• late effects;
  • psychosocial;
  • scholastic achievement;
  • radiation;
  • special education;
  • education attainment;
  • childhood cancer


  1. Top of page
  2. Abstract
  6. Acknowledgements


The objective of the current report was to compare the self-reported rates of special education (SE) and educational attainment among specific groups of childhood cancer survivors and a random sample of sibling controls.


The Childhood Cancer Survivor Study is a retrospective cohort of individuals who were diagnosed with a cancer in childhood and survived at least 5 years postdiagnosis. This analysis includes 12,430 survivors and 3410 full siblings. Reported use of SE services and educational attainment were analyzed within subgroups defined by type of cancer, age at diagnosis, and type of treatment.


The use of SE services was reported in 23% of survivors and 8% of siblings, with the greatest differences observed among survivors who were diagnosed before age 6 years, most notably survivors of central nervous system (CNS) tumors (odds ratio [OR], 18.8; 95% confidence interval [95%CI], 15.01–23.49), leukemia (OR, 4.4; 95%CI, 3.75–5.16), and Hodgkin disease (OR, 4.4; 95%CI, 2.64–7.24). It was found that intrathecal methotrexate (IT MTX) and cranial radiation (CRT), administered alone or in combination, significantly increased the likelihood that a survivor would use SE (IT MTX only: OR, 1.3; 95%CI, 1.09–1.78; CRT only: OR, 7.2; 95%CI, 6.14–8.39; IT MTX and CRT combined: OR, 2.6; 95%CI, 2.30–2.95). A positive dose response was identified between higher doses of CRT and use of SE. It was determined that survivors of leukemia (OR, 1.6; 95%CI, 1.23–2.16), CNS tumors (OR, 2.7; 95%CI, 1.92–3.81), non-Hodgkin lymphoma (OR, 1.8; 95%CI, 1.15–2.78), and neuroblastoma (OR, 1.7; 95%CI, 1.14–2.61) were significantly less likely to finish high school compared with siblings; however, when survivors received SE services, risk estimates approximated those of the sibling SE population.


Children who are diagnosed with cancer should be followed closely during and after treatment to identify early signs of learning disabilities and to maximize intervention strategies for the successful completion of scholastic goals. Cancer 2003;97:1115–26. © 2003 American Cancer Society.

DOI 10.1002/cncr.11117

With the introduction of multimodal therapy in the early 1970s, 5-year survival rates for most childhood cancers improved dramatically. Data from the Surveillance, Epidemiology, and End Results (SEER) Program show that 5-year survival rates for all childhood cancers increased from 56% in 1974 to 75% by 1995. The current 5-year survival rates are now as high as 81% for patients with acute lymphoblastic leukemia (ALL) and 93% for patients with Hodgkin disease and Wilms tumor.1 The need to identify the degree to which late effects of therapy may impact quality of life becomes critical as increasing numbers of children become long-term survivors. The Childhood Cancer Survivor Study (CCSS) was created to serve as a resource for the investigation of a broad spectrum of outcomes among long-term survivors of childhood cancer, such as the risk of second cancers, cardiovascular events, fertility, mortality, and psychosocial factors, including health-related behaviors and educational achievement.

Educational achievement and diminished cognitive functioning are two outcomes that have been reported extensively in the literature on survivors of childhood cancer. To varying degrees, deficits have been noted most often among children diagnosed with ALL and central nervous system (CNS) tumors who received intrathecal methotrexate (IT MTX) and/or cranial radiation (CRT).2–6 The CCSS provides a unique opportunity to investigate these treatment effects by making it possible to explore their potential impact on survivors' long-term educational achievement across all major childhood cancer diagnoses. The purpose of this analysis was twofold: 1) to describe the self-reported utilization and factors associated with the use of special education (SE) services, and 2) to determine the level of education attained among a large population of long-term survivors of childhood cancer.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The Long-Term Follow-Up Study is a retrospective cohort comprised of individuals diagnosed with cancer, leukemia, tumors or other serious illnesses of childhood. Those individuals with a confirmed diagnosis of cancer comprise the CCSS. Details of the study design and CCSS cohort characteristics have been published elsewhere.7 Briefly, study patients were identified from the 25 CCSS institutions from across the United States and Canada. Eligibility criteria for the study included a newly diagnosed cancer (leukemia, CNS tumor, Hodgkin disease, non-Hodgkin lymphoma, kidney, neuroblastoma, soft tissue sarcoma, or bone tumor) between January 1, 1970 and December 31, 1986; survival of 5 years from the date of diagnosis; age < 21 years at the time of diagnosis; English-speaking or Spanish-speaking; and living in the United States or Canada at the time of diagnosis.

