1. Introduction
Research in Developmental Disabilities 33 (2012) 1763–1772
Article history:
Received 16 November 2011
Received in revised form 19 March 2012
Accepted 3 April 2012
Available online 14 June 2012
Keywords:
Early intensive behavioral intervention
Special education
Cost-offset study
Autism spectrum disorder
Early intensive behavioral intervention (EIBI) may result in improved cognitive, adaptive
and social functioning and reductions in autism severity and behavioral problems in
children with Autism Spectrum Disorder (ASD). For a subset of children, normal
functioning may be the result. However, due to the intensity (20–40 h per week for 3 years
with a low child staff ratio) implementation costs are high and can be controversial.
Estimated costs for education, (supported) work and (sheltered) living for individuals with
ASD in the Netherlands are applied in a cost-offset model. A compelling argument for the
provision of EIBI is long term savings which are approximately s 1,103,067 from age 3 to
65 years per individual with ASD. Extending these costs to the whole Dutch ASD
population, cost savings of s 109.2–s 182 billion have been estimated, excluding costs
associated with inflation.
� 2012 Elsevier Ltd. All rights reserved.
Contents lists available at SciVerse ScienceDirect
Research in Developmental Disabilities
Autism spectrum disorder (ASD) is one of the most common developmental disorders. The disorder is characterized by
impairments in communication and social interaction, by repetitive behaviors and by limited areas of interest (American
Psychiatric Association, 2000; Baron-Cohen et al., 2009). In 50–80% of the individuals with ASD an intellectual disability (ID)
is also present (Goldberg Edelson, 2006). Although individuals with ASD present great variability in severity and clinical
picture, their prognosis without treatment is generally poor. ASD is a chronic disability and due to the unique social and
communicative difficulties the majority of individuals involved requires professional care throughout their lives (Billstedt,
Gillberg, & Gillberg, 2005; Mordre et al., 2011). As more people are being diagnosed with ASD and require specialized
services (Wing & Potter, 2002; Yeargin-Allsop et al., 2003), the costs of public health and social welfare programs are
increasing. Ja¨rbrink and Knapp (2001) estimated the lifetime costs (including costs such as family expenses, medication and
daycare) to care for an individual with ASD in Britain more than s 2.5 million, excluding the costs associated with typical
child rearing.
Currently, early intervention based on applied behavior analysis (EIBI) is considered the treatment of choice for children
Cost comparison of early intensive behavioral intervention and treatment
as usual for children with autism spectrum disorder in the Netherlands
Nienke Peters-Scheffer a,b,*, Robert Didden a,c, Hubert Korzilius d, Johnny Matson e
aBehavioural Science Institute, Radboud University Nijmegen, PO Box 9104, 6500 HE Nijmegen, The Netherlands
b Stichting De Driestroom, PO Box 139, 6660 AC Elst, The Netherlands
c Trajectum Zutphen, PO Box 300, 7200 AH Zutphen, The Netherlands
d Institute for Management Research, Radboud University Nijmegen, PO Box 9108, 6500 HE Nijmegen, The Netherlands
eDepartment of Psychology, Louisiana State University, Baton Rouge, 70803, USA
A R T I C L E I N F O A B S T R A C T
with ASD (Eikeseth, 2009). Although EIBI programs vary slightly in their approach, all programs are characterized by the
following essential features: (1) systematic use of behavior analytic principles, (2) treatment is comprehensive, (3)
* Corresponding author.
E-mail address: n.peters@pwo.ru.nl (N. Peters-Scheffer).
