Measures of adherence to epilepsy treatment: Review of present practices and recommendations for future directions
Address correspondence to Dr. Angelia M. Paschal, University of Kansas School of Medicine-Wichita, 1010 North Kansas, Wichita, KS 67214-3199, U.S.A. E-mail: firstname.lastname@example.org
Epilepsy is one of the most common neurological disorders worldwide, and the majority of people with epilepsy who live in developed countries manage their condition with antiseizure medication. Surprisingly, therefore, the literature on epilepsy does not document a comprehensive investigation of patient adherence to medication treatment. This paper reviews existing literature on direct and indirect measures of adherence. Based on this review, areas in need for further research have been identified, including improvement of self-report instruments, consideration of cultural factors, attention to patient literacy or numeracy levels, and inclusion of patient-guided measures. While no single method of determining adherence has proved effective, combining direct and indirect measures in a patient-guided, culturally competent atmosphere may increase adherence to treatment, improving health outcomes for this population.
Epilepsy is one of the most common neurological disorders, with a prevalence of approximately 7.1 per 1,000 people in the United States (Hirtz et al., 2007). The majority of people with epilepsy in developed countries are able to manage their condition by using one or more pharmacological therapies (antiseizure medications) (Sander, 2004). Surprisingly, therefore, the literature on epilepsy does not document a comprehensive review of methods of determining patient adherence to medication treatment, though some studies have researched one or more specific methods of determining adherence (Peterson et al., 1984; Lisk & Greene, 1985; Stanaway et al., 1985; Graves et al., 1988; Cramer et al., 1990; Hazzard et al., 1990; Gomes et al., 1998; Adamolekun et al., 1999; Williams et al., 2001; DiIorio et al., 2003; Kemp et al., 2007). Although some researchers have explored this issue among other health topics or examined adherence to medication in general (Nichol et al., 1999; Osterberg & Blaschke, 2005), these methods may not be applicable for professionals working with patients with epilepsy. This paper outlines those methods that have been utilized by epilepsy clinicians and researchers. The authors also examine what is needed to advance adherence assessments for this condition.
Adherence is generally defined as the extent to which patients' treatment-related behaviors (e.g., taking medication, following a diet, modifying habits, or attending clinics) correspond to health professionals' advice (McDonald et al., 2002; Osterberg & Blaschke, 2005). Williams et al. (2001) posit that adherence measures can be grouped into two categories: direct and indirect. Direct measures of determining adherence to treatment for drug-managed epilepsy involve measurement of drug levels in hair or in body fluids such as blood or saliva. Indirect measures involve nonbiological tools such as self-report measures, pill counts, appointment attendance, medication refills, and seizure frequency (Table 1).
Table 1. Major direct and indirect measures of adherence to epilepsy medication treatment; their advantages and disadvantages
| Plasma or serum antiepileptic drug levels||Blood is drawn at least twice (interval may vary); plasma or serum levels of medication are measured||Commonly used, well understood by clinicians Effective in extreme low adherence situations||Patient factors and drug types can cause measurement variability Less accurate in monotherapy cases|
| Detection in human hair||Hair medication levels analyzed using gas or liquid chromatography||Less invasive than blood testing||Researchers disagree on effectiveness|
| Saliva concentration||Saliva samples taken; levels of medication excreted in saliva are measured||Painless Does not require venous access (good for pediatric and geriatric patients) As accurate as blood plasma or serum monitoring||Measurement must be calibrated to individual saliva production and serum/saliva medication ratios|
| Self-report measures||Patient reports medication adherence through methods such as surveys, interviews||Low cost Not physically invasive Adaptable to target population||No standardized and validated measures exist for epilepsy treatment adherence Can be inaccurate due to patient misperceptions or tendency to give desirable response|
| Pill counts||Doses monitored through counts of remaining pills or use of an event recorder||Not physically invasive Event recorders measure regularity of dosing||Does not guarantee that medication is taken outside of controlled environment|
| Appointment attendance||Regularity of attendance at appointments is documented||Easy to collect information Can be related to other adherence behaviors||Tends to be limited term, while epilepsy is lifelong|
|Not proof of proper medication use|
| Medication refills||Medical/pharmacy records reviewed to see if medication is refilled at appropriate intervals||Easy to collect information in managed care setting Can be correlated to blood serum levels||Difficult to collect information if prescription source is not known Not proof of proper medication use|
| Seizure frequency||Frequency of seizures is logged over time||Essential indication of degree to which epilepsy is managed||Nonadherent patients may not have frequent seizures|
The literature on epilepsy treatment documents three main direct measures of epilepsy medication adherence: the measurement of medication levels in blood plasma or serum, in hair samples, and in saliva. Direct measures lend themselves to evidence-based medical approaches and comparative analyses since they provide quantitative data on the physical presence of medication in a patient's body.
