The Nitrous Oxide TestAn Improved Method for the Detection of Left-to-Right Shunts

By ANDREW G. MORROW, M.D., F.A.C.S., RICHARD J. SANDERS, M.D.,AND EUGENE BRAUNWALD, M.D.

The diagnosis of a left-to-right shunt generally depends upon the demonstration of signifi-cant differences in oxygen content among the venae cavae, the chambers of the right heart,and the pulmonary artery. The inconclusive or misleading results sometimes obtained withthis method have stimulated interest in improved technics. The wide arteriovenous dif-ference that normally exists during the inhalation of nitrous oxide has been studied in thedevelopment of a more accurate diagnostic approach. The superiority of the nitrous oxidetest over the oxygen method is demonstrated.

IN THE past, the detection of intracardiacand extracardiac left-to-right shunts has

been based upon differences in the oxygencontent of blood obtained from the venaecavae, right atrium, right ventricle, and pul-monary artery. Since relatively large dif-ferences in oxygen content may exist inthese areas in subjects without shunts,1-3 theoxygen method sometimes provides inconclu-sive or even misleading diagnostic informa-tion. The inhalation of nitrous oxide, allinert foreign gas, and the subsequent meas-urement of its concentration in arterial andright heart blood provides a new diagnosticapproach in which the difficulties inherent inthe oxygen method are largely obviated. Thiscommunication presents the theory, technic,and diagnostic applications of the nitrousoxide test, and a comparison of the nitrousoxide and oxygen methods in the detectionof left-to-right shunts. The clinical study tobe presented is an extension of the technicdevised and experimentally evaluated byCallaway.4'5Kety and Schmidt6 observed that immedi-

ately following nitrous oxide inhalation,cerebral tissue absorbed large quantities ofthe gas and a substantial arteriovenous dif-ference existed until tissue saturation nearedcompletion. Figure 1 demonstrates that asimilar arteriovenous difference exists when

From the Clinic of Surgery, National Heart Insti-ttite, 'Bethesda, Md.

VOL.% _

N202.0

I.0

00 2 4 6 8 10

TIME -MINUTES

FIG. 1. Nitrous oxide levels in the femoral arteryand pulmonary artery of a patient without a left-to-right shunt. Fiftecn per ecit nitrous oxide was in-lhaled for 10 minutes.

pulmonary artery blood is used instead ofjugular vein blood. During the first minuteof inhalation the concentration of nitrousoxide in right heart blood is low, since itsappearance is delayed by tissue absorption.In the presence of a left-to-right shunt, bloodfrom the left heart, rich in nitrous oxide,raises the level in right heart blood therebymaking the detection of shunts possible.

MATERIAL AND METHODSA total of 208 satisfactory nitrous oxide tests

in 184 patients in whom the diagnosis was con-lirmed, forms the basis of this report. Most of

Circulation, Volume XVII, February 1958284

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

NITROUS OXIDE TEST

the patients were adults; 53 had left-to-rightshunts, 43 at the atrial, 5 at the ventricular, and5 at the pulmonary artery level. The diagnosisin the patients with shunts was established atoperation in 29, by the passage of a catheteracross the defect into the left heart in 19, and byretrograde thoracic aortography or dye-dilutionstudies with left heart injection in 5.7 The 131control patients without shunts all had valvularrheumatic heart disease. This diagnosis wasestablished by clinical examination as well as rightand left heart catheterization.8 ' More than halfof these patients were operated upon and the ab-sence of a shunt was further confirmed. In all 53patients with shunts and in 96 of the 131 patientswithout shunts right heart blood samples wereanalyzed for oxygen content.

