Practice Guidelines for

Direct Attention Training

McKay Moore Sohlberg, Ph.D.

Communication Disorders & S ciences
University of Oregon
Eugene, Oregon

Jack Avery, M.A.

Department of Communication Disorders
Uni oereity of Minne sota
Minneapolis, Minnesota

Mary Kennedy, Ph.D.

Department of Communication Disorders
Un iversity of Minnesota
Minneapolis, Minnesota

Mark Ylvisaker, Ph.D.

Departmen t of Communication Disorders
College of Saint Rose
Albany, New York

Carl Coelh o, Ph.D.

Communication Sciences Department
Uni oereity of Connecticut
Storrs, Connecticut

Lyn Turkstra, Ph.D.

Departmen t of Communication S ciences
Case Wesf.em Reserve Uni versity
Cleveland, Ohio

Kathryn Yorkston, Ph.D.

Department of Rehabilita tion Med icine
Un iversity of Wash ington
S eattle, Washington

J<»,n..,1 <>f M<dk<.l Spft<~ · I"v'(i'_ AuMl'¥l:l

\'olu""" II. Number 3. PI> . i.-uxi:l
Copyri, n' Cl 2003 by no,l",,,, Le~ mi"ll. a d;v;. iun ofThon>tO<l !.earning. 10K.
xx ANCDS BULLETIN BOARDIVOL. 11, NO. 3

Th is a rticle is part of a llllncs of reports from a committee charged with dQVcloping evi-
dence -based practice l EBP ) guide lines fo r rehnbihteticn of cognitivc-cQlnmu nication
deficits following t ra uma tic brai n injury (TBn . We examine t he literature for evidence
of the erteeuvecess of direct attention t raining to treat attention hn llainncntil following
T BI. Evidence is gleaned from the outcomes of nine Class I and Ctaee 11 st udies that
spa n interven tion from acu te to out pat ien t reh abilit ation. Hcsultll and diecuaaion are
or,g"a ni zcd us ing five key ques t ions aa a mech an ism to rev iew the research to det ermine
if the upprcach , outcomes, anti associated recommendations warrant II cha nge in clini-
ca l practice. The key qu estions are: Wh o life the participan\.!l who r eceived the int er-
vention? Wh a t comprises the attenti on trai nin g? What are th e outcomes of tilt" inte r-
veutien? Are there methodological concerns'! ~ th ere clinically <t pflhcable trends
IIC TOSS different attention remediat ion stu dies? The complexities nnd difficulties inher-
em in imp lem en ting clinical t rialli. wit h th e hete rogeneous T BI population a re dis-
cussed . The a rt icle concludes with treatment guidelines and eptlcos suppo rted by t he
research review, Fut ure eeaeareh needs arc highlig hted.