The CCSS protocol and contact documents were reviewed and approved by the Human Subjects Committee at each participating institution. Baseline questionnaires were collected from September, 1994 through November, 2000. Age specific questionnaires were developed for survivors age < 18 years and survivors age ≥ 18 years at the time of contact. Survivors age ≥ 18 years were sent a baseline questionnaire and consent form directly and were asked to complete and return them. For survivors age < 18 years, the questionnaire and consent form were sent to the parent(s) to complete. The next of kin of survivors who subsequently died at least 5 years postdiagnosis were asked to complete a questionnaire for that survivor. After obtaining consent, survivors' medical records, including radiotherapy records, were abstracted by trained data managers at each institution according to a standardized protocol to obtain information such as type, dates, and schedule of treatment. Copies of the baseline questionnaire and the medical record abstraction form used in data collection are available for review online at

Among the 20,276 eligible 5-year survivors, 14,054 survivors completed a questionaire at the time of this analysis, 2996 survivors were classified as lost to follow-up, 3132 survivors did not participate, and 94 survivors were pending inclusion into the cohort. At the time of this analysis 12,431 of the 14,054 participants have signed a medical release allowing for the abstraction of medical records.

Comparisons made to determine whether demographic and/or treatment-related characteristics differed between participants and refusals showed that, in general, participants and refusals were very similar with regard to gender, cancer diagnosis, age at diagnosis, age at contact, and type of cancer treatment. The rate of refusal was significantly higher (P = 0.001) among the next of kin (34%) of patients who had survived at least 5 years from diagnosis but died subsequently compared with patients who currently were living (21%). Comparisons between survivors who completed the baseline questionnaire and survivors who were classified as lost to follow-up did not reveal any statistically significant differences other than vital status.7

During the collection of baseline data, information was obtained from the cancer survivors on their siblings. To establish a sibling cohort, first, a random sample of 50% of participating survivors was selected. Then, if the survivor who was selected through randomization reported a living full brother or sister, the sibling closest in age to the survivor was invited to participate in the study. If the sibling agreed, they or their parents completed an age specific baseline questionnaire similar to that completed by the survivors. The target sample for the sibling cohort is 5000 participants. Recruitment for the sibling cohort was initiated in July, 1996 and is still in progress. At the time of the current analysis, 5800 siblings had been contacted to participate; 3528 siblings completed a questionnaire, 472 siblings chose not to participate, and 1800 siblings were pending. Of the 3528 sibling questionnaires received, 110 questionnaires are being reviewed by the coordinating center for possible content and logic discrepancies and were not included in the current analysis.

Data Analysis

Items from the baseline questionnaire relating to education and school history included the highest grade or level of schooling completed and the type of high school diploma received: standard versus general educational development credential (GED). A GED is a high school equivalency test that is an alternative way to complete high school. Also included was whether the participant received learning-disabled or SE services in elementary, junior, or high school. Possible responses were yes, no, or not sure. For participants who reported receiving SE services, additional information was obtained, including reasons for SE and specific grade levels at which the services were rendered.

When evaluating possible associations between SE placement and exposure to CRT, we examined CRT both as a dichotomous variable and by dose range. CRT doses were coded as the maximum dose to any part of the brain. Assigned CRT dose incorporated all treatments, including those given for recurrences and new malignancies, as long as at least one treatment occurred prior to initiation of SE services. Tumor location was not available, and surgical intervention was not considered in this analysis for survivors of CNS tumors. Survivors who initiated SE services prior to their primary cancer diagnosis were not included in analysis of treatment effects.

Because the age at each grade level was not provided by respondents, the expected age at grade level was calculated, accounting for birthdays during the school year and assuming that participants completed each grade without being held back or skipping grades. Thus, it was considered that age in preschool was < 6 years, age in kindergarten was 6 years, age in first grade was 7 years, etc. Both survivors and siblings age < 6 years at the time of questionnaire completion (not of school age) were excluded from this analysis (one survivor and eight siblings).

Odds ratios (ORs) and 95% confidence intervals (95%CIs) were calculated using Cochran–Mantel–Haenszel statistics for disease specific and treatment specific analyses. Because the basic study design of the CCSS is such that not every survivor has a sibling control, generalized estimating equation models were used to calculate estimates of ORs.8 This method performs logistic regression analyses, controlling for the nonindependence of members of the same family (survivors and their siblings).


  1. Top of page
  2. Abstract
  6. Acknowledgements

The current report includes 12,430 survivors who were age ≥ 6 years at the time of questionnaire completion for whom both baseline questionnaires and medical record abstraction forms were complete at the time of analysis. Sibling data were available for 3410 individuals age ≥ 6 years. Table 1 shows characteristics of the study participants. Compared with cancer survivors, siblings were more likely to be female and were somewhat older. Survivors ranged in age from 6 years to 47 years, with a median age of 23 years, whereas siblings were between the ages of 6 years and 56 years, with a median age of 26 years. The three major cancer diagnoses were leukemia, Hodgkin disease, and CNS tumors, accounting for 60.5% of the survivor population. Twenty-three percent of the survivors indicated they were in SE at least some time during their K–12 schooling, compared with 8% of the sibling cohort (P < 0.01).

