Direct evidence of nonadherence to antiepileptic medication in refractory focal epilepsy

Authors


Address correspondence to Nicolas Carpentier, Neurology Department, University Hospital of Nancy, France. E-mail: nic.carpentier@gmail.com

Summary

The adherence to medication in drug-resistant focal epilepsy (RFE) remains largely unknown. The present work aimed to assess the frequency of recent adherence to antiepileptic drugs (AEDs) in patients with RFE. This prospective observational study screened all patients with RFE, admitted to the Nancy University Hospital between April 2006 and September 2008, for a 5-day hospitalization without AED tapering. The adherence to AEDs was assessed by measuring serum drug levels on day 1 (reflecting the recent “at home” adherence) and day 5 (reflecting the individual reference concentration when drug ingestion was supervised). A patient was considered nonadherent if at least one of their serum drug levels was different between days 1 and 5. The day-1 value was considered different from day 5 when it was at least 30% lower (underdosed) or 30% higher (overdosed). Nonadherent patients were classified as under-consumers in the case of one or more underdosed day-1 values, over-consumers in the case of one or more overdosed day-1 values, or undefined if they exhibited both underdosed and overdosed day-1 values. Forty-four of the 48 screened patients were included. Eighteen (40.9%) of 44 patients were nonadherent. Among them, 12 (66.7%) were over-consumers, 4 (22.2%) were under-consumers, and 2 (11.1%) were undefined nonadherents. The study indicates that recent adherence to antiepileptic medication in this group of patients with RFE is poor. Overconsumption is the most frequent form of nonadherence in this population and should be specifically assessed to prevent its possible consequences in terms of AEDs dose-dependent adverse events.

The assessment of adherence to medication in epilepsy yields a wide range of frequency, between 29% and 79% (Sabaté, 2003). In the subgroup of drug-resistant focal epilepsy (RFE), adherence to antiepileptic drugs (AEDs) has not been specifically studied; however, its study is crucial in order to assess intractability, because seizure relapses due to poor treatment compliance should not be considered as evidence of inadequate seizure control (Kwan et al., 2010). In epilepsy, adherence to medication is still commonly assessed by the direct method of serum drug levels (Sabaté, 2003; Paschal et al., 2008). Nevertheless, previous studies using this procedure have often been compromised by the lack of reliable individual drug reference concentrations, by uncontrolled confounding factors, for example, change in diet and concomitant medications (Gomes et al., 1998), by the selection bias related to the measurement of a limited number of AEDs, or by the heterogeneity of epileptic syndromes among the studied population (Specht et al., 2003). Patsalos et al. (2008) indicated that serum drug levels should not be interpreted based on reference range because of the interindividual variability, and stressed the need to identify each patient’s individual effective drug level. In the present study, our purpose was to prospectively assess recent adherence to AEDs in patients with RFE, using serum drug level measurements. In contrast to previous studies, we measured serum drug levels of all commercially available AEDs and used individual reference concentrations measured under the condition of optimal adherence as controls.

Methods

Study design

This prospective observational study screened all patients with the diagnosis of RFE, consecutively admitted to the Nancy University Hospital between April 2006 and September 2008. These patients were referred for a 5-day hospitalization including a neuropsychological and psychiatric evaluation, brain magnetic resonance imaging (MRI), brain positron emission tomography (PET) scan, and interictal single proton emission computed tomography (SPECT) brain imaging. The diagnosis of RFE was determined before this admission, based on the clinical history, video-EEG recordings, and brain MRI (Kwan et al., 2010). No recordings of seizures were performed at this stage of investigation, so no AED tapering was performed prior to or during the hospitalization. Patients were not aware of this study before the admission. All patients were taking brand-name formulations of AEDs. Diet and concomitant medications remained unchanged the week before and during the hospitalization. A nutritionist approved the diet program for hospitalization. The study was approved by the local institutional ethical committee. All patients provided their written informed consent to participate.

Serum drug levels

Serum levels of all AEDs prescribed to each patient were measured on the first day of hospitalization and again on the fifth day. The measure at day 1 reflected the individual “at home” recent adherence, and the measure at day 5 reflected the individual reference concentration under the condition of “optimal” adherence (Patsalos et al., 2008). Blood sample was collected at day 1 immediately after admission. The delay between the morning drug intake reported by the patient and this first sample was recorded by the nurse in each case. During the hospitalization, the AEDs were administered under nursing supervision at the times specified by the individual medical prescription, which had been delivered before patient admission. The second blood sample at day 5 was collected from each patient at the same time after morning drug intake as on day-1, in order to make the AED concentration comparable between day 1 and day 5 (Patsalos et al., 2008). For each AED, both plasma samples were analyzed in the same laboratory (Williams et al., 2003), either with immunoassays or in-house validated methods using liquid or gas chromatography. All methods had a coefficient of variation ≤13% in the range of observed concentrations (Table S1).

Definition of nonadherence

A patient was considered nonadherent if at least one of their serum drug levels was different between day 1 and day 5. The day-1 value was considered different from that of day 5 when it was at least 30% lower (underdosed) or 30% higher (overdosed). Nonadherent patients were classified as under-consumers in the case of one or more underdosed day-1 value, over-consumers in the case of one or more overdosed day-1 value, or undefined nonadherent if they exhibited both underdosed and overdosed day-1 values. A threshold of 30% was defined because all AEDs except phenytoin have been demonstrated to have <30% fluctuation in plasma concentration at constant dosage (Bialer & Midha, 2010).

