Annals ofthe Rheumatic Diseases 1992; 51: 1173-1178

The venous footpump: influence on tissue perfusionand prevention of venous thrombosis

A M N Gardner, R H Fox

Development of the technique of video phlebo-graphy has enabled the anatomy and physiologyof the venous pump in the human foot to bedemonstrated.' This hitherto largely unrecog-nised organ consists of the abnormally largevenae comites of the lateral plantar artery (fig1). The plantar venous pump has a strokevolume of between 20 and 30 ml and is activatedsimply by weight bearing.There are three misconceptions relating to

the venous return from the foot which havebeen disproved by video phlebography. (a) Theconcept of a peripheral venous pump has alwaysembraced the idea of active muscular contrac-tion; however, video phlebography shows thatthis is not true in the foot. Passive weightbearing flattens the plantar arch, stretching theveins longitudinally, causing them to constrict'neck down' and dislocate the blood withsufficient force to overcome the resistance of

a sphygmomanometer cuff inflated to over100 mm around the calf. This alone is adequateto return blood from the foot to the right atriumin the upright position. The foot pump has beenshown to be functional in paraplegic legs.2 Theblood returns in a jerky fashion with the venousvalves acting as haemodynamic ratchets.2 3(b) The belief that the plantar veins drain ontothe dorsum of the foot is a second fallacy; videophlebography shows that the blood is pumped

5-

NIc:

n

5

NI

0

o 30oTime (s)

Erect non-weight bearing

O 3-0Time (s)

Erect weight bearing

.'~~~~~~~~~~~~O

5

NI

0

Torbay Hospital,Torbay, Devon,United KingdomA M N GardnerR H FoxCorrespondence to:Mr A M N Gardner,The Priory, Ipplepen,South Devon TQ12 SRT,United Kingdom

Figure I Phlebogram ofthefoot ofone ofthe authors(AMNG). The venous footpump consists mainly oftheenlarged venae comites ofthe lateral plantar artery. (A) Non-weight bearing, ankle movement had no effect, dorsiflexionofthe toes merely emptied the small precalcaneal plexus ofveins. (B) Weight bearing emptied the venousfoot pump upinto the calf.

0 3*0Time (s)

Erect after weight bearing

Figure 2 Original Dopplerflow recording ofposterior tibialarterialflow at the ankle (RHF), which showed thephenomenon ofpostcompression ivperaemia. (Top) Erectnon-weight bearing. (Middle) Erect weight bearing, noteslight decrease in forwardflow and increase in reverse flow.(Bottom) Erect after lifting footfrom the ground. There was alarge and persistent increase in forvardflow ofover 100%and an absence ofreversedflow. ATL Doppler recording atCardiff.

.. _ ................................................... ._ .1173

%J

on 15 Septem

ber 2018 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.51.10.1173 on 1 O

ctober 1992. Dow

nloaded from

Gardner, Fox

preferentially up into the deep channels of thecalf.2 (c) A third misconception is that bloodfrom the subcutaneous plantar venous plexus(the semelle veineux plantaire of Winkler4) issucked into the deep veins of the foot byspreading of the metatarsals; such aspiration isimpossible because the veins would simplycollapse.The importance of normal venous return in

resolving disease and injury of the lower limbhas prompted the development of a pneumaticdevice that activates the venous foot pump in aphysiological manner and reduces the risk ofvenous thrombosis without resorting to treat-ment with anticoagulant drugs. The implica-tions are therefore profound in the field ofrheumatology and orthopaedics. The device isdescribed at the end of this review.

140-

130-._

E 120-E 110-

O 100-

.X 90-0t 80-

Dc 70-

L 60-

50

Normal subjects

T

T

T

Patientswith PVD

Rest 1 2 3 4 5Time after pump activation (min)

Figure 3 Intermittent pneumatic compression increasespopliteal arterialflow in nornal subjects and in patients withperipheral vascular disease (PVD). ATL Doppler plusspectrum analyser.

