The economic impact of epilepsy: a systematic review

Background In this review we aimed to determine the economic impact of epilepsy and factors associated with costs to individuals and health systems. Methods A narrative systematic review of incidence and case series studies with prospective consecutive patient recruitment and economic outcomes published before July 2014 were retrieved from Medline, Embase and PsycInfo. Results Of 322 studies reviewed, 22 studies met the inclusion criteria and 14 were from high income country settings. The total costs associated with epilepsy varied significantly in relation to the duration and severity of the condition, response to treatment, and health care setting. Where assessed, ‘out of pocket’ costs and productivity losses were found to create substantial burden on households which may be offset by health insurance. However, populations covered ostensibly for the upfront costs of care can still bear a significant economic burden. Conclusions Epilepsy poses a substantial economic burden for health systems and individuals and their families. There is uncertainty over the degree to which private health insurance or social health insurance coverage provides adequate protection from the costs of epilepsy. Future research is required to examine the role of different models of care and insurance programs in protecting against economic hardship for this condition, particularly in low and middle income settings. Electronic supplementary material The online version of this article (doi:10.1186/s12883-015-0494-y) contains supplementary material, which is available to authorized users.


Background
Epilepsy is one of the most common neurological conditions which occurs in about 5 to 8 cases per 1000 people per annum in developed countries [1,2]. Epilepsy affects people of all ages and although treatable, often requires lifelong medication and sometimes surgery to control seizures [3]. The high health care costs related to assessment and treatment, surgery and hospitalisation for seizures, as well as lost employment, income, and household work, are well recognised [4]. These costs vary according to the severity of the condition, response to treatment, length of time since diagnosis, and the perspective examined (e.g. health systems, societal or individuals and families). However, the economic impact of epilepsy has been poorly quantified and few studies have evaluated strategies to reduce it [5].
A previous review of the economic impact of epilepsy in high and low and middle income countries was undertaken in 2008 and does not incorporate more recent studies [6] Most other reviews were limited to high-income settings and show associations with the temporal stage and severity of the disease, seizure frequency, drug treatment or resistance, hospital admissions and level of disability [7][8][9][10][11]. However, heterogeneous methods are used and study samples are small, raising issues of generalisability in particular towards those in low resource settings where rates of epilepsy are high [1]. Furthermore, the focus of previous reviews has largely been limited to the expenditure from the perspective of the health sector, excluding individual and household impacts. The aim of our review was to provide an update of the evidence and to examine the costs of epilepsy from societal, health system and household perspectives.

Methods
Data were identified from published research articles and abstracts using a manual search in MEDLINE, Embase and PsycINFO databases, searched from inception to July 2014. We included incidence studies and case series with prospective consecutive patient recruitment within clearly defined geographical and time-limited boundaries. There were no restrictions on the basis of language, sample size, or duration of follow-up. Where articles were published in a language other than English, assistance was sought in translating the articles. Studies excluded were those limited to specific patient characteristics such as sex, where the recruitment strategy used convenience sampling with retrospective recruitment, were limited to unstructured assessment of psychosocial outcomes and focussed solely on clinical outcomes. If several articles reported outcomes from the same study population, data were taken from the first publication that referred to each follow-up period.
Two authors (KP, a public health researcher and Associate Professor MH, Head, Mental Health and Chronic Disease Program, Neurological & Mental Health Division) developed the search strategy using relevant terms that included common keywords for 'epilepsy' (e.g. epilepsy, epileptic, seizures, convulsions) combined with common keywords for 'costs' (e.g. economics, income, health care costs, expenses) (See Appendix for the search strategy). Epilepsy was defined as two or more recurrent unprovoked seizures. An epileptic seizure was diagnosed using the International League Against Epilepsy (ILAE) Commission on Epidemiology and Prognosis definition: "a transient occurrence of signs or symptoms due to abnormal excessive or synchronous neuronal activity in the brain." [12] We also accepted any criteria the study authors used for epilepsy, including idiopathic epilepsy and epileptic syndromes with seizures of localized onset; symptomatic epilepsy and epileptic syndromes with simple partial seizures; symptomatic epilepsy and epileptic syndromes with complex partial seizures; and generalized idiopathic epilepsy and epileptic syndromes. Provoked, unprovoked, cryptogenic, remote, status epilepticus, febrile, convulsive, absence were also included.
The outcomes included were the costs to the health system (drugs, hospitalisations, visits to family doctor, etc.), costs to individuals and households (out-of-pocket costs and patient co-payments associated with treatment), and indirect costs in terms of lost income and production. The latter could be measured either in monetary units or some other measure such as time off work.
KP, MH and JM identified and reviewed papers for inclusion based on title and abstract in line with the inclusion criteria. The reference lists of all full text articles were reviewed to identify further articles of relevance that required retrieval of the full text. For all included studies (published full texts), a data extraction form was used to collect information on study design, setting, and outcome measures (Additional file 1) and data were extracted by one author (KA) and verified for accuracy (BE and SJ). The risk of bias in each study was assessed independently by the authors (KA, FS, MH and JM) using criteria based on a standard quality and risk-of-bias assessment [13].
Quantitative analysis was deemed inappropriate due to heterogeneity in the data, study designs and study settings. As this was a systematic review that did not involve data collection from participants, ethics approval was not required for this study.All information has been reported in accordance with PRISMA (Additional file 2) and MOOSE (Additional file 3) guidelines.