0891-4222/$ – see front matter � 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ridd.2012.04.006
N. Peters-Scheffer et al. / Research in Developmental Disabilities 33 (2012) 1763–17721764
systematic development from simple to more complex skills with a transfer to natural settings, (4) functional assessment of
individual behavior and individualization of goals and instructional procedures, (5) the use of scientific methods to evaluate
the effects of treatment, (6) early age of treatment onset, (7) a low child staff ratio, and (8) involvement and training of
parents and significant others. Usually, treatment is implemented intensively (i.e., 20–40 h per week) for a long period of
time (i.e., two years or more; Eikeseth, 2009; Green, Brennan, & Fein, 2002; Leaf & McEachin, 1999; Lovaas, 2003), although
less intensive EIBI has also shown positive outcomes (e.g., Eldevik, Eikeseth, Jahr, & Smith, 2006; Peters-Scheffer, Didden,
Mulders, & Korzilius, 2010).
Four of the five meta-analyses included in an overview of Reichow (2011) concluded that EIBI is an effective intervention
strategy for many children with ASD and results in increased cognitive, social and communication skills and reductions in
challenging behavior. However, great variability in outcome within and between studies is seen, with some children making
rapid and remarkable progress, while other children’s gains are limited (Eikeseth, 2009; Peters-Scheffer, Didden, Korzilius, &
Sturmey, 2011; Reichow & Wolery, 2009). These differences in outcome are influenced by child and family factors (e.g., age of
treatment onset, co-morbid conditions and pre-treatment IQ, autism symptom severity and language) and treatment
characteristics (e.g., treatment intensity, treatment duration, treatment quality, and intensity and quality of supervision;
Allen & Warzak, 2000; Ben-Itzchak & Zachor, 2007; Davis, Smith, & Donahoe, 2002; Eikeseth, Hayward, Gale, Gitlesen, &
Eldevik, 2009; Eldevik et al., 2006; Granpeesheh, Dixon, Tarbox, Kaplan, & Wilke, 2009; Lovaas, 1987; Peters-Scheffer et al.,
2010; Smith, Eikeseth, Klevstrand, & Lovaas, 1997; Smith, Groen, & Wynn, 2000; Smith, Klevstrand, & Lovaas, 1995).
Due to treatment intensity and duration, EIBI is an expensive treatment (albeit highly effective), but only a few studies
have been conducted regarding the financial costs and benefits of EIBI to children with ASD. Based on the outcome of the
studies by Lovaas (1987) and McEachin, Smith, and Lovaas (1993), Jacobson, Mulick, and Green (1998) estimated the costs
and benefits of EIBI for children with ASD in Pennsylvania. In their model, in which EIBI outcome ranged from regular
education without support (47%) and less intensive special education (42%) to intensive special education (11%), cost savings
ranged from $ 187,000 to $ 203,000 per child for ages 3–22 years and from $ 656,000 to $ 1,082,000 per child for ages 3–55
years. Authors concluded that the estimated savings outweigh the differences in initial treatment costs for EIBI. Translating
outcome into a cost dichotomization (i.e., successfully or unsuccessfully mainstreamed in special education), Chasson,
Harris, and Neely (2007) estimated that with a success rate of 72%, $ 208,500 per child would be saved by the state of Texas
across 18 years of education with EIBI. Motiwala, Gupta, and Hon (2006) used more conservative efficacy rates and therefore
estimated savings lower than Jacobson et al. (1998) and Chasson et al. (2007) between 34.479 and 53.720 Canadian Dollars
per individual.
Cost-effectiveness studies by Jacobson et al. (1998) and Chasson et al. (2007) have estimated costs exclusively based on
best outcome studies (i.e., Lovaas, 1987; McEachin et al., 1993; Sallows & Graupner, 2005) and although Motiwala et al.
(2006) used more conservative efficacy rates, none of the studies included less favorable outcomes studies published after
Lovaas (1987). In these studies children received fewer hours than recommended (e.g., Anderson, Avery, DiPietro, Edwards, &
Christian, 1987), program supervision was infrequent or of less quality (e.g., Bibby, Eikeseth, Martin, Mudford, & Reeves,
2002) or children had a higher pretreatment chronological age (e.g., Eikeseth, Smith, Jahr, & Eldevik, 2002; 2007) or lower IQ
(e.g., Smith et al., 1997) than in the Lovaas study (1987). It is unlikely that in community-based programs children obtain the
same outcome as in the Lovaas study (Mudford, Martin, Eikeseth, & Bibby, 2001). Therefore, by including studies with less
favorable outcome our study may provide a more realistic estimate of the outcome of community based EIBI and the
potential costs and savings of EIBI in the Netherlands.
