Int. J. Cancer: Supplement 12, 52–57 (1999) r 1999 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer SOMATIZATION, ANXIETY AND DEPRESSION AS MEASURES OF HEALTH-RELATED QUALITY OF LIFE OF CHILDREN/ADOLESCENTS WITH CANCER Julia M. CHALLINOR1*, Christine A. MIASKOWSKI1, Linda S. FRANCK1, Robert E. SLAUGHTER1, Katherine K. MATTHAY1, Robin F. KRAMER1, Janet J. VEATCH1, Steven M. PAUL1, Michael D. AMYLON2 and Ida M. MOORE3 1University of California, San Francisco, San Francisco, CA, USA 2Lucile Salter Packard Children’s Hospital at Stanford, Stanford, CA, USA 3University of Arizona, Tucson, AZ, USA This descriptive study of health-related quality of life of children with cancer compared children/adolescents’, parents’ and teachers’ ratings for somatization, depression and anxiety to determine if there were significant correlations among respondent scores. In addition, the percentage of agreement among respondents and significant differences based on age, gender, use of cranial radiation and treatment status were measured. Forty-three children/adolescents with cancer, currently receiving therapy for at least 1 year or who had completed therapy for no more than 3 years (excluding children who had received bone marrow transplants or who had brain tumors), were recruited, with a parent and teacher, from 3 university medical centers. The Behavioral Assessment System for Children questionnaires for children/ adolescents, parents and teachers were used. Parents reported a higher level of depression for the children/adolescents with cancer than did the teachers or the children/adolescents themselves. Parents reported a higher level of anxiety for the children/adolescents than did the teachers. High positive correlations were found among scores from parents and teachers and among scores from parents and children/ adolescents for the anxiety and depression but not somatization subscales. Children/adolescents and teachers had high, positively correlated scores only for the depression subscale. High, positive correlations were found between somatization, anxiety and depression within each group of respondents. A significant percentage of agreement between all respondents on ratings for at-risk status was obtained only for the depression subscale. Age was the only variable found to have an influence on scores and only for the anxiety subscale. Int. J. Cancer Suppl. 12:52–57, 1999. r 1999 Wiley-Liss, Inc. Currently in the United States, 1 of every 475 children under the age of 15 years will be diagnosed with cancer (Bleyer, 1990). Treatment for cancer can last as long as 3 years for acute lymphoblastic leukemia (ALL) or be as short as a few months for a solid tumor. However, regardless of the specific cancer diagnosis, treatment may include surgery, chemotherapy, radiation therapy or a combination of these modalities. The health-related quality of life (HRQL) of children/ adolescents during cancer therapy as well as of the 65% of children who will survive the disease is a concern for health-care teams, families and the children/adolescents themselves. Measures of HRQL evaluate the impact of a disease or treatment on the patient (Bradlyn et al., 1996). For children/adolescents with a chronic illness, such as cancer, HRQL is an important criterion that can be used to make patient management decisions and to evaluate the quality of health-care services (Fitzpatrick, 1993). Children/adolescents undergoing cancer therapy experience marked changes in their lives (e.g., schedules, health status, psychological well-being and family roles). Children/adolescents who survive cancer often do not return unmarked to a pre-disease state. However, the severity of late effects related to cancer and its treatment varies according to the diagnosis and treatment regimen and other factors. Researchers are in agreement that the choice of an appropriate instrument to measure HRQL is a challenging task (Grant et al., 1990; Jalowiec, 1990; Mulhern and Bearison, 1994). Evaluations based on the responses of the children themselves can be enhanced by questioning parents and teachers, who compare a particular child/adolescent’s HRQL to that of other children/adolescents. Use of multiple respondents is particularly important when assessing the health and quality of life of children who are very young or of children who are sick regardless of age (Eiser and Jenney, 1996). Observation of children’s behavior has been a traditional method of evaluating concepts such as pain status, developmental status, motor skills, social skills and emotional status (Glascoe, 1995; Eiser and Jenney, 1996). An evaluation of children’s behavior does not directly describe traits or capabilities directly; rather, it allows the researcher to make inferences about the children being studied (Sattler, 1992). Assessments of children’s behavior while on cancer therapy or following treatment of the disease allow the researcher to make inferences about the children’s responses