http://informahealthcare.

com/bij
ISSN: 0269-9052 (print), 1362-301X (electronic)

Brain Inj, 2014; 28(12): 1610–1616

! 2014 Informa UK Ltd. DOI: 10.3109/02699052.2014.934921

ORIGIANL ARTICLE

Efficacy of memory rehabilitation therapy: A meta-analysis of TBI and

stroke cognitive rehabilitation literature
Madison Elliott & Frederick Parente

Towson University, Towson, MD, USA

Abstract Keywords
Objective: To examine the efficacy of cognitive rehabilitation strategies specifically designed Cognitive rehabilitation, memory,
to improve memory after traumatic brain injury (TBI) and stroke vs. memory improvement with meta-analysis, neurorehabilitation,
the passage of time. remediation, stroke, TBI, working memory
Design and methods: A meta-analysis was performed on 26 studies of memory retraining and
recovery that were published between the years of 1985 and 2013. Effect sizes (ESs) from each History
study were calculated and converted to Pearson’s r and then analysed to assess the overall
effect size and the relationship among the ESs, patient demographics and treatment Received 15 May 2013
interventions. Revised 14 March 2014
Results: Results indicated a significant average ES (r ¼ 0.51) in the treatment intervention Accepted 6 June 2014
conditions, as well as a significant average ES (r ¼ 0.31) in the control conditions, in which Published online 21 July 2014
participants did not receive any treatment. The largest ESs occurred in studies of stroke patients
and studies concerning working memory rehabilitation.
Conclusions: Results showed that memory rehabilitation was an effective therapeutic
intervention, especially for stroke patients and for working memory as a treatment
domain. However, the results also indicated that significant memory improvement occurred
spontaneously over time.

Introduction to brain injury attempted to amalgamate this literature via published

rehabilitation practice guidelines for TBI and stroke. These
At least 5.3 million US citizens live with a permanent TBI- or
guidelines are based on findings from recently published
stroke-related disability. Health professionals consider TBI
systematic reviews, which generally examine a wide body of
to be one of the most debilitating injuries relative to
published studies that include diverse research designs [7].
other neurological disorders [1, 2]. Mar et al. [3] detailed
Systematic reviews construct valuable inferences about the
the devastating societal and individual impact of neuro-
efficacy of rehabilitation strategies; but what they achieve
logical trauma. Many factors contribute to a person’s recov-
in variety, they also lack in statistical rigour. Meta-analyses
ery trajectory and, in the past decade, rehabilitation of
must exclude studies that do not provide group data for effect
cognitive processes has become significant in the treatment
size (ES) and, therefore, cannot examine as many publications
course [4, 5].
as a systematic review. Nevertheless, meta-analyses are
Although cognitive rehabilitation is increasingly popular,
becoming increasingly popular in review methodology in
clinicians currently face major difficulties when identifying
psychological and brain sciences and are recognized as a
and treating cognitive impairments caused by brain injury or
statistically reliable manner of testing objective observations
stroke. This is because brain-injured populations typically
across studies [8]. The purpose of the present investigation is
are a heterogeneous group with a multitude of deficits [6].
to provide a meta-analytic review of cognitive rehabilitation
Evidence-based treatment guidelines are only as good as the
strategies for the domain of memory in individuals with TBI
research they rely on and published rehabilitation studies
and stroke. The authors’ goal is to contribute to future
tend to be both elaborate and ambiguous in nature. Several
guidelines, as well as augment existing qualitative review
professional committees, including members of the National
findings from memory rehabilitation literature.
Institutes of Health (NIH) and the Brain Injury-
Interdisciplinary Special Interest Group (BI-ISIG) of the
American Congress of Rehabilitation Medicine, have Cognitive rehabilitation of memory
Cognitive Rehabilitation Therapy (CRT) for brain injury and
stroke is diverse, with numerous treatment and theoretical
Correspondence: Madison Elliott, Psychology Department, Towson
University, 8000 York Road, Towson, MD 21252, USA. E-mail: models. Treatments can be process-specific, aimed at improv-
mellio4@students.towson.edu ing overall performance of a given activity, skill-based,
DOI: 10.3109/02699052.2014.934921
motor-based or restricted to a certain area of cognition like
attention, language, executive function or memory. Overall
goals of treatment and treatment implementation are as varied
as the treatments themselves [9]. This rehabilitation is broad
because of the unique nature of every brain injury. In the case
of memory impairment, long-term, short-term and/or working
memory can be negatively affected. Rehabilitation tasks
and their outcome measures are often developed to improve or
evaluate function within a specific type of memory [10].
