Despite the availability of nearly 30 efficacious antiseizure medications (ASMs), variability in seizure outcomes remains a significant challenge with 30–40% of patients experiencing uncontrolled epilepsy [1,2,3]. However, this number is questioned because it is usually assessed with the use of monotherapy to control seizure in newly diagnosed patients [3], which is a matter of debate. In patients who have not responded to a first and a second antiseizure treatment, the seizure remission rate may be as high as 20% when treatment is changed to an ASM with a different or complementary mechanism of action [4]. In addition, it has become clear that biological factors such as underlying brain disorders, structural abnormalities, or seizure etiology do not entirely explain why people continue to have seizures while using ASMs. Thus, next to these non-modifiable and yet undiscovered biological factors, an explanation for the persistence of seizures must be sought in other determinants such as behavioral factors. For people with epilepsy, an adequate ASM and its dosage, together with regular intake, are mandatory to prevent or minimize seizures. Thus, any behavior that diverges from the recommended medication use, so-called non-adherence, can negatively impact the effect of ASM and seizure control. The direct relationship between non-adherence and epileptic seizures has been demonstrated in various populations such as children and adolescents [5], adults [6], as well as in various countries [7, 8]. In summary, the most frequently self-reported factors of non-adherence to ASM in the literature are related to the patient, the treatment or the circumstances [9]. In parallel, the most frequently cited factors are the presence of side effects, financial problems, or stigmatization [9]. In the field of (pseudo‑)failure of seizure control, the clinician is often confronted by multiple pitfalls, and assessing non-adherence to treatment represents one of them. However, after decades of adherence research, it is still impossible to predict who will be adherent and who will not [10].

In this article, we delineate the main challenges that healthcare providers face when elucidating the reasons for unmet seizure control, and suggest some feasible interventions.

Case presentation

The patient is an 18-year-old woman who presented in February 2022 for the first time with three consecutive unclear episodes of nausea, visual hallucinations, subsequent loss of consciousness, postictal vomiting, fatigue as well as an insidious start of cognitive decline with impaired memory and concentration. Electroencephalography (EEG) at a private neurological practice was slightly abnormal, but cerebral magnetic resonance imaging (MRI) was completely normal. The symptoms abated over a few weeks without therapy. She had no preexisting diagnosis or treatment for epilepsy and no family history of epilepsy. A second bout of these episodes occurred in May 2022, which led to hospitalization. While repeated cerebral MRI revealed no pathological changes and the cerebrospinal fluid (CSF) was normal, the EEG showed multifocal epileptic discharges over the parieto-occipital and frontal regions as well as short runs of focal-to-bilateral epileptic discharges. A comprehensive panel of antineuronal autoantibodies associated with autoimmune encephalitis in both the serum and the CSF was negative. Thus, the patient fulfilled the criteria for possible (“seronegative”) autoimmune encephalitis according to the Graus et al. criteria [11], given the subacute onset of working memory decline, altered mental status, the seizures with most likely parieto-occipital onset, and the swift bilateral spread according to the EEG and semiology (visual hallucinations, nausea, and loss of consciousness) along with the exclusion of another etiology. The patient was treated with intravenous (i.v.) high-dose methylprednisolone (500 mg daily for 5 days) and i.v. immunoglobulins (0.4 g per kg body weight daily for 5 days). Antiseizure medications (levetiracetam, 750 mg twice daily, and increasing doses of lamotrigine, up to 2 × 100 mg/day) were started. She improved upon a tapering scheme of oral steroids within the next few months, although memory complaints and rare seizures persisted for about 6 months. Therefore, we performed long-term video-EEG monitoring over 73 h in October 2022, which showed normal background alpha rhythm with signs of drowsiness and bi-temporal-occipital focal slowing. After rapid spread to bilateral and anterior regions, polyspike-wave complexes were observed in the bi-frontal and left central regions; thus, the EEG was still very active and we increased the dosage of lamotrigine to 200 mg twice daily.

