British Journal of Anaesthesia 1996; 77: 128–132
Pre-emptive analgesia with NSAID—what
does it achieve?
Sir,—We wish to comment on the study of Espinet and colleagues
 who investigated the combined effect of thoracic extradural
local anaesthetic block and administration of diclofenac either
30 min before or 30 min after skin incision on post-hysterectomy
pain and analgesic consumption. They found that the timing of
analgesia made no difference to pain scores or postoperative
analgesic requirements, that is there was no pre-emptive effect of
diclofenac and extradural local anaesthesia. They discussed these
findings in relation to the known actions of non-steroidal anti-
inflammatory drugs (NSAID) in inhibiting peripheral sensitiza-
tion of the primary afferent nerve terminal and so reducing
afferent barrage from the periphery. This afferent barrage
contributes to central sensitization in the spinal cord and the
degree of central sensitization is said to relate to subsequent pain
experience and analgesia . The authors comment that the lack
of pre-emptive analgesia in their study is at variance with that seen
in animal studies using similar paradigms.
While their study was designed carefully to exclude the
confounding effects of systemic opioids, nitrous oxide and local
anaesthetic test doses, the lack of pre-emptive analgesic effect
reported in this study may be explained by central actions of
NSAID. In addition, not all types of injury are sensitive to pre-
emptive treatment, and under certain circumstances central local
anaesthetic block may not prevent the subsequent development
of hyperalgesia. These possibilities are discussed below.
It is now well established that prostaglandins contribute not
only to peripheral but also to central sensitization, and that
NSAID act at central and peripheral sites . In experimental
models of central sensitization, intrathecal administration of
NSAID results in antinociception [4, 5], and critically, their
central action is equally effective whether administered before or
after the noxious stimulus. This is because they act on intracellular
messengers responsible for the maintenance of the persistent
nociceptive state in the spinal cord . In humans, systemically
administered diclofenac crosses the blood–brain barrier and CSF
concentrations may continue to increase for 12 h after a single
dose . Therefore, in the study of Espinet and colleagues , a
central spinal action of diclofenac, reducing central sensitization
with equal efficacy either before or after incision, could potentially
have obscured any pre-emptive action of extradural bupivacaine.
A recent study using pre-emptive vs post-injury ketorolac did
show a small initial opioid sparing effect. However, ketorolac may
cross the blood–brain barrier to only a minimal extent and so
inhibition of peripheral afferent discharge may not have been
obscured by the central actions of the NSAID in this study .
Animal studies also show that pre-emptive analgesia may not be
effective in all types of injury; Dougherty, Garrison and Carlton
 have reported that pre-emptive topical administration of
lignocaine to the sciatic nerve could ameliorate the development of
hyperalgesia in a loose ligation model of nerve injury (a model that
shows a marked response to a number of pre-emptive analgesic
techniques) . However, Dougherty, Garrison and Carlton 
also reported that lignocaine showed no pre-emptive effect in a
partial transection injury of the sciatic nerve. If different types of
injury are differentially sensitive to pre-emptive local anaesthesia,
the implications are that some aspects of surgical trauma may not
be sensitive to pre-emptive local anaesthetic block. Therefore,
differences in operation type and operative technique may
contribute to the different outcomes in clinical studies of pre-
emptive local anaesthesia.
Furthermore, the route of pre-emptive local anaesthetic admin-
istration may influence the development of post-injury hyper-
algesia. For example, Tverskoy and colleagues  reported that
spinal anaesthesia with bupivacaine was less effective than local
infiltration of bupivacaine in reducing late postoperative wound
pain after hernia repair. This is a surprising finding in the light of
the common perception among anaesthesists that spinal an-
aesthesia is more reliable and complete than local infiltration. One
initial suggestion given to explain these findings was that topical
local anaesthesia may have had an anti-inflammatory effect at the
site of the wound . Alternatively, Luo and Wiesenfeld-Hallin
have recently reported  that while intrathecal local anaesthetic
initially produces profound block of sensory transmission, as the
local anaesthetic wears off, a paradoxical state of spinal hyper-
excitability ensues. This may completely mask any previous
benefit. More importantly, no subsequent long-term behavioural
benefit could be shown from pre-emptive intrathecal lignocaine,
even compared with animals which received no intrathecal
lignocaine (although in contrast, topical local anaesthetic to the
sciatic nerve has previously been shown to be beneficial with this
same nerve injury model ). The authors also pointed out that
the marked clinical benefit shown by Bach, Noreng and Tjellen in
their well known study  in amputees was achieved using a
combination of extradural local anaesthetic and opioid.
