Disability and Rehabilitation, May 2006; 28(10): 645 – 651

RESEARCH PAPER

A preliminary economic evaluation of percutaneous neuromuscular

electrical stimulation in the treatment of hemiplegic shoulder pain

JANINE A. VAN TIL1,3, GERBERT J. RENZENBRINK1,2, KARIN GROOTHUIS1 &

MAARTEN J. IJZERMAN1,3
1
Roessingh Research and Development, Enschede, The Netherlands, 2Rehabilitation Center ‘Het Roessingh’, Enschede,
The Netherlands, and 3Biomedical Signals and Systems Group, University of Twente, Enschede, The Netherlands

Accepted July 2005

Abstract
Objective. The objective of this study was to compare the cost-effectiveness of various treatment modalities for hemiplegic
shoulder pain.
Design. A stage II economic evaluation.
Main outcome measures. Incremental cost effectiveness ratio of P-NMES, compared to slings and anti-inflammatory
injections.
Results. The incremental cost effectiveness ratio (ICER) of p-NMES, compared to anti-inflammatory injections is e6,061
(+3,285). The incremental cost of the first quality-adjusted life year after implantation of the P-NMES device compared to
anti-inflammatory injections is e33,007 (+5,434). This decreases to
e7,000 after 5 years, and to
e5,000 after 10 survival
years.
Conclusion. In this early evaluation, P-NMES seems to be cost-effective according to known guidelines. Treatment with
P-NMES is recommended for patients with chronic HSP.

Keywords: Electrical stimulation, hemiplegic shoulder pain, stroke, economic evaluation

multi-factorial, and interventions are therefore aimed

Introduction
Stroke is the fourth highest cause of death and the
most important cause of disability in The Nether-
lands. The incidence of stroke is approximately
30,000 cases a year and approximately 120,000
people live with the mental and/or physical con-
sequences of stroke. Hemiplegic shoulder pain
(HSP) is a common complication of stroke, which
negatively affects the use of the affected limb [1],
compromises functional recovery [2] and delays
discharge from hospital or a rehabilitation centre
[3,4]. In a review performed by Turner-Stokes et al.
[2] the incidence rates of HSP are reported to be
between 5 and 84% after the occurrence of stroke.
HSP thereby influences the quality of life (QoL) in a
substantial percentage of the stroke population.
There is still considerable controversy about the
etiology of HSP and, as a consequence also about
the appropriate treatment [5]. HSP is believed to be
at restoring normal alignment, the preservation or
restoration of both range of motion and muscle func-
tion, and pain reduction. Evidence-based decision-
making with regard to the choice of intervention for
HSP is limited. There is some recently published
evidence supporting the positive influence of elec-
trically stimulated contraction of the shoulder
muscles on external rotation of the shoulder [6],
subluxation, and pain sensation [7]. Other interven-
tions for HSP include careful positioning of the arm,
the use of a sling or lap-board, medication, and
physical therapy [21]. To a lesser extent treatment
for shoulder pain can include anti-inflammatory (AI)
injections, nerve blocks, icing, ultrasound, and
strapping [22 – 24]. Clinical trials were performed
to study the effect of AI injections [25,26] and
slings [9].
Although treatment with electrical stimulation is
considered to be most promising, the implementation