Surprisingly, serum levels of lamotrigine and levetiracetam were repeatedly sub-therapeutic for lamotrigine and below detectable range for levetiracetam since October 2022 (Table 1).

Table 1 Dosing scheme and serum concentrations of lamotrigine and levetiracetam measured 4 and 8 weeks after antiseizure medication initiation
Full size table

The neurologist suspected non-adherence to treatment or altered pharmacokinetics such as increased metabolism.

Facets of non-adherence

The concept of medication adherence has evolved thanks to the contributions from psychosocial models. Non-adherence to treatment is not “now or never” but a complex and multifaceted behavior [12]. Non-adherence occurs at different phases (when?) and with different patterns (how?). Medication adherence consists of three consecutive and intrinsically related phases that are initiation, implementation, (or execution) and persistence (Table 2; [13]). Thus, the consequences will be different if a patient is not starting a new medication, not implementing the treatment into daily routine, or not persisting with the prescribed drug. In addition, evaluation of adherence should be repeatedly performed because behavior may change with time.

Table 2 List of potential barriers encountered by epilepsy patients during the three adherence phases, with corresponding low-threshold strategies to ameliorate medication adherence
Full size table

Although there are myriad adherence patterns and each patient is unique, ten different types of errors have been described while using medications [14], which will be discussed in the following. The first error consists in obtaining a prescription but starting the intake later or never. In this case, it is called “primary non-adherence” [15] and represents the simplest form of non-initiation (i.e., zero use). Second, “medication holidays” means some gaps during the treatment that range from 1 day to several days. Third, the “toothbrush effect” depicts an intensive medication use immediately before or after the medical visit that returns to insignificant use some days after the visit. Taking the wrong medication (fourth) can have dramatic consequences but is generally easily recognized, while overdosing (fifth), underdosing (sixth), random dosing (seventh), and a wrong frequency (eighth) are more difficult to detect. The ninth error, stopping the treatment too early, is equivalent to non-persistence. Finally, patients may take several medications as they think best, which leads to an erratic form of polypharmacy. Patients can make any, from one to all, of the errors during their treatment, and several errors may simultaneously occur with different medications.

Because the overall goal of antiseizure pharmacotherapy is to maintain therapeutic coverage, that is, concentrations within therapeutic ranges, occasionally missing a dose or compensating missed doses by additional doses might result in a variation in concentration levels. The consequences range from loss of effect (resulting in seizures) to overdosing ASM with toxic and severe adverse effects [16]. Thus, determining the phase and the pattern of adherence is crucial in order to establish the adequate intervention needed (Fig. 1).

Fig. 1
figure 1

Graphical representation of a treatment over 4 weeks (from 9 April to 3 May) and a twice daily intake (black dots) with the three adherence phases: initiation, implementation, and persistence. Late initiation (A), implementation errors (B, underdosing), medication holidays (C, 72 h), and early treatment stop (D) are representative of the patterns detected with electronic monitoring. See text “Facet of non-adherence” for details. The grey areas indicate missing doses

Full size image

Non-response, pharmacoresistance, or non-adherence?

Approximately half of the newly treated epilepsy patients fail to successfully respond to the initial antiseizure medication [17]. Various terms such as “intractable,” “refractory,” or “drug resistant” have been used to describe the situation that seizures are not being controlled by two ASMs. Drug-resistant epilepsy in its proper sense was defined in 2010 as “the failure of adequate trials of two tolerated, appropriately chosen and used antiepileptic drug (AED) schedules (whether as monotherapies or in combination) to achieve sustained seizure freedom” [18]. Independently of the term used, a prerequisite to any intervention is to elucidate the underlying reasons for sub-therapeutic concentrations or impaired response to an appropriate antiepileptic treatment with ASMs (i.e., at a correct dose and schedule). The reasons for treatment resistance are still poorly understood [19, 20], and research targets mechanisms related to the disease, to the ASM, and to genetics [21]. It is noteworthy that non-adherence is also mentioned in articles investigating pseudo-refractory epilepsy [22]. We suggest categorizing the reasons for failed seizure control into biological (unmodifiable) reasons and into therapeutic and behavioral (modifiable) reasons.

