Family factors associated with academic achievement deficits in pediatric brain tumor survivors

Family factors associated with academic achievementdeficits in pediatric brain tumor survivors

Emily Ach1,2, Cynthia A. Gerhardt1,2, Maru Barrera3, Mary Jo Kupst4,5, Eugene A. Meyer6, Andrea F. Patenaude7,8 andKathryn Vannatta1,2*1The Research Institute at Nationwide Children’s Hospital, Columbus, OH, USA2The Ohio State University, Columbus, OH, USA3Hospital for Sick Children, Toronto, Ontario, Canada4Children’s Hospital of Wisconsin, Milwaukee, WI, USA5Medical College of Wisconsin, Milwaukee, WI, USA6Albert Einstein College of Medicine, Yeshiva University, Bronx, NY, USA7Dana-Farber Cancer Institute, Boston, MA, USA8Harvard Medical School, Boston, MA, USA

*Correspondence to:The Research Institute atNationwide Children's Hospital,Center for Biobehavioral Health,700 Children's Dr., Columbus,OH 43205, USA. E-mail:[email protected]

Received: 3 November 2011Revised: 31 August 2012Accepted: 8 September 2012

AbstractPurpose: The purpose of this study is to examine whether parental education, socioeconomic status, orfamily environment moderate the extent of academic achievement deficits in pediatric brain tumorsurvivors (PBTS) relative to classmate case-controls. PBTS are known to be at risk for cognitiveand academic impairment; however, the degree of impairment varies. Prior research has focusedon treatment risk, and efforts to examine the protective role of family resources and relationships havebeen lacking.

Methods: Pediatric brain tumor survivors (N = 164), ages 8–15 and 1–5 years posttreatment, wererecruited at five treatment centers in the United States and Canada. A case-control classmate, matchedfor age, gender, and race, was recruited for each survivor. The Wide Range Achievement Test, ademographic form, and the Family Environment Scale were administered in families’ homes. Treatmentdata were abstracted from medical charts.

Results: Pediatric brain tumor survivors demonstrated lower achievement than classmate-controlsin reading, spelling, and arithmetic. Parental education and socioeconomic status were associated withlevels of achievement demonstrated by PBTS but did not account for discrepancies between PBTS andclassmate-controls. Deficits in achievement relative to classmate-controls, across all academic domains,were greater for survivors in families lower in support and higher in conflict. These associationsremained after controlling for age at diagnosis, time since treatment, and whether treatment hadinvolved chemotherapy, focal, or whole brain radiation.

Conclusions: These results support the development of interventions to enhance family functioningas well as educational resources as part of intervention and rehabilitation services to optimizeacademic progress in children who have been treated for brain tumors.Copyright © 2012 John Wiley & Sons, Ltd.

Introduction

Brain tumors are the second most common form of pediat-ric cancer, affecting 2500 children under the age of 19each year in the United States and Canada [1,2]. Treat-ment advances over the past 30 years have led to a 20%decline in mortality [3,4], and overall survival rates nowexceed 65% [2,5]. This heightens concern about thequality of life of pediatric brain tumor survivors (PBTS)and creates an imperative to develop interventions thatcan optimize functional outcomes. These approaches mustrely on empirical evidence of risk factors and mechanismsthat contribute to key outcomes.Neurocognitive and academic impairment are well

documented for PBTS [6], and these morbidities maylead to diminished educational attainment, psychosocialdevelopment, and quality of life into adulthood [7,8].Relative to children never diagnosed with cancer, PBTSare more likely to (i) receive services for learning disabil-ities (19% vs. 7%), (ii) be enrolled in a special education

program (20% vs. 8%), (iii) experience academic pro-blems requiring extra tutoring or academic supportservices (46% vs. 23%), or (iv) repeat a grade (21% vs.9%) [9]. PBTS are less likely to graduate from highschool than other pediatric cancer survivors or healthycontrols [10].Academic outcomes vary between studies and among

PBTS within study samples [6]. Treatment factors havebeen the focus of the efforts to understand these differ-ences [11,12]. Academic deficits have been linked withcranial radiation and CNS-directed chemotherapy, withparticular risk associated with irradiation during earlychildhood [13,14]. Furthermore, core neurocognitivedeficits (e.g., inattention and short-term memory deficits)emerge as late effects and result in declining age-referenced intellectual and academic abilities over time.Attempts to preserve function while still maximizingsurvival by altering the timing or dosing of neurotoxictreatments have become a focus of cooperative groupclinical trials [13,14].

Copyright © 2012 John Wiley & Sons, Ltd.

