Reprinted from

Volume 10, Number { 1, February 2007

Vagus Nerve Stimulation Therapy for Pharmacoresistant Epilepsy: Effect on Health Care Utilization
Allan L. Bernstein, Howard Barkan, and Terry Hess

Also available on
SCIENCE@OIRECTe

www.sciencedirect.com

Available online at www.sciencedirect.com

~~;o~ ScienceDirect
ELSEVIER
Epilepsy & Behavior 10 (2007) 134-137

-d-"~

Epilepsy
Behavior
www.elsevier.comllocate/yebeh

&.

Vagus nerve stimulation therapy for pharmacoresistant Effect on health care utilization *
Allan L. Bernstein
a

epilepsy:
b

a,b,*,

Howard Barkan ", Terry Hess

Department of Neurology, Kaiser Permanente Medical Center, 401 Bicentennial Way, Santa Rosa, CA, USA b Clinical Research Center of the North Bay, Santa Rosa, CA, USA Received 23 June 2006; revised 8 September 2006;1!ccepted 27 September 2006 Available online 3 November 2006

Abstract We retrospectively analyzed the effects of vagus nerve stimulation (VNS) therapy on utilization of medical services by 138 patients in a large staff-model health maintenance organization. We compared average quarterly rates for 12 months before device implantation with quarterly rates during 48 months of follow-up. Wilcoxon matched-pairs signed-ranks tests comparing pre-VNS with post-VNS utilization rates showed statistically significant reductions in numbers of emergency department visits, hospitalizations, and hospital lengths of stay, beginning with the first quarter after implantation (P < 0.05 for all post-implantation quarters for these three aspects). For the first two quarters after implantation, the average number of outpatient visits was significantly greater than the pre-implant quarterly average (quarter 1: P < 0.0001; quarter 2: P = 0.0067), but the average was 12.2% less by the fourth quarter of the first year after implantation and significantly less beginning with the first quarter of the second year (P = 0.0017) and continuing through the end of the study (P < 0.0001 for all subsequent quarters). A comparison of time spent on epilepsy-related tasks during the year before implantation with the year after implantation also revealed significant decreases in the average number of days on which patients could not work because of health-related concerns, from 3.. 67 to 1.04 days (P = 0.002, paired Student's t test) and the average time spent caring for health problems, from 352.6 to 136.1 minutes per week (P < 0.001). VNS therapy had a positive effect on both the utilization of health care services and the time spent on epilepsy-related tasks for these patients with pharmacoresistant epilepsy. . . 2006 Elsevier Inc. All rights reserved.
Keywords: Vagus nerve stimulation therapy; Epilepsy; Utilization; Pharmacoresistant

1. Introduction We studied the effect of vagus nerve stimulation (VNS) therapy on the utilization of medical services by a consecutive surgical series in Kaiser Permanente, a large staff-model health maintenance organization (HMO) in Northern California. To our knowledge, this endeavor is the first utilization study of VNS therapy conducted in the United States, although several cost studies from Europe, the United Kingdom, and Canada have been published [1-8]. .

2. Methods We retrospectively calculated the utilization of all outpatient visits, emergency department visits, hospital lengths of stay, and number of hospital admissions, regardless of whether the utilization was related to epilepsy, for patients implanted with the VNS therapy device (Cyberonics, Inc; Houston, TX, USA). We compared the average quarterly utilization for the 12 months before implantation (pre-VNS) with quarterly rates for the 48 months afterward (post-VNS). Utilization during the month of implantation was not included in the analysis. We performed separate analyses of the frequency of outpatient department visits, frequency of emergency department visits, hospital lengths of stay, and number of hospital admissions. Treating physicians prescribed VNS according to clinical judgment. All implantations were performed at one of two medical centers, and patients were returned to their treating neurologists. A separate analysis of time spent on health-related activities compared data that were collected during telephone interviews with the patients or their earegivers. The interviewer asked these persons how much time

'" Cyberonics, Inc supported this research through an unrestricted grant. Corresponding author. Fax: + I 707 571 4858. E-mail address: Allan.L.Bemstein@kp.org (A.L. Bernstein). 1525-5050/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi: I0.1016/j.yebeh.2006.09 .014