Technic. Nitrous oxide tests were performed byplacing the tip of a cardiac catheter in the rightatrium, right ventricle, or pulmonary artery anda needle in a peripheral artery. Specimens fornitrogen blank determination were obtained beforeadministration of nitrous oxide. The patient wasinstructed to breathe a mixture of 15 per centnitrous oxide, 21 per cent oxygen, and 64 per centnitrogen as deeply and as rapidly as possible. Thegas was administered with a mask or mouthpieceand a Collins 3-way respiratory valve. Patientsunder general anesthesia were hyperventilated bymanual compression of the breathing bag. Five-or 10-ml. blood samples were then simultaneouslydrawn at a constant rate from the right heart andperipheral artery during the 60 seconds of gasinhalation. Early in the series another sampleproximal to any shunt was collected from the venaeava or a peripheral vein. As will be indicated,this sample was found unnecessary. A 10-minuteperiod for desaturation was allowed when the testwas to be repeated in another chamber. A double-lumen catheter was sometimes employed to shortenthe procedure by sampling from 2 chambers simul-taneously. Blood samples for oxygen analysiswere obtained after the nitrous oxide tests hadbeen completed. In general, 3 samples each weredrawn from the pulmonary artery, right ventricle,and right atrium and 2 from each vena cava. In40 of the 149 patients in whom specimens foroxygen content were obtained, the inferior venacava could not be entered and in 10 patients pul-nmonary artery blood was not sampled. Theoxygen contents to be presented in figure 4 rep-resent the mean values of all specimens obtainedin each site. The mixed vena caval oxygen contentin the 109 patients in whom both cavae were en-tered was calculated by weighting the inferior cavalvalues as twice the superior caval values.

Oiled, heparinized syringes were employed andair contamination was prevented with mercury-filled syringe caps. Nitrous oxide content was

450I--o: Non. Shunt Ptemnt

4 0 |*-Sunt Potient

300

2 50

VOLUME4200

150

50N

A Pe.ipheryo Ve B VWnoCoy

FIG. 2. Nitrous oxide samples from the venoussystem, right heart, and a peripheral artery in pa-tients with and without left-to-right shunts. Nitro-gen blank values have been subtracted.

determined by a modification of the Van Slykeinanometric method.0' 10 Oxygen contents were de-termined by the method of Van Slyke andMeNeill.`'

RESULTSNitrogen Blanks. Since the manometric

method does not distinguish dissolved nitro-gen from nitrous oxide, blank samples arenecessary to determine the nitrogen contentof the blood prior to each nitrous oxide test.No systematic difference was found amongnitrogen blank values in systemic artery, pe-ripheral vein, or right heart blood in anygiven patient. The average of the 2 or 3blank values in each patient was thereforeused in all calculations. This average valuefor all patients ranged from 0.50 to 1.97 vol-ume per cent with a mean of 1.30 + 0.18volume per cent. No difference in blankvalues existed between control patients andthose with shunts. Six patients in whom thedifference between simultaneous blanks ex-ceeded 0.40 volume per cent were excludedfrom the study, since this large differencemade sampling or analytic technics suspect.

Nitro Is Oxide Sample. The nitrous oxidesample, as used in this presentation, refers tothe nitrous oxide content of the specimen lessthe blank value. In the control patients therange and scatter of nitrous oxide sampleswere widest in the peripheral vein and pro-gressively narrowed as the pulmonary arterywas reached (fig. 2).

285

11

a

CRightAtrium D.PulmmmyAt.. E-FcowoiocRt Ventrielea Bmchial Ad

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

MORROW, SANDERS, AND BRAUNWALD

450 F

ARTERALF 00Fo3C

300 - 0 0

aMc)2 C-0

so L-20

400

0005 20%

r

4000 00

ARTERIAL 0SAMPLE con /

VOLUME% 2 0S °pN420 230o o °

200 0°

0so 0 10

-20 0 so 100

PAl

000Rj n Shu P 400nt

W PAShunt Patjc.

" S . ~~~~~PAof RV

|52o00o 2 so 300 550 400

FIG. 3. A, Top. Relation between right atrial andarterial nitrous oxide samples in patients with andwithout left-to-right shunts. Broken diagonal lines,RA/A ratios. B. Bottom. Relation between pulmonaryartery or right ventricular and arterial nitrous oxidesamples in patients with and without shunts.

