Therapie Dr Teast oi Plasma Antiepileptic Drug Monitoring in a Neurological Depariment of Medicine, The University of Queensland, Royal Brisbane Hospital, Herston, Brisbane, Australia i ven Piss, Lid, Now York Practice: A 25-Year Experience M. J. Eadie ‘Summary: Rates for being free of all epileptic seizures for at least 1 year and for ceasing therapy after 3-5 years of full seizure control were similar in $48 treated epileptic patients referred from 1962 to 1970, during which period plasma antiepileptic drug concentration monitoring became available (32.7 and 17.0%, respectively), in 960 patients referred from 1971 to 1979, when such ‘monitoring was used increasingly (29.6 and 14.9%, respectively), and in 348 patients referred from 1980 to 1989, when this monitoring was widely available GO.S and 10.1%, respectively), However, monitoring was associated with higher rates of fll seizure control in patients seen within 6 months of their first seizure, even when patients with only a solitary seizure were excluded, Pos- sibly, monitoring then helped achieve early complete suppression of not yet established seizure processes, but had less effect on already entrenched sei- zute mechanisms. Plotting cumulative percentages of patients with fully con- trolled seizures versus plasma antiepileptic drug concentrations yielded data that were broadly in Keeping with conventional ‘therapeutic’ ranges for phe- nytoin, carbamazepine, phenobarbital, and ethosuximide monotherapy. With a ‘given plasma phenobarbital (but not phenytoin or carbamazepine) concentra- tion, rates of full seizure control were appreciably higher for generalized onset than for partial onset seizures. Use of a second antiepileptic drug was associ- ated with improved full seizure control in both generalized onset and partial ‘onset seizures, which had not been controlled by monotherapy; adding & third drug yielded litle further benefit. Key Words: Antiepileptic drug—Control— ‘Monitoring—Therapeutic range. Application of pharmacokinetic principles in con- Junction with plasma antiepileptic drug concentra- tion monitoring has yielded one of the major ad- vances in the management of epileptic seizures in recent years (1). Most medical practitioners who treat patients with seizures probably will consider that the plasma drug concentration measurements have helped in a variety of clinical situations in their practices. However, with few exceptions (2,3), rel- Received November 25, 193; accepted April 11, 196, ‘Address correspondence and reprint requests fo Dr. M. J Eadie at Department of Medicine, The University of Queens: land, Royal Brisbane Hospital, Herston, Brisbane, Australis, 4003, 458 atively little has been published about evaluating the extent to which plasma antiepileptic drug con- centration monitoring better controls epileptic sei- ‘ures in the treated population. It was, therefore, proposed to assess the use of this monitoring pro- cedure on the control of epileptic seizures in a per- sonal series of 1,902 consecutive patients who had had at Ieast one epileptic episode referred over the 30-year period that spanned the introduction of plasma antiepileptic drug concentration monitoring. Relationships between plasma antiepileptic drug concentrations and full seizure control were exam- ined for the major individual antiepileptic drugs and, in relationship to each drug, for generalized onset and partial onset seizures, since it has beenPLASMA ANTIEPILEPTIC DRUG MONITORING 459 suggested that the so-called ‘therapeutic’ range of drug concentrations is not the same for different seizure types (4). MATERIALS AND METHODS The study was based on a series of 1,902 consec- utive persons who had had at least one epileptic event and who had been referred to the author's private neurological consultant practice between 1962. and 1991. No patient seen in hospital-based neurological consulting was included. Members of the seties underwent no selection procedure prior toentry, apart from that exercised by their referring practitioners in deciding whether or not to seek a neurological opinion regarding an episode, ot epi- sodes, that appeared epileptic in nature. Follow-up referral of patients was at the instigation of their local medical practitioners, although opportunity for review at appropriate intervals was requested of these practitioners, The great majority of patients were already receiving antiepileptic drug therapy when first seen. At the first and subsequent consul- tations, drug doses were adjusted when indicated clinically to obtain complete seizure control without unacceptable adverse effects, the conventional lim- its of therapeutic range drug concentrations not constraining such prescription. Drugs being admin- istered were not altered unless inappropriate for the seizure type present. Phenytoin, carbamazepine, and