338

KETOSIS AND COMA IN DIABETES MELLITUSBy B. A. YOUNG, M.D., M.R.C.P.

Physician, St. Alfege's Hospital; Physician-in-Charge, Diabetic Clinics, St. Alfege's and Miller Hospitals, Greenzcich

During recent years new ideas on certainaspects of the mechanism and treatment of diabeticketosis and coma have developed, supported by aconsiderable amount of clinical and experimentalevidence. This paper attempts to bring togethersome of the newer views, and the opportunity istaken of reiterating some old ones.

Mechanism of KetosisKetosis results from the accumulation in the

blood of ketone bodies which are formed duringthe oxidation of fatty acids to produce energy.These substances, beta-hydroxybutyric acid, aceto-acetic acid and, secondarily acetone, are producedduring normal metabolism and are found in theblood of normal subjects (Barnes and Wick, I939).It is only in fasting conditions in the normal andin cases of severe diabetes mellitus that ketonebodies accumulate in significant amounts. Duringfasting ketosis is due to insufficient ingestion ofcarbohydrates; in diabetes, to insufficient carbo-hydrate metabolism arising from lack of insulin.The diabetic ketosis, if uncontrolled, will proceedto a fatal coma, partly caused by ketotic intoxica-tion and partly by the profound changes in electro-lyte balance due to concomitant metabolicdisturbances.

During the past decade theories of themechanism whereby carbohydrate lack causesketosis have changed considerably. The earlierconception of ketosis (Geelmuyden, 1904) was thata chemical union took place during the course ofmetabolism between carbohydrate and ketonebodies, resulting in a compound subsequentlyoxidized; a process harmless to the body. Aketogenic-antiketogenic ratio was postulated im-plying a quantitative relationship between theoxidation of ketones and glucose (Schaffer, 1923).If carbohydrate metabolism were deficient it wasassumed that the chemical combination of glucosewith fatty acid did not take place and that the un-united ketone bodies were liberated into the bloodstream to produce ketosis.

Recent work has shown that the muscles of thebody can absorb and oxidize large quantities ofacetoacetic and beta-hydroxybutyric acid, theketone bodies playing a part in metabolism in-

dependent and not unlike that of glucose (Law-rence, 1942; MacKay, 1943; Young, I944).These substances, formed by partial oxidation offatty acids in the liver, are distributed to thetissues for complete combustion. MacKay (I943)suggested two forms in which fuel is supplied tothe body, one in which fat is used directly by thetissues and the other whereby the glucose andketone bodies secreted by the liver are oxidized toto produce energy. Glucose and ketones hedescribes as ' quick fuel.' Glucose is the onlysource of supply of' quick fuel' to the brain andnervous system, whereas both glucose and ketonescan be utilized by all other tissues. If available,glucose is the ' quick fuel' of choice; if unavail-able the tissues, other than the brain, make up forthe deficiency by oxidizing ketone bodies. Inketogenic states from 30 to 80 per cent. of theenergy requirements of the body may be suppliedby ketones (Barnes, et al., 1940), and during ex-tended fasting it is possible that not more than ioper cent. of the total fuel requirements ofthe tissuesare supplied by glucose (Drury, 1942). Ketonebodies are, therefore, no longer regarded aspoisons necessarily harmful to the organism, beingused equally efficiently as a source of energy inthe ketosis of diabetes, of fasting conditions and innormal metabolism (MacKay, 1943).The danger of ketosis is one of degree and is

dependent on the non-availability of carbohydrateas the ' quick fuel' of choice. In diabetes the in-tensity of the ketosis, particularly if it developsrapidly, contributes to the production of coma.Ketosis is not the only factor in the production ofdiabetic coma, which is the result of a generalizedmetabolic disturbance of which ketosis is a part.Evidence of ketosis and its degree is, however, auseful guide to the control of diabetes and theprevention and treatment of coma.

The Mortality of Diabetic ComaIn considering the mortality rates of diabetic

coma reported by various writers comparison isdifficult owing to the different clinical criteria,degrees of diabetic ketosis and to the many othervariable factors involved. Rabinowitch (I939) andCollen (1942 a, b) analysed large series of patients

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from

fJul 1951 YOUNG : Ketosis and Coma in Diabetes Mellitus 339

suffering from diabetic coma in an effort to com-pile a formula for the calculation of a severityindex on which to estimate the severity of ketosisand coma. Both conclude that the four most im-portant factors affecting prognosis are, in order ofsignificance, age, degree of unconsciousness,diastolic or systolic blood pressure and com-plicating diseases, particularly infection. Of lessimportance are duration of coma, blood non-protein nitrogen and the blood sugar, the latterinfluencing the mortality to a slight degree andonly in unconscious patients. Both writers statethat the CO2 combining power of the plasma is oflittle or no importance in evaluating the severityof coma and bears no relation to mortality. Thiswas confirmed by Nicholson (1947).

