csf hydrodynamic studies in man1 - bmj · journalofneurology,neurosurgery, andpsychiatry, 1977, 40,...

Journal ofNeurology, Neurosurgery, and Psychiatry, 1977, 40, 105-119

CSF hydrodynamic studies in man11. Method of constant pressure CSF infusion

J. EKSTEDT2

From the Department ofNeurology, the University of Uppsala and the Department ofNeurology,the University of Umed, Sweden

SUMMARY The constant pressure method for the study of the hydrodynamics of CSF is presented.By infusing artificial CSF at constant pressures and recording the resultant flow, it is possible toobtain information about the hydrodynamic conductance of the CSF outflow pathways. By loweringthe infusion pressure below the pressure ofthe sagittal sinus all CSF produced can be collected and theCSF formation rate may thus be calculated. There is a rectilinear relationship between CSF pressure

and the flow necessary to maintain the pressure. It is thus concluded that the arachnoidal villi, whenonce opened, are not further distended by pressure. This method makes possible indirect calculationof the pressure of the sagittal sinus and the pressure difference between the subarachnoid space

and the sagittal sinus.

Since the introduction of lumbar puncture byQuincke in 1878 measurement of the lumbar CSFpressure has become a routine procedure in clinicalwork. Numerous attempts have been made toacquire more information about the hydrodynamicsof the CSF than only that related to pressure.Ayala (1923, 1925) drew conclusions concerning thepressure/volume relationship in the craniospinalspace by studying the pressure decrease after with-drawal of a certain volume of CSF. Masserman(1934) withdrew CSF and studied the time forreturn of the CSF pressure to its initial value. Hethereby estimated the rate of production of CSF.By reinjection of the CSF, information aboutelimination was also obtained. Schaltenbrand andWordehoff (1947) further elaborated this methodto the status of a clinical diagnostic procedure.Sokolowski (1974) has also used a single bolusinjection of fluid for studying the CSF hydro-dynamics.

Constant flow rate, subarachnoid infusions werefirst performed in human subjects by Foldes andArrowood (1948). They pointed out that, whenincreasing the flow rate, the CSF pressure rose but

'Presented in part at the Second and Third International Symposia onIntracranial Pressure, Lund, June 1974 and Groningen, June 1976.Supported by grants from Swedish Medical Research CouncilB 74-04X 4125 and B 77-04X-04125-04A.2Address for correspondence and reprint requests: Dr Jan Ekstedt,Department of Neurology, University Hospital, S-901 85 UmeA,Sweden.Accepted 12 October 1976

soon reached a plateau which was higher as theflow rate was higher. The relationship betweenpressure and rate of infusion was, however, notlinear. Rubin et al. (1966), Cutler et al. (1968), andLorenzo et al. (1970) were the first to use the methodof ventriculoventricular or ventriculolumbar per-fusion in human studies. By adding various tracesubstances to the perfusion solution, CSF formationand elimination could be measured at differentpressure levels. The necessity for ventricular puncturerestricts the use of the method to special humancases and to animal experiments.Katzman and Hussey (1970) developed a clinical

test based on constant rate fluid infusion. Theirtest started with removal of 10 ml of CSF. Thepressure return to the initial value was studied byopen manometry in order to determine the rate offormation of CSF. Saline was then infused at aconstant rate of 12.7 mm3/s (0.76 ml/min) during40 to 60 minutes. Sometimes a higher infusion rate of32 mm3/s (1.9 ml/min) was also used. The plateauvalues of the pressure at the different infusion rateswere the parameters used to describe the CSFabsorption. Clinical experience with the method wasdescribed by Hussey and Schanzer (1970).Nelson and Goodman (1971) modified the con-

stant rate infusion method by using the rate of riseof CSF pressure as the measured parameter. Theyalso used artificial CSF instead of saline for infusionand electromanometric recording of the CSFpressure. Martins (1973) used the constant rate

105

guest. Protected by copyright.

on Decem

ber 22, 2020 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.40.2.105 on 1 F

ebruary 1977. Dow

nloaded from

http://jnnp.bmj.com/

106

infusion method but treated the results in anotherway by calculating the resistance to drainage ofCSF. Trotter et al. (1974) also used a modifiedKatzmann and Hussey method. They read theCSF pressure after 20 minutes infusion at a rate of12.7 mm3/s (0.76 ml/min) and five minutes after achange of the rate to 32 mm3/s (1.91 ml/min). DiRocco et al. (1976) have also used the constant rateinfusion test in human studies. A recent extensivereview of the physiology of the CSF, includingnumerous references, is that of Welch (1975).

In the present paper, another technique for thestudy of the hydrodynamics of the CSF is described,the constant pressure infusion method. The infusionpressure is kept at a constant level and the resultingCSF flow recorded. Constant pressure infusion hasbeen used for investigating the hydrodynamics of theanimal eye (Becker and Constant, 1956; Armaly,1959; Bara'ny, 1962, 1964) and of CSF in animalexperiments (Davson et al., 1970), but not previouslyfor human investigations. Constant pressure infusionpermits a detailed analysis of the CSF hydro-dynamics, mainly because much more experimentaldata of pressure/flow can be obtained during a giventime period than with the constant rate infusionmethod. The present study has been preliminarily

J. Ekstedt

reported by Ekstedt (1973, 1975) and Ekstedt andFriden (1976). In a modified version the method hasalso been used by Portnoy and Croissant (1976).

Methods

INVESTIGATION CHAIR/BEDThe chair used was modified from an investigationchair common in Swedish hospitals. Originally, itwas supplied with a back support made of springsand steel bands. This was replaced by a bottom ofdouble canvas reinforced by straps on critical sites.An opening of 150 x 150 cm was made centred aboutthe fourth lumbar spinal process.By means of a sterile tape sheet (Steridrape No.

1092, 3M Company, Minnesota, USA) fastened tothe skin, a sterile field for puncture was obtained.After the lumbar puncture the chair was folded backto a horizontal position. The chair is shown inFig. 1.

LUMBAR PUNCTUREFor pressure recording a 0.9 x 90 mm lumbarneedle was used (conductance 35 mm3 kPa-' * s-1).For infusion a 1.2 x 100 mm needle was used (con-ductance 85 mm3 * kPa-1 * s-1).;-.;g~~~~~~~~~~~~~~~~~~~~~~~~~

.... .....