Plasma or serum antiepileptic drug levels
The most commonly used measure of adherence among patients with epilepsy is the evaluation of medication levels in blood plasma or serum (Leppik et al., 1979; Peterson et al., 1984; Takaki et al., 1985; Graves et al., 1988; Hazzard et al., 1990; Helgeson et al., 1990; Gomes et al., 1998; Adamolekun et al., 1999; Mitchell et al., 2000; Snodgrass et al., 2001; Williams et al., 2001; Specht et al., 2003; Dutta & Reed, 2006; Landmark et al., 2007). This method involves measuring plasma or serum levels for drugs such as phenobarbital at least twice, with intervals as long as several months (Takaki et al., 1985; Williams et al., 2001). A drop in medication level of a certain predetermined percentage, dependent on study and medication being studied, is indicative of noncompliance. Sample sizes for blood plasma or serum studies have ranged from 24 patients (Gomes et al., 1998) to as many as 264 (Landmark et al., 2007).
Gauging blood medication levels is not exact, with the main concern being variability in evaluations. Several factors have been linked to variability in blood levels for a single patient, including age, food intake, and drug interaction (Gomes et al., 1998). Serum levels also vary according to drug taken. For example, Graves et al. (1988) found phenytoin to differ from baseline in compliant patients by approximately 5 μg/ml or less, while carbamazepine varied only 2 μg/ml. Interpatient variation of medication levels could be as high as 30% for phenytoin and 40% for carbamazepine.
Although plasma or serum measurement is effective in assessing drug intake in extreme low adherence situations, some do not consider it to be sufficiently accurate for optimizing treatment (Affolter et al., 2003), especially in newly diagnosed epilepsy and medication monotherapy cases (Tomson et al., 2007). This is presently the most common direct measurement of compliance among epilepsy patients, but researchers are increasingly associating the measurement of blood plasma or serum medication levels with psychological assessments (Jones et al., 2006) or self-report measures (Kemp et al., 2007) in order to gather complete information about patient adherence to treatment.
Detection in human hair
In addition to blood concentration levels, some researchers have analyzed medication levels in human hair in order to assess epilepsy therapeutic compliance (Goulle et al., 1995; Kintz et al., 1995; Mei & Williams, 1997; Williams et al., 2001; Williams et al., 2002). Human hair incorporates all medications taken by a patient, including recreational as well as therapeutic drugs. Typically, antiseizure medication levels are measured in hair by conducting gas chromatography mass spectrometry (Goulle et al., 1995) or isocratic high-performance liquid chromatography assays (Mei & Williams, 1997).
Drug detection in hair is not a very common form of adherence assessment, perhaps because researchers disagree on the effectiveness of this method. According to Kintz et al. (1995), the utilization of hair samples is not suitable for evaluating the quantity of a drug consumed. By contrast, Williams et al. (2001) concluded that this method is easy to perform and has a similar sensitivity to blood plasma for detection and quantification of phenytoin and carbamazepine in hair samples from an epilepsy inpatient population, while Williams et al. (2002) find hair sampling more accurate than either self-report methods or blood plasma monitoring. Mei & Williams (1997) consider this method to be one of the most advanced in measuring adherence among epilepsy patients when used in conjunction with blood plasma monitoring of antiepileptic drug concentrations.
Finally, adherence to treatment has also been directly measured by sampling saliva for anticonvulsant drug levels (Mucklow & Dollery, 1978; Stanaway et al., 1985; Herkes & Eadie, 1990). Researchers have repeatedly found this method of sampling to offer statistically similar results to blood plasma or serum monitoring (McAuliffe et al., 1977; Tokugawa et al., 1986; Tsiropoulos et al., 2000; Vasudev et al., 2002; Grim et al., 2003; Miles et al., 2003; Ryan et al., 2003; Miles et al., 2004; Malone et al., 2006), though limitations of saliva sampling include the need to calibrate measurements to the saliva production of each individual (Malone et al., 2006) as well as wide variability in the ratio of medication levels in serum to saliva. Ryan et al. (2003) found saliva:serum ratios for lamotrigine ranging from 0.40 to 1.26 for a sample of 37 adult and pediatric patients; these ratios could vary still more depending on the medication used. Therefore, this method of measuring adherence to treatment may be most effective when sampling is tailored to each individual, a disadvantage in care settings with large patient volume.
