Journal of Physiology (1989), 419, pp. 509-517 509With 2 text-figuresPrinted in Great Britain

THE STEADY-STATE AND REBREATHING METHODS COMPAREDDURING MORPHINE ADMINISTRATION IN HUMANS

BY DENIS L. BOURKE AND ANTHONY WARLEYFrom the Departments ofAnesthesiology and Pulmonary Medicine, University of

Texas Health Science Center at Houston, Houston, Texas 77030, and the DepartmentofAnesthesiology and Critical Care Medicine, The Johns Hopkins Hospital,

Baltimore, Maryland 21205-2181, USA

(Received 29 June 1989)

SUMMARY

1. We examined the quantitative and qualitative differences between the steady-state and rebreathing methods of determining C02-response curves before and aftereach of two doses of intravenous morphine (0 07 and 0-14 mg kg-') in four healthymale humans.

2. During each study session steady-state and rebreathing C02-response curveswere determined as an ordered pair (separated by 15 min). Carbon dioxide-responsecurves were determined for control, after 0 07 mg kg-' morphine, and after a total of0-21 mg kg-1 morphine. Each subject was studied on a second occasion when theorder of the C02-response pairing was reversed.

3. The results are discussed and related to a model that may account for thedifferences based on the step increase in CO2 at the onset of rebreathing, the rate ofrise of CO2 during rebreathing and the time constant for the central chemoreflex.

4. Our empirical conclusion is that morphine causes a parallel right shift of thesteady-state CO2-response curve and causes a non-specific decrease in the slope of therebreathing C02-response curve. We suggest that the parallel shift of the steady-state C02-response curve is specific to drugs acting on opioid receptors.

INTRODUCTION

The steady-state (SS) technique has been considered the standard for assessingCO2-mediated control of respiration (Read & Leigh, 1967; Linton, Poole-Wilson,Davies & Cameron, 1973; Dempsey, 1976; Jordan, 1982). It has been largelyreplaced, however, by Read's rebreathing (RB) technique (Read, 1967), primarilybecause of the rapidity with which a C02-response curve (VE, C02) can be determined.Using the RB technique, VE, CO2s can be determined in about 5 min compared witha minimum of 20 min for the SS technique. Although there has been somecontroversy over the comparability of the two techniques (Linton et al. 1973;Irsigler, Stafford & Severinghaus, 1984; Berkenbosch, DeGoede, Olievier &Schuitmaker, 1986; Berkenbosch, Bovill, Dahan, DeGoede & Olievier, 1989),differences in VE,CO2s determined under normal conditions by the two techniquesappear inconsequential. VE, C02 determined by the RB technique are usually shiftedMS 7618

D. L. BOURKE AND A. WARLEY

to the right (5-S Torr) and have a slightly greater slope thanVE,CO2s determined bytheSS technique (Read & Leigh, 1967; Dempsey, 1976; Jordan, 1982). Further, ithas been assumed that changes in VE,Co2s caused by various treatments (drugs,exercise, etc.) are similar both quantitatively and qualitatively regardless of whichtechnique is used.Many authors report that analgesic doses of narcotics cause a decrease in the slope

of theVE,Co2 when measured by the RB technique (Jennett, Barker & Forest, 1968;Rebuck, 1976; Rigg, 1978). Our experience and that of others indicates thatwhen measured by theSS technique, narcotics cause a parallel right shift in the

VE,C02 (Prescott, Ransom, Thorp & Wilson, 1949; Loescheke, Sweel, Kough &Lambertsen, 1953; Jordan, 1982). In fact, a valuable feature of the 88 technique isthat narcotics are distinguished by causing a parallel right shift while non-narcoticCNS depressants cause a decrease in the slope of the VEC02Our study was designed to determine whether differences could be observed

between the RB andSS techniques in healthy young males after receiving morphine.