All writers find a negligible mortality in patientswith ketosis without disturbance of consciousness.Rabinowitch (I939) finds a I3.3 per cent. mor-tality with drowsiness, rising through i6 per cent.(semi-conscious) to 53 per cent. (unconscious butresponding to pain) and Ioo per cent. in completelycomatose patients. Colleen (I942) reports a mor-tality rate of 83.8 per cent. in completely un-conscious patients. Nicholson (1947) has a

mortality of 48.8 per cent. in comatcse patientsirrespective of age, but with no complications.Franks (1947), using Collen's severity index(Collen, 1942 a, b), finds a 36 per cent.. and 52 percent. mortality in severe cases of coma treated bytwo different methods. Joslin (I946) reports thelowest mortality of Io per cent. in unconsciouspatients of all ages, with a 33.5 per cent. fatalityrate in 65 patients in the sixth decade, includingconscious, drowsy and unconscious patients.These figures of patients treated by physicians ofconsiderable experience, supported by teams ofwell-trained helpers, clearly demonstrate thatdiabetic coma is still responsible for a largenumber of fatalities, particularly in its later stages.Differential Diagnosis of Diabetic ComaThe investigation of a suspected case of diabetic

coma is not complete until a full and detailedphysical examination of all systems has beenundertaken. Two aspects only of the differentialdiagnosis will'be considered.

Inwulin reaction and diabetic coma. A differentia-tion between insulin and diabetic coma has to bemade in diabetics. From the number of patientssuffering from insulin reactions sent to hospitalwith a diagnosis of diabetic coma tlis would ap-pear to be difficult. This is not so as there is littlesimilarity, other than a state of unconsciousness,in the two conditions. The mode of onset of comais important. The hypoglycaemic reaction issudden, usually occurring in a healthy diabetic.Diabetic coma may advance rapidly but always

supervenes on a previous state of ill health or acuteillness. There is no dehydration in insulin coma,whereas in diabetic ketosis, which has progressedto coma, this should be an obvious feature. Airhunger is a most important sign of ketosis and ischaracteristic of this condition. An extensor-plantar response is an almost constant accompani-ment of insulin coma and is absent in ketosis unlessco-existing cerebral disease is present. The smellof acetone in the breath and the large amount ofketoneuria in the comatose ketotic is never simu-lated by the hypoglycaemic patient. In rare casesof serious doubt a second specimen of urine takena short time after the first, and still showing aheavy ketosis, will put the diagnosis of diabeticcoma beyond doubt.

The diabetic acute abdomen. Abdcminal symp-toms are commonly associated with diabeticketosis. In the absence of any underlying in-flammatory condition nausea, vomiting and in-tense pain occur, the latter often of a colickynature and usually preceding the vomiting. Theremay be general or local abdominal tenderness andrigidity. The temperature is often raised and thepulse rapid. In ketosis a leucocytosis up to 50,000W.B.C. is not uncommon (Lawrence, 1950). Thiscondition, known as the diabetic acute abdomen,may therefore be very difficult to distinguish fromdiabetic ketosis associated with an acute surgicalabdomen. The condition being more prone todevelop in younger patients (Beardsxood, I935),is frequently mistaken for acute appendicitis.Several theories of the cause of the condition havebeen put forward, none of which is convincing.

Active treatment of the dehydration and ketosisresults in a rapid disappearance of abdominalsymptoms as the ketosis improves. Failure ofsuch a response makes laparotomy advisable.Serious though an unnecessary laparotomy duringdiabetic ke'tosis may be, it is far less harmful thanthe results of an untreated peritonitis.Treatment of Diabetic Ketosis and Coma

Diabetic coma is largely a preventable condition.Taken early the abolition of ketosis is relativelysimple, once established, coma is commonly fatal.Prophylaxis consists of (a) the early diagnosis ofthe severe diabetic in whom symptoms and signs,including urinary findings, should make diagnosisa simple matter, and (b) the education andregular supervision of established diabetics, pre-ferably at a diabetic clinic with all the necessaryfacilities for the investigation and treatment ofthe disease and continuity of medical attention.The close co-operation between the clinic doctor,the diabetic and his general practitioner does muchto abolish diabetic coma in a district served by anestablished diabetic clinic.