...:...;......

Fig. 1 Experimental chairlbed.The lumbar needles are insertedvia the opening in the backsupport with the patient in thesitting position. The chair is thenfolded back and the rest of theinvestigation made with thepatient supine.


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


CSF hydrodynamics: constant pressure CSF infusion

The needles were connected to the experimentalequipment with non-elastic PVC tubes filled withartificial CSF. The tubes and stopcocks were dis-posable, replaced for each experiment.

PRESSURE RECORDINGTwo pressure transducers (Hewlett-Packard 1280 C,Waltham, Mass., USA) were used. They wereconnected to carrier frequency amplifiers (Hewlett-Packard 8805 B). All connections between tubeswere made by means of plastic three-position stop-cocks (Viggo, Helsingborg, Sweden). Calibrationwas made by connecting the transducer with either oftwo sterile water columns, one with its water levelat the height of the zero reference level of the patient,the other 306mm above that level, thus correspondingto a pressure of 3 kPa. The accuracy of pressuremeasurement was better than 0.03 kPa. Carefulflushing of tubes, stopcocks, and transducers wasperformed before each experiment to avoid airbubbles. The compliance of the pressure recordingsystem, measured at the tip of the lumbar needle,proved to be less than mm3/kPa. The pressure signalsduring recording were usually low pass filtered with atime constant of 0.05-0.01 s.

CSF INFUSIONDuring the first 50 experiments a 0.9% saline solutionwas used. Nearly all patients experienced some kindof discomfort during the infusion. Sometimes therewere very prominent signs of distress, which neces-sitated termination of the investigation. In a fewcases the patient became very excited, requiringintravenous sedation. In most cases the side-effectstook the form of pain, paraesthesia or paralgesia(especially in the legs and buttocks), urgency forvomiting or defaecation, and symptoms from theautonomous nerve system, such as piloerection orsweating. The side-effects with saline were similarto those that have already been reported by Weedand Weyeforth (1919) and by Nelson and Goodman(1971). It was obvious that an infusion solution thatcould produce such dramatic symptoms should notbe considered suitable for these types of experiments,and that it was potentially dangerous. The side-effects also made it very difficult for the patient tokeep a stable position for any length of time. There-fore, a solution closely resembling the one devised byElliot and Jasper (1949) (the B solution) was made.The first 200 bottles contained glucose but in laterbottles the glucose has been omitted without notableeffect.The solution was prepared in two bottles (90 ml

and 10 ml) which were mixed immediately before theexperiment.3 The solution was made by the hospitalpharmacist and contained Na+ 145.7 mmol/l;

107

K+ 4.0 mmol/l; Ca++ 1.4 mmol/l; Mg++ 0.4 mmol/l;Cl 130.4 mmol/l; HCO3- 21.7 mmol/l; HPO4- 0.6mmol/l; pH = 7.40.The differences in side-effects when using the

artificial CSF instead of saline were considerable.There was no reaction of the type described above.In three cases (of 783) there was a meningitic reactionwith fever for a few days. All patients recoveredcompletely. In these three cases the rest of thecontents of the bottles were sent for bacteriologicalexamination; however, there was no bacterialgrowth. There might have been chemical or viralcontamination.The artificial CSF was at room temperature in both

bottle and tubes. However, about 50 mm of theneedle was within the body and thus heated thesolution. In in vitro experiments when a 50 mmsegment of the needle was heated to 37.0°C it wasfound that when flow was below 25 mm3 S-1 theartificial CSF had a temperature of more than 34°Cwhen leaving the tip of the needle. At 100 mm3 s-'the temperature was 29°C. It has not been consideredessential to preheat the artificial CSF.

INFUSION APPARATUSThe experimental set-up is shown in Fig. 2. The bottlewith artificial CSF hung about 200 mm below thereference level (the cranial mid-point). The infusionpressure was created with a membrane air pump(aquarium pump Rena super 300, Annecy, France)which pumped air into the bottle. An adjustable airleak maximised the air pressure. The voltage to theair pump was automatically adjusted by a servodevice.The air entering the pump was filtered by a sterile

filter (Aga Sterile Filter 323 900 002, Stockholm,Sweden). The air leaving the air leak was again ledinto the pump, so that there was very little intake ofnew air into the pump. Furthermore, the output fromthe pump was twice filtered by means of Munktel airfilters MS 20 (Stora Kopparberg, Grycksbo, Sweden).The air was then free of any particles and sterile.

LIQUID FLOW RECORDINGThe infusion bottle was suspended by a thin nylonwire in a strain gauge transducer. The tubes leadingair into the top of the bottle and taking fluid fromthe bottom of it were pliable, exerted no elasticforces on the bottle and acted on it only by theirweight.The bottle had a volume of 120 ml. Before and after

'Solhtion I (90 ml): NaCI: 7.98 g; KCI: 0.33 g; CaCI, * 2 H,O: 0.22 g;MgCl - 6 H.O: 0.08 g; HCI I mol/l to pH 2.8; pyrogen free distilledwater to 1.0 1. Autoclaved at 1 20°C for 20 minutes. Solution 11 (10 ml):NaHCO,: 18.2 g; Na2HPO4 * 2 H2O: 0.98 g; pyrogen free distilledwater to 1.0 1. Autoclaved at 11I0°C for 35 minutes.


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


J. Ekstedt

Fig. 2 Experimental set-up. Thepatient is lying supine with two lumbar needles(AandB) in the subarachnoid space. A is connectedto transducer Cwith a tubefilled withartificial CSF. Similarly the needleB is connected to transducer D. By means ofstopcock E, connection can be made to bottle F, containing artificial CSF. The bottlecan bepressurisedby airfrom the airpump G. Air is leaking through the adjustablevalve Hin order to maximise thepressure andtopermit asmoother regulation. Thepower to the airpump is deliveredby the cerebrospinalpressure regulator I, governedbyeither Cor D. Flow is calculatedby weighing the bottle in the straingauge balanceJ, inwhich the bottle is hung up with a 0.30 mmfishing line K. A multi-pen recorderLrecords the resultgraphically. The sagittal midpoint ofthe cranium is chosen as thereference level(M).

each experiment the balance was calibrated byapplying a 5 g weight to the bottle.On the ink recorder (Rika Denki KA 4 Tokyo,

Japan) 1 g represented a deflection of 25 mm on the250 mm wide paper. The accuracy of measuredweight was better than 0.03 g.

