Cancer Teenagers Psy

the low incidence of cancer in children, greater na-tional and international collaboration might lead toa clearer understanding the psychological sequelae.This review attempts to synthesize current work inthe area, to stimulate more thorough and defensiblemethodologies in the future.

Due to the number of sequelae associated withdifferent therapies, interest in the long-term physi-cal outcome of cancer treatment has grown steadilyin recent years (Hawkins & Stevens, 1996). How-ever, the long-term psychological and behavioral con-sequences of cancer also merit study. In attempts tosummarize these findings, at least five reviews of

Journal of Pediatric Psychology, Vol. 25, No. 6, 2000, pp. 449460

Examining the Psychological Consequences ofSurviving Childhood Cancer: Systematic Reviewas a Research Method in Pediatric Psychology

Christine Eiser, BSc, PhD, Jacqueline J. Hill, BSc, MSc, and Yvonne H. Vance, BSc, MScUniversity of Sheffield

Objective: To report the results of a systematic review to determine the psychological consequences of sur-

viving childhood cancer.

Methods: Searches were conducted using Psyclit, Medline, Cinahl, and Bids and articles selected on the ba-

sis of predened criteria. Key information was extracted to data sheets and these were rated by two

coders.

Results: Twenty studies were identied, seventeen from the United States. Survivors did not show decits

in measures of anxiety, depression, or self-esteem when compared with population norms or matched con-

trols. Survivors of some cancers (bone tumors) have poorer outcomes.

Conclusions: The results of this review support ndings of previous descriptive reviews. Methodological

problems include poorly reported medical information (for example, time since diagnosis), heterogeneous

samples, self-selection of participants, poorly chosen/lack of suitable measures, and a lack of longitudinal

work. Findings are discussed in terms of the need for cross-cultural work on adjustment to childhood can-

cer, the need for studies to take on a more developmental approach, and for greater national and interna-

tional collaboration.

Key words: childhood cancer; pediatric psychology; systematic review; psychological consequences of cancer.

Improvements in treatment and coordination ofcare have contributed to increased survival follow-ing diagnosis of childhood cancers so that now al-most 70% of children may expect to survive at least5 years following diagnosis (Stiller, Allen, & Eatock,1995). However, because many survivors experienceresidual physical, behavioral, or social sequelae as-sociated with the disease or its treatment, consider-able implications remain both for individuals andhealth services (Hawkins & Stevens, 1996). Given

2000 Society of Pediatric Psychology

All correspondence should be sent to Christine Eiser, Department ofPsychology, University of Sheffield, Western Bank, Sheffield, S10 2TP,England. E-mail: [email protected].

this literature have been published (Chang, 1991;Eiser, 1998; Eiser & Havermans, 1994; Kazak, 1994;Zeltzer, 1993). All identify methodological prob-lems in empirical work, particularly the focus onmeasures of maladjustment and depression ratherthan coping. All also conclude that there is no evi-dence to suggest that survivors inevitably fare badly,but at the same time they point to subgroups whoshow more serious adjustment and emotional prob-lems. The more recent reviews (Eiser, 1998; Kazak,1994) emphasize that most empirical work hasfailed to distinguish between survivors of differentcancers; consequently, more is known about somecancers than others (e.g., Hodgkins disease or acutelymphoblastic leukemia [ALL]). Methodologicallimitations are widely noted in this field, especiallythe reliance on cross-sectional work rather than lon-gitudinal research. Longitudinal studies demandboth time and money but are unique in their abilityto determine causal relations and trace characteris-tics that identify at-risk groups over time.

Given the low incidence of childhood cancer,progress in treatment has been achieved in largepart through national and international collabora-tion which has resulted in more rapid understand-ing of the etiology of cancer and in advances inmedical care. It is disappointing that such collabo-ration has not been evident for more psychosocialissues. Differences in perception of cancer and inattitudes toward individuals with cancer are likelyto enhance understanding of adaptation and cop-ing. As survival increases, issues of coping becomeparamount.

These reviews have raised awareness about therange of outcomes for survivors of childhood cancerbut, like all narrative reviews, are themselves thesource of bias. For example, investigators may failto search the complete spectrum of journals rele-vant to the issue; they may fail to include all studieswithout specifying the criteria on which exclusionsare made; and they may interpret the literature ac-cording to a predefined hypothesis. Failure to spec-ify the specific purpose of the review may limit theextent to which the work has any impact on clinicalpractice and patient management.

For these reasons, systematic reviews are in-creasingly recommended, especially in health sci-ences work, where reviews of research evidence areconsidered important to allow for planning servicesand allocating resources. Systematic reviews are or-ganized round a specific topic and should yield in-

formation of direct relevance to health planners.Systematic reviews differ from other types of re-view in that they adhere to strict scientific design inorder to make them more comprehensive, minimisethe chance of bias, and so ensure their reliability.Rather than reflecting the bias of the authors or be-ing based only on a selection of the published litera-ture, they contain a comprehensive summary of theevidence (Cook, Sackett, & Spitzer, 1995).

A number of guidelines for conducting system-atic reviews have been described (Oxman, 1994). Itis recommended that the work is conducted ineleven phases. Phase 0 includes an assessment ofthe need for the specific review and must involvereference to potential sources including computerdatabases and the Cochrane Collaboration to deter-mine that no comparable review has been, or is be-ing, undertaken. Phases 1 and 2 involve planningthe review. A preliminary assessment of the poten-tial scope of the review needs to be undertaken anda decision made as to whether answering the speci-fied research question is possible. A protocol needsto be written in which questions are defined andmethods determined. The protocol should state thedatabases to be searched and the criteria for inclu-sion of a study into the review. A data extractionsheet must be described that summarizes the maininformation to be extracted from each individualarticle in the review.