In common with other veins in the leg thefoot veins are prone to thrombosis in associationwith thrombosis higher in the limb and inisolation, where it can account for some cases of'plantar fasciitis'. In subjects with marchhaemoglobinuria the plantar veins are theprobable site of haemolysis.2 There is a similarmechanism of venous return in the arm and thisis mirrored in other species.

Blood flow and tissue perfusionHaemodynamic studies of blood flow inducedin the posterior tibial vessels at the ankle byweight bearing have shown that blood flowincreases immediately after the foot is unloaded.Doppler tracing (fig 2) showed that the increasein total blood flow far exceeded the decrease inflow which occurred during weight bearing.This observation was followed by a series ofexperiments by Morgan et al using a pneumaticimpulse device to activate the foot pump in 10patients with vascular disease and 12 normalsubjects.5 The observed hyperaemia caused bythe device was an increase in popliteal bloodflow of 93% (p<0 0001) in normal subjects and84% (p<003%) in patients (fig 3). A placebodevice produced no significant change in flow.Activation of the foot pump did not alter theblood pressure but each impulse decreased theperipheral vascular resistance and increasedblood flow for about 25 seconds. This stilloccurred in ischaemic limbs with arterialobstruction. The observations were made in thefoot down position as the hyperaemic effect wasdependent on venous priming (fig 4).

Increased arterial flow following compressionhas been observed for over a century, initially inthe foot web of frogs6 and latterly in human legsusing xenon-133.' The hyperaemia has untilrecently been assumed to be due to an increase

ADeviceon

I I

Rest 1 2 3 4

Time (min)

Deviceoff

jI I X X

I 15 6 7 8

BDeviceon

Deviceoff

Rest 1 2 3 4 5 6 7 8Time (min)

Figure 4 Effect ofimpulse pumping on common femoral arterialflow in a patient with afemoropopliteal arterial block.(A ) Sitting leg at 45°. (B) Same patient lying supine. The hyperaemic effect ofimpulse compression is dependent onvenous priming.

1000 -

C

E

0._

0

(ra-0-E

0U-

900 -

800 -

700 -

600-

500 -

4U '.

4nnl1w1vv -

1174

on 15 Septem

ber 2018 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.51.10.1173 on 1 O

ctober 1992. Dow

nloaded from

The venous footpump and venous thrombosis

in the arteriovenous pressure gradienexpulsion of venous blood from theconcept of microcirculatory blood Ilonger valid. Arterial inflow and vencare separated by a capillary bed thatcollapsible tubes, with only aboutusually being open at one time.8 Thhigh pressure at one end of a capillFvenous blood out of the other woutenable if the capillaries were rigiccollapsible tubes held fully open,instance Poiseuille's law would appHyperaemia following impulse cc

is not confined to the tissues of the ithe cuff as popliteal blood flow may bIndeed under some circumstancesarterial flow may start during the ccphase (fig 5). The increase appetriggered by the pulsation impartedreturn flow by activation of each c!foot pump. This finding is in keepinexperiments of Lewis and Grant, wishowed hyperaemia following transilocclusion.'0 More recent studies offusion have shown that the rapid iperfusion pressure is mainly responsitemporary decrease in periphera]resistance. " 12

Until recently there has been no s

explanation for these observations. Iopening up of the microcirculationlated and reversed capillary flowobserved by Florey,'3 but it was uncli

40 -

0

.r__1)- E 20-_a

C:

O -

B

0

3cn

C- E 20

'aCN

._

CR

0 2 4 6Seconds

Figure 5 In some circumstances hyperaemia nduring the three second compression period. Impapplied in a glove to the hand. Doppler records Iarterialflozv (A) and synchronous radial venous(B) after hand cooling in a patient with mild Rcdisease. Arrow shows start of three second compperiod. Percompression hyperaemia is potentiatofvenous congestion.