Direct costshealth system
The annual total cost per patient in Italy was €1302 (€ 0.75 = $US 1) [14]. A cost-of-illness study from Spain estimated that the total direct costs per patient for the care of a person with epilepsy was of €1055 per 6 months and the cost of patients attending consultations for epileptic surgery was €2193 [22]. The direct medical costs in France for the first year after newly diagnosed epilepsy were FF14,305 (FF6 = $US1 in 1998) and FF3,766 during second year [20]. One study from the Netherlands [17] and one study from Italy [16] which focused on children and adolescents compared the direct costs among three different health care settings -general practices, university hospitals and an epilepsy centre. They both found that the costs differ significantly between the different settings of health care, even for patients in the same category of epilepsy.

Determinants of costs and variation in direct costs
Nine of the eighteen studies that reported direct costs compared different items of expenditures and found that drugs and hospital services were the major sources of costs [14-17, 19, 23, 24, 27, 32]. Three studies from Italy reported that the costs of epilepsy varied significantly depending on the severity of the disease and the response to treatment [14,16,19]. Two of them found the annual costs of epilepsy were highest in surgical candidates, followed by patients with drug-resistant epilepsy, active non-drug epilepsy, newly diagnosed epilepsy and epilepsy in remission or with occasional seizures [14,19] A Russian study showed that yearly costs for different types of seizures significantly varied for employed and unemployed participants [27]. After therapy had been optimized using new antiepileptic drugs, four (primary generalized, secondary generalized, partial focal and complex focal) out of five types of seizures, with the exception of polymorphic or undifferentiated seizures were of lower mean cost per year to unemployed patients [27]. In France, the first year costs after newly diagnosed seizures were highly associated with aetiological categorisation of seizures at inclusion and to other clinical parameters such as the number of seizures, age and pattern of seizures or being treated or not by antiepileptic drugs [20]. The costs during the second year had lower variance and were highly related to frequency of seizures and whether the patients were treated with antiepileptic drugs. The highest costs were incurred during the first year after a newly diagnosed seizure. Two studies found the health care setting was a significant determinant in the variation of direct costs [16,17]. For instance in the Netherlands, care provided in a general practice setting cost €52 per person per month, whilst in a university hospital it cost €282 and at a specialised epilepsy centre, €357 [17]. These results reflect the varying degrees of severity of epilepsy between the patients treated across the three sites with patients with severe and complex epilepsy most likely treated in the specialised epilepsy centre.
Direct costsout-of-pocket costs As previously reported, six studies from the Netherlands, US, Nigeria, Cameroon, Cote d'Ivoire and India reported out-of-pocket costs. A study from the US [34] and the previously mentioned Dutch study [17] were the only studies that investigated out-of-pocket expenditures in high income countries. In the Netherlands, individuals cared for in a university hospital had the greatest out-ofpocket costs per month (€147) followed by patients cared for in a specialised epilepsy centre (€97) and then general practice (€7).
In the US, out-of-pocket costs for outpatient visits, hospitals stays, emergency department visits and prescription medications were compared between individuals with epilepsy who were uninsured, had Medicaid coverage, had Medicare (<65 and ≥ 65) or private insurance. The uninsured reported the highest out-of-pocket expenditure US $ 1018 for hospital stays. Uninsured individuals also experienced significantly higher per-visit and total costs for emergency department care compared with patients in all other insurance groups. They also paid the most out-of-pocket for prescription medication [34].
In Nigeria and Cote d'Ivoire patients with epilepsy were found to incur substantial out of pocket burdensfor instance, in Nigeria, 50 % of a cohort of children attending a tertiary centre incurred out of pocket costs of over 20 % annual family income [31,32]. The high costs of drugs relative to income can also result in nonadherence to medication, as observed in Cote d'Ivoire [31]. However, a program to support access to medications was found to be effective in ensuring that the costs of treatment to patients in rural Cameroon were manageable [30]. A study set in India looked specifically at treatment discontinuation and found that it was associated with high out-of-pocket costs, unemployment and low socioeconomic status.