As would be anticipated, the patients withleft-to-right shunts had peripheral venous andvena caval samples similar to the controlpatients (fig. 2 A and B). However, 39 ofthe 43 patients with shunts at the right atriallevel had right atrial nitrous oxide samplesbetween 1.06 and 3.39 volumes per cent. Thehighest right atrial sample among the controlpatients was 0.88 volume per cent. Thus, allbut 4 of the patients with shunts had samplesgreater than the right atrial sample of any ofthe 83 control patients (fig. 2C). All 24 pa-

tients with shunts who had pulmonary arteryor right ventricular tests had samples exceed-ing 0.64 volume per cent; the highest pul-monary artery or right ventricular sample ofthe 58 control patients was 0.47 volume percent (fig. 2D). The systemic artery nitrousoxide samples ranged from 1.60 to 4.62 vol-umes per cent with a mean of 3.04 0.59volumes per cent. No systematic difference

-40 -30 -20 -10 0 to 2.0VOL. 02

30 4.0 5.0

FIG. 4. Differences in oxygen content between rightheart chambers and venae cavae in patients with andwithout shunts. IVC (-) SC average = 0.94 +1.51 volume percent (all patients). RA (-) Mixed VCaverage = (-) 0.41 ± 0.96 volume per cent (non-shunt patients). RV (-) RA average = 0.01 ± 0.95volume per cent (non-shunt patients). PA (-) RVaverage = 0.23 ± 0.53 volume per cent (non-shuntpatients).

existed between the arterial nitrous oxidelevels of patients with and without shunts(fig. 2E).Relation between Arterial and Right Heart

Nitrous Oxide Samples. In control patientshigh arterial nitrous oxide levels, reflectingbetter nitrous oxide uptake were, in general,associated with slightly higher right heart lev-els (fig. 3A and B). In patients with shuntsthe right heart nitrous oxide level distal tothe shunt is determined to some extent by thevenous level but primarily upon the quantityof shunted blood and its nitrous oxide content.Since the nitrous oxide content of shuntedblood is identical to that of arterial blood, theratio between right heart and arterial levelsbecomes meaningful and is therefore relatedto the magnitude of the shunt. In the 83control patients, the ratio between right atrialand arterial samples (RA/A X 100) was lessthan 30 per cent in all but 1, in whom it was31 per cent. In 41 of the 43 patients withright atrial shunts the RA/A ratio was30 per cent or more. The other 2 patients hadratios of 29 per cent (fig. 3A). In the 58 con-trol patients with tests in the pulmonary ar-tery or right ventricle the highest PA/A orRV/A ratio was 16 per cent. All 24 patients

286

200

aa

woos

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

NITROUS OXIDE TEST

1° T |TEST+|

&.s NON SHUNT1E *.SWUNT,

a

U a-a:a:a a

a N2N TEST+

* A |N TEST-

A 2TEST

WO2 TIESTÆ

[N2 TEST+|+ Pmu0|NO0TEST-| 0- 0 0 0

o A oa Du%

A A 11 |

_ 0 0,

a %AaCS02aqRfiTEST

-2.0 -1.0 0 1.0 2.0 3.002 STEP-UP INTO RIGHT ATRIUM-VOL.1

N20TEST

(RA)

AND

(RAV

-lo% I.L Lp4.0 5.0 60 -2.0 -1.0 0 1.0 2.0 3.0 4.0 5.0 6.0%02 02 STEP-UP INTO PULM ART AND RIGHT VENTRICLE-VOL%02

FIG. 5. A, Left. Comparison of oxygen differences with atrial nitrous oxide ratios in patients withand without left-to-right shunts in whom both tests were performed. B, Right. Oxygen differencescompared with nitrous oxide ratios in the right ventricle or pulmonary artery in patients with andwithout shunts.

with shunts had a PA/A or RV/A ratio of 20per cent or more (fig. 3B).Oxygen Samples. Figure 4 presents the dif-

ferences in oxygen content between contigu-ous sites of sampling. The oxygen content ofinferior caval blood exceeded that of superiorcaval blood in 83 of 109 patients (fig. 4A). Incontrol patients the range and scatter of oxy-

gen content between consecutive cardiac cham-bers fell progressively as the pulmonary ar-

tery was reached. This range of differencewas greatest between venae cavae and rightatrium (fig. 4B). It diminished slightly be-tween right atrium and right ventricle (fig.4C), and was smallest between right ventricleand pulmonary artery (fig. 4D). These dataconfirm previous observations indicating in-complete mixing of blood in the cavae andright atrium.1-3