Psycho-OncologyPsycho-Oncology 22: 1731–1737 (2013)Published online 12 October 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/pon.3202

Treatment factors may not, however, fully account forthe variability in academic achievement amongst survi-vors [15]. Sociodemographic variables, such as familysocioeconomic status (SES), are predictive of a child’srecovery from traumatic brain injury (TBI) [16], and thequality of parenting and family environment also appearto moderate academic achievement after TBI [17,18].Global indicators of family functioning, which take intoaccount family problem solving, roles, communication,affective responsiveness and involvement, and the extentto which set rules and procedures are used to govern fam-ily life, have been associated with intellectual functioningand memory in children recovering from TBI [19].Despite calls to examine whether family processes mod-

erate functional outcomes such as academic achievementfor cancer survivors, very few studies have performed this[12,20–22]. In a sample of 63 children with heterogeneousbrain tumor diagnoses, Carlson-Green explored whetherfamily variables predicted children’s cognitive and behav-ioral outcomes, above and beyond illness factors [23]. Acombination of illness and family variables (e.g., familystress level, number of parents in the home) was the stron-gest predictor of children’s intellectual functioning [23].Better medical and psychological outcomes for childrenwith cancer and other chronic illnesses have also beenfound among families characterized by high levels of cohe-sion and low levels of conflict [24], suggesting that thequality of the child’s family environment may contributeto academic functioning for PBTS.Previous work with PBTS has relied primarily on com-

parison to siblings or scores from norm-referenced samples.Comparisons with instrument norms fail to account for geo-graphic differences or the unique educational opportunitiesand environments each PBTS has experienced. Although asibling control group may address this shortcoming, it failsto match on the basis of age and possibly gender. Therefore,a matched-pair design, utilizing case-control classmates,was adopted in our research to control for child demo-graphic variables and systematic differences in schoolenvironment and general educational opportunities.On the basis of prior research [11–13], we expected to

find deficits in academic functioning for PBTS relativeto classmate-controls, and that larger deficits relative toclassmate-controls would be found for children who werediagnosed at a younger age, treated with radiation andchemotherapy, and further from treatment completion. Itwas hypothesized though that higher family SES and higherlevels of parental educationwould be associated with smallerdeficits in academic achievement among PBTS [16]. Finally,it was hypothesized that lower levels of family conflict andhigher levels of family support and control would be associ-ated with smaller deficits in academic achievement for PBTSrelative to classmate-controls [17–19,22,23].

Method

Procedures

PBTS were identified using tumor registries at five pediatriconcology centers in the United States and Canada. Childrenwere eligible if they (i) were 8–15 years old, (ii) had beentreated for an intracranial tumor, and (iii) had been off

treatment for 1–5 years without disease progression.Children were excluded if they (i) had a preexisting neuro-behavioral disorder; for example, neurofibromatosis ortuberous sclerosis, (ii) were not fluent in English, (iii) werehome-schooled, or (iv) received full-time special educationservices. These criteria were implemented to assure chil-dren were old enough to complete self-report measuresand belonged to a classroom in which sociometric dataassessing peer relations (not the focus of this paper) wouldbe valid.Parents of eligible children received a letter from their

child’s oncologist or neurosurgeon and were contactedby a study coordinator to explain the study, confirm eligi-bility, and seek permission to work with their child’sschool on a study of children’s behavior and relationshipswith peers at school. Collaborations were establishedwith schools, consents distributed, and data collected withclassmates of PBTS in primary elementary school classesor a required academic class of students in middle or highschool. A second phase involved one on one assessmentswith PBTS and their parents in their homes. The classmateof the same gender, race, and closest in birth date to eachPBTS who participated in each class was identified as acase-control. Parents of this child were contacted andinvited to participate in the second phase of the study. Ifcontrol families declined, the next most closely matchedclassmate was contacted. Potential controls were excludedif they, or another child in the home, had been treated fora serious illness that had lasted at least 6 months andrequired treatment by a medical specialist.During home-based assessment batteries, children in

both groups completed a standardized measure of academicachievement as well as checklists assessing social andemotional adjustment. Each caregiver in the home wasalso asked to complete multiple questionnaires regard-ing child and family functioning. If a parent in a two-caregiver home declined to participate, informationregarding their education and employment was obtainedfrom the participating parent. Parents provided consentand children assent for participation following approvalby local Institutional Review Boards. Families werecompensated for their time.

