicated on the basisof non-response to initial saline infusion, if the ATLSalgorithm was used. Despite the above criticismsthese papers are very useful and, by using a moreappropriate model, they clearly reinforce the well-known fact that the definitive treatment of majorhaemorrhage is hamostasis and not transfusion,and that if ongoing haemorrhage is suspected,cautious fluid administration should be used,avoiding rapid changes in blood pressure. This isespecially important now that modern techniquesenable us to infuse fluids through peripheralintravenous lines at rates exceeding 800 mL min-1.20

'SCOOP AND RUN' OR 'STAYAND PLAY'

Even before the first ATLS course in 1977, whichadvocated the earliest possible fluid resuscitation,7there was considerable debate over the benefit ofprehospital fluid resuscitation.19 There has been noevidence to date suggesting that prehospitaladministration of intravenous fluids is of benefit totrauma patients,21 although ATLS as a package hasbeen shown to be more effective in prehospitaltreatment than Basic Life Support.2223 As describedearlier, experience gained during the two worldwars has suggested that resuscitation should not

169 begin until the surgeon is ready to stop the

bleeding. Experiments such as those describedabove have added fuel to the debate. Thearguments against the ATLS algorithm include notonly the theoretical possibility of blowing off a clot,but also the argument that time spent on on-sceneresuscitation was wasted when patients reallyrequired the definitive treatment provided bysurgery. This argument was strongly supported bya retrospective series of 52 American traumapatients.24 It was found in that study that the meantime taken to establish intravenous access was11.3 min, and on average only 489 mL were infused,whilst the average time for transport to hospitalwas 10.7min. In this case one can stronglyquestion the value of prehospital resuscitation andargue for the 'scoop and run' policy. The case mustbe different in more rural areas, however, wheretransport times to hospital might be much longer.Physiological computer modelling suggests thatprehospital volume resuscitation cannot be usefulunless prehospital time exceeds 30 min.25Kaweski et a/. in 1990 reported a retrospective

study of 6855 trauma patients which suggestedthat, even when there was no delay caused by theprehospital administration of fluid, thisadministration did not improve the outcome.26 Theydivided the patients into three groups dependingon initial blood pressure at the scene, and injuryseverity score, and they found that 56% of thepatients had been given intravenous fluids. Whenthe patients who had been given fluids werecompared with those, within the same group ofinjury severity, who had not been given fluids, therewas no statistically significant difference in theoutcome. The best predictor of outcome was, not

surprisingly, found to be the severity of initialinjuries. In this series the average transport timeto hospital was 37 min, irrespective of whetherfluids were given or not. The authors do not state,however, whether those patients who were notgiven fluids had had an intravenous line placedwhich, for example, did not work, as in the light ofthe previous paper, if this were the case in a

significant proportion of cases, and if time was

wasted inserting intravenous lines that did notwork, then perhaps the results could still beinterpreted as supporting a 'scoop and run policy'.In any case this series showed that no advantagewas to be gained by fluid resuscitation.Owing to the results of the previous paper it

became more ethically justifiable to perform a

prospective trial. The preliminary results (from thefirst 300 patients) of such a trial have been reported

© 1995 Blackwell Science Ltd, Journal ofAccident and Emergency Medicine 12, 165-172


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

by Bickell et al.27 Patients included in the trial wereall those who had gunshot or stab wounds to thetrunk. They were randomized according to thecalendar day of injury such that those injured onan even day of the month received immediate fluidresuscitation, and those injured on an odd day didnot receive any fluid until induction of anaesthesia.As the teams worked one 24-h shift every 72 h, the24-h periods of immediate or delayed resuscitationcorrelated with the shifts, and teams alternatedbetween the two groups. In this study there wasfound to be no significant difference in the rate ofsurvival to hospital discharge, between theimmediate resuscitation group (56%) and thedelayed resuscitation group (69%). The full resultsof this unique trial, which may go against the grainof modern teaching on fluid resuscitation, areeagerly awaited.

'ALL BLEEDING STOPS,EVENTUALLY'

If the experiments of Shaftan and Dronen arereproducible clinically such that bleeding is morelikely to stop without resuscitation, andresuscitation can be shown to have no positiveeffect, then one must ask what are the dangers inhypotension or non-resuscitation. Understandably,the greatest concern of anyone dealing with suchpatients must be that bleeding will stop becausethe patient has died through inadequateresuscitation. It is intuitive and has become almostaxiomatic that if someone is losing volume theyshould be given volume. This is what makes theabove study so unique and so important. Thedanger, however, is that this study may beinterpreted to mean that resuscitation is notindicated in all situations. The results of this studyon trunk-penetrating injuries cannot be correlatedwith other common types of injury, e.g. burns orcrush injuries where considerable fluid loss mayoccur in the absence of severe bleeding, ornecessarily with road traffic accidents, which area major source of trauma. The selection criteriafor this study were such that ongoing haemorrhage,poorly responsive to external control, was likely tooccur. This type of trauma, where surgery is alwaysultimately required, is that most likely to benefitfrom delayed resuscitation.