Table 1. Demographics of the Current Analysis
CharacteristicNo. (%)
  1. CNS: central nervous system.

 Alive11425 (91.9)3410 (100.0)
 Dead1005 (8.1)0
 Male6543 (52.6)1642 (48.2)
 Female5887 (47.4)1768 (51.8)
 White, non-Hispanic10969 (88.2)3087 (90.5)
 Black, non-Hispanic478 (3.9)84 (2.5)
 Hispanic607 (4.9)133 (3.9)
 Other346 (2.8)97 (2.8)
 Not specified30 (0.2)9 (0.2)
Age at questionnaire completion (yrs)  
 6–9154 (1.2)68 (2.0)
 10–141734 (14.0)337 (9.9)
 15–192739 (22.0)587 (17.2)
 20–242622 (21.1)589 (17.3)
 25–292378 (19.1)633 (18.6)
 30–341693 (13.6)563 (16.5)
 35–39813 (6.5)378 (11.1)
 40–44276 (2.2)19.5 (5.7)
 45–5921 (0.2)60 (1.7)
 Leukemia4213 (33.9)
 Hodgkin's disease1680 (13.5)
 CNS tumor1637 (13.2)
 Soft tissue sarcoma1075 (8.6)
 Kidney1063 (8.6)
 Bone cancer1033 (8.3)
 Non-Hodgkin lymphoma908 (7.3)
 Neuroblastoma821 (6.6)
Age at diagnosis (yrs)  
 0–55707 (45.9)
 6–102425 (19.5)
 11–152618 (21.1)
 16–201680 (13.5)
Special education  
 Yes2866 (23.1)282 (8.3)
 No8150 (65.6)2747 (80.6)
 Not sure81 (0.7)24 (0.7)
 Ambiguous/missing data1333 (10.7)357 (10.5)

SE Utilization

Examination of the use of SE services by diagnosis type within age-at-diagnosis strata are presented in Table 2. These findings demonstrate, that after adjusting for age at questionnaire completion, younger age at diagnosis is associated significantly with a higher need for SE across all cancer types. Recognizing that males typically receive SE services more often than females, we also stratified utilization of SE by gender while still controlling for age at questionnaire completion. This analysis shows that, whereas both male and female survivors are more likely to be in SE compared with same-gender siblings, the differences were greater among females. For example, male survivors of CNS tumors who were diagnosed between the ages of 0 years and 5 years were 13.3 times more likely to be in SE compared with male siblings (95%CI, 9.91–17.76), whereas female survivors of CNS tumors who were diagnosed in the same period were 30.5 times more likely to be in SE compared with female siblings (95%CI, 21.40–43.50). Additional analyses were performed restricting the use of SE to the period after completion of therapy. Comparison of siblings with diagnosis specific subgroups of survivors who reported using SE after they completed therapy provided ORs comparable to those obtained when including both the on-therapy and off-therapy periods (data not shown). Analysis of SE utilization by treatment era revealed that the rates of SE differed significantly (P = 0.001) by era, with overall rates of 17%, 22%, and 25% for patients who were diagnosed during 1970–1975, 1976–1980, and 1981–1986, respectively.

Table 2. Special Education Services by Age Diagnosis
DiagnosisAge 0–5 yrsAge 6–10 yrsAge 11–15 yrsAge 16–20 yrs
Yes (%)aORb95%CIYes (%)aORb95%CIYes (%)aORb95%CIYes (%)aORb95%CI
  • Yes: received special education services; OR: odds ratio; 95% CI: 95% confidence interval.

  • a

    Participants who answered “don't know” or did not respond to the question are not represented in this table.