Lt following this was induced, how it could influence thelimb. This precapillary sphincters that are the main deter-flow is no minants of capillary flow.us outflow The discovery of endothelial cell derivativesconsists of which can constrict or dilate vessels'4 has beenone in six pivotal to our understanding of the control ofe model of the microcirculation and it is now possible toary forcing construct a modified view of Starling's conceptld only be of tissue nutrition (fig 6). Most important isi tubes or endothelial relaxing factor (EDRF), which wasin which discovered in 198415 and is now known to be

lIy.9 nitric oxide with a half life of between six and 50)mpression seconds. 16foot within Endothelial relaxing factor mediates these doubled. relaxation of arteriolar smooth muscle byincreased various endogenous hormones including acetyl-

)mpression choline. Its production in endothelium isars to be stimulated by the 'shear stress' caused byto venous sudden pressure change. It is thus a tissue

ycle of the hormone (autocoid): it diffuses locally where itsig with the action is confined to the tissues where it isho in 1925 generated and it is rapidly neutralised byent venous plasma, haemoglobin and myoglobin in thetissue per- presence of oxygen. 17-20 The short half life ofincrease in EDRF corresponds well with the short durationible for the of postimpulse compression hyperaemia (ap-I vascular proximately 20 seconds2). Endothelial relaxing

factor is continually being produced becausesatisfactory if neutralised by its antagonist L-arginine, bloodRetrograde pressure increases greatly.2' It is therefore anwas postu- important element in the control of systemichad been vascular resistance and hence blood pressure,ear, even if and is counterbalanced by the inherent con-

traction of vascular smooth muscle that occursin response to stretch (Bayliss phenomenon).22Significantly EDRF has been found to bedeficient in patients with hypertension.2'The previously unexplained mechanisms

whereby tissue blood flow is matched to theoxygen requirement may be explained by theprolongation of the life of the tissue hormone(autocoid) EDRF in anoxia.20 Likewise EDRFcan explain how blood flow in skeletal muscle israpidly increased following contraction.1 23 24Muscle contraction produces a sudden increasein blood pressure in the intramuscular venules;the resultant shear stress causes the endo-thelium to produce EDRF that rapidly diffusesfrom the venules to the closely adjacent arter-ioles (Windkessel vessels), causing them todilate.'9 25 In support of this concept are theobservations that venous congestion increasespostcontraction hyperaemia26 and exercisetolerance,27 28 whereas in the reverse situationwhen the veins are empty hyperaemia andcapacity for muscular exercise are reduced, asanyone working above their head will confirm.Prostacyclin-like EDRF is produced by shearstress and its longer half life (four minutes) may

.... account for the more persistent component of*r-, impulse compression hyperaemia.8 10 Another autocoid produced by endothelial

shear stress is the peptide endothelial derivedconstricting factor (EDCF) that apparently has

payutrt ng the opposite effect to EDRF on vascular smoothofradial muscle. The physiological role of EDCF has notsflow been defined; in low concentrations it is aaynaud's vasodilator and its vasoconstrictor effect is moretrdby a period marked in veins than arteries (Vane J, personal

communication). As it is a larger molecule than

1175

A^ .0-

on 15 Septem

ber 2018 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.51.10.1173 on 1 O

ctober 1992. Dow

nloaded from

Gardner, Fox

Ir r

a us uo

vasomotC\- \ ' | -l ,X0 ' / lbraeAutonomous 4

--lsmooth Sudden pchange --ishear str(

Vasomnotio i) liberates I

*r',ximal sm-ooth~M uscle under:

! ervous control

Figure 6 Starling's diagram ofthe capillary loop. Suggested modification accordirecent advances in the understanding ofthe microcirculation. Distal arteriolarflowby a balance ofvasoconstriction in response to increased pressure (Bayliss response)dilatation by endothelial relaxingfactor (EDRF) (nitric oxide) produced by endothresponse to sudden pressure change. Intrinsic rhythm ofarteriolar muscle (vasomotilcapillaries intermittently to open and close. Pericapillaryfluid circulation through (capsule is indicated byfluorescin capillaroscopy; note difference in capillary and lyipressures. Endothelial permeability plays a vital part in regulation oftissue fluid atnourishment.