to the physical, mental and social impacts of the illness on their HRQL. The Behavioral Assessment System for Children (BASC) questionnaire is a 5-component assessment instrument designed to measure the emotional and behavioral status of children and adolescents ages 4 to 18 years (Reynolds and Kamphaus, 1992). The system includes separate questionnaires to be completed by the children’s teachers and parents and by the children/adolescents themselves. Items on the questionnaires include positive and negative behavioral observations that are rated by teachers and parents as well as positive and negative feelings and selfperceptions that are rated by the children/adolescents. Normative data for this instrument were based on the responses of 2,401 teachers, 3,483 parents and 9,861 children from 90 geographically diverse sites in the United States and Canada. Three subscales of the questionnaires have particular importance for children with a chronic illness such as cancer. These subscales measure somatization, anxiety and depression and are important indicators of the children/adolescents’ HRQL. These symptoms/problems occur in healthy children as well as illness groups. Although in many ways children with cancer are similar to their healthy peers, they are at increased risk for somatization, anxiety and depression. The results reported here are part of a larger study that examined the general behavioral performance of children/adolescents with cancer and those who had been treated recently for cancer using the BASC questionnaires. The purposes of this descriptive study of HRQL of children with cancer were (i) to compare somatization, depression and anxiety subscale scores of the BASC questionnaire among respondents (i.e., parents, teachers, children/adolescents); (ii) to determine if Grant sponsor: National Cancer Institute; Grant number: CA 78217; Grant sponsor: Oncology Nursing Society/Sigma Theta Tau. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute. *Correspondence to: Pediatrics Specialties Clinic, Box 0314 A 223, 400 Parnassus Avenue, San Francisco, CA 94143, USA. Fax: ⫹1–415–332–2582. SAD MEASURES OF HRQL there were significant correlations among respondent scores; (iii) to determine if there were significant differences in the percentage of agreement among respondents who are considered ‘‘at risk’’ as defined by the authors of the BASC (i.e., at least 1 SD above the mean of the normative data); and (iv) to determine if there were significant differences based on age, gender, use of cranial radiation and treatment status in the percentage of children/adolescents with cancer who were rated ‘‘at risk’’. MATERIAL AND METHODS Setting The study was based in 3 university medical centers: The University of California, San Francisco (UCSF), The Lucile Salter Packard Children’s Hospital at Stanford and the University of Arizona. The study was approved by the Human Subjects Committee at all 3 centers. Sample Children were eligible to participate in the study if they met the following criteria: (i) had a cancer diagnosis (excluding brain tumors as these children have special problems), (ii) were between the ages of 6 and 16 years inclusive (for the present purposes, adolescents were defined as the 12–14 year age group), (iii) were undergoing cancer therapy for at least 1 year or had completed therapy for no more than 3 years, (iv) were currently enrolled in school, (v) were English-speaking and (vi) were receiving cancer treatment involving radiation and/or chemotherapy but not bone marrow transplant (children who receive bone marrow transplants do not return to school for at least 1 year following transplant). One family refused to participate and a second was not approached due to a stressful situation in the family at the time of the study. Of the 45 children who were eligible to participate, only 1 was excluded a priori. The sample consisted of 24 children who were undergoing therapy and 19 children who had completed therapy (i.e., were off therapy for no more than 3 years) for a total sample of 43 children. A parent and a teacher of each child were recruited to participate in the study. Instrument The teacher, parent and child/adolescent self-report BASC questionnaires were used in this study. The parent and teacher forms include 3 age levels (4–5, 6–11 and 12–18 years of age). The child/adolescent self-report form is divided into 2 age groups (6–11 and 12–18 years of age). The parent and teacher forms take approximately 10 to 20 min to complete. The child/adolescent self-report form takes approximately 30 min to complete. For most scales on the BASC questionnaires, a higher score is indicative of pathology. The exception is the adaptation scale, where a lower score is indicative of pathology. Scoring of the BASC questionnaire For this study, general and gender-appropriate T scores were calculated. A T score tells the distance of a raw score from the normal group’s mean raw scale in standard