28 March 1970 BRTISHMEDICAL JOURNAL 779

Potassium Balance and Acid-base Changes in Patients Undergoing RegularHaemodialysis Therapy

A. G. MORGAN,* M.D., M.R.C.P.; L. BURKINSHAW,t M.A., PH.D.; P. J. A. ROBINSON,4 M.B., M.R.C.P.

S. M. ROSEN§ M.B., M.R.C.P., M.R.C.P.ED.

British Medical J7ournal, 1970, 1, 779-783

iSummary: Serial measurements of total body potassiumin 21 patients with chronic renal failure being

treated with three 10-hour periods of dialysis per week,against a dialysate fluid containing 1.5 mEq of potassiumper litre, showed no evidence of potassium depletion. Mildhyperkalaemia was found in some patients before dialysis,correlated with the pre-dialysis hydrogen ion concen-tration. Hypokalaemia occurred during dialysis in almosthalf of the studies made; the plasma potassium concentra-tion, however, rose to normal levels within two to fourhours of stopping dialysis. A delay in the movement ofpotassium from the cells into the extracellular fluid issuggested as a cause for the observed hypokalaemia.In all but one patient the pre-dialysis blood pH was

normal, but rose to alkalaemic levels during dialysis. Apronounced degree of hypocapnia was noted beforedialysis, and this was not altered by a rising blood pHduring dialysis. It is suggested that a stimulus to respirationother than the hydrogen ion gradient between the braincells and cerebral spinal fluid may produce the observedhypocapnia.

IntroductionPotassium depletion has recently been recorded as a compli-cation of regular dialysis therapy. This study was made todetermine whether our standard treatment regimen for ter-minal renal failure, which incorporates three 10-hour dialysisperiods per week against a dialysis fluid containing 1.5 mEqof potassium per litre, would control hyperkalaemia, withoutproducing potassium depletion.

Patients and MethodsPotassium balance was studied in 21 patients undergoing

maintenance dialysis on a single pass, two-layered Kiil dia-lyser, each patient receiving three 10-hour periods of dialysisper week. A dialysate solution containing 15 mEq of potas-sium per litre was used throughout the study (Table I). Nodietary restriction of protein, salt, or potassium was imposed,the potassium intake being about 80 mEq/day. Cation-exchange resins were neither required nor used.* Senior Medical Registrar, St. James's Hospital, Leeds 9.t Research Fellow, M.R.C. Environmental Research Unit, General

Infirmary, Leeds 1.$ Clinical Scientist, Department of Renal Medicine, St. James's Hospi-

tal, Leeds 9.5 Physician in Charge, Department of Renal Medicine, St. James's

Hospital, Leeds 9.

TABLE I.-Dz'alysate CompositionContents Concentrations

Potassium (acetate). 1-5 mEq/l.Magnesium (acetate) .. 0-2 mEq/l.Calcium (acetate) 2-5 mEq/l.Sodium (acetate).. 27 mEq/l.Sodium (chloride) 103 mEq/l.Dextrose monohydrate 220 mg./100 ml.Total acetate .. 39 mEq/l.

Total body potassium measurements were made with awhole-body radiation counter comprising three plastic scin-tillation detectors fixed round a chair, which gave absolutevalues of total body potassium with a standard error of about4% (Burkinshaw, 1967). Repeated measurements on a normalindividual over a three-year period were found to have acoefficient of variation of about 2 %, therefore allowing a changeof 5 % or more to be detected with confidence. The measuredpotassium content of each patient was expressed both in mEqand as a concentration (mEq/kg. body weight) and comparedwith the mean normal concentration for the patient's age.Previous measurements of 91 normal men aged 18 to 39 and45 normal women aged 18 to 29 had shown that the meanvalues for their potassium concentration, calculated for eachage decade, agreed with the values reported by Anderson (1963).Therefore in this study our own mean values for total bodypotassium have been used in the relevant age ranges, and thoseof Anderson for the other age ranges. The lower limit for thenormal range of total body potassium was taken as twice thestandard deviation below the mean. This was calculated fromour own data where applicable, otherwise a standard deviationof 10% was taken (Anderson, 1963).

