Epilepsy & Behavior 63 (2016) 79–88

Contents lists available at ScienceDirect

Epilepsy & Behavior

journal homepage: www.elsevier.com/locate/yebeh

Cost-utility analysis of competing treatment strategies for drug-resistant

epilepsy in children with Tuberous Sclerosis Complex
Aria Fallah a,b,⁎, Alexander G. Weil c, Shelly Wang d,e, Evan Lewis f, Christine B. Baca g,h, Gary W. Mathern a,b
a
Department of Neurosurgery, David Geffen School of Medicine at University of California Los Angeles, Los Angeles, CA, USA
b
Brain Research Institute, University of California Los Angeles, Los Angeles, CA, USA
c
Division of Pediatric Neurosurgery, Department of Surgery, University of Montreal, Montreal, QC, Canada
d
Division of Neurosurgery, Department of Surgery, University of Toronto, Toronto, ON, Canada
e
Department of Biostatistics and Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA
f
Division of Pediatric Neurology, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada
g
Department of Neurology, David Geffen School of Medicine at University of California Los Angeles, Los Angeles, CA, USA
h
Department of Neurology, Veterans Administration Greater Los Angeles Healthcare System, Los Angeles, CA, USA

a r t i c l e i n f o a b s t r a c t

Article history: Background: The management of drug-resistant epilepsy in children with Tuberous Sclerosis Complex (TSC) is
Received 17 May 2016 challenging because of the multitude of treatment options, wide range of associated costs, and uncertainty of sei-
Revised 7 July 2016 zure outcomes. The most cost-effective approach for children whose epilepsy has failed to improve with first-line
Accepted 24 July 2016 medical therapy is uncertain.
Available online xxxx
Methods: A review of MEDLINE from 1990 to 2015 was conducted. A cost-utility analysis, from a third-party payer
perspective, was performed for children with drug-resistant epilepsy that had failed to improve with 2 antisei-
Keywords:
Epilepsy
zure drugs (ASDs) and that was amenable to resective epilepsy surgery, across a time-horizon of 5 years. Four
Seizures strategies were included: (1) resective epilepsy surgery, (2) vagus nerve stimulator (VNS) implantation, (3) ke-
Tuberous Sclerosis Complex togenic diet, and (4) addition of a third ASD (specifically, carbamazepine). The incremental cost per quality-
Antiseizure drugs adjusted life year (QALY) gained was analyzed.
Ketogenic diet Results: Given a willingness-to-pay (WTP) of $100,000 per QALY, the addition of a third ASD ($6600 for a gain of
Vagal nerve stimulator 4.14 QALYs) was the most cost-effective treatment strategy. In a secondary analysis, if the child whose epilepsy
Resective surgery had failed to improve with 3 ASDs, ketogenic diet, addition of a fourth ASD, and resective epilepsy surgery were
incrementally cost-effective treatment strategies. Vagus nerve stimulator implantation was more expensive yet
less effective than alternative strategies and should not be prioritized.
Conclusions: The addition of a third ASD is a universally cost-effective treatment option in the management of
children with drug-resistant epilepsy that has failed to improve with 2 ASDs. For children whose epilepsy has
failed to improve with 3 ASDs, the most cost-effective treatment depends on the health-care resources available
reflected by the WTP.
© 2016 Elsevier Inc. All rights reserved.

1. Introduction economic analysis is a set of formal methods to compare competing

Tuberous Sclerosis Complex (TSC) is an autosomal dominant genetic
multisystem disorder that is variably expressed with a prevalence of
1 in 10,000 and affecting approximately 50,000 individuals in the
United States and over 1 million individuals worldwide [1,2]. It is one
of the leading causes of genetic epilepsy, with seizures affecting almost
90% of children [3]. A third of these patients will be unable to achieve
seizure freedom with antiseizure drugs (ASDs) alone [4,5].
The extended period that pediatric patients will live with epilepsy
accrues to significant economic costs on the health-care system. An
⁎ Corresponding author at: Department of Neurosurgery, David Geffen School of
Medicine at UCLA, 300 Stein Plaza, Suite 525, Los Angeles, CA 90095-6901, USA. Fax: +1
310 794 5028.
E-mail address: afallah@mednet.ucla.edu (A. Fallah).
reatment strategies with respect to their resource use and their ex-
pected health outcomes. Jacoby et al. found that the cost of treatment
is proportional to the severity of epilepsy [6]. In addition, more than half
the total costs of epilepsy care are from the 15% of patients whose sei-
zures are the most refractory to medications [6,7]. Furthermore,
health-care costs for patients whose seizures are drug-resistant are
eightfold higher than for those with controlled epilepsy [6].
Economic analysis in populations with pediatric epilepsy is rare, and
the results from adults cannot be extrapolated to children for several
important reasons, including the following: 1) varying underlying
pathological substrates, 2) more common extratemporal pathology
yielding lower seizure freedom outcomes, and 3) longer life expectancy
increasing the impact of intervention on long-term costs and health
utility. In a cohort of 30 pediatric patients with various underlying epi-
lepsy substrates, Widjaja et al. demonstrated that surgical treatment

