Initial evaluation of low-dose phenobarbital as an indicator of compliance with antimalarial drug treatment.

Initial evaluation of low-dose phenobarbital as an indicator of compliance with antimalarial drug treatment J. Karbwang,' W. Fungladda,2 C.E. Pickard,3 S. Shires,4 A. Hay,4 & M. Feely5 Since poor compliance with antimalarial therapy is often suspected but difficult to prove, this study affempted to establish a model for predicting the plasma concentration of phenobarbital (given in low doses in conjunction with the drug) as an indicator of compliance. Phenobarbital was chosen because its value had been demonstrated as a marker of compliance in long-course therapies, any significant departure from steady-state concentrations (achieved with full compliance) indicating one or more missed doses. Therapy for uncomplicated malaria varies from 5 days with artesunate to 7 days with quinine + tetracycline. Volunteers with confirmed falciparum malaria were randomized into 5 groups and given malaria therapy as well as phenobarbital daily for 3-7 days. Plasma samples for determination ofphenobarbital concentrations were taken just prior to the daily dose ofphenobarbital. Although there was a clearand predictable individual pattem ofblood concentrations following each dose ofphenobarbital, inter-individual variation in blood levels was significant and reduced theirpredictive value beyond the second day's dose. The cause of the variations is not clear; it could be attributable to different sources of the drug, previous intake of phenobarbital by the patient, or differences in drug absorption and disposition in malaria patients. Results for the 5-day artesunate regimen suggest that phenobarbital may be useful as a marker of compliance if the patient stops medication after 3 days; clear differences were evident at the end of the course of treatment between plasma phenobarbital concentrations in individuals completing the 5-day course and those who stopped after 3 days. For the quinine-tetracycline regimen, results suggest that itmay be possible to discriminate between subjects where there is a 3-day difference in treatment. Phenobarbital is a better discriminant when dosing is every 24 hours as with artesunate, rather than the 8-hourly regimen for quinine-tetracycline. When measuring compliance for malaria treatment, if it is important to know what proportion ofpatients reach 3, 5 or 7 days of compliance, then phenobarbital might have a role to play in this assessment, but further investigations in more patients would be required. Altematively, different markers could be used for the doses to be given on these days and, as long as the patient does not mix the doses for the different days, sequential doses and detennination of compliance could be based on an "all or none" detection of the marker rather than on drug levels. Introduction During the last 12 years new methods of measuring patient compliance with drug therapy have been introduced. Comparisons of these methods with I Professor, Department of Tropical Medicine, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand. 2 Associate Professor, Department of Social and Environmental Medicine, Faculty of Tropical Medicine, Mahidol University. Bang- kok, Thailand. 3 Department of Chemical Pathology, Leeds General Infirmary. Leeds, England. 4 Division of Clinical Sciences, School of Medicine, University of Leeds, Leeds, England. 5 Division of Medicine, School of Medicine, University of Leeds, Leeds, England. Requests for reprints should be sent to Dr M. Gomes at the address shown on p. 9. Repnnt No. 5853 patient interviews and tablet counts, which have traditionally been employed in clinical trials, suggest that the older methods greatly underestimated the extent of poor compliance. Such is the extent to which the traditional methods are inadequate that an editorial in the Lanicet (1) suggested that they be abandoned and that electronic medication monitors or a pharmacological indicator, such as low-dose phenobarbital. or both, be used instead to measure compliance during clinical trials. Low-dose phenobarbital was introduced as a pharmacological indicator of compliance in 1987 (2). Since then. it has been employed to evaluate compli- ance with a number of quite diverse treatments and been compared with other methods of evaluating compliance (1, 4-7). However, virtually all of these applications and evaluations have been directed to- wards the measurement of compliance with treat- Bulletn of the World Health O,ganizaOon, 1998, 76 (Suppi. 