Among the therapeutic factors, special attention should be given to contra-indications due to drug–drug interactions. The interaction check should include the multiple ASMs themselves, also including plant products such as cannabidiol, which is a powerful inhibitor of cytochrome CYP 2C19 (increasing markedly the half-life of the active metabolite of clobazam, desmethylclobazam, and thus, inadvertently leading to increased drowsiness, as an example), and antiviral and immunomodulatory medications, including anti-COVID-19 agents [23]. The list of drug–drug interactions with ASMs is exhaustive [24] and all may influence the pharmacokinetics [23, 25,26,27]. However, there are some interactions that are clinically relevant, namely, those mediated by induction or inhibition of drug-metabolizing enzymes [28] or by inhibition of gastrointestinal absorption through modulation of the expression of drug transporters [29]. As a rule of thumb, the interaction is greatest with substances that undergo extensive first-pass metabolism (e.g., itraconazole, indinavir, dihydropyridine calcium antagonists) or with a narrow therapeutic index (e.g., oral contraceptives, oral anticoagulants, immunosuppressants, chemotherapeutic agents [28]). A less clinically important interaction consists in the displacement from protein-binding sites (e.g., phenytoin with valproic acid [28]). Regarding the newest ASMs, particular caution should be taken because our understanding of the mechanisms responsible for their interactions is evolving slowly [27]. In addition, interaction checks should include food, herbal health products, and diet supplements. Only few case reports exist for interactions between Ginkgo biloba and valproic acid, carbamazepine, and lamotrigine [30]; the same is true for glucosamine supplements (at least 1500 mg) with valproic acid [30] and for noni juice fruit with phenytoin [31], for which the exact mechanism of interaction is still unknown [32]. However, because “absence of evidence is not evidence of absence” [33], caution is recommended with the co-administration of alternative medicines especially because patients may use them on a regular basis and over a long time [31].

Among the multitude of individual reasons that exist for not taking medication [9], only a few are modifiable.

Modifiable behavioral factors and interventions

Identifying the current modifiable factors of non-adherence is a prerequisite to developing tailored interventions that will encourage adherence to reliable long-term medication use (see Fig. 2). In addition, tailored interventions are likely to have the greatest impact or effect [34]. Among the modifiable causes for non-adherence, forgetfulness, or carelessness are indubitably non-intentional [35]. The distinction between non-intentional and intentional (i.e., a patient’s active decision not to adhere to the therapeutic recommendation) medication non-adherence has been extensively investigated [36]; however, the studies were seldom conducted in light of modifiable factors of non-adherence [37, 38]. Because individuals who take their medicines (and those who do not take their medicines) do it with a health-related intention, all modifiable causes should be categorized as intentional when the behavior differs from the recommendation.

Fig. 2
figure 2

Flowchart that tackles modifiable reasons for unreached seizure control according to therapeutic, behavioral, and other factors. PGx pharmacogenetic testing

Full size image

Among the modifiable reasons, we distinguish three categories: the medication management; the beliefs/concerns regarding the disease and medications; and the price of the treatment.

Forgetfulness

Forgetfulness is the most often mentioned reason for not using medications [9] and might consist in neglecting to take the medications or to refill the prescription. It is easily overcome by alarms and reminders.

Medication management

In order to facilitate the medication management, the prescriber should consider the patient’s current intellectual and physical deficits when choosing a medication. The following should be avoided: complicated instructions for use, complex medication regimen (high dosing frequencies, halving tablets), large tablets, and tablets with rough surface or unpleasant taste. Further, difficulties with popping medications out of a package or swallowing difficulties might represent another severe barrier to oral intake. Modified formulations (soluble, liquid, granular, powder) are appreciated alternatives. Aversion to medication can be overcome with alternative routes of administration such as transdermal systems, nasal sprays, or intramuscular depot preparations of ASMs, as is done with certain antipsychotic drugs.