Memory impairment is commonly reported after brain injury
or stroke and individuals who experience trauma localized in
the temporal lobes, hippocampus and amygdala are particu-
larly susceptible to memory impairment [10]. However,
the locus of the brain injury is not the sole determinant of
memory problems; many factors contribute to each person’s
unique outcome post-brain injury [2]. To explain the varying
nature of TBI, Kay [4] and McCrea et al. [5] have proposed
integrated, multi-factorial neuropsychological models of brain
injury outcome, including the determination of factors such
as neurological health, psychological health, physiological
health, as well as their interactions with individual subjective
or objective cognitive functions.
Various behavioural treatment strategies for memory
rehabilitation have proven to be efficacious, such as the
internet-based tasks used in Bergquist et al. [11]. Other forms
of computerized training [12], awareness questionnaires [13]
and motive or incentive-based tasks [14] have also resulted
in effective treatment. However, relatively few articles have
been published that have documented successful pharmaceut-
ical treatments for memory enhancement [15].
The typical behavioural rehabilitation efficacy study
evaluates memory differences for a control group vs. experi-
mental group, pre- and post-intervention design. Outcome
measures vary greatly and include neuropsychological
batteries, motor-function tests, brain imaging and various
self-report instruments. Observer versions of the self-report
measures have been administered to participants’ family
members, professional supervisors or close friends to obtain
a more diverse understanding of their daily behavioural
functioning [2].
Although there are published reports of potentially effect-
ive treatments, there is no conclusive evidence of an effective
standardized treatment programme for memory dysfunction.
Most treatments produce improvement in memory function-
ing, but it is unclear whether some treatments are more
efficacious than others [16]. Furthermore, there is little
research to document enduring improvement over time for
any treatment modality. There is some evidence, however,
for treatment efficacy that exceeds the level of improvement
that occurs with the passage of time. In Rohling et al.’s [8]
meta-analytic re-examination of Cicerone et al.’s [10, 17]
review, the authors found a larger effect size for memory
improvement in TBI and stroke patients after clinical
rehabilitation than in TBI and stroke patients who received
no treatment.
Memory rehabilitation efficacy
CRT after TBI and stroke has been the topic of a number
of major reviews evaluating treatment effectiveness [8, 10,
Efficacy of memory rehabilitation therapy 1611
17–19]. In their extensive review, Cicerone et al. [10, 17]
evaluated 118 studies that used CRT, classifying them first
in terms of methodological quality and then summarizing
consistency of findings. The highest ranked studies were
those including randomized control trials (RCTs) (I), followed
by studies including non-randomized control prospective or
retrospective case-controls (II). The lowest ranked studies
were those not including control groups or those that used
data from case studies of single subjects (III). The study also
categorized the research into various domains: attention,
vision, perception, apraxia, language and communication,
working memory, executive functioning, problem-solving and
awareness and comprehensive holistic cognitive rehabilita-
tion. Some of the cognitive rehabilitation domains, such
as attention training, visual-spatial training and language-
based interventions produced larger treatment effects than
others [17, 20].
More recently, Rohling et al. [8] performed a quantitative
meta-analysis of a select, well-controlled, literature set that
derived from Cicerone et al. [10, 17]. Rohling et al. [8]
concluded there was ‘modest qualitative support’ for CRT
efficacy, as asserted by Cicerone et al. [10, 17] asserted in
their original paper. Rohling et al. [8] found sufficient
evidence that attention training after TBI, language and
visual-spatial training for symptoms of stroke were effective.
The authors concluded that, out of five major treatment-
effectiveness findings in Cicerone et al. [10, 17], their meta-
analysis results only supported three: attention-training,
language-based training and visual-spatial training.
Rohling et al. [8] cautioned readers that Cicerone et al.
[10, 17] did not adequately estimate re-test effects from
studies with uncontrolled designs. One of the two major
claims Rohling et al. [8] questioned was the efficacy of
memory rehabilitation. Although their data showed a moder-
ate effect size (ES) for memory rehabilitation, it was not
significant. This suggested that general neuroplasticity and
natural recovery were responsible for much of the docu-
mented memory recovery, even in the treatment conditions.
Due to the lack of significant ESs, the authors also could not
definitively say that the memory rehabilitation had the desired
effect. However, very few of the 115 articles evaluated in
Rohling et al. [8] involved memory rehabilitation, so it is
difficult to derive any conclusive evidence about the overall
efficacy of memory therapy.
The present meta-analysis reviewed studies published in
1985–2013 about memory rehabilitation in participants who
suffered a TBI or stroke, in order to build upon the
aforementioned findings by Cicerone et al. [10, 17] and
Rohling et al. [8], regarding the domain of memory rehabili-
tation. The primary goal of the present meta-analysis was
to update Rohling et al.’s [8] work by adding new studies.