Participants

Across the five data collection sites, parents of 306 eligi-ble children were contacted for the first phase of theresearch. Data were collected in the classrooms of 216PBTS (71%). Subsequently, the families of 188 PBTS(87%) participated in the second phase of the study. Homevisits were completed with a classmate-control for 173(92%) of these children. In four cases, the PBTS wasunable to complete the primary measure of achievementbecause of sensory-motor impairment, and in five addi-tional instances, data for either the PBTS or the classmate-control was missing, for example, unable to be scored orthe child refused to complete the measure. This resulted ina final sample of 164 PBTS and case-control pairs with dataon the primary outcome measure. The mean age of PBTSand controls was 11.3 years (SD= 2.3), t (326)= 0.246,p> 0.05. Both groups included 76 girls (46%). Eighty-fivepercent of PBTS and 88% of controls were White, 7%

1732 E. Ach et al.

Copyright © 2012 John Wiley & Sons, Ltd. Psycho-Oncology 22: 1731–1737 (2013)DOI: 10.1002/pon

of PBTS and 8% of controls were Black, and 8% of PBTSand 4% of controls belonged to another racial group, w2

(2, 325) = 0.247; p> 0.05.PBTS (N= 164) were, on average, 7.8 years old at

diagnosis (SD = 3.0), and home data collection occurredan average of 3.1 years (SD= 1.7) after completion oftreatment. The sample included children treated for lowgrade astrocytoma (49%), medulloblastoma (22%), ep-endymoma (9%), craniopharyngioma (7%), germ cell(5%), and other tumors (7%). Most children (89%) hadundergone surgical resection, 43% had received chemo-therapy, and 43% cranial radiation. Thirty-three percentof the PBTS received both chemotherapy and radiationas part of their treatment. Focal radiation was documentedfor 34% of PBTS with a dosage range of 1800–5580 cGy,MDose = 4105 cGy, SD= 1421. Whole brain or craniosp-inal radiation was documented for 22% of PBTS with adosage range of 2300–4680 cGy, MDose = 3084 cGy,SD= 949.For the purposes of this study, the terms ‘mother’ and

‘father’ are used to refer to the child’s primary femaleand male caregiver living in the home. Data were providedby 153 mothers and 101 fathers of PBTS and 158 mothersand 92 fathers of controls. The majority of the participat-ing mothers (90%) and fathers (81%) were biological par-ents. PBTS (76%; n= 123) and classmate-controls (83%;n= 135) were equally likely to live in two-parent versussingle-parent families, w2 (1, 164) = 2.42; p= 0.120.

Child measures

Wide Range Achievement Test-3

The Wide Range Achievement Test-3 (WRAT-3) [25] is astandardized, performance-based measure that producesage-standard scores (M=100, SD=15) on three subscalesreflecting Reading, Spelling, and Arithmetic skills. Excel-lent split-half reliability (0.94 to 0.97) are reported andscores correlate (r=0.40 to 0.70) with other widely usedtests of academic achievement [25,26].

Parent measures

Demographic Questionnaire

This form records information about background charac-teristics of the participating child and each caregiver inthe home, including occupation and highest level of edu-cation completed. SES was computed using the RevisedDuncan Scale [27], on the basis of occupational prestige.When both parents were employed, the SES used to char-acterize the family reflects the higher of their two scores.

Family Environment Scale

The Family Environment Scale assesses the social environ-ment within the family using 90 ‘true’ or ‘false’ questions[28]. Factor analysis has found three higher order factors:Supportive, Conflicted, and Controlling [29]. The Support-ive factor taps mutual concern, shared interests, cohesion,and expressiveness. The Conflicted factor represents a lackof cohesion and organization as well as conflict in thehome. The Controlling factor reflects emphasis on control,achievement, morality and religion, and less independenceamong family members.

Analysis plan

Nondirectional, matched-pairs t-tests (a=0.05) comparedPBTS and classmate-controls in terms of sociodemographicand family environment variables as well as child perfor-mance in Reading, Arithmetic, and Spelling. Correlationsbetween PBTS and their classmate-controls on family vari-ables and achievement scores examined whether studyprocedures had resulted in dependent samples. Discrep-ancy scores were computed between each PBTS and theirclassmate-control for each academic domain by subtractingWRAT age-standard scores for classmate-controls from thatof PBTS, for example, ReadingDiscrepancy=ReadingBT�ReadingCC. Negative discrepancy scores indicated anachievement deficit for a PBTS, whereas a positive discrep-ancy indicated better achievement by the survivor relativeto their classmate-control. Independent group t-tests andPearson correlations examined whether achievement dis-crepancy scores varied as function of discrete and continu-ous variables reflecting treatment, sociodemographic (i.e.,family SES, maternal education, and paternal education),and family environment (i.e., Supportive, Conflicted, andControlling) risk factors for PBTS. This approach wastaken to allow moderator variables to remain continuouswhile accounting for the dependence of data created by thematched-pairs design. Finally, partial correlations assessedthe association of family factors and achievement deficitsafter controlling treatment effects. The sample of 164 PBTSand classmate-control pairs had sufficient power to detectsmall effect sizes for group differences (1�b= 0.72) andcorrelations (1�b= 0.74).