In recent years, t he leadership of the Academy of
Neu ro logic Comm un ication Disorde rs and Sci-
ences (A..N C DS), t he American Speech-La nguage-
Hearing Associa tion (AS HA), the Special Interest
Division 2 (SID 2-Neuro physiology and Neuro-
genic Speech and Lan guage Disorders ) of ASHA,
and t h e Vete rans Administration recogni zed the
tren d toward refe rencing research evidence to sup-
port clinica l decisi on making in t he ma nagement
of medical conditions. In 1997, ANCUS emba rked
on the ta sk of es tablishing committees of experts
to develop evidence-based practice ( EBP) gu ide-
lines for the following areas; dysarthria, aphasia,
de mentia. a praxia , an d cognitive rehabilitation for
t ra um atic brain injury. This article is part of a se-
ries of reports from the commit tee charged with de-
veloping EBP gui delines for rehabilitation of cogni-
tive-communicative defici ts follo wing traumatic
bra in inj ury. For an overview of the committee
process, t he reader is referred to the project intro-
duction and initial com mittee report (Golper et al .,
2001; Kennedy et al., 2002)
The in itial task of our subco mmittee was to ou t-
line our philosophy and assumptions relev an t to
EBP a nd clinical decision making, This required us
to define "evidence." Evidence is the reason a clini-
cian would pursue a s pecifi c treatmen t decision.
Central to our posi tion is the proposition t ha t t he
science of clinical decision making for complex and
diverse populations includes, but is not restricted to,
informat ion generated by randomized con trolled
t rials and other clinical experiments. We described
our bas ic premises in an initial report and positi on
paper (Ylvisa ker et aI., 2002). An early task was to
define the proj ect scope by identifying the major ar-
eas within the field of cognitive-communicative re-
h abilita t ion wem for which we would develop
guidelines. These a reas include cognitive-communi-
cation assessment, at tention training, management
of memory impa irments, social skillslbehavior regu-
lation management, and in terven tion for metacog-
nitiv e and executive func tion deficits. This current
article addresses attention training. Specifica lly, we
discuss the evidence relevant to t reat ment ou t-
comes associated with direct or s t ructured atten tion
t raining fo r attenti on impairments followin g TBL
THE EVIDENCE·BASED
PRACTICE MOVEMENT
In an attem pt to establish defensible, vali dated
intervention pra ct ices, the field of cognitive rehabil-
itation has joined the healthcare moveme nt's vigor-
ous commitment to developing EBP. The primary
goal of this bu rgeoning reform is to ensure thai clin-
ical decisions are guided by empirical evidence a nd,
ideally, evidence from well-controlled s t udies that
sys tema tically evaluate ou tcom e, efficacy, and effec·
t ivenees of s pecific interventi ons (Golper et at,
200 1). More s pecifica lly, t he charge is to develop evi-
dence-based gu idelines that causally link treatment
protocols to expected clinical outcomes (Robey,
2001). Increasingly. however; it is recognized t ha t
strict adherence to a preordain ed experimental
model as a filter to accept or reject particula r clini-
ca l pract ices in a heterogeneous populat ion te.g., the
TBI population ) may result in unsuccessfu l gener-
alizations from resea rch to treatmen t outcomes. Sci-
entifically sound clinical decision making includes
a na lyzing exist in g experimental evidence in add i.-
tiou: to a diverse set of individual patient coneidera-
PRACTICE GUIDE LINES FOR DIRECT ATIENTION TRAINING
lions and possibly t he res ult of client-specific exper-
ime ntal intervention trials (Montgomery & Turkstra,
2003; Ylvisaker et al., 2002).
The development of EB P guidelines for CCR in
general , and specifically for direct attention train-
ing, required us to become conscious of our a s-
su mptions and biases relevant to analyzing reha-
bilitation resea rch. This process made us mindful
of t h ree fundam ental be liefs. F irst, when reviewing
the literatu re, we realized the narrow qu estion
"Does it work?" oversimplifi ed the com plexities be-
h ind direct a t tent ion train in g delivered to TBr s ur-
vivors. Second, we fou nd it crit ical to remember th e
bidirectional nature of the relationship between
practice a nd resea rch . Third, we maintain that in-
terven tion outcomes are rela tive to persona l goals
and va lues.
Researchers are cha rged with the task of con-
du cting clin ica l research tha t effect ively eva lua tes
specific aspects of particula r in terventions in light
of previous wor k. To mak e well-founded clinical
j udgments, practitioners are challenged with the
respons ibility of in tegrating existing research find-
ings with their own clinica l experience, while re-
specting the uniq ue cha llenges presented by their
clients. We boldly suggest tha t it may not be possi-
ble to a nswer t he q uestion "Does a ttention remed i-
a tion work?" using a n unequivocal yes/no form at .
S uch a question assumes that atte nt ion remedia-
tion is a uniformly delivered in tervention that can
be compared across stud ies a ud that the myriad of
relevant contextual variables presen t in a clinical
practice can be dismissed .
Our concern is tha t. the unequivocal "Does it
work?" question is divisive and encourages a battle
tha t is ne ither helpful nor illumi nating. For
decades, the fi eld of CCR seems to have been
trapped in all interna l debate over whether it is bet-
te r to foc us on training processes, skills, or complex
functional abilities, and in what ways and in wh at
contexts that training might best be accomplished .
Although the battle is not over, it is no longer dis-
puted tha t functional changes must be the goal of
treatment and that there are many ways to facili-
tate functional changes. with an equal number of
ways to measure them. The fi eld continues to strug-
gle with a quagmire of measurement issues. We
have. however, learned that a one-size-fits -a ll solu-
tion does not work. Individ uals and fami lies re-
spond to different interventions in different ways
and at different times after injury. The response dif-
ference is in part demonstrated by the widely d iver-
gent treatment results reported in the literature,
which range from no change in performance after
treatment to significant improvements (Sohlberg &
Ma teer, 2001 ). As described in this current article,
the wide swings of outcome in the attention rem e-
diation literature suggest tha t the Question "Does it
wor k?" may be better replaced by "When does it
work best and for whom?" Then we can move on to
such q uestions as "Is it worth the effort?"
No one disputes the va lue gained wh en research
advances and im proves clinical practice. However,
the investigation process is not unidir ectional.
Clinical practice a lso sha pes research questions.
The patient-specific hypothesis testing CPSHT) ap-
proach discussed by Ylvisaker and Feeney ( 998 )
reinforces the sym biot ic relationship between re-
search a nd practi ce. Neither service delivery nor
applied research exists in isolation. Clin icia ns rely
on we ll-founded research to genera te clinical proce-
d ures that can be defended and supported within
their delivery contexts. They must a lso often em -
ploy diagnostic therapy to determine treatmen t e f-
fects in specifi c pa tients. In turn , researchers' study
of cur rent clinical practices gene rates new research
questions and highlights issues relevan t to treat-
ment efficacy, particularly when the research repli-
cates the rea lities of clin ical delivery (e.g., pa tient
demographics, dose, an d duration limits). This iter-
ative process is critical to the development of EB P.
Oue im portant assumption in our view of devel-
oping EBP from existing Literature is tha t this
process necessarily draws on personal goals and
va lues. Montgomery a nd Turkstra (2003) stated
that a problem with developing evidence-based
guidelines is that wha t constitutes a meaningful
clinical outcome is a personal and social judgmen t.
As s uch, the ability to scru tinize the lite rature fo r
eviden ce will never be a wholly scien tific process in
the traditional sense of science. The ever-evolving
definitions describing different types of outcomes
promoted by the World Hea lth Organ ization
(WHO) highlight the fact that this construct is a re-
fl ection of social va lue. Consider the scena rio of two
CCR experts evaluating a research outcome for an
article investiga ting the effectiveness of a ttention
remediation. One analyst may exa mine the im-
provements in neuropsychological test scores fol-
lowing attention training as proof of treatment s uc-
cess, while the other interprets it a s a dismissable
treatment artifact. This is indeed the case within
t he research (Pa r k, Proulx, & Towers, 1999 ;
Sohlbe rg; McLa ughlin , Pavese, Heidr ich, & Posne r,
2001). Acknowledgm ent of the subjective and val-
ue-laden nature inherent in analyzing outcomes is
cr itical to tile development of useful guidelines that
help clinicians nav igate th rough the literat ure.
xxu
ESTABUSHING GUIDELINES FOR
ATTENTION T RAINING
Establis hing pr actice gui deli nes for attent ion
t ra ini ng is a curren t foc us in t he fi eld of CCR (C i-
cerone et aI., 2000). The committee adopted a broad
definition of a tten tion. We were interested in inter-
ventions that addressed a wide assort ment of
skills, processes, a nd cogn itive states tha t rela te to
the ability to focus and process incomin g informa-
tion. At first blush, it may a ppear a st raightforwa rd
process to measure and analyze the a ttention out-
comes in individua ls with brain injury wh o have re-
ceived attention tra ining. U nfortunately, close in-
s pection revea ls thi s is no t the case.
Direct Attention Training as an Interve ntion
Attention training is based on the premise that at-
tent iona l a bilit ies ca n be im proved by act ivating
particula r aspects of attention through a stim ulus
drill a pproach . The repea ted stimulation of etten-
t ional systems via graded a ttention exercises is hy-
pothesized to facilitate cha nges in attentiona l func-
tioning (Cicerone et al. , 2000; Sohlberg & Ma teer,
2001). Most a ttention training program s assume
tha t aspects of cognition can be isola ted and dis-
cretely ta rgeted wi th t raining exercises. The as -
pects of attention tha t are addressed vary widely
among interventions a nd freq uently depend on a
t heoretical model of attention. Atten tion mod els,
regardless of their ope ra tiona l fra mework, appear
to include funct ions related to susta in ing attention
ove r time (vigila nce), ca pacity for inform a tion ,
s hifting attention, speed of processing, and sc reen-
ing out d istractions. Some a ttention efficacy stud-
ies eva luate a ttention interven tions that focus on
pa rticula r attention components s uch as reaction
time a nd sus ta ined a ttention for visual inform a tion
(e.g., Pons ford & Kin sella , L988). Ot he r efficacy
st udies use a ttention training programs that in-
cl ude hierarchical tasks to address a continuum of
attention components from basic sustained at ten-
tion to more complex men tal control (e.g., Park et
al . 1999; Sohlberg et al. , 200 1).
Tradi tiona lly, stud ies em ploy a un iq ue battery of
measures to assess possible changes related to in-
tervention. Some stud ies limit their outcome mea-
sures to indi vidually selected neuropsychological
tests te.g., Park et a l. , 1999; Sturm , WiJlmes, Or-
gass, & Hartje, 1997). Alterna tively studies may
utilize patients' percep tion of cha nge (e.g., Ci-
cerone, 2002; Sohlberg et a l., 2(01) or their tas k
ANCDS BULLETIN BOARDNOL . 11. NO. 3
performance in an activity assumed to require a t-
tentional processing (Kewruan et al., 1985) as an
indicator of attention function ing. The dispa rate
outcome ma rkers render it difficult to equate suc-
cesses and therefore compa re interven tions.
There a re severa l additional complications that
must be considered when reviewing the attention
literature. One conside ration is the rich opportuni .
t ies provided by cross-populat ion inferences. As dis-
cussed by Ylvisaker et a l. (2002 ), there is a wealth
of information on populations who are ne urologi-
ca lly andlor symptomatically close to indivi duals
with brain inju ry. Although our review is limited to
the a tte nt ion train ing literature on individ uals