Serial pre-dialysis total body potassium measurementswere made on all the patients included in this study, andpost-dialysis measurements were also made in some. Bio-chemical analyses were carried out on blood taken from thearterial side of a modified Scribner shunt. For most patientspre-dialysis and post-dialysis potassium and acid-basemeasurements were also made. Measurements of plasmapotassium, blood sugar, and plasma insulin were made at two-hourly intervals during dialysis on four occasions, and on afurther three occasions when the dextrose content of thedialysate solution had been reduced to 110 mg./100 ml. Plasmapotassium was estimated on a Technicon AutoAnalyzer andblood sugar by an AutoAnalyzer glucose oxidase technique(Morley et al., 1968). Immunoreactive insulin was estimated induplicate by a modification of the method of Hales and Ran-dle (1963), the reagents and Oxoid membrane filter suppliedby the Radio Chemical Centre at Amersham being used.Acid-base measurements were made on an E.I.L. pH meter.

Results

Total body potassium was measured immediately beforedialysis on 46 different occasions in the 21 patients studied,and the mean value recorded for each patient is shown inTable II. These mean values all lie within the normal range.The average duration of maintenance dialysis for this groupof patients was nine months, but with a range of up to 26months.

In two patients in whom the initial total body potassiummeasurements were made within the first two months of dia-lysis low normal values were obtained (79% and 81% of themean normal). Serial measurements, however, showed a pro-gressive rise over the following few months towards the mean

TABLE II.-Effect of Regular Dialysis Therapy on Mean Total BodyPotassium (T.B.K.)

CaseNo.

123456789101112131415161718192021

SexandAge

M. 19M. 33M. 34M. 27F. 40F. 54M. 41M. 39M. 35F. 53M. 21M. 42F. 35F. 26F. 38M. 32F. 29M. 37F. 26M. 42M. 44

Durationof

Dialysis

3 weeks6 weeks7 weeks3 months3 months3 months4 months4 months4 months4 months6 months7 months10 months10 months13 months15 months15 months17 months19 months21 months26 months

Mean Pre-dialysisT.B.K.

mEq mEq/kg. % MeanNormal*

3420 51 5 942990 50 5 962690 51 0 983185 55 0 1021660 45 2 1151755 31 1 872690 46 5 953220 47 1 903600 49 6 952095 34 4 932300 46 1 872940 50 1 1032120 510 1272155 43 6 1022265 45 0 1063755 53 5 1022290 42 0 1003265 49*4 972310 49 7 1163320 53 7 1102405 49 6 102

Lower Limitof Normal

Range for T.B.K.

858282818080808282808580809080829082908080

* See text for definition of the term "normal."

BRITISHMEDICAL JOURNAL

before and after dialysis. In all, 37 paired measurements weremade, but in only six cases did a significantly measurabledrop (<50/0 of initial total body potassium) occur, and in nocase did this exceed 8.50 (197 mEq). No correlation wasfound between the initial potassium concentration (totalbody potassium/kg. body weight) and the change intotal body potassium during dialysis. The mean change in totalbody potassium for the 37 dialysis periods studied was a dropof 48 mEq, or 2 %.Plasma potassium was measured before and after dialysis on

30 occasions (Table III), the mean levels being 4 9 mEq/l. and3.4 mEq/l., respectively. In 16 of these studies a mild degreeof hypokalaemia (plasma potassium <3.5 mEq/l.) was pro-duced, but in only two cases was this accompanied by asignificant change in total body potassium (>5 %). No statisti-cally significant correlation could be found between thechange in total body potassium and the change in plasmapotassium. In five patients plasma potassium measurementswere also made two and four hours after dialysis, and a rapidcorrection of hypokalaemia was found to occur (Fig. 1).The values for pre-dialysis and post-dialysis pH, Pco2,

and bicarbonate are shown in Table III. Initial blood pH valuesfor all but one of the patients lay within the normal range, butboth the Pco2 and bicarbonate levels were low. Post-dialysisacid-base measurements showed no significant change inPco2; therefore the small but satistically significant rise inplasma bicarbonate produced by dialysis resulted in a rise ofblood pH in most patients to alkalaemic levels. The meanpre-dialysis and post-dialysis values for pH, Pco2, and bi-carbonate are shown in Table IV. The regression equations