An advantage of this method is that it is the most painless of direct measures, an important consideration for pediatric patients (Zysset et al., 1981; Baumann et al., 2004). Also, it does not depend on venous access, a benefit for pediatric or geriatric patients who may have poor veins (Grim et al., 2003). Still, very few physicians are familiar with saliva concentration measurement for evaluating treatment adherence (Baumann et al., 2004). Relatively few studies have assessed adherence to treatment using this method; most have focused on adherence only indirectly by studying the relationship between blood and saliva medication levels.
The literature on epilepsy treatment points to five major indirect measures of treatment adherence: self-report measures, pill counts, appointment attendance, medication refills, and seizure frequency. An evidence base is more difficult to document for indirect measures than for direct due to variations in methodology used, although these measures also yield quantitative data about patient adherence.
Self-report measures are the most commonly used measurement in most studies of medication adherence (Osterberg & Blaschke, 2005), and self-report methods such as surveys and interviews have been widely used to study adherence in epilepsy (Loiseau & Marchal, 1988; DiIorio & Henry, 1995; Gomes & Maia Filho, 1998; Mitchell et al., 2000; Paschal et al., 2005; Jones et al., 2006). Self-report measures have the advantage of being low cost, noninvasive, and easily adaptable to a target population. However, these measures vary greatly in terms of how they are developed, whether they have been validated, and to whom they are administered. Researchers have developed several self-report measures specifically for adult and youth populations (Mitchell et al., 2000; Ball & Taderera, 2003). Even within age groups, sampling methods, survey questions, and analytical methods vary. In terms of study populations, some researchers have studied outpatients (Loiseau & Marchal, 1988), others have recruited from clinics (Jones et al., 2006), and still others have been drawn from ongoing studies (DiIorio & Henry, 1995).
Despite the common use of self-report measures to determine treatment adherence, few measures have been validated specifically for the study of epilepsy. A notable exception is the QOLIE-AD-48, which measures health-related quality of life for adolescents with epilepsy and has been found to be reliable and valid (Cramer et al., 1999). Instruments that specifically measure epilepsy treatment adherence are lacking, however, although more general self-report measures may include items relevant to treatment adherence. Gomes & Maia Filho (1998) used a questionnaire that included one adherence question (“Did you forget or miss any of your medicine last week?”). Although judgment of adherence was based on that single question, other questions on the survey also addressed patient medication use behaviors.
While simple to implement (a reason for their common use), problems inherent to self-reported measures may further undermine results. Patient misperception or the tendency to give socially desirable responses might lead to over-reporting of adherence (Lisk & Greene, 1985; Mitchell et al., 2000). Recent studies have combined self-report measures with direct measures of adherence (Kemp et al., 2007) in order to confirm self-report data. However, a standardized, validated tool for measuring epilepsy patient adherence to treatment does not exist at this time.
Osterberg & Blaschke (2005) indicate that pill counts, or counting the remaining pills in a patient's bottle or vial, are the second most common method (after self-report measures) used to judge adherence to treatment for many medication-dependent health conditions. Pill counts are a noninvasive measure, but this method has rarely been used to determine adherence to epilepsy treatment (Lisk & Greene, 1985; Weber & Beran, 2004). One reason is that pill counts are most useful in research settings where control over medication dispensing can take place; the method might not be as useful in assessing adherence in clinical practice settings (Mitchell et al., 2000). This method might be used more frequently in practice than is documented in the literature, although this seems unlikely since patients could easily alter pill counts outside of a controlled environment.
A related form of adherence measurement is the use of a Medication Event Monitor System, or event recorder. These are standard pill bottles with microprocessors in the caps that record bottle openings, each of which is considered a dose. This method was piloted on epilepsy patients and was found to be more accurate at determining daily compliance than either pill counts or blood serum concentrations, which do not measure the regularity of dosing (Cramer et al., 1989). However, event recording does not guarantee that medication is actually taken, and patients may not perceive their medication adherence accurately as determined by the event recorder (Buelow & Smith, 2004).
Documentation of appointment attendance is another method that has been used to measure adherence among epilepsy patients (Peterson et al., 1984; Mattson et al., 1988; Thorbecke, 1988; Cramer et al., 1990; Hazzard et al., 1990; Tsai & Cheng, 1992; Adamolekun et al., 1999; Mitchell et al., 2000; Al-Faris et al., 2002). For example, Mitchell et al. (2000) measured adherence to treatment among children aged 4–13 years of age over a 6-month period, using as their outcome measure whether families returned to the clinic on scheduled dates. The study concluded that seizure frequency was unrelated to adherence, but that families with more stressors also adhered more closely to treatment in terms of clinical visits. Other researchers have related appointment attendance to caregiver behavior (Adamolekun et al., 1999) or to pharmacological adherence (Mattson et al., 1988).