Results

Analyses of similarities and differences in the sociodemo-graphic and family environment characteristics variables

Table 1. Comparison of family demographic characteristics forpediatric brain tumor survivors (PBTS) and classmate-controls

PBTS Controls

ta p db rcM SD M SD

MotherEducation 14.3 2.1 14.7 2.2 �1.58 0.117 �0.19 0.25**Family Environment Scale

Supportive 30.0 7.0 32.4 6.5 �3.95 0.000 �0.36 0.02Conflicted �9.5 4.7 �10.4 4.5 2.09 0.038 �0.20 0.11Controlling 10.2 4.4 9.6 4.5 1.52 0.130 �0.13 0.20*

FatherEducation 14.4 2.5 14.7 2.5 �0.43 0.671 �0.12 0.25*Family Environment Scale

Supportive 29.5 6.2 29.8 7.1 �0.62 0.541 �0.04 0.12Conflicted �9.9 4.8 �9.9 5.3 �0.29 0.777 0.00 0.01Controlling 10.8 4.2 10.8 3.7 0.25 0.804 0.00 0.21

Family DemographicsSESd 53.4 21.2 58.3 21.9 �2.19 0.030 �0.23 0.22**

aTwo-tailed, dependent groups t-tests, df= 152 to 160 for mothers and df= 91 to 127for fathers, p denotes significance of t.bCohen’s d estimating effect size assuming independent groups [30].cCorrelation between scores for matched-pairs of PBTS and classmate-control familiesto evaluate group dependency.dRevisedDuncan scores of occupational prestige; when two parents reported occupations,the higher was used.*p≤ 0.05.**p≤ 0.01.***p ≤ 0.001.

1733Academic achievement in survivors


for families of matched pairs of PBTS and classmate-controls are summarized in Tables 1. Groups differed onfamily SES with significantly lower SES scores for PBTSfamilies, but the groups did not differ on parental education.Means for both groups reflect occupations in the technical,sales, and administrative support sector. According tomothers, the families of PBTS were also characterized bylower levels of support. The two groups did not differsignificantly on the amount of conflict or control reportedby either parent. Significant correlations between PBTSand classmate-controls for achievement, parental education,and family SES demonstrated dependency of data for thesegroups created by the matched-pairs design.Significant correlations were also found between PBTS

and their classmate-controls in all three domains of achieve-ment (Table 2). Matched-pair t-tests demonstrated thatachievement was lower in Reading, Spelling, and Arithmeticfor PBTS, and themagnitude of these deficits did not vary bydomain, F (2, 326) = 1.59; p=0.21. Conservative effect sizecomputations, assuming independent samples, yieldedvalues of Cohen’s d=�0.32 to �0.45 (Table 2) [30].Effect size estimates were larger when assuming groupdependence (d=�0.57 for Reading, d=�0.53 for Spelling,and d=�0.76 for Arithmetic) [31].

Comparisons of achievement discrepancy scores as afunction of treatment variables are summarized in Table 3.Significantly, larger discrepancies, reflecting deficits forPBTS relative to classmate-controls, were found in readingand arithmetic for children treated with chemotherapy aswell as those who received whole brain radiation. Largerdeficits for reading and spelling, but not arithmetic, wereassociated with younger age at diagnosis. There was no ev-idence that deficits in any academic domain were larger forchildren who were further from treatment completion.Contrary to hypotheses, neither maternal nor paternal

education level was associated with the magnitude ofdiscrepancies found between the academic achievementscores of PBTS and classsmate-controls (Table 4). A sig-nificant positive correlation was found between SES forfamilies of survivors and reading discrepancy scores.Negative discrepancy scores, reflecting deficits in readingfor PBTS, became smaller and approached zero at higherlevels of SES. Post-hoc analyses found significant corre-lations (p< 0.05) between Reading, Spelling, and Arith-metic achievement scores with family SES, r(163) = 0.28,0.23, and 0.22, maternal education, r(157) = 0.26, 0.20,and 0.21, and paternal education, r(122) = 0.21, 0.28, and0.23, within the PBTS sample. All nine correlations

Table 2. Comparison of academic achievement for pediatric brain tumor survivors (PBTS, n= 164) and classmate-controls (n= 164)

WRATa subscale

PBTS ControlsMean

Discrepancyb tc p dd reM SD M SD

Reading 100.8 15.2 105.8 13.2 �4.98 �3.63 0.000 �0.35 0.24**Spelling 99.3 15.5 104.1 14.2 �4.82 �3.38 0.001 �0.32 0.24**Arithmetic 93.5 17.0 100.5 13.9 �6.93 �4.85 0.000 �0.45 0.32***

aWRAT, Wide Range Achievement Test.bDifference between age-referenced standard scores on the WRAT for survivor-control pairs (e.g, ReadingBT�ReadingCC).cTwo-tailed, dependent groups t-tests, df = 163.dCohen’s d estimating effect size assuming independent groups [30].eCorrelation between scores for matched pairs of PBTS and classmate-control families to evaluate group dependency.*p≤ 0.05.**p≤ 0.01.***p≤ 0.001.

