871

administering methionine or ammonium chloride. Reduc-tion of dietary protein and the administration of neomycinusually improved an. abnormal electroencephalogram.The electroencephalographic abnormality was related

to liver-cell function, the portal collateral circulation, andthe amount of nitrogen in the intestine.The electroencephalogram showed an ill-defined

correlation with the blood or the c.s.F. ammonium level.The mechanism and specificity of the electroencephalo-

graphic changes and their value in the diagnosis of neuro-psychiatric changes in patients with liver disease and inthe assessment of patients for portacaval anastomosisare discussed.

We wish to thank Miss J. M. Atwell and Miss A. M. Leesfor technical assistance, and the Dan Mason Research Founda-tion of the West London Hospital Medical School and theMedical Research Council for financial grants. Messrs. E. R.

Squibb and Sons gave generous gifts of neomycin and financialsupport.

REFERENCES

Adams, R. D., Foley, J. M. (1953) Proc. Ass. Res. nerv. Dis. 32, 198.Atkinson, M., Barnett, E., Sherlock, S., Steiner, R. E. (1955)

Quart. J. Med. 24, 77.Bickford, R. G., Butt, H. R. (1955) J. clin. Invest. 34, 790.Conway, E. J. (1950) Microdiffusion Analysis and Volumetric Error.

3rd ed., London.Davidson, E. A., Summerskill, W. H. J. (1956) Postgrad. med. J.

32, 487.Fazekas, J. F., Ticktin, H. E., Ehrmantraut, W. R., Alman, R. W.

(1956) Amer. J. Med. 21, 843.Foley, J. M., Watson, C. W., Adams, R. W. (1950) Trans. Amer.

neurol. Ass. 75, 161.Friedlander, W. J. (1956) Electroenceph. clin. Neurophysiol. 8, 513.McDermott, W. V. Jr., Adams, R. D. (1954) J. clin. Invest. 33, 1.Martini, G. A., Phear, E. A., Ruebner, B., Sherlock, S. (1957)

Clin. Sci. 16, 35.Phear, E. A., Ruebner, B., Sherlock, S., Summerskill, W. H. J.

(1956) ibid. 15, 93.— Sherlock, S., Summerskill, W. H. J. (1955) Lancet, i, 836.

Phillips, G. B., Schwartz, R., Gabuzda, G. J. Jr., Davidson, C. S.(1952) New Engl. J. Med. 247, 239.

Sherlock, S., Summerskill, W. H. J., White, L. P., Phear, E. A.(1954) Lancet, ii, 453.

Summerskill, W. H. J., Davidson, E. A., Sherlock, S., Steiner, R. E.(1956) Quart. J. Med. 25, 245.

Traeger, H. S., Gabuzda, G. J. Jr., Ballou, A. N., Davidson, C. S.(1954) Metabolism, 3, 99.

Van Caulaert, C., Deviller, C., Halff (1932) C.R. Soc. Biol. Paris,111, 739.

White, L. P., Phear, E. A., Summerskill, W. H. J., Sherlock, S.(1955) J. clin. Invest. 34, 158.

OUTPATIENT P.A.S. THERAPY

W. M. DIXONM.B. Lond., M.R.C.P.

ASSISTANT CHEST PHYSICIAN

PETER STRADLINGM.D. Lond., M.R.C.P.

CHEST PHYSICIAN AND LECTURER IN TUBERCULOSIS

I. D. P. WOOTTONM.B., Ph.D. Cantab., F.R.I.C.

LECTURER IN CHEMICAL PATHOLOGY

THE HAMMERSMITH CHEST CLINIC AND POSTGRADUATE

MEDICAL SCHOOL OF LONDON

Matilda told such Dreadful Lies,It made one Gasp and Stretch one’s Eyes;Her Aunt, who, from her Earliest Youth,Had kept a Strict Regard for Truth,Attempted to Believe Matilda:The effort very nearly killed her.

Hilaire Belloc: Cautionary Tales.

THE use of dual chemotherapy regimes to avoid bac-terial resistance in patients with tuberculosis is now anaccepted principle ; and chest physicians often prescribea combination of isoniazid and p-aminosalicylic acid(P.A.S.) as long-term ambulant chemotherapy for manypatients.To prevent the emergence of resistant strains of Myco-

bacterium tuberculosis, P.A.S. must unfortunately be takenin doses near to the limit of tolerance, and there is thusareal risk that patients may not take enough of the drug.Subjective assessment of the patients’ reliability in this

respect is open to grave error, so that an objectiveestimation of P.A.S. consumption is of considerable value.Since the drug is eliminated via the kidneys (Venkatara-man et al. 1948, Way et al. 1948), indirect evidence ofconsumption can be obtained by chemical tests forP.A.S. in the urine. Occasional use of such a test in ourroutine work suggested that many apparently reliablepatients were untrustworthy, and this supported a previ-ous finding in 100 Scottish outpatients (Simpson 1956).Since no previous trial had been carried out to determinehow reliable a complete population of outpatients isin taking P.A.S., we decided to make such an objectiveassessment of our patients.