Correspondence: Janine van Til, PT, MSc. Roessingh Research and Development, PO Box 310, 7500 AH Enschede, The Netherlands. Tel: 31 53 4850739.
Fax: 31 53 4340849. E-mail: j.vantil@rrd.nl
ISSN 0963-8288 print/ISSN 1464-5165 online ª 2006 Taylor & Francis
DOI: 10.1080/09638280500277057
646 J. A. van Til et al.
of neuromuscular electrical stimulation (NMES) in
daily practice is hampered by the practical problems
that are involved in putting on and taking off the
necessary equipment and the pain that can be
induced by electrical surface stimulation [8]. A newly
developed system bypasses these problems by deli-
vering the stimulus percutaneously (P-NMES) via
electrodes implanted through the skin into the
muscle. In recently published stage II trials, P-NMES
seemed to be effective in the treatment of shoulder
subluxation and shoulder pain [9,10]. The percuta-
neous intramuscular approach has also been
reported to be less painful, more selective and easier
for daily use [7].
However, with the restraints that have been
imposed on the funding of the health care system
worldwide, costs play an important role in the
reimbursement of p-NMES. As explorative trials to
evaluate the effectiveness of P-NMES have recently
been performed [9 – 11], the purpose of the current
study is to obtain a preliminary indication of the cost-
effectiveness of the interventions available for
chronic hemiplegic shoulder pain, to help decide
whether a p-NMES is a feasible alternative compared
to more common interventions. The main outcome
measures in this study are the incremental costs and
effectiveness of P-NMES, compared to alternative
interventions, in the treatment of shoulder pain.
Methods
Study design
The primary audience for this study is the national
health insurance board investigating the effectivity of
P-NMES in the treatment of HSP. This study is a
stage II economic evaluation according to the criteria
of Sculpher et al. [12]. Sculpher et al. state that an
economic evaluation can be performed during each
phase of a clinical trial, with accompanying certainty
of results, depending on strength of evidence of the
clinical trial. This study will also provide an
indication of continuing research on the effectivity
of p-NMES. The methods section begins with a
presentation of the outcome measures for economic
evaluation, after which the short- and long-term
analysis of cost-effectivity data is discussed. The
requisite data gathering for both P-NMES and
alternative interventions are discussed subsequently,
and the Methods Section ends with a description of
applied statistics and sensitivity analysis.
Available interventions for HSP
The effectiveness of P-NMES was measured in two
small uncontrolled pilot studies [9,19], an uncon-
trolled trial [10] and a randomized controlled trial
[11]. These trials were conducted in our rehabilita-
tion center in The Netherlands and in the United
States. The methods, the inclusion and exclusion
criteria and the results of these studies are presented
in detail elsewhere [9,10], but were comparable. For
pain, a weighted average of success was calculated
over four studies at 6 months of follow-up. In the
Dutch trial, health-related QoL was measured with
the Short Form 36 (SF-36). Health utility was
calculated from the SF-36 data according to the
method developed by Brazier et al. [20]. The utility
gain of the treatment is the difference between the
pre-treatment health status utility and the 6-month
follow-up health status utility.
Three-month follow-up data were extracted from
the available studies concerning the effectiveness of
AI injections [25,26] and slings [9]. No literature
with regard to the effect of slings and IA injections on
QoL was identified. Therefore, possible increases in
QALY due to interventions with slings and IA
injections were estimated with the Brazier et al.
[20] algorithm.
Measures of effectiveness
The first measure of the effectiveness of a treatment
aimed at pain reduction is the number of patients
who experience diminished pain experience (per-
centage successful treatments). Subjective pain
experience is commonly scored by means of a visual
analogue scale (VAS) or a numerical rating scale
(NRS). A decrease of three points on a NRS has
been shown to represent a true change in pain
experience and therefore indicates a successful
intervention [13].
Most treatment modalities ultimately aim to
improve the QoL of patients. Successful treatment
is one that increases QoL in the remaining life-years
and/or lengthens the life of the patients. The
secondary measure of effectivity is therefore the
number of quality-adjusted life-years (QALY);
the expected survival years multiplied by the increase
in QoL of patients as a result of the intervention.
QoL is usually expressed as a utility, a number
between 0 (dead) and 1 (perfect health).
Pain experience has a maximal impact of 0.136 and
an average influence of 0.072 on the SF-36 overall
health status utility. The average change in health
status value of a successful treatment is multiplied by
the effectivity of treatment at the 6-month follow-up
and multiplied by the estimated duration of the effect
to obtain an estimate of QALY gain after an
intervention with slings of IA injections.
In the short-term analysis the cost-effectiveness
figures of treatment are calculated for a period of
one year. The incremental cost-effectiveness ratio
(ICER) is the ratio of the difference in cost and the
 Economic evaluation of P-NMES 647

