Societal costs and quality of life analysis in patients undergoing resective epilepsy surgery: A one-year follow-up

Highlights • Societal costs decrease over the first year after resective epilepsy surgery.• Quality of life increases after resective epilepsy surgery.• Seizure freedom is reached in 73% of cases one year after resective epilepsy surgery.


Background
Epilepsy reduces Quality of life (QoL) and increases the emotional, physical, and financial burden on patients with epilepsy and society considerably.The extent of the burden depends on aspects such as seizure type and response to anti-seizure medication (ASM) [1].Epilepsy affects around 50 million people worldwide [2].In the Netherlands, fifty-five per 100,000 people are diagnosed with epilepsy annually, which amounts to 5,000-10,000 new patients annually [3][4][5].Twenty to forty percent of epilepsy patients will develop drug resistance and about 20-50% of these drug-resistant epilepsy (DRE) patients are eligible for Resective Epilepsy Surgery (RES) [6][7][8][9][10].RES has shown great potential for treatment of DRE patients and may offer a solution for (a) those who continue to circulate as patients in the healthcare system due to their drug-resistance and (b) the costly care-as-usual of DRE patients of €20,751 a year and its economic effect on society [11][12][13][14][15][16].In order to calculate the effect of DRE on societal costs (SC) of epilepsy patients, data are needed, but are scarcely available.Data that are necessary for conducting cost-effectiveness analyses encompasses clinical outcomes such as QoL and Quality Adjusted Life Years (QALY) data, in addition to economic outcomes, which comprise both direct (healthcare) and indirect (non-healthcare) costs.
Widely available are data on the clinical effectiveness of RES, with 50-68% of patients becoming seizure free (for a Cochrane review, see [17]).In addition, various studies have reported on QoL of RES in epilepsy, including study designs comparing pre-and post-operative QoL, randomized clinical trials [14,18,19], and retrospective and prospective cohort studies with ASM control groups, suggesting a significant QoL increase after surgery and this increase is related to seizure reduction [18,[20][21][22].
In contrast, few papers have been published focusing on costs or costeffectiveness of RES.Choi et al. and Sheik et al. presented a costeffectiveness analysis of RES using decision analytic models [23,24].A study conducted in France collected real-world data from 15 centers over 2 years and demonstrated cost-effectiveness of RES between 9 and 10 years after surgery [25].Another American cost study reported that annually RES is $6,800 lower than care-as-usual and demonstrated RES to be cost-saving after 4 years [26].Although these trial-based studies present cost-effectiveness of RES on real-world data, none of the studies have incorporated indirect cost valuation to determine their costeffectiveness.Furthermore, these economic studies did not use a societal perspective and few included both costs and disease-specific QoL data in the same study.
Therefore, the aim of this study was to provide an analysis that covers both economic and disease-specific clinical outcomes by offering an overview of the societal (i.e., direct and indirect) costs directly related to RES.The costs of diagnostic/presurgical evaluation necessary for identifying patients as suitable candidates for resective brain surgery (including neuropsychological evaluation, diagnostic Magnetic Resonance Imaging (MRI), video-electroencephalography (EEG), positron emission tomography (PET), single-photon emission computed tomography (SPECT), magnetoencephalography (MEG), and stereo-EEG) are not included in this analysis.This project serves as a pilot project to offer an up-to-date model for larger cost-effectiveness studies.We hypothesize that (i) SC decrease within the first year after RES; (ii) RES enhances QoL to a clinically important degree and reduced seizure frequency in DRE patients in the first year after surgery; (iii) patients with more comorbidities would benefit less from RES compared to patients with less comorbidities; and (iv) seizure reduction subsequently leads to an increase in QoL in DRE patients in the first year after RES.

Materials & methods
A single center prospective cohort study was conducted.A cohort study design was chosen due to the small population size of the Netherlands and recent failures surrounding inclusion of sufficient patients in randomized controlled trials in larger countries such as the United States [11].This study was approved by the Medical Ethics Review Committee (METC) of Maastricht UMC+ (2019-1134) and registered at the Dutch Trial Register as RESQUE study (Resective Epilepsy Surgery, Quality of life and Economic evaluation) (NL8278).The STROBE and CHEERS checklists were used to strengthen the reporting of our study (Supplementary Table 1 & 2) [27,28].