  • b

    ORs are adjusted by age at questionnaire completion

 Leukemia853 (35.9)4.43.75–5.1619.6 (26.2)3.32.66–4.0065(13.3)1.71.26–2.2416(9.5)1.40.80–2.33
 Central nervous system tumor437 (70.3)18.815.01–23.49235 (55.8)11.99.48–14.98109 (32.3)5.34.08–6.9227(23.9)4.12.55–6.55
 Hodgkin disease25(33.8)4.42.64–7.2433(14.7)1.81.22–2.6180(13.6)1.91.42–2.4534(5.7)0.90.61–1.31
 Non-Hodgkin lymphoma46(26.3)2.92.01–4.1452(21.8)2.61.84–3.5737(14.3)1.91.32–2.788(6.6)1.00.46–2.11
 Kidney148 (18.6)1.81.44–2.2415(12.8)1.20.70–2.2100
 Neuroblastoma169 (24.8)2.52.02–3.138(23.5)2.51.15–5.3100
 Soft tissue sarcoma95(23.7)2.61.96–3.3333(19.6)2.31.52–3.4020(9.0)1.10.70–1.8416(5.6)1.40.80–2.32
 Bone malignancy10(20.0)2.11.01–4.1632(17.0)1.91.28–2.8741(10.6)1.30.94–1.9123(8.1)1.20.78–1.94
 Leukemia428 (34.5)2.92.37–3.58112 (28.6)2.72.03–3.4831(12.7)1.10.76–1.7113(12.9)1.30.68–2.35
 Central nervous system tumor233 (70.4)13.39.91–17.76140 (60.6)10.67.79–14.3969(35.8)4.43.14–6.1919(29.2)3.62.03–6.38
 Hodgkin disease22(38.6)3.82.13–6.6530(19.0)1.71.13–2.6649(16.7)1.71.17–2.3626(9.2)1.00.62–1.53
 Non-Hodgkin lymphoma29(24.2)1.91.21–2.9444(24.4)2.21.51–3.2031(17.4)1.71.14–2.667(8.2)0.80.38–1.88
 Soft tissue sarcoma60(26.7)2.21.55–3.0120(20.2)1.81.06–2.9811(10.6)1.00.51–1.888(9.0)0.90.44–2.01
 Bone malignancy4(19.1)1.40.45–4.2516(16.8)1.40.83–2.5117(9.2)0.80.48–1.3918(11.1)1.20.68–1.98
 Siblings184 (12.3)1.0184 (12.3)1.0184 (12.3)1.0184 (12.3)1.0
 Leukemia425 (37.4)7.65.78–9.9184(23.6)4.43.17–5.9934(13.9)2.71.77–4.133(4.4)1.10.37–3.56
 Central nervous system tumor204 (70.1)30.521.40–43.5095(50.0)13.79.51–19.6640(27.8)6.74.36–10.318(16.7)4.42.00–9.63
 Hodgkin disease3(17.7)2.90.83–10.473(4.55)0.80.24–2.3631(10.5)2.31.44–3.518(2.6)0.70.31–1.49
 Non-Hodgkin lymphoma17(30.9)5.52.94–10.308(13.6)2.20.94–4.926(7.4)1.40.60–3.351(2.7)0.40.01–14.69
 Soft tissue sarcoma35(19.9)3.22.07–4.9613(18.8)2.91.49–5.809(7.6)1.50.73–3.058(8.2)2.00.94–4.19
 Bone malignancy6(20.7)3.31.29–8.4516(17.2)3.01.73–5.3424(11.9)2.41.48–3.905(4.1)1.00.42–2.30

Categories of treatment type were assigned to examine the possible effect of CRT alone or combined with IT MTX on the utilization of SE services (Fig. 1A), which shows that SE use was associated significantly with all three treatment categories: IT MTX only (OR, 1.3; 95%CI, 1.09–1.78), CRT only (OR, 7.2; 95%CI, 6.14–8.39), and CRT plus IT MTX (OR, 2.6; 95%CI, 2.30–2.95). Although both males and females showed increased risk for utilizing SE, the risk estimates for females were markedly higher compared with males in the groups that received CRT alone or in combination with IT MTX. Because the majority of survivors who received both IT MTX and CRT were leukemia survivors, a logistic regression analysis using leukemia survivors only was performed to identify a possible interaction between the two treatment modalities. Dichotomous variables for IT MTX and CRT were measured against the combined effect of the two treatments. After adjusting for age at diagnosis and age at questionnaire completion, our data supported the appearance of a multiplicative effect between IT MTX and CRT. The risk of needing SE associated with the combined effect of CRT and IT MTX was 2.7 times the risk relative to what would be expected if the two treatments were independent of each other (OR, 2.7; 95%CI 1.50–4.76; data not shown).

thumbnail image

Figure 1. Estimated odds ratios (OR) and 95% confidence intervals for receiving special education according to groups defined by type of treatment (A) and dose of cranial radiation (B) in centigrays (cGy). All analyses were adjusted for age at the time of diagnosis and at the time the questionnaire was completed. Note that an odds ratio of 1.0 indicates no difference in the standard error between the study group and the referent group. When a confidence interval spanned the value of 1.0, we were unable to conclude that our findings were statistically significant. IT MTX: intrathecal methotrexate; CRT: cranial radiation.

Download figure to PowerPoint

ORs for the use of SE also were calculated according to CRT dose (Fig. 1B). The risk of using SE services increased significantly as the level of CRT increased. It was found that dose ranges that included both 1800 centigrays (cGy) and 2400 cGy increased a survivor's risk for entering SE, although the risk was somewhat attenuated for the lower dose. Females accounted for a significantly larger portion of survivors who received CRT and used SE, with a clear dose response to increasing levels of CRT.