difference in effectiveness between P-NMES and
alternative interventions.
The effectiveness of the treatment is the ratio of the
number of patients who receive successful treatment
and the total number of patients. It was assumed that
the effectiveness of treatment is stable up to one year.
Short-term analysis is performed for the effect of the
intervention on both pain and QALY.
The long-term consequences of the intervention
were calculated for the effect of the intervention on
QALY. Mortality is not expected to be influenced by
HSP, and therefore it has no influence on the cost-
effectiveness of the treatment. The effect of decreas-
ing effectiveness of each intervention is calculated up
to 10 years after intervention.
To obtain data on expected survival after stroke a
literature search was carried out with the search
terms ‘survival’ and ‘stroke’. After excluding early
case fatality an average mortality rate of 10% per year
is assumed for stroke patients [14]. Excluding early
case fatality is justified, because most severely
impaired patients will not be eligible for treatment,
due to more urgent co-morbidity [15 – 18]. The
effectiveness of the treatment is expected to decline
over time and as a result of a variety of age-related
adverse effects on health-related QoL. Three differ-
ent scenarios for a decline in effectiveness (from 0 to
50%) of all the interventions after a variable time-
span (3 months to 2 years after intervention) were
chosen after discussions with a panel of rehabilitation
physicians. Future health gains are discounted at 5%
in the long-term analysis.
Measurement of costs
This study assesses direct cost associated with
p-NMES, slings and IA injections. These mainly
include the time invested by physician, occupational
and physical therapist fees and associated costs in a
rehabilitation setting and costs of the device as the
major contributors to treatment costs. Non-direct
costs were not assessed in this analysis because of the
difficulty of obtaining these costs retrospectively.
Figures used are presented in Table I. Unit costs are
summed to get total treatment costs.
Statistics
Incremental cost-effectiveness ratio ðICERÞ
ðcosts pNMESÞ  ðcosts alternativeÞ
¼
ðeffectivity pNMESÞ  ðeffectivity alternativeÞ
Depending on the measure of effectivity, the ICER
reflects either the cost for each additional patient
treated successfully, or the cost per additional QALY
gained.
The sensitivity analysis
The effect of uncertainty on health gains after
treatment was modelled, and included in the
calculation of long-term cost-effectiveness. Probabil-
istic sensitivity analysis, based on Monte Carlo
simulation [27], was performed to study the simul-
taneous impact of uncertainty of the costs and
efficacy on the cost-effectiveness results for short-
and long-term economic evaluation. Both first-order
uncertainty (variability per patient in the costs,
number of units of therapy received and the survival
of patients over 10 years) and second-order uncer-
tainty (varying effectivity data, the decrement in
effectivity, QALY gain from treatment and survival
rate of patients) are accounted for [28,29]. First
the second-order parameters are selected from

Table I. Cost associated with a single unit of treatment. The sources for the pricing data are the Advisory Board for Health Care Pricing, the
manufacturer of materials and the medical compass 2003. The sum costs reflects the minimal costs of treatment options.

Costs of treatment P-NMES Sling Intra-articular Injections

Referral
GP e23.20 e23.20 e23.20
Consultations 1 1 3
Or
Rehabilitation setting e104.00 e104.00 e104.00
Time 15 minutes 15 minutes 15 minutes
Material e1,500 e105 e5.54
Amount 1 1 3