Study population
All Dutch-speaking DRE patients equal to or over the age of 16 who were referred to the Department of Neurosurgery of the Academic Center for Epileptology, Maastricht and Heeze (ACE) for RES between January 2019 and December 2022 were deemed eligible.To measure a Minimal Clinically Important Change (MCIC) in disease-specific QoL with a power of 80%, at least 25 patients had to be included [29].People with a total intelligence quotient (TIQ) score of < 70 were excluded, which was determined through a neuropsychological assessment by a neuropsychologist.In case of disharmonic intelligence profiles (i.e., discrepancy between verbal and non-verbal/visual-spatial TIQ), the verbal TIQ was leading.

Questionnaires
Participants were asked to complete standardized validated questionnaires two weeks before surgery, and 3-, 6-, and 12-months after surgery.The questionnaires referred to the previous 3 months prior to the time of questionnaire completion.The primary outcome was the change in SC.Secondary outcomes were (i) disease-specific QoL measure by the Quality of Life in Epilepsy 31P (QOLIE-31P) questionnaire [30] (ii) generic health-related QoL measured by (European Quality of Life 5 Dimension (EQ-5D-5L)) [31], and (iii) seizure outcome according to the International League Against Epilepsy (ILAE) outcome scale [32].

Costs
To calculate SC, validated questionnaires for medical consumption (using iMTA Medical Consumption Questionnaire iMCQ) [33] and productivity losses (using iMTA Productivity Cost Questionnaire iPCQ) [34] were used to collect data 3 months before surgery, and then were administered at 3, 6, and 12 months after surgery.The answers to these questionnaires were supplemented with their corresponding patient files from SAP General User Interface for Windows 7.70 (SAP GUI 770) where necessary.In the context of our study, it is important to acknowledge the disparity between the diagnostic/presurgical workup phases and medical consumption immediately prior to RES.The former frequently extends over several months to even years and refers to medical consumption related to identifying DRE and eligibility of those DRE patients for RES.The latter primarily occurs in the immediate time frame prior to RES and refers to all medical consumption directly related to RES of patients already diagnosed with DRE and deemed eligible for RES.Given this distinction and alignment of our objective on reporting costs of RES with the validated questionnaires, the chosen 3-month timeframe before surgery was expected to provide adequate information for this study.A societal perspective was chosen in line with the Dutch guideline for economic evaluations, as it provides a broader scope of cost analyses in comparison with a healthcare perspective that only incorporates direct medical costs [35].This broader scope entails the use of resources outside the healthcare sector, e.g.productivity losses, which may also impact society.
Cost valuation was done using the Dutch guideline for direct (healthcare) and indirect (non-healthcare) costs for one year post surgery [36].All costs are reported in euros.Discounting (i.e., converting future costs to relative value) was not performed as the follow-up period did not exceed one year.Travel costs were calculated by multiplying the average distance in kilometers with standard price including parking costs if applicable.Supplementary Table 3 provides an overview of all reference prices per cost category.

Data collection
Patients received an information letter including the informed consent form by mail or email, along with their invitation for their first neurosurgical consultation.If patients were willing to participate, they either sent their informed consent form back after reading and signing it or agreed to participate and gave informed consent after contact with the researcher during the outpatient clinic visit.
Patients received their survey package timely by e-mail via Castor Electronic Data Capture (EDC).Additional demographic and clinical data were collected using patient files from SAP GUI 770, i.e.: age, date of surgery, sex, body mass index (BMI), hemisphere side of surgery, type of surgery, surgery complications, comorbidities, previous resection, previous neuromodulation, history of ketogenic diet, age of epilepsy onset, number of ASM currently using, and total number of ASM used in the past (simultaneously or consequently).Duration of epilepsy was calculated by subtracting patients' age at onset from their current age.