Table 3. Differences in academic achievement deficits as a function of age at brain tumor diagnosis, treatment, and time since treatment completion

WRAT reading discrepancya WRAT spelling discrepancya WRAT arithmetic discrepancya

M SD tb dc M SD tb dc M SD tb dc

ChemotherapyNo (n = 91) �1.92 16.34 2.57** 0.41 �3.15 16.81 1.35 0.21 �4.19 17.85 2.18* 0.35Yes (n= 69) �9.06 18.74 �7.09 20.01 �10.49 18.56

Whole brain radiationNo (n = 119) �2.86 16.51 2.58** 0.46 �3.98 17.30 1.24 0.22 �4.37 18.50 3.51*** 0.71Yes (n= 35) �11.60 21.06 �8.40 22.03 �16.46 15.63

Focal radiationNo (n = 99) �4.30 16.85 0.52 0.09 �4.28 18.30 0.53 0.09 �5.88 17.96 1.09 0.18Yes (n= 56) �5.88 19.73 �5.93 18.93 �9.23 19.38

r r rAge at diagnosis 0.23** 0.18* 0.00Time since treatment �0.04 �0.07 0.02

WRAT, Wide Range Achievement Test.aDifference between age-referenced standard scores on the WRAT for survivor-control pairs (e.g., ReadingBT�ReadingCC).bTwo-tailed independent group t-tests.cCohen’s d estimating effect size assuming independent groups [30].*p≤ 0.05.**p≤ 0.01.***p≤ 0.001.

1734 E. Ach et al.


remained significant (p< 0.05) even after controlling fortreatment risk factors, that is, administration of chemother-apy, focal or whole brain radiation, age at diagnosis, andtime since treatment completion. Although three of thesecorrelations did not reach statistical significance withinthe classmate-control sample, no significant differenceswere found between groups in the magnitude of thesecorrelations.Finally, significant correlations were found between

achievement discrepancy scores in all three academicdomains and both mother and father ratings of supportand conflict on the Family Environment Scale, as wellas for arithmetic and father report of control (Table 4).The more support and less conflict parents reported inthe family, the less academic impairment PBTS demon-strated relative to classmate-controls. In most instances,these correlations remained significant after controllingfor treatment effects.

Discussion

Increasing numbers of children treated for brain tumors nowbecome long-term survivors [5]. Many, but not all, of thesesurvivors will evidence deficits in academic achievementand progression that may have far-reaching consequencesfor employment, independent living, and quality of life[11,15]. The purpose of the current study was to investigatewhether deficits in academic achievement are moderated byfamily resources and processes that could be used to identifychildren at heightened risk and be possible targets innew, expanded intervention paradigms. This effort wasstrengthened by inclusion of a large, demographically, andmedically diverse sample of PBTS drawn from multipleinstitutions. Academic achievement was assessed using awell-established, standardized, performance-based measureof reading, spelling, and math skills rather than gradesor caregiver ratings of academic performance. Classmate-controls provided a comparison point for achievement levelsfor PBTS that controlled, at least in part, for variations ineducational opportunities and resources.

Consistent with previous research [6], PBTS in thisstudy demonstrated lower levels of achievement in mul-tiple domains relative to classmate-controls, althoughobtained scores on the WRAT for the full sample of PBTSwere not strikingly different than age-based norms. Previ-ously reported findings that achievement in PBTS variesas function of age at diagnosis and treatment with chemo-therapy or whole brain radiation [13,14] were replicated.Consequently, we evaluated whether achievement deficitsrelative to classmate-controls varied as a function of socio-demographic factors and family environment both with andwithout controlling for treatment variables. Interestingly,family resources that accounted for variance in achievementor achievement deficits did so whether treatment-associatedrisk for accounted for or not.Consistent with previous work [12,32–35], we found that

parental education and family SES were correlated withachievement scores obtained by PBTS; however, thesedifferences did not account for discrepancies betweensurvivors and their classmate-controls. Similarity in thefamily demographics of children attending the same schoolexplains these results. By using a design that attempted tocontrol for shared school environment, we also limited thevariance in achievement discrepancy scores that could beattributed to individual family demographic characteristics.One significant relationship was found; higher family SESwas correlated with smaller reading deficits for PBTS rela-tive to their classmate-controls. Reading deficits in PBTSmay reflect poor auditory attention, phonological proces-sing, and sequencing, leading to suggestions that remedialtraining and repetition of phonologic skills may reducereading deficits [36]. Higher SES families may read andallocate time to promoting skills that underlie reading withtheir children.In contrast, family environment did moderate the mag-

nitude of achievement deficits among PBTS relative totheir classmate-controls, and families of children attendingthe same school did not tend to be similar in family environ-ment. Reading, arithmetic, and spelling deficits all varied asa function of the degree of family support and conflict