deviations. The subscale scores for somatization, depression and anxiety from the teacher, parent and child/adolescent self-report forms are reported. At-risk status According to the manual, BASC questionnaire T-subscale scores for the problem subscales above 60 are considered ‘‘at-risk’’ scores (i.e., significant problems that may require treatment but do not indicate a formal diagnosis). Scores for somatization and depression above 70 and anxiety scores above 65 are considered clinically significant (i.e., maladaptive behavior). However, analysis for this study did not discriminate between ‘‘at-risk’’ and ‘‘clinically significant’’ scores. Therefore, results are reported for children with T scores above 60 whose status is referred to as ‘‘at-risk’’. Reliability and validity The 3 BASC forms have excellent reliability and validity (Hoza, 1994). Merenda (1996) reported that the coefficients of internal 53 consistency of the BASC questionnaire were excellent. The internal consistency scores of the teacher and parent forms average above 0.80 and the self-report forms average 0.80 according to the manual. Data analysis Repeated-measures ANOVA was used to evaluate for differences in somatization, anxiety and depression scores among parents, teachers and the children/adolescents themselves. Pearson product moment correlation coefficients were used to examine the correlations between teacher, parent and child/adolescent ratings of somatization, anxiety and depression. Percent agreement and Cohen’s were used to compare the percentage of children with cancer who were ‘‘at risk’’ according to somatization, depression and anxiety scores among the teachers’, parents’ and child/adolescents’ ratings. 2 analysis was performed to determine if age, gender, treatment status or cranial radiation status had an effect on the somatization, depression or anxiety subscale scores of the children/adolescents with cancer. A p value of ⬍0.05 was considered statistically significant. RESULTS Demographic data Table I provides a summary of the characteristics of the children/adolescents who participated in this study. The majority (60%) were female. Figure 1 includes a frequency of age and gender for the sample. Almost 3/4 were Caucasian. The majority of the children/adolescents were from a 2-parent household, and the commonest diagnosis by far was ALL. Most of the children/adolescents did not receive cranial radiation. Approximately half were undergoing therapy and the rest had completed treatment. More than two-thirds of the children/adolescents were from UCSF. All parent report forms were completed by mothers except for 1 father and 1 grandmother. Comparisons and correlations Comparison of teacher, parent and child/adolescent somatization, depression and anxiety subscale scores. A 1-way repeatedmeasures ANOVA was performed, and significant differences were TABLE I – DESCRIPTIVE CHARACTERISTICS OF THE SAMPLE (n ⫽ 43) Age: median (range) Gender (M:F) Ethnicity Caucasian African American Mixed Filipino Hispanic Asian Marital status of parents Married Single Unknown Cancer diagnosis Acute lymphoblastic leukemia Wilms’ tumor Acute non-lymphoblastic leukemia Rhabdomyosarcoma Lymphomatous leukemia Cranial radiation Yes No Therapy status On therapy Off therapy Sample by collection site UCSF1 University of Arizona LPCH at Stanford2 9 years (6–14 years) 17:26 32 (74.4%) 1 (2.3%) 2 (4.7%) 1 (2.3%) 6 (13.9%) 1 (2.3%) 26 (60%) 14 (33%) 3 (7%) 38 (89%) 1 (2%) 1 (2%) 1 (2%) 2 (5%) 9 (21%) 34 (79%) 24 (56%) 19 (44%) 29 (68%) 7 (16%) 7 (16%) 1University of California at San Francisco.–2Lucille Salter Packer Children’s Hospital. CHALLINOR ET AL. 54 TABLE III – PEARSON PRODUCT MOMENT CORRELATIONS AMONG THE TEACHER, PARENT AND CHILDREN/ADOLESCENT SUBSCALE SCORES Parent Child/adolescent (n ⫽ 13) Teacher r ⫽ 0.50 (p ⫽ 0.082) Parent — Depression (n ⫽ 32) Teacher r ⫽ 0.52 (p ⫽ 0.003) Parent — Anxiety (n ⫽ 30) Teacher r ⫽ 0.56 (p ⫽ 0.001) Parent — Somatization1 r ⫽ ⫺0.02 (p ⫽ 0.955) r ⫽ 0.02 (p ⫽ 0.959) r ⫽ 0.62 (p ⬍ 0.0005) r ⫽ 0.52 (p ⫽ 0.002) r ⫽ 0.23 (p ⫽ 0.227) r ⫽ 0.39 (p ⫽ 0.031) 1Only adolescents have a somatization subscale on the self-report form. FIGURE 1 – Frequency distribution of the age and gender of the sample (n ⫽ 43). found among the 3 respondent groups in the depression scores (Table II). Post hoc contrasts, using the Bonferroni criteria, revealed that parents reported significantly higher depression scores for children/adolescents with cancer compared with the teachers and the children/adolescents themselves. In addition, significant differences were found in the anxiety scores. Parents reported significantly higher anxiety for their children with cancer than did the teachers. A similar pattern was noted for the somatization scores (not included on the child self-report form), though the results were not statistically significant. Correlation among the respondents’ subscale scores Pearson product moment correlation coefficients were