Indirect costsproductivity loss
Ten studies from nine countries (Germany, Sweden, Spain, China, India, Netherlands, Nigeria, Ecuador and Bulgaria) estimated the indirect costs related to epilepsy. These studies used a variety of ways to identify indirect costs. Three studies measured the costs in monetary units [22,23,32]. A Spanish study provided an average annual cost estimate at €1528 for lost production [22]. This was measured on the basis of lost employment to patients and caregivers. A German study also measured on the basis of lost employment to patients, estimating the total cost of lost production to be €1610 over a 3 month period [23]. Finally, a study in the Netherlands showed that the loss of productivity due to illness and found that loss of productivity was greatest for patients cared in a specialised epilepsy centre (0.26 days per month) and there was no loss of productive days for those cared for in a university hospital and general practice [17].

Indirect costsemployment status
Five of the studies evaluated the employment circumstances of patients with epilepsy [17,21,23,31,33] and two studies evaluated the circumstances of parents of children with epilepsy, including work capacity, source of income and incapacity rates [28,32]. A study from Sweden focused on the incapacity rate and the source of income of patients with epilepsy compared to a control group and found that after a newly diagnosed unprovoked epileptic seizure, no negative impact in terms of employment status. However it found that income was lower in patients with epilepsy than in controls [21].

Determinants of costs and variation in indirect costs
The Spanish study found that the costs varied significantly between different patients [22] 46 % of the patients had no indirect costs at all whereas 30 % of the patients faced costs of between €3001 and €4000 due to loss of employment.
The Dutch study found that patients from an epilepsy centre reported the highest productivity changes and unemployment rates compared with patients from the university hospital and from the general practices [17].
A German study found that the unemployment rate due to epilepsy was higher for non-surgically treated patients with drug-resistant temporal lobe epilepsy compared to surgically treated patients. Freedom from seizures was found to be a significant determinant for socioeconomic outcomes -64 % of the surgical patients became seizurefree whereas 23 % of the non-surgically treated group achieved freedom from seizures due to modifications in antiepileptic drug treatment [33].

Outcome measurement
Seven studies assessed direct costs through questionnaires, [14,16,17,23,24,32] five obtained data from medical records [15,18,27,30,31] and five studies used an ad hoc diary to detail information regarding epilepsy care (laboratory and diagnostic tests, outpatient evaluations, hospital admissions) [16,17,19,23,24] Three studies used a hospital database to analyse the use of the various health-care services [22,27,29]. Two studies obtained data from a survey which conducted in-person interviews using standard case report forms [20,34] and one from a semi-structured interview [19].
The method of ascertainment was similar amongst the ten studies estimating indirect costs. All of them measured their outcome using self-reported data. Seven studies used a questionnaire [17, 21-23, 26, 28, 32], three used a seizure and cost diary [23][24][25], and one conducted a semi-structured psychosocial interview [21]. One of the studies also collected information on indirect costs from a social insurance database [21].

Quality of studies
Overall there was a low risk of bias in the studies reviewed ( Table 4). The inclusion criteria were clearly defined for over 90 % of studies. Confounders were only accounted for in 50 % of studies so this may have resulted in an over- estimation of the effects reported in the other studies. For the most part, outcomes were assessed using objective criteria and measured in a reliable way indicating a low risk of detection bias in the studies reviewed. Attrition bias may be an issue as 15 of the studies did not report outcome data for participants lost to follow-up. Finally, while appropriate statistical analyses were used in most studies, it is worth noting that the analyses were generally limited to descriptive and univariate analyses.