METHODS

With institutional review and approval, four informed and consenting healthy male voluniteersbetween the ages of18 and 32 years participated in this study. Each subject was studied oIn twoseparate occasions at least1 week apart. The only differencebetween the two studv sessions wasthe sequence of techniques used throughout a single study session to determine the 1',EcO2s.Throughout all study sessions, we attempted to maintain a comfortable but low level of externalstimulation (auditory, visual, etc.).RB determinations of the VE,C02 were made using a modification of Read's method (Rebuck,

1976). SS determinations of the VE co2 were made by first adding 4% CO2 to inspired gasses for 10min, recording expired minute ventilation (VE) and end-tidal CO; (PET,Co2) values between 10 and12 min, then increasing inspired CO2 to 7% and again recording VE andPET, co2 after10 mi.Each study session began at 8.00 a.m. Subjects had fasted for the previous 8 h and refrained

from drug use (including tobacco and caffeine) for the previous 24 h.Subjects were seated in acomfortable semi-recumbent position. After applying a blood pressure cuff, ECG electrodes and aprecordial stethoscope, an intravenous infusion of normal saline was begun. Subjects then breathedthrough a breathing circuit for a 13 min acclimatization period.The breathing circuit (Fig. 1) was designed to permit the determination of VE, co s by either the

RB orSS technique. Respired gasses were sampled at the mouth piece. Oxygen, inspired CO2 andPET,CO2 were continuously measured and recorded (Gould Godert Capnograph, Holland). Expiredtidal volume (VT) and respiratory rate (f) were continuously measured and recorded (Mled-ScienceWedge Spirometer*, USA). Precordial sounds and ECG were monitored and blood pressure wastaken every 5 min.

After the 13 min acclimatization period, resting VE (rVE) and PET,C2 were recorded for 2 min.Next, a VE CO2 curve was determined by either the SS or RB technique after which the subjectcontinued to breathe through the circuit for 13 min with 0 00% inspired CO2. Then 2 min of restingrespiratory measurements were made and another VE,C02 determined by the alternate techniique.The order of the VE,o2 determination techniques was maintained throughout a subject's studysession and reversed when he returned for his second study session.

Following the second VE, co2 determination, each subject received 007 mg kg-1 morphinesulphate (MS) intravenously. After a 10 min rest period, the subject again began breathing throughthe circuit for 10 min. At this time, unstimulated VE andPET, C02 observations were made. Then theidentical sequence of events that was used for the first two VE, co determinations was repeated.

Following the fourth VE co2 determination, the subject was given 0-14 mg kg-' MS intravenouslyand the sequence of events that had followed the earlier morphine administration was repeated. Atthe end of this period, the subject remained for observation for 1 h and was then permitted to leave.After at least one week, the subject returned for a second study session, identical to the first exceptthat the order of the two VE, co2 determination techniques was reversed.

510

STEADY-STATE AND REBREATHING METHODS

The sequence for subjects 1 and 3 was SS followed by RB during the first session and RB followedby SS during the second session. This sequence was reversed for subjects 2 and 4.

Ventilation and CO2 data were corrected to BTPS (body pressure and temperature whensaturated with water vapour). VE, CO2 lines for RB data were calculated using linear least-meansquares regression. VE CO2 lines for SS data were calculated as the straight line passing through thetwo points determined during breathing 4 and 7 % C02. Statistical analysis included regressionanalysis, analysis of variance and paired t tests with correction for multiple comparisons (Snedecor& Cochran. 1980). The probability of a type I error being less than 005 was considered to besignificant. Power analysis estimates ranged from 0-80 to 0 93. Results are reported as mean + S.D.unless otherwise noted.

Fig. 1. Breathing circuit for determining both SS and RB VE, co2s A, subject mouthpiece. B,control value, C, Sadd valve. D. Lloyd valve. E. mixing chamber. F, Wedge spirometer.H, 3-way valve. I, rebreathing bag.

RESULTS

All subjects tolerated the study sessions well. Heart rate (HR) and blood pressure(BR) remained within 20% of control throughout. There were no abnormal ECGevents. In all instances subjects were able to leave the lab within 1 h of the end ofthe study.

Unstimulated, or resting, VE (rVE) at the beginning of the study was 651 ±104 1min'1 and PET, CO2 was 433 ± 2@1 Torr. Data for rVE preceding and between each pairof VE, CO2S are shown in Table 1. At no time was there a significant difference in therVE observed before a pair of VE, co2S or the rVE observed between members of thepair. All rVE measurements were below control value after the first and final doses ofMS.-PET, CO2 followed the same pattern in the opposite direction.