A.L Bernstein et al. / Epilepsy & Behaoior had been spent performing health-related activities, as well as how many days they had not been able to work because of health-related concerns, during the year before and the year after implantation with the VNS device. Patients were not screened for any particular demographic variable. Data were available for 140 patients who had been implanted with the VNS device, but 2 patients who were considered outliers were excluded from the analysis. One of the outliers incurred high utilization related to a catastrophic motor vehicle crash, and the other outlier had an extended hospitalization unrelated to epilepsy. A third patient incurred high utilization related to epilepsy and comorbid conditions. To illustrate the effect of high utilization by a single patient, analyses were computed and graphed with (/1 = 138) and without (/1 = 137) the utilization of this third patient. In addition to compiling summary statistics, we used the Wilcoxon matched-pairs signed-ranks test to compare pre-VNS (the quarterly average during the year before implantation with the VNS device) outpatient visits, emergency room visits, hospital lengths of stay, and number of hospital admissions with quarterly utilization of the same items for the 48 months after implantation of the VNS device. Differences from baseline (pre-VNS) were used in the statistical significance tests; however, results are presented graphically as percentage change in mean values from baseline to provide a more intuitive appreciation of the time trends. We used a paired Student I test to compare the amount of time spent performing health care-related activities and days on which patients could not work because of health-related reasons during the year before the patient was implanted and the year afterward. A P value ";0.05 was con" sidered significant. This study is an initial exploration using a surgical series design; therefore, statistical analyses were not corrected for multiple comparisons. The Institutional Review Board of the Kaiser Foundation Research Institute approved this study.

la

(2007) 134-137

135

the first quarter of the second year after implantation, the average number of quarterly outpatient visits was significantly less than pre-VNS, with a mean of2.9 3.1 (median 2.0, range 0-20), a decrease of 16.3% (P = 0.0017). For the remaining quarters of the study, the average number of outpatient visits per quarter was significantly less than that pre-VNS (P < 0.0001) (Fig. 1). During each of the 16 quarters after implantation, the rate of emergency department visits (Fig. 2), hospital lengths of stay (Fig. 3), and number of hospital admissions (Fig. 4) per quarter were significantly less than those preVNS. To illustrate the influence of a single patient with greater utilization than the other patients, each graph shows utilization by 138 patients, which includes the . "patient with greater utilization, and by 137 patients, which excludes the patient.
Outpatient Visits 150

'" z :::f .

100 50

s:

~ ., <: '" .

-,

3. Results 3.1. Demographics Of the 140 patients who were implanted, 73 (52.1%) were females and 67 (47.9%) were males. Average age at implantation was 30.2 16.9 years (range 3-81 years). 3.2. Utilization During the year before implantation, the number of outpatient visits per patient per quarter (n = 138) ranged from o to 20, with the mean number of visits per patient per quarter increasing from 3.0 3.4 (median 2.0, range 0-18) during the first quarter to 4.3 3.5 (median 3.5, range 0-20) during the fourth quarter, just before implantation of the VNS device (see Appendix A). For the first two quarters after implantation, the outpatient visit rate was significantly greater than pre-VNS, with a mean of 7.2 2.9 (median 7.0, range 2.0-20.0), an increase of 103.7% over pre-VNS (P < 0.0001, Wilcoxon) visits during the first quarter; a mean of 4.2 3.2 (median 3.0, range 018), an increase of 19.4% over pre-VNS (P = 0.0067) during the second quarter; a mean of 3.6 2.8 (median 3.0, range 0-12), an increase of 3.2% over pre-VNS (P = 0.7014, not statistically significant) during the third quarter. During the fourth quarter after implantation, the outpatient visit rate was less than that pre-VNS, with a mean of 3.1 2.9 (median 2.0, range 0-14), a decrease of 12.2% (P = 0.0516, not statistically significant). Beginning with

:. -100 -150 6
Quarter

;: e

----.... ~ ~

-50

~ ~

----12 13 14 15 16

10

11

Post-VNS

l-n=138

__ n:137I

Fig. I. Percentage change in mean number of outpatient visits per quarter from pre-VNS through 4 years ofVNS therapy. To illustrate the effect of a single patient with high utilization, the line denoting n = 138 includes all patients in the analysis, and the line denoting n = 137 excludes the patient with high utilization.