with traumatic brain inj ury, we wa nt to ack nowl-
edge extens ive informat ion releva nt to attention
remed iation availa ble in other literature domains
such as special ed ucation. Addit ionally, from a n
ecological stand point, it may be remiss to eva lua te
one particula r a pproach for addressing e ttentional
im pa irments when in reality cl in icians often com-
bine interventions. For example, direct tra ining is
ofte n pai red with pha rmacological management,
training in the use of extern al aids, and/or meta-
cogn itive train ing.
Scr utin izin g the Atten tion
Literature Using Key Que s tions
As discussed , we submit tha t answering a binary
qu estion que rying whether attention training does
or does not work is not possible, give n the com plex-
ity of client, trea t ment, and research varia bles cou-
pled with the value-laden nature of ou tcomes. We
offer, as an al te rnative, a templa te of five key ques-
tions tha t professionals can em ploy to eva luate the
literature. This a pproach offe rs a mechanism to fi l-
ter the past , cur rent, an d fu ture a ttention inter-
vention lite rature and help clinicians decide if the
approach, outcomes, and associated recommen da-
tions a re a "good fit" for their practice. The hope is
that the templa te will al low clinicians to scrutinize
the research to determine if research results war-
rant a change in clinical practice a nd a llow them to
deve lop an evolving portfo lio of ev ide nce-based
treatment practices . We further hope that the tem-
pla te will assist resea rchers in designing studies
tha t reflect clinical reali ties. Key guiding questions
for examining the attention training literature a re:
1. Who a re the participants who received the
intervention?
• What is the diagn osis or etiology?
,. What is the injury severity?
PRACTICE GUIDELINES FOR DIRECT ATTEN TION TRAINING
• What is the participant's age?
• What. is the participant's level of education?
• What is the time postonset of injury?
• Are there dual diagnosis or comorbidity factors?
• Wha t is the participant's cognitive profile
postinjury?
2. What comprises the attention training?
• Wh at are the focus and rationale?
• What are the treatment duration and frequency?
• Where is the treat ment setting?
• Who are the providers?
• Are training programs personalized to match
client skills andlor needs?
• Are other in tervent ions incor pora ted into or in
addition to the delivery of attention exercises
(e.g., reinforcement or strategy training)?
3. What are the outcomes of the intervention?
• Are there measures suggesting changes in
at ten tion impairment (e.g., psychometric tes ts)
following t rea tment?
• Are there measures suggesting changes in
activity/participation (e.g., changes in an atten-
tion -demanding skill such as driving, changes
in perceptions/rating of ability by client and/or
caregiver)?
• Are reported chan ges clinically meaningful?
• Is there maintenance or generalization of any
reported changes?
4. Are there methodological concerns? Are there
other explanations for given outcomes, and may
results be either exaggerated or hidden?
• What is the s tudy design?
• Are treatments compared to an alternative or
no treatment condition?
• Are reliability and/or validity iss ues addressed?
5. Are there clinically applicable trends across dif-
ferent a ttention remediation st udies?
• Are there ro bust findings t ha t warrant a
change in practice?
A retrospect ive application of these key ques-
tions to a former cognitive re habilitation interven-
t ion outside of attention remediation serves as an
illustration. Conside r the example of memory drills
that were a comm on memory im pairment inter-
vent-ion during the 1980s. Using t his a pproach ,
clinician s gave clients mult iple tria ls of list learn-
ing or paragraph listening tasks as a means to im-
prove recall a bility (e.g., Glisky & Schecter, 1986).
Even t ua lly, t his a pproach fell out of favor when
both clinicians and researchers found no improve-
ment in eit her clients' or s ubjects' mem ory func-
tioni ng. Had clinicians s u bjected the literat ure to
the proposed quest ioning sequence in the template,
xxnr
the field may have been q uicker to discard an inef-
fective t rea tmen t. Specifically, if clinicians had fo-
cused their reading of the litera ture on the mea -
s urement outcomes (q uest ion 3), they would have
noted tha t memory dr ills occasionally resulted in
improved list learning for s pecifi c lis ts, but that
generalization was poor and no improvements
we re seen in other memory meas urements (God -
frey & Kni ght, 1985; Moffat , 1992). Using the WHO
outcome nomenclature, there was :::I s pecific im-
provement in a constrained activity, but no changes
in the underlying impairment or activities/partici-
patory performance on rela ted activities.
What Do E xist i ng E viden ce-Base d Repor ts
Reveal About Attentio n Rem e diatio n ?
At the t ime of this art icle, there are t wo compre-
hensive reviews examining the attention remedia-
t ion literature (C ice rone et al., 2000; Park & I ngles,
2001). Park and Ingles (200 1) reported on a meta-
analysi s of the attention reh abilitation literature.
T hey coded a nd analyzed 26 s tudies as direct re-
training s t udies and concluded t h at subject perfor-
mance im proved significantly on tasks t ha t were
trained, but found no evidence of t rea t ment effects
on tas ks that were different from those trained. For
some of the st ud ies re porting a s tatist ical ly sigmfi -
cant im provemen t in one or more measures of at-
tention (e.g., Gray, Robertson, Pentland , & Ander-
son 1992; Niemann, Ruff. & Baser, 1990; Park et al.,
1999; S turm et al., 1997), Pa rk and Ingles s uggest.-
ed t hat t he pattern of im provemen t was a tt ribu t-
able to the acquisit ion of specific sk ills rat her than
to the t r ain ing of attention . It is n ot poss ible to
scr ut ini ze t heir findings using our five key ques-
t ions beca u se the research methodology conscli-
dates stu dies with different subj ect profiles, t reat-
me nt a p proa ches, and outcom e meas ures. Fo r
exa mple, ma ny of the participan ts in t he reviewed
s t udi es h ad severe bra in inj uries res ult in g in im-
pairment to basi c at tentiona l processing; b ut be-
ca use grouped data are presented, it is diffi cult to
examine individual profiles. Similarl y, the grouped
data render it difficult to determine potent ially im-
portant intervention characteris t ics. For exam ple,
the t re atmen ts they a na lyzed were admin istered
an average of 3 1.2 hr; but with a standard deviation
of 32.7 hr. Nonetheless, this s tudy reminds us of the
impo rta nce of analyzing the potentia l transfer of
a ny observed t raining effects to related tasks. It al-
so encourages the field to a ddress definit ion a l am-
biguities s uch as the dis tinction between attention
skills and attention processes . Further, t h e meta-
m v
analysis highlights the question of effect size-if
t here are observed cha nges in eit her skill or
process performance, we must look at the magni-
tude of cha nge to determine clinical relevance.
Cicerone et a l. (2000 ) reported the fi ndings of a
s ubcommittee of the American Congress of Reba-
bilitation Medicine (ACRM) that ana lyzed existing
research addressing CCR interventions for pe rsons
with TBI a nd/or st roke. They selected both t reat-
ment effi cacy st udies (highly constrained, time-lim -
ited research with mostly homogenous sa mples)
and st udies of clinical effectiveness (empirical t re at-
ment evaluations within clinical settings, which re-
flected the actual use of an intervention). They used
a scree ning process resultin g in the selection of 171
articles for inclusion in their review. Thi rteen of
t hese s tu dies were assigned to the ca tegory of re-
med iation of a tten tion deficits. Of t hese 13 st udies .
3 were classified as we ll-des igned, prospective, ran-
domized con t rolled t ria ls (Class I st udies ); 4 were
classified as us ing prospective, nonra ndcmized con-
trols or a clinical series with cont ro ls (Class II st ud-
ies ); and 6 em ployed single-s ubject design or clini-
cal re ports wit hout controls (Class UI st udies).
Trends across st udies were s um marized, describ-
ing the clinical implications of t heir literature re-
view (Cicerone et a l., 2000):
Evidence . . . supports th(~ effecnvenese of attention trai n-
ing beyond the effects of nonspecific cognitive stimulation
fQr subjects ....-ith TB I or stroke during th e poetecurc phase
of recovery li nd reh abilitation .. . Interve ntions should
Include traini ng with different stimulus modali ties. levels
I)f complexity lind response demands. The interven tion
should include thera pist aeti"'iue8 such as mooitoring eub-
jecte' pe rformance, previdlng feedback, and teaching lIu at-
egies. Attention traini ng appears to be more effective when
directed at im proving the !lubj ect:lI performa nce on more
complex, fun ctional tusks. However, th e effecte of t rent-
ment m ay be relatively sm all or Ulsk specifi c. and an eddi-
t ional need exists to examine the im llUet of attention treat-
ment of ADL,I; or fun cnonnl outcomes . (p. 1600)
S ummary of Stu d ie s I nclude d
i n Exi sting R eview P a p ers
In t his section, we utili ze our fi ve key questions as
t he organ izing matrix to review the fi ndings from
t hose st udies reported to ha ve the s t ronges t. re-
search methodology in the existing comprehensive
reviews of at tention remedia tion (i.e., t he Class I
and Class fI st udies discussed in t he review s by Ci-
ce rone et a l., 2000, a nd Park & Ingles, 2001 ). We se-
ANCDS BULLETIN BOARDN OL. 11, NO. 3
lected only those st udies that s pecifica lly evaluated
t he d irect t ra ining of a t tention to s ubject pools that
included s urvivors of t raumatic brain inj ury. Table
1 presents the Class I and Class Il st udies for s u b-
j ects in the acute reha bilitation phase.Table 2 s um -
marizes the Class 1 and Cla ss II st udies for s u b-
j ects in t h e postacu te rehabili tation phase. We
review this older body of literature collectively a nd
then examine th e recen t. litera t u re by d iscussing
the most recen t st udies individuall y.
Reliability for codi ng the six st udies us ing t he
fi ve key qu estions was examined . The fir st t.wo au-
thors independently reviewed four of t.he six stud -
ies (.66) and compared comp leted tables. There was
100% reliability on key poi nts , with two exceptions.
There wa s not a consistent description of the focus
and rationale of treatment (key question 2) be-
tween t he two raters. Fur ther, for several st udies ,
th ere was disa greement abou t whether outcomes
(e.g., res ponses to questionnaires) reflected impair-
ment or activity/ participa tion level ch anges. Wit h
minima l discussion, consens us w as reach ed on
both iss ues.
Who Recei ved th e Interven tion ?
Cons isten t w it h ou r prem ise t.h a t read ers mu s t
scru tin ize resea rch to dete rmin e its re levance to
t heir clinical practice, the wide va riety of s ubject
cha racteristics does not merge nea tly into a singu-
lar pool of s h ar ed demographics. Two st udies (No-
va ck , Ca ld we ll. Du ke, Bergq uist, & Gage, 1996;
Ponsford & Kin sella , 1988) focused on acute reha -
bilitation patien ts, and the rema inin g four (Gray et
al., 1992 ; Niemann et a l., 1990; Soh lbe rg & Mateer,
1987; Strache. 1987) focused on out pa tient clien ts.
On ly t hree of the six a rticles (Nie mann et a l., 1990;
Novack et al., 1996; Ponsford & Kinsella, 1988) re-
ported on s u bjects with only TEl or closed head in-
j ury (C J-U); t he other st ud ies' s ubject pools included
un s pecified "non-t ra u matic brain inj ury" (Gray et
al., 1992), e VA (St rache, 1987), and a neurysm and
penetrating h ead injury (Sohlbe rg & Mateer, 1987).
T hree of t he six st udies exa mined perform ance of
s u bj ects wit h severe or ve ry severe inj u ries (No-
vack et el ., 1996 ; Ponsford & Kinsella , 1988;
Soh lbe rg & Mateer, 1987). On e s t ud y's s ubj ec ts
(Niema n n et a l., 1990) s panned moderate to severe
inj uries, and a not her st udy (Gray et al., 1992 ) de-
scribed t heir participants a s h avin g mild to moder-
a te difficul t ies with a ttention, but d id not provide
information on t he na t ure or degree of actua l brain
inj ury. No inju ry severity in formation wa s provided
in the sixt h study (S t rache, 1987).
PRACTICE GU IDE LINES FOR DIRECT ATTENTION TRAINING xxv