agents yielding phenobarbital were considered appropriate forall partial onset seizures, convulsive seizures of generalized onset epilepsy, and, until the 1980s, juvenile myoclonic seizures, although af- ter this time, barbiturate anticonvulsants were pre- ferred (ethosuximide was considered appropriate for absence seizures of generalized onset epilepsy, although clonazepam and valproate were regarded as suitable if the patient was already receiving them; these latter agents were used in myoclonic and other generalized onset seizures that began be- fore adolescence). Clonazepam or valproate, if al- ready in use, was continued in all varieties of sei- zure disorder. Antiepileptic drugs were also changed if they were causing significant adverse ef- fects or had failed to achieve full seizure control at maximum tolerated dosage. Patients were asked to keep a written record of all seizures, although this was not always available when patients presented again. In all patients’ records, relevant data were noted systematically from the onset, with a view to their subsequent analysis. Characteristics of the members of this series have been described in greater detail elsewhere (5). Some 28.6% of its members were <12 years of age at the time of presentation, Generalized onset epileptic seizures accounted for 33.3% of the series, partial onset seizures for 53.9%, seizures of unclassifiable type for 10.6%, and uncommon seizure types for 2.2%. Only a solitary epileptic event had occurred when 13.9% of patients first presented, but 89.0% of these were already receiving treatment at this time. Because of this and the possible medico-legal con- sequences of withdrawing therapy at this stage, 89,8% of solitary seizure patients continued to re- ceive antiepileptic drug therapy. Plasma antiepileptic drug concentration measure- ments first became available for use in members of the series in 1968, when the Wallace (6) method for determination of whole blood (later plasma) phenyt- in levels was set up for research purposes. Within the next 4 years, plasma phenobarbital and carbam- azepine gas-liquid chromatographic assays were de- veloped and validated (again initially for research purposes), and by the mid 1970s, antiepileptic drug assay services were being provided by local private and institutional pathologists, most of whom used commercially available immune assay systems of various kinds and observed adequate quality con- trol procedures, Antiepileptic drug concentration data from the various assays became available for many patients in the series after this time, Once available, measurements were carried out at the time of consultation whenever they appeared likely to yield clinically useful information, were practica- ble, and were recommended at no longer than 6-month intervals if the patient was not to be seen at a shorter interval. Some patients were seen before assays were functioning for the drug or drugs that they were taking, while for a few patients, knowl- edge of the plasma antiepileptic drug concentration would not have altered clinical management since the patients were resistant to the idea of any change in therapy. No conscious decision was made to carry out measurements primarily in patients whose seizures were difficult to manage. It was recognized that knowledge of the plasma drug concentration at which a patient's seizures were controlled was at least as valuable as knowledge of the drug concen- tration when seizures were continuing. Sometimes insufficient clinical data were available to correlate with the plasma drug concentration values because, for various reasons, patients could not be followed for an adequate period, As a matter of practical policy, it was decided to take at least I year's con- Ther Drug Mont, Vo. 16, No.5, 1996460 M. J. EADIE tinuous freedom from all known epileptic events as the criterion of ‘full control’ of epileptic seizures. In all calculations considering the relationships be- tween plasma drug concentrations and seizure con- trol, the plasma concentration value used was the highest tolerated one that applied during, or as near as possible to, the earlier part of the period of such seizure control. Drug concentrations were measured in whole plasma, not plasma water. All measurements were under steady-state conditions but predose (trough) measurements were not necessarily required except for valproate. Other antiepileptic substances stud- ied had long enough elimination half lives for the timings of the measurements in relationship to drug intake not to be clinically important under steady- state conditions, Unlike the situation that appar- ently obtained in other countries, rapidly absorbed carbamazepine preparations were not available in Australia during the period of the study, although there is some evidence that the absorption