Extensive literature searching is the main task inphase 3. Both computerized and manual searchesare recommended to ensure comprehensive and un-biased study identification. In phase 4, decisions aremade about which studies will appear in the review.Selection criteria need to be identified and pilotedto ensure they can be reliably interpreted. There areinevitable decisions to be made about how narrowinclusion criteria should be. Generalizability of theresults will be affected by the strictness of the crite-ria adopted. Too liberal inclusion criteria can makeinterpretation of the results more difficult. Tables ofstudies included and excluded from the reviewshould be constructed. In phase 5, the validity ofstudies needs to be determined. This reflects the de-gree to which a study is subject to bias. Decisionsabout validity are best made with reference to a spe-cially developed checklist. Data extraction is carriedout at phase 6. Reference should be made to theprotocol to determine the information to be ex-tracted from studies: bibliographic details, descrip-tions of the setting, study population, the nature

450 Eiser, Hill, and Vance

6. Are there differences in psychological out-comes dependent on the nature of the specific can-cer diagnosis?

Method

We had few difficulties in phases 0 and 1 aboutmaking a decision about the scope of the review. Asargued above, the decision to conduct a systematicreview of survivors of cancer was based on increasedsurvival and recognition of late psychological andphysical effects.

An initial search of the literature suggested thatpsychosocial effects of childhood cancer have beenassessed in a number of ways: (1) learning, intelli-gence, and academic achievements (Mulhern, 1994);(2) lifestyles, especially health behaviors such assmoking or drinking (Haupt et al., 1992), or occupa-tional and vocational status (Green, Zevon, & Hall,1990); and (3) general mental health issues, includ-ing anxiety and depression. The literature on learn-ing has been reviewed elsewhere (cf. Mulhern,1994). The focus on health behaviors is relativelyrecent, and we felt there was not sufficient evidenceto justify a review. This review therefore focuses spe-cifically on studies involving general mental healthissues.

As recommended, we followed the guidelinesdescribed above in executing the search. The reviewwas carried out using the following databases:Psyclit, Medline, Cinahl, and Bids. Cancerlit andPub.Med. were also accessed but revealed very fewarticles not covered by the preceding databases. Thefollowing key words were used as the search criteria;childhood survivors of cancer, adolescent survivorsof cancer, quality of life and children with cancer,outcome measureschildrencancer. In addition,we hand-searched recent issues of key journals andidentified additional references from those providedin identified papers.

Inclusion Criteria

Articles were included on the basis of (1) year of pub-lication: studies were included if published withinthe last decade (i.e., since 1990); (2) language: En-glish language articles only; and (3) method: studieswere included if they used standardized measuresand conducted statistical tests to compare scoreswith population norms or matched controls. Refer-

and delivery of the intervention, outcome measuresused, and results. Information about sample size, at-trition rates, and aspects of the methodology thatmight affect results also need to be recorded. Dataextraction should be performed by at least two cod-ers and disagreements discussed. Where data ismissing or unclear, attempts should be made tocontact the authors for further information.

The aim of phase 7 is to draw together resultsof the primary studies identified in phase 6 usingquantitative or qualitative methods. Where appro-priate, meta-analyses should be conducted. Phase 8involves writing the report, which should includereasons for the review, methods used, tables of stud-ies included and excluded, discussion of thestrength of the evidence, practical implications, andsuggestions for dissemination of the findings. Inphase 9, the review is assessed by a panel of expertsand meetings held to discuss the relevance of thework. The submission of the final report occurs inphase 10, which may include recommendations forfurther work.

The purpose of this article is to report our expe-rience in conducting a systematic review concernedwith psychological outcomes for survivors of child-hood cancer. In accordance with recommendationsin phase 0, we first identified the specific questionsto be asked and made sure that no comparable re-view was being undertaken. The review was justifiedon the following grounds: first, current literature ispublished in medical, nursing, and psychologyjournals and it was considered useful to provide asingle source; second, it is important to determinethe psychological and behavioral outcomes amongsurvivors in order to provide information for newfamilies; third, such information is useful at a pub-lic health level in determining the extent and typeof support likely to benefit families. The specificquestions to be addressed were:

1. What are the main outcome measures used toassess the impact of cancer on survivors?

2. What does this literature tell us about the so-cial and psychological outcomes for survivors ofchildhood cancer?

3. Do parents and teachers differ from survivorsin their views of the impact of cancer?

4 Can clinical or demographic factors modifythe outcomes of treatment?

5. Is there any evidence of posttraumatic stressdisorder (PTSD) among survivors?

Surviving Childhood Cancer: Systematic Review 451

ence Manager (Institute for Scientific Information,1997) was used as the bibliographic software pack-age to organize the relevant references.

Results

Twenty studies met the inclusion criteria and aresummarized in Table I. Two further articles (Bo-man & Bodegard, 1995; Greenberg & Meadows,1991) were excluded on the grounds that theysolely used interview methodologies and no stan-dardized measures. Of the included articles, 17 werefrom the United States, 2 from the United Kingdom,and 1 from the Netherlands. Fourteen focused ongeneral measures of adjustment and mental health,two focused specifically on social functioning, andfour on posttraumatic stress disorder. There werethree prospective studies (Kazak, Christakis, Alder-fer, & Coiro, 1994; Kupst et al., 1995; Noll, Bukow-ski, Davies, Koontz, & Kulkarni, 1993), the re-mainder were cross-sectional.

Age of subjects ranged from 3 (Butler, Rizzi, &Handwerger, 1996) to 37 years (Gray et al., 1992a,1992b). Sample sizes varied from 20 (Olson, Boyle,Evans, & Zug, 1993) to 309 (Barakat et al., 1997).

