Annals of Internal Medicine | 2019
Self-management of Epilepsy
Abstract
Epilepsy affects about 50 million people worldwide (1), with the highest rates in children and older adults. Persons with epilepsy have higher rates of injury and premature death than the general population, and many face diminished quality of life (QOL), even if their seizures are controlled (24). Seizure control and medication adherence are common challenges among patients (58). Some forms of epilepsy and many anticonvulsants are associated with cognitive impairment or behavioral issues, complicating care plans and medication adherence. Patient self-management behaviors are important in controlling epilepsy, because decreased patient participation in treatment regimens is a major cause of breakthrough seizures, leading to increased hospital use and mortality (9, 10). In 2003, the Institute of Medicine defined self-management support as the systematic provision of education and supportive interventions by health care staff to increase patients skills and confidence in managing their health problems, including regular assessment of progress and problems, goal setting, and problem-solving support (11). Self-management support is a core component of care delivery models for chronic disease and a requirement for participation in some Medicare alternative payment programs (1216). Systematic reviews have shown that self-management support for patients with chronic illness improves symptoms and role function, but these positive effects are influenced by the type of chronic illness and the self-management skills taught (1719). Further, the effectiveness of self-management may be diminished by conditions associated with epilepsy, such as traumatic brain injury or depressive disorders, and by low health literacy. For patients with epilepsy, robust self-management skills may improve self-efficacy and medication adherence, prevent seizure triggers, and increase patient and family knowledge regarding when to seek urgent medical care. Self-management interventions hold promise for patients with epilepsy, although the paroxysmal nature of seizures, along with prevalent comorbid conditions, may attenuate the benefit. We conducted a systematic review to examine the effect of self-management in patients with epilepsy. Methods Study Design This work is part of a Veterans Health Administration (VHA)funded report available online (www.hsrd.research.va.gov/publications/esp). The present analysis addresses 2 questions: For adults with epilepsy, what are the most commonly used components of self-management interventions, and what are the effects of these interventions on self-management skills and self-efficacy, clinical outcomes, and health care use? This review followed a published protocol (PROSPERO: CRD42018098604) developed with input from stakeholders and follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement (20). We developed a conceptual framework that outlines the population, intervention, outcomes, and potential effect moderators (Appendix Figure 1). Self-management was defined as interventions that aim to equip patients with skills to actively participate in and take responsibility for managing their epilepsy. We adapted an existing operational definition (21) to increase the breadth of eligible interventions (requiring only 1 component beyond knowledge acquisition, instead of 2) to specify that decision-making skills should be aimed at epilepsy-relevant behaviors. Appendix Figure 1. Conceptual framework. ED = emergency department. Data Sources and Searches In collaboration with an expert reference librarian, we searched MEDLINE (via PubMed), the Cochrane Central Register of Controlled Trials, PsycINFO, and CINAHL in April 2018. The MEDLINE search was updated on 25 March 2019 (Appendix Table 1). We also screened reference lists from published reviews. Appendix Table 1. Search Strategy* Study Selection Our prespecified eligibility criteria are listed in Appendix Table 2. Major eligibility criteria were a comparative evaluation of a self-management intervention for adults with epilepsy, delivered in the outpatient setting or by remote technology and using a randomized trial, nonrandomized trial, controlled beforeafter, or prospective cohort study design. We used artificial intelligence technology (DistillerAI; Evidence Partners) to assist with screening abstracts (22). All citations classified by DistillerAI with certainty (that is, eligible or ineligible) were reviewed by 1 investigator (23). All other citations (50%) underwent abstract screening by 2 investigators. Articles included by an investigator or artificial intelligence algorithm underwent full-text screening by 2 investigators. Disagreements were resolved by consensus between the investigators or by a third investigator. Appendix Table 2. Eligibility Criteria Data Extraction and Quality Assessment Data from published reports were abstracted into a customized DistillerSR database by 