TABLE 1. e las!! I a nd Class n s t ud ies for participants in t he acute re habilitat ion phase.

N ovack e t aI., 1996 Ponsford & Kinsella, 1988

Reference C laBs I C la88 11
C LienblS u bje<:ta
Num ber of eubjecte 22 pairs 10
Cont rols Random assignment to "focused" vs. 16 ortho pat ients (eval ua tion not
"une u-uctu red" remediation treat ment controls )
Diagnosis/et iology T81 CH I
Severi ty "severe" ""very seve re"
Ago m '" 27 .8 (13.2) yrs VB. 26.4 (10.9) yn m '" 24.4 (8.7J yrs vs. 25.8 (7.8) yra,
Education m '" 11.5 (2.4) yra v s, 11.8 (l .G) yrs m '" 11.0 0 .9) yrs vs . 11 .8 (2.0) yrs.
T ime postonset m .. 5.9 (3.3) wks V8. 6.4 (4.9) wke range e 6-34 wks
Dual dxlco-morbidity Not provided Not provided
Cognit ive profile Not provided S ufficie nt senso ry a nd motor function
postinjury for computer tasks
Attention Trai n ing
Focus Focused vs. unstructured computerized Com puter tasks: directed st imulation
activities. Focused tasks targeted vs. independent work. Tasks
vigilance e mphasized vigilance
Rauon ale Focused program will promote more Cognitive benefit of boosting speed and
extensive recovery of nttentional skills selectivi ty of infor matioo processing
Dura tion/freq uency m '" 20.5 (8.81 sessions va. 20.9 () 1.7) 30 min for 15 days without feedback;
sessions . 30 min . 5 x wk 30 min for 15 days w ith reinforce ment
T rea tment setting Rehab hospital Rehab hospital
Providers Masters level educator or Not provided
psycho metrician
Program ind ividualizat ion Advancement for focused group based
on level of accuracy
Add itiona l or concu rrent Cu ing d uring focused activit ies to Feed back, reinforcement, and graphing
cogni tive intervent ion ensu re success of results for one phase
Ou tcome s
Im pa irme ntJpsychom. No s ignifica nt diffe rences in attentional No s ignifi ca nt treatment effects; trends
chan ge skills or general cognitive abilities noted on single s ubject performance
level
Participation level changes No s ignifica nt difference of AOL or No s ignifica nt change on Ra ting scale.
cognit ive FIMs Ceiling effect seen on video
meas ureme nts
Methodology rev iew
Study design PTe- to posttrea t me nt com pa rison Mul tiple baseline across subjects
Relia bility Not provided Not provided
VaJidity Not provided Not provided

A general trend was seen for age, with the ma - m e, 1987), however, included subjects above the
jori ty of subjects falling between the range 0[ 25 to age of 60. Education trends were also seen. The
35 years. Two s t u dies (N ie m ann e t al., 1990; S rr a - s u bj ects of five s t ud ies h ad a t least 9 years of edu-
~. TABLE 2. Class I a nd Class II studies for partid panUi in the POSLacut.e re ha bilitat ion phnse.
G ray e t a J. 1992
Reference Class I
Clien ts/Su bjects
Num ber of s ubjects
Cootrols
17
I'
Diagnosislet iology TB I or nOI1 ·'1'81
Severi ty "Mild- modera te to severe"
attenuonal dysfuncti on
(not severi ty of inj ury)
Age m = 26. 18 (7 .58) yrs vs .
34. 14 08.44} yrs.
Educat ion Not provided
Ti me postcnset Range.? weeks to 10 yrs
Dua l dia gnollisl Not provided
co-morbidity
Cognitive profil e Subjective repo rts of poor
postinj ury concent ration for real-life
situa tions
A t tenti on Train i.n g
F""", Microcom puter delivered
task.: s peed of a ttending
and inform ation processing
Rationale Increased s kills will reflect
improvement in si milar
ne uropsych tests
Nieman n e t al. 1990
C lass I
13
13
'I'Bt
Moderate to severe
28.9 (8. 2) y ra VB.
34 .3 (12.0) yrs.
13.8 (l.8) yra va.
13.7 (:l.5) yrs.
41.0 (2 J.5) months vs.
37. 1 (20.1) months.
Range : 12-72 mon ths
No eubatance abuse iss ues.
No premorbid psych.
admissions
DRS sco re of 100 .
No severe a pahasia
Computer assisted attention
retrain ing program
Ccmpu teriecd attention
tr aining will improve
measures of atten tion but
not memory
Soh lberg & I\laleer. 1987 S t r ache. 1987
C lass D C lass II
• None
Two groups of 15
15
CH I, PH I, ane u rysm H ead tra uma , e VA, other
etiologies
24 hrs to i w ecke LOC Not provided
25-.1 0 yrs. m = 32 yrs (range 20 to 70)
11- 13 yrs 9 or more yrs
12- 72 months Less than 6 months to more
tha n 3 yrs
Not provided Not provided
VIQ a t 80-87; P IQ a t 74-98; Not provided
FSIQ a t 77-85
APT and other com merciaJ Apparatus s upported
produC1.S training ror deficits of
conce nt ra t ion
Repeated st-im ulation of Increased attention skills
attention will im prove will im prove physica l,
im pa ired systems psychological, social, and
vocat ional process or reha b
(ext ra polnted-c-not di rectly
stated )
(co n /ill/US)
 TAB LE 2. continued.

