U.S. flag

An official website of the United States government

Skip to main content
Return to topReturn to top



Arterial Aneurysms and Arteriovenous Fistulas

Circulatory Effects of Arteriovenous Fistulas 1

Daniel C. Elkin, M. D.


  As in lesions observed in civilian practice, the local evidences of an arteriovenous fistula were similar regardless of the site of the lesion. The most common sign was the bruit, or murmur, which could be heard with a stethoscope in the region of the fistula. In contrast to the purely systolic sound of an arterial aneurysm, that produced by an arteriovenous fistula was a continuous murmur extending through both systole and diastole. Although in general the larger the fistula the louder the murmur, this was not always the case.  At times small fistulas produced loud murmurs. These murmurs, like those arising from the heart, were transmitted a considerable distance from the site of origin. The chief cause of failure to hear the murmur was simply failure to listen for it.

  When the murmur was of sufficient intensity, a thrill was generally felt. Again, although somewhat indicative of the size of the fistula, the intensity of the thrill did not offer completely reliable evidence of the size since it could be modified by the position of the fistula and the structures overlying it. The thrill was less well transmitted than the murmur, and was localized to an area near the lesion.

  Some arteriovenous fistulas were associated with no external evidence of either arterial or venous dilatation. Occasionally, however, there was a pulsating mass of variable size in the affected area caused by aneurysmal dilatation of either the involved artery or vein, or by a false sac in the region of the fistula. Superficial veins near the fistula or distal to it were usually dilated and tortuous (Fig.15). This dilatation disappeared soon after the artery was repaired. The pressure in the veins in the vicinity of the fistula was elevated but the systemic venous pressure was normal. Blood withdrawn from veins near the fistula always showed an abnormally high oxygen content.

  In the region of the fistula itself the skin was warmer than normal but in the area distal to it there were sometimes signs of circulatory insufficiency ranging from local coolness and pallor to such severe manifestations as anemia or gangrene. Edema sometimes developed in the area distal to the lesion.

1 The special studies reported in this chapter were made by James V. Warren, M. D., Emory University Hospital, Atlanta, Ga.


Figure 15  A. Plantar arteriovenous fistula. Note dilatation and tortuosity of superficial veins before operation.  B. Same  foot 2 weeks after the excision of the fistula.

  Skin temperature studies and oscillometric records confirmed these clinical observations (Chart 19). Oscillometric studies revealed increased pulsations at the site of the fistula which diminished gradually as they were measured proximally and distally.

  When the vessels were exposed at operation the proximal artery and vein were usually dilated. The artery distal to the fistula was likely to be small, its size depending upon the amount of blood shunted away from it through the fistulous communication.  The vein distal to the fistula was usually dilated because of the arterial blood shunted into it, though the enlargement was greater on the proximal side. All collateral branches in the region of the fistula, whether arterial or venous, were increased in size. Serial sections of the vessel walls did not usually reveal significant changes.


  Patients with arteriovenous fistulas resulting from battle injuries usually presented no cardiac abnormalities upon physical examination. Alterations in size were sometimes observed, but only occasionally was the increase greatly in excess of normal standards. Whatever increase might be present could be detected only by the subsequent decrease in the transverse diameter of the heart as measured in teleoroentgenograms taken after operative eradication of the fistula (Fig. 16A and B). Only on relatively infrequent occasions was this increased size in excess of the normal standards; therefore it was noted only by the comparison of films made before and after the removal of the fistula. Occasionally, however, the heart became considerably enlarged. Since the alteration was reversible and disappeared promptly after removal of the fistula this enlargement was assumed to represent dilatation rather than true hypertrophy. The irreversible hypertrophy occasionally seen in patients


Chart 19. Oscillometric and skintemperature determinations in left plantar arteriovenous fistula. Notethatoscillations and skin temperatures are reduced distal to the lesion.

with longstanding lesions was practicallynever observed in patients in military service sincemen were not inducted into the Army with such lesions.

  It is generally recognized that in patients with arteriovenous fistulas congestive heart failure may develop, though the mechanism producing the cardiac decompensation is in dispute. It has been variously ascribed to the greatly increased demands on the heart and to the diminution of coronary blood flow secondary to the lowered mean arterial pressure.

Prior to World War II studies on the cardiac output in arteriovenous fistulas (a knowledge of which is essential in the elucidation of the altered circulatory dynamics in these patients) had been few in number and not


Figure 16. Serial teleoroetgenograms in femoral arteriovenous fistula.  A. Preoperative view shows heart of increased size.  B. Postoperative view shows decrease in cardiac size.

entirely consistent in results. Lewis andDrury, 2 whose studies were made with a glasscardiometer inclosing the heart, reported in 1923 that an increase incardiac output wassometimes observed when an arteriovenous fistula was open and that anincreased central venouspressure was observed in association with the elevated cardiac output.They therefore assumedthat in human subjects the cardiac output was normal when the venouspressure was normal; noactual measurements were made. In 1924, Harrison, Dock, and Holman, 3who used the directFick technique, reported the elevation of cardiac output in animalswith experimentalarteriovenous fistulas; their observations were confirmed by a numberof other observers. Thefew studies made on clinical subjects were much less conclusive,although in most instances thecardiac output was found to be increased. The possibility exists thatthe foreign gas methodsoften used in clinical studies were invalidated by the abnormalcirculatory dynamics present inpatients with arteriovenous fistulas.

  Investigations conducted in peacetime were necessarily limited to experimental animals with artifically created fistulas and to clinical cases as they were encountered in individual experiences. During World War II the concentration of patients with these lesions at vascular centers in the Zone of Interior permitted studies which would be impossible except in time of war, and the investigations carried out throw some light, both positively and negatively, upon the alteration in circulatory dynamics caused by the presence of arteriovenous fistulas.