EDRF it is less likely to diffuse fromand influence the smooth muscle of accing arterioles.

It is easy to see how a systemic coraction limited to the venous systeibe important in controlling blooddistribution in shock and stress, bimportantly could be a local action coivenular smooth muscle, causingcongestion, decreasing the capillaryvenous pressure gradient, increasingproduction, and improving tissue nSuch an action might be most impccontragravitational areas of the cirparticularly the brain. Support for ththesis is the histological finding of an insmooth muscle in cerebral venules.

Venous thrombosis in the legsThe importance of the unsolved prcthromboembolism cannot be overemjanticoagulant drugs save lives butproblems limit their use, particularly i

paedic and neurosurgery. The most dangerousthromboses are those in the femoropopliteal andlarger veins, emboli from which may be fatal.Least dangerous are emboli from the musclesinuses which give rise to emboli that are smallenough to be lysed within the lung with no latesequelae.2 Thromboses in the main venouschannels commonly lead to the postphlebiticsyndrome and chronic ulcers; these are theresult of chronic venous hypertension caused byobstruction to flow and valve damage that

;/ impairs the function of venous pumps. In thelong term (four to five years) fibrin accumulatesaround the capillaries29 30 leading to nutritionalchanges in skin and other tissues: vein wallsbecome weakened and often become varicose.2The association between skeletal injury and

venous thrombosis is well known and maylCapsule influence the decision to submit patients for

joint replacement. The incidence of majorproximal deep vein thrombosis after hip opera-tions without anticoagulant drugs is around50%.3'3 A related problem is the delayeddevelopment of postphlebitic problems in abouthalf the patients after tibial fracture (table).34

ressu re Physical prophylaxis for venous thrombosis_w- after an operation avoids the complications ofess' pharmacological methods and if effective isEDRF clearly more desirable. It is to this end that

several centres around the United Kingdomhave taken part in controlled trials of pneumaticimpulse pumping of the foot and the prophy-lactic effect is now proved (fig 7).3The mechanisms whereby impulse pumping

prevents thrombosis are thought to be: (a) theturbulence, seen on video phlebography, which

ing to is induced in valve pockets where thrombosiscontrol is commonly begins; (b) release of EDRF, which

and not only produces hyperaemia but also inhibitsMehum in and even reverses platelet aggregation36 37;on)causes (c) release of prostacyclin38; and (d) enhance-mphatic ment of fibrinolysis after tissue compres-id so.39-43

If impulse pumping is used in establishedthrombosis there is the theoretical risk oftriggering pulmonary embolism. Emboli so

i venules dislodged are small and clinically insignificantompany- and the risk is certainly less than that of

ambulation where contraction of the thighistricting muscles risks dislodgement of thrombi largem could enough to block pulmonary circulation.volume

ut moreistricting Effects of impulse compression on compartmental pressures incapillary II patients with severe tibial fractures. In three patients anarterio- increase in pressure occurred when pumping was discontinued.arERF Two patients were considered to be saved from fasciotomy

Effect Pressure 7ime needed for changechange (hours)(mmHg)

Decreased from>40 mmHg (n=4)

Decreased (n=2)

Increased (n= 1)Unchanged (n=4)

40-1045-3647-3649Q3231-2738-27

34792

263

10-14 19

272865

24 Hours after treatment

TIense haemarthrosis;fasciotomy after 30 minutec

1176

on 15 Septem

ber 2018 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.51.10.1173 on 1 O

ctober 1992. Dow

nloaded from

The venousfootpump and venous thrombosis

A B

Con AVI Con AVI

C

Con AVI

D 40-

E

0

0

20-c.5

L-

oE 0O0

Con AVI

Figure 7 Effect ofimpulse pumping in reducing risk ofmajor thrombosis (shaded areas ofcolumns) after operations on hips and knees. Con=control; A VI=patients treated bypneumatic impulse pumping. (A) Exeter. Total hip replacement, phlebographic study.(B) Scarborough. Total hip replacement, phlebographic study. (C) Portsmouth. Neck offemur afterfracture, Doppler study. (D) King's College Hospital (London). Total kneereplacement, phlebographic study. Results ofA, B, and C were statistically significant withrespect tofemoropopliteal thrombosis andD likewise when distal thromboses were included.