Treatment hour institution e104.00 e104.00 e104.00
Total 5 hour 0 30 minutes
Prescription e11.60 e11.60 e11.60
Total 0 0 1
Occupational therapist e12.50 e12.50 e12.50
Total 0 3 0
Sum costs e2041 e164 e98
648 J. A. van Til et al.

appropriate Beta-distributions and then, holding
these parameters constant for the patient group,
1000 patients were simulated and the mean cost and
effectiveness calculated for each treatment. This
simulation is performed 1000 times, resulting in
1000 values of the mean cost, mean effectiveness and
mean utilities. Both short- and long-term simulations
were performed in S-Plus 6.1. Estimated costs
and mean simulated costs may differ, due to some
skewed distribution assumptions used in the
simulation.
Results
Short-term analysis
Cost-effectiveness assessment. Using the three-point
decrease definition of success, the average percen-
tage of subjects treated successfully was 73% for the
P-NMES group, 28% for the sling intervention and
42% for intervention with the AI injections. The
costs of a 6-week course of P-NMES treatment were
estimated at e2,041, compared to e164 for a fitted
sling and e98 for three AI injections. The results of
calculated ICER are presented in the text, and the
results for both modeled and calculated costs are
presented in Table II. The incremental costs per
successfully treated patient with P-NMES are e6,268
when compared to IA injections, and e4,171 for P-
NMES compared to slings.
Cost QALY assessment. The average gain in the
Qol 24 weeks after intervention with P-NMES
treatment was 0.089 (8.9%) in the Dutch trial. The
increase in QALY as a result of treatment with
AI injections was estimated to be 0.0309 (3%) and,
as a result of treatment with slings it was 0.017 (2%).
The ICER for P-NMES when compared to IA
injections is estimated at e32,821 and when
P-NMES is compared to slings the ICER is
estimated at e27,085.
Sensitivity analysis. The estimated average cost of
p-NMES, slings and AI injections are reasonably
similar to the costs calculated in the first year analysis.
When varying both effectiveness and cost in the
manner described in the Methods section, results of
the short-term cost-effectiveness analysis seem to be
stable and proof against both changes in costs and
changes in effectiveness of the intervention, as is
shown in Figure 1. The variation in costs is minor,
and limited to a maximum of e25, which reflects
approximately 1 – 2% of additional costs. The range
between minimal and maximal effect (QALY gained)
is somewhat greater, and the maximum variation can
be as great as 25% of additional effect.
Long-term consequences. For the long-term analysis the
treatment effects were discounted, and varied ac-
cording to different scenarios. The results of the
comparison between the most positive scenarios
(minimal decrease in effectivity) and the most
negative scenario (decrease in effectivity of 50%)
are presented in Figure 2, but the trend remains the
same. The results of long-term cost-effectiveness
seem to be stable and proof against both changes in
costs and changes in the effectiveness of the
intervention and less than e20,000 between 1 and
2 years after the intervention with P-NMES.
Discussion
The first and most important conclusion that can be
drawn from this study is that although P-NMES
is more costly compared to the alternative interven-
tions, it also seems to be more effective in the
treatment of HSP. The incremental cost of P-NMES
is
e6,000 per patient treated successfully and an
incremental
e30,000 per QALY gained. However,
no gold standard exists for cost-effectiveness analysis
[30]. In general, cost-effectiveness (cost/QALY) of
e20,000 – 40,000/QALY is consistent with that of
other society-funded medical programs [31]. A cost