1. What Are the Main Outcome MeasuresUsed to Assess the Impact of Cancer onSurvivors?

There has been little consensus regarding the mostappropriate measures, which included assessmentof self-esteem (11), anxiety (5), depression (3), socialskills (2), body image (2), loneliness (2), mood (2),personality (3), coping (1), health status/quality oflife (1), social desirability (2), intrusiveness (4), andsymptoms (2). These are all generic measures. Nostudies used cancer-specific outcome measures.Three studies included specially developed inter-views with standardized questionnaires. In fourstudies, PTSD was determined. A smaller number ofmeasures were used for completion by parents:child behavior (4), family functioning (3), parentaldistress (1), parental social support (1), parent anxi-ety (1), and parenting stress (1). Measures used forteachers included the Child Behavior Checklist(teacher report form; Achenbach, 1991); the taxon-omy of problem situations (Dodge, McClaskey, &Feldman, 1985) and the cancer-specific teacher be-havior rating scale (Deasey-Spinetta & Spinetta,1980).

2. What Are the Social and PsychologicalOutcomes for Survivors of ChildhoodCancer?

Comparisons with population norms. Seven studies in-cluded comparisons with population norms, six in-cluded both population norms and a control group,and seven relied on a control group only. Only onestudy (Eiser et al., 1997) reported more symptoms insurvivors when compared with norms; the majority(n 5) reported no differences. One study (Elkin,Phipps, Mulhern, & Fairclough, 1997) reported thatsurvivors had fewer symptoms compared withnorms.

Eiser et al. (1997) studied 41 patients all ofwhom had a lower limb tumor and had beentreated by limb salvage. The group had worse scoresthan population norms (Jenkinson, Coulter, &Wright, 1993) and these were significant on sub-scales measuring physical functioning, physical roleperformance, pain, general health, and social func-tioning. This may reflect the fact that many bonetumor survivors continue to have problems withmobility and pain and need further treatment. Thisstudy also included some survivors who were con-siderably older than many of those in other studies.These outcomes may therefore partly depend on thefact that patients were treated before the introduc-tion of modern protocols, or that the effects of can-cer survival become cumulative over time.

The remaining studies reported no differencesbetween survivors and population norms. In termsof self-esteem, no differences between survivors andage-appropriate norms were reported by Anholt,Fritz, and Keener, (1993). Noll et al. (1993) reportedsimilar results for depression, and Radcliffe, Ben-nett, Kazak, Foley, and Phillips (1996) for self-perception. This latter study is particularly impor-tant as it focuses on survivors of CNS tumors, whomight be expected to have poorer outcomes. Kupstet al. (1995) studied a group of survivors (16 males,12 females; mean age 19 years) for 10 years fol-lowing diagnosis. These survivors represented 64children originally recruited on diagnosis of ALLand assessed at intervals throughout treatment(Kupst & Schulman, 1988). Symptom scores (Dero-gatis, 1992) and 10-year follow-up assessments werewithin normal ranges.

Based on a standardized measure of symptomreport, Elkin et al. (1997) found that survivors re-ported fewer symptoms (i.e., had better health) incomparison with population norms.



Table I. Summary of Studies Included in the Review

Age of Time sinceSample children Study Parent diagnosis Comparison

Study (n) (range; yrs) origin sample (mean) group Results

Anholt et al. 62 718 USA No 20 monthsa 120 healthy Self concept was similar to population

(1993) children; norms norms.

Survivors rated their school status,

behavior, overall happiness, and

satisfaction more positively than

controls.

Barakat et 309 820 USA Yes 5.86 yearsa 219 healthy No difference in PTSD from population

al. (1997) families; norms norms.

Parents of survivors reported more PTSD

than control parents.

Butler et al. 42 316 USA Yes 35 monthsa Norms Incidence of Posttraumatic Stress

(1996) Disorder (PTSD) no different from

population norms.

Eiser et al. 41 828 UK Yes 6.8 years Norms Scores below population norms

(1997) especially on measures of physical

functioning, physical role performance,

pain, general health, and social

functioning.

Elkin et al. 161 14.530.9 USA No 9.5 years Norms 75% showed some residual cosmetic

(1997) impairments.

64% demonstrated physical

impairment.

Gray et al. 62 1837 USA No 14.6 years Norms; peers No signicant differences between

(1992a, (selected by survivors and peers on standardized

1992b) survivors) inventories, story-telling, or physical

symptoms.

Compared with peers, survivors were

less satised with social relationships;

showed greater concern about

infertility; expressed more perceived

control and more satisfaction with their

degree of autonomy; were more likely

to prefer interacting with others.

Kazak et al. 59 1015 USA Yes 96.20 months Norms No differences from population norms

(1994) in levels of adjustment.

Few changes over 1 year testing.

Survivors with learning difculties had

more adjustment problems.

Kazak et al. 130 819 USA Yes 5.79 yearsa 155 healthy No differences in PTSD symptoms

(1997) children; norms between survivors and controls.

Predictors of PTSD included family

functioning and social support; no

effects of demographic or clinic

variables on PTSD.

Kupst et al. 28 1430 USA Yes 10 years Norms Survivors and parents were generally

(1995) well-adjusted.

Survivors coping and perceived

adjustment were positively related to

SES and mothers coping and negatively

related to academic problems.

No differences from population norms.


Madan- 25 1218 USA Yes 8.4 years 16 healthy children No major difculties on measures of

Swain et al. social competence, overall coping, and

(1994) family communication.

Survivors reported body image and

adjustment difculties.

Noll et al. 19 (time 1); 1118 USA Yes 62.5 months 17 healthy children; No differences between survivors and

(1993) 17 (time 2) norms controls in psychological functioning.

Depression scores were not signicantly

different from population norms.

No differences were found between the

groups on teacher reports.