profile of the Australian carbamazepine market leader changed around 1982 (7). Any plasma carbam- azepine concentrations measured during the period of likely temporary post-intake toxicity at steady state were not included in the data analyzed. Plasma phenobarbital concentrations were taken as adequate measures of the anticonvulsant effect of the parent substance and of methylphenobarbital and primidone (8), although the latter drug was not often prescribed The full data from the series were stored in a database running under the program DBase III Plas on an IBM compatible microcomputer. Particular data items were analyzed by means of this program, with further assessment by means of the statistical ‘component of the AXUM graphics package (9). Sta- ical significances of differences were assessed by means of the confidence interval analysis pro- ‘gram of Gardner and Altman (10). The study was carried out after approval from the University of Queensland Human and Animal Ex- perimentation Ethics Committee. RESULTS Outcome of Plasma Antiepileptic Drug Monitoring Patients were divided into those who first pre- sented <1970 (when antiepileptic drug monitoring was beginning to become generally available); those who first presented between 1970 and 1979, a de- Ther Drug Mont, Vo. 1, No.5, 1996 cade during which antiepileptic drug concentration monitoring came into increasing use; and those who presented from 1980 to 1989, a period during which plasma antiepileptic drug monitoring had become routinely available. Of the 548 antiepileptic drug- treated patients first seen <1970, 179 G2.7%, or 47.4% of those whose outcomes were known) achieved at least I year’s full seizure control while receiving drug therapy and 199 (36.37%) were known not to have achieved this end point. For the 960 patients first seen between 1970 and 1979, the cor- responding figures were 284 (29.6%, or 47.2%) and 329 (34.3%): for those 348 first seen >1979 but <1990, full control for at least 1 year was achieved {in 106 (30.5 or 38.7%), and was not achieved in 168 (48.3%). Thus, the increasing availability of plasma antiepileptic drug concentration monitoring did not appear to be associated with improvement in sei- zure ‘control,’ as defined above. Rates of reaching the stage of a planned with- drawal of antiepileptic therapy after 3-5 years full freedom from seizures and preferably with electro- encephalograms (EEG) no longer showing paroxys- mal features, were considered. Some 93 of 548 pa- tients (17.0%) first seen <1970, some 143 of 960 (14.9%) seen between 1970 and 1979, and some 35 of 348 (10.176) seen > 1979 and <1989, achieved this end point, although some of the latter group had not been followed long enough at the time of data anal- ysis for cessation of treatment to be considered This set of results also provided no evidence of im- proved seizure control during the period in which plasma antiepileptic drug concentration monitoring became increasingly available. However, the above analyses did not compare outcomes of those whose plasma drug levels had been monitored with those whose levels had not been measured. As well, at least two potentially confounding factors began to apply in the above analyses with the passage of time. Firstly, from 1975 on, the proportion of children newly referred each year declined progressively, as pediatric neu- rologists began to practice in Brisbane (5). For the whole series, the prospects for attaining seizure control and arriving at the point of withdrawing an- tiepileptic therapy, were better for children than adults (5). Secondly, as the author became more senior, there was a tendency for more referrals of difficult-to-control epileptic patients, who would be less likely to attain as long as 1 year's freedom from seizures. To try to minimize the effects of these factors, only patients first seen <1976 who hadPLASMA ANTIEPILEPTIC DRUG MONITORING 461 been followed long enough for their outcomes to be known were next studied. In this group, rates of achieving at least I year of seizure control were ‘compared with those who had their plasma antiep- ileptic drug concentrations monitored and those who had not. Results are shown in Table I, in which ‘outcomes for certain subdivisions within the popu: lation studied are also set down. In those who pre- sented within 6 months of the onset of their seizure disorders, there was statistically significant evi- dence that plasma antiepileptic drug concentration monitoring was associated with a better chance of achieving at least 1 year’s total seizure control. In view of this result, and because it could be argued that patients with solitary seizures were included in the analysis and had a better prognosis that those with multiple seizures (5), the outcome of monitor- ing was assessed in all patients in the entire series ‘who were seen early in the course of their seizure disorder (in the first 6 months), but who had already had more than one seizure, Of those monitored, with known outcomes, 47 of 78 (60.3%) went at least 1 year seizure+ree: of those not monitored, only 49 of 114 (43.0%) achieved this (difference 17.396; 95% Cul. 3.11-31.4%).. Antiepileptic Drug Monotherapy Ethosuximide was used only to treat absence sei- zures and then was always employed in conjunction with an anticonvulsant prescribed to prevent tonic- clonic seizures, and the combination was therefore regarded as tantamount to ethosuximide monother- apy in relationship to absence seizures. Table 2 shows the proportions of patients with generalized onset and with partial onset seizures who achieved at least 1 year of complete freedom from attacks while receiving monotherapy with phenytoin, car- bamazepine, phenobarbital (or a drug metabolized to phenobarbital, i.e. methylphenobarbital or primi- done), or ethosuximide, and who had their plasma antiepileptic drug concentrations monitored. The mean plasma concentration of each drug in those whose seizures were controlled and in those whose seizures were not controlled are also shown. A se- ies of confidence interval analyses showed that there were no statistically significant differences be- tween the I-year rate of full seizure control achieved by monotherapy with phenytoin, carbam- azepine, or phenobarbital (or drugs metabolized to it) in generalized onset or partial onset epileptic sei- zures, or in the I-year control rates in generalized ‘onset and partial onset seizures for any one of these rugs. With the exception of ethosuximide, mean plasma drug concentrations were higher in unsue- cessfully treated patients, suggesting that antiepi- leptic drug underdosage was unlikely to have been a ‘major factor in treatment failure, ‘The cumulative percentages of patients whose epileptic seizures were controlled for at least 1 year are shown plotted (spline fits) against the plasma drug concentrations by which (not ar which) such control was attained in Figs. 1-4 for persons taking, respectively phenytoin, carbamazepine, phenobar- bital (or a congener), or ethosuximide, each as monotherapy. Overall, the conventional upper lim- its of the therapeutic ranges for the drugs corre- sponded reasonably well with the plasma concen- trations above which there were only few additional patients who attained seizure control. From the data in Figs. 1-4, the upper limits of the ranges would have been 70-80 (but with relatively little gain from achieving levels >60), 40-50, 120-140, and (less definitely) 800 umoV/L for phenytoin, car- bamazepine, phenobarbital, and ethosuximide, re- spectively. There might have been an argument for setting the upper useful limit of phenobarbital con- centrations for the drug used as monotherapy in generalized onset seizures (apart from absences) at ~100 pmol/L, with the higher value of 140 amoV/L TABLE 1. Percentages of patients with known outcomes seen prior to 1976 who achieved at least I year of full seizure control and whose plasma antiepileptic drug concentrations were, or were not, monitored See Difference Confidence interval Subjects studied Monitored Not monitored @, 05%) All subjecte 154330 46.7% Tiss S0.ae) 376 3.28-108% Sin years 95223 42.6%) 1227293 41.6%) 0.36, = 11649.569% S12 years 59/107 (53.15) H3I62 (65.3%) oz 12-21.0% Presented <6 months from onset 95/198 48.0%) Bu 105-25.19 5895 (611%) Results are shown forthe fl set of patients, for those aged 12 years, for those $12 years, ofthe onset oftheir seizure disorder * Statistically significant diference nd sso for those seen within 6 months Ther Dr Monit, Vo. 16, No. $, 1984462 M. J. EADIE TABLE 2. Outcome of monotherapy with phenytoin, carbamazepine, phenobarbital (or methylphenobarbital or primidone), and ethosuximide in patients with generalized onset and with partial onset epileptic seizures Controlled Phenyioin No. a9 % 538 Mean concentration" no sD 287 Carbamazepine No. 20 % 0 Mean concentration* 308 SD 182 Phenobarbital No. ares % 65.1 “Mean concentration” 483 SD 12 thosuximide No. a2 % 50 ‘Mean concentration” 4547 8D 1797 res Paria epilepi Not controlled Gontroted 1809 sano szini0 462 327 73 509 06 S18 97 a4 403 20 29152 26152 0 30 50 Prey 304 3a 93 84 ns sss ure nin 349 50 50 0s ra Ho 3 a8 490 um 50 a 212 Drug concentrations given in moll, applying for partial onset seizures. The upper limit of the ranges for phenytoin and carbamazepine did not seem to differ appreciably between generalized ‘onset (excluding absence and preadolescent myo- clonic) seizures and partial onset seizures. There ‘were no sharp lower limits for the ranges of bene ficial concentrations for phenytoin, phenobarbital, and ethosuximide, there being progressively in- creasing percentages of patients with fully