A secondary goal was to investigate new factors that may
contribute to the recovery effects.
Many neuro-rehabilitation techniques have changed and
advanced since the Rohling et al. [8] meta-analysis. However,
the Rohling et al. [8] methodology for meta-analytic research
in this area is exemplary. Therefore, the methods for the
current study were adapted from Rohling et al. [8]. These
procedures include adding the most recent publications, using
the most rigorous and careful methods of the four major
1612 M. Elliott & F. Parente
reviews listed and replicating the Rohling et al. [8] selection
criteria.
A re-examination of new memory rehabilitation studies,
and those in Rohling et al.’s [8] database, revealed several
new variables that were significant factors during a patients’
recovery. The following variables were identified based
on their prevalence and statistical availability: average age
of the patients (paediatric vs. adult), year of publication,
type of brain injury, type of memory targeted by the
rehabilitation strategy, type of intervention (computerized
vs. not-computerized) and sample size.
Meta-analysis and effect size measures
A meta-analytic review is a manner of statistically aggregat-
ing and evaluating previously conducted studies. Generally,
the purpose of a meta-analysis is to examine the overall effect
of a specific variable in available literature. Meta-analyses use
research studies as a unit of measure to evaluate central
tendency and variability of the ESs across the sample of
chosen studies. Rohling et al. [8] used a widely accepted
procedure extracting ESs from published studies; the ESs
measure of the amount of variance the effect controls in a
given experiment. Their meta-analysis was a quantitative
evaluation of the extent to which covariates of interest control
significant portions of the ESs in the study sample. According
to the standards used by Rohling et al.’s [8], the overall effect
of the variable is considered reliable if the average ES is
significantly different from zero. The goal of the present
study was to follow Rohling et al.’s [8] example and to reduce
the relevant statistical information in each published study
to standard units of ES and then evaluate whether the ESs
co-vary with any of several other variables that describe the
conditions of the original experiments.
Method
Sample of studies
Studies were identified using combinations of the search
terms: ‘working memory’, ‘rehabilitation’, ‘remediation’,
‘memory’, ‘training’, ‘brain injury’, ‘TBI’ and ‘stroke’ in
the following research databases: Medline, Pub Med,
PsycINFO, PsycArticles and Google Scholar. Once pertinent
articles were identified through these databases, their refer-
ence lists were reviewed to locate other potentially relevant
studies for inclusion. Based on Rohling et al.’s [8] criteria,
excluded studies (1) did not contain a memory-rehabilitative
intervention; (2) merely described treatment approaches or
theories; (3) were review articles; (4) presented unspecified
or unmeasured interventions; (5) lacked a diagnosis or
assessment of TBI or stroke; (6) were case studies of a
single participant with no empirical data; (7) were non-peer
reviewed articles; (8) included exclusively-pharmacological
interventions; and (9) were not written in English. Rohling
et al. [8] mention that single papers may include multiple
analysable studies, with ‘unique non-overlapping samples
of participants’ (p. 22). In total, 26 published memory
rehabilitation studies met selection standards and could be
used for the analysis. Fourteen of the studies compared
a treatment intervention condition to a control group.
Brain Inj, 2014; 28(12): 1610–1616
The other 12 studies contained a treatment intervention, but
no control condition. Studies not cited in article, but included
in the meta-analysis are listed [30–51].
Procedural analysis
Meta-analytic procedures using the MIX 2.0 for Microsoft
Excel meta-analysis software package were used to evaluate
the ESs computed from the available statistics in each
published study. These ESs (calculated as the Pearson’s r)
were subsequently analysed to determine whether the average
ESs were significantly different from zero. The researchers
chose to use r because of several advantages listed in
Rosenthal and Dimatteo [21]. These authors stated that
converting to r is advantageous because this ES, in its point
biserial form, better represents the relationship between levels
of the independent variable and dependent variable scores.
Additionally, converting a continuous r to dichotomous d
results in a loss of statistical information. Researchers may
use degrees of freedom contrast techniques to analyse trends
across a greater number of groups with r and it requires
no computational adjustments when examining different
types of t-test samples. There is also a well-accepted public
familiarity with r and its practical importance is established in
the field of psychology [21]. Magnitude of ES (small,
moderate, large) was interpreted according to the guidelines
published in Cohen [22].