Pre-emptive analgesia as a concept has essentially been assumed
to relate to inhibition of the afferent barrage, especially the initial
“injury discharge” associated with the primary event such as
nerve injury. As has been emphasized previously , the injury
discharge may outlast the treatment and so obscure the benefit
implied by some animal studies. Furthermore the following
factors: (i) giving drugs that act not only on injury discharge but
also on the substrate of the central sensitized state, such as
NSAID; (ii) the differential sensitivity, of different types of nerve
injury to pre-emptive analgesia; and (iii) the recent experimental
evidence that intrathecal local anaesthetic alone may have little
benefit in terms of reducing subsequent central sensitization; all
help to explain some of the disappointing findings in clinical
studies of pre-emptive analgesia but conversely these same
observations will help in the design of better regimens to prevent
University Department of Anaesthesia
1. Espinet A, Henderson DJ, Faccenda KA, Morrison LMM.
Does pre-incisional thoracic extradural block combined with
diclofenac reduce postoperative pain after abdominal hys-
terectomy? British Journal of Anaesthesia 1996; 76: 209–213.
2. Munglani R, Hunt S, Jones JG. Spinal cord and chronic pain.
In: Kaufmann L, ed. Anaesthesia Review. Edinburgh:
Churchill Livingstone, 1996; 53–76.
3. McCormack K. Non-steroidal anti-inflammatory drugs and
spinal nociceptive processing. Pain 1994; 59: 9–43.
4. Malmberg AB, Yaksh TL. Hyperalgesia mediated by spinal
glutamate or substance P receptor blocked by spinal cyclo-
oxygenase inhibition. Science 1992; 257: 1276–1279.
5. Malmberg AB, Yaksh TL. Antinociceptive actions of spinal
nonsteroidal anti-inflammatory agents on the formalin test in
the rat. Journal of Pharmacology and Experimental Thera-
peutics 1992; 263: 136–146.
6. Coderre TJ, Yashpal K. Intracellular messengers contribu-
ting to persistent nociception and hyperalgesia induced by L-
glutamate and substance P in the rat formalin pain model.
European Journal of Neuroscience 1994; 6: 1328–1334.
7. Zecca L, Ferrario P, Costi P. Determination of diclofenac and
its metabolites in plasma and cerebrospinal fluid by high-
performance liquid chromatography with electrochemical
detection. Journal of Chromatography 1991; 567: 425–432.
8. Fletcher D, Zetlaoui P, Monin S, Bombart M, Samii K.
Influence of timing on the analgesic effect of intravenous
ketorolac after orthopaedic surgery. Pain 1995; 61: 291–297.
9. Dougherty PM, Garrison CJ, Carlton SM. Differential
influence of local anaesthetic upon two models of exper-
imentally induced peripheral mononeuropathy in the rat.
Brain Research 1992; 570: 109–115.
10. Munglani R, Bond A, Smith G, Harrison S, Elliot PJ, Birch
PJ, Hunt SP. Changes in neuronal markers in a mononeuro-
pathic rat model: relationship between neuropeptide Y, pre-
emptive drug treatment and long term mechanical hyper-
algesia. Pain 1995; 63: 21–31.
11. Tverskoy M, Cozacov C, Ayache M, Bradley EL jr, Kissin I.
Postoperative pain after inguinal herniorrhaphy with different
types of anesthesia. Anesthesia and Analgesia 1990; 70: 29–35.
12. Woolf CJ, Chong M. Pre-emptive analgesia-treating post-
operative pain by preventing the establishment of central
sensitization. Anesthesia and Analgesia 1993; 77: 362–379.
13. Luo L, Wiesefeld-Hallin Z. Effects of intrathecal local
anaesthetics on spinal excitability and on the development of
autotomy. Pain 1995; 63: 173–179.
14. Gonzalez-Darder J, Barbera J, Abellan M. Effects of prior
anaesthesia on autotomy following sciatic nerve transection in
rats. Pain 1986; 24: 87–91.
15. Bach S, Noreng MF, Tjellen NU. Phantom limb pain in
amputees during the first 12 months following limb am-
putation after preoperative lumbar epidural blockade. Pain
1988; 33: 297–301.
16. Munglani R, Jones JG, Hunt S. Pre-emptive analgesia—use
of immediate early genes expression as markers of neuronal
stimulation. British Journal of Anaesthesia 1993; 71: 458.