The present study was designed to provide a cost-offset analysis of EIBI relative to treatment as usual for children with
ASD in the Netherlands. An estimate of the effects of EIBI on financial costs was investigated. To date only one study
regarding the effectiveness of EIBI has been published in the Netherlands (Peters-Scheffer et al., 2010). Therefore, potential
cost savings including different outcomes of EIBI were based on a number of international studies and presented per child
and extended to the ASD population in the Netherlands.
2. Cost-offset analysis
To analyze the cost and benefits of EIBI in the Netherlands, efficacy rates of EIBI and treatment as usual based on meta-
analytic studies were determined first. Next, costs were identified for individuals with ASD from age 3 to 65 years including
costs for education, work and living and total costs were calculated for individuals with ASD who received EIBI or treatment
as usual including different outcomes (i.e., normal functioning, reduced dependency, or dependency). Lastly, avoided costs
for the Netherlands through the provision of EIBI were calculated per child and for the Dutch ASD population.
2.1. Efficacy rates
As still relatively few children in the Netherlands receive EIBI and pre-treatment and/or post treatment data of most
children is lacking, efficacy rates of both the EIBI group and the treatment as usual group were based on published literature.
Following Jacobson et al. (1998), children were categorized into three groups according to their level of functioning. The first
group was comprised of children who achieve normal functioning, participate in regular education with little or no support
and who are vocationally productive adult workers. The second group consisted of children who participate in less intensive
special education and evince reduced dependency throughout their lives, while the third group requires continuing
specialized and intensive educational and adult services.
N. Peters-Scheffer et al. / Research in Developmental Disabilities 33 (2012) 1763–1772 1765
2.1.1. Behavioral intervention
Given the controversy (e.g., Schopler, Short, & Mesibov, 1989) regarding the reported efficacy of the Lovaas study (1987)
and several replications (e.g., Sallows & Graupner, 2005), the efficacy figures used in our study are based on the results
reported in six meta-analytic studies regarding EIBI that were recently published (i.e., Eldevik, Hastings, Hughes, Jahr,
Eikeseth, & Cross, 2009; Makrygianni & Reed, 2010; Peters-Scheffer et al., 2011; Reichow & Wolery, 2009; Spreckley & Boyd,
2009; Virue´s-Ortega, 2010). Common measures reported in the included studies are IQ, adaptive behavior, language and
school placement. Although subject to parental advocacy and school policy, school placements seem to be the best real world
efficacy measurement of academic and social competence (Kazdin, 1993) and are therefore used in our analysis.
Characteristics of the studies included in the meta-analyses are displayed in Table 1. As the study of Matos and Mustaca
(2005) was in Spanish, it was excluded from the analysis. Also, the control groups of the studies of Ben-Itzchak, Lahat, Burgin,
and Zachor (2008) and Harris, Handleman, Gordon, Kristoff, and Fuentes (1991) were excluded, as they were not comprised
of children with ASD. Sixteen of the studies reported school placements. However, the follow-up studies by Lovaas (1987),
i.e., McEachin et al. (1993) and Eikeseth et al. (2002), i.e., Eikeseth, Smith, Jahr, & Eldevik (2007) used the same participants as
the original studies and were therefore combined with the original papers. The actual treatment hours for the children in the
low intensity group of Lovaas (1987) and Smith et al. (2000) are not reported. As these children received minimal EIBI
treatment (i.e., less than 10 h per week), school placements of these groups of children were excluded from the analyses.
Therefore, 14 studies were included in the analysis. There were 292 children with a mean chronological age of 41.45 months
(30.2–66.3) and a mean pretreatment IQ of 60.17 (50.5–83.0). On average, children received 32.54 h of treatment per week
(20–40) for 27.01 months (12–36). After treatment, 29% were placed in regular treatment, 34% were placed in less intensive
special education and 37% were placed in special education.