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from

340 POSTGRADUATE MEDICAL JOURNAL July 1951

The main principles in the treatment of diabeticketosis and coma are: (I) to provide insulin inoptimum dosage; (2) to replenish the dehydratedtissues with adequate amounts of fluid; and (3)to replace the electrolyte loss from the body, inparticular those of chloride, potassium andphosphorus.The treatment of diabetic ketosis (pre-coma)

and of fully developed diabetic coma are twodifferent problems and will be discussed separately.

r. Diabetic ketosis (pre-coma). Mild cases ofketosis without symptoms and signs, apart from apositive Rothera's test in the urine, may be treatedas out patients by re-adjustment of the insulindose. They should, however, be seen frequentlyand admitted to hospital should the condition notrespond promptly.More severe cases which have progressed to

dehydration and drowsiness and show a heavyketosis by the ferric chloride test require admissionfor intensive treatment. Such patients are con-scious and can swallow, a fact which simplifiestreatment considerably. If vomiting is severe in-travenous fluids are necessary as for the treatmentof coma; if slight it will usually cease with treat-ment. Gastric lavage is very valuable in persistentcases. Most ketotic patients are constipated andrequire an enema.

It is my practice immediately to institute afour-hourly regime of insulin injection withgfucose solution and fluids by mouth. In an adultan insulin dose of 24 to 40 units of soluble insulinis given followed by 50 g. of glucose in -2 pt. ofwater flavoured with ' diabetic' fruit juice. Theinsulin dose varies with the intensity of the ketosisand in very severe cases an initial dose of up toioo units may be necessary. In the case of childrenhalf the adult dose is sufficient. Thereafter thebladder is emptied three hours following an in-sulin injection and again shortly before the nextinjection is due at the fourth hour. The secondspecimen is tested for sugar and ketones and theinsulin dose adjusted accordingly. The followingschedule is prescribed, according to the severity ofthe case, for the guidance of the nursing staff:

Urine(Benedict's Red Yellow Green Blue

Test)Soluble insulin 40-24 32-16 i6-8 NIil

units units unitg

The insulin injection is followed by o5 g.glucose as before.

Between these feeds the patient is encouraged totake water flavoured with ' diabetic' fruit juice,tea with a little milk and Marmite made into soup

or clear soup with the addition of salt. The pro-gress of the patient is judged by the diminishingketosis and no effort is made to render the urinesugar-free. Blood sugar control, although of in-terest, is not necessary.

In my experience the large majority of cases,even of severe ketosis, give no anxiety in thisregime and lose all ketoneuria within 24 hours.

For the second 24 hours a high carbohydrate,low fat diet containing 200 g. carbohydrate and notmore than 15 g. fat is prescribed, and soluble in-sulin given in three equally spaced injectionsduring the day in a dose based on the patient'sprevious insulin requirements and needs duringthe period of ketosis. I do not use protamine zincinsulin in the treatment of severe ketosis, butduring the third 24 hours a before-breakfastmixture of 2 to I soluble to protamine zinc in-sulin is given supplemented, if necessary, bysoluble insulin before two other main meals.

This scheme of treatment is based on Hims-worth's (I932) experience and has been used byme for the last I9 years. I have had no cause toamend it and have never failed to avoid the de-velopment of coma by its use in a ketotic, consciousdiabetic patient.

2. Diabetic coma. The main general measuresto be adopted in diabetic coma include the treat-ment of shock with warmth and stimulants. Anenema should be given and a stomach wash,outshould not be omitted.

It is when insulin dosage, parenteral fluid ad-ministration and the replacement of depleted bodyelectrolytes are considered that differences ofopinion occur, and these aspects of treatment willbe discussed separately.

(a) Insulin dosage. All are agreed that insulinshould be given as early as possible in the treatmentof diabetic coma. Root (i945) states that insulingiven six hours after the onset of coma i worth,unit for unit, less than one-third of its value duringthe first hour. For this reason, provided thegeneral practitioner is sure of the diagnosis, agreat deal of good is done by the injection of 40units of insulin while the patient is awaiting thearrival of an ambulance.