Olson et al. 20 616 USA Yes 40 healthy children Survivors showed poorer social

(1993) competence, school performance, more

behavioral and academic problems than

controls.

No differences on self-esteem, family

conict, physical functioning, social

skills, sense of control over health.

Pendley et 21 1121 USA Yes 17 monthsa Healthy children Healthy controls and survivors did not

al. (1997) recruited from local differ on measures of body image,

advertisements attractiveness, loneliness, social anxiety,

and school absenteeism.

Radcliffe et 38 618 USA Yes 25 years Norms Compared with norms, children rated

al. (1996) themselves as less anxious, depressed,

and athletically competent, but similar

in terms of self-perception.

Mothers rated their children as less

competent than childrens self-report.

Teachers did not report differences

between survivors and norms.

Sloper et al. 31 818 UK Yes 5 years Healthy children No differences in self-ratings of anxiety

(1994) or self-esteem with controls.

Teachers rated survivors to have greater

difculties than controls with

concentration, academic progress, and

popularity.

Stern et al. 48 1423 USA No 2.79 yearsa 40 healthy Survivors were relatively well-adjusted,

(1993) adolescents but had a less positive self-image in

terms of their social and sexual selves.

Stuber et al. 64 719 USA Yes 6.7 years Norms Compared with norms, 12.5% survivors

(1996) showed severe PTSD symptoms.

van 95 812 Netherlands Yes 2 yrs 32% 90 children from Survivors showed more social and

Dongen- 25yrs 42% local schools internalizing problems than controls.

Melman et 5yrs 26%a On most measures, female survivors did

al. (1995) not show signicantly more serious

adjustment problems than healthy

controls.

Vannatta et 28 818 USA No 3 years 28 classroom years Compared with peers, survivors received

al. (1998) fewer friendship nominations.

Peers perceived survivors to be more

sick, fatigued, and absent from school.

aTime since treatment completed.

Table I. Continued

Age of Time sinceSample children Study Parent diagnosis Comparison

Study (n) (range; yrs) origin sample (mean) group Results

4. Can Clinical or Demographic FactorsModify the Outcomes of Treatment?

It might be expected that outcomes would be mod-erated by family, clinical, or demographic factors,so that children with poorer clinical status wouldhave poorer psychological outcomes. In addition,social and family circumstances may be expected tomoderate outcomes. Consistent findings have notbeen reported.

Family variables were described by Sloper et al.(1994) and Kupst et al. (1995). Sloper et al. (1994)found that parents who report more psychologicaldistress themselves also report more behavior prob-lems in their child. Kupst et al. reported that highersocial class and better maternal coping were associ-ated with better adjustment among survivors. Thosewith school problems had poorer adjustment thanthose without. Age, gender, and previous copingwere not associated with any of the outcome vari-ables.

Other work has assessed the contribution ofdemographic or clinical factors. Van Dongen-Melman et al. (1995) reported worse outcome forthose treated by CNS radiation and for those whoare overweight. Particular problems were noted formale survivors both from parent and self-report.Elkin et al. (1997) reported that older patients hadworse outcomes in terms of somatization, anxiety,hostility, phobic anxiety, and psychoticism, as wellas a global index of symptomatology (Derogatis,1977). A single relapse more than doubled the likeli-hood of higher scores on the obsessive-compulsiveand paranoid ideation scales. There is thus a sugges-tion from two studies that family factors are impor-tant, and three others implicate treatment variables.

5. Is There Any Evidence of PTSD AmongSurvivors?

Incidence of PTSD. Four studies have included assess-ments of the incidence of PTSD among survivors.Suggestions that survivors reported more symptomsof PTSD than would be expected from populationnorms were made by Stuber, Christakis, Houskamp,and Kazak (1996). They reported that more thanhalf their sample of survivors reported specificsymptoms of PTSD. These included bad dreams,feeling afraid or upset when they think about can-cer, feeling alone inside, and feeling nervous. In to-tal, 12.5% of survivors scored at a level that would

Comparisons with control groups. In addition tothe comparisons made with population norms, An-holt et al. (1993) compared survivors with a normalgroup matched for chronological age. Survivors hadmore positive self-esteem, especially in terms ofschool status, behavior, and overall satisfaction.However, no differences between survivors and con-trols were reported by Gray et al. (1992a, 1992b),Stern, Norman, and Zevon (1993), Sloper, Lar-combe, and Charlton (1994), Madan-Swain et al.(1994), and Pendley, Dahlquist, and Dreyer (1997).Noll et al. (1993) compared adaptation in survivorswith matched controls (17 dyads) in a 2-year longi-tudinal study. Data were collected from teachers,peers, and patients themselves. There were no dif-ferences between children with cancer and controlson any measures of friendship or popularity. Chil-dren treated for cancer rated themselves and weremore likely to be viewed by classmates as sensitiveand isolated. No differences between the groupswere found on teacher reports.

The findings from these studies are broadly inline with those where comparisons are madeagainst population norms: few measurable differ-ences on standardized measures between survivorsand controls are noted, but some discrepancies doexist. For example, based on interview data, survi-vors were more likely to have repeated a schoolgrade, less likely to be drinkers, and less likely tohave experienced black-outs following a drinkingepisode (Gray et al., 1992a, 1992b). Again from in-terview data, Madan-Swain et al. (1994) found thatsurvivors were more likely to report body image dis-turbance and adjustment difficulties, and Pendleyet al. (1997) reported that survivors participated infewer activities than controls.