REFERENCE LEVEL FOR PRESSUREIn 10 patients the intraspinal pressure was firstrecorded in the lateral recumbent position with thereference line along the spine. The patient was thenturned to the supine position and the pressure againmeasured in relation to the external auditorymeatus. The procedure was repeated a few times.It was then found that a reference level at the cranialmidpoint between inion and glabella best corres-ponded to the reference level of the spine in thelateral recumbent position. This point was thereforechosen as reference point for all measurements.On the exterior, it corresponds to a level through themandibular joint and to the anterior part of theosseous channel of the external auditory meatus.It also corresponds to the level of the foramina ofMonro chosen by Janny (1974) as his reference point.In seven patients radiographs were taken of the skulland thorax with the patient in a supine position. The

chosen point of reference then proved to correspondreasonably well with the centre of the right atrium.If another point of reference is chosen, the pressurewill change with the weight of the water column thatseparates the reference points.

UNITS OF MEASUREMENTSSI units have been used throughout. Pressure: Ikilopascal (1 kPa)= 7.5 mmHg= 102 mmH2O.The hydrodynamic conductance of the CSF outflowpathways has the dimension mm3 kPa-1 s-1.

1 mm3 kPa-1 s-1= 6 10-3 ml (cmH2O)'1 * min'-=8 . 10-3 ml. (mmHg)-l min-'.

EXPERIMENTAL PROCEDURESWith the patient in the sitting position the needleswere inserted in the L3-L4 or L4-L5 intervertebralspace after local anaesthesia of the skin and theintraspinal ligament with 1 % lidocaine (Xylocaine,Astra S6dertalje, Sweden). Care was taken to insertthe needles with minimal trauma. Multiple puncturesof the dura mater were avoided. Two millilitres ofCSF were withdrawn with a syringe for proteindetermination and cell count and immediatelyreplaced by the same amount of artificial CSF.This also served as a check of correct needle position

108


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


CSF hydrodynamics: constantpressure CSFinfusion

in the subarachnoid space. The amount lost duringthe needle insertion was also replaced. The patientwas then folded back to the supine position and thepressure transducers were adjusted to the referencelevel. The different steps of the procedures are shownin Fig. 3.The resting pressure was recorded continuously.

Usually 60 minutes of recording was aimed at, butsometimes a shorter time could be allowed if thecurve was steady. The resting pressure was notmeasured until there had been a period of at least15 minutes without any upward or downward trend.In order to detect whether there was any leakage

around the lumbar needles 20 experiments wereperformed in the following way: the resting pressurewas recorded during 30 minutes and, if constantduring the last 15 minutes, the chair/bed was foldedto the sitting position and kept so for 15 minutes.Thereafter, the chair was again folded back to thesupine position and recording of the resting pressurecontinued for a further 30 minutes. During sitting,the hydrostatic pressure on the needle insertionarea was increased by the weight of the column of thespinal liquid and the flow through a leak shouldhave increased. The resting pressure when thepatient was then folded back to the supine positionshould have been lower if there was a CSF leakaround the needles.The infusion was then started. The first infusion

pressure was chosen to be about 0.5 kPa abovethe resting pressure. The pressure was kept atthis level until the rate offlow had levelled off after theinitial high rate and the CSF pressure was constant.Then the pressure level was increased by a further0.5 kPa and the same procedure was repeated.The pressure was kept at a certain level for at leastfive minutes. In most cases a final pressure of 6 kPawas aimed at, but sometimes, when there was a very

T Weight of infusion bottle

7.5 kPa~~/CSFpessure .5k P

'Resting pressure > Influsion 4Post- ;10 pressure infusionlevels

0 60 miR 120 ml 1501L---b T;

ime

m

good drainage, the content (120 ml) of the bottle withartificial CSF was not sufficient to attain this level.Eight to 15 corresponding values of pressure flowwere obtained in each case. A part of a recording isshown in Fig. 4.The next step was the 'post-infusion'. The con-

nection to the bottle was shut off and the pressurewas allowed to return spontaneously by eliminationof CSF through the physiological pathways. This wascontinued until a steady plateau value was recorded,which, in most cases, closely corresponded to theinitial resting value. If this level was different fromthe initial resting value, there was a strong suspicionthat the resting pressure had been wrongly recordedand the investigation was therefore often repeatedsome days later. In some 20 experiments the infusionwas also continued after the maximum pressure wasreached. Infusion pressure was lowered in steps of1 kPa. This was done in order to detect any hysteresisin the pressure/flow relationship.

In 10 experiments the pressure was, for a period ofone hour, switched between two high values everyfive to 10 minutes, to detect possible changes in flowwith time.The next step in the investigation was then to

perform the drainage. The pressure in the infusionbottle circuit was lowered to 0.25 kPa. This caused anoutflow of CSF from the subarachnoid space intothe bottle. Initially, the rate of outflow was high,but within 10 to 15 minutes the rate levelled offwhen the intraspinal pressure approached thepressure in the infusion circuit. The drainage periodwas planned to last one hour but sometimes had to beshortened because the investigation had persistedfor more than three hours and the patient might behungry and unwilling to cooperate. In some cases thedrainage was postponed to a separate investigation afew days or some weeks later. This part of the

Fig. 3 Schematic layout oftheinvestigation. During30-60minutes the restingpressure isrecorded. The CSFpressure is thenincreased in steps of0.5 kPa up to amaximum of7.5 kPa. The resultingdecrease in the bottle weight isshown in the upper trace. Afterobtaining the highestpressure thebottle is shut offandthepressure

0.25 k P a returnfollowedduring the_ postinfusionperiod. During theC periodofCSFdrainage the CSF

'CSF drainage pressure is loweredbelow that ofthevenouspressure in the sagittalsinus.

ini~om When thepressure has stabilisedtheoutflow is equal to the CSFproduction rate.