Studies that measured appointment attendance observed time spans ranging from 6 months (Mitchell et al., 2000) to 18 months (Thorbecke, 1988). Appointment attendance is easily determined from medical records; thus this is a simple measure to implement. However, this method fails to capture some aspects of adherence since epilepsy treatment adherence may be a lifelong issue for patients. Sample sizes in such studies have ranged from as few as 35 (Hazzard et al., 1990) to as many as 238 (Tsai & Cheng, 1992). Although appointment adherence may be a good indicator of a patient's general adherence to epilepsy treatment, it does not necessarily mean that patients are also taking their medications or consuming them properly.
Adherence to treatment may also be measured by determining whether patients fill and/or refill their prescriptions (Peterson et al., 1984; Rochat et al., 2001). In a study on adherence among children with epilepsy, Mitchell et al. (2000) reviewed medical records to assess whether patients requested medication refills. Stanaway et al. (1985) also considered intervals between collections of drugs from pharmacies on their 95 subjects. Ball & Taderera (2003) determined the number of antiepileptic drugs prescriptions that were written upon discharge of their patients, and determined how many prescriptions were refilled by their hospital.
Similarly, Steiner et al. (1988) measured adherence by examining prescription refill records of pharmacies and checking this data against patients' blood levels, finding significant correlation. They concluded that this method was feasible in managed care settings. Consulting pharmacy prescription records for epilepsy patients works well in a closed pharmacy system or among centralized pharmacies (Osterberg & Blaschke, 2005). However, with recent, growing use of Canadian pharmacists and on-line prescriptions, this method may not be efficacious for most settings, and it is not proof that medications are properly taken.
One of the least common methods in assessing adherence to treatment is measuring seizure frequency over time, perhaps due to the fact that even nonadherent patients may experience seizures only rarely (Snodgrass et al., 2001). Only a few studies reported the use of this approach, and these did so in combination with other direct and indirect adherence measurements (Peterson et al., 1984; Adamolekun et al., 1999; Jones et al., 2006). Still, seizure frequency is an essential manifestation of the degree to which epilepsy is managed, and clinicians are likely to consider frequency of epilepsy seizures in everyday practice as a valid measure of adherence.
Each direct or indirect measure of adherence to a course of epilepsy medication treatment contains particular advantages and disadvantages for the patient or caregiver in terms of accuracy, practicality, and sustainability (Table 1). Research has yet to yield an indicator of adherence or nonadherence for epilepsy patients that is completely adequate in itself (Mitchell et al., 2000). However, some methods are clearly utilized more frequently than others in relation to epilepsy treatment, particularly blood plasma and serum concentration levels and self-report instruments. Plasma and serum monitoring in particular remains the standard for judging adherence to epilepsy medication regimens, and this is unlikely to change in the future.
In order to address the shortcomings of particular adherence measurements, several investigators have used a combination of adherence assessments in their studies (Peterson et al., 1984; Lisk & Greene, 1985; Stanaway et al., 1985; Hazzard et al., 1990; Adamolekun et al., 1999; Kemp et al., 2007). Combining direct and indirect methods of adherence assessment may represent the best avenue for accurately assessing the degree to which a patient adheres to pharmacological treatment. Based on the research reviewed above, particular areas that seem best suited for further research include: improvement of self-report instruments, consideration of cultural factors, attention to patient literacy or numeracy levels, and inclusion of patient-guided measures (Table 2).
Table 2. Recommendations for future study of epilepsy treatment adherence methods
|Validation of self-report instruments||Review of existing instruments for best practices, structure, and content|
|Culturally sensitive measurements||Develop culturally competent content for self-report instruments|
|Educate providers as well as patients on issues of cultural competency related to treatment adherence|
|Literacy- and numeracy-appropriate measurements||Develop medication information and instructions appropriate to a variety of health literacy or numeracy levels|
|Measures from patient perspectives||Educational interventions to improve patient-provider communication|
|Measures from patient perspectives||Use community participatory research methods to design instruments or develop other adherence measures|
Validated self-report instruments
Self-report measures are the most commonly used methods to assess adherence in other medical fields, but research into epilepsy treatment adherence using self-report instruments has been inconsistently carried out. Present needs include the development and validation of reliable self-report measures of adherence to epilepsy treatment that are appropriate for different demographic groups; that is, determining whether such instruments actually measure adherence to treatment, and whether they repeatedly return consistent results for a target population. Existing instruments and treatment outcomes could be reviewed to determine best practices for the development of measures, as well as means of validating structure and content.