Valuation
Medical consumption and productivity costs were calculated for the first year after RES along with these costs 3 months before the surgery in order to include costs directly related to surgery, e.g.neurosurgical and anesthesia consultation and MRI-scanning of the brain used for intraoperative neuronavigation.Costs were divided into direct costs (including medication, general practitioner, practice nurse, social worker, physiotherapist, occupational therapist, speech therapist, dietician, homeopath, psychologist, occupational physician, emergency department, ambulance, diagnostics such as MRI scans, specialist or epilepsy nurse consultations, hospital or epilepsy centre admissions) and indirect costs (including informal care by family or friends, paid home care, unpaid labor absenteeism, paid work absenteeism).Direct and indirect costs make up the SC.Due to the sensitive nature of the hospital's financial information, costs of the surgery itself were not available, and instead were calculated by multiplying the cost of surgery per minute by the average operation duration in minutes.The cost per minute was provided by the financial control department of ACE and consists of the amount of money attributed to using a fully equipped and staffed neurosurgical operating room, with specific hardware available, e.g.microscope and navigation equipment.Direct and indirect costs were based on 2014 reference values [37,38] and were adjusted to 2022 with data from the Centraal Bureau van de Statistiek (CBS) [39].The friction cost method was used to calculate loss of paid and unpaid work [36].For unemployed or retired patients or students, loss of paid work was not included, however their daily occupation was recognized as unpaid work.

Statistical analysis
For baseline characteristics, continuous variables were represented as mean and standard deviation (SD).Categorical variables are presented as frequencies and percentages.
To account for missing data, multiple imputation was used according to the method described by Van Buuren et al. [40,41].Thirty imputed datasets were created and predictive mean matching was used for continuous, polytomous regression for categorical, and logistic regression for dichotomous variables.Prior to analysis, total and subscores were calculated for disease-specific QoL scores and generic healthrelated QoL utility scores were calculated for generic health-related QoL based on the Dutch Valuation set [37,38].Seizure frequency was dichotomized for statistical testing into a low (ILAE class 1-2) and high (ILAE class 3-5) seizure frequency group.A MCIC was validated and considered to be 11.8 points for disease-specific QoL total scores [11] and half the standard deviation of the sample for generic health-related QoL utility scores [42].
Linear mixed-effects models were used to analyze (i) change of QoL measures and seizure frequency over time; and (ii) the association between disease-specific or generic health-related QoL with demographic variables (i.e.: age, date of surgery, sex, BMI, hemisphere side of surgery, type of surgery, surgery complications, comorbidities, previous resection, previous neuromodulation, history of ketogenic diet, age of epilepsy onset, number of ASM currently using, and total number of ASM used in the past).To determine the association between seizure frequency and QoL, we used linear regression.Where applicable, the best model fit was determined using the Akaike Information Criterion (AIC) which would later be used in the multivariable analyses [43].
Results from linear mixed-effects models were presented as coefficient with its 95% confidence interval (CI) and p-value.Results from logistic regressions were presented as odds ratio, including 95% CI and p-value.When presenting MCIC QoL scores, non-responders referred to patients who decreased in their disease-specific QoL scores after surgery, intermediate responders referred to patients who increased in their disease-specific QoL scores after surgery but not as much as the MCIC, and responders referred to patients who reached a MCIC increase or more in their disease-specific QoL scores after surgery.For seizure frequency, responders were classified as patient who reached ILAE class 1 or 2, intermediate responders were classified as patients who reached ILAE class 3 or 4, and non-responders were classified as patients who reached ILAE 5 or 6, twelve months after surgery.

Patients' characteristics
Thirty DRE patients who were deemed eligible for RES were included.Patient characteristics are shown in Table 1.There were 3.3-17.5%missing data per variable (Supplementary Fig. 1).Information on the current number of ASM was not retrievable for 3 patients.ASM history was not retrievable for one patient.
Eighteen patients did not suffer from comorbidities.The remainder of the patients suffered from a combination of several diseases (basilar artery aneurysm, endometriosis, hypertension, vascular encephalopathy, hyperthyroidism, diabetes, leukemia, meningoencephalitis, osteoporosis, dyspepsia, depression, irritable bowel syndrome (IBS), orthopnea, addiction, transient ischemic attack (TIA), and/or heart attack).In the 3 months prior to RES (temporal or extra-temporal, see Table 1 for types of resection), none of the patients received subdural grids or sEEG evaluation as part of their diagnostic evaluation.