109


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


J. Ekstedt

K~~~~

Artificial CSFbottle weight

Infusion pressure

./~~~~/

/~~~~~

483 1974Male 6

- 10-23 18 years

X,

I

I

/~~~~~_. A _1

CSF pressure / Low pass filteroff

_ I * *I

0 1 2 3 4 5 6 7 8 9 10 I, 12 13 14Time (min)

Fig. 4 Part ofan actualrecording during infusion. AtA the infusionpressure is changedby 0.5 kPa,which causes an increase offlow measurablefrom the bottle weight curve. A differentplateau valueforthe infusionpressure and CSFpressure is causedby the resistance in the lumbar needle through whichinfusion is made. This is the reason why a separate needlefor CSFpressure recording is convenient.

investigation was done in order to determine the rateof formation of the CSF. A part of a recording isshown in Fig. 5.

If the intracranial resting pressure was considerablyincreased from the beginning of the experiment or ifthe investigation had given a suspicion of an intra-cranial, space-occupying lesion, then no drainagewas performed because of the risk of tentorialherniation.

In order to minimise post-lumbar puncture head-ache, the drained CSF was in most investigationsreinfused until a pressure was attained which ex-ceeded the resting pressure by 0.5 kPa. Reinfusionwas not made in those cases where the clinical resultof a pressure lowering was to be investigated.Once every 10 minutes, or at the end of each pres-

sure level, during the infusion period, the bloodpressure and pulse rate were measured with asemiautomatic sphygmanometer cuff recorder(Metronik A 900 Koln, W-Germany).

CALCULATIONSAll experimental data were measured from themultichannel ink recorder. The CSF pressure wassubject to variations with time. The arterial pulsesynchronous and respiration synchronous varia-tions were low pass filtered with a time constant of0.05 s. There were still slower variations of thepressure, with a frequency of one or more waves per

minute. When selecting a part of the recording formeasurement, the mean pressure had to be constantduring several minutes. During this time the slope ofthe weight curve had to be as constant as possible andapproximated by a straight line. At the time ofexperiment a check was always made that there was a

well measurable part of the curve before proceedingto a new pressure level.

All calculations were made on a small calculator,Compucorp 445-444 (Computer Design Corporation,Los Angeles, Calif., USA). The output from thecalculator is shown in Fig. 6.

Three different methods were employed for thecalculation of the curves. In method 1 the value of theresting pressure is given a very strong weighting. It isthus considered that the curve should pass throughthe point of the resting pressure and the regressionline was calculated according to Snedecor andCochran (197 1, p. 166).

L[(pcl-pclr) * q]

Gop = ,2pel-pcir) 2q Formula 1

(Gop= the hydrodynamic conductance of CSFoutflow pathways, pet = the actual CSF pressure,

P clr = the resting CSF pressure, q = the actual flow ofartificial CSF).When there is an experimental error both in the X

and Y measurement values the following formula'The programme was written by Ulf Ekstedt.

kPa9 1

8 -

7 -

6-

5-~

4-t

110

I _

0


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from



r483 1974-10-23

Male 68years

CSF bottle weightT

/1cm3- CSF outflow rate

CSF pressure

7 3.027 30 3~~~33 36

11-\ CSFoutf lowrate

[10--- 20

Drainagepressure

-. - .-I

42 45 48 51 5 min42 45 48 51 54min

Fig. 5 Part ofan actualrecording during drainage. The drainagepressure isO.25 kPa. The CSFpressure has stabilisedata slightly higher value because ofthepressure drop causedby the CSFflow in the needle resistance. The CSFformation rate is calculatedfrom the increase ofthebottle weight. For monitoringpurposes during the investigation the instantaneousformation rateis calculatedanddisplayedon the recording.

may be theoretically more suitable. It has been cal- been performed strictly according to the proceduresculated for the last 300 investigations but proved to described here. All investigations were made forgive values for Gop that differed very little from those diagnostic purposes because of a suspicion of alteredobtained by formula 1. CSF hydrodynamics and none on volunteers. The

[(pci -pcer) * q2] project was approved by the ethics committee ofGop = E[(pel pclr)2 . q] Formula 2 Uppsala University. The findings in pathological

cases will be reported separately. In the material 58If using the constant rate infusion method the follow- patients (aged 17-83 years) have been selected as

ing formula will probably be more accurate. probably not having any intracranial or circulatory'Eq2 disease, judged by medical history, clinical neuro-

p = *Eq Formula 3 logical investigation, and further medical develop-,E[(P,clpelr) -

qj ment. These patients constitute the normal material.In method 2 the point for the resting pressure is The CSF hydrodynamic parameters in these patients

omitted entirely from the calculation. All the other will be reported separately by Ekstedt. In the presentexperimental points were used for a conventional paper this group of patients is also used as a referenceregression analysis by the method of least squares. A group for normality. Of course, this post-value of the resting pressure was then calculated from experimental method of deciding which patient shallthis regression line and the calculated value was com- be considered as normal or not is open to criticism.pared with the actually recorded value. However, in the absence ofan investigation on healthy

In method 3 each experimental point is taken in volunteers, which is, for obvious reasons, difficult torelation to the resting pressure. The slope of the line do, the present 58 patients must serve as a provisionaljoining the experimental point and the resting pres- normal sample.sure point was calculated. The mean value of all thesevalues was then calculated. Results

SUBJECTS In the present paper the results will be presented fromTo date 783 investigations have been performed in the qualitative point of view only. The general705 patients aged between 3 days and 92 years during character of the resting CSF pressure recording wasa period of six years. About 500 investigations have the same for the supine position as for the lateral

kPa7 -

6-

5-

4-

3-

2 -

I-

0

III


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


112 J. Ekstedt

3 13

147I -, -31i 7 4 -7

3o e0O0.

0 * 5 7* c;I 660 1 C * 5 1 6 C 2 *2.J0 16.0U 17.12.* ,O 23 *7 1 5* c23.4 26*0 13.3 oc3.30 34.0 1.4* 53.70 35.0 12.7 44

2 04 75 1 4 .

t 4 74.50 4;15 14.0*o 65.00 5* C 14.*3 045 * 5 0 v45 0 14S * o6 - JO ? 3 e z l 4 - 56.30 14.0,,

0 16.3345 /

; 4 4 3 P S3 *7 4 pv104643

455;o

3.51327 j~B2.