ESs from control conditions estimated how much memory
improvement was not attributable to an experimental inter-
vention, but instead attributable to the passage of time and
non-specific neuroplasticity. ESs from intervention conditions
were subsequently analysed and used to estimate how much
memory improvement could be attributed to a cognitive
rehabilitation effort. The final analytic procedure compared
the first (control analysis) to the second (intervention
analysis) in order to determine whether cognitive rehabilita-
tion interventions yield significantly larger ESs compared to
control conditions, where no intervention has taken place.
The Q statistic was calculated to evaluate the significance
of the overall effect size and to assess the significance of the
difference between the treatment ES once the control ES was
removed. The Q statistic and forest plots were also calculated
initially to assess the homogeneity of the ESs across the
various studies. The observed heterogeneity of ESs in both
control and intervention group conditions was used to justify
the selection and use of a random effects model for the
analysis. This model allows for comparisons to be generalized
beyond studies in the current sample and studies identical to
those in the current sample [21]. A random effects model
was considered superior to a fixed effects model for analysing
the current set of studies because fixed effect models are
used when the sample of studies are considered homogenous
and of the same population [23]. Although MIX 2.0 analytic
procedures weighted ESs by their sample size, Begg’s
Test statistics also were calculated to assess whether or
not sample sizes affected the reported ESs. The degree
of dissemination bias, i.e. the extent to which the published
studies over-estimate the size of the effect, was also examined
using a funnel plot. Covariates of the ESs, including
the average age of the patients (paediatric vs. adult), ratio
DOI: 10.3109/02699052.2014.934921
of males to females in the sample, year of publication, type
of brain injury, type of memory targeted by the rehabilita-
tion strategy, type of intervention (computerized vs. not-
computerized) and sample size were also evaluated to
determine whether or not they predicted memory recovery
in either group.
Results
Control conditions analysis
The average effect size for this condition of r ¼ 0.31 was
significant (Z ¼ 10.00, p50.05), which indicated a moderate
and significant improvement in memory, that was not
attributable to a rehabilitation intervention. Figure 1 is a
forest plot displaying the ESs for all included control studies.
The plot displays the ES and 95% confidence interval for each
included study. The extent to which the horizontal lines
overlap is an indication of the homogeneity of the ESs across
studies. This visual representation of the data demonstrates
that the studies were generally not homogenous in their
reported ESs, with two possible outliers. The outliers may
be due to different methodology or sample characteristics,
as suggested by Control Group Q statistic and selectivity
funnel plot results.
A significant Q statistic (73.259, p ¼ 0.00) also indicates
that the studies included are heterogeneous. The I2 statistic
estimates the percentage of within-study variance among the
studies due to non-random factors. The I2 statistic 82.25%
also suggests heterogeneity among the ESs. It is important
to note that when k520, as in the present analysis, both the Q
and I2 tests should be interpreted with caution [24].
Nevertheless, this statistic suggests that it may be worthwhile
to explore characteristics of the various experiments that may
Figure 1. Forest plot of ESs in control group studies.
Efficacy of memory rehabilitation therapy 1613
have contributed to the inconsistency of methods used among
the control studies.
Begg’s Test statistics were computed to determine whether
publication bias was present in the sample of studies used.
Smaller samples with fewer degrees of freedom often produce
inflated ESs and may indicate the existence of unpublished
studies with lower ESs. The Begg’s statistics computed on
these data indicated the presence of publication bias among
the chosen studies.
Funnel plots can be used to illustrate dissemination bias.
If there is little or no dissemination bias, a funnel plot will
show all the dots distributed randomly around the synthesis
estimate line and they will also fall between the confidence
interval funnel lines on each side of the plot. The funnel
plot from the present analysis, as seen in Figure 2, indicates
the presence of dissemination bias in control condition
studies. This also means that there is a possibility of
unpublished studies with lower ESs existing and reinforces
the Begg’s test results that published studies contain inflated
ESs which are possibly over-estimating the effect of recovery
without an intervention condition.
Intervention conditions analysis
The average effect size for this group of studies was r ¼ 0.51
(Z ¼ 20.62, p50.05), which was significantly different
from zero. The confidence interval for the overall ES
(0.462–0.558), did not include the average ES for the control
studies (r ¼ 0.31). The results, therefore, indicated that,
overall, a moderate and significant improvement in memory
was apparent in the cognitive measures used in these studies.
This change in memory function cannot be attributed solely to
the passage of time.
Figure 3 is a forest plot for the intervention condition
studies. The visual representation of the data here demon-
strates intervention studies were somewhat less heterogeneous
than the controls in their reported ESs. These outliers may
be due to different methodologies or publication bias,
as suggested by the Intervention Group Q statistic and
selectivity funnel plot results.
The significant Q statistic indicated significant heterogen-
eity of the ESs (Q ¼ 81.50, p50.05). The I2 statistic of
Figure 2. Funnel plot for control group studies.