VE, CO slopes and the interpolated PET, CO at 15 1 min-1 (P15) are shown in Table 2.VE, co2S are plotted for comparison in Fig. 2. Control VE, CO2 slopes for both SS and RBVE,CO2 were within the normal range (Dempsey, 1976; Irsigler, 1976; Rebuck, 1976;Goodman & Curnow, 1985). Regardless of VE, CO2 sequence, control RB VE, co2S were

511

D. L. BOURKE AND A. WARLEY

TABLE 1. Resting PE and PET CO, preceding each VE, CO2rVE (1 min-')

SS RB

PET, CO2 (Torr)

SS RB

6-88+1 09 6-61 + 0-85 43-7 + 27 434+ 2-25-86+0-73 6-69+1-47 43-4+P13 42-6+1-7

5.50+0.71* 4.98+0.74* 46-1 +2'1 46-4+ 1j7*5-66+0-73 5-61 +0.59* 45 9+ 18 46-1 +2.1*

4 61 + 0.70* 5-25+ 1.13* 48-1 + 2.9* 47-8+ 35*4-72 + 0.99* 515 + 039* 49 0+3 7* 48-7 + 29*

Values are means+ S.D. * Indicates value significantly different from respective control value(paired t test with correction for multiple comparisons). Within the control and each treatmentgroup no member of a pair of values was significantly different from the other membrane (power> 0-80).

SS VE, C2

2ndControl

3rd

I I I I~~~~~~~~~~~~~~~~~

RB VE, CO2

Control 2nd

3rd

40 50 60

End-tidal PCO2 (Torr)70

Fig. 2. VE, co2s Top: SS curves; second and third slopes are displaced parallel from control.Bottom: RB curves; second and third slopes are decreased from control.

512

Condition/sequenceControlSS-RBRB-SSAfter 0 07mg kg-' MSSS-RBRB-SSAfter 0-21mg kg-' MSSS-RBRB-SS

60 -

40-

20-

.

0)

60 -

40 -

20 -

STEADY-STATE AND REBREATHING METHODS

steeper than the corresponding SS VE, Co2s. Overall, this difference was 1-3+0-86 1min-' Torr-1. P15 for the RB VE,Co2s was 45 + 1-2 Torr greater than for the SSVE, CO2S-In all eight sessions the first dose of MS caused the slope of the RB VE Co s to

decrease 0-76+ 1-03 1 min-' Torr-' and the P,5 to shift rightward 39 ± 24 Torr. Afterthe final dose of MS, the slope of the RB VE, Co2 was decreased by 1-30 + 0-55 1 min-'Torr-' (P < 0-001), and the Pl5 was displaced rightward 7-3 + 3-7 Torr.

TABLE 2. rE,CO, slopes and P,,sSlope (1 min-' Torr-')

Ss RB SS

P,5 (Torr)

RB

ControlSS-RB 1-43+0-58 2-50+ 1-14RB-SS 1-82+0058 3-35+ 1-00After 0-07mg kg-' MSSS-RB 1-98+ 093 2-24+ 1-00RB-SS 1-91+0-98 2.07+1-03*After 0-21mg kg-' MSSS-RB 1-47+0-51 1.57 + 092*RB-SS 1-71+0-71 1-69 + 050*

Values are means+ S.D. * Indicates value significantl,(paired t test with correction for multiple comparisons).

47-0+3-9 52-3+0-847-7+3-0 51-3+1-8

52-3 + 3.8* 56-6+ 1-2*51.5+3-4* 54.9+2-5*

57.9+ 3-4* 59.7+ 4.5*57-2 + 4-2* 58-4+ 1.6*

Ly different from respective control value

During the eight study sessions the first MS dose did not affect the slopes of theSS VE, C02 (mean difference 0-31 + 0 53 1 min-' Torr-', power > 0 85). However, SSVE, Co P15s were shifted to the right 4-6+ 1-8 Torr (P < 0-001). The second MS dosecaused a further rightward shift of the P,5 from control (10-2 + 1-9 Torr, P < 0-001),again without a detectable change in the SS VE, Co2 slope (mean difference 0 04+ 0'421 min-' Torr-', power > 0 80).The effects ofMS on the slope and P,5 of both the SS and RB VE, Co2s were the same

regardless of the order of VE, Co2 determinations during a session. Linear regression ofthe SS PE, CO slope vs. MS dose yielded a regression line slope of -0-42 1 min-' Torr-'mg-' kg-1, which was not significantly different from zero (power > 0 80). The slopeof the regression line for the RB VE, co2S VS. MS dose was - 5-67 1 min-' Torr-1 mg-'kg-' which was significantly different from zero (P < 0-001).