Sir,—Thank you for the opportunity to reply to Hudspith and
Munglani. On the whole we concur with their comments. The
complexity of pain production and the number of mechanisms
involved make it difficult to block or reduce the nociceptive input.
This almost certainly accounts for the lack of a useful pre-emptive
analgesic effect in the clinical setting. It is unlikely that one single
analgesic method will be sufficient and we should look towards a
multi-modal approach to pain relief/reduction. In a small part
this is what we tried to achieve in our study. Further research into
this field is needed.
West Lothian NHS Trust
St John’s Hospital at Howden
Cytokine balance and immunosuppressive
changes at cardiac surgery
Sir—The study of McBride and colleagues contains much useful
information on the immunological changes associated with surgery
, but omitted several important aspects. We would welcome the
authors’ comments on the following points.
The authors mention briefly that beta adrenoceptor blocking
drugs may alter the pro-inflammatory cytokine response, but the
possible effect of other cardiac drugs is not considered. Could
calcium channel antagonists also have changed the cytokine
response? It has been shown in an animal model that admin-
istration of calcium antagonists, either before or after endotoxin
administration, exerted a protective effect and decreased mortality
[2, 3]. Some cytokines exert their effects by increasing free
intracellular calcium, especially IL-4, which is one of the principal
cytokines in TH2 lymphocyte development . A decrease in
intracellular calcium concentration after calcium channel block
may decrease amplification of the initial cytokine signal by
interfering with the IP3–calcium–calmodulin pathway. In con-
trast, it has recently been shown that diltiazem enhanced
production of cytokines and maintained immune function after
haemorrhage . It seems that cations, and Ca2? in particular,
may have important intracellular influences on immune function,
and therefore any comparisons between McBride’s in vivo and in
vitro cytokine changes must be interpreted with great caution.
In describing the activation of TH2 cells and the increase in IL-
10 as key events in an anti-inflammatory response to cardiac
surgery, the authors failed to mention that the principal
immunological fault after major surgery and trauma is impaired
production of TH1 cytokines (interferon gamma, IL-2). These
cytokines are crucial for the development of TH1 lymphocytes,
which are largely responsible for cell-mediated immunity [6, 7].
Development of TH2 lymphocytes is dependent on the early
production of IL-4 (TH2 cytokine) in conjunction with down-
regulation of interferon gamma and IL-2 [6, 8]. Whether TH2
lymphocyte activity is favoured when TH1 lymphocyte pro-
duction is suppressed is unclear, but no increase in IL-2 or IL-4
was found during or after cardiopulmonary bypass .
The statement that oxpentifylline induces TH2 activity is
misleading . This phosphodiesterase inhibitor increases
intracellular cAMP concentration which stimulates PGE2 syn-
thesis, resulting in impaired IL-2 synthesis and abnormal TH1
development. Because IL-4 synthesis is not affected, TH2 activity
is unaffected. Therefore, preferential TH2 development occurs as
a result of inhibition of TH1 lymphocytes . Indeed, failure of
IL-2 production is considered to be a major factor in the impaired
cell-mediated responses after surgery and trauma [11, 12].
It is clear that there are many factors involved in the
immunosuppression after cardiac surgery and the changes in IL-
10, IL-1ra and sTNF alpha observed by the authors only play a
G. M. HALL
Department of Anaesthesia
St George’s Hospital Medical School
1. McBride WT, Armstrong MA, Crockard AD, McMurray
TJ, Rea JM. Cytokine balance and immunosuppressive
changes at cardiac surgery: contrasting response between
patients and isolated CPB circuits. British Journal of Anaes-
thesia 1995; 75: 724–733.
2. Lee HC, Lum BKB. Protective action of calcium entry
blockers in endotoxin shock. Circulatory Shock 1986; 18:
3. Bosson S, Kuenzig M, Schwartz SI. Increased survival with
calcium antagonists in antibiotic-treated bacteremia. Circu-
latory Shock 1986; 19: 69–74.
4. Clemens MJ. How do cytokines work? 3.2.5 Inositol
phosphates and control of intracellular calcium. In: Clemens
MJ, ed. Cytokines. Oxford: Bios Scientific Publishers, 1991;
5. Meldrum DR, Ayala A, Chaudry IH. Mechanism of dil-
tiazem’s immunomodulatory effects after hemorrhage and
resuscitation. American Journal of Physiology 1993; 265:
6. Faist E, Kupper TS, Baker CC, Chaudry IH, Dwyer J, Baue
AE. Depression of cellular immunity after major injury.