2 Lewis, T., and Drury, A. N.:  Observations relating to arteriovenous aneurism.  Heart 10: 301-389, Oct 1923.

3 Harrison, T.R.:  Dock, W., and Holman, E.:  Experimental studies in arterio-venous fistulae; cardiac output. Heart 11: 337-341, Dec 1924.


Materialsand Methods

  At Ashford General Hospital the cardiac output was studied in 47 patients with arteriovenous fistulas by means of the low frequency, critically damped ballistocardiograph (Table 13). All but 1 of the 47 were males. The patients were presumably normal prior to their injuries and none had evidence of frank congestive heart failure. The interval between the injury which resulted in the arteriovenous fistula and the operative repair varied from 2 months to slightly over 2 years. Definite local signs of the fistula, including a thrill or murmur near its site, were present in all patients, and in all the fistula was demonstrated and eradicated successfully by surgical intervention. The postoperative course was without serious complications in all instances, and no patient presented evidence of recurrence of the lesion during the period of observation.

  The ballistocardiograph, which was utilized to determine the cardiac output from the ballistic recoil of the body with each heart beat, is a table or bed so suspended on four steel strips or springs that movement only in the longitudi -



nal direction is possible. The length of thesprings is adjustable. and the inherent frequency ofthe bed can therefore be kept constant in spite of variations in theweights of patients. When thebed is displaced it returns to its original resting position withoutovershoot because it is criticallydamped. Movements of the bed are recorded photographically, by means ofa light beam andmirror, with a camera similar to that used in an electrocardiograph.The output of the heart wascalculated in the studies at Ashford General Hospital from themeasurement of several featuresof the ballistocardiographic tracing and the determination of thepatient's blood pressure (Chart20). The ballistocardiographic measurements were checked againstcomparative studies by thedirect Fick technique. The observations were recorded in terms of thecardiac index, that is, theoutput of the heart in liters per minute per square meter of bodysurface, in order to facilitatecomparisons between individuals of various sizes.

All determinations were made 3 hours or more after the previous meal and after the patient had relaxed on the ballistocardiographic table for 15 minutes or more. Arterial pressure was obtained by the auscultatory method with an ordinary mercury manometer. The transverse cardiac diameter as measured from teleoroentgenograms was used as an index of changes in the size of the heart.

  Although several studies had previously been made to determine the correlation between the ballistocardiographic method and the Fick catheter

Chart 20.Effects of temporaryocclusion of arteriovenous fistula on arterial pressure, stroke volume, heartrate,cardiac output, and atrial pressure.


method of measuring the cardiac output,additional studies were made as a part of thisinvestigation to establish the validity of the ballistocardiographicmethod under the conditions ofthe investigation. The control studies were made on 3 patients witharteriovenous fistulas and on3 others with large areas of reactive hyperemia. Reactive hyperemia wasproduced by applyingblood pressure cuffs around the upper parts of both legs and theninflating them above arterialpressure for 20 minutes. The state of reactive hyperemia which ensuedwhen the cuffs werereleased was for all practical purposes a temporary arteriovenousfistula. With the release of thecuffs the cardiac output in each case rapidly rose to about twice thenormal level.


  The control studies described established the validity of the ballistocardiograph as a means of measuring the output of the heart. Excellent correlation was obtained.

The blood pressure showed little variation from the preoperative to the postoperative period. In the cases in which a change occurred, the tendency was toward diminution of the pulse pressure after operation chiefly because of an elevation of the diastolic pressure.

  When the normal resting cardiac output, that is the postoperative value, was compared with the preoperative cardiac output, the latter was found to range from 21 percent below, to 127 percent above the former (Chart 21). (For correct interpretation of these results it is necessary to bear in mind that variations of as much as 25 percent above or below the value selected as the resting cardiac output can be expected in normal subjects when repeated measurements are made at different times.)

  The postoperative alteration in cardiac output was obviously the result of a change in stroke volume rather than of a change in pulse rate. When the fistula was intact the basal pulse rate was above 85 in only 7 of the 47 patients, the vessels involved in these 7 patients were the femoral vessels in 4 patients, the iliacs in 2, and the subclavian in 1. The cardiac output in this group of patients before operation ranged from 32 to 127 percent (average 82 percent) above the postoperative level (Chart 22).


It is fortunate that the ballistocardiographic method of studying the cardiac output is apparently accurate since it possesses several important advantages over other methods now available: (1) From the patient's standpoint this is the simplest of all methods since he merely lies on the table completely relaxed and does not have to endure the psychologically disturbing effects of the introduction of needles or the use of a rebreathing apparatus. (2) Repeated determinations can easily be made on the same subject. (3) Extremely rapid changes in cardiac output, which cannot be observed by other methods, can be studied by this method.


Chart 21. Effects (in percentages of normal values) of arteriovenous fistula on cardiac output. The shaded bars represent the variation of increase and the black bars the variation of decrease.  A change of less than 25 percent in cardiac output is considered within the range of normal variation.  Note that only 40 of the 47 patients studied are depicted in the chart inasmuch as readings were not obtained on 7.

Chart 22. Ballistocardiographic observations before and after excision of femoral arteriovenous fistula.


  The person-to-person variability in cardiac output is large and in this series the postoperative cardiac index varied considerably from patient to patient but the range was similar to that found in comparable normal subjects studied by the catheter method by Stead, Warren, Merrill, and Brannon.If necessary, the investigation at Ashford General Hospital could have been carried out by using the average normal value obtained by this study, but all the patients in the series returned to a normal state after operation without evidence of residual abnormality in the circulation, and there seemed full justification for utilizing individual postoperative values as the basis for estimating the load placed on the heart by the fistula before operation.