1): 67-73 0 World Health Organization 1998 67 J. Karbwang et al. ments lasting over several weeks or even longer. This paper reports an evaluation of this indicator for measuring compliance with short (5 or 7 days) courses of antibiotics or antimalarials. Methods Subjects and study protocol. A total of 50 patients with acute, uncomplicated Plasmodium falciparum malaria who attended a clinic for treatment at the Mae Sod Hospital in Thailand were recruited into the study as unpaid volunteers. The subjects were male or female, aged 13-49 years, and weighed 39.5- 69 kg. Patients who were pregnant or had evidence of severe infection with falciparum malaria were excluded. The 50 patients were divided into five groups, as follows. Group 1:10 patients, aged 16-36 years (mean, 23.9 t 6.8 years) and weighing 45-69kg (mean weight 54.1 ± 7.1kg), received 4mg phenobarbital daily for 5 days, which was taken in conjunction with artesunate (300mg on day 1 and 100mg daily for the next 4 days)." Group 2: 5 patients, aged 15-30 years (mean, 19 ± 6.3 years) and weighing 41-53kg (mean, 46.4 ± 5kg), received the same regimen as patients in group l.a Group 3: 10 patients, aged 13-25 years (mean, 19.8 ± 4 years) and weighing 42-57kg (mean, 51.7 ± 5.4kg), received 4mg phenobarbital daily for 3 days, as well as artesunate (300mg on day 1 and 100mg daily for the next 4 days). Group 4: 15 patients, aged 13-48 years (mean, 27 + 11 years) and weighing 40-42kg (mean, 50.9 + 7.6kg), received three separate doses of 2mg pheno- barbital daily (total, 6mg daily) for 7 days, together with quinine + tetracycline (quinine: 600mg X 3 times per day + .tetracycline: 250mg x 4 times per day for 7 days). Group 5:10 patients, aged 16-49 years (mean, 28.8 ± 12.1 years) and weighing 41-63kg (mean, 51.5 ± 6.9 kg), received 6mg phenobarbital daily (3 sepa- rate doses of 2mg each) for 3 days, together with quinine + tetracycline using the same regimen as group 4. Blood (1 ml) was drawn by venepuncture into heparinized tubes from patients in groups .1-3 before aThe protocal was identical for group 1 and group 2; however,.the phenobarbita levels were significantly different in the two groups and they were therefore analysed separately. the consumption of the daily dose of phenobarbital/ artesunate mixture, and daily before the next dose from patients in groups 4 and 5 receiving the quinine/ tetracycline regimen. After centrifugation at 25°C for 15min and 1500g the plasma was separated and transferred to clean plastic 100 il tubes and stored at -70°C before trans-shipment to Leeds, England, for analysis of phenobarbital. Samples were air- freighted on dry ice (-70°C) and remained frozen during shipment. On receipt in Leeds, samples were stored at -20 IC until analysis. Analytical mefftods. Phenobarbital and cyclobarbital (internal standard) were obtained from May& Baker Ltd, Dagenham, England. Methanol, acetronitrile, hexane and ethyl acetate (all HPLC grade) were obtained from Rathburn Chemicals Ltd, Walker- burn, Scotland. Nitric acid (Analar) and sodium dihydrogen orthophosphate were obtained from British Drug Houses Ltd, Poole, Dorset, England. Liquid chromatography. Phenobarbital in plasma and saliva was determined by high performance liq- uid chromatography (HPLC). The HPLC system consisted of a model III Constametic solvent deliv- ery system (LDC/Milton Roy, Stone, England) and a model 460 autosampler (Kontron, Watford, Eng- land) with a 100 >l sample loop. Separation was per- formed using a Partisphere C8 column (12.5 x 0.6 cm (internal diameter); particle size, 5 ,um (Whatman Scientific, Maidstone, England)) in con- junction with a C18 guard column (Waters Chroma- tography Division, Harrow, Middlesex, England). Phenobarbital was detected at X = 205 nim with a NAUR variable wavelength detector (Knauer, Aschaffenburg, Germany) coupled to a Philips PU 4811 computing integrator (Pye Unicam, Cam- bridge, England). Sample treatment. Plasma phenobarbital concentra- tions were measured using a method adapted from Peaker et al. (8) and Wolff et al. (9). Briefly, the phenobarbital method required patient's plasma, quality control (QC),. and plasma phenobarbital standards (0.2 ml) to be added to conical glass flasks (15 ml) with glass stoppers (Orme Scientific, Man- chester, England) that had been washed with nitric acid of the internal standard cyclobarbital (4mg/1). A further 0.05 ml of cyclobarbital (internal standard, 4 mg/I) in aqueous methanol (1 : 1 vol/vol) was added to each conical flask before the contents were thoroughly mixed on a vortex mixer (5 sec) and al- lowed to stand for 5 min. A mixture of ethyl acetate and hexane (3:2 vol/ vol) was prepared and 2.5ml added to each flask. The flasks were stoppered and vortex mixed (5 sec) 68 WHO Bulletin OMS. Vol 76 (Suppl. 1) 1998 Evaluation of phenobarbital as an indicator of compliance before centrifuging (700g, 4°C) for 5 minutes. The supernatant was transferred to acid-washed borosilicate tubes (5ml, Orme Scientific, Manches- ter, England) and taken down to dryness in a Univap model GV2 refrigerated solvent trap and centrifugal evaporator (Uniscience Ltd, London, England) at 40°C for 35mi. Extracts were reconstituted with 0.1 ml of HPLC mobile phase in acid-washed glass vials (0.3-ml, crimped top; Orme Scientific, Man- chester, England), capped, and either analysed im- mediately or stored at -20 IC overnight. Plasma extract (20 1.d) was injected onto the Partisphere C18 column and eluted using a mobile phase consisting of water, methanol, acetonitrile (330:105:65, vol/vol/ vol). The flow-rate was 1.5 mi/ and the retention times were 4.7 min and 6.5 min for phenobarbital and cyclobarbital, respectively. Assay validation. Day-to-day assay reproducibility of phenobarbital was assessed by analysis of QC samples. Two different QC were used during the study. The standard deviation and coefficient of vari- ation (CV) of the two QC samples were as follows: QC-1 was 1.05 ± 0.05 mg/l (CV, 5.0%; n = 8); and QC-2 was 0.48 t 0.093 mg/l (CV, 5.7%; n = 10). The within-day assay CV was 1.5% (n = 10). Accuracy was assessed by measuring plasma fortified with phenobarbital over the concentration range 0.125- 4 mg/l and the relationship between the quantity of phenobarbital found to that added, was linear (re- gression, y = 0.0179 + 0.9949x; r = 0.99), Mean recovery of phenobarbital was 90.4 ± 2.9%. The sensitivity of the HPLC assay for phenobar- bital was 0.03 mg/l (signal-to-noise ratio = 3). Level-to-dose ratio. Concentrations of phenobarbi- tal were converted to a level-to-dose ratio (LDR), which takes into account the daily dose of the patient and the individual's weight. LDR was calculated from the following equation: LDR = phenobarbital concentration (mg/l)/prescribed phenobarbital dose (mg-kg-' per 24h) Statistics. Statistical tests were performed using the Astute program for Microsoft Excel (DDU software, University of Leeds). Results and discussion The results from groups 1 to 5 are summarized in Tables 1 and 2 and all the results are shown graphi- cally in Fig. 1 to 5. All of the results are expressed as level-to-dose ratios. In Tables 1 and 2, the level-to- dose ratios obtained on different days were compared with the peak values vis-a-vis day 6 (groups 1 and 2), day 4 (group 3), day 8 (group 4), and day 4 (group 5). Phenobarbital levels have been shown to be a useful marker of drug compliance for long-term ad- ministration of medication, once steady-state pheno- barbital levels have been reached (2). Under these conditions, the steady-state level reflects the cumula- tive dose. The relatively low inter-individual varia- tion at the steady-state allows the demonstration of a difference between the steady-state and non-steady- state reversion achieved