Discussion
Most studies of the economic impact of epilepsy have focused primarily on the direct costs of treatment and have been conducted in high income country settings. In the small number of studies where patients were followed up from initial diagnosis, such costs were found to peak in the initial year of diagnosis, due mainly to surgery or vagus nerve stimulation (VNS). All studies generally reported significant ongoing costs incurred from medications and outpatient medical consultations, with substantially higher costs associated with ongoing seizures. Out-of-pocket costs were assessed in six studies. One study from the Netherlands assessed out-of-pocket alongside total costs of treatment and found a significantly large component of direct costs (13 % of costs for patients managed by GPs; 52 % managed though a university hospital; 27 % for epilepsy centre) based mainly on the costs of unpaid care and transportation [17]. This finding suggests that much of the direct costs of treatment and ongoing management for patients with epilepsy may slip through the safety net of existing social health insurance schemes. Thus further research in other settings, including those where ostensible universal coverage arrangements are in place, needs to be undertaken to explore the burden of out-ofpocket costs. Such evidence will most likely be specific to health systems and arrangements within them for the reimbursement of health care costs associated with epilepsy.
In Cote d'Ivoire such costs were found to be a major contributor to non-adherence to medications [31]. In such resource-poor settings, where the availability of safety nets in terms of social health insurance and social welfare are limited or non-existent, the potential economic burden on households is likely to manifest in financial catastrophe, under-treatment, poor adherence and treatment abandonment. Other factors beyond costs also come into play, such as the lack of availability of drugs, stigma associated with epilepsy and negative attitudes towards western medicine.
One study that examined the role of private health insurance from the US found that lack of health insurance coverage was associated with fewer visits to neurologists and greater out-of-pockets costs of medicines, compared to those with insurance. Medicaid coverage was found to offer financial protection in terms of lower out-of-pocket payments despite the same health care utilisation as those with private insurance [34].
Where assessed, indirect costs associated with loss of productivity and employment, were shown to have been major sources of burden associated with this condition. These were reflected in reduced employment/productivity, school attendance and income; as very few of these studies provided monetary estimates of these effects, it is not possible to draw conclusions about the relative burden of indirect from direct costs. The conclusions within individual studies were that indirect costs were overwhelming and constituted a significant burden to individuals and societies, and tended to be greater when there were ongoing seizures particularly among those in lower socioeconomic groups. More data are required to confirm the robustness of these findings in other settings.
We recognise that as the majority of the studies relied exclusively on self-reported costs, there is the potential for recall bias to affect the findings. While prospective diaries may mitigate some of the problems associated with self-report by reducing reliance on patient recall, these are often unreliable when patients lack information about the nature of procedures and tests being carried out. Ideally, future research would involve the use of linked administrative data sets for the assessment of direct health care costs. However, where self-reported is likely to be the only feasible means of collecting data on out-of-pocket and indirect costs, the use of prospective diaries and minimising the time between follow-up interviews may improve the reliability of such data.
Nineteen of the twenty-two studies we identified were cohort studies; the others were a case-control and quasiexperimental studies. The length of follow-up in these studies ranged from 3 months to 6 years; most were 12 to 24 months. Whilst the evidence suggests that long-term direct costs of treatment tends to remain steady (at least for those with control over recurrent seizures), data from patients with longer follow-up will provide a better understanding of treatment compliance and its relationship with costs over time.
A major limitation of this review is the poor comparability of findings across studies due to differences in methods and scope. This limited our ability to make direct comparisons of the size of the burden of epilepsy across different populations and health care settings. On the other hand, the merits of such an exercise may be questionable as it may be more relevant to examine the factors associated with variations in cost, such as insurance coverage and category of illness, as a means of providing guidance for policy and the development and targeting of interventions. It was not possible to provide conclusive assessment of risk of bias due to variation in study questions and study designs. Given that the studies included in this review were largely observational and the objective was not to generate an estimate of a pooled treatment effect, such concerns over potential bias can to a large extent be discounted.
A further limitation of this review is that the majority of epilepsy sufferers are found in low and middle income countries in Africa, Central and South America (>80 %) however most of the published research on this topic and so included in this review have been conducted in higher income countries where epilepsy may be a comparably lower burden. It is likely that this review underestimates the true economic burden faced by households in settings where there are fewer resources and a weaker health system capacity to support people with epilepsy and their households. This misalignment between the regions with the greatest disease burden and populations most researched highlights an urgent need for more research in low and middle income countries to guide policy and planning initiatives to address the economic burden of epilepsy in these settings.

Conclusions
Our review uncovered a small number of varied studies that have examined the costs associated with treatment of epilepsy. The focus of most of the papers was on the direct costs to health systems but a small number also addressed out-of-pocket and indirect costs associated with loss of income and employment. The main findings are that the key drivers of cost are costs of surgery/VNS, and severity and degree of seizure control. In the Netherlands, where patients are ostensibly protected by universal health care programs, significant out-of-pocket costs were evident due to unpaid care and patient transportation. While in the US, health insurance may offset the costs of treatment and enabling access but does not liberate individuals from bearing a significant burden of out-of-pocket costs associated with ongoing costs of managing illness. Given that the magnitude of such costs is inherently context-specific, there is considerable scope for future research in this aspect of epilepsy, particularly in low and middle income countries.

Additional files
As an appendix, we have provided the search strategy used to obtained the results of this systematic review. As additional files, we have included completed widely accepted systematic review guidelines: the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and Guidelines for Meta-Analyses and Systematic Reviews of Observational Studies (MOOSE). In addition, we have provided the data extraction form employed by the authors to undertake data extraction for each included paper.