Reasons for Entry into SE

A unique feature of this cohort is that participants were asked to give the underlying reasons for being placed in SE programs. A mark all that apply question was given with possible responses of no, yes, or not sure for the following reasons: missing school, low scores on tests, problems with learning or concentrating, and emotional or behavioral problems. Data for reasons given by age at diagnosis and diagnosis type are provided in Table 3. Survivors who were diagnosed between the ages of 0 and 15 years reported needing SE services due to missed school. Low test scores also were associated significantly with survivors who were between the ages of 0 and 10 years at the time of diagnosis. Compared with the sibling cohort, survivors of all cancer diagnoses used SE services at an increased rate due to missing school, with survivors of bone cancer (OR, 7.4; 95%CI, 3.95–13.92) and soft tissue sarcoma (OR, 6.2; 95%CI, 3.34–11.33) showing the strongest associations. Survivors of leukemia, CNS tumor, non-Hodgkin lymphoma, kidney, and neuroblastoma showed significant increased use of SE due to low test scores. Ideally, it would have been preferable to show diagnosis specific and gender specific reasons for entering SE within age groups stratified by diagnosis; however, unfortunately, even with a population this large, small cell sizes became problematic.

Table 3. Reasons for Special Education Servicesa
CharacteristicMissed schoolLow testsLearning/concentratingEmotional/behavioral
Yes (%)bORc95%CIYes (%)bORc95%CIYes (%)bORc95%CIYes (%)bORc95%CI
  • OR: odds ratio; 95%CI: 95% confidence interval.

  • a

    Results were based on 2,866 participants who indicated they were in special education.

  • b

    The Yes (%) column indicates those who answered “yes” within each strata. Participants who answered “don't know” or who did not indicate specific reasons are not represented in this table.

  • c

    ORs are adjusted by age at questionnaire completion.

Age at diagnosis (yrs)            
 0–5306 (18.3)2.21.20–4.05908 (54.8)2.41.48–3.811461 (86.8)1.50.89–2.53219 (13.4)1.50.69–3.27
 6–10214 (38.8)5.12.84–9.16272 (50.4)2.01.28–3.22421(76.7)0.90.56–1.5262(11.7)1.30.62–2.85
 11–15114 (37.5)4.02.24–7.2494(31.5)0.90.59–1.52213(71.2)0.80.51–1.3845(15.2)1.90.87–3.96
 Leukemia164 (29.3)5.03.08–8.10558 (53.4)1.81.35–2.38908(85.3)1.30.94–1.91113 (11.0)0.60.43–0.94
 Central nervous system tumor109 (24.0)3.62.21–5.94376 (50.8)1.71.26–2.26659(88.2)1.91.27–2.7087(12.0)0.70.47–1.06
 Hodgkin disease49(37.4)6.13.42–11.0054(40.0)1.30.85–2.0293(66.0)0.60.39–1.0118(13.2)0.80.46–1.51
 Non-Hodgkin lymphoma35(35.4)6.13.33–11.0665(48.5)1.71.09–2.4994(72.3)0.70.43–1.1719(14.7)0.90.51–1.65
 Soft tissue sarcoma36(36.0)6.23.34–11.3359(38.8)1.10.71–1.62102(65.8)0.50.32–0.7818(11.8)0.70.39–1.28
 Bone cancer35(41.2)7.43.95–13.9233(36.7)1.10.67–1.8156(62.2)0.50.30–0.8615(16.7)1.10.56–2.08

Reasons for SE were also examined by treatment data, adjusting for age at diagnosis and age at questionnaire completion (data not shown). Survivors who received CRT alone or in combination with IT MTX demonstrated the most difficulty with low test scores (CRT only: OR, 1.4; 95%CI, 1.07–1.71; CRT plus IT MTX: OR, 1.4; 95%CI, 1.13–1.73) and problems with learning and concentrating (CRT only: OR, 3.2; 95%CI, 2.26–4.42; CRT plus IT MTX: OR, 2.0; 95%CI, 1.51–2.67). When analyzing dose ranges, problems with learning and concentrating were most evident among those who received the highest CRT doses (> 5000 cGy: OR, 2.7; 95%CI, 1.44–4.89). Female gender, again, was associated with the highest risk of utilizing SE within these treatment groups.

Duration of SE Services

In an effort to determine whether reported utilization of SE was either an acute intervention strategy for survivors of childhood cancer or a long-term component of their education experience, we examined duration of SE services relative to grade level at diagnosis. Figure 2 represents the overall frequency of SE by grade level at diagnosis for survivors and siblings. Children who were diagnosed at younger ages tended to be placed in SE services and remained in those programs for a longer time. Duration of SE services also was examined comparing survivors who received CRT with survivors who did not receive CRT (data not shown). The mean number of years in SE was 5.7 years for the group that received CRT and 4.7 for the group that did not receive CRT (t = 6.77; P < 0.01). Furthermore, a statistically significant difference was found in the duration of SE services between survivors who received no CRT, survivors who received low-dose CRT(1–1999 cGy), and survivors who received high-dose CRT (≥ 2000 cGy) using a one way analysis of variance (F = 25.91; 2 degrees of freedom; P < 0.01). The means for all three groups were significantly different (P = 0.05) from each other, with survivors who received high-dose CRT needing SE services for the longest time (5.8 years) and survivors who never received CRT needing them for the shortest time (4.7 years).

thumbnail image

Figure 2. Overall frequencies of special education by grade at the time of diagnosis (patients with all cancer diagnoses and siblings).