, 1 4 *1 4 4, a -71614

a.73, 7a34 73 'P 4.U4Jl4 Fs 06 7 r

.4,22 i' L~

3.L740 ;,

4 3.5 1 34 'I50vj. 12'3

3 662 1 7 _I j;0 30054493 In.

B3 .............14.6411 F34

o 6 00.3431 5 1 v

3.-3334 4Z) .4 7* i53

* 4 7 5 6 ;

3 7 3250526-753 0

1 374-01-311474-05-120

3 c.

O: 2

SI*2 3 4 5 6 *7

0.~~~~~~~

14.*23 744* @7i! 6 * )6 3 3C 33 t, 01

Y 0.4463 p1t,

Fig. 6 Outputfrom the calculator. First the originaldataare listed(A). Then the regression line ofthe experimentalpoints is calculated by three different methods (BJ, B2, B3,see text). The experimentalpoints are then automaticallyplotted with pressure in kPa as the abscissa andflow ratein mm3s-1 as the ordinate (C).

recumbent. It was found to be very important not toturn the head, as this invariably increased the pressure.

In three investigations, where the intraventricularpressure was recorded during the whole experimentalprocedure, the arterial pulse synchronous variationswere about 10% higher intraventricularly, but in allother respects the pressure was always equal. Thisheld true both for the spontaneous variations inpressure and the variations induced during the in-fusion. There is thus no reason to believe that a CSFhydrodynamic investigation should give differentresults when performed via the ventricles than whenperformed by the lumbar route.

In 20 experiments, when raising the patient to asitting position for 15 minutes and thereafter re-suming the supine position, there was only one casewhich showed decrease of the pressure by 0.4 kPa. In

all other cases the post-sitting value within a fewminutes approached the pre-sitting by ±0.1 kPa(Fig. 7). Hydrostatically increasing the pressure on adural rift should induce a leakage. It can then beassumed that a leakage around the needle, as des-cribed by Lundberg and West (1965), did not occurfrequently in the present experiments.During the infusion period the patient usually ex-

perienced only slight unpleasantness. Sometimes hebecame sleepy and fell asleep at pressure levels above5 kPa and then woke up again when the pressuredecreased. Pulse and blood pressure were usuallyunaffected.

All spontaneous variations in the CSF pressurewere exaggerated at increasing mean pressure. Inspite of this it was nearly always possible to obtainwell measurable sections of the pressure and weightcurves even at the highest pressure levels.

In Fig. 8a is shown a pressure flow diagram froman investigation which was, from the technical pointof view, very successful. There is no doubt that therelation between pressure and flow is linear. Theconductance, calculated by the three methods des-cribed in the Method section, were all numericallyclose to each other. It has been postulated that if thethree methods give values that differ by less than 15%,the pressure/flow relation can be consideredrectilinear.

In the majority of the several hundred recordingsthere seems to be a rectilinear relation betweenpressure and flow. This holds true for patients con-sidered to be normal from the point of view of CSFhydrodynamics (Fig. 9) as well as for patients withabnormal CSF hydrodynamics (Fig. 10). In a fewcases, however, an obviously nonlinear relationshipbetween pressure and flow was obtained whichcould not be explained by technical errors (Fig. I 1).In many early experiments there was a tendency forlower pressure values to give unexpectedly high flowvalues, and this was dependent upon the fact thatthere was not enough time left for the flow curve tolevel off. When this fact was noticed, this source oferror was eliminated.

In 20 experiments the infusion pressure was in-creased in steps of 0.5 kPa up to a pressure of 6.5 kPaand thereafter again decreased in steps of 1.0 kPa.The flow values obtained with increasing pressuredid not differ significantly from those obtained withdecreasing pressure.

In five experiments the pressure was alternativelyheld at 4.5 or 5.5 kPa in periods of five minutes for 60to 90 minutes in order to detect possible changes inthe outflow pathways with time-for example, adistention of the arachnoid villi. There was nosystematic change in the outflow conductance de-tectable in these experiments (Fig. 12).


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from



737 1976-04-08Male 571ears

Infusion pressurea 1, iA.1 hiL^AII ..~ .iA 112.pre e v WI . -w , "

2 CSF pressure--,rI--, ,^J9-4 zJ%_}<__I

18 il 24 2 30 33 36 39 42 45 48 SI 54 57 60 63 min

Fig. 7 CSFpressure recordedduringsupine-sitting-supine position. The curve with the large arterialpulse synchronouspressure variations is unfiltered, the other curve is lowpassfiltered with -3dbfrequency 0.05 Hz. The baselines ofthe curves are separatedby I kPa in order not to interfere with eachother. The scale ofthe ordinate refers to thefiltered curve. After resuming the supineposition there is nopressure decrease to indicate a CSFleakage during sitting.

CSFmm

91FLOW3. sG(method 1) 14.337' mm3 kPa1s1

method 2) 14.147.-method 3) 14.641 "'3

373

PCIr 0.95 kPaqf 6.96 mm3 S-1

PRESSUREo /1 2 3 4 5 6 7 k Pa- S -10 2O 30 4i 5o mmHg

(a)

CSF FLOWmm3 s71

A

0

17908

PCIr 1.35 kPaqf 5.83 mmY

2O 3o 40 5o mm Hg(b)

Fig. 8 Relation betweenpressure andflow in twopatients. The CSFpressure is the abscissa. The inflow through the lumbarneedle ofartificial CSFwhen thepressure has stabilised is the ordinate. The symbols are defined in Fig. 9(a) isfrom atechnically very successful experiment (the same asshown in Fig. 6). There seems to be no doubt that the experimentalpointscan be approximatedwith a straight line. The three different methodsfor calculation ofthe regrevsion coefficients allgivepractically the same value. (b) is,from a technicalpoint ofview, a less satisfying experiment. The three calculation methodsallgive badfitsfor the regression lines and the regression coefficients are all different.

B

kPa81

7.-

6

5-

4.-

3 .

0

80

70

60

50

40

30

20

10

LA.,"L.LJA JJICIA

-TV,-rw

1- AL-M M. AL AM .M MA kfAillolildoA

113

-r r 'r ir,rrjwqm,-.I

I .