Impulse pumping has been used successfully inconjunction with streptokinase in threatenedvenous gangrene (A M N Gardner, unpublishedobservations).

Pneumatic impulse activation of the venousfoot pump: A-V impulse systemThe impulse compression device (fig 8) differs

Figure 8 AVI device. The main body ofthe machinecontains an electrical compressor, an air reservoir, valves,and timing mechanism that intermittently and rapidly inflatesa cuffapplied around thefoot. To be effective the venousfootpump must be primed, so a graduated stocking is worn onthe calfand the leg positioned below heart level. This systemis manufactured and supplied by Novamedix Ltd, ViscountCourt, South Way, Walworth, Andover, Hampshire SPIOSNW, United Kingdom.

V.I I I2

"I

4 6 8Seconds

10 12

Figure 9 Femoral veinflow using the device wtth a leg in aplaster cast (tracing ofSonicaid Vasoflow 2 recording).

from all previous compression devices in that itis applied only to the mid foot and inflatesrapidly, thus mimicking the haemodynamicaction of normal ambulation. It consists of a

pneumatic cuff wrapped around the foot andworn beneath a purpose made slipper. It can

also fit within a plaster cast and can be appliedto the hand when an appropriately shapedinflation pad is confined within a glove. Infla-tion is rapid (<0 4 seconds) and intermittent(20 second cycle), metered from an air reservoirand compressor. This rapid inflation imitatesthe effect of normal gait by generating vigoroushaemodynamic impulses throughout the venousconduits (fig 9)2 in addition to turbulence in thevalve pockets where many major thrombosesoriginate. In clinical practice the hyperaemiceffect of impulse compression has proveduseful. In a controlled study it has been shownto increase blood flow and to alleviate rest painin ischaemic limbs2 and may also limit tissuenecrosis in patients with arterial insufficiencyduring investigations while waiting for an

operation.5 Similarly, a significant reduction ofthese two parameters occurs with impulsecompression in patients with limb pain andswelling after trauma."The beneficial effect of impulse compression

on pain and swelling after trauma has beenproved.43 This may be due in part to theliberation of EDRF which itself has an analgesiceffect,45 but more importantly may be theclearance of kinins from tissues and reduction ofvenous congestion with neutralisation of theresulting acidosis that probably increases sensi-tivity to pain.46 In compartmental syndromesimpulse compression of the foot can oftenreduce dangerously high pressures therebyallowing capillaries to reopen and thus avoid theneed for fasciotomy (table)." Studies are inprogress to determine whether impulse com-pression can improve bone healing; acceleratedrecovery from fractures of the os calcis has beenshown.47 Also encouraging are initial experi-ences in treating algodystrophy and facilitatingphysiotherapy.As it has been suggested that rheumatoid

arthritis may be an example of reperfusioninjury with hypoxia,48 it is possible that altera-tion of blood flow could influence the patho-logical process within the joint. As blood flowcan be improved in ischaemic limbs it isnot beyond the realms of possibility thatjoint perfusion kinetics could be similarlymanipulated.

40-

30-

0

cor.,0

z

10-

0-

1177

on 15 Septem

ber 2018 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.51.10.1173 on 1 O

ctober 1992. Dow

nloaded from

Gardner, Fox

I Gardner A M N, Fox R H. The venous pump of the humanfoot. Bristol Medico-chirurgical J7ournal 1983; 98: 109-12.

2 Gardner A M N, Fox R H. The return of blood to the heart.London: Libbey, 1989.

3 Gardner A M N, Fox R H. The return of blood to the heartagainst the force of gravity. In: Negus D, Jantet G, eds.Phkebology '85. London: Libbey, 1986: 68-71.