G ray e t aJ. 1992 N iem ann et a l., 1990 Sohlber g & Ma teer, 1987 St rac he, 1987
Reference Clbs I Class I Class II Class U
Duration/frequency m _ 15.35 (2.06) hr over Two 2· hour sessions! 7- 9 training sessions! Twenty 30-min sessions
3-9 wks VB. 12.7(3.8) hra wk x 9 weeks wk for 4-8 weeks ove r 4 wks
recreational com put ing
T rea tment setting Ou tpa tient/postacute Out patient program Post-acute day rrea i me r a Neuro logic rehabilitation
clinic program ce nter
Providers Not provided Not provided Not provided Trained assistants unde r
instruction of
neuropsyc hologist
Program individc alieation Not provided Advancement through Treatment tasks and degree Assigned to group re:
menu of tas ks pe r of d ifficulty re: intake intake test ing. One group
predetermined criteria testing. Advancemen t. per with s tanda rdized
individual criteria tra ining, one group wi th
progress-dependent
ad va ncement
St ra tegy t ra ining or Ye, Ye, Not provided Not provided
feedback
O utcome s
Impairmentlpsychometric improvement on 2 measures S ign ifican t improvement Sign ificant gains in Improvement for bot h
Change of a ttention and in audito ry on 4 measures of a ttention. a ttention skills but not in groups beyond
.....orking memory at No treatment effect visua l processing s pontaneous recovery test
6 months follow-up. generalization to repetition , or direct
dependent variables training of tested
functi ons. ~Sti ght but d ear
advanta ge" for the
baseline and progress-
dependent grou p on
mnemomic functions
Pa rticipat ion level changes Not provided Not provided All s ubjects improved in Not provided
independent living or
return to work stat us
Methodology r eview
Stud y design Pre- a nd posttreatment Pre- and pos ttreatment Single subject mu lt iple P re- lind postt reatment
comparison ccmpariaon baseline across behaviors com pa rison
design
Relia bility Not provided Not provided Not provided Not provided
~._. Validity Not provided Not provided Not provided Not provided
xxviii
cat ion, a nd two s tud ies (Novack et e l., 1996 ;
Sohlberg & Mateer, 1987) reported subj ects with a t
least I year of college. Gray et al. ( 992) did not pro-
vide premorbid education demographics.
Given the heterogeneous na tu re of survivors of
TBI. it. was st riking that only one of the s ix studies
provided comorbidi ty information (Niemann et al..
1990). Descriptions of cogni tive profiles a t the onset
of trea tmen t varied widely. Specific markers such
as IQ (Soh lberg & Ma teer, 1987) and subjective de-
scriptions of distractibi lity (Gray et a l., 1992) were
used. Studies al so reported results of in take test-
ing, bu t s pecifi c deta ils were often lacking and in -
terpretation was left to the reader (e.g., Srrache,
1987).
What Com p r ised the A ttenrian Remediation ?
Details of treatme nt s pecifics va ried widely a mong
the six stud ies and ranged from specific descrip-
tions of tasks (Pons ford & Kinsella , 1988) to gener-
al trea tment profi les (Strache, 1987). Compute r-
ized attention programs were used in a ll six studies
a nd were eithe r the sole focus of the t rea t me nt
(Gray et aI., 1992; Novack e t a l., 1996; Ponsford &
Kinsella , 1988; S trache, 1987) or were su pplement-
ed wit h other tasks (Niemann et e l., 1990; Sohlberg
& Mateer, 1987).
Studies also varied in the explicitness of descrip-
t ions of the rationale and focus of t he treat men t. In
broad strokes, each of the studies used exercises de-
sign ed to stim ula te discrete types of attention, with
the expectation tha t im proved perfor mance would
follow repetitive dri ll. The d uration a nd freq uency
of intervent ions, however, varied between the st ud-
ies. Specific attention tasks in at least one study
lasted for 5- 10 min (Nieman n et a l., 1990). Trea t-
ment sessions ranged from 30 min (Novack et al.,
1996; Ponsford & Kinsella , 1988) to 120 min per
session (N iema nn et a I., 1990) a nd varied from one
to two sessions da ily (Sohlberg & Ma teer, 1987) to
twice per week (Gray et al. , 1992; Niemann et nl..
1990 ). Ove ra ll length of trea tmen t varied from 4
(Niema n n et al ., 1990) to 9 (Novack et al., 1996)
weeks. As mentioned, trea tment occu rred in acute
rehabilitation settings for two studies (Novack et
aI., 1996; Pon sford & Kinsella , 1988) and in ou tpa-
tient clinics or settings for t he remaining four.
Trea tment providers ranged from "tra ined assis-
tants" (Niema n n et a l., 1990) to teachers with grad-
uate degrees (Novack et a l., 1996). No information
was given abo ut the providers in the othe r fou r
stud ies.
AN CDS BUL LETIN BOARDIVOL. 11. NO. 3
Individ ualized trea t ment plans fo llowing screen-
ings or extensive eva luations were created in three
st udies (Novack et 81.. 1996; Sohlberg & Mateer,
1987 ; Strache, 1987). A hierarchy of exe rcises based
on assumed in creases in diffic ulty was used in an
six st ud ies, although criteria for advancemen t to
more d ifficult tasks varied from ope ra tiona lly de-
fined parameters (N iema nn et nl., 1990; Sohlberg
& Matee r, 1987) to ind ividua l clinician decision
(Novack et al., 1996 ) to ad vancemen t cont ingen t on
success at each level of diffic ulty (St rache, 1987).
S peed of processing was a key compone nt in t he
tasks of a t least four of the studies (Gray et el..
] 992: Ponsford & Kinsella, 1988; Sohlberg & Ma-
teer; 1987; Strache. 1987). 'Two st udies (Novack et
a1., 1996; Ponsford & Kinsella, 1988) focused on vig-
ilan ce and se lective attention skills, al though A
small percentage of Novack et al.'s su bjects ad-
vanced to alternating atten tion dr-ill. Su bjects from
the re ma ining fou r studi es were exposed to alte r-
nating and/or divided a ttention d rill for at least
portions of their treatment protocol.
S tudies varied on rein fo rcemen t, feedback, and
s tra tegy t raining. Novack et al. (1996) strategically
ad ded cues to guara ntee subjec t success. Scores re-
ported to s ubjects included formal re porting {Nie-
mann et al.. 1990), discrete comp ute r screen dis-
plays of task scores (Pon sford & Kin sella , 1988),
com prehensive feedback (Niema n n et al., 1990 ).
a nd graphi ng of per formance changes over time
(Ponsford & Kinsella, 1988). At least two studies
used tasks as a springboard for strategy training
(Gray et nl., 1992; N iemann et al., 1990). One task
in Gray et al'e. study focused on overt "ver bal regu -
la tion" train ing (i.e., coaching th e client to ve rbally
self-cue as an attention strategy) during a n a lter-
na ting a tten tion task.
What Are the Ouscomee of Intervention?
The s ix a rticles offer excellen t opportu nit ies to ex-
a mine the compl ica ted issues related to the su bjec-
tive va lues u nderlying the cl aim of a successful or
un successful outcome. In genera l, the stud ies tend-
ed to e mploy a ba ttery of measu remen ts to assess
cha nges rel a ted to in terven tio n. Before we ca n
ma ke sense of the complexities behind outcome de-
cis ions. we must discuss the differen t tools a nd a p-
proaches used for measurement.
All six studies assessed some aspect of im pa ir-
ment via standardized testin g, and t his ranged
from extens ive neu ropsychologica l test batte ries
(Gray et al., 1992; Sohlberg & Ma teer, 1987; Stru-
che, 1987) to a sma ll number of s pecific tests su p-
PRACTICE GUmELINES FOR DI RECT ATTE NTION TRAINING
plemented by speed and accu racy measurements of
nonspecified tasks (Novack et al., 1996). Use of
these cognitive measures also varied and ranged
from pretreatment/postt rea tment. comparisons to
re peat administration of specific tests over the
span of treatment (Ponsford & Kinsella . 1988;
Sohlberg & Ma teer, 1987J. Some stud ies used tests
assu med to predict real world dem ands (Niemann
et a l., 1990). Other tests were selected specifically
because they were similar to treatment tasks (Gray
eL al., 1992; Niemann et al., 1990). Four of the six
studies used the neuropsychological tes t PASAT as
part of their measurement tools (Gray et al., 1992;
Niemann et al., 1990; Novack et a l., 1996; Sohlbcrg
& Mateer, 1987), al though the manner of lest de-
livery (e.g., taped presentation vs. live voice vs. re-
vised tests) was not specifi ed. Clear compa risons
with othe r outcomes are fu rt her tempered by the
fact that. testing was incomplete (Novack et a t..
1996) and that raw d ata were analyzed differently
between stud ies (e.g., Gray et a l., 1992; Sohlberg &
Mateer, 1987). Other incons istent use of impair-
ment leve l testing was also eviden t, Studies dif-
fered in using parallel versions of repeated tests,
includi ng the use of versions supplied by the pub-
lisher (Sohl berg & Mateer, 1987) or created by the
resea rch a dminist rators (Ponsford & Kinsella,
1988). One st udy created a unique way to score lest
data (ra t ios of accuracy scores to completion time)
in an a ttem pt to ca pture s ubjects' changes of per-
formance that were not reflected in formal scoring
(e.g., Novack et al., 1996).
Finding effective methods to measure relevant
change poses a sign ifica nt challenge. As discussed ,
we need to understand the relevance of neuropsy-
chological (i.e., im pai rmen t level) testing for deter-
mining success or fail ure of interve ntion. The re-
port of positive gains following direct attentio n
training has been criticized and dismissed as treat-
men t artifact by some a uthors (Park et al., 1999),
whereas others (Gray et al., 1992; Sohlberg & Ma-
teer, 1987) specifically chose neu ropsychological
measurements fo r thei r simila rity to the training
tasks and based their interpretation of find ings on
test score im provement. Additionally, philosophical
differences are seen when results of cognitive tests
are used to validate or reject the effectiveness of
the training. Two stud ies (Gray et al., 1992; Nie-
mann et al ., 1990) repo rted or implied poor gene r-
a lization oftraining skills when secondary tests did
not improve over the pretreatment-posttreatment
cycle. Sohlberg and Mateer (l987) viewed the a b-
sence of change in a n untreated skill area (i.e.. the
double disassociation observed with un changed vi -
seal-processing scores a nd improved attention
scores) as positive proof of the effectiveness of at-
tention skills training.
In addition to the impairment level testing, three
s tud ies incor pora ted activity/participation level in -
fo rmation in their assessments. The F unctional In-
depende nce Measure was used on a su bse t of one
study (Nova ck et al., 1996), Ponsford and Kinsella
(1988) created a rating sca le of dist ractibility a nd
also scored subject distractibility during a n un-
structured work task. Sohlberg and Mateer (1987)
provided anecdotal info rmation abou t community
rein tegration status after treat ment was delive red .
Althou gh it may be impossible to combin e the
outcomes in to a bin ary decision on the efficacy
a nd/or effectiveness of a tten tio n training, it is help-
ful to look a t t he outcomes of the ind ividual studies.
The acute rehabiliuntion studies <Novack et a l.,
1996; Ponsford & Kinse lla , 1988) reported improve-
ment in their subjects, bu t based on control group
performance or subject. changes during baseline
phases, attributed it to spon taneous recovery. and
not to treatment. Both studies' participation level
testing fai led to reflect benefit, although ceiling ef-
fect problems were reported for Ponsford and Kin-
sella's use of videoua ped measurements.
Successful postint.ervention improvement was
reported in the ou tpatient studies. Gray et al .
(J992) reported their subjects s howed improve-
ment in storage an d manipulation of numerical
materia l in working memory, bu t. the s uccess
emerged at follow-up testi ng only. They reported
improvements in pictu re completion and speed of
processing skills, but admitted that cont ributions
of p remorbid IQ a nd lengt h of time s ince injury
could not be ruled out. They a lso reported that pe r-
fo rmance decline in the cont rol group could have
artificially s kewed the improvements of the expe r-
imental grou p. Strache ( 1987) reported tha t su b-
jects demonstrated progress in concentration, psy-
chomotor function. and intellectual and memory
function, and reported even stronger memory im -
provement for a s ubset of subjects whose progres-
s ion through thei r t rea tment hierarchy was contin-
gent on their accu racy of performance, and not
a rbit rarily decided. Niemann et al . (1990) reported
their subjects perform ed better on four measures of
a ttentio n in their evaluation battery, includi ng Tri-
a ls B and a cancellation Lest. Sohlberg and Mateer
(1987) reported success ful gains by all subjects on
the PASAT a nd anecdotal reports of success in re-
tuming to work or ind ependent living.