Method

Three tests have been used to detect p.A.s. or itsderivatives in urine, depending on colour produced withEhrlich’s reagent, with hypochlorite, or with ferricchloride (Penman and Wraith 1956, Simpson 1956, Ruiz1957). An intense violet colour is produced when urinecontaining P.A.s. is treated with ferric-chloride solution,but this test is difficult to read when it is done in theusual way on undiluted urine. The violet tint is obscuredby the brown colours of the urine and the reagent, andprecipitation of ferric salts usually causes turbidity. Thedifficulty can be overcome if the reaction is carried out inhigh dilutions by mixing water (5 ml.), urine (6 dropsor 0-1 ml.), and 3% ferric-chloride solution (1 drop).With these quantities, urine containing 100 mg. of r.A.s.per 100 ml. produces a distinct violet colour, whichbecomes deeper as p.A.s. content increases. In a negativetest the solution is colourless and only slightly opalescent.

This modified ferric-chloride test has the further

advantage that false-positive reactions are greatlyreduced. False reactions most commonly follow the

ingestion of salicylates other than p.A.s., usuallyaspirin. Tests were done on urine specimens from 8patients with rheumatoid arthritis who were receivingup to 100 gr. of aspirin daily. With the test as described,negative results were obtained in all cases, althoughviolet colours were readily produced if the amount offerric-chloride solution was increased. It follows thataspirin preparations taken by the patients are unlikelyto invalidate the test. A further small trial on 6 inpatientsshowed that the reaction remained positive for up totwelve hours after the ingestion of 5 g. of P.A.S. butbecame negative thereafter. Urine stored for up to

twenty-four hours after voiding remained positive to thetest.The trial took place during two weeks in the spring of

1957. If a patient attended the clinic during that periodthe urine was tested at this visit. Otherwise all specimenswere collected at the patients’ homes by the healthvisitors, who called without prior notice. A great dealof this work was done in the evenings when patientswho were working came home ; the collections weremade on different evenings, but always on days whenpatients should have been taking the drugs. The speci-mens collected were tested on return to the clinic. A

patient whose test was negative at the time of a clinicvisit did not have a further specimen collected at home.

Results

Of 151 patients (59 women and 92 men) only 76 (50%)gave a positive reaction for P.A.S. in their urine. Youngwomen had the lowest proportion of positive reactions(see table). From the 99 patients who were working, 57urines (57%) gave a positive result, whereas only 19

(36%) from the 52 not working were positive. The

working men, with 66% of positive urine tests, yieldeda higher figure than the working women (42%).

Duration of chemotherapy made little difference tothe results. About half the urines gave negative results,regardless of the time the patients had been receivingP.A.S.

872

RESULTS OF URINE TESTS FOR P.A.S. IN 151 PATIENTS, ACCORDINGTO SEX AND AGE

Several preparations, including calcium benzoyl P.A.S.,are in use at the clinic, and an attempt is made in everycase to find a preparation which will cause least upsetto an individual patient. We found, however, that thepreparation used had little or no effect on the results.

Before the results of the urine tests were known, thehealth visitors were asked to judge whether their patientswere taking the prescribed P.A.S. or not. They were notprepared to give an opinion in 48 cases. In the remainder,their opinions disagreed with the urine tests in 14 of the44 women and in 15 of the 59 men. In all, their opinionsproved unreliable in over a third of the cases.

Discussion

The success of P.A.S. therapy has always been threat-ened by its troublesome toxic effects. It has been esti-mated that over 30% of the patients taking the drughave nausea (Medical Research Council 1950), andanorexia is even more common. 33% of patients in theMedical Research Council trial had diarrhcea, but for-tunately the more alarming hypersensitivity reactionsare less common. Various attempts to reduce the gastro-intestinal irritation and to improve or disguise the

unpleasant taste of P.A.s. have met with little success.It is not surprising, therefore, that we have found thathalf our patients, while professing to take P.A.s. in one ofits various forms, were not in fact taking the drug.