1 reviewer; a second investigator reviewed these data for accuracy. Data elements included descriptors to assess applicability, study quality, outcomes, and intervention details. Intervention characteristics included interventionist training, family member or caregiver involvement, delivery method, duration or intensity, and peer support. With input from stakeholders, we specified the primary outcomes as self-management behaviors, QOL, and seizure rates. We contacted 1 author to request missing study details but did not receive a response. Two investigators independently assessed study quality, and disagreements were resolved by consensus or through arbitration by a third investigator. We used the Cochrane Effective Practice and Organisation of Care Risk of Bias (ROB) Tool, which is applicable to randomized and nonrandomized studies (24), and assigned a summary ROB score (low, unclear, or high) to individual studies separately for nonpatient-reported outcomes, hereafter referred to as objective outcomes (such as emergency department visits), and patient-reported outcomes (such as QOL). Data Synthesis and Analysis We developed summary tables to describe study characteristics. Initially, we planned to categorize studies as those meeting the full definition of self-management (21) and those with fewer components. However, studies were classified more naturally into 2 categories: those emphasizing education and those emphasizing skill acquisition from psychosocial therapy approaches. Although we planned to evaluate the consistency of effects by components of the intervention, the number of studies was insufficient to perform these analyses. We aggregated outcomes when at least 3 studies were conceptually similar in terms of design, population, intervention, and outcomes. Analyses were stratified by study design (randomized vs. nonrandomized) and by intervention category. When meta-analysis was appropriate, we used random-effects models (DerSimonianLaird estimator with KnappHartung SE adjustment) to generate a pooled mean difference (MD) or the standardized MD (SMD) when studies used different measures for the same outcome. The SMD is the difference in outcome means between the intervention and the comparator divided by the pooled SD. Cohen (25) suggested the following guidelines for interpreting the magnitude of the SMD: small, 0.2; medium, 0.5; and large, 0.8. We evaluated statistical heterogeneity by using Cochran Q and I 2 statistics. Test statistics for publication bias (such as Begg or Egger regression statistics) perform adequately only if more than 10 studies are included in an analysis. Because no analyses met this threshold, formal analysis for publication bias was not performed. Certainty of Evidence The certainty of evidence was assessed by using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach (26). We limited GRADE ratings to outcomes identified by the stakeholder and technical expert panel as critical to decision making. In brief, this approach requires assessment of 4 domains: ROB, consistency, directness, and precision. When applicable, we considered the effect of plausible residual confounders, strength of association (magnitude of effect), and publication bias. These domains were considered qualitatively, and a summary ratinghigh, moderate, or low certainty of evidencewas assigned after discussion by 2 investigators. Role of the Funding Source The U.S. Department of Veterans Affairs was not involved in the design, conduct, or interpretation of the analyses. Results From 2700 citations, we reviewed 166 full-text articles and identified 15 eligible studies (2741): 13 randomized and 2 nonrandomized (Figure 1). Appendix Tables 3 and 4 summarize the characteristics of the included studies. Figure 1. Evidence search and selection. OECD= Organisation for Economic Co-operation and Development. Appendix Table 3. Intervention Characteristics Appendix Table 4. Study Characteristics Self-management Intervention Components Interventions were mapped to the 6 components described in the operational definition: knowledge acquisition; stimulation of independent sign or symptom monitoring; medication management; enhancement of problem-solving and decision-making skills for medical treatment management; safety promotion; and changes in physical activity, dietary, or smoking behaviors. Because some studies had more than 1 active intervention group, 18 intervention groups are described across the 15 studies. Each intervention group had a median of 4 self-management components (range, 2 to 6 components). Medication management and safety promotion were the least frequently addressed components (Appendix Tables 3, 4 and 5). Appendix Table 5. Intervention Components Across Studies We identified 2 distinct groups of interventions, classified by emergent criteria: intervention focus (educational vs. psychosocial therapy) and intervention development (created vs. adapted for patients with epilepsy). The first group f