In considering the insulin dose to be given in theearly hours of coma a balance must be struck be-tween the optimum amount which can be utilizedby the patient and the dose which might pre-cipitate hypoglycaemia. In the past it is probablethat too low doses have been used initially. Onthe other hand recent papers have suggested highinsulin doses which appear neither necessary norsafe for general use. Micks (1948) uses as much as500 units of soluble insulin as the first dose,followed by Ioo units as frequently as every iominutes if improvement does not occur within an

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from

July I951 YOUNG: Ketosis and Coma in Diabetes Mellitus 341

hour. Even pre-coma cases receive ioo units re-peated half-hourly until improvement is obvious.Such doses have been criticized by Lawrence andOakley (1948) and Dunlop and Donald (1948).Although the results of treatment have been goodin the hands of this writer (Micks), the danger ofsevere hypoglycaemia would be very great if suchdosage were adopted generally. Root (1945) andJoslin (1948) recommend a scale of dosage basedon blood sugar levels giving Ioo units of crystallineinsulin subcutaneously if the blood sugar exceeds300 mgm. per Ioo cc., with an additional 200 unitsif it is between 600 and I,ooo mgm. and a further300 units if it is over 1,ooo mgm. A proportion ofthe additional insulin is given intravenously insevere cases. At the third and sixth hours, with arising blood sugar, 50 to 200 units are given bythese authors according to the physician's judg-ment of progress. In stressing the need for highinsulin dosage in the first three hours, Joslin et al.(1942) reports a mortality of i per cent. in a seriesof 525 cases treated with an average of 83 unitsduring the first three hours, and a subsequentseries of I23 cases with a mortality of 1.6 per cent.treated with an average of 2I6 units in the firstthree hours. Not all these patients were inprofound coma.

Lawrence's (1950) recommended dose is not solarge, the initial injection in a very severe comabeing ioo units repeated four hourly with a fallingblood sugar, but increasing to 200 units with noblood sugar response. He states that he is notconvinced that larger, or more frequently repeateddoses, are more effective, and that the restoration ofcirculation by intravenous therapy, making theusual dose effective, is of more importance.Colwell (1950) stresses the need for higher initialinsulin dosage, prescribing 80 to 200 units in thepresence of severe dyspnoea and extreme lethargyand higher doses in cases with deep coma and fail-ing circulation. He agrees with Lawrence in stat-ing that if therapeutically effective doses are giveninitially it is unnecessary to repeat its administra-tion until several hours have elapsed. Severeacidosis, he states, will not respond to insulin forthree to five hours, even though it be given intra-venously. He goes so far as to recommend a six-hourly administration schedule if response, asshown by physical signs, is favourable, and afour-hourly review and repeat of initial dose ifsymptoms fail to improve.My own initial dose in coma cases with blood

sugars of over 500 mgm. per Ioo cc. is o00 to I50units of soluble insulin subcutaneously and 50units intravenously. The patient is reviewed atthe end of three hours in the light of improvingcirculation, diminishing ketosis, returning con-sciousness and, less important, falling blood sugar.

In the absence of response the insulin dose isrepeated, and in severe cases may be doubled.Thereafter a three-hourly review is made and theoriginal subcutaneous dose repeated or reducedaccording to the response to treatment. With thereturn of consciousness and the ability to retainfluids by mouth, the four-hourly regime outlinedunder pre-coma is commenced.

It is important to appreciate that a patient indiabetic coma has suffered from profound circu-latory failure and that complete bed-rest isindicated for at least a week after recovery (Law-rence, I950; Cooke, I948).

(b) Correction of dehydration and chloride loss;saline and glucose administration. In recentyears the question as to whether glucose or salineshould be the initial parenteral fluid in coma caseshas been the subject of much discussion. Hims-worth (i931, 1932) laid particular stress on theneed for the early administration of glucose in thetreatment of pre-coma with vomiting and coma,advocating the use of a Io per cent. solution. Ex-perimental evidence in favour of early intravenousglucose has been advanc'ed by Soskin et al. (I937),who showed that depancreatized, hepatectomizedanimals can metabolize large amounts of carbo-hydrate with glucose oxidation and abolition ofketosis provided blood sugars remain at highlevels. It was shown (Mirsky et al., 1941) thatketosis was diminished or abolished in de-pancreatized animals, and in humans withdiabetic acidosis, when sufficient carbohydrate wasgiven. These observations suggested. that freecarbohydrate administration was beneficial indiabetic coma.