Statistical analyses

Nonadherent and adherent patients were compared using Pearson’s chi-square test for qualitative values, Fisher’s exact test for qualitative values with restrictive population, and Student’s t-test for quantitative values (with Bonferroni correction if indicated).

Results

Among 48 screened patients, three patients refused to participate. One additional patient was excluded because of missing AEDs measurements. The clinical characteristics of the 44 included patients are summarized in Table 1. For every patient, all AEDs were studied. The posologies of all medical prescriptions complied with international recommendations. Dose frequency of all AEDs was twice daily, except for benzodiazepines (one patient) and phenobarbital once daily. No concomitant non-AEDs known to interact with cytochrome P450 (CYP) hepatic enzyme systems were reported. Among all the patients, delays between drug intake and blood samples ranged from 135 to 955 min. The difference in each individual delay for collecting blood for the same AED after drug intake between day 1 and day 5 did not exceed 30 min.

Table 1.   Clinical characteristics of the population and comparison of the clinical characteristics of the nonadherent and adherent patients
 Total populationNonadherent patientsAdherent patients
Median (min–max)Median (min–max)Median (min–max)
  1. No comparisons had statistically significant differences (with a threshold of p-value set at 0.05).

Age (years)30 (14–63)27 (14–61)34 (17–63)
AEDs per patient2 (1–5)2 (2–4)2 (1–5)
 n (%)n (%)n (%)
Male22 (50)8 (44)14 (54)
Localization   
 Temporal31 (71)10 (56)21 (81)
 Frontal8 (18)4 (22)4 (15)
 Posterior1 (2)1 (5)0 (0)
 Nonlocated4 (9)3 (17)1 (4)
Type of AEDs   
 Benzodiazepines1587
 Carbamazepine1679
 Lamotrigine1596
 Levetiracetam221012
 Oxcarbazepine1129
 Phenobarbital404
 Topiramate1385
 Valproic acid606

Eighteen (40.9%) of the 44 patients were nonadherent. Of the nonadherent patients, 12 (66.7%) were over-consumers, four (22.2%) were under-consumers, and two (11.1%) were undefined nonadherents. The clinical characteristics of nonadherent and adherent patients did not differ significantly. No type of AEDs was significantly more taken by nonadherent patients compared to adherent patients (Table 1). Among all the over- and under-consumed AEDs, only one was an enzyme inducer (carbamazepine) and it concerned only one of the 18 nonadherent patients .

Discussion

The rate of nonadherence was 40.9%. This value is comparable with that observed in previous studies in epilepsy (Sabaté, 2003). This high rate might even have been underestimated because concentrations of long half-life AEDs such as phenobarbital and benzodiazepines at day 5 were potentially not yet at steady-state (Patsalos et al., 2008). We used individual reference concentrations measured under the condition of optimal adherence as controls. This allowed the issue of interindividual variability, and especially those related to the effect of complex polytherapy-related pharmacokinetic interactions on serum drug levels to be overcome. Except for carbamazepine, which was over-consumed in only one patient, none of the over- and under-consumed drugs were enzyme inducers or inhibitors. Complex polytherapy-related pharmacokinetic interactions cannot explain all these differences observed between day 1 and day 5 plasma AED concentrations. The threshold of 30% took into account the coefficient of variation of each analytical method and intraindividual variability of each AED (Bialer & Midha, 2010). Confounding factors such as diet, concomitant medications (Gomes et al., 1998), and delays of collecting blood after medication intake were controlled. For all these reasons, the 30% variability of AEDs levels observed between day 1 and day 5 can be attributed only to recent nonadherence to AED.

Four (22.2%) of the non-adherent patients were under-consumers. Under-consumption can lead to seizure relapses due to inadequate dosage and plasma concentration. Moreover, under-consumption can falsely indicate that an AED “failed” because of lack of efficacy (Kwan et al., 2010).

Nonadherent patients were most often over-consumers (66.7%). This has been previously observed, but never pointed out. This overconsumption should be differentiated from overtreatment due to excessive medical prescription (Perucca, 2002). This overconsumption is probably underestimated in everyday practice in patients with RFE. Overconsumption is important to consider because it exposes patients to higher risks of AED dose-dependent cognitive adverse events (Kwan & Brodie, 2001), and poorer quality of life (Luoni et al., 2011). Altogether, the nonadherence might in fact reflect an erratic AED consumption with a possible alternation of under- and overconsumption. The population studied comprised a majority of temporal lobe epilepsy (71%), which is associated with long-term memory impairment (Helmstaedter et al., 2003), possibly contributing to erratic consumption.

To conclude, this study indicates that nonadherence to AEDs is frequent in RFE and that overconsumption is most frequent. Overconsumption should therefore be specifically and repeatedly assessed to prevent potential AED dose-dependent adverse events.

Acknowledgments

The authors thank patients and caregivers for their willingness to participate in this study, Dr. Nicolas Gambier for interpretation of the data, Dr. Séverine Lhote for collection of the data, and Elizabeth Dulcich for her careful review of English.

Disclosure

Dr. Vignal has received support and has served as a paid consultant for UCB, Pfizer, Esai, Gsk, and Sanofi-Aventis. Pr Vespignani has received support and has served as a paid consultant for Ucb and Gsk. Pr Maillard has received support and has served as a paid consultant for Esai and Gsk. The remaining authors have no conflicts of interest. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Ancillary