http://dx.doi.org/10.1016/j.yebeh.2016.07.034
1525-5050/© 2016 Elsevier Inc. All rights reserved.
80 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88
in pediatric epilepsy was a cost-effective treatment strategy compared
with continuing medical therapy [8].
The most effective and cost-effective treatment strategies for seizure
control in children with DRE secondary to TSC must be established. This
population is distinctively challenging to treat, because of a multitude
of management options with significant clinical equipoise. The com-
plexity of DRE treatment in children with TSC is compounded by the
relatively modest surgical outcomes [9–11]. A cost-utility analysis
(CUA) is warranted to evaluate the current interventions and to
provide guidance to clinicians in treatment decision-making. Utilities
capture the preferences individuals place for a particular health state,
and the adjusted life expectancy (denominator in a CUA) is quality-
adjusted life years (QALYs). Given the uncertainty on how best to select
treatment strategies, the results of this study will assist clinicians in
everyday decision-making and decision-makers in answering critical
health policy questions. Some examples of health policy questions
that can be answered include maximizing the benefits of health-care
spending, containing costs, and providing bargaining power with sup-
pliers of health-care products. Health-care resource-poor countries
may decide not to fund some treatment strategies altogether given its
high cost and low efficacy in consideration of their willingness to pay.
In this study, we compared 4 competing treatment strategies for DRE,
as defined by the International League Against Epilepsy [12] in TSC:
resective epilepsy surgery, vagus nerve stimulator (VNS) implantation,
ketogenic diet treatment, and addition of a third ASD (specifically, carba-
mazepine). All these strategies are indicated treatment options of DRE in
the pediatric population with TSC. Resective surgery is indicated for focal
epilepsy, ideally in noneloquent cortex. Vagus nerve stimulator (VNS) im-
plantation, approved for adolescents older than 12 years of age, is an ad-
junct palliative therapy to decrease seizure frequency. Ketogenic diet, a
high fat, adequate protein, and low carbohydrate diet that leads to ketosis,
is approved for children as a palliative approach. Additionally, in a second-
ary analysis, we investigated the role of mammalian target of rapamycin
(mTOR) inhibitor as a potential treatment strategy for DRE in children
with TSC. Although not clinically used for the treatment of DRE, mTOR in-
hibitors decrease proteins of the gene products regulated by TSC and are
approved for the treatment of subependymal giant cell astrocytomas
(SEGA) in patients with TSC. Preliminary data in Phase II trials demon-
strated SEGA regression and modest improvement in seizures [13].
1.1. Objectives and hypothesis
Our primary objective was to evaluate the cost-utility of 4 competing
antiseizure treatment strategies for children with focal DRE secondary
to TSC that is amenable to surgery. Our secondary objective was to
evaluate the cost-utility of treatment strategies for the same population
of children, if a third ASD failed. In addition to ketogenic diet, VNS inser-
tion, and surgical resection, our secondary analysis includes the addition
of a fourth ASD and treatment with mTOR inhibitor. We sought to deter-
mine whether and under what circumstances one treatment strategy
would be more cost-effective than another. We hypothesized that
resective surgical treatment would be the costliest treatment initially
but, in a 5-year timespan, would have the greatest health-utility com-
pared with the alternative treatments.
2. Methods
2.1. Model overview
Using decision analysis software (TreeAge Software, Inc.,
Williamstown, Massachusetts, USA), our primary analysis evaluated a
hypothetical cohort of children, under 18 years of age and being treated
at a tertiary care hospital, with focal DRE secondary to TSC that is ame-
nable to resective surgery. Patients entered the model having had sei-
zures that did not improve from treatment with 2 first-line ASDs
(valproic acid and levetiracetam) and receiving 1 of 4 competing
treatment strategies for epilepsy: 1) addition of a third ASD (carbamaze-
pine), 2) ketogenic diet, 3) vagus nerve stimulator (VNS) implantation, or
4) resective epilepsy surgery. The model included children whose epilep-
sy had failed to improve with valproic acid and levetiracetam, as these are
first-line therapies for epilepsy associated with TSC because of high effica-
cy and tolerable side effect profiles. Our secondary analysis evaluated the
same cohort of children with seizures refractory to 3 first-line ASDs; the
analysis additionally included a fourth ASD (clobazam) and a fifth treat-
ment: mammalian target of rapamycin (mTOR) inhibitor. We followed
the hypothetical cohort over a 5-year time horizon following the initia-
tion of treatment. We did not perform our analysis over a longer time pe-
riod given the greater uncertainties associated with our key parameter
estimates. The decision tree structure is described in detail in Appendix
1. The study was performed in accordance with the guidelines
established by the US Public Health Service [14].
2.2. Model assumptions
For the reference case analysis, several assumptions regarding the
treatment strategies, outcomes, and adherence were required. For
model simplicity, we assumed full medical adherence, no adverse
events from medical or interventional procedures, and no major com-
plications or death. Furthermore, despite significant pharmacotherapy
clinical equipoise, we assume that the first-line ASDs are valproic
acid and levetiracetam, the third ASD is carbamazepine, and the fourth
ASD is clobazam. This decision was informed by a study that described
the most commonly used ASDs, excluding vigabatrin as it is commonly
utilized for infantile spasms, in the treatment of epilepsy in children
with TSC [15]. These assumptions with accompanying rationales are
presented in Appendix 2. Some of these assumptions, as indicated in
Table 6, were tested through sensitivity analysis.
2.3. Clinical probability estimates
Outcome probability estimates were retrieved using a review of the
medical literature using MEDLINE that aimed to identify relevant publi-
cations between January 2000 and September 2015 in any language. We
used keyword searches and reviewed the bibliographies of relevant ar-
ticles to identify additional relevant articles. We utilized meta-analyses
preferentially when available and, alternatively, large observational co-
hort studies to inform our parameter estimates. The key parameters
used in the decision model and their respective citations are provided
in Table 2.
2.4. Outcomes
In this CUA, our only outcome was QALYs since this is a measure
that can be used to compare different interventions in medicine.
Quality-adjusted life years (QALYs) were calculated using health utilities
(corresponding to Engel classification at 1-year following the initiation of
therapy) multiplied over a 5-year time horizon [16]. The basic event
pathway created for this study assumes 4 possible outcomes following
each treatment (i.e., Engel classes I, II, III, and IV outcomes) with the ex-
ception of the addition of a third ASD that leads to only 2 possible out-
comes (i.e., seizure freedom and continuation of seizures). Our analysis
reports the incremental cost-utility among the treatment strategies.
2.5. Utilities
Utility estimates were derived using the Cost Effectiveness Analysis
Registry from the Tufts Medical Center website [17]. Table 3 contains
the specific health utility estimates that were used.
 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88 81

2.6. Cost estimates Table 2

Probabilities of outcome (Engel classification) following various treatment strategies for
drug-resistant epilepsy in TSC.
The analysis was conducted using the perspective of a third-party
payer and incorporated the direct health-care costs of therapy, subse- Outcome Probability Range in Range test in Reference
quent hospitalizations, and ASD treatment. Historic costs were adjust- literature sensitivity analysis