Root (i945), the leading opponent of the earlyuse of glucose, objects to its use on the groundsthat (a) no more than 5 to 0o g. of carbohydrateneed be oxidized per hour in or4er to check ketoneformation, (b) excessive hyperglycaemia is harmfulto the pancreas, and (c) excessive glucose con-centration damages the liver (Astwood et al., 1942).Franks et al. (I947) in investigating two groups ofketotic patients treated with early intravenousglucose or saline, found that while they showed nodifference in the rate of disappearance ofketoneuria, comparable doses of glucose were re-tained more efficiently, using a smaller insulin dose,with the lower blood sugar levels of the salinegroup. Further, they state, that owing to thehyperglycaemia, glycosuria and polyuria of theglucose series, the urinary output was nearly fourtimes that of the saline group, resulting in a de-layed restoration of the failing circulation and ofcellular dehydration. A lower fatality rate in casestreated with saline only is reported in this paper,and it is suggested that glucose should be withheld

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from

342 POSTGRADUATE MEDICAL JOURNAI, July 1951

for at least four hours from the beginning oftreatment.As will be discussed in the next section,

potassium and phosphorus diuresis is known tooccur during ketosis. It is possible that the con-tinued diuresis produced by early glucose ad-ministration may increase this loss (Storey, I950;Butler et al., I947).That glucose should be administered after the

initial period of treatment is not disputed evenby those who strongly deprecate its use during theearly hours. Joslin (I945) has stated that patientsin diabetic coma should receive not less than o00g. of glucose during the first 24 hours. Root(1945) gives carbohydrate four to. six hours afterthe beginning of treatment. Opinion, though re-maining divided on the danger of the use ofglucose in the early stage of diabetic coma, appears,in recent years, to have swung in favour of itsavoidance during the first hours of coma both inAmerica and in this country (Lawrence, quoted byOakley, 1949; Dunlop and Donald, 1948; Dunlopet al., I950; Brit. med. J., Editorial, I948).

It is my practice to administer 2 to 3 1. of normalsaline during the first three hours, thereafter con-tinuing with a 4.3 per cent. solution of glucose innormal saline or a 5 per cent. solution of glucoseaccording to whether or not the chloride deficiencyhas been remedied, as shown by the appearance ofchloride in the urine. Large quantities of intra-venous fluid may be necessary during the first 24hours of treatment to combat chloride and fluidloss in patients who may have a deficit of 26 g.of sodium chloride and io per cent. of the bodyweight of water (Butler et al., I947).

(c) Consideration of other electrolytes; potassiumand phosphorus. During the development ofdiabetic ketosis, polyuria, often accompanied byvomiting, results in the well recognized diminutionof extracellular fluid volume, with lowering of theserum sodium and chloride. Recently attentionhas been drawn to the large loss of potassium andphosphorus which also occurs. It has beenestimated that in addition to 26 g. of sodiumchloride and Io per cent. of body weight of water,the average case of coma also loses 5 g. of phos-phorus (Butler et al., I947) and 3 to 6 g. ofpotassium (Danowski, I949). In the initial stagesof diabetic coma, however, the cellular deficit ofpotassium is frequently marked by a rise in theserum potassium. This condition is broughtabout when the increased renal excretion ofpotassium fails to keep pace with the increasedrate of release of the ion from the body cells andwith contraction of extracellular volume. Withthe administration of insulin and parenteral fluidin the treatment of coma, entry of potassium intothe cells occurs associated with improved glucose

utilization. Potassium excretion decreases, extra-cellular volume expands and the rate of potassiumliberation from the cells diminishes (Seldin et al.,I949). The result is a considerable fall in serumpotassium wii bin four to 24 hours of the beginningof treatment.

Aitken et al. (I937) demonstrated in a case offamilial periodic paralysis, where the serumpotassium level is low during paralytic attacks,that the serum potassium can be lowered to pro-duce symptoms by the administration of largeamounts of glucose by mouth, by the injection ofinsulin and especially by the combined injectionof insulin and ingestion of glucose. Symptoms ofpotassium deficiency were produced when theserum potassium fell to below 12 mg. per Ioo cc.,the normal being i6 to 22 mg. Further studies ofthese cases (Allott and McArdle, 1938) showedthat changes in phosphorus metabolism ranapproximately parallel to those of potassium.