DISCUSSION

Our goal was to confirm empirically our impression that the effect of analgesicdoses of narcotics, in particular morphine, on the C02-response curve is qualitativelydifferent depending on whether the method used is the standard steady-statetechnique or the more convenient rebreathing technique developed by Read (1967).Our results confirm that morphine causes a dose-related parallel rightward shift inthe SS VE, CO2, whereas the RB VE, Co2 undergoes a dose-related decrease in slope. Allof our C02-response data, regardless of how determined (SS or RB) or treatment

Condition/sequence

513

Pri Y 41917

D. L. BOURKE AND A. WARLEY

conditions (control, low MS or high MS), fell within the expected ranges (Prescottet al. 1949; Read, 1967; Irsigler, 1976; Goodman & Curnow, 1985).Our protocol was also designed to examine the possibility that determination of a

VE,CO2 by either method might, only 15 min later, affect the VE,CO2 using thealternative method. Although several reports confirm the short-term reproducibilityof VE,co s determined by either method, none specifically addresses the possibleeffects o? alternating the methods (Strachova & Plum, 1973; Irsigler, 1976; Sahn,Zwillich, Dick, McCullough, Lakshminarayan & Weil, 1977; Berkenbosch et al.1989). Based on these reports, it seemed unlikely that the proximate determinationof VE, CO2S by different methods would alter the results. However, we endeavoured inour experimental design to eliminate or at least to determine the possible effects ofinteraction. One approach was to measure resting VE in the 2 min period immediatelypreceding each VE,co measurement. These results and the associated PET, O valuesdo not indicate that the determination of a VE, co2 by either technique affected in anysignificant or consistent way resting VE 15 min later. Second, each subjectparticipated in two separate sessions. In one session, the SS technique was followedby the RB throughout; in the other, the sequence was RB first and SS second'. Twosubjects had the SS-RB sequence during the first session and the RB-SS during thesecond session. The other two subjects had the reverse pattern. Neither the within-session sequence of VE, CO2s nor the order of the SS-RB and RB-SS sessions had anydiscernible effect on our overall results. Moreover, for every subject in every session,regardless of the SS-RB sequence, SS VE, Co2s were progressively displaced to theright without significant changes in slope. However, the RB VE, CO2s had progressivelylower slopes in addition to rightward displacement measured at a VE = 15 1 min-'.Although our report specifically examines only morphine, it is likely that the SS

technique, unlike the RB technique, permits the discrimination of respiratorydepression caused by analgesic doses of opiates and more generally those drugsacting on opioid receptors, from the respiratory depression caused by other classesof drugs (benzodiazepines, barbiturates, anaesthetics etc.). For example, one of theauthors participated in a study of the respiratory effects of ketamine in 1972(Bourke, Malit & Smith, 1987). It was found that ketamine depressed respiration andcaused a parallel rightward shift of the SS VE,CO . At the time, we were not aware ofketamine activity at opioid receptors (Smith, Pe?oe, Martin & Coalgate, 1980; Finck& Ngai, 1982). We have since postulated that the respiratory depression cause byketamine (in doses up to 3 mg kg-1) is mediated by opioid receptors.The properties of the SS VE, Co2 described above may in many instances not be of

sufficient value or interest to an investigator to warrant the extra timne and effortrequired to study respiration by the SS technique. However, a recently describedtechnique for reducing the time required to achieve steady-state conditions maymake the SS method of determining VE,cO2s more practicable when the extraqualitative information would be of value (Poon & Olsen, 1985).Although our study confirms our hypothesis that there is a qualitative difference

in the results obtained from the study of respiratory depression caused by narcoticsdepending on whether the SS or RB technique is used, our study was not designedto elucidate the reasons for this difference. Recently, however, Berkenbosch et al.(1986) examined the differences in the SS and RB technique using cats during normal