ed to present value while future costs were discounted at 3% per year Resective surgery
[18]. Discounting accounts for people's preferences for, in general, Engel class I 0.55 0.50–0.65 0.45–0.70 [9–11,35,36]
Engel class II 0.13 0.13–0.18 0.08–0.23 [9–11,35,36]
things in the immediate future than those available in the distant fu-
Engel class III 0.15 0.12–0.25 0.07–0.30 [9–11,35,36]
ture. All costs were converted to US Dollars and adjusted to 2016 to
obtain the net present value. Willingness-to-pay (WTP) was set at VNS implantation
Engel class I 0.12 0.05–0.19 0.00–0.24 [37–41]
$100,000 per gain in QALY [19]. We obtained costs for physician ser-
Engel class II 0.11 0.08–0.31 0.03–0.36 [37–39,41]
vices and procedures through a combination of a literature review of Engel class III 0.42 0.13–0.64 0.08–0.71 [37–39,41]
MEDLINE, 2016 American Medical Association Current Procedural Ter-
Ketogenic diet
minology codebook, ICD-9-CM codes, and Medical Expenditure Panel
Engel class I 0.00 0.00–0.11 0.00–0.16 [42–45]
Survey [20], Agency for Healthcare Research and Quality Website [21]. Engel class II 0.07 0.07–0.35 0.02–0.40 [42–45]
Microcosting methods were only used for outpatient ASD costs. Engel class III 0.32 0.10–0.32 0.05–0.37 [42,44,46]
To estimate the lowest costs, drug unit costs were retrieved from
mTOR inhibitor
a discounted online website [22]. Microcosting is useful when the Engel class I 0.20 0.00–0.20 0.00–0.25 [47–49]
analysis is centered on the way a resource is delivered or when a partic- Engel class II 0.35 0.00–0.35 0.00–0.40 [47–49]
ular cost is not available in the medical literature or electronic data. The Engel class III 0.05 0.05–0.57 0.00–0.62 [47–49]
major disadvantages of microcosting are that work is increased, costs Third ASD (carbamazepine)
can be missed, and the costs incorporated often lack external validity. Seizure-free 0.06 N/A 0.01–0.11 [15]

2.7. Sensitivity analysis

Key data inputs entered into the decision tree may be inherently biased
or inaccurate. We aimed to obtain a best estimate, low estimate, and high
estimate based on our data search. To test the influence of all variables on
the model results, we performed multiple exploratory one-way sensitivity
analyses (“tornado analysis”) over a wide range of extreme but plausible
values for costs, probabilities, and utilities (Tables 1–3) [23]. This analysis
yielded a visual representation of the range of expected values for the de-
cision tree; we rank ordered the variables and selected all critical variables
that appeared to have a substantial impact on the expected value. We per-
formed a Monte Carlo simulation, assuming that all critical variables follow
a triangular distribution with base-case, minimum and maximum values
from Table 1 [24]. We generated an incremental cost-utility plot. We also
created a WTP curve to identify critical thresholds where one treatment
strategy is preferable over another.
2.8. Post hoc analysis
Sensitivity analysis was performed to answer two secondary policy
questions: What should the cost of VNS implantation or mTOR inhibitor,
respectively, be for it to be a cost-effective treatment strategy?
3. Results
3.1. Base-case results
3.1.1. Four treatment strategies for children with TSC whose seizures had
failed to improve with 2 ASDs
The CUA is displayed in Fig. 1. Treatment with a third ASD (specifi-
cally, carbamazepine) is the least costly (net expected cost of $6568
over 5 years) while treatment with resective surgery is the most costly
(net expected cost of $73,383 over 5 years) (Table 4). The least effective
treatment is ketogenic diet (net expected utility is 3.60 from a total of
5 QALYs) while the most effective treatment is resective surgery (net
expected utility of 4.38 from a total of 5 QALYs) (Table 2). With a WTP
of $100,000 per QALY, the most cost-effective treatment option is the
addition of a third ASD ($6568 for 4.14 QALYs). Resective surgery
($73,383 for 4.38 QALYs) is incrementally more cost-effective but has
an incremental cost-utility ratio (ICUR) of $285,068 per additional
QALY, which exceeds the WTP of the model. The treatment strategies
of ketogenic diet and VNS implantation are dominated (i.e., costlier
and less effective than alternative treatment strategies and, therefore,
not cost-effective), as displayed in Table 4.

Table 1
Base-case estimates of costs, updated to 2016 US Dollars.

Parameter Best base-case Range in literature Range tested in Reference

estimate sensitivity analysis

Cost of levetiracetam $96.00 $72.72–$390.31 $69.08–$409.83 [22,25,26]

Cost of valproic acid $645.49 $259.82–$647.27 $246.83–$679.63 [22,25,26]
Cost of resective surgery $46,778.00 $24,449.00–$49,871.51 $23,226.55–$52,365.09 [17,27,28]
Evaluation cost — resective surgery $12,355.39 $9982.31–$14,728.46 $9483.19–$15,464.89 [17]
Cost of VNS insertion $17,938.31 $8295.87–$17,938.31 $7881.08–$18,835.22 [29–31]
Cost of VNS battery replacement $7994.25 N/A $7594.54–$8393.96 [31]
Evaluation cost — no resective surgery $8135.00 $6287.69–$9982.31 $5973.31–10,481.43 [17]
Follow-up costs following surgical treatment — first 2 years $4783.85 N/A $4544.66–$5023.04 [8]
Cost of mTOR $134,436.00 $150,249.16–$152,821.24 $142,736.70–$160,462.30 [22]
Cost of ketogenic diet — initiation $4824.43 N/A $4583.21–$5065.65 [30]
Cost of ketogenic diet — ongoing $2737.50 N/A $2600.63–$2874.38 [30]
Cost of third ASD (carbamazepine) $52.00 $52.00–$1662.24 $49.40–$1745.35 [22,25,26,32]
Cost of fourth ASD (clobazam) $9301.38 $9301.38–$9792.00 $8836.31–$10,281.60 [22,25,26]
Follow-up costs following medical treatment — first 2 years $3560.77 N/A $3382.73–$3738.81 [8]
Follow-up hospitalization costs following surgical treatment — after 2 years (seizure-free) $0.00 N/A – [17]
Follow-up hospitalization costs following surgical treatment — after 2 years (reduction
$593.08 N/A $563.43–$622.73 [17]
in seizures)
Follow-up hospitalization costs following surgical treatment — after 2 years (no response) $2280.00 N/A $2166.00–$2394.00 [17,33,34]
82 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88

Table 3
Health state utilities of treatment outcomes, with perfect health state utility of 1.0.

