Ann Rheum Dis: first published as 10.1136/ard.26.5.373 on 1 September 1967. Downloaded from Ann. rheum. Dis. (1967), 26, 373 ASSESSMENT OF DRUGS IN OUT-PATIENTS WITH RHEUMATOID ARTHRITIS EVALUATION OF METHODS AND A COMPARISON OF MEFENAMIC AND FLUFENAMIC ACIDS WITH PHENYLBUTAZONE AND ASPIRIN BY R. M. MASON, D. E. BARNARDO*, W. R. FOXt, AND M. WEATHERALL+ From the Department ofPhysical Medicine and Rheumatology, the London Hospital, and the Department ofPharmacology, the London Hospital Medical College Relief of symptoms in patients with chronic The methods of clinical trials are still capable of rheumatoid arthritis is difficult. Conventional much improvement. Particularly in trials on out- analgesics such as aspirin, paracetamol, and phenyl- patients, the opportunities for undetected errors are butazone have limited efficacy and are not free from very great. Problems of measurement are also toxic effects. Any alternative which was more considerable, at least in rheumatoid arthritis, in effective, safer, or both would be welcome, but most which all of the many measurements which can be copyright. new remedies owe such success as they achieve to made appear to vary somewhat independently of therapeutic optimism which is a powerful but each other (American Rheumatism Association, transient potentiator of pharmacological effects. 1965), so that confidence in any one is limited. The accurate assessment of new drugs by properly One cannot decide which is the best of a number of controlled trials is indispensable. Two drugs which alternative procedures without testing all of them in have shown promise in laboratory studies and early parallel. The present trial has been extended clinical trials are mefenamic and flufenamic acids considerably beyond the needs of a simple compari- (Winder, Wax, Scotti, Scherrer, Jones, and Short, son of drugs in order also to be informative about 1962; Goodley, 1963; Young, 1962), derivatives of problems of method. xylylanthranilic acid with actions broadly resembling http://ard.bmj.com/ those of aspirin. We have submitted these drugs Methods to a controlled trial, which has already been briefly (1) Patients reported (Barnardo, Currey, Mason, Fox, and All were females over 18 years of age and were selected Weatherall, 1966). Since our trial began, several by the six participating physicians from the attenders other trials have been reported. Flufenamic acid at the Department of Physical Medicine and Rheuma- (600 mg./day) has been found to be less effective tology of the London Hospital between October, 1964, than prednisone (Fearnley and Masheter, 1966), and and September, 1965. Each had "definite" or "classical" on October 2, 2021 by guest. Protected comparable in effect to aspirin (2,700 mg./day) over rheumatoid arthritis (Ropes, Bennett, Cobb, Jacox, and to 6 months and Jessar, 1959) for more than one year, was not pregnant periods up (Simpson, Simpson, nor expected to become so during the trial, and had no Masheter, 1966) and also comparable with phenyl- history of cardiac failure, hepatic disease, or proven butazone (300 mg./day) over shorter periods peptic ulceration. Patients sensitive, or reacting un- (Rajan, Hill, Barr, and Whitwell, 1967). Mefenamic favourably to, either phenylbutazone or aspirin were acid (1,500 mg./day) has been found effective also in admitted and a provision was made as described below to osteo-arthritis (Cahill, Hill, Jessop, and Hume avoid their receiving the drug to which they were sensitive. Kendall, 1965). Patients receiving other well-stabilized treatment such as steroids or antimalarial drugs were not excluded and their treatment with these drugs was continued while they took * Present address: Gastro-intestinal Research Unit, Mayo Clinic, Rochester, Minn. part in the trial. The purpose ofthe trial was explained to t Present address: Research Dzpt., Potato Marketing Board, each patient and anyone who did not wish to take part or London, S.W.1. who would have had practical difficulties in co-operating + Present address: Wellcome Research Laboratories, Beckenham, Kent. was excluded. 373 Ann Rheum Dis: first published as 10.1136/ard.26.5.373 on 1 September 1967. Downloaded from 374 ANNALS OF THE RHEUMATIC DISEASES (2) Design and Procedure replication was achieved by an appropriate instruction to the pharmacist and did not involve the physician in Patients differ considerably from one another. Varia- breaking the code. tion in the severity of any given patient's arthritis is The trial proper was preceded by a short pilot study in likely to be less than the variation between patients. which the trial procedure was tested and practised and The trial was therefore designed so that drugs were the physicians obtained experience in the completion of compared within patients. Each patient was to receive the forms designed for keeping records in the trial. both new drugs and an appropriate standard treatment. Seven patients were admitted to the pilot study, which The choice of a standard presented some