1614 M. Elliott & F. Parente
Figure 3. Forest plot of ESs in intervention group studies.
Figure 4. Funnel plot for intervention studies.
69.31% indicates that there was somewhat less heterogeneity
than the control conditions where I2 ¼ 82.25%. These statis-
tics should be considered more reliable for the intervention
analysis (k ¼ 26) according to sample size criteria in Higgins
et al. [24].
The Begg’s test statistics computed on these data showed
that dissemination bias was present in the sample of studies
used for the analysis. The intervention analysis funnel plot
(Figure 4) is an additional indication of the presence of
dissemination bias in intervention studies. This shows that,
similar to published control studies, published intervention
studies have inflated ESs and may be over-estimating the
effect of recovery due to rehabilitation.
Two of the study covariates predicted the ESs. The 95%
confidence intervals computed on the ESs in the intervention
Brain Inj, 2014; 28(12): 1610–1616
group conditions indicated that studies of stroke rehabilitation
produced significantly larger ESs compared to mixed brain
injury and TBI studies (p50.05). Confidence intervals were
also used to assess ES differences between the intervention
(experimental group) and the control group in studies that
included both. The discrepancy between these confidence
intervals indicated a significant difference between the
experimental and the control group ESs (p50.05). A third
confidence interval analysis indicated that studies of working
memory produced significantly higher ESs relative to studies
of other memory processes. There was no significant differ-
ence in ESs for intervention studies (experimental groups)
that contained a control group vs. intervention studies that did
not contain a control group, for average age of the patients in
control and intervention studies (paediatric vs. adult) or type
of intervention (computerized vs. not-computerized). There
was no significant correlation between ES and the ratio of
males-to-females in the sample or for the year of publication
in control and intervention studies.
Post-hoc correlation analyses investigating the relationship
between dependent variables in the control and intervention
conditions and ESs did not reveal any significant relationship
(p40.05) between the ESs and the number of dependent
variables in either the control or the experimental conditions.
Additional post-hoc correlation analyses show that there was
no significant correlation (p40.05) between sample sizes
and ESs for intervention (experiment) group studies, but that
there was a significant correlation (p50.05) between sample
sizes and ESs for control group studies. There were no other
significant covariate/ES correlations.
Discussion
This meta-analytic review investigated differences in studies
examining the influence of memory rehabilitation interven-
tions for individuals who suffered brain damage as a result
of TBI or stroke. One goal of this meta-analysis was to
supplement the work published by Cicerone et al. [10, 17] and
Rohling et al. [8]. The Cicerone et al. [10, 17] study did not
quantitatively evaluate ESs; the Rohling et al. [8] study had a
relatively small sample size for memory studies. The present
study also examined more covariates than the Rohling et al.
[8] meta-analysis. The results obtained in this investigation
for the intervention group show a significant moderate effect
(r ¼ 0.51) of the interventions which could not be attributed
to the passage of time. However, the control analyses show
a smaller, but also significant and moderate effect (r ¼ 0.31)
for recovery without a treatment intervention. This pattern of
results replicated the ES relationship found in Rohling et al.
[8]. In both studies, there was a small ES that could not be
the result of spontaneous natural recovery, which suggested
that the interventions used in the examined studies were
responsible for the treatment effect. Unlike Rohling et al. [8],
the present analysis found that the difference in ESs between
control and intervention conditions was significant. This
shows that, overall, the memory rehabilitation strategies
produced the desired effect. The confidence intervals
computed around the ESs in the control groups indicated
that the average ES was significantly greater than zero;
this suggests that significant spontaneous improvement in
DOI: 10.3109/02699052.2014.934921
memory occurs after brain injury and stroke. The fact that
some amount of the ES in the experimental treatment
condition cannot be attributed to the passage of time suggests
that therapeutic intervention may accelerate this
improvement.
The heterogeneity of ESs in these studies could be
accounted for partially by a few moderating variables.
Analyses show that studies of stroke recovery produced
significantly larger ESs than were apparent in studies of TBI
recovery or mixed injury recovery. This ES difference
occurred across intervention and control group studies,
which suggests that stroke patients experienced more
change in their functional status over time and were more
responsive to treatment. This finding is inconsistent with
published research [25, 26]. Persons with TBI are typically
younger than stroke patients and, thus, typically have better
prognosis for recovery than stroke patients. Individuals with
TBI are more frequently discharged from in-patient facilities
and make larger gains in treatment [27]. Their actual recovery,
however, may last a lifetime [28, 29]. The larger ESs for
stroke studies reported here may be directly related to the
particular behavioural and medical treatment approaches
implemented with the stroke samples in the studies examined;
for example, the use of pharmaceutical treatments [16] to
stop haemorrhaging quickly and preventative measures that
decreased the level of brain damage when the person had
the stroke [27].