3. Do Parents and Teachers Differ FromSurvivors in Their Views of the Impactof Cancer?

Parents and teachers report that survivors havemore problems than would be expected from popu-lation norms or in comparison with control groups(Radcliffe et al., 1996; Van Dongen-Melman, DeGroot, Kahlen, & Verhulst, 1995). Teacher reportshave been mixed, with most findings showing nodifferences between survivors and norms or con-trols (Noll et al., 1993; Radcliffe et al., 1996). How-ever, no studies compared parent and child ratingson the same measures.


indicate a clinical diagnosis of PTSD. These findingshave not been supported when comparisons aremade with control groups. Butler et al. (1996) foundlittle evidence of PTSD in a group of 42 survivorscompared with children on treatment. Both Kazaket al. (1997) and Barakat et al. (1997) reported thatthe incidence of PTSD among survivors was similarto that for healthy controls.

Moderators of PTSD. A number of variables havebeen reported to contribute to PTSD. These includefamily functioning and social support (Kazak et al.,1997). In the study described above, Barakat et al.(1997) also found that perceived life threat and so-cial and family resources moderated the incidenceof PSTD symptoms.

6. Are There Differences in PsychologicalOutcomes Dependent on the Nature of theSpecific Cancer Diagnosis?

The majority (n 14) of studies involved samplesincluding survivors of different cancers and did notattempt to distinguish between diagnostic groups interms of outcomes. Two studies focused specificallyon survivors of a CNS tumor (Radcliffe et al., 1996;Vannatta, Gartstein, Short, & Noll, 1998), and Eiseret al. (1997) included only those treated for a bonetumor. Three studies looked specifically at childrenwith leukemia (Kazak et al., 1997; Kupst et al., 1995;Stuber et al., 1996). Although outcomes likely willdepend on the specific cancer, the failure of anycomparative study means that it is not currentlypossible to make any definitive comments aboutthis issue.

Discussion

One might argue that these findings reflect the con-clusions of previous narrative reviews. However, therequirements for methodological rigor, which arean integral part of the systematic review process,mean that one can be more confident about thelimitations of previous studies and about the re-quirements for greater quality in future work.

Simple comparisons of survivors against popula-tion norms on standardized measures of anxiety,depression, and self-esteem point to few differences.These conclusions are also true where comparisonsare made against matched control groups. How-ever, there are suggestions that groups differ when

other methodologies are used; for example, inter-view data can highlight problems not included onquestionnaire measures (Gray et al., 1992a, 1992b).

Parents and teachers rate survivors as havingmore problems than controls or norms, but as yetno comparisons have been made between survivorsand parents or teachers on the same measure. Cer-tain family, clinical, and demographic variableshave been described as moderating outcomes, in-cluding parental distress (Sloper et al., 1994), socialclass and coping (Kupst et al., 1995), and incidenceof recurrence (Elkin et al., 1997). Finally, reports ofPTSD incidence have been mixed. When compari-sons have been made against norms, PTSD in survi-vors has been higher (Stuber et al., 1996), whereaswhen control groups are used, few differences arefound (Barakat et al., 1997; Butler et al., 1996; Kazaket al., 1997).

Evaluations of a systematic review should in-clude comments about the adequacy of the litera-ture and especially any methodology that mightreduce the validity of the conclusions. First, criticalclinical information such as time since completionof treatment can be poorly reported. Some authorsadopted a general requirement that patients shouldbe at least 5 years from diagnosis, although otherswere less specific. Thus, it is not clear how far recov-ery continues with increasing time from diagnosis.Second, studies routinely include survivors with dif-ferent cancers, some of whom may have experi-enced less aggressive treatments compared withothers. For example, it is inappropriate to includein the same study survivors of stage I Hodgkins dis-ease, who may have experienced relatively brieftreatments, with those treated more aggressively, asis the case for a bone tumor or CNS tumor. Inconsis-tent results may therefore be partly attributable todisease or treatment effects, but have received littleattention. Details of treatment protocols are rarelyprovided so that there is little evidence regardingthe relationship between psychological outcomeand treatment.

Third, problems in much of this work includethe self-selection of participants. Response rates canbe poor and can introduce bias into a study. Theextent to which survivors who take part in researchare representative of the total cohort needs to beconsidered. Inevitably this is a mobile populationand some are lost to follow-up. Methods of recruit-ment of the control groups are sometimes question-able. Pendley et al. (1997) recruited volunteers fromadvertisements; Gray et al. (1992a, 1992b), asked


Implications

Theoretical. Our review emphasizes how far researchcontinues to adopt a deficit-centered theoreticalperspective and to focus on maladjustment ratherthan coping. New theoretical frameworks are cru-cial. At the least, theories are needed to direct as-sumptions about the ways in which developmentmight be affected by surviving a life-threateningdisease and, consequently, the behaviors to be mea-sured. At the same time, it must be recognized thatthe goals of the individual child also include thoseof leading a normal life. For this reason, theoreti-cal frameworks that focus on the developmentaltasks to be achieved during adolescence (Havig-hurst, 1953) may also be important. Future workneeds to address how normative developmentaltasks can be attained with as little psychological dis-tress as possible.

Clinical. A priority of a systematic review is toprovide information relevant to hospital managersso as to influence priority setting and decisionsabout resources. Unfortunately, methodologicallimitations restrict the generalizability of findingsand their use in clinical settings. Survivors with pro-nounced difficulties are excluded from much re-search with the result that perceptions of the needfor services for survivors is likely to be underesti-mated. Survivors with obvious physical difficultiesare routinely excluded, and there has been verylittle follow-up of survivors of CNS tumors who aremore likely to need educational support.

A major issue for providers concerns the appro-priateness of follow-up care. Follow-up clinics arerecommended in order to gain information aboutthe toxicity of different treatments and directchanges in subsequent protocols. For survivorsthemselves, the requirement for follow-up mayseem to contradict their otherwise healthy statusand lack of awareness of the possibility of futurehealth complications. Thus, attendance at follow-up can be poor and confuses the issue of how it maybe affected by provisions of adult- or pediatric-basedservices. In either case, survivors are likely to begiven threatening information regarding their cur-rent and future health, and information about thebest way to do this would be useful. Again this liter-ature offers no guidance to those concerned withhow to give potentially threatening information insuch a way that survivors respond pragmaticallyrather than by denying the existence of problems.