In Bickell's study it must also be noted that thegroup of patients who received no fluid prior toinduction of anaesthesia received on averageabout one 1 L on induction. This highlights an

important point. During hypovolaemia bloodpressure is maintained to a limited extent by thecardiovascular responses of vasoconstriction andtachycardia. On induction of general anaesthesiathese reflexes are lost,28 and the danger is thatblood pressure may plummet. This is one of themajor risks of not resuscitating until induction ofanaesthesia, and the report provides noinformation about the anaesthetic experiencesduring the study. Another effect that might beexpected if the blood pressure is allowed to staylow would be an increase in the incidence of endorgan failure. This does not appear to have been aproblem in the study, but it does highlight the pointthat what is of critical importance is not bloodpressure but tissue perfusion, and if bleeding isallowed to stop through hypotension, then a directindex for measurement of tissue perfusion may beneeded. Possible indices might be arterial lactate,end tidal carbon dioxide or conjunctival oxygen,all of which have been found to show a degree ofcorrelation with the level of tissue perfusion.929These parameters clearly require further study, andat present are probably only useful as an adjunct.Urine production has always been an extremelyuseful index of renal perfusion, but the obviousproblem with this is the time delay before the urineis produced and a reading is obtained. Dopplerand oscillometric techniques of measuring bloodflow tend to produce errors of a clinically significantmagnitude when used in clinical practice. Perhapsif Bickell's study, when completed, clearly showsa requirement for hypotensive resuscitation in thetype of trauma patient being studied, then greateremphasis will be placed on finding other more

direct means of monitoring the tissue perfusion ofthe patient. This will enable the blood pressure tobe dropped, ensuring minimal blood loss but stillallowing confidence that the vital organs are beingadequately perfused.

Other differences between Dronen's swine modeland that of the trauma patient must be considered,such as the variable spectrum of injuries, degreeof tissue damage and nociceptive effect, thevariable delay before treatment can begin, and theincomplete knowledge of the emergency clinicianwith regard to exactly what has occurred. This maybe most clearly evident in the ATLS protocol inpatients who respond transiently to 2 L of Ringer'ssolution. These patients may have stoppedbleeding, but have lost more fluid than has beenreplaced, in which case the treatment is to givethem more fluid and return their blood pressure to

1995 Blackwell Science Ltd, Journal of Accident and Emergency Medicine 12,165-172

R. Cutress

170


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

Fluid resuscitationin traumatichaemorrhage

normal, maximizing tissue perfusion. On the otherhand, they may still be bleeding, in which case, asdiscussed above, the experiments of Dronensuggest that a balance must be struck, selectinga blood pressure at which bleeding is minimizedbut the vital organs are adequately perfused. Whatis clearly needed is a means of determining veryrapidly which of those two categories a 'transientresponder' belongs. At present there does notappear to be any index that indicates whether apatient is actively bleeding. Until that time, it is fairto say that the ATLS algorithm, probably correctly,errs on the 'safe' side.

CONCLUSIONS

There is no doubt that as a package, ATLS providesextremely useful guidelines, and it has been shownto make a substantial contribution to themanagement of trauma. One of the positivefeatures of the controversy regarding fluidresuscitation that it has shown is that, althoughATLS has been of enormous benefit, to improve it,each of its parts must be taken and studiedselectively, and be shown to be experimentally andclinically validated.

In conclusion, four key points arise. First, as themodel used to support the initial use of crystalloidin haemorrhagic shock is not appropriate in thecase of ongoing haemorrhage, and as the meta-analysis of crystalloid against colloid is onlysuggestive but not significant, a large prospectivetrial is needed to compare mortality with initial fluidresuscitation using 0-negative blood, Ringer'slactate solution, and a colloid. Secondly, whilst twoof the studies described show no statisticaladvantage gained from fluid resuscitation, due totheir limitations, and other advantages gained fromATLS as a whole, it is still safer to resuscitate asdescribed in the ATLS protocol, followingconventional wisdom, but ensuring that this doesnot cause any delay in transport and definitivetreatment. Thirdly, even if Bickell's study whencompleted shows that there is an advantage in notresuscitating those with the type of trauma studied,an alternative direct means of measuring tissueperfusion, and a means of establishing whether apatient is actively bleeding, will be needed in orderto allow the use of non-resuscitation with fluid, asan option prior to surgery, in some cases. Fourthly,the estimated transit time to hospital will alwaysplay an important part in the decision made on howto resuscitate.

One of the fascinating things about medicine isthat there can be no rules, but guidelines are clearlyneeded, and these must be shown to beappropriate. ATLS is appropriate at the moment,but if we are not to fall into the same traps as thosewho followed Galen for 2000 years we mustcontinue to put it to the test and not acceptconventional wisdom unconditionally.