Societal costs
The SC of the first year after RES were €54,376.These costs included surgery costs and all direct and indirect costs, such as hospitalization days, consultations, transportation to hospital, etc. Considering the average costs of one minute of a neurosurgical operation and calculating the average duration of RES procedure, the average costs of surgery were €3,545.Table 2 and Fig. 1 illustrate the SC per time point, along with the difference in costs.Sixty-nine percent of non-employed patients suffered from comorbidities compared to 29% of the employed patients (Supplementary Table 4).There were no changes in individual employment status over time.

Quality of Life, seizure frequency
Fig. 2 presents the average disease-specific QoL and generic healthrelated QoL scores, which increased over time (59.3 to 64.1 and 0.78 to 0.82 respectively) however demonstrated no statistical significance (Table 3).Most generic health-related and disease-specific QoL subcategory scores increased over time (Supplementary Fig. 2, Supplementary Table 5, and Supplementary Fig. 3).
Across all time points, seizure frequency decreased in most patients after undergoing surgery compared to baseline (p-value < 0.000) (see Table 3).Patients whose seizure frequency decreased experienced a statistically higher disease-specific QoL score over time (p-value = 0.037).Moreover, patients who tried less ASM types in the past (ASM history) experienced a higher disease-specific QoL change after RES compared to patients who had an ASM history of ≥ 5 ASM (p-value = 0.034).
The significant results mentioned above combined with the relatively little change of the total scores of disease-specific QoL over time raised questions about the clinical significance of the disease-specific QoL scores.This suggested the need for further investigation to better understand the nuanced interplay between seizure frequency, QoL domains, and overall QoL in individual patients.
In total, seizure frequency non-responders consisted of 2 patients, intermediate responders consisted of 4 patients, and responders consisted of 23 patients.Before surgery, 13% of patients reported having 4 to 12 seizure days annually and 83% of patients reported daily seizures (Supplementary Fig. 6).From 3 months to 12 months after surgery, in total 4 patients lost their seizure freedom and shifted from ILAE class 1 to a higher classification, leading to 73% of patients maintaining seizure freedom 12 months after RES.

Economic and clinical outcomes
To our knowledge, this is the first study that offers an overview of the SC of RES and provides an analysis covering both economic and diseasespecific clinical outcomes.Our study demonstrates that the average SC of one patient from 3 months prior to surgery until the first year postsurgery entailed €54,376 and decreases over time.A French study of 2016 confirms that total costs-of-disease decrease after RES and that the procedure is safe, effective, and cost-saving [25].It is important to acknowledge the comparison of the SC presented in this study that entailed the first year after RES to annual care-as-usual SC in a similar patient population.A Dutch study from 2014 investigating SC for 203 DRE patients on ASM, including costs of possible side effects, estimated the SC per patient to be €20,751 annually [16].Our results show that the first year after RES is higher SC per patient compared to these annual SC of ASM treatment.Nevertheless, our study presents valuable information on the trend of SC after RES.
Our clinical outcomes, namely 50% of patients increasing in diseasespecific QoL, 53% of patients increasing in generic health-related QoL, and 73% of patients maintaining seizure freedom, add onto the current body of literature.For instance, a randomized controlled trial demonstrated that care-as-usual led to lower and less increase in QoL in comparison to patients undergoing RES [11].Moreover, prospective cohort studies demonstrated that disease-specific QoL increases in 58-61% of patients undergoing RES [11,12,44].Postoperative seizure frequency decrease after RES also corresponds with findings of others, who have reported seizure freedom in 50-68% of surgical candidates [17,45].Lastly, our result which shows that seizure reduction was associated with an increase in disease-specific QoL is in line with several studies with the same research design [21,22,46].