Although the aim of a systematic review is to

patients to recommend friends. Both approachesmay introduce bias in the direction of recruiting in-dividuals who are more willing to help. Where thecontrol group is recruited from among survivorsfriends, fewer differences between survivors andhealthy peers may be expected, since friends arelikely to share common interests. Survivors withany obvious handicap or visible disfigurement areroutinely excluded from most published work (e.g.,Pendley et al., 1997), perhaps resulting in an under-estimation of psychological problems. At a minimallevel, researchers should attempt to describe anydifferences between those who agree to participateand those who refuse or cannot be traced, to ensurethat some idea about representativeness of survivorsis possible.

Fourth, issues of measurement are not satisfac-torily resolved. There are huge obstacles to choosingmeasures that are suitable across as wide age rangesas are customarily described. Many measures incommon use were developed to assess symptomsgenerally and are not cancer-specific. Some ques-tionnaires are potentially distressing or intrusive(Stuber et al., 1997). These methodological prob-lems, including use of postal surveys and poten-tially distressing questionnaires, may contribute topoor completion rates. The solution has to be morefocused and developmentally appropriate assess-ments. This can be achieved as the numbers of sur-vivors grows, and if centers are prepared tocollaborate. Pediatricians and child psychologistsneed to be prepared to draw on the expertise ofadult clinicians to select measures appropriate foryoung adult survivors.

Fifth, only Kupst and her colleagues have tracedoutcomes over a significant period of time (from di-agnosis for more than 10 years). Longitudinal workhas the important advantage over cross-sectionalstudies of charting changes in adjustment and func-tioning over time. However, using longitudinal datacan also have severe shortcomings. Comparisons ofgroup means on a variety of measures can give theimpression that adjustment does not change overtime, while inspection of individual scores (for ex-ample, individuals who move from scoring at a clin-ically elevated level to a nonclinical level on ameasure) can show a completely different pattern(cf. Thompson, Gustafson, George, & Spock, 1994).Researchers should attempt more longitudinal de-signs, while remaining alert to individual differ-ences within the data.


increase our ability to assess research in an objectivemanner, the method itself is inevitably associatedwith problems. As with any review, a systematic re-view can be no better than the empirical work onwhich it is based. The process of conducting a sys-tematic review in some contexts (e.g., randomizedcontrol trials; RCTs) has been well described, andguidelines are available about the most appropriatesearch strategies. Much less advice is available whenconducting a systematic review of literature not in-volving RCTs. Decisions need to be made in de-termining the criteria for inclusion of a study intothe review. No studies reported power or effect sizes,making judgments about appropriateness of samplesizes impossible. (Attempts at producing power cal-culations might reduce the confusion in the litera-ture but could not be conducted here due to lack ofthe relevant data.) In attempting to determine thevalidity of studies, critical information, such as timesince treatment, was often missing. Yet, clearly, ifstrict inclusion criteria were adopted, almost nostudy would be satisfactory.

A problem with defining strict inclusion criteriais that the potential exists for a single study (orgroup of studies) to be used in policy decision-making and resource allocation. By relaxing the cri-teria for inclusion in our review, there is the poten-tial for misuse of these data in policy development.However, in Table II we define methodological crite-ria for inclusion that should guide future studies.With this standard established, researchers can bebetter aware of the public policy implications anduses of their research, and improvements in re-ported studies can evolve more effectively and effi-ciently.

Although our conclusions do not differ substan-tially from those made in previous narrative re-views, the attempt to conduct a systematic reviewfocuses attention on the methodological issues in-volved in this work. At the least, this exerciseshould be used to establish a degree of quality con-trol in the area and alert researchers to the level ofsophistication that needs to be more routinely inte-grated in psychosocial work in pediatric oncology.This is especially pertinent considering the ex-tremely stretched resources in medical servicesacross most, if not all countries around the world.Research findings can potentially be used to guideallocations of resources and planning services, sothat it makes sense to increase the quality of ourwork in order to use these resources wisely. System-atic reviews offer a methodology that emphasizes

that only quality work should be used to guidethese resources. Thus, Table II offers a guide for fu-ture research. These standards need to be upheld byforthcoming researchers in order to set the tone andquality for future research in this area.

It is generally recognized that improvements insurvival in childhood cancer have been achieved atleast partly as a result of collaboration achievedthrough national and international clinical trials.This collaboration raises questions about cross-cultural implications of cancer. With respect to lateconsequences of surviving childhood cancer, the lit-erature is predominantly conducted in the UnitedStates (we identified only three studies publishedelsewhere). No studies were included in this reviewfrom anywhere outside the United States and Eu-rope. This may be partly due to a publication biasin American and European journals and also be-cause of our criterion that each study be publishedin English. However, considering the differences inadjustment to cancer across cultures, it raises issuesabout the functioning of childhood cancer survi-vors in other countries. For instance, with regard todesired medical communication or disclosure, thetraditional North American or West European re-sponse is one of full disclosure. However, the tradi-tional Asian response is one of patient protection(Gotay, 1996). Another example of cross-culturaldifferences lies in attitudes toward autonomy. InNorth American or West European cultures, individ-ual rights and individual independence are of greatimportance, whereas Asian cultures tend to stressinterconnectedness and interdependence. These is-


Table II. Suggested Methodological Criteria

Issue Criterion

Measurements Well-validated and reliable measures.

Respondents Both parent and child (if child is age-

appropriate and in good health).