Table 4. Association of achievement deficits with family environment and demographic characteristics reported by parents of braintumor survivors

WRAT reading discrepancya WRAT spelling discrepancya WRAT arithmetic discrepancya

Family Environment Scale r Partial correlationb r Partial correlationb r Partial correlationb

Mother reportSupportive 0.19* 0.18* 0.16* 0.14 0.17* 0.16Conflicted �0.30*** �0.29*** �0.26*** �0.26*** �0.17* �0.18*Controlling 0.04 �0.01 0.00 0.01 �0.03 �0.08

Father report 0.22* 0.30** 0.21* 0.28** 0.23* 0.29**SupportiveConflicted �0.25* �0.27** �0.23* �0.27** �0.21* �0.25*Controlling 0.16 0.08 0.19 0.17 0.25* 0.14

Demographic factorsMaternal education 0.04 0.02 0.07 0.04 0.08 0.01Paternal education 0.07 0.08 0.13 0.14 0.10 0.13Family SESc 0.19** 0.16* 0.12 0.11 0.11 0.11

WRAT, Wide Range Achievement Test.aDifference between age-referenced standard scores on the WRAT for survivor-control pairs (e.g., ReadingBT�ReadingCC).bPartial correlations controlling for treatment, (chemotherapy, focal, and whole brain radiation), age at diagnosis, and time since treatment.cSocioeconomic status (SES) based on Revised Duncan Scores of occupational prestige; when both parents reported occupations, the higher of the two was used.*p≤ 0.05.**p≤ 0.01.***p≤ 0.001, two-tailed; df = 157 (mothers), df = 124 (fathers).



reported by mothers and fathers. With increasing amountsof supportiveness; reflecting cohesion, expressiveness,shared interests, and mutual concern in the home, survivorsdemonstrated smaller achievement deficits. Conversely,larger deficits were demonstrated by children in homescharacterized by greater conflict, less organization, andcohesion. These results are consistent with literature high-lighting the importance of family environment for childrenrecovering from TBI and expand upon limited relatedresearch with PBTS [18,19,23].These data do not provide detail about the mechanisms

by which family environment may preserve or enhanceacademic achievement for PBTS. It is possible that sup-portive families provide extra time, attention, and assistancein developing academic skills and completing schoolwork,thereby compensating for neurocognitive deficits thatresult from treatment. Conflict in the home may representcompeting demands for caregiver time or disagreementsabout how to respond to child impairment and needs.Less conflicted and more supportive families may spendmore time engaged in activities that foster a child’sintellectual development, self-esteem, and self-efficacy,factors that may generally promote academic engage-ment and performance [37].There are limitations to this research, including the

cross-sectional design and possible bias introduced bythe eligibility criteria. In addition to examining mechanismsof influence, a longitudinal design would be better ableidentify bidirectional relationships between achievementdeficits and family processes. Our approach emphasizedthe associations of individual family characteristics withacademic deficits and did not examine unique contributionsthat took into account correlated indicators of family envi-ronment or resources. Rather, we focused on analyses thatexamined whether each family indicator accounted forachievement while controlling for treatment risk and sharedschool environment. Our assessment of family variableswas limited to questionnaires completed by caregivers.Future studies might consider assessments of family inter-actions from other perspectives (e.g., child-report) or meth-ods (e.g., in-home or laboratory-based observations offamily interactions) [38]. The current study excluded PBTSif they did not spend part of their school day in a main-stream classroom, perhaps excluding children with thegreatest academic difficulties. Current education policiesfavor mainstreaming whenever possible; therefore, our

participants represented a broad range of abilities andincluded children who received less restrictive special edu-cation services. Of the PBTS initially identified as eligible,approximately 10% were excluded because they werehome-schooled or in self-contained, special educationsettings. Had the sample included these more impairedchildren, effects might have been larger.Increased understanding of which PBTS are at greatest