Since a variation of 25 percent above and below the postoperative cardiac index could be expected in any group of subjects, a change after operation within this range could not be attributed to the effect of the fistula on the circulation. In 22 of the 47 patients, approximately 47 percent, the postoperative alteration was within this range. In the other 25 patients the cardiac output when the fistula was open was more than 25 percent greater than the output after it had been closed at operation, the altered values ranging from 25 to 127 percent. In other words, slightly more than half of the 47 patients with arteriovenous fistulas studied by the ballistocardiographic method had a significantly elevated cardiac output before operation as measured under basal conditions.

  It was difficult to estimate the functional size of the fistula from either clinical or pathologic evidence. The size of the vessel involved and the intensity of the thrill or the murmur were not entirely reliable in predicting the increase in the cardiac output, though in general, clinical evidence of a large fistula or of the presence of a fistula located in a large vessel was associated with the greatest increase in cardiac output. An analysis of the 47 cases showed that while all of the patients with the most appreciable increase in cardiac output had fistulas involving relatively large vessels, the relationship appeared to be no more specific. A fistula involving the femoral vessels might be associated with a large increase in the cardiac output, but did not necessarily cause it. On the other hand a fistula fed by a small vessel, because of the limited arterial inflow, produced at the most a small increase in the cardiac output. The increase in the cardiac output is perhaps, as Starr's studies 5 indicate, the best index of the load placed on the circulation by an arteriovenous fistula.

  When an attempt was made to correlate the duration of the fistula with the increase in cardiac output it was found that in general the greatest increases in cardiac output occurred in association with fistulas of relatively short duration, and vice versa. A reasonable explanation of this observation is that larger fistulas were promptly diagnosed (and the patients sent immediately to the vascular centers), while small fistulas were frequently overlooked for long periods of time.

Stead, E. A., Jr.; Warren, J. V.; Merrill, A. J., and Brannon, E. S.: Cardiac output in male subjects as measured by the technique of right atrial catheterization; normal values with observations on the effect of anxiety and tilting. J. Clin. Investigation 24: 326-331, May 1945.

Starr, L:  Clinical studies; with the ballistocardiograph in congestive failure, on digitalis action, on changes in ballistic form, and certain acute experiments. Am. J. M. Sc. 202: 469-485, Oct 1941.


  Cardiac enlargement, as demonstrated by an increased transverse cardiac diameter in teleoroentgenograms, is recognized as a frequent accompaniment of arteriovenous fistula. The increase is reversible and the heart returns to its normal size after the fistula is corrected. An attempt to correlate the change in the transverse diameter of the heart following operation on the patients in this series with the degree of increase in cardiac output resulted in a correlation coefficient of 0.12, which is not significant. This lack of correlation suggests comparison with the observations made in arterial hypertension in which the amount of cardiac enlargement cannot be directly correlated with either the degree or the duration of the increased arterial pressure. Given 2 patients with the same degree of hypertension, 1 may have marked cardiac enlargement and the other no enlargement at all. Such was the case in instances of cardiac enlargement associated with arteriovenous fistula in this series.

  From these studies it may therefore by concluded that many, though by no means all, patients with arteriovenous fistulas have distinct elevations of the cardiac output as a result of the lesion, with a return to values apparently normal after correction of the fistula. These observations are in accord with previous observations.

  Why the increased cardiac output occurs is a question of considerable interest. Most textbooks stress the importance of the filling pressure of the ventricles, that is, the pressure in the atria and the great veins, but this is not an adequate explanation: The studies by Starr and his associates,6 corroborated in this series, show that in patients with large areas of reactive hyperemia there may be a decided change in cardiac output without any change at all in right atrial pressure.

With this change eliminated as the explanation of the increased cardiac output in association with arteriovenous fistulas, changes in the arterial tree offer a possible explanation. When the fistula is open there is a definite reduction in peripheral resistance. Ballistocardiographic studies show that when the fistula is compressed and then released, changes in stroke volume and heart rate are almost instantaneous. The evidence indicated that the alteration in cardiac output is predominantly the result of change in stroke volume rather than pulse rate and is related to change in the arterial tree rather than to an increase in filling pressure presented to the right side of the heart.

  An increase in cardiac output also occurs in clinical conditions in which the circulatory dynamics are altered as they are altered in arteriovenous fistulas. In patent ductus arteriosus, for instance, which is actually an arteriovenous fistula between the aorta and the pulmonary artery, such an increase was reported by Eppinger, Burwell, and Gross.7 During the later months of preg-

Starr, I.; Rawson, A. J.; Schroeder, H. A., and Joseph, N. R.: Studies on the estimation of the cardiac output in man, and of abnormalities in cardiac function, from the heart's recoil and the blood's impacts; the ballistocardiogram. Am. J. Physiol. 127: 1-28, Aug 1939.

Eppinger, E. C.; Burwell, C. S., and Gross, R. E.: Effects of the patent ductus arteriosus on circulation. J. Clin. Investigation 20: 127-143, Mar 1941.


nancy the cardiac output is almost alwayselevated, and Burwell 8 has pointed outthat in manyways the placenta behaves like a modified arteriovenous fistula. Thestudies carried out atAshford General Hospital furnish no direct evidence concerning cardiacfailure in patients witharteriovenous fistulas. It seems possible, however, that in thiscondition just as in patent ductusarteriosus, severe anemia, and thyrotoxicosis, failure occurs when thecardiac output is far abovenormal. In arteriovenous fistula one would not necessarily expect a lowcardiac output toindicate circulatory inadequacy, though cardiac failure would notnecessarily develop in patientswith the highest cardiac outputs. The resistance of the cardiovascularsystem to stress varieswidely from person to person and the amount of overwork which causesfailure in one subjectwould not necessarily cause it in another.

  The explanation is probably along the lines suggested by Starr and Jonas 9 for cardiac failure in thyrotoxicosis: It occurs in patients in whom the heart is not able to increase its output to meet increased demands, which means that a normal cardiac output in a patient with extreme hyperthyroidism may be taken to indicate the possibility of cardiac failure. Similarly, cardiac failure might be expected to develop in patients with arteriovenous fistulas in whom the cardiac output does not increase to compensate for the loss of effective output by way of the fistula.