514

STEADY-STATE AND REBREATHIVG METHODS

conditions as well as during metabolic acidosis and alkalosis. In a second study theycompared the two techniques in humans under normal conditions only (Berkenboschet al. 1989). A generalization of their conclusions is that if the ratio of the stepincrease in PET, Co2 at the initiation of rebreathing (A) to the rise of CO2 duringrebreathing (R) is equal to the time constant of the central chemoreflex (tc) then theslopes of the SS and RB curves will be equal. Further, a change in the ratio, A/R,relative to t. will cause the RB slope to differ from the SS slope.An a posteriori review of our data indicated that R did not change significantly

from control after either dose of MS. On the other hand, Table 1 indicates that thePET, Co2 preceding each RB curve increased progressively following each dose of MS.Therefore with a fixed concentration of 7% CO2 in the rebreathing bag at thebeginning of each RB determination, A progressively decreased after each MS dose.With a constant R and a decreasing A, the ratio A/R would decrease. Assuming t,constant, a decreasing RB slope relative to the SS slope should be observed. This wasthe situation in our study. In fact, using the formula derived from the Berkenboschet al. (1986) model, we were able to account for about 80% of the differences betweenSS and RB slopes that occurred in our study.Although we do not have sufficient data to determine t. independently, we propose

that a small MS dose-related increase in tc would almost completely reconcile ourresults with the model proposed by Berkenbosch et al. (1986). After the final dose ofMS, a modest increase in t. along with our experimentally observed decrease in Awould be such that the ratio A/R would nearly equal t, and the calculated slopes forthe RB and SS curves would be virtually equal. This was the case with ourexperimental results for the SS and RB curves after a total of 0-21 mg kg-' MS.Morphine may alter t, by one or more of several mechanisms. A direct narcotic effecton neuronal dynamics may increase the central time constant. We are unable to findany specific data to support this hypothesis. However, the central time constant isknown to be inversely proportional to cerebral blood flow and, despite somecontroversy, there is considerable evidence that narcotics reduce cerebral blood flow(Takeshita, Michenfelder & Theye, 1972; Vernhiet, Renou, Orgogozo, Constant &Caille, 1978; Carlsson, Smith, Keykhah, Englebach & Harp, 1982). It should benoted that most of the research relating narcotics to reductions in cerebral blood flowhas been done with much higher, that is, anaesthetic, doses (1-3 mg kg-') ofmorphine than the analgesic doses (0 07 and 0-21 mg kg-1) used in our study.Although the effect of narcotics on CO2 regulation of cerebral blood flow has not beenstudied extensively, one report (Vernhiet et al. 1978) suggests that it remains intact.None the less, based on the available studies, one can estimate that 0-21 mg kg-1 MScould reduce cerebral blood flow by as much as 15% (Takeshita et al. 1972). Such areduction in cerebral blood flow may sufficiently increase t, so that our data wouldbe consistent with the model proposed by Berkenbosch et al. (1986) for explaining thedifferences between the SS and RB C02-response curves.Our study and the work of Berkenbosch et al. (Berkenbosch et al. 1986), indicate

that the results of C02-response curve measurements by either method are dependenton a number of variables. The RB method is not only sensitive to the centralchemoreceptors but is also affected in complex ways by cerebral blood flow, transporttime, t., cerebral and whole-body CO2 production, and other technical factors. By its

17-2

515

D. L. BOURKE AND A. WARLEY

nature the SS method is more resistant to changes in these variables and is mostsensitive to cerebral blood flow and its effects on the relationship between PET, Co2

and tissue Pco .

As Berkentosch et al. (1986) suggest and our results confirm, repeatedmeasurements of ventilatory CO2 sensitivity under changing conditions by the RBmethod may produce considerably different results as compared with similarmeasurements using the SS method. For these reasons we believe that the intuitiveappeal of the model developed by Berkenbosch et al. (1986) combined with itsremarkable ability to explain our empirical results are impetus for further study ofboth the validity of the model and the differences between the SS and RB methodsof assessing respiratory control under a variety of conditions.

REFERENCES

BERKENBOSCH, A., BOVILL, A. D., DAHAN, A., DEGOEDE, J. & OLIEVIER, I. C. W. (1989). Theventilatory CO2 sensitivities from Read's rebreathing method and the steady-state method are

not equal in man. Journal of Physiology 411, 367-377.BERKENBOSCH, A., DEGOEDE, J., OLIEVIER, C. N. & SCHUITMAKER, J. J. (1986). A pseudo-

rebreathing technique for assessing the ventilatory response to carbon dioxide in cats. Journalof Physiology 381, 483-495.

BOURKE, D. L., MALIT, L. A. & SMITH, T. C. (1987). Respiratory interactions of ketamine andmorphine. Anaesthesiology 66, 153-156.

CARLSSON, C., SMITH, D. S., KEYKHAH, M. M., ENGLEBACH, B. S. & HARP, J. R. (1982). The effectsof high-dose fentanyl on cerebral circulation and metabolism in rats. Anaesthesiology 57,375-380.