TABLE IV.-Mean Values with Standard Deviation for Pre-dialyjsis andPost-dialysis pH, Pco,, and Bicarbonate

value for the group (mean= 100.6°0(group, S.D.= 10-3; S.E.=2-3).

of the normal control

In an attempt to ascertain potassium loss during dialysis,total body potassium measurements were made immediately

Pre-dialysis (S.D.)Post-dialysis (S.D.) ..

pH Pco2(mm. Hg)

7 438 (0 04) 28-75 (3-31)7 494 (0 03) 29-00 (2 21)

TABLE III-Effect of Dialysis on Total Body Potassium (T.B.K.), Plasma Potassium, and Arterial pH, Pco2, and (HCO3)

Pre-dialysis T.B.K.

mEq/kg.Body Wt.

51 550 550 551 055*055.745-245 229131 146 546 047 133 634.446 151 648 553 642 044 642 344.542 047.551 554.453 149 848-1

0O of MeanNormal*

9496969810210311511579879588908993841061181331001041011081009112011210910699

Plasma K(mEq/l.)

Pre-dialysis

4.55 85.75 04.44.94.44 64-14-84-25-45 -74-35 14-75 -34.94.55.53 84-26-25.44.35-64.94-85 24.9

Post-dialysis

3 63.93 23 23.53 83 02 72 52-63 23 63.93.33.73.34 03.43.43 62 63 23.74.33.43.93 03.53 63 2

Change in T.B.K.During Dialysis

mEq

+ 121- 86- 49-89+ 46+ 44-69-8-68- 101-44-18+ 51+ 88- 1034±8- 160- 10-89-38-180-41- 49- 12+ 73- 197-46H 18- 150-8

0/ ofPre-dialysisT.B.K.

+ 3 6- 29-1 6-- 3 2+ 14+ 13- 39- 0 5- 43- 5 8-1-6- 06+ 16+ 44-5 0+ 0 4- 52-05

4 0-177.91-92-10-5

+ 23-831-3

+ 0-6-5 8- 03

Arterial Acid-base Measprements

Pre-dialysis

pH

7-447 407.397-457-46

7.497.477-487-477-487-377-377-48

7-417-42

7-48

7-42

7-387.497-42

Pco2

2921203330

2827303130293228

2833

29

29

3328

27

(HCO3)

1913122121

2119192222161820

1820

21

18

1922

17

pH

7-487-477-527-467.537-547-467-517 507-477-487.497-52

7-537-51

7 51

7 50

7-457 48

7.47

Post-dialysis

Pco2 (HCO3)

2826233228

2828282932283128

2932

30

31

3229

28

*See text for definition of the term "normal."

780 28 March 1970 Haemodialysis Therapy-Morgan et al.

Bicarbonate(mEq/l.)

18-9 (2 7)21-7 (1-95)

CaseNo.

1

34

5

678

10111213

14

151718192021

2118182024

2320202323212123

2425

24

23

2221

20

-il-

*____ ~~~~~~II_-

* I _ --_ I

Il_ I1~ I_

28 March 1970 Haemodialysis Therapy-Morgan et al.F gure 2Figure 1

E 4*0

0._

c,30

2-0Pre-dia Iysis

40

a,

EE

0a-

Post-dialysis 2 hours 4 hoursPost-dialysis

0E

E

2, 5.0.

s 0

4,4v

CLX 4-0

II Is 20Plasma bicarbonate (mEq / 1.)