In all but 1 of the 43 patients with left-to-right shunts at the atrial level the oxygencontent of right atrial blood exceeded that

of vena caval blood. However, in 11 of these

43 patients the oxygen content of right atrial

blood did not exceed that of caval blood by1.5 volumes per cent or more. Among 93 con-

trol patients 6 had a right atrial blood oxygen

content exceeding caval content by 1.5 vol-umes per cent or more (fig. 4B). Two of the 5patients with left-to-right shunts at the rightventricular level and 1 of the 5 patients withshunts into the pulmonary artery did nothave oxygen differences exceeding 1.0 volumeper cent (fig. 4C). Among 94 control patients,11 had increases of 1.0 volume per cent or

more between the right atrium and right ven-

tricle and 4 of 86 controls had increases of thismagnitude between the right ventricle andpulmonary artery.Comparison between the Oxygen and Nitrous

Oxide Methods. In figure 5A the differencesin oxygen content between vena caval andright atrial blood have been plotted againstthe nitrous oxide RA/A ratios in the 122 pa-tients in whom both tests were performed.For purposes of comparison, an oxygen in-crease of 1.5 volumes per cent between thecaval and right atrial blood and an RA/A

100% H

90% -

80% h

287

70%

60%ATRIALN20TEST

(f)50% h40%

30%

20%

10%

0

-10% Oh-3.I

a

A

a:A* 1

A

0

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

MORROW, SANDERS, AND BRAUNWALD

ratio of 30 per cent have been employed asstandards for the diagnosis of a shunt at theatrial level. With these criteria there were5 false positive tests with the oxygen methodand 1 false positive test with the nitrousoxide method in the 79 control patients. Inthe 43 patients with proved shunts at theatrial level there were 11 false negative testswith the oxygen method and 2 false negativenitrous oxide tests. Thus, the oxygen methodresulted in 16 diagnostic errors and thenitrous oxide method in 3.

It is also apparent from figure 5A that all3 diagnostic errors in the nitrous oxide testsoccurred iil patients with RA/A ratios be-tween 29 and 31 per cent, close to the diagnos-tic level set at 30 per cent. In contrast, 9 of the15 diagnostic errors in the oxygen method oc-curred in patients in whom the oxygen dif-ferences were more than 0.5 volume per centfrom the somewhat arbitrary diagnostic line.A (liference of 1.5 volumes per cent was se-lected because it is the standard generallyaccepted and represented a judicious balancebetween false positive and false negative oxy-geim results. It may be noted that if an oxy-gen increase of 1.0 volume per cent had beenemploye(l as the diagnostic criterion therewould be a total of 13 diagnostic errors, andif 2.0 volumes per ceimt was chosen, therewould be 22 errors.

In figure 5B the differences in oxygen con-tent between right atrial and right ventricularblood and between right ventricular and pul-monary artery blood are plotted against thenitrous oxide ratios in the 27 patients who hadboth tests. Diagnostic criteria selected forthese areas were an oxygen increase of 1.0volume per cent or more and an RV/A orPA/A ratio of 20 per cent or more. In the17 control patients there were 3 false positivetests with the oxygen method and none withnitrous oxide. In the 10 patients with shuntsthere were 3 false negative tests with the oxy-gen method and none with nitrous oxide. Thus,in these 27 patients the oxygen method re-sulted in 6 diagnostic errors and the nitrousoxide test in none.

DISCUSSION-Errors in the performance of the nitrous

oxide test relate to the administration anduptake of the gas as well as to the collectionand chemical analysis of the blood samples.An adequate uptake of nitrous oxide is essen-tial but may be prevented by a large leakaround the mouthpiece, hypoventilation, orsevere pulmonary parenchymal disease. Whenthe test is performed in an anesthetized pa-tient, the anesthesiologist must provide hyper-ventilation and of course cannot employnitrous oxide as the anesthetic agent prior tothe test.

Small differences in the nitrous oxide con-tent of superior caval, inferior eaval, and cor-onary sinus blood and and incomplete mixingof these streams in the right atrium intro-duce an element of variability into the resultsof the test. However, this source of error isnot so great as in the oxygen method, sincethe differences in the oxygen content of theright atrial tributaries are greater than thevariations in their nitrous oxide content.An important potential source of error is

improper positioning of the catheter. Whenithe test is performed in the right atrium, thecatheter tip must be placed close to the tri-cuspid valve to insure that the sample will bedrawn from an area distal to the entrance ofany shunt at this level. Should the catheterinadvertently be passed through a defect intothe left atrium, a nitrous oxide sample equalto the arterial will be obtained. Another testin the pulmonary artery or right ventricle willindicate whether a true atrial septal defectexists or whether the catheter passed througha nonshunting patent foramen ovale.