risk for achievement deficits and which processes accountfor these different outcomes could help healthcare provi-ders, schools, and families to titrate appropriate servicestoward those at greatest need. The current results reinforceconcerns about academic risk associated with socioeco-nomic disadvantage for PBTS, even if this risk is sharedby classmates experiencing the same classroom environ-ment. Identification of mechanisms that account for differ-ences in academic outcomes will inform the developmentof family-based interventions aimed at shaping the homeenvironment and altering family dynamics to best facili-tate a child’s rehabilitation. Previous intervention researchhas focused on pharmacotherapy [39] and cognitive reha-bilitation programs aimed at remediating core neurocogni-tive (e.g., attention and working memory deficits) [40,41]or phonological skills [12] thought to underlie pooracademic performance for PBTS. Intervention programstargeting characteristics of the family environment couldnot be identified in the current literature and should beconsidered in future work. Attention to family environ-ment in future interventions may be particularly warrantedgiven evidence in this research that mothers, if not fathers,of PBTS report that their families demonstrate less supportand more conflict than families of control children. Ideally,intervention programs might integrate remediation ofchild skills and enhancement of family resources thatcan facilitate learning and academic engagement. Mostof all, continued research is needed to develop and estab-lish the efficacy of integrative, creative treatmentapproaches that will maximize functioning and enhancethe opportunity for PBTS to lead fulfilling, independent,and productive lives.

Acknowledgements

This work was supported by awards from the American CancerSociety (RSGPB-03-098-01-PBP) and the National Cancer Institute(RO3 CA097740-02).

References

1. Canadian Cancer Society. Canadian CancerStatistics, 2008. Available from: http://www.]cancer.ca/Canada-wide/About%20cancer/Cancer%20statistics/Cancer%20in%20young%20adults.aspx?sc_lang=en [April 14, 2011].

2. National Cancer Institute. SEER Cancer Statis-tics Review, 1975-2007. National Cancer Insti-tute, 2009. Available from: http://seer.cancer.gov/csr/1975_2007/ [February 6, 2010].

3. Gurney JG, Wall DA, Jukich PJ et al. The con-tribution of nonmalignant tumors to CNS tumorincidence rates among children in the UnitedStates.Cancer Causes Control 1999;10:101–105.

4. Packer RJ. Progress and challenges in child-hood brain tumors. J Neurooncol 2005;75:239–242.

5. Jemal A, Siegel R, Ward E et al. Cancer statis-tics 2009.CACancer J Clin 2009; 59:225–249.

6. Robinson KE, Kuttesch JF, Champion JEet al. A quantitative meta-analysis of neuro-cognitive sequelae in survivors of pediatricbrain tumors. Pediatr Blood Cancer 2010;55:525–531.

7. Mostow EN, Byrne J, Connelly RR et al.Quality of life in long-term survivors ofCNS tumors of childhood and adolescence. JClin Oncol 1991;9:592–599.

8. Kieran MW, Walker D, Frappaz D et al. Braintumors: from childhood through adolescenceinto adulthood. J Clin Oncol 2010;28:4783–4789.

9. Barrera M, Shaw AK, Speechley KN et al.Educational and social late effects of child-hood cancer and related clinical, personal,

and familial characteristics. Cancer 2005;104:1751–1760.

10. Mitby PA, Robison LL, Whitton JA et al.Utilization of special education services andeducational attainment among long-term sur-vivors of childhood cancer: a report from theChildhood Cancer Survivor Study. Cancer2003;97:1115–1126.

11. Mulhern RK, Butler RW. Neurocognitivesequelae of childhood cancers and their treat-ment. Pediatr Rehabil 2004;7:1–14.

12. Palmer S. Neurodevelopmental Impact onChildren Treated for Medulloblastoma: AReview and Proposed Conceptual Model.Dev Disabil Res Rev 2008;14:203–210.

13. Mulhern RK, Palmer SL, Merchant TEet al. Neurocognitive consequences ofrisk-adapted therapy for childhood

1736 E. Ach et al.


http://seer.cancer.gov/csr/1975_2007/

http://seer.cancer.gov/csr/1975_2007/

medulloblastoma. J Clin Oncol 2005;23:5511–5519.

14. Askins MA, Moore BD, 3rd. Preventing neuro-cognitive late effects in childhood cancer survi-vors. J Child Neurol 2008;23:1160–1171.

15. Mulhern RK, Merchant TE, Gajjar A et al.Late neurocognitive sequelae in survivors ofbrain tumours in childhood. Lancet Oncol2004;5:399–408.

16. Taylor HG, Yeates KO, Wade SL et al. Aprospective study of short- and long-term out-comes after traumatic brain injury in children:behavior and achievement. Neuropsychology2002;16:15–27.

17. Yeates KO, Swift E, Taylor HG et al. Short-and long-term social outcomes followingpediatric traumatic brain injury. J Int Neurop-sychol Soc 2004;10:412–426.