BIUTISHMEDICAL JOURNAL

Figure 30

00 0 .

0

0

14 patients23 readingsr=+066P<0-001K=0-09 + 0-13 [ H+]

A 44 4 .25 44 40Pre-dialysis [H+] ngEq/l.

FIG. 1.-Plasma potassium levels measured in five patients immediately before and after dialysis, and also at two and four hours after dialysis. FIG. 2.-Regression lines for the correlation between Pco2 and plasma bicarbonate. Line A, obtained from normal controls given either an acid or alkali load (Lennonand Lemann, 1966). Line B, obtained from patients with untreated renal failure (Lennon and Lemann, 1966). Lines C and E, obtained before and afterdialysis from patients on regular haemodialysis (Earnest et al., 1968). Line D, pre-dialysis regression line obtained from our data, Pco2= 11-96 (±3-81) +0-89(+0-20) [HCOA;] r= +0-72, P<0-001. Line F, post-dialysis regression line obtained from our data, Pc02= 15-33 (+4-84)+0-63 (±0-22) [HCO3];r=0-56, 0-01 >P>O-001. FIG. 3.-Relation of pre-dialysis [H* ] to pre-dialysis plasma potassium.

for the correlation between Pco2 and bicarbonate have beencalculated for both the pre-dialysis and post-dialysis measure-ments, and are shown graphically in Fig. 2 along withdata obtained by other workers. The level of Pco2 for anygiven level of plasma bicarbonate was found to be lower thanexpected. A highly significant correlation (P<0.001) wasfound to exist between the pre-dialysis blood pH hydrogenion concentration and plasma potassium (Fig. 3), but no cor-relation was found between the post-dialysis blood pHhydrogen ion concentration and the plasma potassium.The blood sugar and insulin levels measured during dialysis

in the seven patients studied were all within normal physio-logical limits. On three occasions dialysis was performed withthe dextrose content of the dialysate reduced to 110 mg./100ml. and hypokalaemia was still observed (3-2, 2.8, and 3.0mEq/l. respectively).

DiscussionThe control of hyperkalaemia in patients undergoing

maintenance dialysis has recently relied on both increasing thefrequency of dialysis and lowering the potassium concentra-tion in the dialysate fluid, rather than using ion exchangeresins. At this moment, however, there is a debate about theoptimum concentration of potassium in the dialysate fluidwhich will allow control of hyperkalaemia without producingpotassium depletion (Table V). Johnson et al. (1969) andNovak et al. (1969) showed that if dialysis is limited to twoi2-hour dialysis periods per week potassium depletion is notproduced even if a potassium-free dialysate is used. If,however, dialysis against a potassium-free fluid is increased

to thrice-weekly, falls in total body potassium of over 20°%may be produced. Seedat (1969) likewise found evidence ofpotassium depletion in all his patients dialysed three times perweek against a potassium concentration of 1 mEq/l. In ourunit, using three 10-hour dialysis periods per week, it wasfound that dialysis against a fluid containing 0.5 mEq ofpotassium per litre produced severe and symptomatic hypoka-laemia, while dialysis against a 2.5 mEq/l. potassium concen-tration did not adequately control hyperkalaemia. This studywas therefore undertaken, and it confirms that patients on afree potassium intake (about 80 mEq/day) and dialysed forthree 10-hour periods per week against a potassium con-centration of 1-5 mEq/l. do not develop total body potassiumdepletion, yet adequate control of hyperkalaemia is stillachieved.One of the main advantages of thrice-weekly dialysis for

the patient with terminal renal failure is that a free proteinintake can be allowed, and in consequence a pronounced im-provement in nutritional state occurs. In two of the patients inthis study total body potassium measurements were madeshortly after starting maintenance dialysis and were low nor-mal, but as their nutritional state improved and lean bodymass increased, the values progressively rose towards themean value for the group (mean=100-6% of the normal con-trols). This return of total body potassium concentration(measured in mEq/kg. body weight) to normal in patientstreated with adequate dialysis and increased dietary proteinhas also been commented on by other workers (Ram andChisholm, 1969).The mild degree of hyperkalaemia (mean 4.9 mEq/l.) seen

in our patients would seem to be a function of the partially

TABLE V.-Summary of Some Recent Publications Giving Details of Potassium Balance in Patients with Chronic Renal Failure on RegularHaemodialysis Therapy

Authorof

Publication

Ram and Chisholm (1969)

Novak et al. (1969) . .