Table II. The calculated (c) and modelled (m) short-term analysis data. The average costs, average effectiveness and average 1-year QALY
gain for each individual treatment options are presented. Using both the effectiveness results and the QALY gain data are used to calculate
cost-effectiveness results.

Cost-effectiveness of Average cost of Effectiveness of One year QALY Incremental Incremental cost-effectiveness
P-NMES treatment (e) treatment (%) gain of treatment cost-effectiveness ratio (QALY) ratio

p-NMES C 2041 73 0.0893

M 2036 73 0.0891
sd. 2.76 6 0.003
IA-Injection C 98 42 0.0301 6,268 32,821
M 100 37 0.0309 6,061 33,997
sd. 0.03 9 0.008 3,285 5,434
Slings C 164 28 0.02 4,171 27,085
M 210 20 0.017 3,587 25,605
sd. 1.43 7.5 0.006 799 2,660
 Economic evaluation of P-NMES 649

Figure 1. Sensitivity analysis. The influence of uncertainty of both costs and QALY gain on first year cost-effectiveness results is presented. It
can be seen that although there is some change in cost-effectiveness due to the modelled uncertainty in costs and QALY gain, the changes are
minor and do not influence cost-effectiveness of P-NMES as a whole.

Figure 2. Future year analysis. The modelled future year development in cost-effectiveness of P-NMES, compared to injections and slings, is
presented against varying declines in effectiveness. Irrespective of assumed declines in effectiveness, all comparisons drop below the e20,000
cut-off point between 1 and 2 years after the intervention.
650 J. A. van Til et al.
of 5e20,000/QALY would be considered as parti-
cularly cost-effective, while a cost of 4e60,000/
QALY would be considered expensive. Therefore,
the incremental costs of P-NMES fall below the
e40,000/QALY threshold. In the long-term analysis
the incremental costs of P-NMES fall below the
particularly cost-effective threshold of e20.000/
QALY between 1 and 2 years after the intervention.
Both the effectivity and cost estimates resulting from
this economic evaluation must be interpreted with
caution, as a result of the preliminary nature of this
analysis. Data on the effectivity of treatment options
are mainly obtained either from small trials
[9,10,19,26] or from slightly larger randomized
controlled trials [11]. Moreover, the data on the
costs of all interventions and the estimates of QALY
gained from the alternative interventions are based
on assumptions, although plausible, made by the
authors of the article in discussions with the medical
staff of the local rehabilitation center. To investigate
the influence of uncertainty on both costs and
effectivity, a sensitivity analysis was performed to
study the effect of varying effectivity and costs on the
results. This effect seems to be negligible, and it can
be concluded from this preliminary analysis that the
use of P-NMES in the treatment of HSP seems to be
cost-effective.
One difficulty that is encountered when interpret-
ing the results of this study is the categorization of
patients with acute and chronic HSP. No accurate
comparison can be made between P-NMES and
alternative interventions, because they are each
intended for different stages of HSP. As yet,
P-NMES has only been studied in the treatment of
patients with chronic HSP, which the majority of
whom have already received most treatment options
available with no lasting results, and their shoulder
pain has therefore become chronic. For these
patients, P-NMES might be the only remaining
treatment option.
In contrast, the implementation of P-NMES in
the acute stage of HSP might be cost-effective
according to the decision rules described above.
However, it is also much more expensive than to
the alternative interventions, and would therefore
seems to be unnecessary for patients who respond
well to conventional treatment. Our expectation,
which is supported by the literature [32], is that
NMES is more effective and has a more lasting
effect on HSP than alternative methods of treat-
ment. This pleads for the identification of the
subgroup of patients with acute HSP who are most
at risk of developing conventional therapy-resistant
shoulder pain. Costs could be reduced by treating
these patients with P-NMES, thus avoiding the
treatment with ineffective conventional therapy.
Budgetary impact could even decrease further if
earlier intervention with P-NMES would lead to a
decreased the utilization of resources after treat-
ment, earlier recovery and earlier discharge from
hospital. There are no medical reasons that prohibit
earlier intervention with p-NMES, but the difficulty
lies in identifying the patient subgroup most at risk
of developing chronic HSP.
In conclusion, although more expensive than the
alternative interventions, based on both short- and
long-term economic evaluation, P-NMES seems to
be cost-effective in the treatment of HSP. P-NMES
intervention thereby seems to be justified for
patients with chronic HSP, moreover because
P-NMES is the only treatment option left for the
patients in this extensively treated group. Future
research should aim at identifying the subgroup of
patients with acute HSP who are most at risk of
developing chronic HSP. This would enable inter-
vention with P-NMES in an acute stage, resulting
in possible cost savings. Outcome measures for
future studies should include a reliable measure for
pain, an index for health-related QoL and a
sensitive measure for motor function. Based on this
study, an extensive stage IV budget-impact study
[12] is justified, alongside a randomized controlled
trial of P-NMES intervention for patients with both
acute and chronic HSP.
Acknowledgements
This research was sponsored by NeuroControl
Corporation (NCC), Cleveland, Ohio. NCC intends
to commercialize the device evaluated in this paper.
NCC had no influence at all in relation to this work.
NCC provided financial support for the execution
of this study and allowed a discount on the purchase
of P-NMES devices for the clinical trial that was
performed in our rehabilitation center.
References
1. Teasell R. Musculoskeletal complications of hemiplegia
following stroke. Seminars in Arthritis and Rheumatism
1991;20(6):385 – 395.
2. Turner-Stokes L, Jackson D. Shoulder pain after stroke: A
review of the evidence base to inform the development of
an integrated care pathway. Clinical Rehabilitation 2002;
16(3):276 – 298.
3. Roy CW, Sands MR, Hill LD. Shoulder pain in
acutely admitted hemiplegics. Clinical Rehabilitation 1994;
8:334 – 340.
4. Roy CW, Sands MR, Hill LD. The effect of shoulder pain on
outcome of acute hemiplegia. Clinical Rehabilitation
1995;9:21 – 27.
5. Joynt RL. The source of shoulder pain in hemiplegia.
Archives of Physical Medicine & Rehabilitation 1992;73(5):
409 – 413.
6. Price CI, Pandyan AD. Electrical stimulation for preventing
and treating post-stroke shoulder pain: A systematic Cochrane
review. Clinical Rehabilitation 2001;15(1):5 – 19.