Health state Health state utility Range in literature Range test in sensitivity analysis Reference

Seizure-free state (nonsurgical treatment) 0.94 N/A 0.92–0.96 [50]

No change in seizure frequency (nonsurgical treatment) 0.84 N/A 0.82–0.86 [50]
Less than 50% reduction in seizure frequency (nonsurgical treatment) 0.82 N/A 0.80–0.84 [51]
Engel class I 0.96 N/A 0.94–0.98 [52,53]
Engel class II 0.91 N/A 0.89–0.93 [52,53]
Engel class III 0.79 N/A 0.77–0.81 [52,53]
Engel class IV 0.66 N/A 0.64–0.68 [52,53]

3.1.2. Five treatment strategies for children with TSC whose seizures had
failed to improve with 3 ASDs
The CUA of the 5 treatment strategies, including mTOR inhibitor, is
displayed in Fig. 2. Treatment with ketogenic diet is the least costly
(net expected cost of $16,227 over 5 years) while treatment with
mTOR inhibitor is the most costly (net expected cost of $646,046 over
5 years), as displayed in Table 5. The least effective treatment is keto-
genic diet (net expected utility is 3.60 from a total of 5 QALYs) while
the most effective treatment is resective surgery (net expected utility
of 4.38 from a total of 5 QALYs). With a WTP of $100,000 per QALY,
the most cost-effective treatment option is resective surgery ($77,675
for 4.38 QALYs). This treatment strategy is incrementally more cost-
effective than the addition of a fourth ASD ($50,862 for 4.11 QALYs)
and ketogenic diet treatment ($16,228 for 3.60 QALYs). The treatment
strategies of VNS implantation and mTOR inhibitor are both dominated
(Table 5).
3.2. Base-case sensitivity analyses
3.2.1. Four treatment strategies for children with TSC whose seizures had
failed to improve with 2 ASDs
Using a tornado diagram, we identified several variables tested
over a range of plausible values through sequential one-way sensitivity
analyses that had a profound effect on the expected value (Fig. 3).
However, the range of tested values for these variables did not
affect the dominance of treatment strategies. From these, we selected
the following variables for probabilistic sensitivity analysis: cost of
carbamazepine years 1 to 5, cost of valproic acid and levetiracetam
years 1 to 5, probability of a third ASD resulting in seizure freedom,
utility of seizure-free state, and utility of less than 50% improvement
in seizures.
3.2.2. Five treatment strategies for children with TSC whose seizures had
failed to improve with 3 ASDs
In a secondary analysis, we analyzed the hypothetical scenario of
children whose seizures had failed to improve with 3 ASDs, by modeling
the effects of a fourth ASD (clobazam) and the experimental medication
mTOR inhibitor everolimus. In this scenario, we selected the following
variables from the tornado diagram for probabilistic sensitivity analysis,
each having a profound impact on expected value: cost of resective sur-
gery year 1; probability of resective surgery resulting in Engel classes I,
II, and III outcome; probability of ketogenic diet resulting in Engel clas-
ses I and II outcome; probability of VNS implantation resulting in Engel
class II outcome; cost of valproic acid and levetiracetam years 1–5; cost
of carbamazepine years 1–5, cost of hospitalization following surgery
year 1; and utility of Engel classes I, II, III, and IV (Fig. 4). From these,
one-way sensitivity analyses found several important thresholds that
identified a dominant treatment strategy (Table 6).

Fig. 1. Results of cost utility analysis for 4 treatment strategies in drug-resistant epilepsy in TSC after failure of 2 ASDs.
 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88 83

Table 4
Cost-utility league table ranking 4 treatment strategies in children with drug-resistant epilepsy secondary to TSC who have failed 2 ASDs (referencing common baseline).

Treatment strategy 5-Year total cost Incremental costa 5-Year total utility (QALY) Incremental utility (QALY)a Incremental cost/utility ratio (ICUR)b

Third ASD (carbamazepine) $6568.49 – 4.14 – –

Ketogenic diet $13,458.85 $6890.36 3.60 −0.54 Dominatedc
VNS implantation $50,742.96 $44,174.47 3.89 −0.25 Dominatedc
Resective surgery $73,383.93 $66,815.44 4.38 0.25 $268,335.11/QALY
a
Incremental cost and utility represent the difference between the strategy and the next best nondominated strategy.
b
Incremental cost/utility ratio represent the difference in cost divided by the difference in quality-adjusted life years for each strategy compared with the next best nondominated
strategy. The cost and QALY values are rounded in the calculations.
c
Strategies that are “dominated” are more costly and less effective than alternative treatment strategies, and, therefore, not cost-effective.

3.3. Monte Carlo analyses
3.3.1. Four treatment strategies for children with TSC whose seizures had
failed to improve with 2 ASDs
Using a triangular distribution and the same range for the critical
values identified in the one-way sensitivity analysis, we performed
1000 sampling points in a Monte Carlo simulation. No critical thresholds
were identified as we varied the cost and probabilities of outcome sug-
gesting that our model is robust and the addition of a third ASD is al-
ways the dominant (i.e., most cost-effective) treatment strategy with
a WTP of $100,000 per QALY over the course of 5 years.
A cost-effectiveness acceptability curve shows the probability that a
treatment strategy is cost-effective compared with the alternatives for a
range of maximum values that a decision-maker might be willing to
pay for a particular unit change in outcome. We found a threshold of
$266,000 for WTP over which resective surgery becomes more cost-
effective than the addition of a third ASD (Fig. 5). The simulation
shows that, when WTP is less than $184,000 per additional QALY, the
addition of a third ASD is the most cost-effective treatment strategy
every time. In contrast when the WTP is greater than $420,000 per ad-
ditional QALY, resective surgery is the most cost-effective treatment
strategy every time. As the WTP increases between $184,000 and
$420,000 per additional QALY, resective surgery gradually becomes
the more cost-effective option.
3.3.2. Five treatment strategies for children with TSC whose seizures had
failed to improve with 3 ASDs
We repeated the Monte Carlo simulation with the same parameters
now adding a fourth ASD (clobazam) and mTOR inhibitor as treatment
options in children whose seizures had failed to improve with 3 ASDs.
The cost-utility acceptability curve found a threshold for WTP of
$67,500 over which the addition of a fourth ASD becomes more cost-
effective than the ketogenic diet. A second threshold for WTP is of
$97,000 over which resective surgery becomes more cost-effective
than the addition of a fourth ASD (Fig. 6). The simulation shows that,
when WTP is between $67,500 and $97,000, the addition of a fourth
ASD is more commonly the cost-effective treatment strategy.
3.3.3. Post hoc analysis
There is no cost for VNS implantation or mTOR inhibitor that would
make these treatment strategies more cost-effective than the addition
of a third ASD (these strategies are more costly and less effective than
alternative strategies and, therefore, dominated). In children whose
seizures had failed to improve with 3 ASDs, the cost of mTOR inhibitor

Fig. 2. Results of cost utility analysis for 5 treatment strategies in drug-resistant epilepsy in TSC after failure of 3 ASDs.
84 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88

Table 5
Cost-utility league table ranking 5 treatment strategies in children with drug-resistant epilepsy secondary to TSC despite trials of 3 ASDs (referencing common baseline).