  The combination of a normal cardiac output and a large fistula might therefore be more indicative of cardiac insufficiency than a large cardiac output per se. If this reasoning is correct, it furnishes an additional explanation of the inability in this investigation to correlate cardiac enlargement or elevated blood volume directly with the increase in cardiac output.


  Holman,10 in 1924, seems to have been the first to report an increase in blood volume in animals in which arteriovenous fistulas had been produced, with a return to normal level following the removal of these lesions. In his method the dye brilliant vital red was used to determine the blood volume. Four minutes after the dye was injected a sample of blood was drawn. The five animals in which he had produced particularly large arteriovenous communications all showed an increased blood volume, and in general the increase appeared to be related to the size of the fistula. Similar blood volume studies were reported later in clinical cases of arteriovenous fistulas, the increased blood volume in most instances returning to normal after correction of the lesions.11

8 Burwell, C. S.:  Placenta as a modified arteriovenous fistula considered in relation to the circulatory adjustments to pregnancy.  Am. J. M. Sc. 195: 1-7, Jan 1938.

9 Starr, I., and Jonas, L.: Supernormal circulation in resting subjects (hyperkinemia), with a study of the relation of kinemic abnormalities to the basal metabolic rate.Arch. Int. Med. 71: 1-22, Jan 1943.

10  Holman, E.: Experimental studies in arteriovenous fistulas; blood volume variation (pt 1).  Arch. Surg. 9: 822-836, Nov 1924.

11(1)Holman, Emile: Arteriovenous Aneurysm.Abnormal Communication Between Arterial and VenousCirculations.New York, The Macmillan Company, 1937.

(2) Holman, E.:  Clinical and experimental observations on arteriovenous fistulae. Ann. Surg. 112: 840-875, Nov 1940.


  In Rowntree and Brown's blood volume studies12 on 7 patients with Arteriovenous fistulas, 3 of which were congenital, the average value for the group was slightly above normal but only 1 patient showed a distinct elevation. Pemberton and Saint 13 reported finding a somewhat elevated value in a single patient with a congenital arteriovenous communication.  In view of their own observation on 3 patients, studied before and after operation, Reid and McGuire14 expressed some doubt concerning changes in the blood volume in arteriovenous fistulas. In the first case there was no essential change. In the second there was a decrease of 1,000 cc. after operation, but 9 months later the value was the same as before operation. The same authors reported a slight increase in the blood volume in 2 experimental animals with arteriovenous fistulas.Ellis and Weiss,15 who used Evans blue dye T-1824 to study a single patient before operation, found the blood volume within normal limits. Kennedy and Burwell 16 found a greatly elevated blood volume before operation in a patient with multiple arteriovenous communications; the fistulas recurred in this case and there was considerable fluctuation in the postoperative results.

The methods used in the early studies of the blood volume in arteriovenous fistulas are open to many theoretical and practical objections. The later observations mentioned, which are not open to these objections, are limited in number. For these reasons, advantage was taken of the concentration of patients with arteriovenous fistulas in the vascular centers of the Zone of Interior to attempt more conclusive studies on the blood volume in the presence of these lesions.

Materials and Methods

  The blood volume was studied at Ashford General Hospital in 41 patients with arteriovenous fistulas by means of the blue dye T-1824 (Table 14).  All the lesions followed trauma. The patients, 1 of whom was a woman, were all young adults and all apparently healthy prior to injury. At the time of the investigation none had complicating injuries or illnesses which might have altered the blood volume.

  Blood vessels in all parts of the body were involved. Every patient presented the physical signs characteristic of arteriovenous fistulas and in no instance was there any evidence of frank cardiac decompensation. The interval between wounding and operation varied from 2 months to a little over 2 years.

12  Rowntree, Leonard G., and Brown, George E.:  The Volume of the Blood and Plasma in Health and Disease.  Philadephia, W. B. Saunders Company, 1929.
13  Pemberton, J. de J., and Saint, J. H.:  Congenital arteriovenous communications.  Surg., Gynec. & Obst. 46: 470-483, Apr 1928.
14  Reid, M. R. and McGuire, J.:  Arteriovenous aneurysms.  Ann. Surg. 108: 643-693, Oct  1938.
15  Ellis, L. B. , and Weiss, S.:  Local and systemic effects of arteriovenous fistula on the circulation in man.  Am. Heart J. 5: 635-647, Jun 1930.
16  Kennedy, J. A., and Burwell,  C.S.: Measurements of the circulation in a patient with multiple arteriovenous connections.  Am. Heart J. 28: 133-148, Aug 1944.



  In every case the fistula was successfully excised, with no complications postoperatively. Thiopental sodium anesthesia or continuous spinal analgesia was used.

  Determinations of the blood volume were made under basal conditions, at least 12 hours after the last meal and before patients had risen in the morning. With the exception of the determinations made on the third postoperative day, all preoperative and postoperative studies were made on patients who were ambulatory.   The plasma volume was measured with blue dye T-1824, as recommended by Gregersen, Gibson, and Stead.17  Blood samples were drawn from the antecubital veins. The syringes used were dry and in order to avoid hemoconcentration, tourniquets were not used. Following withdrawal of the dye-free sample, 3 cc. of the dye (0.1-percent solution) were injected from a calibrated

17 Gregersen, M. L; Gibson, J. J., and Stead, E. A.:  Plasma volume determination with dyes; errors in colorimetry; use of the blue dye T-1824. Am. J. Physiol. 113: 54-55, Sep 1935.


syringe which was then rinsed three times with blood. Care was taken to prevent extravascular loss of the dye. If any loss occurred, a note was made of it.  Exactly 10 minutes after the dye had been injected, a sample of blood was obtained with the proper precautions against hemolysis and hemoconcentration. In some cases an indwelling needle was left in place and serial samples were obtained to determine the rate of disappearance of the dye.