Contamination of the blood samples by airand excessive suction on the sampling syringeshould be avoided, since either can falselylower the nitrous oxide content of the sample.Blood must be drawn at a constant rate dur-imig the sampling period, sinee both the ar-terial and venous nitrous oxide contents arechanging throughout this period (fig. 1). Thedifference between duplicate Van Slykenitrous oxide determinations in this laboratory

288

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

NITROUS OXIDE TEST

is ordinarily less than 0.10 volume per cent.Since the errors inherent in the technics of

sampling and analysis are independent ofthe arterial nitrous oxide level, they assumegreater significance when the ratio is calcu-lated with a low arterial sample. For ex-ample, an error of 0.20 volume per cent in aright heart sample represents 10 per cent ofan arterial level of 2.00 volumes per cent, butonly 5 per cent of an arterial level of 4.00volumes per cent. Fifteen nitrous oxide testswere eliminated from the series because ofarterial levels less than 1.50 volumes per cent.Seven of these unsatisfactory tests were at-tributed to hypoventilation. The other 8tests were performed in cyanotic patientswith right-to-left shunts. In these patientspulmonary venous blood, with a high nitrousoxide level, was presumably diluted by rightheart blood with a low level and thereby low-ered the arterial content. Thus, in the pres-ence of a large right-to-left shunt, the ar-terial nitrous oxide content may not ac-curately reflect the content of blood that hasbeen shunted from left to right. Therefore,the right heart arterial nitrous oxide ratio(fig. 3) can no longer be relied upon. How-ever, in such circumstances, the absolute levelof nitrous oxide in right heart blood (fig. 2)may still be useful in the detection of left-to-right shunt.The vena caval and peripheral venous

nitrous oxide samples were originally ob-tained for comparison with the sample fromthe right heart, much as caval and rightatrial oxygen contents are compared. How-ever, it was apparent that in the absence ofa shunt the nitrous oxide content of rightatrial blood was relatively constant and lessthan 30 per cent of the arterial content (fig.3). When this maximum value had beenestablished, it was no longer necessary tosample the cava or peripheral vein. Further-more, the inclusion of caval and peripheralvenous samples did not improve the diagnosticaccuracy of the method.The superiority of the nitrous oxide test

over the oxygen method, as demonstrated in

figure 5, is related to several factors. Theoxygen content of right heart blood variesmarkedly among different individuals and inthe same subject at different times. There-fore, blood proximal to any shunt must al-ways be sampled. This may lead to errorsin the diagnosis of atrial septal defects be-cause caval blood is poorly mixed and lam-inar flow is present, particularly from therenal veins.12 The sampling of renal veinblood with its high oxygen content may maskthe presence of a shunt into the right atrium.In some patients the catheter cannot bepassed into the inferior vena cava. It hasbeen demonstrated (fig. 4A) that the oxygencontent of superior caval blood is generallylower than inferior caval blood. Hence, acomparison of superior caval and atrial bloodcould lead to the false diagnosis of an atria]shunt. In 3 of the 5 false positive oxygentests only the superior cava could be sampled(fig. 5A). Further error in the oxygenmethod is introduced by the variations in thepatient's physiologic state that occur duringthe time required for complete sampling.This source of error is also obviated with thenitrous oxide tests, since the 2 samples re-quired are drawn simultaneously. An addedpractical advantage of the nitrous oxide overthe oxygen method is that fewer samples andanalyses are necessary.

Modifications of the nitrous oxide test in-volving different gas concentrations, sam-pling periods, and ventilatory states are be-ing investigated. The test may be furtherimproved by the use of other inert gases thatare presently being evaluated. Since it isapparent that the nitrous oxide ratio is afunction of the contribution of the shunt tototal pulmonary flow, consideration is beinggiven to the application of nitrous oxide in thequantitation as well as the detection of left-to-right shunts.