Treatment strategy 5-Year total cost Incremental costa 5-Year total utility (QALY) Incremental utility (QALY)a Incremental cost/utility ratio (ICUR)b

Ketogenic diet $16,227.58 – 3.60 –

Fourth ASD (clobazam) $50,861.83 $34,634.25 4.11 0.51 $67,579.03/QALY
VNS implantation $53,511.68 $37,284.10 3.89 0.30 Dominatedc
Resective surgery $77,675.46 $61,447.88 4.38 0.79 $77,831.39/QALY
mTOR inhibitor (everolimus) $646,045.93 $629,818.35 4.07 0.47 Dominatedc
a
Incremental cost and utility represent the difference between the strategy and the next best nondominated strategy.
b
Incremental cost/utility ratio represent the difference in cost divided by the difference in quality-adjusted life years for each strategy compared with the next best nondominated
strategy. The cost and QALY values are rounded in the calculations.
c
Strategies that are “dominated” are more costly and less effective than alternative treatment strategies, and, therefore, not cost-effective.

would have to be less than $800 per year to be a cost-effective treatment
strategy. Assuming that the cost of a battery replacement is one-quarter
the cost of a new VNS implantation, two-way sensitivity analysis
demonstrates that a cost combination of $7000 and $1750 or lower for
VNS implantation and battery replacement, respectively, will make
this treatment strategy cost-effective compared with the alternatives
in children whose seizures had failed to improve with 3 ASDs (Fig. 7).
4. Discussion
4.1. Key findings
We have 5 novel findings from this study that can be applied to the
management of children with DRE secondary to TSC: 1) the addition of a
third ASD (with similar cost as carbamazepine) is the most cost-
effective treatment strategy in children whose epilepsy has failed to im-
prove with two ASDs; 2) VNS implantation should be considered as a
second-line palliative treatment strategy in children who are not candi-
dates for resective epilepsy surgery and do not respond to or tolerate
the ketogenic diet; 3) mTOR inhibitor treatment for epilepsy alone is
not a cost-effective treatment strategy at the present time; 4) for the
child whose epilepsy has failed to improve with 3 ASDs and lives in a
resource-rich country (i.e., a WTP of $100,000 per additional QALY),
resective epilepsy surgery is the most cost-effective treatment strategy;
and 5) for the child whose epilepsy has failed to improve with 3 ASDs
and lives in a resource-limited country (i.e., a WTP of $50,000 per addi-
tional QALY) or is not a candidate for resective epilepsy surgery, the ad-
dition of the ketogenic diet is the most cost-effective treatment strategy.
4.1.1. Interpretation
Given the greater upfront costs of resective surgery, VNS implanta-
tion, and ketogenic diet, these treatment strategies are likely to become
more cost-effective over time. Therefore, extrapolating the findings of
this short-term study beyond the 5-year time horizon may result in mis-
guided conclusions. Furthermore, the sensitivity and threshold analyses
provide insight on the conditions of costs, outcome probabilities, and
health state utilities that influence decision analyses. In children whose
epilepsy has failed to improve with 3 ASDs, if the probability of Engel
class I or II outcome for ketogenic diet or Engel class II outcome for
VNS implantation was higher than estimated, these treatment options
could be more cost-effective than resective surgery. Until we are able
to obtain estimates with greater accuracy, this lowers the confidence
in our conclusion that resective epilepsy surgery is more cost-effective
than VNS implantation or ketogenic diet. Furthermore, if resective sur-
gery costs more than $60,020 or if the probabilities of achieving Engel
class I, II, or III outcome for resective surgery were lower than estimated,
then the addition of a fourth ASD (with similar cost and benefits as
clobazam) is the most cost-effective treatment option.

Fig. 3. Tornado analysis for net costs of 5 competing treatment strategies in TSC after failure of 3 ASDs. Central line is the base-case cost.
 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88 85

Fig. 4. Tornado analysis for net costs of 5 competing treatment strategies in TSC after failure of 3 ASDs. Central line is the base-case cost.

4.1.2. Strengths 4.1.3. Limitations

There are several strengths to this study. It presents a clear, logical
framework for organizing essential information for health-care
decision-making in this clinical context. We used the best available
evidence to inform the key model inputs. We tested these point esti-
mates through sensitivity analyses to identify critical thresholds that
determine the dominant treatment strategy. Through Monte Carlo
simulations, we obtained probabilistic results which not only identifies
what could happen but how likely each outcome is. Since each indi-
vidual carries a unique combination of probability estimates, adopting
a uniform set of values for all patients is fallacious and fails to capture
reality.
There are several important limitations to this study: 1) We had to
make several assumptions in this study (i.e., full medication adherence,
no major complications or death associated with the treatments, candi-
dates for resective epilepsy surgery, seizure outcomes do not change
over the time horizon of the study), and the degree to which these as-
sumptions are true reflects the confidence of the conclusions; 2) Base-
case probability estimates were commonly attained using retrospective
study data and may not be accurate, although some of the input param-
eter uncertainties were overcome using sensitivity analyses (Table 6);
3) Utilities were reflective of the seizure outcomes exclusively and did
not evaluate the impact or side effects of treatment itself;

Table 6
Results of 1-way sensitivity analyses for 5 treatment strategies (willingness-to-pay set at $100,000 per additional QALY).

Variable Base-case estimate Threshold Comment

Cost of resective surgery year 1 $46,778.00 $60,019.76 If cost exceeds this threshold, then the addition of a fourth ASD (clobazam) is more
cost-effective than resective surgery.
Probability of Engel class I seizure outcome 0.55 0.54 If probability falls below this threshold, then the addition of a fourth ASD (clobazam)
with resective surgery is more cost-effective than resective surgery.
Probability of Engel class II seizure outcome 0.13 0.12 If probability falls below this threshold, then the addition of a fourth ASD (clobazam)
with resective surgery is more cost-effective than resective surgery.
Probability of Engel class III seizure outcome 0.15 0.14 If probability falls below this threshold, then the addition of a fourth ASD (clobazam)
with resective surgery is more cost-effective than resective surgery.
Probability of Engel class II seizure outcome 0.11 0.31 If probability exceeds this threshold, then VNS implantation is more cost-effective than
with VNS implantation resective surgery
Probability of Engel class I seizure outcome 0.00 0.12 If probability exceeds this threshold, then ketogenic diet is more cost-effective than
with ketogenic diet treatment resective surgery
Probability of Engel class II seizure outcome 0.07 0.21 If probability exceeds this threshold, then ketogenic diet is more cost-effective than
with ketogenic diet treatment resective surgery
Utility of Engel class I seizure outcome 0.96 0.956 If utility falls below this threshold, then the addition of a fourth ASD (clobazam) is more
cost-effective than resective surgery
Utility of Engel class II seizure outcome 0.91 0.896 If utility falls below this threshold, then the addition of a fourth ASD (clobazam) is more
cost-effective than resective surgery.
Utility of Engel class III seizure outcome 0.79 0.778 If utility falls below this threshold, then the addition of a fourth ASD (clobazam) is more
cost-effective than resective surgery.
Utility of Engel class IV seizure outcome 0.66 0.650 If utility falls below this threshold, then the addition of a fourth ASD (clobazam) is more
cost-effective than resective surgery.
Utility of less than 50% reduction in 0.82 0.822 If probability exceeds this threshold, then the addition of a fourth ASD (clobazam) is
seizure frequency more cost-effective than resective surgery.
86 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88

Fig. 5. Monte-Carlo simulations demonstrating cost-utility acceptability curves for 4 competing treatment strategies in TSC after failure of 2 ASDs.