by non-compliers. For malaria, treatment regimens of anti- malarials are administered over a much shorter period (maximum of one week), i.e. the regimen is completed long before steady-state phenobarbital levels are attained. Furthermore, the ability to clear parasitaemia changes significantly when 2 or 3 days Table 1: Comparison of phenobarbital plasma level-to-dose ratios on different days after treatment with artesunate in groups 1, 2 and 3. Figures in bold are the reference group in the comparison. Day (number of doses) Mean S.E. Range S.D. Range P value Group 1 (n = 10) Day 4 (3 doses) 3.48 0.14 3.0-4.4 0.43 2.61-4.34 <0.0001 Day 5 (4 doses) 3.98 0.19 3.25-4.95 0.60 2.79-5.15 0.0006 Day 6 (5 doses) 4.74 0.20 4.12-5.9 0.64 3.4-6.0 Day 7 (5+0 doses) 3.78 0.17 3.32-4.81 0.55 2.7-4.86 0.0007 Day 8 (5-0 doses) 3.13 0.18 2.25-3.99 0.56 2.03-4.23 <0.0001 Group 2 (n = 5) Day 4 (3 doses) 2.74 0.15 2.25-3.03 0.33 2.09-3.39 <0.0007 Day 5 (4 doses) 3.17 0.03 3.10-3.26 0.07 3.03-3.30 0.0027 Day 6 (5 doses) 3.68 0.1 3.39-3.92 0.22 3.25-4.11 Day 7 (5+0 doses) 3.09 0.27 2.52-4.10 0.60 1.91-4.27 0.064 Day 8 (5+0 doses) 2.92 0.13 2.46-3.18 0.29 2.35-3.48 0.0038 Group3(n= 10) Day 4 (3 doses) 3 0.15 2.30-3.78 0.49 2.03-3.96 Day 8 (3+0+0+°0.0 doses) 1.32 0.13 0.92-2.28 0.42 0.50-2.14 <0.0001 WHO Bulletin OMS. Vol 76 (Suppl. 1) 1998 69 J. Karbwang et al. Table 2: Comparison of phenobarbital plasma level-to-dose ratios on different days after treatment with a quinine + tetracycline (td.s) regimen In studies 1 and 2. Figures in bold are the reference group in the comparson. Day (number of doses) Mean S.E. Range S.D. Range Pvalue Group 4 (n = 15) Day 4 (9 doses) 3.09 0.014 2.05-3.95 0.54 2.03-4.15 <0.0001 Day 5 (12 doses) 3.79 0.15 2.97-4.94 0.58 2.66-4.92 <0.0001 Day 6 (15 doses) 4.31 0.24 2.19-5.8 0.93 2.49-6.13 <0.001 Day 7 (18 doses) 4.76 0.28 2.31-6.05 1.08 2.64-6.88 <0.004 Day 8 (21 doss) 5.25 0.29 2.33-6.77 1.11 3.07-7.43 Group 5 (n = 10) Day4 (9 doses) 3.36 0.16 2.30-4.09 0.49 2.36-4.35 Day 8 (9+0+0+0+0 doses) 1.84 0.20 1.05-2.84 0.66 0.55-3.13 <0.0001 of medication are missed. If phenobarbital is to be used as a relevant marker of compliance for antima- larial regimens it is therefore important to be able to establish that at least 2 days of medication were missed, i.e. it must be possible to distinguish a differ- ence between levels reached with full compliance compared with levels of two missed days of treat- ment. This would enable one to rule out non- compliance as a cause of treatment failures. An examination of the data we have reported here indicates that there is significant variation be- tween the levels of the different groups studied, sig- nificant inter-individual variation at each time point observed, and significant overlap in the distribution of phenobarbital levels when the total dosage period differs by as little as 2 days. Quantitatively, this is reflected (assuming a normal distribution) in the mean and SD (standard deviation) reported for each set of data points. For each individual there is a clear trend and daily difference in drug concentration lev- els. However, for the group as a whole the variation is large. This leads to levels that can be distinguished only between the last day of treatment and the first two days of treatment; for example, in the artesunate treatment group 1, the levels after 5 days of treat- ment were different from those at day 3; and be- tween day 3 and day 5 there, was overlap in drug concentration levels (Fig. 1). The principal reason for the overlap in the range of phenobarbital level-to-dose ratios (LDRs) be- tween 3 days of treatment (day 4 of levels) and 5 days of treatment (day 6) was the high phenobarbital con- centrations in one individual. This individual had consistently higher phenobarbital concentrations than the other patients (Fig. 1) and this is likely to be a reflection of the biological variation between indi- viduals in the clearance of the drug. To obtain a more representative picture of the population distri- bution, the means ± 1.96 SD were calculated and these are shown in the tables