DEMPSEY, J. A. (1976). CO2 response: stimulus definition and limitations. Chest 70,114-118.FINCK, A. D. & NGAI, S. H. (1982). Opiate receptor mediation of ketamine analgesia.

Anaesthesiology 56, 291-297.GOODMAN, N. W. & CURNOW, J. S. H. (1985). The ventilatory response to carbon dioxide. British

Journal of Anaesthesia 57, 311-318.IRSIGLER, G. B. (1976). Carbon dioxide response lines in young adults: the limits of the normal

response. American Review ofRespiratory Disease 114, 529-536.IRSIGLER, G. B., STAFFORD, M. J. & SEVERINGHAUS, J. W. (1984). CO2 response lines by steady-

state and rebreathing techniques in metabolic acidosis and alkalosis, and their relation toestimated medullary chemoreceptor pH. American Review ofRespiratory Disease 129(4), A253.

JENNETT, S., BARKER, J. G. & FOREST, J. B. (1968). A double-blind controlled study of the effectson respiration of pentazocine, phenoperidine, and morphine in normal man. British Journal ofAnaesthesia 40, 864-875.

JORDAN, C. (1982). Assessment of the effect of drugs on respiration. British Journal of Anaesthesia54, 763-782.

LINTON, R. A. F., POOLE-WILSON, P. A., DAVIES, R. J. & CAMERON,I. R. (1973). A comparison ofthe ventilatory response to carbon dioxide by steady-state and rebreathing methods duringmetabolic acidosis and alkalosis. Clinical Science and Molecular Medicine 45, 239-249.

LOESCHEKE, H. H., SWEEL, A., KOUGH, R. H. & LAMBERTSEN, C. J. (1953). The effect of morphineand of meperidine (Dolantin, Demerol) upon the respiratory response of normal men to lowconcentrations of inspired carbon dioxide. Journal of Pharmacology and ExperimentalTherapeutics 108, 376-383.

POON, C. S. & OLSEN, R. J. (1985). A simple quasi-steady technique for accelerated determinationof C02-response. Federation Proceedings 44, 832.

PRESCOTT, F., RANSOM, S. G., THORP, R. J. & WILSON, A. (1949). Effect of analgesics on

respiratory response to carbon dioxide in man. Lancet i, 340-344.READ, D. J. C. (1967). A clinical method for assessing the ventilatory response to carbon dioxide.Australian Annals ofMedicine 16, 20-32.

READ, D. J. C. & LEIGH, J. (1967). Blood-brain tissuePCo2 relationships and ventilation duringrebreathing. Journal ofApplied Physiology 23, 53-69.

516

STEADY-STATE AND REBREATHING METHODS 517

REBUCK, A. S. (1976). Measurement of ventilatory response to C02 by rebreathing. Chest 70,118-121.

RIGG, J. R. A. (1978). Ventilatory effects and plasma concentration of morphine in man. BritishJournal of Anaesthesia 50, 759-764.

SAHN, S. A., ZWILLICH, C. W., DICK, N., MCCULLOUGH, R. E., LAKSHMINARAYAN, S. & WEIL, J. V.(1977). Variability of ventilatory responses to hypoxia and hypercapnia. Journal of AppliedPhysiology 43, 1019-1025.

SMITH, D. J., PEKOE, G. M., MARTIN, L. L. & COALGATE, B. (1980). The interaction of ketaminewith the opiate receptor. Life Sciences 26, 789-795.

SNEDECOR, G. WV. & COCHRAN, W. C. (1980). Statistical Methods. 7th edn.. pp. 83-106. Ames. Iowa.USA: The Iowa State University Press.

STRACHOVA, Z. & PLUM, F. (1973). Reproducibility of the rebreathing carbon dioxide response testusing an improved method. American Review of Respiratory Disease 107, 864-869.

TAKESHITA, H., MICHENFELDER, J. D. & THEYE, R. A. (1972). The effects of morphine and N-allylnormorphine on canine cerebral metabolism and circulation. Anaesthesiology 37, 605-612.

VERNHIET, J., RENOU, A. M., ORGOGOZO, J. M., CONSTANT, P. & CAILLE, J. M. (1978). Effects of adiazepam-fentanyl mixture on cerebral blood flow and oxygen consumption in man. BritishJournal ofAnaesthesiology 50, 165-169.