Archives of Surgery 1986; 121: 1000–1005.
7. Faist E, Markewitz A, Schinkel C, Zimmer S, Ertel W.
Trauma-induced alterations of cytokine synthesis patterns:
mechanisms and modulatory interventions. In: Oppenheim
JJ, Rossio JL, Gearing AJH, eds. Clinical Applications of
Cytokines. New York: Oxford University Press, 1993;
8. O’Riordan M, Mannick JA, Rodrick ML. Cytokine response
to surgery, trauma, and burns. In: Oppenheim JJ, Rossio JL,
Gearing AJH, eds Clinical Applications of Cytokines. New
York: Oxford University Press, 1993; 301–305.
9. Steinberg JB, Kapienski DP, Olson JD, Weller JM. Cytokine
and complement levels in patients undergoing cardiopulmon-
ary bypass. Journal of Thoracic and Cardiovascular Surgery
1993; 106: 1008–1016.
10. Rott O, Cash E, Fleischer B. Phosphodiesterase inhibitor
pentoxifylline, a selective suppressor of T helper type-1 but
not type-2 associated lymphokine production, prevents
induction of experimental autoimmune encephalomyelitis in
Lewis rats. European Journal of Immunology 1993; 23:
11. Chouaib S, Chatenoud L, Klatzman D, Fradelizi D. The
mechanisms of inhibition of human IL-2 production. II.
PGE2 induction of suppressor lymphocyte. Journal of Im-
munology 1984; 132: 1851–1857.
12. Aussel MD, Ferrue C, Fehlmann B. Regulation of inter-
leukin-2 synthesis by cAMP in human T-cells. Journal of
Immunology 1987; 139: 1179–1184.
Sir,—Sheeran and Hall sought to highlight the potential immuno-
modulatory effects of calcium antagonists at cardiac surgery.
Their comments suggest that calcium antagonists may have
influenced the cytokine response.
We have since investigated the balance of pro- and anti-
inflammatory cytokines in plasma and bronchoalveolar lavage of
children undergoing surgery for congenital heart disease. None of
these children received preoperative calcium antagonists. Never-
theless, their cytokine response was similar to that observed in
adults, with a phased anti-inflammatory response developing,
British Journal of Anaesthesia
beginning with IL-10, 10 min after aortic cross-clamp release and
followed by IL-1ra and the TNF soluble receptors over the next
24 h. In the absence of calcium antagonists these children were
able to mount a phased anti-inflammatory response . The
existence of a phased anti-inflammatory cytokine response cannot
therefore be attributed to the presence of calcium antagonists.
It is not yet known what factors if any influence the magnitude
of the phased anti-inflammatory cytokine response and what
clinical significance variations in magnitude of this phased anti-
inflammatory cytokine response may have. Our group is actively
investigating this, and it is indeed possible that calcium antagon-
ists (including volatile anaesthetic agents) may influence the
magnitude of this response.
The cytokine response in patients and the isolated CPB system
differed in that there was no increase in anti-inflammatory
cytokines in vitro. . Because of the known effects of Ca2? on
immune function, Sheeran and Hall advise “great caution” in
comparison of the cytokine response within patients and the
isolated CPB circuits. Cautious comparison of the two systems
indicates that the major difference between patients and the
isolated CPB circuits is not the absence of calcium antagonists
but rather the fact that within the isolated CPB circuit blood does
not have an opportunity to escape from the activating non-
endothelialized surface of the CPB apparatus and enter the
vasculature of the patient. We believe that this is vital to any
interpretation of these differences.
Sheeran and Hall described impaired production of TH1
cytokines as “the principal immunological fault” at major surgery
and claimed that we failed to mention in our discussion such an
“immunological fault”. There is nothing new in describing the
immune response at cardiac surgery as “being at fault”. For years
authors have concentrated on the pro-inflammatory response and
immunosuppression at cardiac surgery as two aspects of dis-
ordered immunological homeostasis. Sheeran and Hall have
simply gone further than most in speculating on the “principal”
of these so called “immunological faults”.
We recommend a fresh approach to investigating immuno-
logical change at cardiac surgery. We believe that an alternative
and helpful way to investigate the immune changes at elective
cardiac surgery is to view the immune changes therein as a model
of a successful and indeed appropriate response. Most patients for
elective coronary artery bypass grafting (CABG) surgery now go
home 5 or 6 days later. Surely this is an example of outstanding
immunological success! Let clinical outcome be the arbiter as to
which aspects of the immune response at cardiac surgery are at
The cardiac surgical patient with good clinical outcome presents
an excellent model for study of a successful immune response. We
have been the first to highlight the phased anti-inflammatory
response as a possible contributor to this success. Other workers
have later described an IL-10 response after cardiac surgery .