Download figure to PowerPoint

Educational Attainment

Educational attainment of survivors, compared with members of the sibling cohort, is presented in Table 4. Logistic regression analysis showed that survivors of leukemia, CNS tumors, non-Hodgkin lymphoma, and neuroblastoma were significantly less likely to complete high school compared with siblings. No significant difference in high school completion was identified between siblings and survivors of Hodgkin disease, kidney, soft tissue sarcoma, and bone tumors. Furthermore, comparison of treatment types (no CRT or IT MTX, CRT only, IT MTX only, and CRT plus IT MTX) showed that all survivors, regardless of their type of treatment, were significantly less likely to complete high school compared with their siblings. College graduates showed a pattern similar to those who completed high school: Typically, the same diagnostic groups were less likely to complete college compared with siblings.

Table 4. Educational Attainment Overall and by Enrollment into Special Educationa
Characteristic% Non-high school completersbOR95%CI% Non-college graduatesaOR95%CI% SE enrollees who did not complete high schoolOR95%CI% SE enrollees who did not graduate collegeOR95%CI
  • OR: odds ratio; 95%CI: 95% confidence interval; SE: special education; CRT: cranial radiation; IT MTX: intrathecal methotrexate.

  • a

    The ORs for this table are adjusted by age at questionnaire completion. They reflect the probability that of survivors are more likely than siblings to be non-completers of high school and college.

  • b

    Calculated using patients age 18–24 years at questionnaire completion for comparison with national data.

  • c Calculated using the whole adult cohort.

 Sibling controls8.
 Central nervous system tumor18.12.71.92–3.8177.11.61.38–1.9229.42.01.04–3.7323.42.31.36–3.91
 Hodgkin disease7.11.20.71–2.1360.71.10.97–1.2512.00.80.30–2.1976.60.90.49–1.47
 Non-Hodgkin lymphoma13.11.81.15–2.7868.61.20.97–1.3728.61.50.64–3.6184.01.00.54–1.95
 Soft tissue sarcoma9.11.10.71–1.8364.31.00.84–1.1621.71.40.53–3.4983.51.00.55–2.15
 Bone cancer9.01.50.89–2.5261.00.90.81–1.1016.71.20.36–3.8379.80.90.48–1.77
 Sibling controls8.
 Neither CRT nor IT MTX11.21.51.10–1.9465.81.10.99–1.2026.31.70.88–3.0983.71.10.68–1.68
 CRT only16.82.51.76–3.6778.01.71.41–2.0327.51.80.92–3.4391.01.91.08–3.17
 IT MTX only14.21.61.11–2.1872.81.10.97–1.3134.61.90.90–3.7890.21.50.81–2.90
 Both CRT and IT MTX14.61.81.33–2.4277.41.41.19–1.5824.81.30.70–2.4388.01.10.68–1.82

A slightly different picture emerged when we examined high school and college completion for those who had ever received SE. The only survivors who received SE services and were significantly less likely to complete high school compared with siblings were patients who were diagnosed with CNS tumors (OR, 2.0; 95%CI, 1.04–3.73) and kidney cancer (OR, 2.6; 95%CI, 1.13–5.93). Survivors of CNS tumors who received SE also were less likely to finish college compared with siblings, but patients with all other diagnoses showed no significant difference with siblings in terms of college completion. Treatment type did not show a significant correlation with high school graduation for survivors who were in SE programs, and only survivors who were in SE and received CRT were significantly less likely to complete college compared with siblings.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Long-term sequelae of childhood cancer have been the focus of a large amount of research over the past 3 decades. Unfortunately, issues such as small sample sizes, single-institution populations, and short length of follow-up often have hindered the generalizibility of important findings among this group. Most of the research involving educational difficulties has focused on cognitive testing through various intelligence quotient (IQ) measurements, whereas little attention has been paid to actual SE placement. In one of the few published reports that examined SE, Haupt et al. studied survivors of ALL who were more likely than their siblings to require SE services.9 Similarly, Mulhern et al. also studied ALL survivors and found they had a greater frequency of SE intervention compared with survivors of Wilms tumor.10 The CCSS is able to expand on previous literature not only by determining utilization of SE and educational attainment for survivors of ALL but also by describing and comparing these outcomes among a much broader spectrum of childhood cancer survivors.