G(IG(


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


114

CSF FLOWmm3.*l

CSF FLOWmm3 s$

90

234 SM 1973-05-1101 04 d" 72 years

PCIr 1.3 kPaqf 4.2 mm3s'Gop12.2 mm3kPi!s',s 1.o kPaPdop 0.3 kPa

0~~~0>/

Az"" 2 31o io

430(a)

80170k

60

50'

40'

30

20

10

PRESSURE5 6 7 kPa

io eo mmHg

506 1975-11-189 65 years

PCIr 1.4 kPa5.3 mm3s /

dO 20.3 mm3 kpi'Pss 1.i kPaPdop 0.3 kPa

03&1 2 310 20

4sb(b)

PRESSURE5 6 7 kPa

io so mmHg

CSF FLOWmm3 s7i

90

80'I~~~~

0

287154(

~~~~Pcl,'1/ ~~~qf!'~I:

~'pI

/0 pdop0o

70'

60

50

7 AE 1973-10-040416 ci 58 years1.2 kPa

3 -15.3 mm s13.8 mm3-kPi"-iO.s kPa,0.4 kPa

40

30

20

10'

PRESSURE7 kPa

30 40 SO mm Hg

(c)

Pcl,qfGodPss

/ ~~~Pdc

0 1 2

51 1 H 1972-11-17dc 58 years

,r0.8 kPa5.8 mm3- s-'

,pl8.3 mm3. kPas-'; 0.4 kPabopO.4 kPa

PRESSURE_ 4 5 6 7 kPa

io 30 40 5o mm Hg(d)

Fig. 9 Pressure/flowdiagramsfrompatients, normalasregard CSFhydrodynamics. CSFpressure onabcissa. CSFinflow through the needle on ordinate. The numbers denote: investigation number, date ofinvestigation, sex, andage ofpatient, pclr: resting CSFlumbarpressure, qf: CSFformation rate, Gop: conductance ofCSFoutflowpathways, p88:

sagittalsinuspressure,ppdop:pressure difference across CSFoutflowpathways. In this andallothergraphs the line is theregression line ofpressure onflow calculatedaccording to method l (see text,formula 1). There is a rectilinear relationbetweenpressure andflow. The horizontalline under the abscissa denotes the CSFformation rate. It represents the ordinatevaluefor zero outflow ofCSFthrough the arachnoid villi. The regression line is linearly extrapolated to the zero outflow line.The intersectionpoint represents the CSFpressure that shouldhave existedifthere were no internalproduction ofCSF-that is, the sagittalsinuspressure (seeformula 4).

J. Ek-stedt

90

80

70

60

50

40

30

20

10

CSF FLOWmm3 sci

90

80

70

60

50

40

30

10'

O > 2 310 2b


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from



CSF FLOW..3 .*1

70

201 1973-03-06cf 65 years

Pclr 2.2 klqf 4.2 miGo 5.2 mAPSS 1. 3 kI

0op.9 klI

Pdo

Pa3 -

m3- kPi-s'PaPa

60

50

40

30

20

10

PRESSURE4 5 6 7 kPa3b 40 5o mmHg(a)

498 1974-11-076 59 years

Pclr 2.2 kPaq, 6.5 mm3g'aop 4.7 mm3kPci slP& 0.8 kPaPdop 1.4 kPa

0

CSF FLOWmm3s-1

341 1973-12-06cd 28 years

/ Pclr 2.6 kPaqf 5.8 mm3 s-'

0 Gop49.1 mm3 kPisPS 2.4 kPaPdopO.2 kPa

90

80

70

60

50,

20

10

io 4o0 So mm Hg

(c)

523 1974-12-059 55years

19 04 13 - 892e?lr 3.7 kPaqf 6.0 mm sl

15.7mm3 kPaT!s'Pss 3.3.kPaPdop 0.4 kPa

0

0/0

0 1 2 3j/-4 5 6o L -- 30 40

(d)Fig. 10 Pressure/flow diagramsforpatients with increased CSFpressure due to decreasedconductance (a, b) or increasedsagittalsinuspressure (c, d). Symbols as in Fig. 7. There is a rectilinear relation betweenpressure andflow.

CSF FLOWmm3s-1

90

80

70

60

50

40

30

20

10

O 1 -,110

CSF FLOWmm3. fl

20

PRESSUREW4 5 8 7 kPa3b 'o 5o mm Hg

(b)

PRESSURE7 kPa

5o mm Hg

115


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


J. Ekstedt

365 9

.. 0

..

P RESSUREi kPa

50 mm Hg

379

SP

PRESSURE

CSF FLOWmm3. s7

90

80

70

60

50

40

30

20

10

0 1 2.. 3io

'.

io

CSF FLOWmm3. s1

90

80

70

60

50

40

30

20

10-

4370

(b)

..

70::

,..382 9

...:

* ~~~~~PRESSURE5 6 7 kPa

io 50 mm Hg

188 9

47 11 19I~~~~

:::

...._.

PRESSURE0 2 3 4 5 6 7 kPa O .. 2 3 4 5 6 7 k Pa

10 so30 50so mm Hg io io 30 io 50 mmNH(c) (d)

Fig. 11 Pressure/flow diagrams that were exceptions to the usual rule ofrectilinear relationship betweenpressure andflow.The recordings (a) and(b)frompatients with sinus thrombosis. Patient (c) hadsequelae after a traumatic intracranialhaemorrhage with secondaryporencephaly. Patient (d) /adhadapneumoencephalogram three days earlier. A ta laterinvestigation herpressure/flow curve was again straight.

116

CSF FLOWmm3. i-I

90

80

70

60

50

40

30

20

10

0 i 2 3 4 5 610 2o' 3s 40

(a)

CSF FLOWmm3.*l

90

80

70

60

50

40

30

20

10

L


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from



301 Conductance (Gp P -pmm! kPas c,,/drPI/

20-

10

738 1976-04-12d 52 years

'hr 1.5 kPaq, 8.6 M s-'a 12.2 mWkPi's'Pss 0.7 kPaPdop 0.7 kPa

10 20 30 40 50 60 70 80 90 100 lO 120 min

Fig. 12 Extended CSF infusion. Every five minutes theCSFpressure changed between 4.5 kPa and 5.5 kPa. Timeon abscissa. On the ordinate the conductance was calculatedaccording to theformula in the Figure. There was nosystematic change in the conductance during the 115minute period. In the post-infusion period the pressurereturned to the pre-infusion value.