Emergency 0
en

Room Visits

:::-

20 -40 -60 -SO

0.

s: o
G

. '" ~
E

<c

.A
~,,--. -

;:

~ ~
10
Qua.rter PostNNS

:. -100

120 234 11 12 13 14 15 16

Fig. 2. Percentage change in mean number of emergency department visits per quarter from pre- VNS through 4 years of VNS therapy. To illustrate the effect of a single patient with high utilization, the line denoting n = 138 includes all patients included in the analysis, and the line denoting n = 137 excludes the patient with high utilization.

136

A.L. Bernstein et al. / Epilepsy & Behaoior la (2007) 134-137

Hospital Length of Stay

00
60

'" ~
et

40 20

i',

i
o

0
20 -40

I
/
/
2

,
\
.,

o
c,

r-: ">< A.

/\

\
\ \/

., fA ij -, \

~ .so
00
-100

\''i
5 6 7

Y
9 10 11

~.
~.

12

13

14

15

16

Quarter

Post-VNS

Fig. 3. Percentage change in mean hospital lengths of stay per quarter from pre-VNS through 4 years of VNS therapy. To illustrate the effect of a single patient with high utilization, the line denoting n = 138 includes all patients in the analysis, and the line denoting n = 137 excludes the patient with high utilization.

Hospital Admits

o
-10

e c,
GO

'" ~
::

-20 -30
I

r--: ,I

\
\

0
iD

E '" " ~
!:

-40 -SO -60 -TO -80 90 -100

F'

...:::::-.,

GO c,

jf <, "\ /\ ~' .. 11 ",,\, V \ I ~ \ J V

9
Quarter

10

11

12

13

14

15

16

Post-VNS

Fig. 4. Percentage change in mean number of hospital admissions per quarter from pre-VNS through 4 years of VNS therapy. To illustrate the effect of a single patient with high utilization, the line denoting n = 138 includes all patients in the analysis, and the line denoting n = 137 excludes the patient with high utilization.

A comparison of time spent on epilepsy-related tasks during the year before implantation with the year after implantation also revealed significant decreases in the average number of days on which work was not possible because of health-related concerns, from 3.67 to 1.04 days (P = 0.002, paired Student t test), and the average time spent caring for health problems, from 352.6 to 136.1 minutes per week (P < 0.001).

4. Di cussion This retrospective study of VNS therapy documents statistically significant reductions in health care utilization, time spent on health-related activities, and time lost from work for reasons related to epilepsy for a group of clinically selected patients with pharmacoresistant epilepsy.

The increase in outpatient department services, which extended into the second quarter post- VNS, can most likely be attributed to the series of office visits required for the initiation of VNS therapy (usually about 2 weeks after implantation) and adjustment of device parameters (typically done incrementally as the patient's tolerance to stimulation increases) [9]. By the fifth quarter post- VNS, the number of outpatient visits per month had decreased to the point that it was significantly less than that preV S, and that difference remained significant through the remainder of the study. Compared with utilization during the year before implantation, results during the fourth quarter of Year 4 revealed impressive decreases in utilization of all four aspects measured: a 91% decrease in outpatient visits, a 99% <1ecrease in emergency department visits, a 67% decrease in hospital lengths of stay, and a 70% decrease in number of hospital admissions. Interestingly, no visits to the emergency department occurred during the final quarter of the study. During approximately the same period, the number of emergency department visits also decreased within the Northern California Kaiser system, although not as notably as in this study; the 2002 rate was 88.9% of the 1998 rate (data on file, Kaiser Permanente Healthcare System, Northern California). Previous studies of costs and health care utilization by patients receiving VNS therapy have described much smaller cohorts from Europe, the United Kingdom, and Canada, and provided data covering a much shorter duration [1-8]. Ben Menachem et al., whose cost-effectiveness study described intensive care, emergency room, and hospital ward costs of 43 patients over 18 months, estimated that the purchase price of the VNS Therapy System could be absorbed within 2 to 3 years [1]. In two reports of cost studies ofVNS therapy, Boon et al. list savings in direct costs of approximately $3000 per year with VNS [3,4]- Another study by Boon et al. compared three modalities for the treatment of pharrnacoresistant epilepsy-surgery, antiepileptic drugs, and VNS-and showed a cost savings of more than $2000 annually for the 25 patients treated with VNS and followed for an average of 26 months [2]. Majorie et al. compared costs 6 months pre-VNS with costs 6 months post-VS for 16 children with Lennox-Gastaut syndrome and reported a savings of about 2800 [6]. Forbes et al. estimated a savings of 745 per year with VNS, but this estimate was drawn from publications and not from actual experience at an epilepsy center [5]. Marrow et al. stated that the cost of VNS did not compare unfavorably with the newer antiepiJeptic drugs [7], and Upton et al. ranked VNS as comparing favorably with the cost of three antiepileptic drugs [8]. Unlike the previously reported studies that dealt with some utilization and costs associated with pharrnacoresistant epilepsy and VNS therapy, this study features quarterly changes in utilization over 4 years after implantation of the VNS therapy device in 138 patients. The applicability of data describing costs is subject to fluctuations in