Seedat (1969)Johnson et al. (1969) . .

Seedat (1969)Johny et al. (1969)

No. ofCases

No. ofDialyses

perWeek

Lengthof

Dialysis(Hours)

DialysatePotassium

Concentration(mEq/l.)

- _ I-7

7

1010216136

15

22222233

121212128-5

121212

1*52-72-0Nil1.0NilNil1.0

1-1-4

781

* v.

0S

3b 32

DietaryPotassiumIntake(mEq)

7070FreeFree58FreeFree58

Incidenceof

PotassiumDepletion

NoneNoneNoneNone

1 patientNone

6 patientsAll patients11 out of 29

studies

EstimatedPotassium

Loss DuringDialysis(mEq)

163 (5-7%)124 (43%)

220

. . . . . . .

compensated metabolic acidosis seen before dialysis. A highlysignificant (P<0-001) correlation was found between the pre-

dialysis plasma potassium and hydrogen ion, and it can be

calculated from the regression equation that a 0-1 unit fall in

pH would raise the plasma potassium by 1.2 mEq/l. This

movement of potassium out of the cells in exchange for extra-

cellular hydrogen ions in an acidosis is extremely well docu-

mented (Burnell et al., 1956). Siggard Anderson (1962)found a similar relationship between pH and plasma potassiumand showed that a fall of 01 unit of pH may raise the

plasma potassium by about 1 mEq/l., which corresponds

closely to our findings in this study. Thus adequate maintenance

dialysis, by preventing the development of any pronounced

degree of acidosis, will help to prevent pre-dialysis hyper-kalaemia.

Hypokalaemia was encountered at the end of dialysis in

almost half of the dialysis periods studied. This could not be

explained by potassium removal during dialysis, since the

mean change in total body potassium produced by dialysiswas extremely small (-48 mEq). No correlation was found

between the change in total body potassium and the fall in

plasma potassium (P>0.05). According to the results of

Scribner and Burnell (1956) a loss of between 100 and 200

mEq of potassium is required to reduce the serum potassiumlevel by 1 mEq/l. when the initial serum concentration was

more than 3 mEq/l. Similar results have also been publishedby many other workers.

Siggard Anderson (1962) showed that an uncompensatedrespiratory alkalosis can produce a fall of up to 3 mEq/l. foreach unit rise in pH. In our patients, during dialysis a rise of

pH did occur (mean pre-dialysis pH 7.438; mean post-dialysis pH 7.494), but in spite of this no correlation could be

found between either the change in pH or the final pH, andthe change in plasma potassium or the final plasma potassiumlevel.

Seedat (1968) reported that some of his patients withchronic renal failure developed hyperglycaemia, hyperinsu-linaemia, and hypokalaemia during dialysis against a dialysatefluid with a high dextrose content. It was concluded that thesefindings could be due to coexisting hypokalaemia. Serialmeasurements of blood sugar and plasma insulin were madeduring seven dialysis periods in order to exclude abnormalcarbohydrate metabolism as a possible cause for the hypo-kalaemia produced by dialysis. All measurements were found tobe within the normal physiological range, therefore suggestingthat in our patients carbohydrate metabolism was essentiallynormal. To exclude further any effect of the absorbed dex-trose from the dialysis fluid on plasma potassium, however, thedialysate dextrose concentration was reduced to 110 mg./100ml. for three dialyses. Hypokalaemia still occurred duringdialysis, again suggesting some other causative mechanism. Bythe same token, this degree of hypokalaemia in the absence oftotal body potassium depletion does not result in abnormalcarbohydrate metabolism. It is well documented that theimpairment of carbohydrate metabolism seen in chronicuraemia is potentially reversible if the patients are maintainedon regular dialysis (Hutchings et al., 1966; Losowsky andKenward, 1968; Seedat, 1969).About 97% of the total body potassium is intracellular; thus