The current analysis was conducted to identify 1) the utilization of SE services by patient and disease characteristics and 2) the level of education attained among survivors and siblings. However, the results of this analysis must be interpreted within the confines of the study's limitations. Again, it is important to mention that the data used for this analysis were self-reported and were not validated through school records. Whether or not a participant indicated they received SE services depended on their perception of what SE is. Furthermore, there is potential for bias when examining the reasons for entering SE, in that it probably is more desirable socially to report the need for SE due to missing school rather than low test scores. It also is possible that some survivors may have been placed into SE programs preemptively by school staff in an effort to provide as much assistance as possible to these children, a benefit probably not offered equally to siblings. In addition, approximately 10% of survivors in the CCSS are from racial minority groups, which is slightly lower compared with the published SEER data for this same period (16%).11 Because minority students typically receive SE at higher rates compared with white students,12 we may be underestimating the utilization of SE within the survivor population. Moreover, although it has been shown that parental education level and SES affect a child's educational attainment, these variables were not available for the current analysis. One method to control for parental education and SES effects would be to use a 1:1 match of full siblings. Because this analysis included closer to a 4:1 match, we were not able to control completely for these effects on the utilization of SE and educational attainment in our cohort. Finally, because information relating to SE was obtained from two different sources (i.e., directly from survivors age ≥ 18 years and from parents if the survivor was age < 18 years or had died subsequently), there is the potential for introduction of a reporting bias. We investigated the frequency of reported SE utilization by year of cancer diagnosis and found no consistent pattern suggestive of under-reporting or over-reporting according to survivor or parental report.

The baseline questionnaire for CCSS asked about placement into advanced program classes. Because these types of programs vary greatly in terms of content and accessibility, not only within each state but across the country, we felt that an analysis for advanced programs would not be appropriate. SE services, conversely, are at least somewhat consistent between states, because federal funding is appropriated for such classes for all school districts, although additional state funding may vary. Neither state of residence nor school district information was included in the baseline survey and, thus, was not available for analysis. The rate of SE utilization among survivors did vary by treatment era in this analysis, with rates lower in the earlier periods. It is unclear how much of this effect may be attributable to changes in treatment and how much may be attributable to increases in educational funding during this era. It is a fact that federal support for SE has increased steadily during the period under study.

The current analysis supports previous research showing that survivors of leukemia are at increased risk for utilizing SE services. However, an important distinction between this study and previous research is that not only were survivors of leukemia more likely than siblings to utilize SE services, but survivors of all the major childhood cancer diagnoses utilized these services more often compared with siblings. This includes survivors with diagnoses for which CNS treatment is administered rarely, if ever.

Risk factors previously identified to affect cognition and poorer educational outcomes include early age of treatment,3, 6, 10, 13, 14 female gender,3, 13, 15, 16 and the administration of CRT.2–4, 9–11, 14, 17 In the current study, 72% of survivors of leukemia and 58% of survivors of CNS tumors were diagnosed before age 11 years, whereas the majority of survivors of other cancers (excluding kidney and neuroblastoma) were diagnosed later in adolescence. In terms of treatment, 94% of survivors of leukemia received IT MTX and/or CRT, and 59% of survivors of CNS tumors were treated with CRT. Although the percentage of survivors of CNS tumors who received CRT may seem high by current treatment standards, it is reflective of the overall population of patients with CNS tumors who were treated during the study period of 1970–1986.

Because both diagnosis and treatment are correlated highly with age at diagnosis, SE use was examined by diagnosis within age at diagnosis strata. This analysis showed a significant increase in the utilization of SE services for survivors who were diagnosed at younger ages, with the strongest association found in females. We found a > 4-fold total increased risk of entering an SE program for survivors of leukemia who were diagnosed between the ages of 0 and 5 years (2.9 times higher in males and 7.6 times higher in females).

Although it is widely accepted that CRT is associated with diminished cognitive functioning, the dose level at which these problems occur seems to be inconclusive. A large proportion of previous studies found that patients who received at least 2400 cGy were at the greatest risk;3, 4, 9, 10, 13, 14 while a few studies also have reported this association in individuals who received a lower dose of radiation: typically, 1800 cGy.2, 17 In addition, a possible synergistic effect between the administration of IT MTX in combination with CRT also has been implicated.18, 19 The current analysis shows an association between the utilization of SE services and survivors who were treated with CRT. This association is evident when examining CRT as a dichotomous variable and by dose range. We found that survivors who received CRT alone were seven times more likely to utilize SE services compared with survivors who did not receive either CRT or IT MTX. It is important to note that the dose level of CRT may be a better predictor of SE compared with whether a survivor ever received CRT. Nearly 74% of survivors who received CRT alone are survivors of CNS tumors. Because these patients typically receive higher doses of CRT, this association may be due to a higher radiation dose within the CRT-only group. However, when CRT is administered in combination with IT MTX, the deleterious effects of CRT seem to be compounded. It also is evident that, although a lower dose of CRT seems to modify the risk of utilizing SE services, the risk remains significantly elevated compared with the risk in survivors who did not receive any CRT.