A.L. Bernstein et al. / Epilepsy & Behauior 10 (2007) 134-137

137

monetary exchange rates and inflation, but utilization data can be applied to a multitude of health care settings among which costs and charges may vary considerably for the same procedure or item. Because the patients enrolled in the Northern California Kaiser system are representative of the socioeconomic makeup of the community except for the highest and lowest levels [10], investigators can be further assured of the suitability of these data. Limitations of this study include the retrospective design, the lack of a contemporaneous control group with pharmacoresistant epilepsy, and the possibility that the impending implantation could have contributed to the number of outpatient visits during the pre-implant year. On the other hand, this study has two major strengths: the well-documented quarterly changes in utilization of the largest number of patients over the longest period reported to date, and the ease with which these data can be applied to a variety of settings. The significantly reduced numbers of emergency department visits, outpatient visits, and hospitalizations, as well as the decreased lengths of stay in the hospital, are consistent with improved control of epilepsy. In addition, the significantly decreased time spent on epilepsy-related matters and days missed from work further reflect positive changes in the lives of both caregivers and persons with epilepsy. In summary, these results illustrate a distinctly positive influence of VNS therapy on the health care utilization of patients with pharmacoresistant epilepsy.
Acknowledgments

with the development of the article. The authors maintained complete control over the direction and content of this article.
Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at doi:lO.lOI6/j.yebeh.
006.09.014. References
[1) Ben-Menachem E, Hellstri:im K, Verstappen D. Analysis of direct hospital costs before and 18 months after treatment with vagus nerve stimulation therapy in. 43 patients. Neurol_.ogy 2002;59(SuppJ. 4):S44--7. (2) Boon P, D'Have M, Van Walleghem P, et aI. Direct medical costs of refractory epilepsy incurred by three different treatment modalities: a prospective assessment. Epilepsia 2002;43:96--102. (3) Boon P, Vonck K, D'Have M, O'Connor S, Vandekerckhove T, De Reuck J. Cost-benefit of vagus nerve stimulation for refractory epilepsy. Acta Neurol Belg 1999;99:275-80. (4) Boon P, Vonck K, Vandekerckhove T, et aJ. Vagus nerve stimulation for medically refractory epilepsy: efficacy and cost-benefit analysis. Acta Neurochir (Wien) 1999;141:447-52. discussion 453. (5) 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. (6) Majoie HJ, Berfelo MW, Aldenkamp AP, Evers SM, Kessels AG, Renier WO. Vagus nerve stimulation in children with therapy-r resistant epilepsy diagnosed as Lennox-Gastaut syndrome: clinical" results, neuropsychological effects, and cost-effectiveness. J Clin NeurophysioI2001;18:419-28. (7) Morrow JI, Bingham E, Craig JJ, Gray WJ. Vagal nerve stimulation in patients with refractory epilepsy: effect on seizure frequency, severity and quality of life. Seizure 2000;9:442-5. (8) Upton A. Vagal stimulation for intractable seizures. Adv Exp Med Bioi 2002;497:233-9. (9) Heck C, Helmers SL, DeGiorgio CM. Vagus nerve stimulation therapy, epilepsy, and device parameters: scientific basis and recommendations for use. Neurology 2002;59(SuppJ. 4):S31-7. (10) Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. Am J Public Health 1992;82:703-10.

Preparation of this article was supported by an unrestricted educational grant from Cyberonics, Inc., manufacturer of the VNS Therapy System, to Kaiser Permanente. Dr. Barkan received compensation from Cyberonics for performing preliminary data analyses. John AlIen, Ph.D., and Amara Jayewardene, M.S., employees of Cyberonics, performed the statistical analyses presented in this article. Susan E. Siefert, ELS, CBC, also of Cyberonics, assisted