any removal of potassium from the extracellular fluid must becompensated for by the movement of potassium from cells. Ifthere is a delay in movement of potassium across the cellmembrane during dialysis the fall in potassium will be greater

than anticipated. This thesis is supported by our results, as

the hypokalaemia produced in our patients, in spite of a nor-

mal total body potassium, is rapidly corrected once dialysis isstopped.

Pre-dialysis acid-base measurements showed that, withthe exception of one patient, the blood pH was within the

BRITISHMEDICAL JOURNAL

normal limits, but that both PCO2 and bicarbonate were low.Dialysis was performed against a dialysate containing 39 mEqof acetate per litre, and it has been shown that it is rapidlymetabolized to bicarbonate and, in consequence, significantarterial acetate levels are not encountered (Earnest et al., 1968).Post-dialysis acid-base measurements showed no change in thePCo2 concentration, a small but statistically significant rise inbicarbonate, and a greater rise in blood pH to alkalaemic levels.

Correlation between PCO2 and Plasma Bicarbonate

Anderton et al. (1965) noted that in patients with renalfailure the ventilatory threshold to PCo2 was significantlyreduced and that this did not become normal even after restora-tion of the plasma bicarbonate to normal values or above. Asimilar abnormality in the ratio of bicarbonate to PCo2 hasbeen found in our study (Fig. 2). The regression equations forthe relationship between PCo2 and bicarbonate agrees closelywith those obtained by Earnest et al. (1968). They differ,however, very considerably from those obtained in eitheracidotic or alkalotic normal controls, or in untreated uraemicpatients (Lennon and Lemann, 1966).

It has been known for many years that during the course ofrecovery from a metabolic acidosis many patients will passthrough a phase of respiratory alkalosis as the concentrationof bicarbonate in the blood approaches normal values (Peters,1917). This phase of respiratory alkalosis has also been notedto occur after haemodialysis for renal failure (Weller et al.,1953; Cowie et al., 1962). Lambie et al. (1965) showed that inhaemodialysis of acidotic patients with renal failure, once theextracellular fluid bicarbonate level was restored to normallevels the gradient for hydrogen ions between the extracellularfluid and intracellular hydrogen ions was reduced to normal.It has been shown that bicarbonate ions pass extremely slowlyinto the cerebrospinal fluid from plasma, and in consequence,though the plasma pH may be restored to normal, the C.S.F.pH remains low (Cowie et al., 1962; Rosen et al., 1964). It wastherefore postulated by Lambie et al. that, provided the changein hydrogen ion concentration in the brain cells follows thealteration in reaction of the body cells, there must be an alteredhydrogen ion gradient between the brain cells and theC.S.F., and it is this gradient which may be responsible forthe persistent hyperventilation. In all but one of the patientsin this study, however, and in all those in that of Earnest et al.(1968) the pre-dialysis pH values were normal. A rise inblood pH during dialysis did occur in both studies, and thusthe production of an abnormal hydrogen ion gradient mayexplain the continuing hypocapnia after dialysis, but in ouropinion it cannot explain the hypocapnia seen before dialysis.

It is also of interest that Page and Kallmeyer xl969), studyingthe acid-base changes produced by a 72-hour peritonealdialysis in patients with renal failure, found that, thoughthe blood pH rose from a mean initial value of pH 7.31 to a

post-dialysis mean of pH 7-50, the C.S.F. pH remainedwithin almost normal limits. At the end of dialysis, however,the hypocapnia had been virtually abolished. It must thereforebe reconsidered whether some stimulus other than hydrogenion is driving respiration in these patients with chronic renalfailure (Pauli et al., 1963; Lennon and Lemann, 1966; Earnestet al., 1968).