Holler (I946) was the first to describe a case ofdiabetic coma in which paralysis of the musclesof respiration ensued after treatment with largedoses of insulin and parenteral 5 per cent. glucosein normal saline. In this case the serum potassiumfell to 9.8 mg. per ioo cc. and recovery occurred,with a rising serum potassium, following parenteraland oral potassium administration. Frenkel et al.(I947) report a case similarly treated who de-veloped circulatory collapse and respiratory failurewith a serum potassium of 6.9 mg. per ioo cc.Further cases have been described by Martin et al.(I947), Nicholson et al. (I947) and Tuynman andWilhelm (I948).

Butler (I947) demonstrated a loss of serumphosphorus and potassium during investigationson a case of developing ketosis in a diabetic. Hestressed the accentuation of this los; by increasingthe metabolism of glucose and deposition ofglycogen du.ring treatment with insulin and glu-cose. Franks (1948) reports an initial rise inplasma inorganic phosphorus and in the excretionof phosphorus followed by a rapid fall with treat-ment in diabetic coma. The change in phos-phorus serum levels ran parallel with those ofpotassium. He states that low phosphorus as wellas low potassium serum levels may be responsiblefor circulatory collapse occurring after theabolition of ketosis, restoration of normal bloodsugar, water and sodium chloride levels in diabeticcoma.The cycle of events taking place during the

onset and treatment of diabetic ketosis may besummed up as follows:

(a) Loss of potassium by cells caused by vomit-ing, diuresis and dehydration. This may beaccompanied by a high serum potassium.

(b) Insulin, glucose and saline therapy causing

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from

July 1951 YOUNG: Ketosis and Coma in Diabetes Mellitus 343

a shift of potassium from extracellular to intra-cellular spaces with a resulting low serumpotassium level.The movement of phosphorus in and out of the

cells follows closely that of potassium.Studies of electrocardiographic changes in re-

lation to serum potassium levels have confirmedthe occurrence of low values in diabetic acidosis.Falling potassium levels in man are associated witha progressive lowering and broadening of the Twave and lengthening of the Q-T interval. Thismay proceed to a further lowering of the T waveand depression of the S-T segment leading toheart block and failure in systole (Stewart et al.,1940; Stoll and Nisnewitz, I94i). Raised levelsof serum potassium produce high, peaked T waves,increase in the duration of the Q.R.S. complex andheart failure in diastole (Winkler et al., 1938).Serious disturbances may occur at concentrationsabove 7.o m.eq. per 1. (28 mg. per Ioo ml.)(Tarail, 1948). In a study of 44 patients withdiabetic acidosis Nadler et al. (1948) correlatedelectrocardiographic changes with serum potassiumlevels. This paper describes the high or normalserum potassium E.C.G. curve before therapywith reversal to the low potassium type aftertreatment with insulin and glucose. It is suggestedthat electrocardiographic studies are useful infollowing serum-potassium levels during thetreatment of acidosis.The natural development of these biochemical

and electrocardiographic findings has been theadministration of potassium in cases of diabeticcoma. Mention has been made above to casessuccessfully treated in this way. Emphasis should,however, be placed on the dangers of potassiumintoxication which may follow its administration,especially by the parenteral route. The control ofintravenous potassium therapy should be exercisedby serial determinations of serum levels, and forquick results a flame photometer is necessary.Few hospitals in this country are so equipped, andelectrocardiographic control may be a suitablealternative. Elkington and Tarail (1950), in anexcellent survey of the present status of potassiumtherapy, give the following contraindications topotassium administration:

(a) Patients with renal insufficiency who mayexcrete potassium poorly.

(b) Patients with oliguria or anuria.(c) Patients with raised serum potassium, as

may occur in diabetic coma before treatment.They stress that the danger of administering

potassium to correct or prevent potassium loss isincreased if the serum potassium level is notpreviously known. Joslin (1949) suggests thatpotassium administration should be limited toclinics where quick tests for potassium are avail-

able. In my view this is wise in the case ofparenteral administration.