We have never suggested that the phased anti-inflammatory
response which we have described is limited to IL-10, IL-1ra and
TNF soluble receptors. In fact it is likely that IL-4, TGF-? and
indeed other as yet undiscovered anti-inflammatory cytokines
may play an important role in the success of our patients.
Only when the ingredients of a successful immune response
such as occurs at cardiac surgery are understood shall we
understand when the immune response is “at fault”. We believe
that it is important to know when the immune response is at fault.
This information will only come through comparing the profiles
of the immune response of cardiac patients who do well with those
who (rarely) go on to develop the systemic inflammatory response
syndrome (SIRS) or indeed infective complications such as
mediastinitis. This knowledge may then enable us to devise
effective strategies in the prevention and early treatment of SIRS
in the context of major trauma and non-cardiac surgery. We may
even be able to predict those patients at risk of developing severe
Sheeran and Hall pointed out that our statement that oxpenti-
fylline induces TH2 activity is misleading. We thank them for this
and highlighting that the preferential TH2 development after
oxpentifylline treatment is an indirect result of inhibition of TH1
development by the drug.
Immunomodulation in the treatment of SIRS in the ICU has
been largely unsuccessful. An air of despondency has descended
upon the literature . This may be because we have been too
quick to modulate the immune response without fully under-
standing what is an appropriate response to trauma and sepsis.
Perhaps it is time to re-evaluate the cardiac surgical patient as a
model of success and then re-design our immunomodulatory
W. T. MCBRIDE
Duke University Medical Center
Durham, NC, USA
M. A. ARMSTRONG
Department of Microbiology and Immunobiology
Queen’s University of Belfast
T. J. MCMURRAY
Royal Victoria Hospital
1. McBride WT, Armstrong MA, Gilliland H, McMurray TJ.
The balance of pro and anti-inflammatory cytokines in plasma
and broncho-alveolar lavage at paediatric cardiac surgery.
Cytokine 1996 (in press).
2. McBride WT, Armstrong MA, Crockard AD, McMurray TJ,
Rea JM. Cytokine balance and immunosuppressive changes at
cardiac surgery: contrasting response between patients and
isolated CPB circuits. British Journal of Anaesthesia 1995; 75:
3. Wan S, Marchant A, DeSmet J-M, Antoine M, Zhang H,
Vachiery J-L, Goldman M, Vincent J-L, LeClerc J-L. Human
cytokine responses to cardiac transplantation and coronary
artery bypass grafting. Journal of Thoracic and Cardiovascular
Surgery 1996; 111: 469–477.
4. Petros AJ, Marshall JC, Van Saene HKF. Should morbidity
replace mortality as an endpoint for clinical trials in intensive
care? Lancet 1995; 345: 369–371.
Perioperative changes in ? ? ? ?1-acid glycoprotein
Sir,—Booker, Taylor and Saba  suggested that preoperative
measurement of ?1-acid glycoprotein (AAG) concentration should
be performed in all infants undergoing major elective surgery who
are expected to receive prolonged bupivacaine infusions in the
postoperative period. This is based on concern that bupivacaine
toxicity may be related to low plasma concentrations of the drug
binding protein. Unfortunately, this concern is, in turn, based on
several dubious premises: (1) that plasma protein binding of drugs
may have significant pharmacodynamic implications; (2) that
changes in protein binding are important clinically for drugs
which are highly bound, such as bupivacaine; and (3) that if
variations occur in plasma concentrations of AAG, then free
plasma concentrations of the drug can vary considerably,
whereas the total concentration of the drug in plasma is only
slightly affected. To explain these misconceptions is a long
story—I refer the authors to a few recent publications [2, 3].
Premise (3) is the incorrect way round; if the fraction of
bupivacaine bound in plasma changes as a result of a change in
AAG concentration, classical pharmacokinetic theory predicts
that, in the case of a low hepatic extraction drug such as
bupivacaine, total plasma drug concentration varies but the
unbound drug concentration does not. As it is the latter which
determines pharmacological effect at steady state, there would be
no clinical consequence of a change in AAG concentration.