7. Chae J, Yu D. A critical review of neuromuscular electrical
stimulation for treatment of motor dysfunction in hemiplegia.
Assistive Technology 2000;12(1):33 – 49.
8. Delitto A, Strube MJ, Shulman AD, Minor SD. A study of
discomfort with electrical stimulation. Physical Therapy
1992;72(6):410 – 421; discussion on 421 – 424.
9. Yu DT, Chae J, Walker ME, Fang ZP. Percutaneous
intramuscular neuromuscular electric stimulation for the
treatment of shoulder subluxation and pain in patients with
chronic hemiplegia: A pilot study. Archives of Physical
Medicine & Rehabilitation 2001;82(1):20 – 25.
10. Renzenbrink GJ, IJzerman MJ. Percutaneous Neuromuscular
Electrical Stimulation (P-NMES) for treating shoulder pain in
chronic hemiplegia. Effects on shoulder pain and quality of
life. Clinical Rehabilitation 2004;18:359 – 365.
11. Yu DT, Chae J, Walker ME, et al. Intramuscular neuromus-
cular electric stimulation for poststroke shoulder pain: A
multicenter randomized clinical trial. Archives of Physical
Medicine & Rehabilitation 2004;85(5):695 – 704.
12. Sculpher M, Drummond M, Buxton M. The interative use of
economic evaluation as part of the process of health
technology assesment. Journal of Health Services Research
Policy 1997;2(1):26 – 30.
13. Stratford PW, Spadoni G. The reliability, consistency, and
clinical application of a numeric pain rating scale. Physiother-
apy Canada 2001;53:88 – 91.
14. Bonita R, Solomon N, Broad JB. Prevalence of stroke and
stroke-related disability. Estimates from the Auckland stroke
studies. Stroke 1997;28(10):1898 – 1902.
15. Bronnum-Hansen H, Davidsen M, Thorvaldsen P. Long-
term survival and causes of death after stroke. Stroke
2001;32(9):2131 – 2136.
16. Hankey GJ, Jamrozik K, Broadhurst RJ, et al. Five-year
survival after first-ever stroke and related prognostic factors
in the Perth Community Stroke Study. Stroke 2000;31(9):
2080 – 2086.
17. Petty GW, Brown RD Jr, Whisnant JP, Sicks JD, O’Fallon
WM, Wiebers DO. Ischemic stroke subtypes: A population-
based study of functional outcome, survival, and recurrence.
Stroke 2000;31(5):1062 – 1068.
18. van Straten A, Reitsma JB, Limburg M, van den Bos GA, de
Haan RJ. Impact of stroke type on survival and functional
health. Cerebrovascular Disease 2001;12(1):27 – 33.
19. Renzenbrink GJ, Nijlant JMM, Ijzerman MJ, StIM system for
the treatment of painfull shoulder subluxation in chronic
hemiplegia; the first European experience. Proceedings of
1st ISPRM Congress, Bologna; Monduzzi Edit, 2001.
pp 725 – 728.
Economic evaluation of P-NMES 651
20. Brazier J, Roberts J, Deverill M. The estimation of a
preference-based measure of health from the SF-36. Journal
of Health Economics 2002;21(2):271 – 292.
21. Jackson D, Turner-Stokes L, Khatoon A, Stern H, Knight L,
O’Connell A. Development of an integrated care pathway for
the management of hemiplegic shoulder pain. Disability &
Rehabilitation 2002;24(7):390 – 398.
22. Broeks JG, Lankhorst GJ, Rumping K, Prevo AJ. The long-
term outcome of arm function after stroke: Results of a
follow-up study. Disability & Rehabilitation 1999;21(8):
357 – 364.
23. Snels IA, Beckerman H, Lankhorst GJ, Bouter LM. Treatment
of hemiplegic shoulder pain in the Netherlands: Results of a
national survey. Clinical Rehabilitation 2000;14(1):20 – 27.
24. Philadelphia Panel evidence-based clinical practice guidelines
on selected rehabilitation interventions for shoulder pain.
Physical Therapy 2001;81(10):1719 – 1730.
25. Snels IA, Beckerman H, Twisk JW, et al. Effect of triam-
cinolone acetonide injections on hemiplegic shoulder pain: A
randomized clinical trial. Stroke 2000;31(10):2396 – 2401.
26. Dekker JH, Wagenaar RC, Lankhorst GJ, de Jong BA. The
painful hemiplegic shoulder: Effects of intra-articular triamci-
nolone acetonide. American Journal of Physical Medicine &
Rehabilitation 1997;76(1):43 – 48.
27. Doubilet P, Begg CB, Weinstein MC, Braun P, McNeil BJ.
Probabilistic sensitivity analysis using Monte Carlo simulation.
A practical approach. Medical Decision Making 1985;5(2):
157 – 177.
28. Briggs AH, Goeree R, Blackhouse G, O’Brien BJ. Probabil-
istic analysis of cost-effectiveness models: Choosing between
treatment strategies for gastroesophageal reflux disease.
Medical Decision Making 2002;22(4):290 – 308.
29. Halpern EF, Weinstein MC, Hunink MG, Gazelle GS.
Representing both first- and second-order uncertainties by
Monte Carlo simulation for groups of patients. Medical
Decision Making 2000;20(3):314 – 322.
30. Briggs AH, Gray A. Using cost effectiveness information.
British Medical Journal 2000;320:246.
31. Kupersmith J, Holmes-Rovner M, Hogan A, Rovner D,
Gardiner J. Cost-effectiveness analysis in heart disease.
Part II: Preventive therapies. Progress in Cardiovascular
Disease 1995;37(4):243 – 271.
32. Baker LL, Parker K. Neuromuscular electrical stimulation
of the muscles surrounding the shoulder. Physical Therapy
1986;66(12):1930 – 1937.