Des pite the successes of the four studies, im pair-
ment level improvement does not. easily transla te
into clinica lly meaningfu l improvement. Even if
impairment test ing should prove to be ecologica lly
va lid in predicting improvements in independence
or community rein tegration , clini cian judgment
and client/family input would be requir ed to deter-
mine if the cost-benefit ratio was acceptable. This
posthoc information is not provided in the studies.
Similarly, the re was a lmost no attempt to measure
mainte nance of effects over time in any of the stud-
ies. Decisions about clinical relevance and further
confidence in the success of the intervention would
be easier if activity/participation-level assessme nt
tools had been used.
Are There Methodological
or Cli n ica l Concer n s ?
We acknowledge that it is easier to "cr iticize than
to do" and that a ll studies, regardless of flaws, con-
tribu te to the developme n t of E BP. None theless,
sa vvy consumers mus t. ident ify t.he stre ngths a nd
weaknesses of the research as they conside r incor-
porating new trea t ment ide as in to t.heir clinical
pract ice.
We classified the six studies in to three Class I
(Gray et al., 1992; Niemann et a l., 1990; Novack et
al., 1996) and three Class n (Pons ford & Kinsella ,
1988; Sohlberg & Mateer, 1987; S trache, 1987)
st udies. The Class I studies centered on pretreat-
ment versus postt rea t men t measurement compar-
isons; although controls were used, none of the
three had a "no-trea tment" group. Novack et al.
(1996 ) compared their experi me ntal ' focused stim-
ula tion group" with an "u ns truct ured intervention
program ." Niemann et a l. (1990) com pared their ex-
per imental attention training gro up wit h a memo-
ry training group in a repeated measurement de-
sign, whereas Gray et a l. (1 992) compared a gro up
who received compute rized trai ning to a grou p that
used computers fo r recrea tion only. The Class n
studies included a multiple-baseline , across-sub-
jects design (Ponsfo rd & Kinsella , 1988; Sohl berg &
Mateer, 1987). Strache (1987) used two experimen-
tal groups and a control gro up tha t received "nor-
ma l clinica l intervent ion." The two experimental
groups received additiona l computerized trainin g
but with different advancement criteria (arbitrnri-
Iy deter mined vs. con tinge nt on accu rate pe rfor-
mance). Subjects were assigned to the groups based
on performance in "extensive diagnostic pretest-
in g." In a ll the studies, the absence of a no-treat-
ment group may undermine the r igor of the design,
ANe DS BULLETIN BOARDNOL. 11, NO.3
especially when Questions of spontaneous recovery
or trea tmen t effects are rai sed .
All six st udies referenced their attempts to con-
trol for the many variab les that complicate clinical
research for TBI su bjects. Typica l controls includ ed
gender, age, seve rity, education, premorbid IQ, a nd
so forth. Time postoneet was a lso routinely ad-
dressed, a nd markers r anged from wee ks to
months to years. As previously noted, the studies
reported no improvements in attention training in
the acute rehabilitation phase. Inspection, howev-
er, shows that one st udy (Ponsford & Kinsella ,
1988) included acute car e su bjects who received in-
tervention between 7-8.5 months following their
inj ury. This time frame over laps with that of sub-
jects of a t least two of the four postacute stud ies
(Gray et al ., 1992; Strache, 1987).
Further methodological anomalies are noted in
the acute rehabilitation studies. Novack et al. (996)
were unable to collect baseline measurements on
their subjects with the exception of two tests, for
which they devised "new methods of scoring" to cap-
ture performance. Also, their participation level
measurement, FI}.f , was collected in the m.iddle of
the intervention phase; it was unclear if the da ta
were collected on 12 or 24 of the 44 subjects. Pons-
ford and Kinsella (1988) created a ra ting s urvey of
attentional behaviors in da ily activities and incorpo-
rated video-based analysis of d istraction in a wor k
module. Ceiling effects un dermined the effective-
ness of the video task, and no informa tion is provid-
ed on the design and reliability of the ra ting scale.
Measure ment issues also a rise when the absence of
norma tive data is noted for the newly crea ted par-
allel versions or one of their key measuremen ts .
Finally, despite both studies' efforts at control-
lin g for d ifferences between treatment and control
groups, info rmation is not provid ed a bout the be-
ha viora l level or com pliance of the s ubjects at the
time of the interve ntion, al though Pons ford and
Kinsella (1988) imply that their subj ects were no
longer in posttra umatic a mnesia . Even with con-
trols for age. sever ity, a nd time postcnset, pa rti ci-
pation in struct ured tasks could vary widely de-
pending on t.he level of agitation, confusion, a nd
fatigue. It is easy to imagine very different out-
comes for s ubjects who received the same treat -
ment but differed greatly in their motivation a nd
poten tial for treatmen t participation.
Su bj ect recruitment and pa rticipation issues
must. be addressed for the four outpa tient studies
as well. Although we a re in terested in attention
training delivered to surv ivor s ofTBI, we must ac-
cept tha t researchers in only one of the studies
PRACTI CE GUIDE LINES FOR DIRECT ATTENTION TRAININ G
used a subject pool com prised solely of persons with
that etiology (Niemann et al ., 1990 ). Subjects fo r
the other stud ies included survivors of aneurysm or
other CVAs (Gray et el., 1992; Sohlberg & Ma teer;
1987; Stracbe, 1987), those with pe net rating hea d
woun ds (Sohl berg & Mateer, 1987), a nd those who
had othe r "neuros urgical procedures" (Gray et aI.,
] 992), Non-TBI su rvivors provid ed 50% or more of
the s ubj ect pools of th e t h ree s t ud ies. If we assume
d ifferences in presentation of inj ury, pa rticipation
in therapy, and courses of recovery between the eti-
ologies, Doe mu st ackn owledge that evidence out-
comes resul ts may be skewed or a t best may be di f-
fi cult to in terpre t. Con trols' age may also need to be
conside red. In Gray et al. (1992), the control group
is slightly older, but more im portan t, the standa rd
deviatio n spread is signifi ca ntly wider. Given the
authors' in terpretation that some positive results
of the study may be ca used by a declin e in the con-
trol group's performa nce, this disparity must be
ackn owledged.
A disappointing methodological fl aw present in
all six reports reviewed in this subsection is the
lack of inter- a nd in tra ra ter reliability. This infor-
mation is lacking in the scoring of objective a nd
subject ive assessmen t tools, the collection of data
during treatment sessions, a nd j udgments rela ted
to the interpreta tion of outcomes. Validity issues
are broader in scope and more diffi cult to qu antify,
ye t the a bse nce of informa tion from the a uthors is
once agai n noted. To be valid clinica l intervent ions,
the treatment tasks should be replicable across set-
tin gs . The six studies here diffe r in the details pro-
vided and range from a sentence-length descri ption
of the interven tion (Novac k et a l., 1996; Strache,
1987 ) to a menu of treatment areas from wh ich a
subject's persona l treatmen t plan was created
(Sohl berg & Ma teer, 1987). Replicability of the
specifics of eac h of the stud ies would be impossible.
Simila rly, re-creating the dose (e.g., 120-m in ses-
sions in Nie mann et al., 1990) or length of trea t-
ment (e.g., 9 weeks in Gr ay et a l., 1992) fo r a clien t
who was dependent on third-party payment of re-
ha bilitation se rvices might be di ffi cult.
Are Th ere Trends A cross S tud ies
Publi.s h ed Prior t o 1999?
Our review, in addition to Cicerone et al.s (2000) and
Park and Ingles' (200 1), suggests there is evidence in
the literatu re of improvement in a ttention-based
skills with direct training; however, the studies that
reported improvements are open to interpretation.
Exe rcises may promote the acquisition of specific
skills and outcomes may be task-specific (Cicerone
et al., 2000; Park & Ingles, 2001). The review of the
older a tten tion literatu re and t he eq uivoca l out-
comes directed us to scrutinize specific intervention
characteristics and outcomes in more recen t efficacy
articles. For instance, we looked to more contempo-
rary stud ies to eval ua te whether treatment effects
were associ ated with. incorpora ting strategy training
in the a tten tion remedia tion, or whether those stud-
ies that ma tched a ttention intervent ion tasks to in-
dividual subject pro files had different outcomes t.han
the studies tha t delivered a s ta nda rd program to
every subject. S imila r ly, we were in te rested in
whether more recent studies provided outcome re-
ports that shed ligh t 0 11 t.he task-specific nature of
attention training.
Recent Attention Efficacy Studi es
A literatu re search was cond ucted to identify arti-
cles published a fte r 1999 (i .e. , after the Cicerone et
al ., 2000 report) that eva lua ted the effi cacy or ef-
fectiveness of a ttention t rai ni ng. Data ba ses that
we re sea rched incl uded PSYCH INF O, MED·
LINElPubMed, Eric, and CfNAHL, us ing combina-
tions of these key words: bra in inj ury, closed head
inj ury, a ttention, remedia tion, reha bilitation, and
t ra i ning. Tw en ty-seven a rticles were iden ti fi ed .
These studies were reviewed a nd se lected based on
the following crite ria : (a) writte n in English, (b) ex-
perimentally eval uated the direct tra ining of atten-
t ion rem edia tion to ad ul ts, (c) excluded s tud ies
dealing with left. hemispa tial ina tten tion, (d) s ub-
jects included people with t raumatic bra in inj ury,
a nd (e) outcome da ta were reported in the study.
This screen ing process revea led three studies: Ci-
cero ne , 2002; Pa rk et a l., 1999; Sohl berg et al.,
(2001) (see Table 3).
The rust two a uthors independently revi ewed
the a rticles and coded them as Class II st udies. The
a rticles were reviewed using the fiv e key qu estions.
Again , there was h igh agreemen t between the two
readers in the codin g. The re wa s discussion about
how to describe particular feat ures (e.g., wh ether
outcomes were classified as impairment or partici-
pation ), bu t the conten t of classifi cation was consis-
tent, and reliability was deemed acceptable. Ta bles
1-3 summ ar ize literatu re according to the first
four key qu estions. The fi fth key question (Are
t here clinically a pplicable trends across the litera -
tu re, is addressed in Ta ble 4, with the generat ion of
practice guidelines based on the evidence.)
Park, Proulx, and Towers (1999). These investi-
gators evaluated the effectiveness of the com mer-
TABLE 3. Recent s t udies l post- l 999J.
Parka e t a l , 1999
Refe rence C la 8s 11
Clie n tslS u bjeets
Numbe r of s ubjects ,.
Controls Culled fro m previously
collected data. Selectively
ma tched for age and
education
Diagnoaisse ric logy TBI
Severity Sev ere
Age m = 37.3 yra (2.66 S E)
Education No repo rted d ifference
betwee n gro ups
Time postonset Less than Iyr to 4 ynJ
Dual diagnosis! Not provided
co-mo rbidi ty
Cognitive pro file "Slightly above ave rage"
postinjury
Attention Training
F'ocu. APT with hie rarchi ca l
exercises
Rationale Repeated s timula rien of
attention win improve
impaired systems
Du ration/frequency 40 hours (media n) for
7.2 mon ths (mea n ) "a bout
2 hours" pe r session
Treatment sett in g Not provided
Providers Psychologist
Program individ ualiztion Repetition of aercises
when 3 or more e rrors.
Adjunct counselin g re:
salie nt cli nical issues
St ra tegy t rai ning or Yes.
feedback
Out com e s
ImpairmentJpsycbom . No significant diffe rence
<hange in a tten tion or working
memory im provements
between 2 groups
No change in depreeeicn .
xxxii
Sohlberg e t aI., 200 1
Class 1J
1.
CTOS8Qver design
TBI, a noxia, tumor
LOC range : Null to 7
months
m = 33. 1 yra va. 38. 1 yra
11 yrs ve. 13 yrs
l to 5 yrs
3 with cd iaeuee: 4 with
mood medica dcna
Large ra nge o n
neuropsychological testing
APT w ith h ierarchical
exe rcises
Repea ted stimulation of
a ttention will im prove
impaired systems
APT: 3 hrslwk X 10 wkB;
education: 1 hTI
wk X 10 wits
Un iversity clinic
Certified S LP or
supervised graduate
s t ude nt
Tasks chosen to m a tch
a tte ntion profile
No d uring APT ; yes d uring
brain inj ury ed ucation
At tention im proved efter
APT (PASA~ Stroo p a nd
Trails b for low vigilence
subjects). Significant
Im prove me nt for aspects
of working memory
Cicerone, 2002
C lass U
•
•
TB I
Mild
m = 31 VB. 34 .75 yra
m = 15.25 yrs
m = 8 .25 mon ths
Bxclueion for those with
significant history
Im pa irme n ts on 216
ad mi t teste plus
s ubjective complain t
Working me mory tasks
combined with s trategy
t raining
Teaching conscious use of
strategies to boost
allocation of resources
and manage speed
demands
60 mnlwk X 11- 27 wks
Outpatie nt clinic
Not provided
Timing and tasks varied
ret client need
Three q ua rters of clie nts
improved on atte ntio n
tests
(con t i n ues)
PRACTICE GUIDELINES FOR DIRE CT ATTENTION TRAIN ING
TABLE S. (COll ti nucd)
P a rks e t a l., 1999
Refe r e n ce Cluss II
Participation level Not provided
changes
Meth odology review
St udy Des igo Pre- postcom pa riecn
~li al>i l it)l Not provided
Not provid ed
cially available Attention Training Program (APT)
(Sohl berg & Ma teer, 1987). Da ta were presented 0 11
16 individuals with severe TEl who were beyond
the period of spontaneous recovery. Subjects were
identified by "specia lists" as having a profile sug-
gesti ng they would benefit from treat ment. De-
scri ptions of individual s ubjects we re not provided .
Remediation consisted of struct ured API' exer-
cises that were administered until a subject made
no more than two errors. Program administration
was standa rd, and exercises were not selected to
match a s ubject's ind ivid ual attention profi le.
Treatment included the provision of feedback
about performance and d iscussion of error pat-
terns. Ai; the program proceeded, participants were
also educated about different types of atten tion,
and pa rall els between difficulties of daily living
and problems performing particular APT exercises
were highlighted. Su bjects typi ca Uy received twen-
ty 120-min sessions spread over 29 weeks; thus , th e
trainin g generally occurred less than once a week
for more than 7 months.
Outcomes (question 3) were measured by com-
paring pre- and posttreatment performance on two
neuropsychological tests (PASAT a nd Consonant
Trigrams). The PASAT was used because it is as--
sum ed to be sensitive to attention impairments;
however, the a uthors hypoth esized that the recall
measure of the Consonant Trigra ms would n ot be
affected by the types of attention addressed in the
API' program . The Beck Depress ion Inventory was
used to assess the im pact of attention training 011
mood. Results showed that the TBI grou p who re-