While self-report instruments will always be limited in the degree to which they can accurately assess adherence to treatment in clinical practice settings, they would be a valuable tool for clinicians nonetheless. With the aid of a validated self-report tool, epilepsy clinicians will be able to assess compliance more effectively among the patients they serve and determine if additional efforts are needed to increase adherence.
Culturally sensitive measurements
Few researchers discussed the need to consider culturally sensitive instruments or approaches in measuring adherence to treatment. As Tsai & Cheng (1992) mentioned, adherence might be influenced by different medical and sociocultural factors that might create barriers. These sociocultural barriers might lead to difficulties in scheduling appointments, misunderstandings between clinicians and patients, misdiagnoses, and low adherence to treatment (Hazzard et al., 1990; Paschal et al., 2005). Other research has shown that culturally relevant approaches can be very effective in assuring successful medical treatment (Wells & Black, 2000). Therefore, issues of cultural competency should be taken into consideration in developing self-report instruments or in adopting the most effective combination of methods to assess adherence for a particular patient. Linguistically and culturally appropriate educational materials should be made available not only to patients, but also to the physicians and care providers who determine treatment regimens and gauge adherence for diverse patients.
Literacy and numeracy-appropriate measurements
Existing research also demonstrated a lack of attention to health literacy and numeracy levels when discussing adherence measures. These related concepts deal with the degree to which patients can understand basic and quantitative health information in order to make decisions about their health (Golbeck et al., 2005). Health literacy and numeracy issues represent a major research need for epilepsy treatment adherence, since Ball & Taderera (2003) found that almost 30% of patients did not understand key aspects of their regimens of medication.
Some researchers did comment on issues of health literacy and numeracy; for example, DiIorio et al. (2003) developed an instrument to gauge the complexity of epilepsy medication regimens, with implications for adherence. Providing medication information and instructions that are appropriate to a patient's level of health literacy and numeracy may increase the patient's understanding of the reasons for maintaining a schedule of medications. This, in turn, may increase adherence, especially in combination with literacy- and numeracy-appropriate self-report measures and physician visits.
Measurements considered from patient perspectives
Finally, few studies considered treatment adherence from patients' perspectives (Trostle et al., 1983; Chappell, 1992; Fisher et al., 2000; Swarztrauber et al., 2003). Those studies that did revealed a variety of nonpharmacological factors that influenced a patient's satisfaction with treatment, including increased communication and emotional support from care providers (Chappell, 1992). Fisher et al. (2000) found that patients considered seizure control the most important aspect of their medication, and that those who were dissatisfied with treatment tended to adjust their medications independently of physician advice. More positively, Peterson et al. (1984) indicated that clinicians were able to improve both communication and adherence through the use of a variety of medication reminder strategies.
Studies of patient adherence to psychotropic medications (a population that tends to have lower adherence than patients taking medication for physical health conditions) have found that patient tendencies to nonadherence or self-medication were mitigated by communication from the care provider (Mitchell, 2007) and family involvement in patient education (Azrin & Teichner, 1998). In order to assist patients with achieving the treatment satisfaction that will ensure their adherence and their best level of health, adherence measurements should be combined with educational interventions aimed at improving patient-provider communication and enlisting the patient's support structure. Using a community participatory approach to designing or selecting instruments and other adherence measurements might be helpful in creating measures that are more relevant and meaningful.
Research into measuring epilepsy treatment adherence remains limited, and many studies documenting the use of adherence measurements among epilepsy patients are dated. Additionally, much research in this area was conducted internationally, and the development of best practices may be confounded by unique cultural factors. United States-based researchers should endeavor to advance this body of work, particularly in the areas indicated above. Only by developing effective methods to gauge adherence to a prescribed treatment can researchers and clinicians ensure the consistent and appropriate management of epilepsy, and therefore a better level of health, for these patients.
The authors thank Kore Liow, MD, an epileptologist and Medical Director at the Via Christi Comprehensive Epilepsy Center in Wichita, Kansas and Professor of Internal Medicine at the University of Kansas School of Medicine-Wichita, as well as Craig A. Molgaard, PhD, MPH, a Chair, Professor, and neuroepidemiologist at the School of Public and Community Health Sciences at the University of Montana, for their expertise and support.
Conflict of interest: The authors confirm that they have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. They have no conflicts of interest associated with the content of this manuscript.