2.1.2. Treatment as usual
Six studies also report school placements of children in a control group who had a mean chronological age of 42.88
months (33.2–65.0) and mean pretreatment IQ of 62.73 (59.4–65.2). After eclectic treatment or treatment as usual, 11% of
the children were placed in regular treatment, 8% were placed in less intensive special education and 81% were placed in
special education. These rates are roughly in line with studies on outcome of adolescents and adults with autism (e.g., Levy &
Perry, 2011).
However, the rates noted above are more pessimistic than rates provided by the Dutch Association for Autism
(Nederlanse Vereniging voor Autisme, 2008). They reported that 36% of the adults lived independently (18% with a partner)
and 10% lived independently but with support. Twenty-nine percent had a paid job (25% with sufficient income to provide in
their living), 35% worked voluntary (e.g., voluntary job, traineeship, sheltered work with support), 13% worked in a sheltered
environment, and 13% participated in structured daytime activities. About 20% of the adults had no structured day care or
(supported) work.1 Half of the adults received a security income from the Dutch Government (so-called WAJONG) as major
source of income and the majority (74%) indicated they needed professional support in conducting their work with 55%
actually receiving this support. Therefore, emulating Motiwala et al. (2006), also the most positive figures (Freeman, 1997)
were included to estimate costs for children who receive standard care in the Netherlands. In Freeman’s study, 25% of the
participants attained normal functioning, while 25% evinced semi-independent living, and 50% were very dependent at
adulthood.
2.2. Costs
Costs were calculated from age 3 to 65 years for individuals with ASD who received EIBI or treatment as usual including
different outcomes (e.g., normal functioning, reduced dependency, or dependency). ASD can be reliably diagnosed between
two and three years of age (Kleinman et al., 2008) and costs after 65 years of age are difficult to estimate due to health costs,
retirement and pension. Also, some researchers assume a higher mortality rate among individuals with ASD (Mouridsen,
Brønnum-Hansen, Rich, & Isager, 2008; Picket, Xiu, Tuchman, Dawson, & Lajonchere, 2011). Estimated costs in Euros are
displayed in Table 2.
2.2.1. Education
In the Netherlands, typically developing children receive, on average, 8 years of primary education, four to six years
secondary education, and four years of intermediate or higher vocational education or university. Attending school is
compulsory from the age of five, but most children start primary school when they are 4 years old (student staff ratio 14.6
to 1) and graduate between 20 and 22 years of age (Minne, Webbink, & van der Wiel, 2009; OCW, 2008).
Approximately 5% of the children in primary school attend special education (Centraal Bureau voor de Statistiek, 2009),
divided into less intensive special education (student staff ratio 5.9 to 1; Minne et al., 2009) and intensive special education
1 Some of the participants mentioned multiple jobs.
Table 1
Characteristics of the studies included in themeta-analysis (i.e., Eldevik et al., 2009; Makrygianni & Reed, 2010; Peters-Scheffer et al., 2011; Reichow &Wolery, 2009; Spreckley & Boyd, 2009; Virue´s-Ortega, 2010).