ceived a tte nt ion t raining improved on both of the
neuropsychological tests, but not on t he Beck De-
pression In ventory Measurement of maintenance
of effects was not addressed.
xxxm
Sohlberg et aI., 200 1 C icerone, 2002
Cla ss II C la8s D
Memory/attention AJl of experimental group
improvements per surveys resumed vocation or
and in terviews social roles
Crossover with in subject Pre- postcomparison
experimental design
Provided for rati ng scales Not provided
and interviews
Not provided Not provided
The study design (q ues tion 4 ) compared pre- and
posttrainin g test scores to test scores of age- and
educa tion-matched controls culled from d ata from
an unrela ted research project undertaken 11 years
prior to th e study. These control s ubjects were ad -
ministered the two neuropsychological tests on two
se pa rate occasions over the course of a week in -
stead of the 7-month in terval of the su bj ects. The
con trol gro up data revealed improvement on the
PASAT but not on the Consonant Trigrams. where-
as the subjects with brain injury who had received
APT im proved on aspects of both tests. individual
test performance da ta were not provided; thus, it is
not possible to a nalyze performance within s ub-
jects. The a uthors interp re t their fin dings to sug-
gest that APT facilitated learning of specific ski lls,
bu t not im provement of damaged a tte ntion fu nc-
tions.Reliability a nd va lidity conce rns, and mea-
suremen t issues, were not discussed .
S oh/berg, McLaughlin, Pavese, Heidrich, and
Posner (2001). Th is grou p of investigato rs pub-
lished an effi cacy study of 14 postacute clients with
mild -severe brain inj uries. who exhibited impaired
attention a bilities as determined by neuropsycho-
logical evaluation. Group heterogeneity was report-
ed, including h isto ries of substance abuse, depres-
sion, and/or ongoing litigation.
ln tervention consisted of APT attention exercis -
es selected for each su bject based on the results of
their particular neuropsychological profile. For ex-
a m ple, a s ubject dis playin g particul ar di fficulty
with s ustained and selective atten tion worked on
therapy tasks designed to target these areas at an
initial level where he achieved 70-80% accuracy.
Individu a ls received 24 hr of attention train ing ad-
ministered in three GO-min sessions each week for
a total of 10 weeks. Attention drills were grouped
by specific attention a bilities (e.g., s ustained etten-
XXXI V
tion ). Explicit performance st ra tegies were not pro-
v ided , a nd instructions or activities to foster gener-
alization to real world tasks were not offered. Sub-
jects a lso received a single 60 -min session each
week devoted to brain inj ury education for t he
same number of weeks as the APT atte ntion drill
work. Education consisted of a combi nation of top-
ics selected from a menu of choices in combina tion
with s upportive listening (a "check-in" for how the
week was going) a nd relaxa tion training.
Outcomes fo r the two intervent ion programs
(APT and Brain Inj ury Educa tion) we re measured
and compa red using both impairment- and activi-
ty/participa nt-based measures. Impairment-level
me asur es we re obtained us ing a battery of neu-
ropsychological tests selected to assess different. a t-
tention networks (vigilance, orienting, working
me mory, and executive functions ). Tests were ad-
minis tered before and after each of the two in te r-
vention phases. Activity/participa tion measures
were obtained usin g sta ndardized question naires
and stru ctured in terviews to assess subjects' per-
cept ions of t he ir neuropsychologica l and psychoso-
cial performance in daily life.
A crossover design wa s used, in wh ich hal f of the
subj ects received the attention training prior to the
Brain Inj ury Education and ha lf of the subjects re-
ceived the opposite orde r of treatment. This design
a llowed subjects to be used as their own controls.
Based on responses to structured interview and
perform ance on neuropsychological testing, im-
proveme nt in complex attention abil it ies te.g..
working memory, alternating attention) were see n
following APT. In contrast, there was little s pecific
improveme nt in basic a ttent ion (e .g., vigilance or
orienting abilit ies ) following APT ad ministration.
Brain Inj ury Education was most effect ive in im-
provin g self-re ports of psychosocial function . lm-
proved PASAT scores were found to be correla ted
with self-reports of im proved attention on stru c-
tured interviews. It was noted tha t vigilance level
influenced the improvement resulting from atten-
tion training, in that those clients with higher ini-
tial vigilan ce improved more on mea sures of execu-
tive a ttention. The au thors in terpreted t heir
findings as suggestive that APT was effective in
improving working memory or complex attention
in clients with intact vigilance. They do not address
other poss ibilit ies fo r improvements, such as sub-
jects learning some type of behavior or skill from
performing APT exercises that resulted in in-
creased performance on specific neuropsychological
tests.
ANCDS BULLETIN BOARDNOL. I I, NO. 3
Relia bility and va lidity issues regarding the
sta ndardized qu estionnaires and structured inter-
views were indirectly and di rectly addressed. The
three surveys administered to the s ubjects were
from previously published articles and, in two of
the three, a ns wers were collected from the subjects
and a member of their famil y. Specific information
on the structure of the interviews, the transcrip-
tion, coding by najv e read ers, and data analysis a re
desc ribed .
Cicerone (2002). This investigator recently evalu-
ated t he effectiveness of a n intervention designed
to address attention deficits following mild trau-
matic brain injury (M'rBI). Treatment pa rt icipa nts
consisted of a convenie nce sam ple of patients re-
ferred to a postacute brai n inj ury rehabilitat ion
progra m based on a diagnosis of MTBI. Four sub-
jects rece ived a ttention remedia tion a nd four
served as a control group based on thei r inability to
receive trea tment . Neithe r criterion for MTBI nor
individ ua l s ubject data wa s presented. The subjects
were ma tched closely for age, ge nde r, educa tion,
and months postinj ury, with all a t least 3 months
postinjury. AU s ubjects had to meet cri teria for sig-
nificant im pa irment on two out of six attention
measures; however, the treatment group partici-
pants were init ially tess impai red than the control
grou p.
The treatment foc us in this study varied from
the other two intervention studies reviewed in this
section. S imilar to the Sohlberg et. at. (2001) study,
the author employed h ierarchically orga nized a t-
tention remediation tasks targeting complex a tten-
tion skills via working memory tasks that we re tai-
lored to match the specifi c attentioual profi les of
the individ ua l clients. However, unlike both of the
afore men tioned studies, the focus was on using the
atte ntional tasks as a method for training in the
use of metacognitive strategies such as verbal me-
diation. rehearsal , anticipat ion of task demands,
self-pacing, and self-monitoring. The interven tion
emphasized the conscious a nd deliberate use of
such strategies to increase the pa rticipa nts' ability
to all ocate their a ttention resour ces and control the
pacing of task performance. Hence, although there
was repetitive administration of attention exercis-
es, stra tegy training was a primary emphasis. The
schedule of trea tment was 1 hr per week for 11-27
weeks.
Treatme nt outco mes included neu ropsychological
measures, self- rating for perception of change, and
informal reports of changes in status fo r vocational
and social roles. Impairment-based measures com-
pared pre- and posttreat ment scores on a number of
PRACTICE GUIDELINES FOR DfRECT ATIENTION TRAINING
attention tests with the performance of the DO--
treatme nt control group. Res ults showed tha t three
of the four participants improved significantly on
the a ttenti on tests, but none in the cont rol group did
so. The t rea tment group also improved significantly
on their se lf-re port of a greater reduction of atten-
tion difficulties in comparison to no change in the
cont rol grou p. The author anecdotally reported that
a ll of the treatment gro up participants returned to
previous vocational a nd social roles, but none of the
comparison gro up participants did so during the
same period. Change was attributed to im proved
st rategy use ra t her than im proved unde rlyin g a t -
tentiona! processing, although there is no attempt
to parse out these facto rs.
The st udy used a prospective, case-comparison de-
sign with groups of fo ur individuals. StatisticaJ
analyses were thus performed on a very small sam-
ple. The author reported that the rating scale was
developed for t he st udy and that formaJ psychomet-
ric review had not been completed before use. Relia-
bility of measuremen ts was not discussed. Individ-
ual subject profiles were not provided; th us, it is
difficult to analyze poss ible threats to internal valid-
ity. One threat to interna l validity resulted from the
fact that the treatment. group performed better ini-
tially on attention measures. which could indicate
they had more cognitive resources and, t herefore.
j ust needed the strategies to facilitate imp rovement.
S u m mary of R ecent. Effica cy S t ud ies
S ubject variabi lity was quite large across st udies.
Park et a l. (1999) stated t hat s ubjects had seve re
bra in injuries, whereas the s ubj ects in Cicero ne's
study (2002) were all d iagnosed with MTBT. The
subjects in t he Soh lberg et 31. (2000) st udy spa nned
the range from mildly to severely impaired. All su b-
jecta we re reportedly beyond t he period of sponta-
neous recovery. All three st udies ad ministered hi -
erarchically organized a t tent ion drills, with two of
them (Cicerone, 2002; Park et al .. 1999) adding a
s t rategy feedback component. One of the studies
reported a low intensive t herapy regime (less t han
once per week), with a protracted time of service
(Park et al., 1999). In the other two studies, su b-
jects were treated at least week ly, with the su bj ects
in the Sohlberg et a l. (200 1) study receiving t hera -
py 3 hr per week, whereas Cicerone (2002) treated
subjects about 1 br per week. These same two st.ud-
ies individ ualized t he selection of a ttention tasks.
To measure outco mes, Cice ro ne (2002) a nd
Sohlberg et al. (2001) used im pairment- and activ-
ity-based measures, whereas Park et a1. ( 1999)
xxxv
used impairment measures only. All three st udies
reported changes on attention tests in the group re-
ceiving attention t raining. The interpretation of
these findings, however, differs widely between the
t h ree st ud ies. Park et al. suggested that the
changes demonstrated by th eir subjects were not
signifi cantly di fferent from t hose in the nonbrain -
inj ured control group and t hat the changes were
most likely d ue to s pecific practice on tasks t hat re-
semble t he ou tcome measures. Sohlberg et al. sug-
gested tha t t he profile of change on tests in t he
neuropsychologica l battery su pports improved cog-
ni tive function ing, s pecifically in complex atten-
tion/exec utive fu nct.ions a nd working memory
processes. Cicerone cla imed tha t. t he benefits of h is
trea tment were due to pa rticipants' improved a bil-
ity to compensate for residual deficits a nd adopt
stra tegies for more effective allocation of t heir re-
maining a ttentional resou rces .
The two st udies reporting more robust changes
following t rea t ment, incl uding improvement in sub-
jects' daily functioning (Cicerone. 2002; Sohlberg et
al ., 2000, shared the following features: (a) individu-
alized attention exercises. (b) treatment sessions
that were 1 hr (vs. 2 hr) in duration, (c) at least
weekly treatment sessions, (d) outcome measures
that included a range of different tests sensitive to
attention and working memory. and (e) outcome
measures that included activity-based measures us-
ing client self-report. d ata. Additionally. examination
of olde r literature in conj un ct ion with this current
lite rature suggests t hat t he inclusion of strategy or
metacognitive tra ining, as part of direct a ttention
training, increases trea tment effectiveness.
WHAT HAVE WE LEARNED
FROM THE LITERATURE?
The answe r to t h is q uestion lies in our in terpreta -
tion of t he evidence. Unfortun ately, studies evalu -
ating outcomes following cognitive intervention are
not as straightforward as pharmacologic st ud ies.
T he heterogeneity inherent in the TEl population,
cou pled with t he s trengths and limitations unique
to each se tting and practitioner and the range of
opin ions regarding what constit u tes meaningfu l
cha nge, makes it di ffi cult, perha ps impossible, to
design st udies with clean, unequivocal outcomes. It
is hoped that because this a rticle has been written
by a committee of resea rchers and clinicians incor -
porating a b roa d range of perspectives, the bi as
tha t occu rs when in terpret ing ev idence will be tem -
pered . In general , we fool con fiden t in ou r assess-
xxxvr
ment that cer tain aspects of a t ten tion training a re
he lpful in imp rovin g attention performance in
some adu lts with TB I.
Recogni zing the difficulty inherent in interpret-
ing the evidence, our ta sk bas been to examine t he
a tte ntio n t raini ng literatu re for evidence that reo
sponds to the ques tion "when does attention train-
ing facilitate the greatest cha nge and fo r who m?"
We conclude with a set of recomm endations fo r im-
plementin g direct atten ti on train ing . Pr actice
guidelines are recommendations for pa tient man -
age ment reflecting moderate clin ica l certainty, usu-
ally evidence fro m Class H experiments or a strong
conse nsus of Class III evidence (Miller, Rosenberg,
Gelinas. & the ALS Pract ice Parameters Tas k
Force, 1999). The suggestions offered in Ta ble 4
were generated from our review of t he attentio n
training literature. Given t he un even and incom-
plete nat ur e of the experi me ntal literatu re , indi-
vidua l st udies addressed different clinical pract ice
questions. We selected aspects of ou r fi ve key ques-
tions t hat we believe are sufficien tly s uppo rted by
evi dence from one or more Class n st ud ies to en-
courage clinicians to adopt a particular clinica l
practice. Table 4 ehould be viewed as our interpre-
tation of the cu rrent lite rat ur e in conjunct ion wit h
our own collective experience a s cognitive rehabili-
tation practitioners.
Examining the literature for evide nce t hat re-
vea ls t he so u rce of observed changes in a tten tion
performance highligh ts the in terpre ta tion chel -
lcnge. For exam ple, Park et al . ( 1999) hypothesized
that the im proved performance on t he Consonant
T rigrams test su ggested ch anges in attention be-
haviors or s kills rather t han improved cognitive