  The blood samples were placed in small tubes where they were covered with mineral oil and allowed to clot and retract before they were gently centrifuged. Any sample which showed evidence of lipemia or hemolysis was discarded.

  The concentration of the dye T-1824 was determined by measuring the light absorption of the dye-containing serum sample in relation to the dye-free sample at a wave length of 620 millimicrons with a Coleman Junior spectro photometer or with a Nickerson decade photometer.18  The apparatus had previously been calibrated with known amounts of dye in serum.

  The plasma volume was calculated from the single 10-minute point as suggested by Gregersen and Rawson,19 or from the dye concentration value obtained from extrapolation of the disappearance curve plotted on a semilog arithmic scale. The total blood volume was determined from the plasma volume and the hematocrit reading. To facilitate comparisons between patients of different sizes, the blood volume per square meter of body surface was also calculated.

When the dye-free blood specimen was withdrawn, sufficient blood was also obtained for hematocrit readings, and determinations of hemoglobin and total protein. Heparin was used as the anticoagulant in the hematocrit tubes. The hemoglobin concentration was determined by measuring the optical density of a dilute alkaline solution of blood. The total protein value was determined by the density method of Barbour and Hamilton.20

  Many of the patients in whom blood volume studies were made underwent additional studies including determination of the cardiac output by means of the ballistocardiograph and estimations of the size of the heart by measurement of the transverse cardiac diameter on a teleoroentgenogram.


  In the 41 patients studied, the blood volume per square meter of body surface ranged before operation from 2,400 cc. to 4,030 cubic centimeters. Ten days or more after operation the values ranged from 2,330 cc. to 3,380 cc., the alterations varying from a decrease of 1,060 cc. to an increase of 190 cubic centimeters. In 23 cases, approximately 56 percent of the series, the change was less than 200 cubic centimeters. In 18 patients, 44 percent of the series,

18  Nickerson, J. L.:  Portable decade photometer for determination of blood volume with the dye T-1824.  Rev. Scient. Instruments 15: 69-72, Mar 1944.

19 Gregersen, M. I., and Rawson, R. A.:  Disappearance of T-1824 and structurally related dyes from the blood stream. Am. J. Physiol. 138: 698-707, Apr 1943.

20 Barbour, H. G., and Hamilton, W. F.: Blood specific gravity; its significance and a new method for its determination. Am. J. Physiol. 69: 654-661, Aug 1924.


there was a postoperative decrease in theblood volume which ranged from 200 to 1,060 cc. persquare meter. In no patient was the rise following operation greaterthan 200 cc. (Chart 23).These data were confirmed in most instances by serial observations.

Plasma volume fluctuated as did the total blood volume. Hematocrit readings varied only slightly but more often than not showed a tendency to fall slightly after operation. Calculated red blood cell volume underwent similar changes. Changes in hemoglobin values and total protein concentration were not significant.


  Technical Methods. The blue dye T-1824 had been widely used for the determination of the plasma volume. In this series two methods were originally employed, the semilogarithmic plotting of disappearance of the dye, as recommended by Gregersen and Rawson, 21 and the examination of the single dyed specimen obtained 10 minutes after injection of the dye as recommended by Gregersen 22 for certain situations. The latter method, simpler to perform, proved as reliable as an extrapolated value for dye concentration obtained from the disappearance curve of the dye. Gregersen's results with the two methods were obtained in normal subjects and in subjects in shock. In the cases in this series in which both methods were used there was no evidence that the disappearance rate of the dye was affected by the presence of the arteriovenous fistula, and the plasma volume obtained by both methods was substantially the same. After the reliability of both methods had been established by comparative studies, Gregersen's 10-minute method was used almost exclusively because its simplicity permitted serial determinations of the plasma volume of a large number of patients who were available for study for a limited time only. No patient in this series had frank congestive heart failure; however, there is a possibility that the prolonged mixing time of the dye in this type of case might invalidate the 10-minute method.

  The calculation of total blood volume from the hematocrit reading and plasma volume has been criticized adversely by some observers on two counts: (1) the imperfection of the peripheral venous hematocrit reading as evidence of the proportion of cells in the blood throughout the body, and (2) the failure of the hematocrit reading to indicate the true cell-plasma ratio because of plasma trapped among the cells. These determinations proved useful in this series, however, because comparative observations were made in the same persons with minimal hematocrit changes.

  In this study total blood volume was determined instead of plasma volume because the body appears to maintain it at a constant level at the expense of the cell-plasma ratio and hence it is a more stable value. Plasma volume ap-

21  See footnote 19, p. 194.

22 Gregersen, M. I.:  Practical method for determination of blood volume with the dye T-1824. J. Lab & Clin. Med. 29: 1266-1286, Dec 1944.


Chart 23.  Effects on blood volume (expressed as changes per square meter of body surface) ofexcision of arteriovenous fistula. The shaded bars represent the variation of increase and the black bars the variation of decrease. A change of 200 cc. in blood volume is considered within the range of normal variations. Note that only 35 of the 41 patients are depicted in the chart inasmuch as readings were not obtained for 6.

parently varies rapidly to compensate forchanges in red cell volume because the body canmobilize fluid and protein with relative rapidity. Studies on theeffect of simple hemorrhage byEbert, Stead, and Gibson 23 indicate that changes in plasma volumeoccur with a slowreadjustment of the red cell volume and hematocrit reading to thenormal value. Similar changesmight be seen soon after an operation in which blood loss anddehydration have occurred. Forthese reasons plasma volume is subject to much more fluctuation than istotal blood volume, andinformation concerning it was of less value for the purposes of thisstudy than evidence ofchanges in total blood volume.