Beliefs/concerns

Among the concerns, taking too many medications and becoming dependent on the medications distort the balance between the worries and the need for treatment. This is especially true in epilepsy where multiple agents are often the standard of care. Sometimes, the benefit of ASM combination therapy (as “appropriate polypharmacy”), often in the lower dose range, may be perceived as dangerous by the patients [39]. In this case, physicians should intervene upfront by explaining the evidence for prescribing multiple ASMs to the patients. Patients also may take personal decisions and actions without informing the treating physician [39]. Of note, in cases of a complex medication regimen, reducing simply the number of medications might be insufficient for decreasing the overall treatment burden.

Even if the newer generations of ASMs are better tolerated when compared with the old ASMs, short- or long-term side effects always remain a big concern [40] because they substantially influence the quality of life. Most healthcare providers aim for a balance between good seizure control and side effects, and usually start at a low dose with slow escalation. However, most patients are unaware of the individualized approach of the treatment based on the epilepsy syndrome and the potential adverse effects, among others [41]. Thus, if patients do not recognize the dose-related side effects, the symptoms can drive them to stop the medication early or, more dramatically, to refuse the treatment at lower dose. In this case, explaining the long-term treatment plan and involving the patient in the therapeutic decision (so-called shared decision-making) remains the best motivator for good adherence [42].

Finally, some patients believe they have no benefit from the medications. Especially in seizure-free patients, occasionally missing a dose may have no other consequences than lowering the belief in the necessity of treatment. Although some reasons might be more obscure such as perceived personal sensitivity to medications, healthcare providers must strive to discover the patients’ inner beliefs and their understanding of disease and medications. Questions from some validated scales might be adapted to assess the beliefs, for example, “How ill would you be without your ASM?” [40]. The delivery of targeted information, the readjustment of knowledge, and the increase in health literacy [43] might motivate a patient to recognize that the benefits of medication outweigh the perceived risks. Ultimately, the personal need and necessity for ASMs might be accepted [40].

Costs, generics, and supply problems

Even if rare in Switzerland, the cost of the medicines might be a barrier to the availability of ASM, especially for people with a moderate income [43]. The prescription of generic medications is promoted by healthcare organizations. Because it is recommended to not change ASM in any direction (from brand to generic, generic to brand, or between generics) if seizures are controlled well, the use of generics is legitimate when given in the first place to control seizure. In such cases, prescribing physicians and delivering pharmacists should be extra careful to stick to this generic ASM and not replace it over time. Many neurologists are reluctant to switch from brand to generic ASMs (and even more so between generic AMSs) because of presumed unmet bioequivalence and fear of breakthrough seizures [44]. This fear is not unsubstantiated since evidence exists that switching ASMs can cause breakthrough seizures and lead to higher costs. Thus, and in contrast to other classes of medication, such as cholesterol-lowering or antihypertensive drugs, switching ASMs to generics should be avoided for patients with well-controlled epilepsy [45].

Unfortunately, the recent supply difficulties encountered in many countries also affected certain ASM. As an example, up to 11.3% of epilepsy patients in Germany were forced to switch ASM because of supply problems [46]. Thus, a further factor disrupting adherence has emerged that is unpredictable.

Status epilepticus

Despite the fact that seizures are often triggered by a cascade of events (e.g., emotional stress, sleep deprivation, alcohol), the most common reason for an epileptic emergency in patients with known epilepsy is related to low levels of ASMs, and thus, unequivocally to non-adherence or to the addition of co-medication that interacts with the ASMs. In a study in the emergency room, missed medication was demonstrated in 38% of patients on the basis of low serum concentration/dose ratios from therapeutic drug monitoring [47], and in 71% of patients with self-reported failure to take the correct dose of their medication [48]. Unfortunately, non-adherence is rarely identified in the emergency setting because patients’ personal medication history is mostly unavailable.