con- trolled seizures once the relatively low plasma con- centrations of, respectively, 10, 10, and 80 umoV/L 0 60 so PERCENT CONTROLLED 30 fp eee ete ey 30 60 70 80 90 100 110 120 ‘0 1020-30 40 were exceeded. All these concentrations were dis- tinctly below the usually quoted lower limits of therapeutic ranges for these agents (respectively, 40, 25, and 300 mol/L.) In contrast, there was relatively sharp lower limit to the range of beneficial monotherapy, viz. 20 pmol/L (not diss conventional therapeutic range lower limit of 20-25 mol/L). Beyond this threshold, success rates in fully controlling seizures increased rather abruptly, with little further dividend to be obtained from FIG. 1. Percentages of patients ‘whose seizures were controlled for atleast 1 year plotted (spline fits) against plasma drug concentrations for 39 patients with generalized ep Hleptic seizures and 110 patients with partial epileptic seizures treated sith phenytoin monother apy (upper pair of curves, with Solid symbol), and corresponding data for 96 patents with general ized epileptic seizures and 173 pa tients with partial epileptic seizures Who received phenytoin as part of multiple antiepileptic drug regimens (lower pair of curves with open symbol), PLASMA CONCENTRATION (micromole /L) ‘Ther Drug Mont, Vo. 1, No 5, 1998PLASMA ANTIEPILEPTIC DRUG MONITORING 463 60 FIG. 2. Percentages of patients hose seizures were controlled for atleast 1 year plotted (spline fis) ‘against plasma drug concentrations for 20 patients with generalized ep- PERCENT ileptic Seizures and 32 patients with Partial epileptic seizures who re- CONTROLLED. Eeived carbamazepine mogother py (upper pat of curves with solid symbols), and corresponding data for 78 patents with generalized ep ileplie seizures and IIB patients ‘With partial epileptic seizures ven carbamazepine as part of multiple antiepileptic drug regimens (lower Dir of curves with open symbols) achieving plasma levels of carbamazepine >40 uumol/L. There were statistically significantly better rates of full seizure control-drug concentration re- lationships for generalized onset than for partial on- set epileptic seizures treated with carbamazepine or with phenobarbital in monotherapy, using sign tests (both p < 0.02). This did not apply for phenytoin in these two major types of epileptic scizure. The mag- nitude of the differences in rate of full seizure con- trol would have been of little clinical consequence for carbamazepine. However, appreciably lower plasma phenobarbital concentrations were associ- ated with a similar chance of full seizure control, generalized onset as compared with partial onset seizures, e.g., to achieve complete seizure control in 40% of treated cases, the plasma phenobarbital PERCENT CONTROLLED PLASMA CONCENTRATION (micromole/L) i020 30 40~~60 mol/L, car- ‘bamazepine concentrations >40 moVL., and phe- nobarbital concentrations above 100 umol/L, were associated with relatively little improvement in the rate of achieving full seizure control, although drug doses were raised to the patient's limit of tolerance when seizures were not controlled by lower doses. Ther Drug Mont, Vo. 1, No.5, 1996 While the upper limit of the conventional therapeu- tic range of plasma phenytoin concentrations is of- ten set at 80 pmol/L, this value seems to have been determined from work such as that of Lund (12), who suggested that, at least up to this concentra: tion, seizure frequency decreased, but did not show that seizures in more patients became completely controlled. During monotherapy, the behavior of the full sei zure control—drug concentration relationships in the region of the conventional lower limits of the therapeutic ranges was interesting. In the case of carbamazepine monotherapy, there was a reason- ably distinct plasma drug concentration (~20 umol/ L), above which the rate of full seizure control in- creased steeply, suggesting that there was a definite lower beneficial concentration for this drug. How- ever, this was not the case for phenytoin, pheno- barbital, or ethosuximide. For these agents, rate of full seizure control improved progressively with in- creasing plasma drug concentrations from concen- trations substantially below those which are taken as conventional lower limits of the respective ther- peutic ranges. It could be suggested that use of the usual therapeutic range values as a guide to initial carbamazepine dosage adjustment would have tended to ensure that the apparent lower limit of the antiseizure response-concentration relationship corresponded with the lower limit of the drug's con- ventional therapeutic range. If this were the case, the same lines of argument should also have applied to phenytoin, phenobarbital, and ethosuximide, since the same drug dosage determination practices were used for these agents. Yet, for