23 Ebert, R. V.; Stead, E. A., Jr., and Gibson, J. G., Jr.:   Response of normal subjects to acute blood loss, with special reference to the mechanism of restoration of blood volume.  Arch. Int. Med. 68: 578-590, Sep 1941.


  Although some of the postoperative measurements calculated on these patients were made relatively soon after operation, they were delayed long enough to escape any immediate effects of the procedure, for example, blood loss and dehydration. Only observations made 10 days or more after operation were used in the calculation of normal values; by this time almost all the patients were ambulatory. The average blood volume per square meter of body surface was approximately 2,800 cc., which is within the same range as the values reported for normal male subjects by Gibson and Evans.24

  Results.  The majority (23) of the 41 patients in this series showed a difference of less than 200 cc. between the preoperative and postoperative blood volumes per square meter of body surface. Such a change, which amounted to only about 7 percent of the normal value, might have been expected as a result of technical errors in the determinations, i. e., factors introduced by the operation and bed rest without regard to the presence of the arteriovenous fistula and its eradication. On the other hand, although in no instance was the postoperative increase in blood volume greater than 200 cc., 18 patients, almost 44 percent, showed a postoperative decrease greater than 200 cubic centimeters. In several instances serial observations showed the decline to be in progress by the third day after operation and to be apparently complete by the 10th day.

  Observations made on these patients were in agreement with previous observations to the effect that blood volume may be significantly elevated in patients with arteriovenous fistulas. This is not, however, an inevitable reaction, since elevations were observed in only 18 of the 41 patients studied. Previous studies had supplied no satisfactory explanation of the alteration, and an attempt was therefore made to relate it to other alterations in the circulation observed in patients with arteriovenous fistulas at the vascular center of Ashford General Hospital.

  The possibility that the increase in blood volume might be related to the size of the fistula was investigated. In this series 3 or 4 patients with the most extreme elevations in blood volume had fistulas which involved the iliac vessels. It was difficult, however, to establish this explanation because the exact size of the fistula could rarely be determined. Clinical evidence, such as the size of the vessels involved and the intensity of associated thrills and murmurs, supplied some data concerning the magnitude of the shunt of blood, but was obviously not accurate. Actual measurement of the fistulous opening at time of operation did not prove successful because of vascular spasm, and measurement of the specimen after removal was unsatisfactory for the same reason and because tissue fixation resulted in shrinkage.

  As was pointed out previously in this chapter, the resting level of cardiac output is above normal in many patients with arteriovenous fistulas. Since it is possible that the elevation of the cardiac output above normal may be

24  Gibson, J. G., and Evans, W. A., Jr.:  Clinical studies of the blood volume; relation of plasma and total blood volume to venous pressure, blood velocity rate, physical measurements, age and sex in 90 normal humans. J. Clin. Investigation 16: 317-328, May 1937.


proportional to the size of the fistula, anattempt was made to correlate this increase with theincrease in the blood volume. In order to avoid the difficultiesinherent in the range in basalcardiac output in the series, blood volume change was plotted againstpercentage deviation fromnormal in cardiac output. In other words, if cardiac output beforeoperation was twice the basalvalue determined after operation, an increase in cardiac output of 100percent was plotted. Thecoefficient thus obtained, 0.26, did not indicate any significantcorrelation.

  In spite of the negative results just reported, it was still thought possible that the size of the fistula might be at least one of several factors responsible for the increased blood volume observed in some patients with arteriovenous fistulas. It has been well established by roentgenologic methods that in this lesion the heart may increase in size with the development of the fistula but return to normal size after the fistula is removed. In view of the evidence already existing that the size of the heart is related to the magnitude of the blood volume, an attempt was made to correlate these measurements in the 29 patients in the series who had preoperative and postoperative teleoroentgenograms, which in some instances were serial. Changes in the blood volume were plotted against changes in the transverse diameter of the heart as measured on the teleoroentgenogram. The correlation coefficient thus obtained, 0.36, indicated that the degree of correlation between these values was not significant.

  It cannot be said that these results were surprising. Earlier studies by Weens, Brannon, and Warren 25 had shown that in normal subjects a rapid increase in blood volume of about 1,000 cc. fails to produce any distinct change in the heart size. It is true that Holman 26 was able to produce changes in the size of the heart in dogs when he increased or decreased the blood volume, but these experimental changes were of much greater magnitude than those observed clinically in arteriovenous fistulas. A change of 500 cc. in dogs would be, roughly equivalent to a change of 4,000 cc. in a human subject. In neither of the studies just mentioned was the increase in blood volume accompanied by an increase in extracellular fluid volume, which might alter the situation considerably. It seems likely that other factors than cardiac size, such as the increased demand placed on the heart, may play a major role in the increase in blood volume in arteriovenous fistulas.

An attempt to relate the increase in the blood volume to the duration of the arteriovenous fistula proved unsuccessful because few patients in this series had large fistulas of long duration. The lesions of longest duration, as already mentioned, were usually of small size, for which reason many of them had been overlooked originally by both patients and medical attendants.

  The results of the study of arteriovenous fistulas at Ashford General Hospital thus produced no explanation for the elevation sometimes observed in blood volumes with the onset of the lesion or for the return to normal levels after its eradication. This is scarcely surprising, since the basic factors con-

25  Weens, H. S.; Brannon, E. S., and Warren, J. V.: Unpublished data.
26  See footnote 10, p. 191.


trolling the size of the blood volume innormal subjects are still imperfectly understood.

  Holman 27 advanced the opinion that the increase in blood volume merely compensates for the loss of blood through the fistula.  His apparently interchangeable use of the terms blood volume and volume of blood flow somewhat confuses the explanation.