4) Neuropsychological, psychosocial, quality of life, and psychiatric out-
comes, which are other patient-important outcome measures, were not
evaluated; 5) These results have to be cautiously interpreted if making
treatment decisions in other countries, given the variability in costs and
health outcome values, which limits external generalizability; 6) Seizure
outcomes were limited to five years because of the lack of long-term ev-
idence, which limits conclusions beyond the study's time horizon.
4.1.4. Clinical implications
Cost-utility analysis is a methodology for evaluating the costs and
utility of health-care interventions. It acts as an aid in decision-making
but cannot be used in isolation, as it does not incorporate individual
center or health-care provider expertise and patient values and
preferences [54]. They do not reflect societal judgments on how
health resources should be allocated. This study suggests that in
general, for children with TSC whose epilepsy has failed to improve
with 2 ASDs, the addition of a third ASD is a universally cost-
effective treatment strategy. It is important to note, however, that
this may not apply to newer ASDs which are generally more expen-
sive. In resource-rich health-care settings with a greater WTP for
each additional QALY, resective epilepsy surgery is an incrementally
cost-effective treatment strategy following failure of 3 ASDs. In

Fig. 6. Monte-Carlo simulations demonstrating cost-utility acceptability curves for 5 competing treatment strategies in TSC after failure of 3 ASDs.
 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88
Fig. 7. Two-way cost-utility sensitivity analysis for the cost of VNS insertion and battery replacement.
children in resource-limited health-care settings or when resective
epilepsy surgery is not an option, the addition of a ketogenic diet is
the most cost-effective treatment strategy. Clinicians should be cau-
tious in recommending first-line palliative treatment with a VNS im-
plantation in children who can tolerate the ketogenic diet.
Treatment with mTOR inhibitor for DRE independent of SEGA is
very costly and relatively ineffective.
4.1.5. Implications for decision-makers
In children whose epilepsy has failed to improve with 3 ASDs, mTOR
inhibitor should cost less than $6300 a year to be a cost-effective treat-
ment option. Lowering the cost of VNS implantation and battery re-
placement may make this treatment option cost-effective. Assuming
the cost of battery replacement is one-quarter the cost of VNS implanta-
tion, a combination of $1500 and $6000 or lower, respectively, would
make this treatment option cost-effective.
4.1.6. Research implications
To improve upon the shortcomings of this decision analysis and CUA,
long-term prognostic studies of each treatment option are warranted in
homogenous cohorts of children with TSC and medically intractable
epilepsy. This can be facilitated with multi-institutional patient regis-
tries given the infeasibility of clinical trials for a rare disease. In addition,
capturing accurate cost data in patient registries would enhance the
precision of our estimates. Further research is required to increase the
accuracy of the range of acceptable values for the outcome states, health
utilities, and costs.
4.1.7. Future directions
Future cost-utility analyses in children with DRE should be per-
formed using Markov modeling and informed by actual patient
data using registries. Markov modeling would allow us to incorpo-
rate change in seizure control over time, which can occur over a lon-
ger follow-up period irrespective of whether the patient received
medical or surgical therapy. From a pharmacotherapy perspective,
mTOR inhibitors show significant promise in the treatment of
SEGAs and reduce seizure frequency in preliminary studies of adults.
As this medication is extremely expensive, our secondary analysis
demonstrated that it will not be a cost-effective treatment strategy
87
for DRE alone. Other novel or experimental interventions can also
be evaluated using our project. Responsive neurostimulation
(RNS®) is a surgically implanted device that uses electroencephalo-
graphic recordings to provide responsive neurostimulation when
detecting abnormal electrical events. It was recently approved by
the FDA as an adjunct palliative therapy for adults with partial-
onset epilepsy. Additionally, MR-guided laser induced thermal ther-
apy (MRgLITT) is a new curative technology in the treatment of focal
epilepsy. Compared with open resective surgery, it involves signifi-
cantly smaller incisions and fewer perioperative risks, with promis-
ing preliminary results in the pediatric population. In a resource-
rich health-care setting, novel strategies such as RNS and MRgLITT can
be considered. From a cost-utility perspective, these novel therapies
would be the dominant strategy for children whose epilepsy has failed
to improve with 3 ASDs, if they exhibit greater efficacy (improved health
states) than resective surgery and cost less than resective surgery
($77,675 over a 5-year period). However, if these therapies are less effi-
cacious and less costly than surgery, or more efficacious and costlier than
surgery, then they may be indicated, depending on the health-care re-
sources of that setting.
5. Conclusions
With the rising costs of health care, the importance of economic
efficiency is progressively relevant. Decision analysis and CUA are
increasingly required to not only guide the clinician in health-care deci-
sions but also to facilitate the efficient delivery of health care. These