Pharmacogenetic testing

With the emerging evidence of genetic determination of variant alleles in transporter genes or cytochrome families, new reasons for non-response to antiseizure pharmacotherapy are now accessible [49]. Today, the evolving field of pharmacogenetics (PGx) addresses the genetic variations that are important for the metabolism of one specific substance in humans. These PGx tests are used to predict how an individual may respond to a specific pharmacotherapy. The evidence regarding the effect of genetic variations on clinical response is sufficient for only a few ASMs and mainly concerns toxicity (Table 3). For example, adverse effects and toxicity (Stevens–Johnson syndrome) are to be expected with lamotrigine and carbamazepine in patients with the HLA‑B variant*15:02, which is more frequently prevalent in South-East Asia, while the HLA‑A variant 31:01 is more prevalent in countries of Northern Europe [50, 51]. High-level evidence for genetic variations leading to alterations in pharmacokinetic metabolism currently exists only for phenytoin. Therefore, most questions about optimizing antiseizure therapies cannot be answered by PGx analysis alone [52, 53]. The PGx information might contribute to answering specific questions in particular cases only.

Table 3 Actionable PGx for ASMs [52, 53]
Full size table

Results—case outcome

The patient in our case is a young woman with possible autoimmune encephalitis and recovery of seizures and cognitive decline 1 year after onset of symptoms. She had no family history of epilepsy and was started on a dual ASM treatment with twice daily intake and the halving of one tablet, which are clear barriers to high adherence. The patient has obtained regular and sufficient refills from the pharmacy. She did not report side effects from the ASM, which may contribute to poor adherence. However, during a medical visit, it became evident that the patient had an aversion to swallowing the large levetiracetam tablets and skipped doses. A liquid formulation was prescribed and dispensed. At a visit in February 2023, serum concentrations were still below the therapeutic range, indicating potential non-adherence. PGx testing was considered to investigate potentially altered pharmacokinetic metabolism of lamotrigine and levetiracetam. However, after consulting with a clinical pharmacist, no PGx testing was performed because known genetic variations cannot currently explain sub-therapeutic concentrations of the two substances.

The pharmacists suggested the use of an electronic smartphone application with reminders including the monitoring of the intakes, together with a pillbox. The patient accepted to use them for 3 weeks. The analysis showed a perfect adherence with 100% of doses taken. Despite perfect adherence, ASM serum levels remained unchanged. One explanation is an adherent behavior to the smartphone application without ingestion of the medicines, which is unsatisfactory. The dual ASM has been kept unchanged. A direct method to assess medicine intake, that is, directly observed intake twice daily in the community pharmacy, was abandoned in order to maintain the trustful relationship with the neurologist. In June 2023, an EEG including blood concentration measurement indicated the absence of activity and confirmed subtherapeutic blood concentrations. With respect to the clinical normalization of her health state as well as the EEG despite the insufficient ASM levels, one might argue that the immunosuppressive treatment led to the resolution of the suspected, possible autoimmune encephalitis with its two main manifestations, cognitive decline and seizures. Thus, it is planned to taper the ASMs when the patient returns from a stay abroad during the summer of 2023.

Practical conclusion

  • Non-adherence is an important cause of epileptic treatment failure and emergency department visits. Reasons for non-adherence are manifold and interrelated. Thus, elucidating the modifiable reasons for unmet seizure control is recommended alongside therapeutic and behavioral factors.

  • The intervention should focus on the individual non-adherence pattern, tailoring the intervention to the patient profile. Against the general belief, epilepsy pharmacogenetics deliver results slowly, with few results of use in the clinic.

  • We recommend the following steps to improve non-adherence to antiseizure medications (ASMs):

    1. Consider medication non-adherence alongside other therapy-related factors (interactions, contraindications) as a reason for uncontrolled epilepsy.

    2. Identify the modifiable reasons for why patients are not using their ASMs as recommended.

    3. Thank the patient for sharing their behavior.

    4. Tailor the adherence intervention to the individual patient profile.

    5. Ask patients routinely about medication use.