calculated by subscales (i.e., somatization, depression and anxiety) for the respondent groups. Table III lists the results. Teachers’ and parents’ scores were significantly positively correlated for anxiety and depression subscales, but there was no significant correlation for the somatization subscale. There was a significant positive correlation between parents’ and children/adolescents’ ratings for anxiety and depression subscales, but there was no significant correlation for the somatization subscale. Teachers’ and children/adolescents’ scores were significantly positively correlated for the depression subscale, but no significant correlation was found for the anxiety or somatization subscales. Correlation among the somatization, depression and anxiety subscale scores Pearson product moment correlation coefficients were calculated by respondent groups for the somatization, depression and anxiety subscale scores. Results are summarized in Table IV. Significant positive correlations were found for all of the subscale scores for each of the 3 groups (i.e., teachers, parents and children/ adolescents). At-risk analyses Analysis of percentage of agreement between teacher, parent and children/adolescent ratings of somatization, depression and anxiety subscale T scores that are considered at-risk scores Scores for the somatization, depression and anxiety subscales were dichotomized into ‘‘not at risk’’ and ‘‘at risk’’ categories. Subjects with scores ⬍60 are considered not at risk; those with scores ⱖ61 are TABLE IV – PEARSON PRODUCT MOMENT CORRELATIONS FOR SOMATIZATION, DEPRESSION AND ANXIETY SUBSCALE T SCORES FOR TEACHER REPORT, PARENT REPORT AND CHILD/ADOLESCENT SELF-REPORT Somatization Depression (n ⫽ 42) Anxiety r ⫽ 0.52 (p ⬍ 0.0005) Depression r ⫽ 0.35 (p ⫽ 0.021) Parent (n ⫽ 43) Anxiety r ⫽ 0.36 (p ⫽ 0.018) Depression r ⫽ 0.31 (p ⫽ 0.042) Children/adolescents2 (n ⫽ 30) Anxiety Teacher1 r ⫽ 0.53 (p ⬍ 0.0005) — r ⫽ 0.55 (p ⬍ 0.0005) — r ⫽ 0.63 (p ⬍ 0.0005) 1One respondent had missing data for this item.–2Children 6–7 years of age do not have a self-report form, only adolescents have a somatization subscale on the self-report form and 1 respondent had missing data for this item. considered at risk. Table V provides the percentages of children/ adolescents who are at risk based on the respondents’ scores. For each of the 3 subscales (i.e., somatization, depression and anxiety), the agreement between parents and children/adolescents, between teachers and parents and between teachers and children/adolescents was measured on the dichotomous risk variables. Results are included in Table VI. Somatization subscale As shown in Table VI, cross-tabulations were performed for teacher, parent and adolescent scores and no statistically significant agreement between the respondent groups was found regarding children with at-risk scores on the somatization subscale. Depression subscale As shown in Table VI, cross-tabulations demonstrated significant agreement between the teachers’ and the parents’ ratings of depression in children/adolescents with cancer. Also significant were the levels of agreement between the teachers and the children/adolescents and between the parents and the children/ adolescents on ratings of depression Anxiety subscales As shown in Table VI, no significant level of agreement was found between the respondent groups for children/adolescents with anxiety scores considered at risk. TABLE II – DIFFERENCES IN TEACHER, PARENT AND CHILD/ADOLESCENT SUBSCALE T SCORES FOR SOMATIZATION, DEPRESSION AND ANXIETY Subscales Teacher mean (SD) Parent mean (SD) Child/adol. mean (SD) Statistic F p value Somatization (n ⫽ 13)1 Depression (n ⫽ 32)2 Anxiety (n ⫽ 30)3 63.85 (20.94) 48.91 (7.81) 51.50 (9.37) 65.31 (12.39) 55.59 (13.84) 56.20 (11.49) 53.85 (13.78) 49.28 (10.86) 50.47 (11.25) 2.32 7.38 3.99 0.120 0.001 0.024 1The somatization subscale is not included in the child self-report form.–2Children 6–7 years of age do not have a self-report form.–3One respondent had missing data for this item. SAD MEASURES OF HRQL 55 TABLE V – PERCENTAGE OF CHILDREN/ADOLESCENTS WITH CANCER WHO WERE RATED BY TEACHERS, PARENTS OR THEMSELVES AS HAVING SOMATIZATION, DEPRESSION AND ANXIETY SUBSCALE SCORES THAT ARE CONSIDERED AT RISK Respondents Percentage of children/adolescents Somatization Depression Anxiety Teacher 37.2% 11.6% 11.6% 95% CI (0.23 ⬍ p ⬍ 0.52) 95% CI (0.02 ⬍ p ⬍ 0.21) 95% CI (0.02 ⬍ p ⬍ 0.21) (n ⫽ 43) (n ⫽ 43) (n ⫽ 42)1 Parents 46.5% 20.9% 27.9% 95% CI (0.32 ⬍ p ⬍ 0.61) 95% CI (0.09 ⬍ p ⬍ 0.33) 95% CI (0.14 ⬍ p ⬍ 0.41) (n ⫽ 43) (n ⫽ 43) (n ⫽ 43) Children/adolescents 30.8% 18.7% 32.3% 95% CI (0.06 ⬍ p ⬍ 0.56) 95% CI (0.05 ⬍ p ⬍ 0.32) 95% CI (0.16 ⬍ p ⬍ 0.49) 2 3 (n ⫽ 32) (n ⫽ 31)4 (n ⫽ 13) 1One respondent had missing data for this item.–2The somatization subscale is not included in the child self-report form.–3Children 6–7 years of age do not have a self-report form.