DIAGNOSTIC CRITERIAThe diagnostic criteria currently employed

in the application of the nitrous oxide testmay be summarized. In the right atrium,

289

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

(MORROW, SANDERS, AND BRAUNWALD

an RA/A ratio above 30 per cent is consid-ered diagiiostic of a shunt. An RA/A ratiobelow 25 per cent indicates the absence of ashunt; ratios between 25 per cent and 30 percent are considered inconclusive. A PA/Aor RAT/A ratio above 20 per cent is diagnos-tie, of a shunt, while the absence of a shuntis indicated by a ratio below 15 per cent.Pulmonary artery or right ventricular testswith ratios between 15 per cent and 20 percent are considered inconclusive. An arteriallevel of at least 1.5 volumes per cent is neces-sary for a valid test.

SUMMARY

An improved method for the detection ofleft-to-right cardiac shunts is presented. Fif-teen per cent nitrous oxide is inhaled for 1minute while integrated blood samples aredrawn simultaneously from the right heartand a systemic artery. The technic, diagnos-tic criteria, and sources of error of thismethod are presented in detail.

In 41 of 43 patients with proved shunts atthe atrial level, the ratio of right atrial toarterial nitrous oxide content (RA/A) ex-ceeded 31 per cent, the highest value observedin 83 control patients. In all 24 patients withproved shunts the ratio PA/A or RV/A ex-ceeded 20 per cent. The highest PA/A orRV/'A ratio in 58 patients without shuntswas 16 per cent.The nitrous oxide test was demonstrated to

be distinctly superior to the oxygen method.In 149 patients in whom both tests were per-formed, there were 22 diagiiostic errors onthe basis of oxygen differences and 3 errorswith the nitrous oxide test.

SUMMARIO IN INTERLINGUA

Es presentate un meliorate methodo prole detection de derivation cardiac sinistro-dextere. Oxygen nitrose in un concentrationde 15 pro ceiito es inhalate durante 1 minuta,e simultaneemente specimens de sanguine in-tegrate es obtenite ab le corde dextere e abun arteria systemic. Le technica, le criterios

diagnostic, e le eausas de error in istemethodo es presentate in detalio.

In 41 ex 43 patientes con demonstrate de-rivationes al nivello atrial, le proportion in-ter le oxydo nitrose dextero-atrial e le oxydonitrose arterial excedeva 31 pro cento (levalor maximal observate in 83 patientes decontrolo). In 24 ex 24 patientes con demon-strate derivationes le proportion inter le oxy-do nitrose pulmono-arterial e le oxydo nitrosearterial o inter le oxydo nitrose dextero-ven-tricular e le oxydo nitrose arterial excedeva20 pro cento. (Le valor maximal observatein 58 patientes sin derivationes esseva 16pro cento.)Le test a oxydo nitrose se monstrava clar-

mente superior al methodo a oxygeno. In149 patientes in qui ambe tests esseva execu-tate, 22 errores diagnostic haberea resultatedel test a oxygeno, 3 del test a oxydo nitrose.

REFERENCES

1. COURNAND, A., RILEY, R. L., BREED, E. S.,BALDWIN, E. DEF., AND RICHARDS, D. W.:Measurement of cardiac output in man, usingthe technique of catheterization of the rightauricle or ventricle. J. Clin. Invest. 24: 106,1945.

2. WARREN, J. V., STEAD, E. A., JR., AND BRAN-NON, E. S.: The cardiac output in man: Astudy of some of the errors of right heartcatheterization. Am. J. Physiol. 145: 458,1946.

3. DEXTER, L., HAYNES, F. W., BURWELL, C. S.,EPPINGER, E. C., SAGERSON, R. P., AND EVANS,J. M.: Studies of congenital heart disease: II.J. Clin. Invest. 26: 554, 1947.

4. MORROW, A. G.: New methods in the diag-nosis of interatrial septal defects. Cardio-vascular Surgery, International Symposiumat the Henry Ford Hospital. Philadelphia,W. B. Saunders Co., 1955, p. 345.

5. CALLAWAY, J. J.: Atrial septal defects: Theuse of nitrous oxide in the diagnosis. To bepublished.

6. KETY, S. S., AND SCHMIDT, C. F.: Determina-tion of cerebral blood flow in man by use ofnitrous oxide in low concentrations. Am. J.Physiol. 143: 53, 1945.