Therefore, on this basis, measurement of this protein would be a
waste of time and resources. One caveat applies to this conclusion,
if AAG concentration decreases in proportion to the degree of
hepatic dysfunction, then its measurement may be an indirect
predictor of the ability of the liver to clear free bupivacaine by
metabolism. However, such an association would be independent
of the degree of plasma drug binding.
G. T. TUCKER
Department of Medicine and Pharmacology
University of Sheffield
Royal Hallamshire Hospital, Sheffield
1. Booker PD, Taylor C, Saba G. Perioperative changes in ?1-
acid glycoprotein concentrations in infants undergoing major
surgery. British Journal of Anaesthesia 1996; 76: 365–68.
2. Tucker GT. Safety in numbers. The role of pharmacokinetics
in local anesthetic toxicity. Regional Anesthesia 1994; 19:
3. Rolan PE. Plasma protein binding displacement interactions—
why are they still regarded as clinically important? British
Journal of Clinical Pharmacology 1994; 37: 125–128.
CBF in adults using near infrared
spectroscopy (NIRS): potential for bedside
Sir,—We read with interest the article by Owen-Reece and
colleagues  and suggest that the findings on correct scaling of
NIRS cerebral tissue measurements warrant further discussion.
Comparison between scalp and dural recordings in patients
allowed conclusions to be made regarding the percentage of the
illuminated tissue volume which was cerebral. Measurement of
cerebral blood flow (CBF) with the optodes on the dura compared
well with values obtained by invasive techniques, but CBF was
underestimated by a factor of 3 with the optodes positioned on the
scalp. This can be explained by the cerebral tissues contributing
only 30–40 % to the total pathlength. Scalp and skull have been
shown to have low blood flow and act as static “deadspace” .
The chromophore concentration changes detected by NIRS are
mostly cerebral, but they must be seen and expressed as
concentration changes occurring in the total illuminated tissue
volume, 60–70 % of which is extracerebral in the adult. In a rat
model  and in neonates  it has been postulated that the
extracerebral component is less than 20 % of the total optical
pathlength and this has been accepted as a minor limitation in
these measurements. Previous studies by Elwell and co-workers in
adults have alluded to the problem of the relatively small
component that the cerebral tissue contributes to the total optical
pathlength, but none the less they achieved results for cerebral
blood flow comparable with invasive techniques despite using
scalp recording . The small cerebral component of the total
interrogated volume has major implications for the clinical use of
such a system as a small alteration in the cerebral tissue volume
has a large effect on the proportions of cerebral and extracerebral
tissue in the total illuminated tissue volume and thereby has a
major effect on measurements of changes in chromophore
concentration. The proportion of cerebral tissue may vary between
subjects or even within a subject because of changes in
optode/scalp coupling or pathological changes, for example in
cerebral oedema when the proportion of brain in the illuminated
tissue volume may increase as the brain is forced against the skull.
In these circumstances, is the instrument measuring quantified
changes in chromophore concentrations occurring in the brain?
It is not specified that figure 1 is a dural recording, but using the
values shown to make an estimate of cerebral blood volume this is
likely to be the case. If this is a dural recording, the same re-
saturation event performed with the optodes placed on the scalp
would cause a corresponding change in [HbO2] over the initial 4 s
of only 0.6 ?mol litre?1. This value would be of a similar
magnitude to the baseline variability of the trace.
Desaturation to 95 % is unlikely to prove harmful and has been
used in neonatal intensive care practice. However, many patients
in whom measurement of CBF would be particularly useful are
likely to be at risk from secondary brain injury, for example head-
injured patients. We have reservations concerning the clinical
acceptance of a desaturation method in this population.
The potential of NIRS to measure cerebral blood flow remains
an exciting possibility, but we would question whether a
desaturation/resaturation technique in combination with a scalp
recording will ever be satisfactory for this purpose in adults.
T. S. WALSH
Department of Anaesthetics
1. Owen-Reece H, Elwell CE, Harkness W, Goldstone J, Delpy
DT, Wyatt JS, Smith M. Use of near infrared spectroscopy to
estimate cerebral blood flow in conscious and anaesthetized
adult subjects. British Journal of Anaesthesia 1996; 76: 43–48.
2. Elwell CE, Cope M, Edwards AD, Wyatt JS, Delpy DT,
Reynolds EOR. Quantification of adult cerebral hemodyna-
mics by near-infrared spectroscopy. Journal of Applied Physi-
ology 1994; 77: 2753–2760.