these drugs, the lower limits of the ranges of beneficial concentra- tions were less than the lower limits of their con- ventional therapeutic ranges. Hence, use of con- ventional therapeutic ranges as guides in the initial stages of adjusting antiepileptic drug dosages prob- ably did not explain the rather sharp lower limit of carbamazepine concentrations at which full seizure control occurred. It should be pointed out that once a successful drug dose was found, the dose was not titrated downwards again to find the threshold at which seizures would recur, as this was clinically unjustifiable if the patient was well on the dose used, Therefore, the apparent lower limits of the ranges may be higher than their true values. The shapes of the response-concentration relationships suggested the possibility of thinking not so much in terms of therapeutic ranges as of plasma drug con- centrations likely to be associated with variousPLASMA ANTIEPILEPTIC DRUG MONITORING 467 probabilities of obtaining the maximal benefit that a drugiis capable of achieving ina given epileptic pop- ulation, From the data of Fig. 1, it could be said that a plasma phenytoin concentration of 60 umol/L had offered nearly all patients the maximum prospect of full seizure control of which the drug was capable, while a plasma phenytoin concentration of 30 mol/L had offered only 50% of patients with po- tentially responsive forms of epileptic seizures this benefit. For carbamazepine, the corresponding val- tues were 35 and 20-25 pmol/L, respectively, and for phenobarbital, 90 and 30-35 mol/L, respec- tively, when these drugs were used as monotherapy in types of seizure disorders likely to respond to them. Attack control rates tended to be higher for po- tentially responsive varieties of generalized onset than for partial onset epileptic seizures when car- bamazepine or phenobarbital was present as a re- sult of antiepileptic drug monotherapy or combina- tion therapy. However, any tendency towards a better full control rate in responsive types of gen- eralized onset seizures was not statistically signifi cant in the case of phenytoin monotherapy, while the respective roles of the agents involved in drug ‘combination therapy was difficult to disentangle. Callaghan et al. (13) had earlier found that phenyt- in and carbamazepine were more effective in gen- etalized than in partial epileptic seizures. The dif- ferences in attack control rates found in the present study for the two major types of epileptic seizures would have been of little clinical importance except in relationship to phenobarbital, for which rates of full seizure control were distinctly higher at a given plasma phenobarbital concentration in patients with generalized onset epileptic seizures of a type likely to respond to the drug (i.e., not absences or myo- clonic seizures with a preadolescent onset) than in partial seizures. A greater efficacy of phenobarbital in generalized than in partial onset epileptic sei- zures has been noted previously (4), although the data in that study were presented in terms of differ- ent therapeutic ranges for the drug in the two types of seizure. Rates of full seizure control were appreciably lower when phenytoin, carbamazepine, or pheno- barbital was present in plasma as part of an antiep- ileptic drug combination. Drug combinations were prescribed only when one antiepileptic agent in its maximum tolerated dose had already failed to stop seizures. Therefore, the combinations were used to treat more therapeutically resistant epileptic sei- zures, for which a lower success rate might have ‘been anticipated and for which contributions of the individual agents would be very difficult to deter- ‘mine, When carbamazepine was present as part of an antiepileptic drug combination, the antiseizure- response-concentration relationship no longer showed the distinct lower threshold above which the rate of full seizure control rose quickly when the drug was used in monotherapy. Rather, the thresh- old was a more gradual one. When carbamazepine was used in combination with phenytoin or pheno- barbital, proportionately more of the drug was me- tabolized to its epoxide (14,15). Carbamazepine- epoxide is an antiepileptic agent (16) that was not accounted for in the present study, and the pres- ence of increased concentrations of this substance may have been relevant to the different shape of the carbamazepine seizure-response-concentration.re- lationship when the drug was used in combination with other agents In theory, combining antiepileptic agents with different mechanisms of anticonvulsant action ‘might be expected to produce an additional antisei- zure effect. In practice, some clinical studies have suggested that this prediction does not apply when human epileptic seizures are treated (17,18), alk though other investigations