Studies addressing working memory produced signifi-
cantly larger effect sizes relative to studies that involved
rehabilitation of other types of memory problems. These
results suggest that cognitive rehabilitation therapy may be
most effective when applied to stroke rehabilitation and
when the goal of the therapy is to improve working memory
functions. The fact that significant improvement in memory
occurred in the control condition suggests the need for
baseline evaluations immediately for individuals who suffered
brain damage as a result of TBI or stroke.
The results are also interesting regarding which covariates
did not predict the ESs. The Beggs’s test suggests that ESs in
the selected studies may be over-estimated. Other character-
istics of the experimental designs, for example, the number of
dependent variables that were measured during the experi-
ment, did not seem to affect the ESs. Surprisingly, 95%
confidence intervals computed around ESs for average age
of participants, gender, year of publication and whether or not
the treatment was computerized showed that none of these
variables predicted memory recovery, without a rehabilitation
measure. Despite their presence in every study examined, the
covariate analyses showed that these factors did not result
in statistically significant differences across studies. This does
not support findings suggesting that stroke and TBI aetiology
varies by age [1]. In addition, the analysis does not support
the claim some studies have made suggesting that newer
computerized or internet-based rehabilitation strategies
demonstrate better overall efficacy relative to traditional
methods [14, 15]. Finally, there were no significant differ-
ences between the average ES for the treatment and control
conditions in the current analysis relative to those reported
by Rohling et al. [8], which suggests that the findings reported
in both of these meta-analyses are stable.
Efficacy of memory rehabilitation therapy 1615
The present study investigated only memory. It is,
therefore, possible that additional meta-analyses of attention,
visual spatial and language also may reveal efficacy in
those domains as well. Limitations of the current study are
primarily due to a lack of available information in the
published studies examined. For example, it was generally
impossible to describe exactly what the authors had provided
relative to treatment intervention. Furthermore, there was
generally no discussion of the locus of the neurological insult
in any of the studies. The average level of education for the
participants was seldom discussed.
The present paper suggests that the ESs in studies of stroke
survivors are larger than those obtained in studies of TBI.
Future studies should provide more rigorous statistical
investigations as to why cognitive rehabilitation of memory
produces a larger effect in stroke patients. For instance, neuro-
imaging of brain areas associated with memory in TBI
and stroke patients should be documented before and after
treatment, as should a standardized qualitative account of the
injury profile. Most of the studies examined provided vague
descriptions and classifications of stroke or TBI damage
in their sample. Many times the only information about
participants’ TBI variability was a GCS range. Even this
information was impossible to code across studies, as some
investigators did not use GCS to define severity and some
used it in combination with other variables like neuro-
imaging and post-traumatic amnesia. Demographic data such
as hemispheric location of injury and cognitive deficits of
participants were rarely available. The results from the
present analysis demonstrate hopeful statistical findings for
individuals recovering from TBI and stroke. The available
memory rehabilitation treatments appear to be effective;
additionally, to a lesser degree, memory also recovers without
rehabilitation over time.
Declaration of interest
The authors report no conflicts of interest. The authors alone
are responsible for the content and writing of the paper.
References
1. Collins M, Iverson G, Lovell M, McKeag D, Norwig J, Maroon J.
On-field predictors of neuropsychological and symptom deficit
following sports-related concussion. Clinical Journal of Sport
Medicine 2003;13:222–229.
2. Englander J, Hall K, Stimpson T, Chaffin S. Mild traumatic brain
injury in an insured population: Subjective complaints and return to
employment. Brain Injury 1992;6:161–166.
3. Mar J, Arrospide A, Begiristain J, Larrañaga I, Elosegui E,
Oliva-Moreno J. The impact of acquired brain damage in terms
of epidemiology, economics and loss in quality of life. BMC
Neurology 2011;11:11–46.
4. Kay T. Neuropsychological treatment of mild traumatic brain
injury. Journal of Head Trauma Rehabilitation 2003;8:74–75.
5. McCrea M, Iverson G, McAllister T, Hammeke A, Powell M,
Barr W, Kelly J. An integrated review of recovery after mild
traumatic brain injury (MTBI): Implications for clinical manage-
ment. The Clinical Neuropsychologist 2009;23:1368–1390.
6. Wolfe CDA. The impact of stroke. British Medical Bulletin 2000;
56:275–286.
7. Cappa SF, Benke T, Clarke S, Rossi B, Stemmer B, van Heugten C.
EFNS guidelines on cognitive rehabilitation: Report of an EFNS
taskforce. European Journal Neurology 2005;12:665–680.