Study EIBI group Control group
Participants Treatment characteris-
tics
Outcome Participants Treatment characteristics Outcome
N CA IQ Hrs Dur. 1 2 3 N CA IQ Type Hrs Dur. 1 2 3
Anan, Warner, McGillivary,
Chong, & Hines (2008)
72 44.0 51.69 15 3
Anderson et al. (1987) 14 42.79 57.26 15–25 12–24 0 31 69
Baker-Ericze´n, Stahmer,
and Burns (2007)
158 49.36 3
Ben-Itzchak et al. (2008) 44 27.29 74.84 45 12
Ben-Itzchak and Zachor (2007) 25 26.6 70.67 > 35 12
Bibby et al. (2002) 66 45.0 50.8 30.3 31.6 5 53 42
Birnbauer and Leach (1993) 9 38.1 45.3 18.72 21.6 5 33.2 45 22
Boyd and Corley (2001) 22 41.3 – 30–40 23 0 41 59
Cohen, Amerine-Dickens,
& Smith (2006)
21 30.2 61.6 35–40 36 48 33 19 21 33.2 59.4 Ecl. 36 5 0 95
Eikeseth et al. (2002, 2007) 13 66.31 61.92 28.00 31.4 38 0 62 12 65.00 65.17 Ecl. 29.08 33.3 8 0 92
Eldevik et al. (2006) 13 53.0 41.0 12.5 20.3 15 49.0 47.2 Ecl. 12 21.4
Harris and Handleman (2000) 27 49.0 59.33 35–45 41 0 59
Harris et al. (1991) 9 50.11 67.56 35–45 12 0 89 11
Howard, Sparkman, Cohen,
Green, & Stanislaw (2005)
29 30.86 58.54 25–40 14.21 16
16
37.44
34.56
53.69
59.88
Ecl.
Ecl.
25–30
15
13.25
14.75
Lovaas (1987); McEachin et al. (1993) 19
19
34.6
40.9
53.0
46.0
40
<10
24+
24+
47
0
42
42
11
58
21 <42 > 40 Usual – 24+ 5 48 48
Magiati, Charman, & Howlin (2007) 28 38.0 83 32.8 24 0 82 18 16 42.5 65.2 Ecl. 26.5 26 0 0 100
Reed, Osborne, & Corness (2007) 12 40 55.6 30.4 9 20
16
43
38
51.9
53.3
Ecl.
Port.
12.7
8.5
9
9
Reed, Osborne, & Corness (2007b) 14
13
42.9
40.8
57.21
49.3
30.4
12.6
9–10
9–10
Remmington et al. (2007) 23 35.7 61.43 25.6 24 74 0 26 21 38.4 62.33 Usual 24 48 0 52
Sallows and Graupner (2005) 13
10
35.0
37.1
50.85
52.10
37.58
31.28
48
48
48 43 9
Sheinkopf and Siegel (1998) 11 33.8 62.8 27.02 15.73 30 20 50 11 35.3 61.7 Usual 11.13 0 0 100
Smith et al. (1997) 11
10
36
38
28
27
30
<10
24
24
Smith et al. (2000) 15
13
36.07
35.77
50.53
50.69
24.52 33.44
24
27
0
13
21
60
79
Weiss (1999) 20 41.5 40 24 50 25 25
Note. CA = average age in months; Hrs = average number of hours per week of treatment; Dur. = average number of months of treatment; 1 = percentage of children placed regular education with no support,
minimal support (e.g., part-time support with shadow tutor, fading the shading tutor) or unknown support; 2 = percentage children with regular school placement with full-time individual support or part-time
EIBI, less intensive special education (e.g., for childrenwith communication impairments ormild ID), mixture of special education and regular education placement, private school with small classes; 3 = percentage
children receiving one-to-one-treatment and special education (e.g., autism specific schools, generic special needs schools, self-contained classes); usual = treatment as usual; ecl = Eclectic treatment; port = Portage
program. Since Sallows and Graupner (2005) report educational placement for the parent-directed and clinic-directed group together, placements are reported for the two groups together.
N
.
P
eters-Sch
effer
et
a
l.
/
R
esea
rch
in
D
ev
elo
p
m
en
ta
l
D
isa
b
ilities
3
3
(2
0
1
2
)
1
7
6
3
–
1
7
7
2
1
7
6
6
N. Peters-Scheffer et al. / Research in Developmental Disabilities 33 (2012) 1763–1772 1767
(student staff ratio 3.5 to 1; Minne et al., 2009). Less intensive special education is attended by students with learning
difficulties, while intensive special education serves children with visual or hearing impairments, children with severe
communication impairments (including hearing problems), children with ID and children with psychiatric disorders. Some
children with ASD are enrolled in special c