We are grateful to Dr. G. R. Dykes, of the chemical pathologydepartment at the Leeds General Infirmary, for the insulin assays,and to Miss D. W. Krupoweiz for the total body potassium

measurements. We would also like to thank the medical and nurs-ing staff of the renal dialysis unit at St. James's Hospital for theirco-operation and devoted care of the patients.This work was assisted by a grant from the Hospital Trust Fund

of the Leeds Regional Hospital Board.

782 28 March 1970 Haemodialysis Therapy-Morgan et al.

28 March 1970 Haemodialysis Therapy-Morgan et al. BRITISH 783

REFERENCES

Anderson, E. C. (1963). Annals of the New York Academy of Sciences,110, 189.

Anderson, 0. S. (1962). Scandinavian Journal of Clinical LaboratoryInvestigation, 14, Suppl. No. 66, p. 1.

Anderton, J. L., Harris, E. A., and Robson, J. S. (1965). ClinicalScience, 28, 251.

Burkinshaw, L. (1967). Physics in Medicine and Biology, 12, 477.Burnell, J. M., Villamil, M. F., Uyeno, B. T., and Scribner, B. H.

(1956). Yournal of Clinical Investigation, 35, 935.Cowie, J., Lambie, A. T., and Robson, J. S. (1962). Clinical Science, 23,

397.Earnest, D. L., Sadler, J. H., Ingram, R. H., and Macon, E. J. (1968).

Transactions of the American Society for Artificial Internal Organs,14, 434.

Hales, C. N., and Randle, P. J. (1963). Biochemical Journal, 88, 137.Hutchings, R. H., Hegstrom, R. M., and Scribner, B. H. (1966). Annals

of Internal Medicine, 65, 275.Johnson, W. J., Frohnert, P. P., and Novak, P. P. (1969). Paper read at

Fourth International Congress of Nephrology (Stockholm).Johny, K. V., Lawrence, J. R., Halloran, M. W., Wellby, M. L., and

Worthley, B. W. (1969). Paper read at Fourth International Con-gress of Nephrology (Stockholm).

Lambie, A. T., et al. (1965). Clinical Science, 28, 237.Lennon, E. J., and Lemann, J. (1966). Annals of Internal Medicine, 65,

265.Losowsky, M. S., and Kenward, D. H. (1968). Journal of Laboratory

and Clinical Medicine, 71, 736.Morley, G.,, Dawson, A., and Marks, V. (1968). Proceedings of the As-

sociation of Clinical Biochemists, 5, 43.Novak, P. P., Frohnert, P. P., and Johnson, J. W. (1969). Paper read at

Fourth International Congress of Nephrology (Stockholm).Page, P. F., and Kallmeyer, J. C. (1969). South African Medical Journal,

43, 111.Pauli, H. G., Riedwil, H., Reubi, F., and Wegmiller, W. (1963). Clinical

Science, 25, 37.Peters, J. P., jun. (1917). American Yournal of Physiology, 44, 84.Ram, M. D., and Chisholm, G. D. (1969). Lancet, 1, 260.Rosen, S. M., O'Connor, K., and Shaldon, S. (1964). British Medical

Journal, 2, 672.Scribner, B. H., and Burnell, J. M. (1956). Metabolism, 5, 468.Seedat, Y. K. (1968). Lancet, 2, 1166.Seedat, Y. K. (1969). British Medical Journal, 2, 344.Weller, J. M., Swan, R. C., and Merrill, J. P. (1953). Journal of Clinical

Investigation, 32, 729

Cardiovascular State of Newly Discovered Diabetic Women*J. A. WEAVER,t M.D.,M.R.C.P. ; S. K. BHATIA,t M.B., B.S., M.R.C.P.ED. ; D. BOYLE,t M.D., M.R.C.P.