difficulty. revealed no appreciable faults of planning. A further Many drugs have been shown to be more effective than 36 patients were admitted; four among these failed to a placebo in rheumatoid arthritis, so the use of a dummy complete three periods of treatment and a corresponding would have raised ethical problems. Besides, one object additional number were admitted to make a complete of the trial was to compare the new remedies with balanced set as designed. While the results from this standard treatments, of which aspirin and phenylbuta- set were being analysed, further patients were admitted zone seemed most appropriate. Some patients are, or with a view to completing a second complete set of 36. believe themselves to be, sensitive to one or other of these The results on the first 36 indicated that no substantial drugs. It is for such patients that a new alternative difference was appearing between treatments, and the analgesic is particularly needed. The trial therefore had second set was therefore not completed. In all 68 to be arranged so that such patients could be included, patients were admitted, including the seven in the pilot without risk of being exposed to the drug to which they study and eight who completed less than three periods. were perhaps sensitive. This objective was achieved Analyses of comparison between drugs are mostly based by designing the trial so that aspirin served as the standard on the balanced set of 36 patients. Other analyses were treatment for half the patients and phenylbutazone for based on all the patients studied. the other half. Ordinarily the choice depended on the statistical design, but if a patient was believed to react Many clinical trials are conducted with fixed dosage of unfavourably to either drug the physician requested drugs. The conclusions which can be drawn from single instead a sequence omitting the drug in question. This fixed-dose studies are very limited. If one drug produces request was made on four occasions, but the allotted better results than another, the difference may be due entirely to the doses chosen, and an otherwise identical copyright. sequence was, in fact, inappropriate and had to be trial with different doses might produce exactly the altered only once. opposite result. If each drug is administered at two or During three consecutive periods, each of 4 weeks' more levels of dosage, and if the larger dose has larger duration, all patients were supposed to receive mefenamic average effects than the smaller, a reasonable basis of acid, flufenamic acid, and either phenylbutazone or comparison is possible between the dose-response lines aspirin in capsules of a different colour. Three kinds for each drug. But it is seldom possible in practice to of coloured capsule were used (blue, blue and white, achieve a graded response, because the range between the and red and yellow) and each drug was dispensed equally apparently ineffective dose and the evidently toxic dose often in one kind of capsule as another. It was therefore is usually too small. evident to patients that they were receiving different To avoid this difficulty, doses in the present trial were http://ard.bmj.com/ remedies in successive periods, but communication selected not by an arbitrary weight of drug but by the between patients or staff about the efficacy of different- response of the patient to treatment. Initially the doses looking remedies (Asher, 1948) would not lead to prescribed were 720 mg. aspirin, 100 mg. phenylbutazone, collective judgements related to a particular drug. Each 500 mg. mefenamic acid, or 200 mg. flufenamic acid, of the twelve possible orders of treatment was allocated each given three times daily. In order to adjust the dose to three patients and the three capsule colours were to each individual patient, all patients were seen at the arranged in a Latin square design over each such group. end of the first week in each period; their progress was A complete replicate of the design thus required 36 briefly reviewed and the physician had the option of on October 2, 2021 by guest. Protected patients. When a physician admitted a patient to the decreasing or increasing the daily dose by one third (by trial, the pharmacist took note of any specific request to altering the frequency of medication to twice or four avoid a sequence containing either aspirin or phenyl- times daily) if it seemed desirable on grounds of thera- butazone and then gave the patient the next available peutic response and toxic manifestations. Provision suitable trial number. Thereafter the pharmacist was also made for discontinuation of a drug if worsening dispensed the appropriate capsules, leaving both the of symptoms or toxic effects were severe. In such a case patient and the physician unaware of the identity of the the patient proceeded immediately to the drug planned treatment. for the next period. Occasionally a short period without When patients failed to complete three periods, treatment was adopted before resuming the trial sequence.