–4One respondent had missing data for this item. TABLE VI – PERCENTAGE OF AGREEMENT BETWEEN TEACHER, PARENT AND CHILD/ADOLESCENT RATINGS OF SOMATIZATION, DEPRESSION AND ANXIETY SUBSCALE T SCORES FOR NORMAL OR AT-RISK STATUS Parent Child/adolescent Total agreement 63% 0.2423 p ⫽ 0.11 — Total agreement 53% 0.0488 p ⫽ 0.85 Total agreement 46% 0.0619 p ⫽ 0.76 Total agreement 81% 0.3281 p ⫽ 0.02 — Total agreement 90% 0.6191 p ⫽ 0.05 Total agreement 85% 0.5181 p ⬍ 0.005 Somatization1 Teacher Parent Depression Teacher Parent Anxiety Teacher Parent Total agreement 69% 0.0808 p ⫽ 0.55 — Total agreement 67% 0.1177 p ⫽ 0.45 Total agreement 68% 0.2619 p ⫽ 0.14 1Only adolescents have a somatization subscale on the self-report form. Association between age, gender, cranial radiation status, treatment status and subscale T scores 2 analyses were performed to determine if age, gender, treatment status or cranial radiation status had an influence on the somatization, depression and anxiety subscale scores of the children/ adolescents who had at-risk scores. Age was the only variable found to have a significant influence on the anxiety scores. Twice as many adolescents (n ⫽ 8/13) compared to children (n ⫽ 4/30) were scored by parents as having at-risk anxiety scores (Fisher’s exact test, 2-tailed ⬎ 0.0005). Analysis of the somatization subscale of the adolescent self-report form The somatization subscale (only on the adolescent self-report form) includes 11 items, 3 of which are of particular interest when evaluating children/adolescents with cancer: ‘‘I am a healthy person’’, ‘‘I am afraid I have cancer’’ and ‘‘Other people are healthier than I am’’. Adolescents receiving therapy for cancer might be expected to answer ‘‘true’’ to all 3 of these items since they are receiving treatment for their disease and, therefore logically, other people are healthier than they are. Seven of the 13 adolescents surveyed were on therapy for cancer. Yet all 7 believed they were ‘‘healthy’’ and only 2 were afraid they had cancer; 3 did not believe that others were healthier than they were. This analysis may explain why the adolescents’ somatization subscale scores showed no significant differences compared to the normative data for this subscale (t ⫽ 1.01, p ⫽ 0.334). The 6 adolescents who had completed therapy did not have the same responses to the 3 questions listed above. Only 3 of the survivors believed they were healthy and 5 of them thought that others were healthier than they were. Two were afraid they had cancer. It is interesting to note that, although these adolescents had completed their therapy, more adolescents in this cohort believed that others were healthier than was true of the adolescents who were continuing on therapy. It is also of interest that only half the adolescents who had completed therapy believed they were healthy in comparison to this self-assessment by all of the adolescents who were on active treatment. DISCUSSION This study evaluates the HRQL of children/adolescents with cancer using measures of somatization, depression and anxiety as reported by teachers, parents and the children/adolescents themselves. The generalizability of the study findings is enhanced because this is a diverse sample of children/adolescents undergoing therapy for cancer and early survivors that was obtained from 3 tertiary-care medical centers in 2 states. Our observations are in agreement with the work of Worchel et al. (1988), who found that parents of children with cancer reported higher levels of depression in their children than did the children themselves. Parents may experience personal stress related to the crisis of their child’s cancer diagnosis and treatment that could influence their assessments of children’s level of depression or anxiety. Brown et al. (1992) found evidence of anxiety and depressive symptoms in the fathers of children who had been on treatment for cancer for 1 year. One might expect that these symptoms could be projected to their children. However, Watson et al. (1999) found no significant impact of parents’ mental health on rating of their child’s HRQL in a study of 90 parents of children with cancer both on and off treatment. An alternative hypothesis to explain the differences between parents’ and teachers’ ratings of depression may be that children/adolescents who return home after a school day may be more depressed when the distraction of the classroom and schoolmates is absent, in which case the parents’ reports could be accurate. Significant differences were found in teachers’ and parents’ ratings of anxiety, with parents reporting higher levels of anxiety. It would be reasonable to expect that the parent of a child undergoing therapy for cancer would be in a highly anxious state and that parents whose children have recently completed therapy may experience anxiety related to the possibility of a relapse. Other researchers (e.g., Brown et al., 1992) found that a significant percentage of mothers of children who had been on therapy for cancer (34%) were classified as anxious by their own reports. 