of results for each of the studies. The mean ± 1.96 SD would represent the range of values that would be observed in 95% of the study population assuming a normal distribution. If the results for the one individual with the high phenobarbital levels were removed, there would be no overlap in the range of values between day 4 and day 6. In the second smaller artesunate study (Fig. 2) there was no overlap in the range of phenobarbital LDRs between day 4 and day 6, nor was there an overlap between day 5 and day 6. One explanation for the variation among groups may be that different formulations of phenobarbital (and perhaps different storage conditions) were used among the. various groups. Also, the groups may have differed in their use of phenobarbital (or simi- lar compounds) at entry into the study; this is un- likely in. the present study because phenobarbital is not a drug approved f6r use in Thailand, and special clearance arrangements had to be made to import the drug for this study. A third explanation is that the disease itself (patient's temperature, inter alia) makes a difference to drug absorption and disposi- tion and hence the levels observed. Whatever the reason, within most groups the inter-individual variation is too large for compliance Fig. 1. Phenobarbital level-to-dose ratios (LDR) for pa- tients receiving 4mg phenobarbital daily for 5 days (days 1-6) In conjunction with artesunate for treat- ment of malaria (group 1, n = 10). 6. 5. 0 ' 4.(70 a 3. .0 2. cL 1. A Group 1 1 73 4 5 8 Days WHO 98016 WHO Bulletin OMS. Vol 76 (Suppl. 1) 1998 v . 70 Evaluation of phenobarbital as an indicator of compliance Fig. 2. Phenobarbital level-to-dose ratios (LDR) for patients receiving 4mg phenobarbital daily for 5 days (days 1-5) In conjunction with artesunate (group 2, n = 5). Fig. 3. Phenobarbital level-to-dose ratios (LDR) for patients receiving 4mg phenobarbital daily for 3 days (days 1-3) and artesunate daily for 5 days (group 3, n = 10). Group 2 cc a ._ .0 0 c 1 2 3 4 5 6 7 8 Days 4. 3.5. 3.- 2.5. 2. 1.5. 1 . 0.5 0 Group 3 0 4 8 Days WHO 98018 WHO 98017 to be predicted accurately for any difference less than 3 days. While it is reassuring that even a one- day difference in total dose results in significantly different mean phenobarbital levels, the means do not directly reflect our ability to use an individual's level to discriminate the total dose taken. There is a sinple analogy to the use of mean phenobarbital levels versus an individual's phenobarbital levels for the prediction of drug compliance: the mean weights of children differ among different age groups at yearly intervals, yet it is unlikely that an individual's weight can reliably predict his or her age. Two major factors that affect patients' compli- ance with antimalarials are the duration of symp- toms and the ability to purchase full doses. When the clinical symptoms of malaria cease, the patient is likely to assume complete recovery and cease taking further doses of the drug. Alternatively the patient is likely, from previous experience with the illness, to purchase less than the full course or adopt a "wait and see" policy. Thus, the pattern of noncompliance is usually one of stopping a regimen prior to comple- tion rather than one of missing intervening doses. In a 5-day regimen for artesunate, phenobarbital may be useful as a marker if the patient stops at 3 days. The data shown in Fig. 3 represent phenobarbital concentrations after 3 days of dosing (day 4 levels) and concentrations after a further 4 days during which no phenobarbital was consumed (day 8). The range of LDRs on day 8 (mean ± 1.96 SDs) is 0.50- 2.14 and, on the basis of data in Fig. 3, it is unlikely that the LDRs would be greater than 3.0 on day 6, that is 2 days after the last dose had been consumed. The range of LDRs after 5 days of dosing (day 6) for both groups 1 and 2 indicate that the minimum LDR is 3.25 (see Table 1). LDR values on day 6 represent plasma concen- trations of phenobarbital measured 24 hours after the fifth dose of the drug had been taken. Thus, on the basis of these results, it