4 Winkler G. Les veines du pied. Archives of Anatomy(Strasbourg) 1923; 37: 175-84.

5 Morgan R H, Carolan G, Psaila J V, Gardner A M N, FoxR H, Woodcock J P. Arterial flow enhancement by impulsecompression. Vasc Surg 1991; 25: 8-16.

6 Roy C S, Brown J G. The blood pressure and its variations inthe arterioles, capillaries and smaller veins. 7 Physiol 1879;2: 329-59.

7 Gaskell P, Parrot J C W. The effect of a mechanical venouspump on the circulation of the feet in arterial obstruction.Surg Gynecol Obstet 1978; 146: 583-92.

8 Krogh A. Anatomy and physiology of capillaries. New York:Yale University Press, 1929.

9 Whittaker S R F, Winton F R. The apparent viscosity ofblood flowing in the isolated hind limb of the dog. J Phvsiol(Lond) 1933; 339-69.

10 Lewis T, Grant R. Observations on reactive hyperaemia inman. Heart 1925; 12: 73-120.

11 Mohrman D E, Sparks H V. Myogenic hyperemia followingbrief tetanus of canine skeletal muscle. Am 7 Physiol 1974;227: 531-5.

12 Smiesko V. Unidirectional rate sensitivity component in localcontrol of vascular tone. Pflugers Arch 1971; 327: 324-36.

13 Florey H. Microscopic observations on the circulation of theblood in the cerebral cortex. Brain 1925; 48: 43-64.

14 Anonymous. Endothelins [leading article]. Lancet 1991; i:79-81.

15 Furchgott R F, Zawadski J V. The obligatory role ofendothelial cells in the relaxation of arterial smooth muscleby acetylcholine. Nature 1984; 288: 373-6.

16 Palmer R M J, Ferrige A G, Moncada S. Nitric oxide releaseaccounts for the biological activity of endothelium-derivedrelaxing factor. Nature 1987; 327: 524-6.

17 Anonymous. Yin and yang in vasomotor control [leadingarticle]. Lancet 1988; ii: 19-20.

18 Anonymous. EDRF [leading article]. Lancet 1987; ii: 137-8.19 Ignarro L J, Byrns R E, Buga G M, Wood K S. Endothelium-

derived relaxing factor from pulmonary artery and veinpossesses pharmacological and chemical properties identicalto those of nitric oxide radical. Circ Res 1987; 61: 866-79.

20 Fairchild H M, Ross J, Guyton A C. Failure of recovery fromreactive hyperemia in the absence of oxygen. Am J7 Physiol1966; 210: 490-2.

21 Valiance P, Collier J, Moncada S. Effects of endothelium-derived nitric oxide on peripheral arteriolar tone in man.Lancet 1989; ii: 997-1000.

22 Bayliss W M. On the local reactions of the arterial wall tochanges of internal pressure. J Physiol (Lond) 1902; 28:228-31.

23 Anrep G V, Blalock A, Samaan A. The effect of muscularcontraction upon the blood flow in the skeletal muscle. ProcR Soc Lond [Bioll 1934; 114: 223-45.

24 Barcroft H, Millen J L E. The blood flow through muscleduring sustained contraction. 7 Physiol (Lond) 1939; 97:17-31.

25 Zweifach B W, Metz D B. Selective distribution of bloodthrough the terminal vascular bed-mesenteric and skeletal.Angiology 1955; 6: 282-90.

26 Folkow B U, Haglund U, Jodal M, Lundgren 0. Blood flowin the calf muscle of man during heavy rhythmic exercise.Acta P-hvsiol Scand 1971; 81: 157-63.