processing. An alterna tive explanation may be t h at
th is recall measure requi res working memory,
which had improved as a result of at tention t ra in-
ing. Converse ly, Schlberg et al . (200 l) claimed their
training resulted in improved processing ; however.
it is p lausible that the t raining m ay have in part
encou raged adoption of a n assertive a tt it ude or a
reduction in a nxiety while performing t he more
complex tests and daily living tasks (Fasot ti, Ko-
vacs, Eling, & Brouwer, 2000). In short, the ques-
tion of why some indi vidu als' performance changes
with the a tten tion t raining remain s unan swered .
FUTURE RESEARCH DIRECTIONS
Our review process exposed and educated our com-
mittee on the cha llenges to writing EBP as well as
ANCDS B ULLETIN BOARDIVOL. 11. NO. 3
illuminated future research directions. Ooe of t he
difficu lties we encoun tered was how to code the rel -
evant fea t ures of the effi cacy research. We fonnu-
lated our committee with tb e notion of assembling
expert reviewers who both provide cognitive reha -
bilitation services a nd who cond uct rese arch in the
fi eld . Another option would h ave been to use blind
reviewe rs to code the st udies in order to achi eve
more objectivity in t he review process. Instead , we
elected to u se d iscussion and consensus to identify
a nd code relevant research features. Further, bias
was poss ible by h aving rev iewers who h ave con-
t r ibuted to t he a ttent ion resea rch literature and
who have intellectual owne rship of the interven-
tion under question. However, t he committee a p-
proach to this project provided checks and ba lances
to gua rd agai nst t his type of bias. T he re a re pros
and cons of relying on "front-line" experience ver-
sus objective a ppraisal.
A further limitation of our review comes from
s um ma rizing rather than providing the details of
some of t he important methodologica l research is -
sues. For exa m ple, we chose Dot to discuss in detail
the s pecific types of reliability and va lidi ty that
were lacking in stu d ies . T his decision was made
due to s pace considerations and a desire to disc uss
a broad range of research issues.
We look forward to conti nued resea rch tha t. facil-
itates efforts to develop evidence- based guidelines
for attention t rain ing. The most outstanding need
revealed by our work is to develop methods t ha t
meas ure the impact of at ten tio n im pa irments on
daily life for individuals across the lifespan. T he re-
la t.ionship between neuropsychological tests and
attention as depl oyed during real world activiti es is
not clear. The increased availability of functio nal
brain imagin g may help identify neuropsych clogi-
ca l circuits t hat a re a ffected by train ing, but again,
we need lo understand how these circuits relate to
the perform ance of funct ional activities. It, may be
fr uitful to st udy alternative assessment paradigms
used in related fi eld s such as the fu nctional assess-
ment model descr ibed in s pecial ed ucation research
and practice (e.g., Lu cyshtn . Albin , & Nixon, 1997 )
or inter pretive research methods ada pted to eva lu -
a te the impact of impairments in an ind ividual's
daily li fe (e.g., Simmons-Mackie & Da mico, 1996).
Our review process h ighlights the need for re-
search that better describes the s pecific elements of
attention t raining that are most effective in partic-
ular contexts and t he outcomes that result from
such training. For exam ple, t he knowledge gained
from our review encourages fut ure st udies tha t in-
TABLE 4. Clinical recommendations based on t.he review of literature, organized by key questions 1 through 3.
Pra ctice
Ke y Q u e s t io n Recommendations
I . Who is a good G u ideline for pcstacute
ca di d ate fo r direct or mildly injured c1ient.a.
a tte n t ion training? wit h intact vigilance
Ins u fficie nt evidence to
make recommendation
for clients at acute phase
of recovery
Un known for use wit h
clients with severe ly
impai red vigilance
2. What are the c r it ica l G uid e lin e for using di rect
features o f direct attention t raining in
atte ntion training? conjunction with
metacognirive training
(feedback. self- monitori ng,
& s trategy t ra in ing)
G ui deline for program
individ ua liza tion
Guid e lin e for t reat men t
frequ ency
G uideli n e for complex
atten tion tasks
Unknown for im proving
vigilance or reaction time
S. What outcome s c an G u ideline for obtaining
y o u expect. from task specific, impairment
d irect a tte n tion level ou tcomes
training?
Unkno wn for obtaining
generalization to untrained.
impairme nt level tasks
Unce rtain for obtaining
gene ralization to
pa rticipan t level tasks
S ummary o f Evidence
Two CIa!iS I and four
C IR8S " stu dies with
descriptions of
participants
O ne C lass I and one
Cteee II s tudies. but with
ques tio nable internal
va lidi ty (incomplete da ta ,
un re ported va ria ble
cont rols. etc.I
Evide nce provided by
incomplete acute care
studies (one Class I and
one Class II).
Two Class I and one Class
" studies
Th ree Class 11 studies
1'wo Class I and four
Class II studies
Four Class II studies
One Class I and one Class
II s tud ies wit h acute
pa rticipants as described
above
Two Class I and four
Class II studies
One Class 1 and two
Class II studies but
striking differences in
results interpretation
C ri terion-refe renced
ou tcomes in three Cteas
II studies
Clinic ians Should
Scrutinize candidacy and
monitor responses to
training
Scru tin ize cand idacy.
Mon ito r responses to
t raini.ng a nd know that
observed improve me nts
may in part be a res ult of
s ponta neous recove ry
Be ca utious and awa re of
u ncertainties of outcome.
Proceed on case-by-case
ba sis
Use attention training in
combination with self-
reflective logs,
a nti cipat ion/prediction
activities, feedback. &
st ra tegy training
Identify client strengths
and needs prior to
t reame nt; select exercises
to address specific areas
of weakness
Administer treatment at
least once per week
Use a hie ra rchy of tasks
that emphasize working
memory, menta l control.
a nd select ive,
a lternat in g, an d/or
divi ded attent ion
Be cautious when using
remediation programs
that focus on simple
vigilance or reaction time
Iden ti fy desired
outcomes. Meas ure
perfonnance
Identify desired
outcomes. Measure
perfonnance
Ide nti fy desired outcomes.
Use mel-hods that can
reliably measure clin ically
meaningfu l progresa
xxxvii
xxxv...
ves tigate the effects of com bining st rategy-based
t rain in g and d irect attention training. More coo-
trolled studies with larger num bers of participant
groups that are carefully described are also cri t ical
The use of cont rol grou ps receiving serial measure-
ments without any treatment would strengthen ou r
understanding of intervent ion effects, as would the
us e of examiners who are blind to s t udy d es igns a nd
part icipan t group assignment. Controlled studies
would help so rt out questions s uch as the effective-
ness of a ttention training for acute patients.
We recogn ize the challenge of fo rmulating treat-
ment groups within the brain injury pop ulation.
Looking to populations tha t share similar feat ures
with the bra in injury population may mitiga te t his
research cha llenge. For example, a recent st udy
eva luated the effects of a ttention training in a pedi-
atric popula tion with attention defi cits as a result of
brain radiation for cancer treatment (Butler &
Copeland, 2002). Their results were consis te nt with
a number of studies reviewed for this a rticle a nd
showed significant changes on neur opsychologica l
tests measuring attent ion following atte ntion train-
ing, but the changes did not gene ra lize to a func-
tio nal acade mic task. Widening our search to in-
d ude other populations such as individuals with
developmental atte ntion defi cits may be product ive.
It is our hope that summarizing future resea rch
using the five key qu estions presen ted in this a rti-
cle may provide an organizational framewor k tha t
will a llow clinicians to make decisions that a re em-
pirically su pported.
Ackno wledgments: The a ut hors wan t to thank all
the expe rt reviewers who provided valuable feed back
on an earlier ve rsion of thi s manuscript. This project
was s po nsored by the Academy of Ne urologic Comm u-
nicatiou Disorders & Sc iences fANCDS ), the Amer ica n
Speech-language- Heari ng Associ ation (AS HA l, Ne uro-
physiology and Neuroge nic Speech a nd Langu age Dis-
orders (AS HA's Special In terest Division 2), and the
Veteran Affairs Administration. Cor respondence con-
cern ing this article s houl d be add ressed to Ma ry R.T.
Kennedy, Ph.D., Chair, ANCDS Wri ting Com m ittee, evi-
dence-ba sed pra ct ice guidelines for cogn it ive-commu ni-
ca tion di sorders after trau matic brain inju ry, Depa rt-
ment of Comm un ica tion Disorders, University of Min-
n eso ta , 115 Sh evlin Ha ll, 164 Pills bu ry Dr. S.E. , Mi n-
nea polis, Minnesota , 55455.
Addre ss co rre s pon d e n ce to Mc.Kay Moore Sohlberg.
Ph .D., Communication Disorders & Sciences, 528 ] Uni -
ve rsity of Oregon, E uge ne, OR 97403, USA.
e-ma il: mckay@o regou.uo regon .ed u
ANCDS BULLETIN BOARDNOL . n , NO. 3
REFERENCES
But ler, It W., & Copeland . D. It (2002). Attentional
processes and t hei r remed ia ti on in ch ild ren t rea ted
for cancer: A literat ure review a nd the developmen t
of a t herapeutic approach. Jou rnal of l nterna tionai
N europsychological Society. 8, 115--124.
Cice rone, K. D. (2002). Remedia t ion of 'worki ng a tte n-
tion' in mild tra umat ic brain inju ry. Brain Inj ury, 16,
185- 195.
Cicero ne, K D., Dahlberg, C., Kama r, K , La nge nbahn,
D. M., Malee, J . F.. Bergquist, 1'. F. et a l. (200m . Evi-
deuce-based cognitive re habilita tion : Recommenda -
tions fo r cli nical pract ice. Archives of Phy sical Medi -
cine & Rehabilitation, 81 , 3 16-321.
Fasotu, L., Kovacs , F., Eling. P. A. 1'. M., & Brou wer, W.
H . (2000). Ti me pressure management as a cc mpen-
satory st ra tegy t rain ing after closed head inju ry.
Neuropsychological Reha bilita tion, 10,47-65.
Glisky, E . L., & Schacter, D. L. (l 9861. Remediation of
organic memory d isorde rs: Cu r re n t st a t us a nd
fut ure pros pects. -km rnol of Head Trauma R ehobui-
lotion, 1, 54~63.
Godfrey, H . P. D., & Kn ight , R. G. t 1985). Cognitive reha-
bi lita t ion a nd me mory functioning in a m nesi ac a lco-
holics. Journal of Consulting & Clinical Psychology,
53. 555-557.
Col per, L. C., We rtz, R. T. , Frattali, C. M., Yorkston , K
M.. Mye rs, P , Katz, R., et al . (200 1). Eindence.bceed
practice g u idelines for I he management of comm uni-
cation disorders in ncurologically impaired individu-
als: Proj ect introduction . [On-line]. Availa ble: http ://
www.ancds.duq.ed ulPracticeGuidelines
Gray, J. M., Robertson. I., PenUand, B., & Ande rson, S.
( 992). Microcomputer-based a ttent iona l retra in ing
after bra in damage: A randomized gro up controlled
trial . N enropeychalogice i Re tra ining, 2, 97 -115 .
Kennedy, M. R. T., Avery, ,I., Coelbo, J. , Sohlberg. M. M.,
Turkstra, L., Ylvis ake r, M., et a l. (2002). Evidence-
based practice guidelines for cogniti ve-communica-
tion disord ers a fte r trau mati c brain injury: Initia l
committee report. J ournal of Med ica' Speech-Len-
HI/ age Pathology, 10(2). ix-xiii .
Kewman , D. G., Se igerma n, C., Kinter, H. , Chu. S., Hen-
so n, D., & Redder, C. (]985). Simulation training of
psychomoto r skills: Te aching the brain inj ured to
d rive. Reha bilitation Psychology, 30. 11-27.
Lucysh in, J . M., Albin, R. W.• & Nixon, C. D. (1997).
Embed ding comp rehensive behavioral support in
fam ily ecology: An expe rimental, single-case a naly-
sis. J ou rnal of Consulting and Clirricfl l Psychology,
65.24 1- 25 1.
Miller, R. G., Rosenberg, J. A. , Gelinas, D. F.. & the ALS
Practice Para meters Task Force . (1999). Neurology,
52, ] 3 11- 1323.
Moffat , N. (1992). Strategies of me mo ry the rapy. In R A.
Wilson & N. Moffa t (Eds.j, Clinical ma nagement of
PRAC1'ICE GUIDELINES FOR DJREC1' ATTENTION TRAIN ING
memory problems (pp. 83- 116). Sa n Diego: Singular
Publishin g Group.
Montgomery, E. 8. , & Turkan-a, L. S. (2003). Evidence-
based medicine: Let's be reasonable. Journal of Med-
ica l S peech -La ngung e Pathology, 11(2 ), i x-xi i.
Niemann, H., Ruff, R. M. , & Base r, C. A. (]990). Cern -
puier-assiated attention retrai ning in head-inj ured
individuals: A controlled efficacy study of a n outpa-
tien t program . J ournal of COl/s uiting lind Clinical
Psychology, 58, 8 11-817.
Novack. T. A., Caldwe ll, S. G., Duke, L.. Be rgqu ist , 'f. F.,
& Gage , R J. (1 996"1. Focused ve rs us unstructu red
intervention for a ttention deficits after trauma tic
brain injury. Journal of Heed '1Tuuma Rehabilitation,
11, 52-60.
Park. N. W.. & Ingles , J. L. (200 1). Effectiven ess of auen-
tion rehabilita ti on after acq ui red bra in injury: A
meta-a nalysi s. N europ sychology, 15, 199-210.
Park , N. w., Proulx, G., & Towe rs, W. M. (1999). Evalua -
tion of t he {m e nt ion process training programme.
Neuropsychological Rehabilitati<1Jl, 9, 135-1 54.
Ponsford , J. L., & Kin se lla, G. (988). Evaluation of a
re medial progra m med for a ttent iooa l defi cits follow -
ing closed head injury. J Ollrrll1l Qf Cli nical and Esperi-
mental NeuropSJdwlogy, 10, 693-708.
Robey, R. (200 IJ. C EU part II I: Evidence-based practice.
ASHA Special Interest Divi$ioll 2 NewslelJer, 11{]),
10-1 5.
XXX IX
S lm mens-Mae kle, N. M., & Damico. J . S. (1996).
Accoun ting for handica ps in a phasia. Disability and
Reh abil itation, 18, 540-549.
Sohlberg, M. M., & Mateer, C. A. (1987J. Effectiveness of
an attention-training progrnm . J ournal of Clinical
and Experimental N europsychology, 9,117-1 30.
Sohlberg, M . M ., & Ma teer, C. A. (2001). Cognitive rehn -
bititation: All in teg ra ted neuropsychologi cal ap-
proach. New York: Guilford Pub licat ions.
Soh lberg, M . M ., McLaugh lin, K. A., Pavese, A., Hei -
drich, A., & Posner, M. (2oo 1J. Evaluation of etten-
tion p rocess trainin g and brai n injury ed ucation in
persons with acq uired brain inj ury. J ourna l of Clinl-
cal a nd Esperimetu ai N eu rop$)'chology, 22, 656-676.
Srraehe. W. ( 1987). Effectiveness of two modes of t rain -
ing to overcome defi cits of concen tration . Intem a-
tianal J ou rna l of Rehabilitation Resea rch. 10 <S up--
plemen t 5), 1418--1458.
Stu rm, W., WiUmes. K. , Orgl\88, B., & Hartj e, W. (1997).
Do specific attention deficits need specifi c training?
Neu ropsychological Reha bilitation , 7, 8 1- 103.
Ylvisaker, lV1., Coe lho. C., Kenned y, M., Sohl berg. M. M..
T u rkan-a. L.. Avery, J., & Yc rkston. K. M. (2002).
Reflect ions on evidence-ba sed pra cti ce and rat ional
clinica l decision making. Journal of Medical Speech-
La ng uage Pat hology, 10(3), xxv- coodu.
Ylvisakee, M., & Feeney, T. (1998). Couabora tiue brain
inju ry interuenticn: Positive everyday routi nes. San
Diego: S ingu lar Publishing G rou p.