56 CHALLINOR ET AL. Again, parents’ anxiety ratings of their children may reflect their own heightened anxiety (Eiser and Jenney, 1996). However, it is also possible that children/adolescents are more anxious at home than at school, where they experience the diversion of the school environment and, therefore, the parents’ ratings may be accurate. The significant high, positive correlations for teachers’ and parents’ ratings for anxiety and depression denote that these behaviors may be more obvious than somatization, for which no significant correlation was found. Similar significantly positive correlations for parents’ and children/adolescents’ ratings for anxiety and depression indicate that these 2 behaviors, as observed by the parents, may be more illustrative of the children/adolescents’ self-reported state than somatization, for which no significant correlation was found. Worchel et al. (1988) did not find a significant correlation between parents’ scores for depression on the Child Behavior Checklist (CBCL) and children’s self-report on the Children’s Depression Inventory. However, these investigators did find that parents’ and nurses’ ratings of depression were significantly correlated, an observation similar to the parents’ and teachers’ correlations for depression scores in this study. It is of note that teachers’ and children/adolescents’ scores were significantly positively correlated only for the depression subscale. It may be that children/adolescents’ anxiety and somatization are not revealed by behavioral characteristics that are readily apparent to teachers. However, depressive behavior may be more obvious to even a busy teacher; hence, the significant positive correlation between teachers’ and children/adolescents ratings of depression. The highly positive correlations between the somatization, depression and anxiety subscales on teachers’ and parents’ reports indicate that children who exhibit any one of these symptoms should be evaluated for the other 2. Although 1 study found that children/adolescents with cancer did not have increased rates of depression compared to a control group (Noll et al., 1993), it is important to consider that those children/adolescents who have increased levels of somatization or anxiety may also be depressed. In fact, Tebbi et al. (1988), in a study of adolescents with cancer, found that somatic symptoms may be more conspicuous than the psychological symptoms of depression. The percentage of children/adolescents who had somatization scores in the at-risk range as reported by all of the respondents in this study is similar to findings described by Mulhern et al. (1989), who found statistically significant increases (17% to 33% compared to the 7% expected) in the proportions of children who were cancer survivors who had ‘‘clinically elevated’’ scores on the CBCL (i.e., 1.5 SD from the normative mean). In particular, these children had elevated scores for somatization by parental report. Our results contrast with those of previous researchers who found no differences in somatization scores (on the CBCL) from mothers whose children had cancer and mothers of the control group (Sawyer et al., 1995). In fact, in that study, the children with cancer who were 11 to 16 years old and their mothers reported anxiety/depression scores that were lower than the control group. Other researchers have found no difference in levels of anxiety between children with cancer compared with healthy controls using the Revised Children’s Manifest Anxiety Scale (Sloper et al., 1994). Lack of significant agreement between teachers, parents and children/adolescents regarding the somatization and anxiety scores, in contrast to scores for depression for the children/adolescents in this study, suggests that depressive behavior may be easier for individuals to observe. Children may exhibit less somatization at school than they do at home. Another explanation may be that the items on the BASC questionnaires for somatization and anxiety (which are not identical on all forms) may not be reliable measures of somatization or anxiety for children with a chronic illness such as cancer. Verhulst and van der Ende (1992) found that, in the general population, adolescents reported a greater number of problems (particularly anxiety/depression and somatization) than did their parents when using the CBCL. The researchers hypothesized that the increasing physical and psychological independence of adolescents may make parents’ reports of observable behavior and internal concerns of their adolescent children unreliable. Likewise, others have reported a low correlation between self-reports of adolescents with cancer and their parents’ reports of behavioral problems and the impact of the illness on school and social activities (Sawyer et al., 1999). In addition, other investigators have suggested that children/adolescents with cancer or other chronic illnesses