This trial might be criticised because the results arebased on a single urine test for each patient : it couldbe assumed that the test had been taken on the oneoccasion when each patient had defaulted. This criticism

might have been sound had the number of negative testsbeen small, but our results are so clear-cut and thedefault-rate so high that we feel this argument is invalid.It is reasonable to assume that if a patient is unreliableonce he is likely to be so again, and cannot be trusted tofollow instructions meticulously. This is particularly soif the patients with negative tests profess to be regularconsumers. Furthermore, it was felt that the patientswould very soon guess the purpose of the test if we

persisted in our investigations. We are satisfied that thedefault rate is very high, and further discussion is basedon this premise. It seems probable that half the patientsdid not take p.A.s., even in the early days of their treat-ment, and have successfully deceived doctors and nursesfor many months or years.The significance of this fact should be accepted : it is

easy to blind oneself to the obvious and assume that,while a patient progresses well, it is our prescriptions thatare achieving this result. Many patients can heal theirdisease unaided, and often did so before the days ofchemotherapy.On the other hand, there are unexpected relapses, and

if the value of long-term chemotherapy is accepted thedrugs must be not only prescribed but taken : too oftenan unexpected bacterial resistance, revealed during a

relapse, has indicated that P.A.S. was not taken during aprevious course of dual chemotherapy. To replaceindividual drugs by single cachets or powders containing

E P.A.s. and isoniazid, as we have done for some years, isonly a step forward. Although resistance will probablynot develop if the drugs a,re taken, a new danger emergesof taking no chemotherapy, and is well illustrated in ourpatients, many of whom were supposed to be takingPycainisan ’ or ’ ‘ B-I’asinah.’The time has come to reassess P.A.S. as an outpatient

remedy, and we feel that its routine use should beabandoned in outpatients. The risks are too great, and aweekly urine test would rapidly become useless becausethe patient would learn its significance and take thenecessary single dose to produce a positive result. Wedo not think that the use of any physical or chemicalcombination of P.A.s. and isoniazid containing less than12 g. of P.A.S. is likely to prove a suitable alternative,and reasons for this have already been set out by two ofus (Dixon and Stradling 1956). In our experience, dailystreptomycin and isoniazid can be given for very muchlonger than is usually done, and it is now our aim toreplace combinations containing P.A.s. by this regimewherever possible. The nurse giving the injection canthen see that the isoniazid is consumed. For specialcases kept on P.A.8. combinations as outpatients, the

single urine-test here described is available for occasionaluse. It is simply performed, more specific than previouslyreported single tests, and easily read by nurses.Many chemotherapy trials based on unsupervised oral

medication have probably been built on very unsure

foundations. Ilowever carefully they were controlled

statistically and scientifically, the results may have beenvitiated by inadequate consumption of prescribed drugs.The gastro-intestinal side-effects of P.A.s. make this

drug particularly difficult to take, but they may not bethe only reason for our findings. Many patients feel welland see no need for medication, and others, when ques-tioned, will admit to forgetfulness. The low cost to a

patient in this country, suggesting that the drugs are oflow intrinsic value, may also be another factor. But infact one brand of commonly used cachet containing P.A.s,and isoniazid costs the Exchequer about ls. per day perpatient when the usual dose is given. The country isspending substantial sums on unconsumed P.A.S. (andprobably on many other drugs). That there is a further

large field for inquiry is suggested by the many tabletsthat can be found scattered in the flowerbeds surroundingsome of our hospitals. If the patients cannot take theirdrugs, then at least the physicians should be aware ofthis and save them the trouble of throwing them away,

Summary and ConclusionsWith a new and simple modification of the ferric-

chloride test for the detection of urinary p-aminosalicylicacid, false positives are few and a second test is unneces- ;sary.

All outpatients of one chest clinic who were receivingP.A.S. had specimens of urine tested without notice, andit was found that only half the patients had P.A.S. intheir urine.

It is suggested that P.A.S. is not a suitable drug forlong-term treatment of outpatients.Our thanks are due to Mrs. M. Greenwood, Mrs. C. Milborne,

Miss K. Reade, and Mrs. K. Royle, health visitors to the clinic.who undertook the major part of the practical field work inthis trial. We also wish to thank Dr. J. D. Pryce who carriedout some of the initial tests on the modified ferric-chloridereaction.

REFERENCES

Dixon, W. M., Stradling, P. (1956) Lancet, ii, 892.Medical Research Council (1950) Brit. med. J. ii, 1073.Penman, H. G., Wraith, D. G. (1956) Lancet, ii, 552.Ruiz, It. C. (1957) Tubercle, Lond. 38, 152.Simpson, J. McD. (1956) ibid. 37, 333.Venkataraman, A., Venkataraman, P. R., Lewis, H. B. (1948)

J. biol. Chem. 173, 641.Way, E. L., Smith, P. K., Howie, D. L., Weiss, R., Swanson, R.

(1948) J. Pharmacol. 93, 368.