3. Delpy DT, Arridge SR, Cope M, Edwards D, Reynolds EOR,
Richardson CE, Wray S, Wyatt J, van der Zee P. Quantitation
of pathlength in optical spectroscopy. Advances in Experimental
Medicine and Biology 1989; 248: 41–46.
4. Wyatt JS, Cope M, Delpy DT, Richardson CE, Edwards AD,
Wray S, Reynolds EOR. Quantification of cerebral blood
volume in human infants by near-infrared spectroscopy.
Journal of Applied Physiology 1990; 68: 1086–1091.
Sir,—It is true that we have quoted higher values for cerebral
blood flow (CBF) elsewhere . These were obtained with a
different flow analysis method which is described more fully in
that article. Briefly, accumulation of tracer in the brain is biphasic,
with a fast and slow component. Thus in that study, CBF was
59 ml 100 g?1 min?1 in the first 1.5 s with a second peak of 11 ml
100 g?1 min?1 at 1.5–3 s. These values are similar to those
measured by PET.
We decided not to use this type of analysis for our study  for
two reasons. First, high and low CBF components are not widely
used clinically (and are not, in any case, provided by xenon
washout and the Kety–Schmidt method). Second, such treatment
may confuse a simple “before and after” comparison. We
therefore used a single average value calculated over the first 4 s of
tracer arrival which is compared more easily with mean CBF
measured by other techniques.
We believe that NIRS is able to quantify changes in brain
chromophore concentration but these changes must be observed
through the “deadspace” of the non-cerebral tissues. It is true
that at present we have difficulty in making allowances for this
when subsequently handling the data. However, recent computer
modelling shows that the cerebral sulci have little effect on the
proportion of cerebral tissue illuminated and also that when the
optodes are pressed firmly against the scalp, skin blood flow has
negligible influence [Delpy D. T., personal communication]. As
stated in our article, it is necessary to determine the contribution
of extracerebral tissue to the total optical pathlength before
further progress is made. Similar modelling work, which we hope
will assist in quantifying the effects of extracerebral tissue on near
infrared spectroscopic measurements, is also in progress at UCL
Department of Medical Physics and Bioengineering.
The authors are correct in interpreting figure 1 as a dural
recording and we apologize that this was not made clear. As they
imply, the signal-to-noise ratio is a significant problem in
measurements made via the scalp and more scalp measurements
were rejected than those from dura.
Desaturation–resaturation is unlikely to be harmful for the
reasons outlined in the article. However, we agree that it may be
preferable to avoid the technique in brain-injured patients. There
is a range of cerebral haemodynamic variables which can be
observed using NIRS which do not affect arterial oxygen
saturation, for example cerebral blood flow measurement using
indocyanine green as a tracer, spontaneous cerebral blood volume
changes, cerebral vascular response to altered
In view of the progress being made in dealing with the scaling
problems of scalp recording and the scope for other cerebral
measurements which can be made non-invasively with NIRS, we
agree with Hopton, Walsh and Lee that the potential of the
technique is exciting. There is a great deal of work still to be done
before it is suitable for widespread clinical use.
Departments of Anaesthesia, Medical Physics and Paediatrics
Departments of Anaesthesia and Surgery
National Hospital for Neurology and Neurosurgery
Queen Square, London
1. Elwell CE, Cope M, Edwards ADE, Wyatt JS, Delpy DT,
Reynolds EOR. Quantification of adult cerebral haemo-
dynamics by near infrared spectroscopy. Journal of Applied
Physiology 1994; 77: 2753–2760.
2. Owen-Reece H, Elwell CE, Harkness W, Goldstone J, Delpy
DT, Wyatt JS, Smith M. Use of near infrared spectroscopy to
estimate cerebral blood flow in conscious and anaesthetized
adult subjects. British Journal of Anaesthesia 1996; 76: 43–48.
British Journal of Anaesthesia
Hypercoagulability induced by crystalloids
Sir,—The in vitro study by Ruttmann, James and Viljoen  in
which dilution of blood with saline and Haemaccel was shown to
render it hypercoagulable, is fascinating reading. Their results not
only challenge intuitive assumptions but also have possible
implications for our clinical practice and, perhaps more im-
portantly, call into question the methodology of studies designed
to assess the effect of regional anaesthesia on the development of
postoperative thromboembolic disease (TED).
The authors cite the in vivo study of Janvrin, Davies and
Greenhalgh  in which patients given crystalloid in the
perioperative period (“wet” group) had a higher incidence of
deep vein thrombosis (DVT) than those who received none
(“dry” group). While this is of course no criticism of Ruttmann,
James and Viljoen, it is worth pointing out that this frequently
cited study involved only 60 patients and its power was in the
region of 0.65 .