18. Taylor HG, Yeates KO, Wade SL et al. Influ-ences on first-year recovery from traumaticbrain injury in children. Neuropsychology1999;13:76–89.

19. Max JE, Roberts MA,Koele SL et al. Cognitiveoutcome in children and adolescents followingsevere traumatic brain injury: influence ofpsychosocial, psychiatric, and injury-relatedvariables. J Int Neuropsychol Soc 1999;5:58–68.

20. Barrera M, Atenafu E, Pinto J. Behavioral,social, and educational outcomes after pediatricstem cell transplantation and related factors.Cancer 2009;115:880–889.

21. Patel SK, Carlson-Green B. Commentary:toward greater integration and specificity inconceptual models of neurocognitive function-ing in childhood cancer survivors. J PediatrPsychol 2005;30:85–88.

22. Baumrind D. The influence of parenting styleon adolescent competence and substance use.J Early Adolesc 1991;11:56–95.

23. Carlson-Green B, Morris RD, Krawiecki N.Family and illness predictors of outcomein pediatric brain tumors. J Pediatr Psychol1995;20:769–784.

24. Barkat L, Pulgaron E, Daniel L. PositivePsychology in Pediatric Psychology. In Hand-book of Pediatric Psychology, Roberts M,Steele RG (eds.). The Guildford Press: NewYork, 2009; 763–773.

25. Wilkinson GS. Wide Range Achievement Test-Revision 3. Jastak Association: Wilmington,DE, 1993.

26. Woodcock R. In Essentials of WoodcockJohnson III Tests of Achievement Assess-ment, Mather N, Welding B (eds). RiversidePublishing: Rolling Meadows, IL, 2001.

27. Nakao K, Treas J. The 1989 SocioeconomicIndex of Occupations: Construction from the1989 Occupational Prestige Scores, Report.University of Chicago, National OpinionResearch Center: Chicago, 1992.

28. Moos RH, Moos BS. Family environmentscale manual (2nd edn). Consulting Psycholo-gists: Palo Alto, CA, 1986.

29. Kronenberger W, Thompson J. Dimensionsof Family Functioning in Families withChronically Ill Children: A Higher OrderFactor Analysis of the Family EnvironmentScale. J Clin Child Psychol 1990;19:380–388.

30. Dunlap W, Cortina J, Vaslow J et al. Meta-Analysis of Experiments with Matched Groupsor Repeated Measures Designs. PsycholMethods 1996;1:170–177.

31. Cohen J. Statistical power analysis for thebehavioral sciences. Lawrence ErlbaumAssoci-ates: Hillsdale, 1988.

32. Mabbott D, Spiegler B, Greenberg M et al.Serial evaluation of academic and behavioraloutcome after treatment with cranial radiation

in childhood. J Clin Oncol 2005;23:2256–2263.

33. Mulhern R, Palmer S. Neurocognitive lateeffects in pediatric cancer. Curr Probl Cancer2003;27:177–197.

34. Palmer SL, Reddick WE, Gajjar A. Under-standing the cognitive impact on childrenwho are treated for medulloblastoma. J PediatrPsychol 2007;32:1040–1049.

35. Taylor H, Alden J. Age-related differences inoutcomes following childhood brain insults:an introduction and overview. J Int Neuropsy-chol Soc 1997;3:555–567.

36. Palmer S, Mitchell A, Thompson K et al.Unmet needs among adolescent cancer patients:a pilot study. Palliat Support Care 2007;5:127–134.

37. Marsh HW, Martin AJ. Academic self-concept and academic achievement: Relationsand causal ordering. Br J Educ Psychol 2010;81:59–77.

38. Dunn MJ, Rodriguez EM, Miller KS et al.Direct observation of mother-child communi-cation in pediatric cancer: assessment ofverbal and non-verbal behavior and emotion.J Pediatr Psychol 2011;36:565–575.

39. Thompson D, Iachan R, Overpeck M et al.School Connectedness in the Health Behaviorin School-Aged Children Study: The Role ofStudent, School, and School NeighborhoodCharacteristics. J Sch Health 2006;76:379–386.

40. Butler RW, Hill JM, Steinherz PG et al.Neuropsychologic effects of cranial irradiation,intrathecal methotrexate, and systemic meth-otrexate in childhood cancer. J Clin Oncol1994;12:2621–2629.

41. Hardy KK, Bonner MJ, Masi R et al. Psycho-social functioning in parents of adult survi-vors of childhood cancer. J Pediatr HematolOncol 2008;30:153–159.



Documents

Family factors associated with academic achievement deficits in pediatric brain tumor survivors