27 Rodbard S, Farbstein M. Improved exercise tolerance duringvenous congestion. J Appl Physiol 1972; 33: 704-10.

28 Rodbard S, Deliry J L, Jaffe P. Venous pressure and exercisetolerance. Cardiology 1976; 61: 229-31.

29 Burnand K G, Clemenson G. The effect of sustained venoushypertension on the skin capillaries of the canine hind limb.Br .7 Surg 1982; 69: 41-4.

30 Burnand K G, Whimster I. Pericapillary fibrin in the ulcerbearing skin of the leg the cause of lipodermatosclerosis.BM7 1982; 285: 1071-2.

31 Ishak M A, Morley D. Deep venous thrombosis after totalhip arthroplasty. Br 7 Surg 1981; 68: 429-32.

32 Stamatakis J D, Kakkar V V. Femoral vein thrombosis andtotal hip replacement. BM7 1977; ii: 223-5.

33 Lassen M R, Boris L C. Prevention of thromboembolism in190 hip arthroplasties. Comparison of low molecular weightheparin and placebo. Acta Orthop Scand 1991; 62: 33-8.

34 Aitken R J, Mills C, Immelman E J. The postphlebiticsyndrome following tibial shaft fracture. .7 Bone J7oint Surg[Br/ 1987; 69: 775-8.

35 Fordyce M J F, Ling R S M. The prevention of DVTfollowing hip replacement by use of the A-V Impulsesystem. J Bone 7oint Surg [Br/ 1992; 74: 45-9.

36 Griffith T M, Henderson A H. EDRF and the regulation ofvascular tone. Int .7 Microcirc Clin Exp 1989; 8: 383-96.

37 Griffith T M. Endothelium-influenced vasomotion: modelsand measurements. In: Westerhof N, Gross D R, eds.Vascular dynamics-physiological perspectives. New York:Plenum Press, 1989: 177-93 (Nato ISA Series).

38 Guyton D P, Khyat A, Husni E A, Schreiber H. Elevatedlevels of 6-keto-prostaglandin-Fla from a lower extremityduring external pneumatic compression. Surg Gynecol Obstet1988; 166: 338-42.

39 Allenby F, Pflug J J, Boardman L, Calnan J S. Effects otexternal pneumatic compression on fibrinolysis in man.Lancet 1973; ii: 1412-4.

40 Knight M, Dawson R, Ackerley R, Jones L. Intermittentcompression of the arms reduces the incidence of postopera-tive venous thrombosis in the legs: an explanation. Lancet????; ii: 1265-8.

41 Tarnay T J, Rohr P R. Pneumatic calf compression,fibrinolysis and the prevention of deep venous thrombosis.Surgery 1980; 88: 489-96.

42 Altenkamper H. Intluence d'une compression intermittentesur l'activitie fibrinolytique de la paroi veineuse chez despatients souffrant de maladie variqueuse. In: Davy A,Stemmer R, eds. Phlebologie '89, Vol. 2. London: Libbey,1989: 855.

43 Hull R D, Rascob G E. Prophylaxis of venous thrombo-embolic disease following hip and knee surgery. J Bone7oint Surg [AmI 1986; 68: 146-50.

44 Gardner A M N, Fox R H, Lawrence C, Bunker T D, LingR S M, MacEachern A G. Reduction of post-traumaticswelling and compartment pressure by impulse compressionof the foot. .7 Bone Joint Surg [Br/ 1990; 72: 810-5.

45 Griffith T M, Randall M. Nitric oxide comes of age. Lancet1989; ii: 875-6.

46 Lindahal 0. Experimental skin pain. Acta Physiol Scand1961; 51 (suppl 179): 1-89.

47 Erdmann M W H, Richardson J, lempleton J. Os calcisfractures: a randomised trial of conservative treatmentversus impulse compression of the foot. Injuiv 1992; 23:305-7.

48 Stevens C R, Williams R B, Farrell A J, Blake D R. Hypoxiaand inflammatory synovitis: observations and speculation.Ann Rheum Dis 1991; 50: 124-32.

1178

on 15 Septem

ber 2018 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.51.10.1173 on 1 O

ctober 1992. Dow

nloaded from