is unlikely that there would be anv' overlap of LDR values between those who completed a 5-day course of treatment and those who stopped treatment after 3 days. The results for the 7-day regimen of quinine + tetracycline are more difficult to interpret. On the face of it the LDR values shown in Fig. 4 suggest that it is not possible to discriminate beyond the first day. However, a more careful reading of the data suggest that two patients were poorly compliant after day 5. One of these patients may have been partially com- pliant before day 4. Plasma phenobarbital concentra- Fig. 4. Phenobarbital level-to-dose ratios (LDR) for pa- tients receiving 3 separate doses of 2mg phenobarbi- tal daily (total, 6mg daily) for 7 days (days 1-7) In conjunction with quinine + tetracycline for 7 days (group 4, n = 15). 'U +4 4. :0D .0 0 c 2. -C a. 0 Group 4 ;- 2 3 4 5 6 7 8 Days WHO 98019 WHO Bulletin OMS. Vol 76 (Suppl. 1) 1998 4.5 4 3.5 3 2.5 21. 1.5 9-i .0 'c .0 0. 0.5 0 . . . . 71 J. Karbwang et al. tions in these 2 patients up to day 5 rose in parallel with the concentrations in other patients. Had there been no such rise up to day 5, poor compliance could not have been the explanation for the decline in the LDR values in these two patients after this point in the study. If the data for these two patients imply poor compliance, then the remaining data suggest that it may be possible to discriminate between 3 days of treatment. Fig. 5 shows phenobarbital level-to-dose ratios for patients who consume 9 x 2mg doses of phenobarbital daily (day 4) followed by 4 days with- out any of the drug (day 8). The reduction in pheno- barbital concentrations after stopping treatment after 9 x 2mg doses indicates that it is less likely that the phenobarbitat LDRs at day 7 for this group would overlap with those of compliant subjects on day 7. One of the difficulties with the quinine + tetra- cycline regimen is that the drugs are taken 3 times a day and hence blood measurements performed 8 hours after consumption may still be influenced by drug absorption. Measurements performed 24 hours after the last drug has been consumed, as in the artesunate study, are not affected in the same way. There is likely, therefore, to be more variability be- tween individuals in plasma concentrations of phe- nobarbital measured 8 hours after dosing and this may be another explanation for the range of results seen in the quinine + tetracycline study. The value of pharmacological markers depends on the question it is expected to answer. If the clini- cian or malaria control manager is concerned to know (i) when most patients stop their regimen, and (ii) whether a longer half-life drug such as mefloquine should be given in combination with a short half-life drug such as artesunate, it may be Fig. 5. Phonobarbital level-to-dose ratios (LDR) for pa- tients recelving 3 separate doss of 2mg phenobarbl- tal daily (total, 6mg daily) for 3 days (days 1-3) In conjunction wtth quinine + tetracycline for 7 days. :tO .06 0 .CGJ 4.5 4. 3.5. 3.- 2.5. 2 1.5 1 0.5. 0 0 Group 5 4 Days 1 WHO 98020 useful to use a different approach to marking drugs. An alternative approach is to have different drug markers administered on specific days during the course of therapy. In (i) above, for example, if it is important to know what proportion of patients reach 3, 5 and 7 days of compliance, different markers could be used for the doses to be given on these days and, as long as the patient does not mix the doses for the different days, sequential doses and determina- tion of compliance could be based on an "all or none" detection of the marker rather than on drug levels. In this situation, the marker used would be based upon criteria of ease (simple urine or blood tests) and the cost of conducting the test. From a practical point of view, it would be ideal if a single sample of body fluid were taken after the last day of the full course and the sample assayed for all mark- ers from the same sample. This is effectively what was done in the Myanmar case study described in this Supplement of the Bulletin (see pages 