Potassium salts may be administered orally,subcutaneously and intravenously. However dis-guised they are poorly tolerated when taken orallyby patients with anorexia or vomiting. They havebeen given by stomach tube, to patients in diabeticcoma, as potassium chloride in the followingdoses: 2 g. at short intervals (Frenkel, I947);o.6 g. doses to 3.6 g. total (Nicholson et al., I947);5 g. at half-hourly intervals for six doses (Tuyn-man et al., 1948); 4 g. left in the stomach afterlavage (Lee et al., I949). The phosphate salts area more logical form of potassium administrationsince deficiency of extracellular phosphorus mayco-exist with potassium in diabetic ketosis. Allott(personal communication) recommends a solutionof potassium-dihydrogen phosphate (KH2PO4),suitably flavoured, as being the most palatable inhis experience in treating cases of familial periodicparalysis. 2 oz. of a solution containing 35 g.KH2P04 to io oz. (with saccharin sol. gr. i,glycerin 3 iv, sol. essence of lemon 3 i added)will contain 2 g. potassium. It is suggested thatfrom the second or third hours of diabetic coma,provided a reasonable output of urine has beenestablished, 2 oz. of this solution containing 2 g.potassium can be administered by mouth two tothree hourly with little risk of causing potassiumintoxication (see Lancet, Editorial, I951). As therange of cellular potassium deficit in diabetic comahas been calculated to be, on an average, 228 to936 m.eq., or about 8 to 36 g. of potassium in a60 kg. adult (Elkington and Tarail, I950), smallerdoses than the above, as have been suggested byother authors, would appear to be ineffective.

In the administration of parenteral potassiumsalts Elkington and Tarail (I950) recommend the-avoidance of solutions with a concentration ofpotassium over 70 to 80 m.eq. (3 g.) per 1., and arate of infusion of above 20 m.eq. (0.89 g.)potassium per hour. A solution containingK2HPO4 4.5 g., KH2PO4 I.0 g. and NaCl 5.5g. to a litre, contains 60 m.eq. (2.4 g.) of potassiumand should be given at a rate not exceeding 300 ml.per hour.Hawkins et al. (I95i) in treating a case of hypo-

kalaemia associated with pyloric stenosis used anisotonic solution of potassium chloride, withhyaluronidase, subcutaneously to a total of 5.5 g.of the chloride in four hours. When this dose wassubsequently increased to i6.o g. given over eighthours signs of potassium intoxication developed,with a serum potassium level of 35 mgm. per o00ml.

Caution is therefore necessary in the use ofpotassium as a therapeutic agent. As it is probablethat the fatalities which occur in diabetic coma,

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from

344 POSTGRADUATE MEDICAL JOURNAL July 1951

in spite of correction of ketosis, are due to serumand cellular potassium and phosphorus depletion,their use in the treatment of this condition isjustified. If the precautions outlined above are

observed it is probable that this addition to thetherapy of diabetic coma may result in a reductionof the mortality of this still very fatal complicationof diabetes mellitus.

BIBLIOGRAPHY

AITKEN, R. S., ALLOT, E. N., CASTLEDEAN, L. I. M., andWALKER, M. (1937), Clinical Science, 3, 47.

ALLOTT, E. N., and McARDLE, B. (1938), Ibid., 3, 229.ASTWOOD, E. B., FLYNN, J. M., and KREYER, O. (1942),J. Clin. Investigation, 21, 621.BARNES, R. H., and WICK, A. N. (1939), 7. Biol. Chem., 131, 413.BARNES, R. H., DRURY, D. R., GREELEY, P. O., and WICK,A. N. (1940), Amer. J. Physiol., 130, 144.BEARDSWOOD, T. J. (I935), J. Amer. med. Ass., 105, II68.Brit. med. J. (I948), Editorial, 2, 209.BUTLER, A. M., TALBOT, N. B., BURNETT, C. H., STAN-

BURY, J. B., and MAcLACHLAN, E. A. (I947), Tr. A. Amer.Physicians, 60, I02.

COLLEN, M. F. (I942a), Arch. intern. Aled., 70, 347.COLLEN, M. F. (1942b), Ibid., 70, 369.COLWELL, A. R. (I950), 'Types of Diabetes and their Treat-

ment,' Illinois, U.S.A.COOKE, A. (1948), Brit. med. J., ii, 799.DANOWSKI, T. S., PETERS, J. H., RATHBUN, J. C., QUASH-

NOCK, J. M., and GREENMAN, L. (I949), J. Clin. In-vestigation, 28, I.

DRURY, D. R. (1942), J. Clin. Endocrinology, 2, 421.DUNLOP, D. M., and DONALD, J. B. (1948), Brit. med. J., ii,

352.DUNLOP, D. M., DAVIDSON, L. S. P., and McNEE, J. W.

(1950), Textbook of Medical Treatment,' Edinburgh.ELKINGTON, J. R., and TARAIL, R. (I950), Amer. J. Med., 9,

200.