D. R. HADDEN,t M.D., M.R.C.P.ED.; D. A. D. MONTGOMERY,t M.D., F.R.C.P.

British Medical J7ournal, 1970, 1, 783-786

Summary: A cardiovascular study of a group of 90newly diagnosed diabetic women aged 35 to 75 years

was begun in 1965 and a repeat examination was carriedout on the same patients in 1968. A high prevalence ofischaemic heart disease was found in these patients at thetime of diagnosis, and this finding had some predictivevalue as regards. prognosis over the three-year period.A comparative study with general medical outpatients

and long-established diabetics (greater than 10 years' dura-tion of disease) confirmed the high prevalence of ischaemicheart disease in late-onset mild diabetics controlled bydiet or oral drugs. It is suggested that this type of milderdiabetic patient contributes in undue proportion to thehigh prevalence of ischaemic heart disease in diabetes.

IntroductionThe general association between diabetes mellitus and cardio-vascular disease Is well established, both from pathologicalreports (Warren and LeCompte, 1952) and from clinical stu-dies (Bradley and Bryfogle, 1956). These observations haveconcerned groups of diabetic patients with established diseaseof some duration, but the cardiovascular state of diabetics atthe time of initial clinical onset of the disease has not beenstudied in detail. Liebow et al. (1964) found that one-third ofa group of 58 women had definite cardiac abnormalities at thetime of first diagnosis of diabetes mellitus. We have similarlyexamined the cardiovascular system in a group of newlydiagnosed diabetic men and women at their first medicalpresentation in 1965 and have repeated the examination threeyears later. The present report concerns only the womendiabetics-it seems suitable to deal with women and menseparately, since a recognized characteristic of ischaemic heartdisease in the diabetic is the unusually high proportion ofwomen affected.

*Presented as a paper at the Spring Meeting 1969 of the BritishDiabetic Association held in Belfast.

f Consultant Physician.1 Research Fellow.Sir George E. Clark Metabolic Unit, Royal Victoria Hospital, Belfast

BT12 6BA.

Patients and MethodsNinety consecutive women patients aged 35 to 75 years

were first examined in 1965 at the same time as the diagnosisof diabetes mellitus was established (group A). The clinicalhistory paid attention to the occurrence of angina, intermittentclaudication, details of smoking habits, and parity. The ex-amination included the measurement of body weight, bloodpressure in the supine position, and the presence of posteriortibial artery pulsation. A 12-lead electrocardiogram wascarried out in all patients, and the tracing was analysed ac-cording to the Minnesota classification (Blackburn et al., 1960)by a cardiologist (D.B.), who did not know the clinical data.He was also unaware whether the tracing was from a diabeticpatient or a control subject (see below). None of the elec-trocardiograms was recorded on the day of performance of aglucose tolerance test. Venous blood samples were taken forestimation of blood glucose by a ferricyanide method (Auto-Analyzer) and cholesterol (Zlatkis et al., 1953) at the time ofthe initial clinical examination, which was before any dietaryadvice was offered.Two control groups were also studied. The first control

group of 84 non-diabetic general medical outpatients (groupC), selected by a normal response to a 50-g. oral glucosetolerance test (two-hour blood glucose <110 mg./100 ml.),was examined similarly to the new diabetic patients, includingthe electrocardiographic test. These patients had various dis-orders-gastrointestinal tract symptoms, simple goitre, anxi-ety state-and included also some dental clinic outpatients.The second control group consisted of patients with estab-lished diabetic disease of greater duration than 10 years(group D). This group was attending the diabetes outpatientclinic and consecutive patients were taken without regard tocardiovascular status, the only criterion apart from duration ofdisease being the age group 35 to 75 years. All the aboveinvestigations were also undertaken in this group.

In 1968 the 90 newly diagnosed women diabetics of 1965were re-examined and tested in an identical fashion to theirinitial presentation (group B). Thus four groups of electro-cardiograms were available for analysis: (A) newly diagnosed