8. Rohling M, Faust M, Beverly B, Demakis G.
Effectiveness of cognitive rehabilitation following acquired
1616 M. Elliott & F. Parente
brain injury: A meta-analytic re-examination of Cicerone et al.’s
(2000, 2005) systematic reviews. Neuropsychology 2009;23:20–39.
9. Planas AM. Advances in stroke. Translational Medicine 2012;44:
318–319.
10. Cicerone K, Dahlberg C, Kalmar K. Evidence-based cognitive
rehabilitation: Recommendations for clinical practise. Archive
of Physiological Medical Rehabilitation 2000;81:1596–1615.
11. Bergquist T, Gehl C, Mandrekar J, Lepore S, Hanna S, Osten A,
Beaulieu W. The effect of internet-based cognitive rehabilitation
in persons with memory impairments after severe traumatic brain
injury. Brain Injury 2009;23:790–799.
12. Lundqvist A, Grundström K, Samuelsson K, Rönnberg J.
Computerised training of working memory in a group of patients
suffering from acquired brain injury. Brain Injury 2010;24:
1173–1183.
13. Livengood M, Anderson J, Schmitter-Edgecombe M. Assessment
of memory self-awareness following traumatic brain injury. Brain
Injury 2010;24:598–608.
14. McCauley S, Wilde E, Merkley T, Schnelle K, Bigler E, Hunter J,
Vasquez A, Levin H. Patterns of cortical thinning in relation to
event-based prospective memory performance three months after
moderate to severe traumatic brain injury in children.
Developmental Neuropsychology 2010;35:318–332.
15. Kim D, Hung TM, Bae K, Jung J, Lee S, Yoon B, Cheong H, Ko K,
Ryu J. Gomisin A improves scopolamine-induced memory impair-
ment in mice. European Journal of Pharmacology 2006;542:
29–135.
16. Dou Z, Man D, Ou H, Zheng J, Tam S. Computerised errorless
learning-based memory rehabilitation for Chinese patients with
brain injury: A preliminary quasi-experimental clinical design
study. Brain Injury 2006;20:219–225.
17. Cicerone K, Dahlberg C, Malec J. Evidence-based cognitive
rehabilitation: Updated review of the literature from 1998 through
2002. Archive of Physiological Medical Rehabilitation 2005;86:
1681–1692.
18. Park N, Ingles J. Effectiveness of attention rehabilitation after
acquired brain injury: A meta-analysis. Neuropsychology 2001;15:
199–210.
19. Robey R. A meta-analysis of clinical outcomes in the treatment of
aphasia. Journal of Speech, Language, and Hearing Research 1998;
41:172–187.
20. Katz I, Ashley J, O’Shanick J, Connors S. 2006. Cognitive
rehabilitation: The evidence, funding and case for advocacy in
brain injury. McLean, VA: Brain Injury Association of America.
21. Rosenthal R, Dimatteo M. Meta-analysis: Recent developments
in quantitative methods for literature reviews. Annual Review of
Psychology 2001;52:59–82.
22. Cohen J. 1988. Statistical power analysis for the behavioral
sciences. 2nd ed. Hillsdale, NJ: Erlbaum.
23. Cooper H, Hedges L. 1994. The handbook of research synthesis.
US: Russel Sage.
24. Higgins J, Thompson S, Deeks J, Altman D. Measuring inconsist-
ency in meta-analyses. British Medical Journal 2003;327:557–560.
25. McAllister T, Flashman L, Maerlender A, Greenwald R, Beckwith
J, Tosteson T, Crisco J, Brolinson P, Duma S, DuHaime A, et al.
Cognitive effects of one season of head impacts in a cohort of
collegiate contact sport athletes. Neurology 2012;78:1777–1784.
26. Kraus M, Susmaras T, Caughlin B, Walker C, Sweeney J, Little D.
White matter integrity and cognitive in chronic traumatic brain
injury: A diffusion tensor imaging study. Brain 2007;130:
2508–2519.
27. Smania N, Avestani R, Roncari L, Ianes P, Girardi P, Varalta V,
Gambini MG, Fiaschi A, Gandolfi M. Factors predicting functional
and cognitive recovery following severe traumatic, anoxic, and
cerebrovascular brain damage. Journal of Head Trauma
Rehabilitation 2011;15:45.
28. Teassel R, Bayona N, Marshall S, Cullen N, Bayley M,
Chundamala J, Villamere J, Mackie D, Rees L, Hartridge C,
et al. A systematic review of rehabilitation of moderate to severe
acquired brain injuries. Brain Injury 2007;21:107–112.