Control/ Well-matched control group, or compared

comparison group with measurement norms (care must be made

that norms are culturally appropriate).

Information about Age; gender; ethnic composition; location of

sample sample (e.g., rural or urban sample); SES;

respondent rate; time since diagnosis (or time

since treatment ended); exact cancer

diagnosis. Additional information concerning

the parents age, marital status, and

employment status.

Results Appropriate rigorous statistical tests to be

used.

scope and range of remedial resources needed bysome survivors. As with clinical work, these answersare most likely to be achieved through collabora-tion, both between professionals and interna-tionally.

Acknowledgments

This research was supported by a grant from theCancer Research Campaign, (CP 1019/0101 & CP1019/0104). We thank Sarah Bennett for help inconducting the searches.

Received March 5, 1999; revisions received August 14,1999; accepted September 18, 1999

and procedures manualI. Baltimore, MD: Clinical Psy-chometric Research.

Derogatis, L. R. (1992). The Brief Symptom Inventory: Admin-istration, scoring, and procedures manualII. Baltimore,MD: Clinical Psychometric Research.

Dodge, K. A., McClaskey, C. L., & Feldman, E. (1985). Situ-ational approach to the assessment of social compe-tence in children. Journal of Consulting and ClinicalPsychology, 53, 344353.

Eiser, C. (1998). Practitioner review: Long term conse-quences of childhood cancer. Journal of Child Psychol-ogy and Psychiatry, 39, 621633.

Eiser, C., Cool, P., Grimer, R., Carter, S., Ellis, A., Kopel,S., & Eiser, J. R. (1997). The role of monitoring in de-termining quality of life following treatment for abone tumour. International Journal of Behavioural Medi-cine, 4, 398415.

Eiser, C., Cool, P., Grimer, R. J., Carter, S. R., Cotter, I. M.,Ellis, A. J., & Kopel, S. (1997). Quality of life followingtreatment for a malignant primary bone tumouraround the knee. Sarcoma, 1, 3945.

Eiser, C., & Havermans, T. (1994) Long term social adjust-ment after treatment for childhood cancer. Archives ofDisease in Childhood, 70, 6670.

Elkin, T. D., Phipps, S., Mulhern, R. K., & Fairclough, D.(1997). Psychological functioning of adolescent andyoung adult survivors of pediatric malignancy. Medicaland Pediatric Oncology, 29, 582588.

Gotay, C. G. (1996). Cultural variation in family adjust-ment to cancer. In L. Baider, C. L. Cooper, & A.Kaplan-De-Nour (Eds.), Cancer and the family. Chiches-ter, England: John Wiley.

Gray, R. E., Doan, B. D., Shermer, P., Vatter-Fitzgerald, A.,Berry, M. P., Jenkin, R. D. T., & Doherty, M. A. (1992a).

sues are also important when considering measure-ment. For example, with regard to the issue ofindependence, it is not surprising that North Ameri-can quality of life measures stress physical function-ing and a persons ability to act independently(Gotay, 1996). However, in a sample of Indian pa-tients, individual functioning was rated as the leastimportant aspect of their quality of life, with issuessuch as family happiness and spiritual satisfactionrating much higher (Chaturvedi, 1991). Futurework must consider the ethnic and cultural back-ground of survivors, in helping them through theillness and to a cure.

The question of psychological outcome follow-ing the experience of a life-threatening disease andits treatment remains to be determined, but goodstudies are important in order to determine the

References

Achenbach, T. M. (1991). Manual for the child behaviorchecklist/418 and 1991 profile. Burlington: Universityof Vermont, Department of Psychiatry.

Anholt, U. V., Fritz, G. K., & Keener, M. (1993). Self-concept in survivors of childhood and adolescent can-cer. Journal of Psychosocial Oncology, 11, 116.

Barakat, L. P., Kazak, A. E., Meadows, A. T., Casey, R.,Meeske, K. & Stuber, M. L. (1997). Families survivingchildhood cancer: A comparison of posttraumaticstress symptoms with families of healthy children.Journal of Pediatric Psychology, 22, 843859.

Boman, K., & Bodegard, G. (1995). Psychological long-term coping with experiences of disease and treatmentin childhood cancer survivors. Acta Paediatrica, 84,13951402.

Butler, R. W., Rizzi, L. P., & Handwerger, B. A. (1996). Briefreport: The assessment of Posttraumatic Stress Disor-der in pediatric cancer patients and survivors. Journalof Pediatric Psychology, 21, 499504.

Chang, P. (1991). Psychosocial needs of long-term child-hood cancer survivors: A review of literature. Pediatri-cian, 18, 2024.

Chaturvedi, S. (1991). Whats important for quality of lifeto Indiansin relation to cancer. Social Science andMedicine, 33, 9194.

Cook, D. J., Sackett, D. L., & Spitzer, W. O. (1995). Meth-odological guidelines for systematic reviews of ran-domized controlled trials in health care from thePotsdam consultation on meta-analysis. Journal ofClinical Epidemiology, 48, 167171.

Deasey-Spinetta, P., & Spinetta, J. J. (1980). The child withcancer in school: Teachers appraisal. American Journalof Pediatric Haematology and Oncology, 2, 8994.

Derogatis, L. R. (1977). The SCL-90-R administration, scoring


Psychologic adaptation of survivors of childhood can-cer. Cancer, 70, 27132721.

Gray, R. E., Doan, B. D., Shermer, P., Vatter-Fitzgerald, A.,Berry, M. P., Jenkin, D., & Doherty, M. A. (1992b). Sur-viving childhood cancer: A descriptive approach tounderstanding the impact of life-threatening illness.Psycho-Oncology, 1, 235246.