  Reid and McGuire's suggestion 28  was that the increase in blood volume seen in patients with arteriovenous fistulas represents merely the nonspecific increase in blood volume seen in patients with congestive heart failure of any type and is not the specific result of the presence of the arteriovenous fistula. Although none of the patients studied at Ashford General Hospital developed the complete clinical picture of cardiac failure, it seems possible that there might have been a similar mechanism at work in the cases in which an increase in blood volume was observed. Warren and Stead 29 pointed out that in patients with chronic congestive heart failure the increase in blood volume, as well as certain other abnormalities in fluid dynamics, could be explained by a deficient output of the heart rather than, as it is usually explained, by increased venous pressure. If this explanation is correct, it may be that the mechanism of the increase in blood volume in patients with arteriovenous fistulas is similar: In cardiac failure the circulation would be inadequate because of the inability of the heart to pump blood. In arteriovenous fistula the blood flow would be inadequate because much of the cardiac output is shunted through the fistula.

  Previously in this chapter attention was called to the similarity between cardiac output in arteriovenous fistulas and conditions such as patent ductus arteriosus and pregnancy, i. e., increase of cardiac output through alteration of circulatory dynamics. Although the mechanism by which the increase in blood volume observed in arteriovenous fistulas is attained is not fully understood, the basic explanation may be an increase in the production of blood. When some large venous channels are engorged with blood, as they are when this lesion is present, the amount of blood remaining in the rest of the vascular bed would be less than normal unless increased production of blood were stimulated.


  Slowing of the heart rate, the most commonly recognized result of temporary occlusion of an arteriovenous fistula, was first described by Nicoladoni 30

27  See footnote 11 (1), p. 191.

28  See footnote 14, p. 192.

29  Warren, J. V., and Stead, E. A.:  Fluid dynamics in chronic congestive heart failure; interpretation of the mechanisms producing the edema, increased plasma volume and elevated venous pressure in certain patients with prolonged congestive failure.  Arch. Int. Med. 73: 138-147, Feb 1944.

30  Nicoladoni, C.:  Phlebarteriectasie der rechten oberen  Extremit?t.  Arch. f. klin. Chir. 18: 252-274, 1875.


in 1875. In 1890 Branham 31 alsodescribed itwithout knowledge of the previous description,and it is now usually referred to as Branham's sign. Furtherobservations on this sign, and on theeffect on the stroke volume and cardiac output of temporary occlusionof arteriovenous fistulas,were made on a group of patients with these lesions at the vascularcenter of Ashford GeneralHospital (Table 15).


Materialsand Methods

  The investigation was carried out on 25 patients with arteriovenous fistulas of traumatic origin, none of whom presented clinical evidence of cardiac decompensation or of any complicating disease which might be expected to alter the response of the vascular system. They formed part of the group upon whom studies of the cardiac output had been made but were limited to those with lesions of the extremities.

  The patient under investigation was placed in the critically damped ballistocardiograph described earlier in this chapter and a blood pressure cuff bound about the affected extremity in such a way that when it was deflated there was no interference with the venous blood flow. The fistula was occluded by rapidly raising the pressure in the cuff to 200 mm.of mercury from a pressure bottle system. To familiarize the patient with the method and to be certain that no discomfort was connected with it, the procedure was carried out

31  Branham, H. H.: Aneurismal varix of the femoral artery and vein following a gunshot wound. Internat. J. Surg. 3: 250-251, Nov 1890.


several times before the ballistic recordswere begun. Ballistic tracings and arterial bloodpressure determinations by the auscultatory method were obtained withthe fistula open and withit occluded. At least 2 sets of observations were made on each patient.In 5 patients theobservations were repeated after 2 mg. of atropine sulfate had beeninjected intravenously andenough time had elapsed for it to exert its maximum effect (Table 16).In all patients afteroperative removal of the fistula another ballistocardiographic recordwas made.


Ballistocardiographic tracings showed that when the arteriovenous fistula was suddenly occluded, 17 of the 25 patients showed a decrease in the heart rate ranging from 4 to 32 beats per minute. In the other 8 patients a decrease also occurred but it was less than 4 beats.

  In 19 patients the stroke volume of the heart decreased more than 10 cc. when the fistula was suddenly occluded. The effect of occlusion on the cardiac index (the cardiac output in liters per minute per square meter of body surface) varied from no change at all to a decrease of 3.6 liters. In 22 of the 25 patients the decrease was 0.5 liter or more. Under the conditions of the study, 0.5 liter probably represented the minimal detectable change.

  Changes in arterial pressure following occlusion of the fistula were seldom marked. The diastolic pressure increased almost constantly; the increase in the systolic pressure was less frequent.

  In all 5 patients in whom additional studies were made after injections of atropine, an increased pulse rate was observed. In some instances this was accompanied by an increase in the cardiac index over the basal level. When the fistula was occluded, although the change in the pulse rate did not exceed 4 beats per minute in any case, stroke volume was altered in every instance. Decline in the cardiac index was from 1.0 to 2.2 liters per minute per square meter; that in cardiac output was from 1.8 to 4.2 liters (Table 16).


  The ballistocardiograph was not altogether ideal for the purposes of this study because it prevented the use of manual compression for the occlusion of the arteriovenous fistula. Since this compression would have interfered with the movements of the ballistic bed, a blood pressure cuff was used which meant that not only the fistula itself was occluded but all the circulation of the extremity distal to the cuff as well. Observations on normal subjects, however, and on the normal extremities of patients with arteriovenous fistulas elsewhere in the body, showed practically no change in the cardiac output or pulse rate when it was employed. For comparative purposes, the Matas compressor, which occludes only the major vessel entering the fistula, was tested in 1 patient. Response of heart rate and cardiac output was the same by both methods.