29. Gray DS, Burnhanm R. Preliminary outcome analysis of a
long-term rehabilitation program for severe acquired brain injury.
Brain Inj, 2014; 28(12): 1610–1616
Archives of Physical Medicine and Rehabilitation 2000;81:
1447–1145.
30. Anderson J, Schmitter-Edgecomb M. Predictions of episodic
memory following moderate to severe traumatic brain injury
during inpatient rehabilitation. Journal of Clinical Experimental
Neuropsychology 2009;31:425–438.
31. Anderson V, Catroppa C. Memory outcome at 5 years post-
childhood traumatic brain injury. Brain Injury 2007;21:1399–1409.
32. Severity of Brain Injury [Internet]. Brain Injury Association of
America. Available online at: http://biausa.fyrian.com/about-brain-
injury.htm#causes, accessed 19 September 2012.
33. Berg I, Konning-Haanstra M, Deelman B. Long-term effects of
memory rehabilitation. A controlled study. Neuropsychological
Rehabilitation 1991;1:97–111.
34. Bourgeois M, Lenius K, Turkstra L, Camp C. The effects of
cognitive teletherapy on reported everyday memory behaviours of
persons with chronic traumatic brain injury. Brain Injury 2007;21:
1245–1257.
35. Couillet J, Soury S, Lebornec G, Asloun S, Joseph P, Mazaux J,
Azouvi P. Rehabilitation of divided attention after severe traumatic
brain injury: A randomized trial. Neuropsychological
Rehabilitation 2010;20:321–339.
36. Evans J, Wilson B. A memory group for individuals with brain
injury. Clinical Rehabilitation 1996;6:75–81.
37. Freeman M, Mittenberg W, DiCowden M, Bat-Americani M.
Executive and compensatory memory retraining in traumatic brain
injury. Brain Injury 1992;6:65–70.
38. Goldstein G, Beers S, Longmore S, McCue M. Efficacy of memory
training: A technological extension and replication. The Clinical
Neuropsychologist 1996;10:66–72.
39. Goldstein G, McCue M, Turner S. An efficacy study of memory
training for patients with closed head injury. The Clinical
Neuropsychologist 1998;2:251–259.
40. Hildebrandt H, Gehrmann A, Modden C, Eling P. Enhancing
memory performance after organic brain disease relies on retrieval
processes rather than encoding or consolidation. Journal of Clinical
and Experimental Neuropsychology 2011;33:257–270.
41. Kerner M, Acker M. Computer delivery of memory retraining with
head injured patients. Cognitive Rehabilitation 1985;11:26–31.
42. Krause M, Kennedy M. Metamemory adjustments over time in
adults with and without traumatic brain injury. Brain Injury 2009;
23:965–972.
43. Milders M, Deelman B, Berg I. Rehabilitation of memory for
people’s names. Memory 1998;6:21–36.
44. Nadar M, McDowd J. ‘Show me, don’t tell me’; is this a good
approach for rehabilitation? Clinical Rehabilitation 2008;28:
847–855.
45. Newsome M, Steinberg J, Scheibel R, Troyanskaya M, Chu Z,
Hanten G, Levin H. Effects of traumatic brain injury on working
memory-related brain activation in adolescents. Neuropsychology
2008;22:419–425.
46. Schacter D, Rich S, Stampp M. Remediation of memory disorders,
experimental evaluation of the spaced retrieval technique. Journal
of Clinical and Experimental Neuropsychology 1985;7:79–96.
47. Schmitter-Edgecombe M, Fahy J, Whelan J, Long C. Memory
remediation after severe closed head injury: Notebook training
versus supportive therapy. Journal of Consulting and Clinical
Psychology 1995;63:484–489.
48. Serino A, Ciaramelli E, Di Santantonio A, Malagu S, Servadei F,
Ladavas E. A pilot study for rehabilitation of central executive
deficits after traumatic brain injury. Brain Injury 2007;21:11–19.
49. Stringer A. Ecologically-oriented neurorehabilitation of memory;
robustness of outcome across diagnosis and severity. Brain injury
2011;25:169–178.
50. Westerberg H, Jacobaeus H, Hirvikoski T, Clevberger P,
Ostensson M, Bartfai A, Klingberg T. Computerised working
memory training after stroke- a pilot study. Brain Injury 2007;21:
21–29.
51. Wilson B, Evans J, Emslic H, Malinek V. Evaluation of NeuroPage:
A new memory aid. Journal of Neurology, Neurosurgery &
Psychiatry 1997;63:113–115.
Copyright of Brain Injury is the property of Taylor & Francis Ltd and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.