Green, D. M., Zevon, M. A., & Hall, B. (1990). Achieve-ment of life goals by adult survivors of modern treat-ment for childhood cancer. Cancer, 67, 206213.

Greenberg, H. S., & Meadows, A. T. (1991). Psychosocialimpact of cancer survival on school-age children andtheir parents. Journal of Psychosocial Oncology, 9, 4356.

Haupt, R., Byrne, J., Connelly, R. R., Mostow, E. N., Austin,D. F., Holmes, G. R., Holmes, F. F., Latourette, H. B.,Teta, J., Strong, L. C., Myers, M. H., & Mulvihill, J. J.(1992). Smoking habits in survivors of childhood andadolescent cancer. Medical and Pediatric Oncology, 20,301306.

Havighurst, R. J. (1953) Developmental tasks and education.New York, Longmans, Green.

Hawkins, M. M., & Stevens, M. C. (1996). The long termsurvivors. British Medical Bulletin, 52, 898923.

Institute for Scientific Information. (1997). ReferenceManager Version 8 for Windows. Research Informa-tion Systems, US.

Jenkinson, C., Coulter, A., & Wright, L. (1993). Short Form36 (SF 36) Health Survey Questionnaire: Normativedata for adults of working age. British Medical Journal,306, 14371440.

Kazak, A. E. (1994). Implications of survival: Pediatric on-cology patients and their families. In A. Bearison &R. Mulhern (Eds.) Pediatric psychooncology. New York:Oxford University Press.

Kazak, A. E., Barakat, L. P., Meeske, K., Christakis, D.,Meadows, A. T., Casey, R., Penati, B., & Stuber, M. L.(1997). Post traumatic stress, family functioning andsocial support in survivors of childhood leukemia andtheir mothers and fathers. Journal of Consultingand Clinical Psychology, 65, 120129.

Kazak, A. E., Christakis, D., Alderfer, M., & Coiro, M. J.(1994). Young adolescent cancer survivors and theirparents: Adjustment, learning problems, and gender.Journal of Family Psychology, 8, 7484.

Kupst, M. J., Natta, M. B., Richardson, C. C., Schulman,J. L., Lavigne, J. V., & Das, L. (1995). Family Copingwith pediatric leukemia: Ten years after treatment.Journal of Pediatric Psychology, 20, 601617.

Kupst, M. J., & Schulman, J. L. (1988). Long-term copingwith pediatric leukemia: A six year follow-up study.Journal of Pediatric Psychology, 13,722.

Madan-Swain, A., Brown, R. T., Sexson, S. B., Baldwin, K. etal. (1994). Adolescent cancer survivors: Psychosocialand familial adaptation. Psychosomatics, 35, 453459.

Mulhern, R. K. (1994). Neuropsychological late effects. In

A. Bearison & R. Mulhern (Eds.), Pediatric psychooncol-ogy. New York: Oxford University Press.

Noll, R. B., Bulowski, W. M., Davies, W. H., Koontz, K., &Kulkarni, R. (1993). Adjustment in the peer system ofadolescents with cancer: A two-year study. Journal ofPediatric Psychology, 18, 351364.

Olson, A. L., Boyle, W. E., Evans, M. W., & Zug, L. A.(1993). Overall function in rural childhood cancer sur-vivors: the role of social competence and emotionalhealth. Clinical Pediatrics, 32, 334342.

Oxman, A. D. (Ed.). (1994). Preparing and maintaining sys-tematic reviews: The Cochrane collaboration handbook.Oxford: Cochrane Collaboration.

Pendley, J. S., Dahlquist, L. M., & Dreyer, Z. A. (1997).Body image and psychosocial adjustment in adoles-cent cancer survivors. Journal of Pediatric Psychology,22, 2943.

Radcliffe, J., Benettt, D., Kazak, A. E., Foley, B., & Phillips,P. C. (1996). Adjustment in childhood brain tumorsurvival: Child, mother and teacher report. Journal ofPediatric Psychology, 21, 529539.

Sloper, P., Larcombe, I., & Charlton, A. (1994). Psychoso-cial adjustment of five-year survivors of childhoodcancer. Journal of Cancer Education, 9, 581588.

Stern, M., Norman, S. L., & Zevon, M. A. (1993). Adoles-cents with cancer: Self-image and perceived social sup-port as indexes of adaptation. Journal of AdolescentResearch, 8, 124142.

Stiller, C. A., Allen, M. B., & Eatock, E. M. (1995). Child-hood cancer in Britain: The National registry of child-hood tumours and incidence rates 19781987.European Journal of Cancer, 31A, 20282034.

Stuber, M., Christakis, D., Houskamp, B., & Kazak, A.(1996). Posttrauma symptoms in childhood leukemiasurvivors and their parents. Psychosomatics, 37,254261.

Thompson, R. J., Gustafson, K. E., George, L. K., & Spock,A. (1994). Change over a 12 month period in the psy-chological adjustment of children and adolescentswith cystic fibrosis. Journal of Pediatric Psychology, 19,189203.

Van Dongen-Melman, J. E., De Groot, A., Kahlen, K., &Verhulst, F. C. (1995). Psychosocial functioning ofchildren surviving cancer during middle childhood.In J. E. Van Dongen-Melman (Ed.), On surviving child-hood cancer. Alblasserdam, The Netherlands: HavekaBV.

Vannatta, K., Gartstein, M. A., Short, A., & Noll, R. B.(1998). A controlled study of peer relationships ofchildren surviving brain tumors: teacher, peer, andself ratings. Journal of Pediatric Psychology, 23, 279287.

Zeltzer, L. K. (1993). Cancer in adolescents and youngadultspsychosocial aspects: Long-term survivors.Cancer, 71, 34633468.


Cancer Teenagers Psy

Documents

Transcript of Cancer Teenagers Psy