However, the results of the in vitro work of Ruttmann, James
and Viljoen are compelling, and notwithstanding the above
criticism, if the in vivo effect is real then this may present a case
for the use of the hydroxyethyl starches (HES) where there is a
risk of TED, and where plasma volume expansion is required.
This colloid, rather than promoting coagulation, prevents it by
inducing a mild and transient type-I von Willebrand-like
Over the years, several studies have demonstrated to varying
degrees the effect of extradural and spinal anaesthesia on the
reduction in postoperative thromboembolic disease compared
with general anaesthesia. The proposed mechanisms whereby this
is effected include reduced blood loss and transfusion require-
ments, reduced stress response and its concomitant procoagulant
effect, improved lower extremity blood flow, and reduction in
plasminogen activator inhibitor-1 (PAI-1) in those receiving
regional anaesthesia [5–7]. In none of these studies was admin-
istration of crystalloid and colloid considered to exert any specific
effect on coagulation. Only administration of blood was standard-
ized between the groups. In most reports, the design of the study
ensured that haemodynamic variables were controlled tightly and
this usually implied that those receiving extradural or spinal
anaesthesia were preloaded with up to 15 ml kg?1 of lactated
Ringer’s solution and required ongoing plasma volume expansion
with this and similar solutions throughout the case and into the
postoperative period to achieve “target” haemodynamic values
(although the actual volumes were not recorded).
Again, if the procoagulant effect of crystalloid haemodilution is
real then it may to some extent mask the beneficial effects of
regional anaesthesia if crystalloid is used to preload these patients,
and this may provide support for the argument that normotension
should be maintained either with vasopressors or HES. Clearly,
this work needs to be re-evaluated in the light of any new
understanding of the effect of crystalloid haemodilution.
A. D. FARMERY
Department of Anaesthetics
1. Ruttmann TG, James MFM, Viljoen JF. Haemodilution
induces a hypercoagulable state. British Journal of Anaesthesia
1996; 76: 412–414.
2. Janvrin SB, Davis G, Greenhalgh RM. Postoperative deep vein
thrombosis caused by intravenous fluids during surgery.
British Journal of Surgery 1980; 67: 690–693.
3. Altman D. Practical Statistics for Medical Research. London:
Chapman and Hall, 1991.
4. Strauss RG. Review of the effects of hydroxyethyl starch on
the blood coagulation system. Transfusion 1981; 21: 299–302.
5. Modig J, Borg T, Karlström G, Maripuu E, Sahlstedt B.
Thromboembolism after total hip replacement: Role of
epidural and general anesthesia. Anesthesia and Analgesia
1983; 62: 174–180.
6. Rosenfeld BA, Beattie C, Christopherson R, Norris EJ, Frank
SM, Breslow MJ, Rock P, Parker SD, Gottlieb SO, Perler BA,
Melville Williams G, Seidler A, Bell W. The effect of different
anesthetic regimens on fibrinolysis and the development of
postoperative arterial thrombosis. Anesthesiology 1993; 79:
7. Davis FM, Laurenson VG, Gillespie WJ, Wells JE, Foate J,
Newman E. Deep vein thrombosis after total hip replacement.
A comparison between spinal and general anaesthesia. Journal
of Bone and Joint Surgery. British Volume 1989; 71: 181–185.
Sir,—We agree fully with the comments that our results challenge
intuitive assumptions and we agree entirely with the inferences
they draw from these results. Furthermore, as Farmery and Kong
correctly point out, our results may indeed call into question not
only the methodology of studies designed to assess the effect of
regional anaesthesia on the development of postoperative throm-
boembolic disease, but also the methodology of most of the
currently quoted studies designed to assess the effect of any of the
colloids on coagulation, as the majority use a crystalloid control
Whether or not our in vitro results can be translated to the
clinical situation remains to be resolved. However, we are
currently in the process of investigating this effect in human
volunteers and the results appear to be consistent with our in vitro
work. Indeed, using standard tests of coagulation and also
thrombelastography, we have been able to further strengthen our
hypothesis that haemodilution renders blood hypercoaguable.
We can only agree with the comments of Farmery and Kong
that the role of crystalloid solutions needs to be re-evaluated,
especially in those circumstances in which the prevention of
procoagulant states is desirable.
T. G. RUTTMANN
M. F. M. JAMES
J. F. VILJOEN
Department of Anaesthesia
University of Cape Town