Limitations
Firstly, the baseline measurement of QoL in our study may have been over-or underestimated, due to feelings of either hope or anxiety (or both).In addition, the relatively minimal change observed in the average total QoL scores could be attributed to the potential nullification of individual QoL scores.More specifically, the improvements in QoL for some patients may be counterbalanced by deteriorations or limited improvements in others.It is therefore imperative to place emphasis on the clinical significance of QoL change via MCIC value comparisons.As a result, a deeper insight could be provided into the effectiveness of RES and the overall enhancement of QoL in patients undergoing RES and thereby refine the understanding of the nuanced factors influencing QoL in this specific patient population.
Secondly, there was a discrepancy in the interpretations of one question regarding type of medical consumption received.Whilst patients reported to not have received any medical care, patient's files reported differently.Patients were probably not fully informed on their treatment during their visits or misinterpreted the meaning of the question.Therefore, it was decided to solely rely on the cross-check of

Table 1
Demographics and patient characteristics for each patient.Continuous variables are presented by the mean and standard deviation (SD) and binomial variables are presented in frequencies (freq), categorical variables with more than two categories are presented as frequencies and percentages.ATL, anterior temporal lobectomy; AH, amygdalohippocampectomy. BMI, body mass index; Nr, number; ASM, anti-seizure medication, NA, not applicable.

Table 2
Average imputed resource use and costs (euros) per time point of the sample (n = 30).Calculated by the friction cost methods.Diagnostic treatments includes MRI, PET, CT scans, blood work, etc.   the patient files to prevent incomplete data.Although this increased the accuracy of results for regional patients, some information on patients coming from other regions could still have been missed.Nonetheless, we believe this method is more accurate than solely relying on patients' answers.Although power was reached and study design was accounted for, other limitations include the limited sample size and absence of a comparator group, where the latter could offer the ability to perform cost-effectiveness analyses.Finally, baseline was set shortly before surgery and referred to the 3 months prior to surgery as our objective was to study the costs associated with RES.We have chosen this period because within these 3 months, patients go through the process to prepare for surgery including meeting the neurosurgeon, anesthesiologist, and epilepsy nurse.This means the costs of diagnostic evaluation including diagnostic MRI, video-EEG, and other diagnostic/presurgical evaluations were not taken into account.It would be interesting to analyze these costs in future projects; those should then be aimed at analyzing the cost-effectiveness of referral for diagnostic/presurgical evaluation, also taking into account the patients not identified as candidates for resective brain surgery.

Future research
Several steps should be taken to maintain momentum in this field of research.Firstly, by incorporating an additional component of caregivers' perspective.This could help create a better representation on QoL and productivity losses of caregivers, the effect epilepsy has on the patients' environment, and the added costs that accompany epilepsy as a disease to a broader extent.
Secondly, although already set in motion by our research team, this research should be extended to a larger sample size with a longer followup period, to incorporate for instance the effect of decreasing use of ASM and further analyze the trend in SC after RES.Thirdly, the complete diagnostic/presurgical evaluation and care-as-usual group as comparator should be included as well to create a more comprehensive picture of patient costs.Lastly, by incorporating expectations of patients before agreeing to surgery, future research could offer a clear patient profile of DRE patients who would benefit most from RES.When concentrating on the Dutch population, +/− 130 patients undergo RES annually, while 2100 patients are expected to be surgical candidates based on fact that 20-50% of DRE patients are eligible for RES [9].This confirms the findings from peer-reviewed studies that RES is currently underutilized [2][3][4].Underutilization may be explained by the lack of complete economic studies showing the direct and indirect economic benefits which are becoming increasingly important for societies.Since healthcare costs are rising considerably and are expected to rise more in the near future, the Dutch government is becoming increasingly critical concerning reimbursement of costly treatments [47].As DRE patients who are not surgically treated generally continue to be patients and actively consume health care, lifetime costs insinuate no decrease [21,24].Therefore, a study assessing the complete diagnostic/presurgical evaluation, SC, and disease-specific QoL in the same population could provide relevant information for policy makers and researchers [47].

Conclusion
This study presents evidence for impact on SC and disease-specific and generic health-related QoL in the first 12 months after RES in patients with DRE.As DRE patients are committed to lifelong reduced QoL, lifelong ASM use, and yearly SC of €20,751, future research on RES should encompass longer follow-up periods, larger sample size, and a cost-effectiveness analysis with a comparator.Overall, this study reports on the SC directly related to RES and could serve as a pilot project to offer an up-to-date model for larger cost-effectiveness studies in the future.

Fig. 1 .Fig. 2 .
Fig. 1.Total societal costs per time point, in red (increase in costs) and green (decrease in costs) boxes the differences in costs between time points are illustrated.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)