REFERENCES

1. Duin N. & Sutcliffe J. (1992) A History of Medicine,p. 16. Simon and Schuster, London.

2. Ledingham I. & Ramsay G. (1986) Hypovolaemic shock:British Journal ofAnaesthesiology 58, 169-189.

3. Wiggers C.J. (1950) Physiology of Shock. TheCommonwealth Fund, New York.

4. Shires T. (1988) Clinical Critical Care Medicine 15,91-1 01.

5. Velanovich V. (1989) Crystalloid versus colloid fluidresuscitation: a meta-analysis of mortality. Surgery105, 65-71.

6. Skinner D., Driscoll P. & Earlam R. (1991) ABC ofmajor trauma. British Medical Journal ix.

7. American College of Surgeons (1988) AdvancedTrauma Life Support Program, Instructor Manual.

8. Kowalenco T., Stern S., Dronen S. & Xu Wang. (1992)Improved outcome with hypotensive resuscitationof uncontrolled hemorrhagic shock in a swine model.Journal of Trauma, 33, 349-353.

9. Falk J.L., O'Brein J. & Kerr R. (1992) Fluidresuscitation in traumatic haemorrhagic shock.Critical Care Clinics 8, 323-340.

10. Takaori M. & Safar P. (1967) Treatment of massivehaemorrhage with colloid or crystalloid solutions.Studies in dogs. Journal of the American MedicalAssociation 199, 297-302.

11. Diprose P. & Sleet R.A. (1993) How well do weresuscitate patients with haemorrhagic shock?Archives of Emergency Medicine 10, 135-137.

12. Cannon W.B., Fraser J. & Cowell E.M. (1918) Thepreventive treatment of wound shock. Journal of theAmerican Medical Association, 70, 618-621.

13. Bickell W.H., Shaftan G.W. & Mattox K.L. (1989)Intravenous fluid administration and uncontrolledhaemorrhage. Journal of Trauma 29, 409.

14. Shaftan G.W., Chu-Jeng C., Dennis C. & Harris B.(1965) Fundamentals of physiologic control of arterialhaemorrhage. Surgery 58, 851-856.

15. Stern S.A., Dronen S.C., Birrer P. & Xu Wang. (1993)Effect of blood pressure on haemorrhage volumeand survival in a near fatal haemorrhage modelincorporating a vascular injury. Annals ofEmergencyMedicine 22, 155-163.

16. Crawford S. (1991) Ruptured abdominal aorticaneurysm: an editorial. Journal of Vascular Surgery13, 348-350.171

C 1995 Blackwell Science Ltd, Journal ofAccident and Emergency Medicine 12, 165-172


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

R. Cutress 17. Westaby S. (1994) Resuscitation in thoracic trauma.British Journal of Surgery 81, 929-931.

18. Dronen S.C., Stern S., Baldurrson J., Irvin C. &Syverud S. (1992) Improved outcome with earlyblood administration in a near fatal model of porcinehaemorrhagic shock. American Journal ofEmergency Medicine 10, 533-537.

19. Pollack C.V. (1993) Prehospital fluid resuscitation ofthe trauma patient. Emergency Medicine Clinics ofNorth America. 11, 61-70.

20. Milikan J.C., Cain T.L. & Hansbrough J. (1984) Rapidvolume replacement for hypovolaemic shock. Acomparison of techniques and equipment. Journalof Trauma. 24, 428-431.

21. Jacobs L.M., Sinclair A., Betser A. & D' Agostino R.B.(1984) Prehospital advanced life support: benefitsin trauma. Journal of Trauma 24, 8-13.

22. Potter D., Goldstein G., Fung S. C. & Selig M. (1988)A controlled trial of prehospital advanced life supportin trauma. Annals of Emergency Medicine 17, 582-588.

23. Reines H.D., Bartlett R.L., Chundy N.E., Kiragu K. R.

& McKnew M.A. (1988) Is advanced life supportsuitable for victims of motor vehicle accidents?Journal of Trauma 28, 563-570.

24. Smith J.P., Bodai B.l., Hill A.S. & Frey C.F. (1985)Prehospital stabilisation of critically injured patients:a failed concept. Journal of Trauma 25, 65-70.

25. Lewis F.R. (1986) Prehospital intravenous fluidtherapy: physiologic computer modelling. Journalof Trauma 26, 804-811.

26. Kaweksi S.M., Sise M.J. & Virgilio R.W. (1990) Theeffect of prehospital fluids on survival in traumapatients. Journal of Trauma 30,1215-1218.

27. Martin R.R., Bickell W.H., Pepe P.E., Burch T.M. &Mattox K. L. (1992) Prospective evaluation ofpreoperative fluid resuscitation in hypotensivepatients with penetrating truncal injury: a preliminaryreport. Journal of Trauma 33, 354-362.

28. Berthoud M.C. & Reilly C.S. (1992) Adverse effectsof general anaesthetics. Drug Safety, 7, 434-459.

29. Heyworth J. (1992) Conjunctival oxygen monitoringin the initial assessment of trauma. Archives ofEmergency Medicine 9, 274-279.

172

C 1995 Blackwell Science Ltd, Journal of Accident and Emergency Medicine 12, 165-172


http://emj.bm

j.com/

J Accid E

merg M


eptember 1995. D

ownloaded from

http://emj.bmj.com/

fluid resuscitation traumatic haemorrhage

Documents