appear to have shown. that it may (19). In the present study, combining two antiepileptic drugs when one had failed did yield statistically significant additional benefit in re- iationship to full seizure control, but any further gains from adding a third drug were marginal. The data of the present paper were derived from planned retrospective analysis ofa personal series of epileptic patients collected over much of a pro- fessional lifetime, during which therapeutic atti- tudes and practices have changed. Full seizure con- trol has come to be seen as a much more desirable couttome than reduction in seizure frequency, which was often regarded as satisfactory a generation ago. Idiosyncratic factors may have affected some of the outcomes of the study, and patient referral patterns may have influenced the rates of full seizure control found and made them unrepresentative of epileptic seizures in the wider community, although the dis- tribution of seizure types in the present series was generally similar to those of other published series (5). The great majority of outcomes of the study appear to be in general conformity with those in the published literature. Because patients were not for- ‘mally randomized to the antiepileptic drugs studied, although the initial allocation often occurred before Ther Drug Mont, Vol. 1, No.5, 1994468 patients were first seen and reflected the prescrib- ing patterns of the referring practitioner, it may be unwise to draw conclusions about the comparative efficacies of the individual drugs within each major type of epileptic seizure. However, the data ob- tained for the comparative efficacies of the same drug in the two major types of epileptic seizures, i.e., generalized onset and partial onset seizures, are likely to be valid. REFERENCES 1, Browne TR. Pharmacologic principles of antiepileptic drag ministration. In: Browne TR, Feldman RG, eds. Epilepric Seizures: diagnosis and management. Boston: Little, Brown, and Co, 1983:145-60, 2, Fréscher W, Bichelbaum M, Gugler R, Hildenbrand G, Pe nin HA prospective randomised tral on the effet of mon toring plasma anticonvulsant levels in epilepsy. J Newrol 198):204:193-201 3, Beatdsley RS, Freeman JM, Appel FA. Anticonvulsant se ‘um levels ae useful only ifthe physician appropriately uses them: an assessment ofthe impact of providing serum level data to physicians. Epilepsia 1983243305. 4, Schmidt D, Einicke I, Haenel F. ‘The influence of seizure type on the efficacy of phenytoin, phenobarbital and car Dbamazepine. Arch Neurol 1986:43:368-5. 5. Eadie MJ. Epileptic seizures in 1902 patients: a perspective from a consultant neurological practice. Bplepsy Res 1994, 1785-79, 6. Wallace JE, Speetrophotomettic determination of dipheny- hhydantoin. J Forens Sci 1966;11:581-9. 7. Baie MI, Hooper WD. Intermittent carbamazepine intox ‘ation possibly related to altered absorplion characteristics of the drug. Med J Australia 1987;146:313-. ‘Ther Drug Mont, Vol. 1, No.5, 1998 M, J. EADIE Eadie MI, Heazlewood R, Tyrer JH. How worthwhile is plasma primidone measurement? Clin Exp Neurol 1981:18 133 Chaka CA. Axum technical graphs and data analysis. Se atte: TiMetex, Ine, 1989 Gardner MI, Altman DG. Statistics with confidence. Lon: ddon: British Medical Journal, 198. MJ, McKinnon GE, Dunstan PR, MacLaughlin DB, on RG. Valproate metabolism during hepatotoxicity associated with the drug. Quavt J Med 1990;77:1229-40. Lund L. Anticonvulsant effect of diphenythydantoin relative to plasma level, Arch Neurol 197431:289-94, Callaghan N, Kenny RA, O'Neill B, Crowley M, Goggin T ‘A prospective study of carbamazepine, phenytoin and so ‘dum valproate as monotherapy in previously untreated and recenly diggnosed palients with epileptic seizures. J Neurol Neurosurg Paychitry 19858:639-44 Dam M, Jensen A, Christiansen J, Plasma levels and effects of carbamazepine in grand mal and psychomotor epileptic seizures, Acta Neurol Scand 1975,(uppl 60):33-8, Westenbers HGM, van der Klein E, OeiT, de Zeeuw RA. Kinetics of carbamazepine and carbamazepine epoxide de termined by use of plasma and saliva, Cin Pharmacol Ther 1978:23:320-8 Morselli PL, Gemma M, de Maio D, Zanda G, Viani F, Ga- taint S. Pharmacokinetic studies on carbamazepine in vol- ‘unters and epileptic patients. In: Schneider H, Janz D, Gardner-Thorpe C, Meinardi H, Sherwin AL, eds. Clinical pharmacology of antiepileptic drugs. Berlin: Springer Ver- Tag, 1975:166-79, Reynolds EH, Shorvon S. Monotherapy or plytherapy for epileptic seizes? Epilepsia 1981;21:1-10, Schmidt D, Two antiepileptic drugs for intractable epileptic Seizures with complex-patial seizures. J Neurol Neurosurg Psychiaory 1982;48:1118-28, Kutt H, Solomon G, Wasterlain C, Peterson H, Louis $, Carruthers R. Carbamazepine in dificult to coatro epleptic outpatients. Acta Neurol Seand 1977:7(suppl 60):27-32