analyses emphasize how costs and prudent resource allocation can po-
tentially play a large role in clinician decision-making. Our research
demonstrated that the addition of a third ASD or resective epilepsy sur-
gery is the most cost-effective treatment options in children with TSC
whose seizures have not improved with 2 ASDs for DRE and who are
candidates for resective epilepsy surgery. The decision to choose the lat-
ter treatment option preferentially depends on the health-care re-
sources available evidenced by a greater WTP to obtain an additional
QALY.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.yebeh.2016.07.034.
88 A. Fallah et al. / Epilepsy & Behavior 63 (2016) 79–88
Role of funding source
No funding was received for this study.
Acknowledgments
Gary W. Mathern was supported in part by the Davies/Randall
endowed chair for epilepsy research at UCLA and the RE children's
project.
Disclosures of conflicts of interest
None of the authors has any conflict of interest to disclose. None of
the authors have financial relationships relevant to this article.
References
[1] O'Callaghan FJ, Shiell AW, Osborne JP, Martyn CN. Prevalence of tuberous sclerosis
estimated by capture–recapture analysis. Lancet 1998;351:1490.
[2] Wiederholt WC, Gomez MR, Kurland LT. Incidence and prevalence of tuberous
sclerosis in Rochester, Minnesota, 1950 through 1982. Neurology 1985;35:600–3.
[3] Curatolo P, Bombardieri R, Jozwiak S. Tuberous sclerosis. Lancet 2008;372:657–68.
[4] Berg AT, Kelly MM. Defining intractability: comparisons among published defini-
tions. Epilepsia 2006;47:431–6.
[5] Chu-Shore CJ, Major P, Camposano S, Muzykewicz D, Thiele EA. The natural history
of epilepsy in tuberous sclerosis complex. Epilepsia 2010;51:1236–41.
[6] Jacoby A, Buck D, Baker G, McNamee P, Graham-Jones S, Chadwick D. Uptake and
costs of care for epilepsy: findings from a U.K. regional study. Epilepsia 1998;39:
776–86.
[7] Begley CE, Annegers JF, Lairson DR, Reynolds TF, Hauser WA. Cost of epilepsy in the
United States: a model based on incidence and prognosis. Epilepsia 1994;35:
1230–43.
[8] Widjaja E, Li B, Schinkel CD, Puchalski Ritchie L, Weaver J, Snead OC, et al. Cost-
effectiveness of pediatric epilepsy surgery compared to medical treatment in chil-
dren with intractable epilepsy. Epilepsy Res 2011;94:61–8.
[9] Fallah A, Guyatt GH, Snead 3rd OC, Ebrahim S, Ibrahim GM, Mansouri A, et al. Predic-
tors of seizure outcomes in children with tuberous sclerosis complex and intractable
epilepsy undergoing resective epilepsy surgery: an individual participant data meta-
analysis. PLoS One 2013;8, e53565.
[10] Fallah A, Rodgers SD, Weil AG, Vadera S, Mansouri A, Connolly MB, et al. Resective
epilepsy surgery for tuberous sclerosis in children: determining predictors of seizure
outcomes in a multicenter retrospective cohort study. Neurosurgery 2015;77:
517–24.
[11] Jansen FE, van Huffelen AC, Algra A, van Nieuwenhuizen O. Epilepsy surgery in tu-
berous sclerosis: a systematic review. Epilepsia 2007;48:1477–84.
[12] Kwan P, Arzimanoglou A, Berg AT, Brodie MJ, Allen Hauser W, Mathern G, et al.
Definition of drug resistant epilepsy: consensus proposal by the ad hoc task force
of the ILAE commission on therapeutic strategies. Epilepsia 2010;51:1069–77.
[13] Krueger DA, Care MM, Holland K, Agricola K, Tudor C, Mangeshkar P, et al. Everoli-
mus for subependymal giant-cell astrocytomas in tuberous sclerosis. N Engl J Med
2010;363:1801–11.
[14] Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russell LB. Recommendations of the
panel on cost-effectiveness in health and medicine. JAMA 1996;276:1253–8.
[15] Overwater IE, Bindels-de Heus K, Rietman AB, Ten Hoopen LW, Vergouwe Y, Moll
HA, et al. Epilepsy in children with tuberous sclerosis complex: chance of remission
and response to antiepileptic drugs. Epilepsia 2015;56:1239–45.
[16] Engel JJ. Surgical treatment of the epilepsies: outcome with respect to epileptic
seizures. Raven Press; 1987.
[17] Bowen JM, Snead OC, Chandra K, Blackhouse G, Goeree R. Epilepsy care in Ontario:
an economic analysis of increasing access to epilepsy surgery. Ont Health Technol
Assess Ser 2012;12:1–41.
[18] Gold MR, Siegel JE, Russell LB, Weinstein MG. Cost-effectiveness in health and
medicine. New York: Oxford University Pr; 1996.
[19] Neumann PJ, Cohen JT, Weinstein MC. Updating cost-effectiveness—the curious
resilience of the $50,000-per-QALY threshold. N Engl J Med 2014;371:796–7.
[20] Medical Expenditure Panel Survey. http://www.meps.ahrq.gov/mepsweb/data_
stats/MEPSnetHC.jspbhttps://urldefense.proofpoint.com/v2/url?u=http-3A_www.
meps.ahrq.gov_mepsweb_data-5Fstats_MEPSnetHC.jsp&d=CwMGaQ&c=
WknmpdNpvrlj2B5K1aWVqL1SOiF30547pqSuOmtwXTQ&r=
Ke2Ou4y6xYFa6e5FXO0Gpxk9gZqLbQACUg3BDgQOe7E&m=G-
aFPduu4sQrup1Zc8wNrhe7jA0s_1en5yJP5QZxc-A&s=
XAaixD954EucZOv4aRnqE18078Z3BcdodgboWof1aW8&e=N; 2015.
[21] Agency for Healthcare Research and Quality Website. http://hcupnet.ahrq.
govbhttps://urldefense.proofpoint.com/v2/url?u=http-3A__hcupnet.ahrq.gov_
&d=CwMGaQ&c=WknmpdNpvrlj2B5K1aWVqL1SOiF30547pqSuOmtwXTQ&r=
Ke2Ou4y6xYFa6e5FXO0Gpxk9gZqLbQACUg3BDgQOe7E&m=G-aFPduu4sQrup1Z
c8wNrhe7jA0s_1en5yJP5QZxc-A&s=ktVIIT_CJ-AIWaXGjQ6KXiVprFPcQvzsi7n1Fz-
fG8o&e=N; 2015.
[22] GoodRx. http://www.goodrx.com;/; 2015.
[23] Muennig P. Designing and conducting cost-effectiveness analyses in medicine and
health care. San Francisco: Jossey-Bass; 2002.
[24] U.S. Department of Labor BoLS. In; Consumer Price Index Calculator.
[25] Drug price search. https://www.rxpricequotes.com;/; 2015.
[26] Familywize. http://familywize.org/drug-price-lookup; 2015.