FRANKS, M., BERRIS, R. F., KAPLAN, N. O., and MYERS,G. B. (I947), Arch. intern. Med., 80, 739.FRANKS, M., BERRIS, R. F., KAPLAN, N. O., and MYERS,

G. B. (1948), Ibid., 81, 42.FRENKEL, M., GROEN, J., and WILLEBRANDS, A. F. (I947),Ibid., 80, 728.GEELMUYDEN, H. D. (1904), Ztschr. f. physiol. Chem., 41, 128.HAWKINS, T. F., HARDY, T. L., and SAMPSON, H. H. (1951),Lancet, i, 318.HIMSWORTH, H. P. (1931), Ibid., ii, 978.HIMSWORTH, H. P. (1932), Ibid., ii, 166.HOLLER, J. W. (1946), J. Amer. med. Ass., 131, I186.JOSLIN, E. P., ROOT, H. F., WHITE, P., and MARBLE, A.

(1942), 7. Amer. med. Ass., 19, II6o.

JOSLIN, E. P. (1945), New. Engl. J. lMed., 232, 219.JOSLIN, E. P. (1946), 'Treatment of Diabetes Mellitus,' 8th

edition.JOSLIN, E. P. (1948), Brit. med. J., ii, 651.JOSLIN, E. P. (1949), J. Amer. med. Ass., I39, I.Lancet (I95I), Editorial, i, 393.LAWRENCE, R. D. (1942), Proc. R. Soc. Med., 31, I.LAWRENCE, R. D. (I950), ' The Diabetic Life,' London.LAWRENCE, R. D., and OAKLEY, W. (1948), Brit. med. J., ii,

310.LEE, J., NAIDOO, D., and TORRENS, J. A. (1949), Ibid., i, 565.MACKAY, E. M. (1943), J. Clin. Endocrinology, 3, Iox.MARTIN, H. E., and WERTMAN, M. (1947), J. Clin. Investiga-

tion, 26, 217.MICKS, R. H. (1948), Brit. med. J., ii, 200.MIRSKY, I. A., FRANZBLAU, A. N., NELSON, N., and

NELSON, W. E. (1941), J. Clin. Endocrinology, 1, 307.NADLER, C. S., BELLET, M. D., and LANNING, M. (1948),

Amer. J. Med., 5, 838.NICHOLSON, W. M., and BRANNING, W. S. (1947), .7 Amer.

med. Ass., 134, 1272.OAKLEY, W. (1949), Brit. med. J., i, 724.RABINOWITCH, I. M., FOWLER, A. F., and BENSLEY,

E. H. (1939), Ann. Int. Med., 12, 1403.ROOT, H. F. (1945), .. Amer. med. Ass., 127, 557.SCHAFFER, P. A. (1923), Medicine, 2, 375.SELDIN, D. W., and TARAIL, R. (1949), J. Clin. Investigation

(Proc.), 28, 8 o.SOSKIN, S., and LEVINE, R. (1937), Amer. J. Physiol., 120, 761.STEWART, H. J., SMITH, J. J., and MJLHORAT, A. T. (1940),Amer. . led. Sci., 199, 789.STOLL, B., and NISNEWITZ, S. (1941), Arch. intern. Med., 67,

755.STORY, R. D., and ROOT, H. F: (1950), J. Amer. med. Ass., 144,

86.TARAIL, R. (1948), Amer. J. Med., 5, 828.TUYNMAN, P. E., and WILHELM, S. E. (1948), Ann. Int. ,led.

29, 356.W INKLER, A. V., HOFF, H. E., ard SMITH, P. K. (1938Amer. J. Physiol., 124, 478.YOUNG, F. G. (1944), St. Thomas' Hospital Gaz., 42, 64.

RUTHIN CASTLE,NORTH WALESA Clinic for the diagnosis and treatment of Internal Diseases (except Mental or Infectious Diseases). The

Clinic is provided with a staff of doctors, technicians and nurses.The surroundings are beautiful. The climate is mild. There is central heating throughout. The annual

rainfall is 30.5 inches, that is, less than the average for England.The Fees are inclusive and vary according to the room occupied.

For particulars apply to THE SECRETARY, Ruthin Castle, North Wales.Telegrams: Castle, Ruthin. Telephone: Ruthin 66.

copyright. on June 13, 2020 by guest. P

rotected byhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.27.309.338 on 1 July 1951. Dow

nloaded from