In the usual type of experiment the pressure was

allowed to return spontaneously in the post-infusionperiod. The plateau value then obtained was usuallyvery close to or identical with the resting pressure

obtained before the infusion. This held true also in theexperiments where the pressure had been kept at ahigh level for an extended time. During the drainageperiod there were usually no subjective sensationsfelt. Only a few patients developed headache duringthe procedure. On rising after the investigation, how-ever, most patients who were left with a subnormalCSF pressure felt a throbbing headache of the typethat is usual after lumbar puncture.

In Fig. 13 is shown the result of a several-hour longdrainage where the CSF formation rate was deter-mined for each 15 minute period. The variation inrate between the different periods is only slight.

7-

6

5

4,

3

2-

CSF Formation rate (qf)mm3s-I

700 1976-01-269 54 years

30 60 90 120 150 180 210 240 270 minFig. 13 Extended CSFdrainage. When the rate ofoutflowthrough the needle hadstabilised the CSFformation ratewas calculatedjor each l5 minuteperiodand thecalculatedformation rateplotted. qf ± SD = 4.0 ± 0.2.

Discussion

Welch and Friedman (1960) made in vitro perfusionsof small discs of the sagittal sinus in young monkeys.They found that the flow through the arachnoid villistarted at a critical opening pressure. The generalshape of the pressure/flow curve was that of an up-ward bend very similar to the curves from the presentinvestigation shown in Fig. 1 1. The authors concludedthat the drainage of CSF was through open channelsfrom the subarachnoid space into the dural sinuses.Colloid osmotic pressure had no influence on theflow. The upward bend of the pressure/flow curvewas interpreted as caused by an elasticity of thearachnoid villi. Further, in the rabbit experiments ofDavson et al. (1970), made with a constant pressuretechnique, there seems to be an upward bend of thepressure/flow curve. However, the deviation from thestraight line is so slight that it may lie within theexperimental error.

In the present investigation a rectilinear relation-ship was found. This held also for the youngestpatients, the very youngest being 3 days old, and cantherefore not be explained as a decreased elasticity ofthe arachnoid villi due to age. Portnoy and Croissant(1976) have also confirmed a rectilinear relationbetween pressure and flow in human experimentswith a slightly different technique.The finding of equal flow values at a certain pres-

sure level when going from lower to higher pressuresas when going from higher to lower pressures makesit reasonable to assume that the arachnoid villi donot get further distended at increasing pressure, atleast not at pressures below 6 kPa.

Martins et al. (1974) made direct measurements ofthe sagittal sinus pressure in man via median burrholes while they increased the CSF pressure by meansof intraventricular infusion. In most of their experi-ments the sagittal sinus pressure was unaffectedwhen increasing CSF pressure by up to 6 kPa.Altered sagittal sinus pressure can thus not explainthe rectilinearity ofthe ratio.There still remains the possibility of a distension of

the villi and thus increased conductance at increasingpressure if that is compensated by a compression ofthe lateral lacunae of the sinus and thus causes a de-crease of the number of villi that could pass CSF tothe sinus. This viewpoint is purely hypothetical andhas no support in the literature. It is highly unlikelythat two mechanisms should balance each other sowell as to nearly always give a rectilinear relationshipbetween pressure and flow.The regression-line offlow on pressure is the hydro-

dynamic conductance of the CSF outflow pathways.The conductance of the intracranial pathways,especially in the aqueduct and in the basal cisterns

117


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


118

and the subarachnoid space over the hemispheres, isprobably very high in relation to the conductance inthe arachnoid villi. The conductance of the outflowpathways may thus certainly be considered as theconductance of the arachnoid villi only. However, inpathological cases there may be blocking of CSFcirculation in the subarachnoidal space-for example,because of adhesions.Leakage around the needles is probably only

exceptionally an outflow pathway for the CSF in theexperimental situation (Fig. 7).The flow through the arachnoid villi will, ac-

cording to Davson et al. (1970), be determined by thepressure difference between the subarachnoid spaceand the venous sinus. The following relation will thenprobably be correct when the system is in dynamicequilibrium.

pC = Pss + q Formula 4Gopwherepc = the cerebrospinal fluid pressure,pss = the sagittal sinus pressure,q = the rate of flow of CSF through the villi, nor-mally the formation rate ofCSF (qf),Gop = the conductance ofthe arachnoid villi.There may also be an opening pressure of the

arachnoid villi-that is, the pressure necessary toopen up the valves before any fluid can pass at all.This pressure was found by Welch and Friedman(1960) to be 0.1-0.5 kPa (10-50 mmH20) in theirmonkey preparation. However, this opening pressuredisappeared when the detergent polysorbate 80 wasadded to the perfusate. This finding makes it probablethat the opening pressure was caused by stickiness ofthe valvular structure. Once the valve has openedthere ought not to be any force acting against flowwhich should have been the case if there was anopening pressure caused by a 'spring loaded valve'.Valvular stickiness cannot influence the followingcalculation ofthe sagittal sinus pressure.From Formula 4 it can be inferred that, if there

were no CSF flow through the arachnoid villi, theCSF pressure should be equal to the sagittal sinuspressure. In all the pressure/flow diagrams presentedin Figs 7-9 a horizontal line is drawn under the X-axis.The position of this line corresponds to the formationrate of CSF for the particular patient. This line thenrepresents the real zero-flow through the arachnoidvilli. If the pressure/flow curve is extrapolated downto this level, a pressure value is obtained whichshould have existed in the subarachnoid space if therehad been no flow ofCSF through the arachnoid villi.The calculation holds true only under the assumptionthat a rectilinear extrapolation of the pressure/flowcurve can really be made. If the curve should deviate

J. Ekstedt

in either direction, another value for the sagittalsinus pressure should be obtained. Whether or not theassumption of a rectilinear extrapolation is true canonly be proved when the sagittal sinus pressure ismeasured directly during a CSF hydrodynamicexperiment. Discussion of the different CSF hydro-dynamic parameters will be postponed to a later paper.

References

Armaly, M. F. (1959). Studies on intraocular effects of theorbital parasympathetic pathways. Archives ofOphthal-mology, 62,817-827.

Ayala, G. (1923). Uber den diagnosticher Wert desLiquordruckes und einen Apparat zu seines Messung.Zeitschrift.fur die gesamte Neurologie und Psychiatrie,84,42-95.