The results of this investigation further confirmed the well-accepted fact that closure of an arteriovenous fistula is usually accompanied by slowing of the heart rate, though the specific stimulus which results in the bradycardic response is not clearly understood, just as the mechanism producing the change is still to be elucidated. The speed of the reaction suggests that a nervous reflex may be involved. It has been suggested that the vascular bed in and around the fistula may take on a function similar to that assumed in carotid sinus or aortic depressor areas in that local pressure changes are reflected by reflex changes in cardiac function. This does not seem likely. Observations by Gerlach and Harke,32 which were confirmed by observations at the vascular center of Ashford General Hospital, show that Branham's sign can be produced when a patient is under high spinal analgesia and when the area of the fistula

32 Gerlach, F., and Harke, W.: Compression of arteriovenous aneurysms.Klin. Wehnschr. 3: 980-981, May 1924.


is presumably devoid of functional nervousconnections, though it is not certain that functionaldenervation is complete.

Lewis and Drury 33 found it possible to block the slowing of the heart rate produced by occlusion of the fistula by a preliminary administration of atropine and the same result was observed in the 5 patients in this series in which this method was used.  Ellis and Weiss,34 and Kramer and Kahn,35 made the same observations. Rieder's negative results 36 can probably be explained by the use of inadequate dosages. The effect of atropine has been interpreted as evidence that the pulse-slowing response is mediated not only by a nervous mechanism in general but by the vagus nerve in particular.

  According to the results of this investigation, occlusion of the fistula also produces a striking decrease in the stroke output of the heart. This decrease, together with the decrease in the heart rate, results in a reduction of the cardiac output. The stroke volume reduction, however, is independent of the change in the heart rate, and atropine which can abolish the change in the heart rate does not affect the reduction in the stroke volume (Chart 24).

Chart 24.  Ballistocardiographic studies in popliteal arteriovenous fistula before and after intravenous administration of 2 mg. of atropine sulfate. Note that after administration of the drug the pulse rate rises but is not slowed when the fistula is occluded, although the struke volume and cardiac index both decrease.

  Changes in the stroke volume occur very promptly, significant alterations appearing within the first three heart beats after the fistula is occluded. On theoretical grounds three possible mechanisms might be responsible: (1) a change in the emptying of the heart, (2) a variation in the degree of diastolic relaxation of the ventricles, and (3) a change in the pressure gradient producing filling

33  See footnote 2, p. 184.

34  See footnote 15, p. 192.

35 Kramer, M., and Kahn, J. H.:  Unpublished data.

36 Rieder, W.:  Heart impairment due to arteriovenous aneurysm.  Arch. f. klin. Chir. 139: 597-606, 1926.


of the ventricles. If either of the first twoprocesses is responsible, the mechanism must be undernervous control because of the rapidity with which changes in thestroke volume appear. It isnot likely that the vagus forms part of the reflex pathway sinceatropine, which suppresses thevagal control of the heart rate, does not affect the stroke volume. Ifa change in the pressuregradient causing filling of the ventricles is responsible for thechange in the stroke volume, italso occurs rapidly.Studies of catheterization of the right side of theheart carried out byNickerson, Warren, and Brannon 37 indicatethat nosignificant change occurs in the mean rightatrial pressure when the fistula is closed, though there are largechanges in the stroke volume. The conclusion that there is no change inthe pressure gradient causing filling of the heart is notnecessarily correct, however, since a pressure gradient cannot bedetermined by measuring thepressure at a single point in a system.

  Changes in cardiac output and stroke volume similar to those described in these 25 patients with arteriovenous fistulas occurred in normal subjects upon the production of large areas of reactive hyperemia. When blood pressure cuffs were placed about the upper thigh and inflated to 200 mm. of mercury pressure, little change in the heart rate and cardiac output was noted. When the pressure was continued for 20 minutes and observations were made as it was released, increases in heart rate and cardiac output were almost double the prepressure levels. These changes, as in the patients with arteriovenous fistulas, occurred without significant alteration in the atrial pressure.

  In general, the output of the heart following temporary occlusion of an arteriovenous fistula was almost the same as the output under similar conditions several weeks after operative removal of the fistula. Since the cardiac output is a highly variable function changing markedly under the influence of even the emotional content of thought, it is difficult to say whether the lack of exact coincidence in the values represented a real difference in output, dependent upon the length of time the fistula was closed, or merely represented a variation in the conditions under which the values were obtained.

  It was not possible in these studies to confirm Holman's observation38 as that proximity of the fistula to the heart is important in the cardiac output. The size of the fistula and of the vessels feeding it seemed more important than its position in determining the amount of blood flowing through it. Fistulas close to the heart were associated with both large and small increases in cardiac output.


  The concentration of casualties with vascular injuries in vascular centers in general hospitals in the Zone of Interior in World War II presented an unparalleled opportunity for the study of certain phases of arteriovenous

37  Nickerson, J. L.;  Warren, J. V., and Brannon, E. S.:  Cardiac output in man; studies with the low frequency, critically-damped ballistocardiograph and the method of right atrial catheterization. J. Clin. Investigation 26: 1-10, Jan 1947.

38  See footnote 11 (1), p. 191.


fistulas. This chapter is the record of thestudies carried out at the vascular center at the AshfordGeneral Hospital.

The investigation included studies of the local and systemic effects of arteriovenous fistulas on the circulation and of their effect on the size of the heart, the cardiac output, and the blood volume.   In many of the subjects both the cardiac output and blood volume were elevated above normal level and the size of the heart was somewhat increased. After operative eradication of the fistula there was a practically uniform return of all three to normal. None of the patients under investigation developed frank congestive heart failure.

  Studies carried out after temporary occlusion of arteriovenous fistulas included slowing of the pulse rate (Branham's sign), which is apparently mediated by the vagus nerve, and almost immediate lowering of the cardiac output and stroke volume of the heart, the latter reaction apparently being a nervous reflex related to the sudden change in resistance against which the heart must pump. The pulse pressure decreases on occlusion of the fistula because of the rise in diastolic pressure, thus returning the character of the pulse to normal.

  In spite of the information secured by these various studies, a rational and more nearly complete explanation of the genesis of the effects of an arteriovenous fistula on the circulation must await an extension of basic knowledge concerning the circulatory system.