may have adaptation styles that mask distress and, therefore, assessment by multiple respondents regarding behavior is important (Canning et al., 1992). Assessment of HRQL using indicators such as somatization, depression and anxiety has produced conflicting results. It is difficult to determine which study results reflect the reality for the general population of children/adolescents with cancer. However, clinicians must remain cognizant that it is the individual child/ adolescent being treated and it is his/her personal HRQL that is of paramount importance. Therefore, behavioral screening using multiple respondents to identify HRQL problems such as somatization, depression and anxiety is prudent. With respect to the impact of age, it is possible that parents, in general, perceive adolescents as more anxious than children, whether or not they have cancer. Also, it may be that adolescents display increased anxious behavior while enduring the stress of a major developmental transition. This finding is in contrast with that of other investigators who found that mothers of children with cancer reported significantly higher scores for anxiety/depression compared to a control group but no difference for adolescents compared to a control group (Sawyer et al., 1995). As with the findings in the present study, Mulhern et al. (1989) found no effects of cranial radiation or gender on the behavior problems of childhood cancer survivors using the CBCL, which includes scores for somatization and anxiety/depression. In contrast, Sanger et al. (1991), who studied children with cancer, found that boys had an increased rate of problems compared to girls on 2 or more scales of the Personality Inventory for Children. In addition, others have found that age had a significant effect on association with increased somatic complaints in cancer survivors, with adolescents scoring higher than children (Mulhern et al., 1989). It is possible that the small sample size of our study limited the ability to detect significant effects for age, gender, cranial radiation or treatment status on teachers’ or children/adolescents’ self-reports for the 3 subscales or of parental reports for somatization and depression. Furthermore, the small sample size limits the generalizability of the study findings, as does the distribution of diseases (with almost 90% of the sample having ALL). Although 3 sites contributed to the final sample, recruitment was difficult due to the need to collect data during the school year (which limited data collection from teachers), the language diversity of patient populations, relapses (which suspended school attendance) and the young age at diagnosis of many of the children who were not enrolled in school. Another limitation is that the BASC may not provide a comprehensive evaluation of HRQL of children/adolescents with cancer. Some researchers have raised concerns about using behavioral instruments such as the CBCL in populations of children with chronic illnesses and cite possible bias in regard to physical symptoms, limited sensitivity to the severity of behavioral problems and subjectivity of scoring by observers (Perrin et al., 1991). Nevertheless, behavioral questionnaires, such as the BASC, which include data from teachers’ and parents’ observations as well as from children/adolescents’ self-report of ‘‘feelings, attitudes and beliefs’’, can be useful to monitor HRQL of children with a chronic illness such as cancer. However, clinicians must remain cognizant that most standardized behavioral questionnaires for children were not developed for use with a chronically ill population and so must interpret results carefully. SAD MEASURES OF HRQL It appears that, in general, some children/adolescents in our study did not have significant HRQL problems as measured by the BASC questionnaires. With one exception, teachers, parents and the children/adolescents themselves reported subscale scores that were in agreement with normative data. However, the range of scores on the 3 subscales provides evidence of the high burden of morbidity in some children/adolescents. The variables that may determine which children/adolescents have at-risk scores remain to be investigated. Until such time as this information is available for proactive intervention, it is important to identify, as soon as possible, the individual children/adolescents undergoing therapy for cancer and early survivors who are having extreme HRQL problems to initiate appropriate interventions. The HRQL of children/adolescents undergoing therapy for cancer and early survivors is an important component of therapeutic success. A longitudinal measurement approach would allow the 57 health-care team to determine whether findings early in the course of treatment are consistent over time or if predictable changes occur over time that require intervention. The value of this approach would be strengthened by the use of an instrument with an abundance of population-based