7. BRAUNWALD, E., TANENBAUM, H. L., ANDMORROW, A. G.: Dye-dilution curves fromleft heart and aorta for localization of left-to-right shunts and detection of valvular in-

'2()()

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

NITROUS OXIDE TEST

sufficiency. Proc. Soc. Exper. Biol. & Med.94: 510, 1957.

8. Ross, J., JR., BRAUNWALD, E., AND MORROW,A. G.: Clinical and hemodynamic observa-tions in pure mitral insufficiency. Am. J.Cardiol. In press.

9. MORROW, A. G., BRAUNWALD, E., HALLER, J.A., JR., AND SHARP, E. H.: Left heart cathe-terization by the transbronchial route: Tech-nique and applications in physiologic anddiagnostic investigations. Circulation 16:1033, 1957.

10. KETY, S. S.: Quantitative determination ofcerebral blood flow in man. Methods in Med.Res. 1: 204, 1950.

1:1. VAN SLYKE, D. D., AND M11CNEILL, J. M.: Thedetermination of gases in blood and other so-lutions by vacuum extraction and manometricmeasurement. J. Biol. Chem. 61: 523, 1924.

12. SIROTA, J. H., AND GORDON, A. J.: Methodto measure regional contributions to totalcardiac output in man with observations onlaminar flow in the great veins. J. Appl.Physiol. 6: 485, 1954.

9i

Surawicz, B., Braun, H. A., Crum, W. B., Kemp, R. L., Wagner, S., and Bellet, S.:Clinical Manifestations of Hypopotassemia. Am. J. M. Se. 233: 603 (June), 1957.The frequency of signs and symptoms of hypopotassemia was investigated in rela-

tionship to plasma potassium levels with evaluation of the clinical status of the patientsbefore and after administration of potassium salts. During a 7 month period 557 casesmanifesting hypokalemia were encountered in 2786 potassium determinations. Theincidence of plasma potassium levels below 3.5 mEq. per liter was higher in females(22 per cent) than in males (18.3 per cent). The mortality of patients with hypo-potassemia was 49 per cent compared to that of the total hospital fatality rate of 15per cent. Fifty patients were selected for detailed study of serum electrolytes, causesof depletion and other clinical features. Among the disorders associated with potassiumdepletion, hepatic cirrhosis headed the list. Inadequate diet, infusion of potassium-freesolutions, vomiting or gastrointestinal suction, diarrhea and renal disease were factorsoperating to lower the plasma potassium levels. Patients with the lowest potassium levelsalso had the lowest concentrations of chlorides and calcium and were more often alkalotic.The clinical abnormalities encountered in these patients were of the type seen in seriouslyill patients with or without hypopotassemia. In order to determine which signs andsymptoms were attributable to potassium deficiency, a rapid infusion of potassium wasemployed. The most significant changes accompanying infusion were improved mentalstatus and increased peristaltic activity. A decrease in activity of deep tendon reflexesshowed some correlation with the decrease in plasma potassium levels. The nonspecificnature of the clinical aspects of hypopotassemia in seriously ill patients is emphasizedindicating the importance of electrocardiographic and electrolyte determinations in thediagnosis of this disorder.

SHUMAN

291

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from

ANDREW G. MORROW, RICHARD J. SANDERS and EUGENE BRAUNWALDLeft-to-Right Shunts

The Nitrous Oxide Test: An Improved Method for the Detection of

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 1958 American Heart Association, Inc. All rights reserved.

75231is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TXCirculation

doi: 10.1161/01.CIR.17.2.2841958;17:284-291Circulation. 

http://circ.ahajournals.org/content/17/2/284located on the World Wide Web at:

The online version of this article, along with updated information and services, is

  http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer

of the Web page under Services. Further information about this process is available in thewhich permission is being requested is located, click Request Permissions in the middle columnClearance Center, not the Editorial Office. Once the online version of the published article for

can be obtained via RightsLink, a service of the CopyrightCirculationoriginally published in Requests for permissions to reproduce figures, tables, or portions of articlesPermissions:

by guest on June 12, 2018http://circ.ahajournals.org/

Dow

nloaded from