Ayala, G. (1925). Die Physiopathologie der Mechanik desLiquor cerebrospinalis und der Rachidealquotient.MonatschriftfurPsychiatrie und Neurologie, 58,65-101.

Brirny, E. H. (1962). The mode of action of pilocarpineon outflow resistance in the eye of a primate (Cerco-pithecus ethiops). Investigative Ophthalmology, 1,712-727.

BarAny, E. H. (1964). Simultaneous measurement ofchanging intraocular pressure and outflow facility in thevervet monkey by constant pressure infusion. In-vestigative Ophthalmology, 3,136-143.

Becker, B., and Constant, M. A. (1956). The facility ofaqueous outflow: a comparison of tonography and per-fusion measurements in vivo and vitro. Archives ofOphthalmology, 64, 305.

Cutler, R. W. P., Galicich, J., and Watters, V. (1968).Formation and absorption of cerebrospinal fluid inman. Brain, 91,707-720.

Davson, H., Hollingsworth, G., and Segal, M. B. (1970).The mechanism of drainage of the cerebrospinal fluid.Brain, 93, 665-678.

Di Rocco, C., Maira, G., Rossi, G. F., and Vignati, A.(1976). Cerebrospinal fluid pressure studies in normalpressure hydrocephalus and cerebral atrophy. EuropeanNeurology, 14,119-128.

Ekstedt, J. (1973). Cerebrospinal fluid dynamics in adulthydrocephalus. Excerpta Medica Foundation. Interna-tional Congress Series, 296, 126.

Ekstedt, J. (1975). CSF hydrodynamics studies by meansof constant pressure infusion technique. In IntracranialPressure II, pp. 35-41. Edited by N. Lundberg, U.Ponten, and M. Brock. Springer: Berlin.

Ekstedt, J., and Friden, H. (1976). CSF hydrodynamicespecially in the adult hydrocephalus syndrome. InIntracranial Pressure III, pp. 177-185. Edited by J. W. F.Beks, M. Brock, and A. D. Bosh. Springer: Berlin.

Elliott, K. A., and Jasper, H. H. (1949). Physiologicalsalt solution for brain surgery. Journal ofNeurosurgery,6,140-152.

Foldes, F. F., and Arrowood, J. G. (1948). Changes incerebrospinal fluid under the influence of continuoussubarachnoidal infusion of normal saline. Journal ofClinical Investigation, 27, 346-351.


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from



Hussey, F., and Schanzer, B. (1970). A simple constant-infusion manometric test for measurement of CSFabsorption. 2. Clinical studies. Neurology (Minneap.),20,665-680.

Janny, P. (1974). La surveillance de la pression intra-cranienne en neuro-chirurgie. Neurochirurgie, 20,521-550.

Katzman, R., and Hussey, F. (1970). A simple constant-infusion manometric test for measurement of CSFabsorption. 1. Rationale and method. Neurology(Minneap.), 20,534-544.

Lorenzo, A. V., Page, L. K., and Watters, G. V. (1970).Relationship between cerebrospinal fluid formation,absorption and pressure in human hydrocephalus.Brain, 93, 679-692.

Lundberg, N., and West, K. A. (1965). Leakage as asource of error in measurement of the cerebrospinalfluid pressure by lumbar puncture. Acta NeurologicaScandinavica, 41, suppl. 13, 115-121.

Martins, A. N. (1973). Resistance to drainage of cerebro-spinal fluid. Clinical measurement and significance.Journal ofNeurology, Neurosurgery, and Psychiatry, 36,313-318.

Martins, A. N., Kobine, A. I., and Larsen, D. F. (1974).Pressure in the saggital sinus during intracranial hyper-tension in man. JournalofNeurosurgery, 40,603-608.

Masserman, J. H. (1934). Cerebrospinal hydrodynamics.Clinical experimental studies. Archives of NeurologyandPsychiatry (Chic.), 32,523-553.

Nelson, J. R., and Goodman, S. J. (1971) An evaluationof the cerebrospinal fluid infusion test for hydro-cephalus. Neurology (Minneap.), 21,1038-1053.

Portnoy, H. D., and Croissant, P. D. (1976). A practical

method for measuring hydrodynamics of cerebrospinalfluid. Surgical Neurology, 5, 273-277.

Quincke, H. (1878). Ueber den Druck in Transsudaten.Deutsches Archivfuirklinische Medicin, 21, 453-468.

Rubin, R. C., Henderson, E. S., Ommaya, A. K., Walker,M. D., and Rall, D. P. (1966). The production ofcerebrospinal fluid in man and its modification byacetazolamide. Journal ofNeurosurgery, 25, 430-436.

Schaltenbrand, G., and Wordehoff, P. (1947). Eineinfaches Verfahren zur Bestimmung der Liquorpro-duktion und Liquorresorption in der Klinik. Nerven-artze, 18,458-463.

Snedecor, G. W., and Cochran, W. G. (1971). StatisticalMethods. 6th edn., pp. 166-170. Iowa State UniversityPress: Ames, Iowa.

Sokolowski, S. J. (1974). A new quantitative technique forthe assessment ofcerebrospinal fluid absorption in man.Journalofthe Neurological Sciences, 23,37-47.

Trotter, J. L., Luzecky, M., Siegel, B. A., and Gado, M.(1974). Cerebrospinal fluid infusion test. Identificationof artifacts and correlation with cisternography andpneumoencephalography. Neurology (Minneap.), 24,181-186.

Weed, L. H., and Weyeforth, P. (1919). Experimentalirrigation of the subarachnoidal space. Journal ofPharmacology, 13,317-334.

Welch, K. (1975). The principles of physiology of thecerebrospinal fluid in relation to hydrocephalus in-cluding normal pressure hydrocephalus. In Advances inNeurology, vol. 13, pp. 247-332. Edited by W. J.Friedlander. Raven Press: New York.

Welch, K., and Friedman, V. (1960). The cerebrospinalfluid valves. Brain, 83,454-469-

119


on Decem


j.com/

J Neurol N


ebruary 1977. Dow

nloaded from


csf hydrodynamic studies in man1 - bmj · journalofneurology,neurosurgery, andpsychiatry, 1977, 40,...

Documents