[27] Agency of Healthcare Research and Quality. http://www.ahrq.gov/research/publica-
tions/pubcomguide/pcguide1.html-refsbhttps://urldefense.proofpoint.com/v2/url?
u=http-3A__www.ahrq.gov_research_publications_pubcomguide_pcguide1.html-
2522-2520-255Cl-2520-2522refs&d=CwMGaQ&c=WknmpdNpvrlj2B5K1aWVq
L1SOiF30547pqSuOmtwXTQ&r=Ke2Ou4y6xYFa6e5FXO0Gpxk9gZqLbQACUg3BD
gQOe7E&m=G-aFPduu4sQrup1Zc8wNrhe7jA0s_1en5yJP5QZxc-A&s=rtC00DbA
u3A_aBbgQK0AeDNYV8NpZAdJ5OGvujM-EPc&e=N; 2015.
[28] Association AM. 2016 AMA CPT professional edition book. American Medical
Association; 2016.
[29] Aburahma SK, Alzoubi FQ, Hammouri HM, Masri A. Vagus nerve stimulation therapy
in a developing country: a long term follow up study and cost utility analysis. Sei-
zure 2015;25:167–72.
[30] de Kinderen RJ, Postulart D, Aldenkamp AP, Evers SM, Lambrechts DA, Louw AJ, et al.
Cost-effectiveness of the ketogenic diet and vagus nerve stimulation for the treat-
ment of children with intractable epilepsy. Epilepsy Res 2015;110:119–31.
[31] Forbes RB, Macdonald S, Eljamel S, Roberts RC. Cost-utility analysis of vagus
nerve stimulators for adults with medically refractory epilepsy. Seizure 2003;12:
249–56.
[32] HealthTrans. http://cashcard.lc.healthtrans.com/Pages/DrugSearch.aspx; 2015.
[33] Jette N, Quan H, Faris P, Dean S, Li B, Fong A, et al. Health resource use in epilepsy:
significant disparities by age, gender, and aboriginal status. Epilepsia 2008;49:
586–93.
[34] Langfitt JT. Cost-effectiveness of anterotemporal lobectomy in medically intractable
complex partial epilepsy. Epilepsia 1997;38:154–63.
[35] Krsek P, Jahodova A, Kyncl M, Kudr M, Komarek V, Jezdik P, et al. Predictors
of seizure-free outcome after epilepsy surgery for pediatric tuberous sclerosis
complex. Epilepsia 2013;54:1913–21.
[36] Teutonico F, Mai R, Devinsky O, Lo Russo G, Weiner HL, Borrelli P, et al. Epilepsy
surgery in tuberous sclerosis complex: early predictive elements and outcome.
Childs Nerv Syst 2008;24:1437–45.
[37] Elliott RE, Carlson C, Kalhorn SP, Moshel YA, Weiner HL, Devinsky O, et al. Refractory
epilepsy in tuberous sclerosis: vagus nerve stimulation with or without subsequent
resective surgery. Epilepsy Behav 2009;16:454–60.
[38] Elliott RE, Rodgers SD, Bassani L, Morsi A, Geller EB, Carlson C, et al. Vagus nerve
stimulation for children with treatment-resistant epilepsy: a consecutive series of
141 cases. J Neurosurg Pediatr 2011;7:491–500.
[39] Englot DJ, Chang EF, Auguste KI. Vagus nerve stimulation for epilepsy: a meta-
analysis of efficacy and predictors of response. J Neurosurg 2011;115:1248–55.
[40] Lagae L, Verstrepen A, Nada A, Van Loon J, Theys T, Ceulemans B, et al. Vagus nerve
stimulation in children with drug-resistant epilepsy: age at implantation and shorter
duration of epilepsy as predictors of better efficacy? Epileptic Disord 2015;17:
308–14.
[41] Major P, Thiele EA. Vagus nerve stimulation for intractable epilepsy in tuberous
sclerosis complex. Epilepsy Behav 2008;13:357–60.
[42] Caraballo R, Vaccarezza M, Cersosimo R, Rios V, Soraru A, Arroyo H, et al. Long-term
follow-up of the ketogenic diet for refractory epilepsy: multicenter Argentinean ex-
perience in 216 pediatric patients. Seizure 2011;20:640–5.
[43] Ketogenic Diet Study Group of the Subspecialty Group of Neurology tSoPCMA.
Prospective multicenter study on long-term ketogenic diet therapy for intractable
childhood epilepsy. Zhonghua Er Ke Za Zhi 2013;51:276–82.
[44] Neal EG, Chaffe H, Schwartz RH, Lawson MS, Edwards N, Fitzsimmons G, et al. The
ketogenic diet for the treatment of childhood epilepsy: a randomised controlled
trial. Lancet Neurol 2008;7:500–6.
[45] Thammongkol S, Vears DF, Bicknell-Royle J, Nation J, Draffin K, Stewart KG, et al.
Efficacy of the ketogenic diet: which epilepsies respond? Epilepsia 2012;53:
e55–9.
[46] Nangia S, Caraballo RH, Kang HC, Nordli DR, Scheffer IE. Is the ketogenic diet effec-
tive in specific epilepsy syndromes? Epilepsy Res 2012;100:252–7.
[47] Cardamone M, Flanagan D, Mowat D, Kennedy SE, Chopra M, Lawson JA. Mammalian
target of rapamycin inhibitors for intractable epilepsy and subependymal giant cell
astrocytomas in tuberous sclerosis complex. J Pediatr 2014;164:1195–200.
[48] Kotulska K, Chmielewski D, Borkowska J, Jurkiewicz E, Kuczynski D, Kmiec T, et al.
Long-term effect of everolimus on epilepsy and growth in children under 3 years
of age treated for subependymal giant cell astrocytoma associated with tuberous
sclerosis complex. Eur J Paediatr Neurol 2013;17:479–85.
[49] Krueger DA, Wilfong AA, Holland-Bouley K, Anderson AE, Agricola K, Tudor C, et al.
Everolimus treatment of refractory epilepsy in tuberous sclerosis complex. Ann
Neurol 2013;74:679–87.
[50] Hawkins N, Epstein D, Drummond M, Wilby J, Kainth A, Chadwick D, et al. Assessing
the cost-effectiveness of new pharmaceuticals in epilepsy in adults: the results of a
probabilistic decision model. Med Decis Making 2005;25:493–510.
[51] Selai C, Kaiser S, Trimble M, Price M. Evaluation of the relationship between epilepsy
severity and utility. Value Health 2002;5:512–3.
[52] Messori A, Trippoli S, Becagli P, Cincotta M, Labbate MG, Zaccara G. Adjunctive
lamotrigine therapy in patients with refractory seizures: a lifetime cost-utility anal-
ysis. Eur J Clin Pharmacol 1998;53:421–7.
[53] Suh GH, Lee SK. Economic evaluation of add-on levetiracetam for the treatment of
refractory partial epilepsy in Korea. Psychiatry Investig 2009;6:185–93.
[54] Fallah A. Moving beyond evidence-based medicine: incorporating patient values and
preferences. Epilepsy Behav 2015;53:209–10.