normative data such as the BASC. A study that examined parents’ perceptions of their own levels of somatization, depression and anxiety and compared the results with children/adolescents’ self-reports would be helpful. This might determine if the parents’ ratings for their children are affected by their own status. It is certainly reassuring that the HRQL of the majority of the children/adolescents evaluated by the BASC questionnaire is within the range of the normative data. However, those children/ adolescents who scored in the at-risk range on teacher, parent and self-report forms for somatization, depression and anxiety deserve the attention of the health team, family members and school personnel to take measures to mitigate their troubled experience. REFERENCES BLEYER, W.A., The impact of childhood cancer on the United States and the world. CA Cancer J. Clin., 40, 355–367 (1990). BRADLYN, A.S., RITCHEY, A.K., HARRIS, C.V., MOORE, I.M., O’BRIEN, R.T., PARSONS, S.K., PATTERSON, K. and POLLOCK, B.H., Quality of life research in pediatric oncology: research methods and barriers. Cancer, 78, 1332– 1339 (1996). BROWN, R.T., KASLOW, N.J., HAZZARD, A.P., MADAN-SWAIN, A., SEXSON, S.B., LAMBERT, R. and BALDWIN, S., Psychiatric and family functioning in children with leukemia and their parents. J. Amer. Acad. Child adolesc. Psych., 31, 495–502 (1992). CANNING, E.H., CANNING, R.D. and BOYCE, T., Depressive symptoms and adaptive style in children with cancer. J. Amer. Acad. Child adolesc. Psych., 31, 1120–1124 (1992). EISER, C. and JENNEY, M.E.M., Measuring symptomatic benefit and quality of life in pediatric oncology. Brit. J. Cancer, 73, 1313–1316 (1996). FITZPATRICK, R., The measurement of health status and quality of life in rheumatological disorders. Baillière’s clin. Rheumatol., 7, 297–317 (1993). GLASCOE, F.P., Developmental screening. In: S. Parker and B. Zuckerman (eds.), Behavioral and developmental pediatrics, pp. 25–29, Little, Brown, Boston (1995). GRANT, M., PADILLA, G.V., FERRELL, B.R. and RHINER, M., Assessment of quality of life with a single instrument. Semin. Oncol. Nurs., 6, 260–270 (1990). HOZA B., Reviews of the behavior assessment system for children. Child. Assess. News, 4,5, 8–10 (1994) JALOWIEC, A., Issues in using multiple measures of quality of life. Semin. Oncol. Nurs., 6, 271–277 (1990). MERENDA, P.F., BASC, behavior assessment system for children. Meas. Eval. Counsel Develop., 28, 229–232 (1996). MULHERN, R.K. and BEARISON, D.J., Future directions in pediatric psychooncology. In: R.K. Mulhern and D.J. Bearison (eds.), Pediatric psychooncology: psychological perspectives on children with cancer, pp. 215–222, Oxford University Press, New York (1994). MULHERN, R.K., WASSERMAN, A.L., FRIEDMAN, A.G. and FAIRCLOUGH, D., Social competence and behavioral adjustment of children who are longterm survivors of cancer. Pediatrics, 83, 18–25 (1989). NOLL, R.B., BUKOWSKI, W.M., DAVIES, W.M., KOONTZ, K. and KULKARNI, R., Adjustment in the peer system of adolescents with cancer: a two-year study. J. pediatr. Psychol., 18, 351–364 (1993). PERRIN, E.C., STEIN, R.E.K. and DROTAR, D., Cautions using the Child Behavior Checklist: observations based on research about children with a chronic illness. J. pediatr. Psychol., 16, 411–421 (1991). REYNOLDS, C.R. and KAMPHAUS, R.W., BASC: Behavior Assessment System for Children manual, American Guidance Service, Circle Pines, MN (1992). SANGER, M.S., COPELAND, D.R. and DAVIDSON, E.R., Psychosocial adjustment among pediatric cancer patients: a multidimensional assessment. J. pediatr. Psychol., 15, 463–474 (1991). SATTLER, J.M., Assessment of children (3rd ed.), Sattler, San Diego (1992). SAWYER, M., TOOGOOD, I. and RICE, M., A comparison of parent and adolescent reports describing the health-related quality of life of adolescents treated for cancer. Int. J. Cancer, (Suppl. 12), 39–45 (1999). SAWYER, M.G., ANTONIOU, G., NGUYEN, A.-M.T., TOOGOOD, I., RICE, M. and BAGHURST, P.A., Prospective study of the psychological adjustment of children with cancer. Amer. J. pediatr. Hematol. Oncol., 17, 39–45 (1995). SLOPER, T., LARCOMBE, I.J. and CHARLTON, A., Psychosocial adjustment of five-year survivors of childhood cancer. J. Cancer Educ., 9, 163–169 (1994). TEBBI, C.K., BROMBERG, C. and MALLON, J.C., Self-reported depression in adolescent cancer patients. Amer. J. pediatr. Hematol. Oncol., 10, 185–190 (1988). VERHULST, F.C. and VAN DER ENDE, J., Agreement between parents’ reports and adolescents’ self-reports of problem behavior. J. Child Psychol. Psych., 33, 1011–1023 (1992). WATSON, M., EDWARDS, L., VON ESSEN, L., DAVIDSON, J., DAY, R. and PINKERTON, R., Development of the Royal Marsden Hospital Paediatric Oncology Quality of Life Questionnaire. Int. J. Cancer, (Suppl. 12), 65–70 (1999). WORCHEL, F.F., Nolan, B.F., Willson, V.L., Purser, J.S., Copeland, D.R. and Pfefferbaum, B. Assessment of depression in children with cancer. J. pediatr. Psychol., 13, 101–112 (1988).