35-41). Acknowledgement This study was supported by the UNDP/Wodd Bank/WHO Special Programme for Research and Training in Tropical Diseases. Resume Evaluation initiale du phenobarbital employe a faible dose comme indicateur de l'observance du traitement antipaludique La mauvaise observance du traitement anti- paludique est souvent pr6sum6e, mais est difficile a prouver. C'est pourquoi cette 6tude tente de mettre au point un mod6le pr6dictif de la concentration plasmatique du ph6nobarbital (administr6 a faible dose en meme temps que le m6dicament) comme indicateur de l'observance. Le choix s'est porte sur le ph6nobarbital en raison de son int6rdt comme marqueur de l'observance dans les traitements au long cours, tout 6cart important par rapport aux concentrations stables (atteintes avec une obser- vance parfaite) indiquant l'oubli d'une ou de plusieurs doses. Le traitement du paludisme non compliqu6 a une duree qui va de 5 jours avec I'art6sunate a 7 jours avec I'association quinine + tetracycline. Des volontaires ayant un paludisme a falciparum confirm6 ont ete repartis al6atoirement en 5 groupes et ont requ chaque jour pendant 5-7jours le traitement antipaludique ainsi que le ph6nobarbital. Un prelevement de sang a W fait WHO Bulletin OMS. Vol 76 (Suppl. 1) 199872 Evaluation of phenobarbital as an indicator of compliance juste avant la prise quotidienne de phenobarbital pour faire un dosage de la substance. Si la concen- tration sanguine individuelle suit un schema clair et pr6visible apres chaque prise de phdnobarbital, la varation interindividuelle est importante et fait perdre de son interet predictff A la concentration sanguine aprbs la dose du deuxieme jour. La cause des vanations n'est pas claire; celles-ci pourraient ftre imputees & I'origine variable du medicament, a la prise antbrieure de phenobarbital par le patient, ou & des differences d'absorption du produit ou d'6tat des impaludes. 11 est recommand6 d'utiliser d'autres m6thodes pour mesurer l'observance au cours du paludisme. Si, par exemple, il est impor- tant de connaitre la proportion de patients qui respectent le traitement pendant 3, 5, et 7 jours, on pourrait utiliser des marqueurs diff6rents pour les doses A administrer ces jours-la, et dans la mesure ou le patient ne confond pas les doses et les jours correspondants, on serait en mesure, pour les doses s6quentielles et la d6termination de l'observance, de s'appuyer sur Ia detection ((tout ou rieno du marqueur plut6t que sur le dosage du medicament. References 1. Patient compliance in therapeutic trials. Lancet, 1991, 337: 823-824. 2. Feely M et al. Low-dose phenobarbitone as an indica- tor of compliance with drug therapy. British jouumal of clinical pharmacology, 1987, 24: 877-883. 3. Pullar T et al. Use of a pharmacological indicator to compare compliance with tablets prescribed to be taken once, twice or three times daily. Clinical pharmacology and therapeutics, 1988, 44: 540- 545. 4. Penn N et al. Use of a pharmacological indicator to monitor compliance with thyroxine. Europeanjoumal of clinical pharmacology, 1988, 35: 327-329. 5. Pullar T et al. Time to stop counting the tablets? Clini- cal pham7acology and therapeutics, 1989, 46: 163- 168. 6. Kumar S et al. Poor compliance is a major factor In unstable outpatient control of anticoagulant ther- apy. Thrombosis and haemostasis, 1989, 62: 729- 732. 7. Hardy E et al. A comparson of a short half-life marker, a long half-life pharmacological indicator and measurements of controlled release "therapeutic" drug in reflecting incomplete compliance by volunteers. Brit- ish joumal of clinical pharmacology, 1990, 30: 437- 441. 8. Peaker S et al. Measurement of low (sub-therapeutic) phenobarbitone levels in plasma by high performance liquid chromatography: an application to patient compli- ance study. Joumal of chromatography, 1989, 497: 308-312. 9. Wolff K et al. Measuring compliance in methadone maintenance patients: use of a pharmacologic indicator to "estimate' methadone plasma level. Clinical trials and therapeutics, 1991, 50: 199-207. WHO Bulletin OMS. Vol 76 (Suppl. 1) 1998 73