A secondary focus of this analysis was to determine the reasons why survivors of childhood cancer are placed into SE programs. The format of this project was such that we were only able to measure the use of SE and the self-reported reasons for entering such programs. It did not allow us to make any conclusions about cognitive deficits among survivors of childhood cancer. Undoubtedly, some survivors of childhood cancer are placed into SE because of cognitive impairments, but it is also possible that survivors are placed into SE for reasons not necessarily due to cognition. Clearly, when school-age children are diagnosed with cancer or other serious childhood illnesses, absenteeism becomes an important issue during the treatment phase of the disease. Standard therapy for patients with childhood cancers often consist initially of intensive therapy that may last several months to 1 year. This often is followed by a less intensive regimen that may continue for several more years. The length of treatment can have a serious impact on scholastic functioning. When a child is receiving treatment on a regular schedule, it is very likely that the same school subject is missed routinely. We found that missing school was a significant factor in determining placement into SE programs for survivors of all cancer diagnoses, with survivors of bone cancer, soft tissue sarcoma, Hodgkin disease, and non-Hodgkin lymphoma reporting a increased risk of six times the risk for siblings. It was found that this association was significantly higher for the same groups of survivors when examining off-therapy relative odds. However, because the question relating to the reasons for SE was not specific for each year in SE, we were unable to determine whether reasons for continued SE use changed over time.

We also identified low test scores as one of the reasons associated with utilizing SE services for survivors who were diagnosed before age 11 years as well as for survivors of kidney cancer, CNS tumors, and non-Hodgkin lymphoma. The questionnaire, however, did not distinguish between low scores on standardized tests or individual subject tests. Therefore, some of the association between low test scores and the need for SE may be related to missing too much of a particular subject area.

Survivors who received CRT also appear to have problems with learning and concentrating. Only survivors of CNS tumors showed an elevated risk due to this cause. CRT, which is the most common form of treatment for these patients, also was associated significantly with problems of learning and concentrating. When we examined CRT by dose range, we found that the association primarily held true only for survivors who received the highest dose of CRT. It also is possible that specific cognitive deficits among the group with CNS tumors are the result at least in part to the location of the tumor and/or the surgical intervention, two factors that were not included in the current analysis. It is worth noting that emotional and behavioral problems were not associated with the need for SE for any group of survivors in terms of diagnosis category, age at diagnosis, or treatment type.

The current report indicates that survivors of leukemia, CNS tumors, non-Hodgkin lymphoma, and neuroblastoma were significantly less likely to graduate high school compared with siblings. This is in contrast to previous findings. One study found that, although survivors of ALL were more likely than sibling controls to enter an SE or learning-disabled program, most were able to overcome their problems and generally had the same probability as their siblings of finishing high school, entering college, and earning a bachelors degree.5 A similar study that was conducted among patients who were diagnosed with several different types of childhood cancer also found that, except for patients with CNS tumors, survivors of childhood cancer had no impairment of educational attainment compared with their sibling controls.20

An encouraging aspect of this investigation is that when we examined high school graduation rates of survivors who were enrolled in SE programs, we found that, in general, they were no less likely to complete high school compared with other children in SE without a cancer history. We have no clear explanation for the apparently high proportion of survivors of kidney cancer who received SE and did not graduate from high school. The therapy they received would not be expected to affect their scholastic ability, nor would it render them more likely to utilize SE at all. However, the absolute numbers are small and are not very different from the numbers for survivors of neuroblastoma, who have a similar age specific distribution. With the exception of survivors of CNS tumors who used SE, survivors of all other types of cancer also were just as likely to complete college as siblings, suggesting that interventions may be effective among survivors who are less likely to have received treatment involving the CNS.

Although it is true that most long-term survivors of childhood cancer will complete high school successfully, and many will go on to college, some survivors will have significant difficulty doing so. By identifying the individuals most at risk for problems with school performance, clinicians and educators will be able to foresee potential problem areas better and can choose to initiate SE services early on in these children's education. Several long-term follow-up clinics have created models for the interaction between physicians, nurses, and school staff members to better serve the increasing population of cancer patients and survivors at risk for cognitive impairments.21, 22

Recommendations have been made suggesting that baseline IQ and achievement tests and yearly sequential testing should be performed on all children who are treated for CNS cancers because of the high incidence of learning disabilities in this population.23 In light of the current research findings, an argument can be made that these tests should be performed in a much larger segment of children who are diagnosed with cancer. Clearly, children who are treated with CRT and IT MTX should be followed much more carefully to identify early signs of learning disabilities. Children who already are at risk for poorer educational outcomes, such as racial minorities and patients from lower SES families, also should be monitored closely to determine the extent to which the diagnosis of cancer further impedes school performance. Because providing at-risk children with the necessary SE services and guidance from teachers and other mentors early on can be the key to successful completion of scholastic goals. We can also suggest all children diagnosed and treated for cancer should be monitored closely and evaluated to allow for early intervention.

The CCSS is comprised of a wide age range of over 14,000 survivors of childhood cancer and 3500 sibling controls. It includes detailed treatment history and has the largest follow-up of any such study to date. The relative size and scope of the CCSS will prove to be an invaluable resource in better understanding the potential late effects of treating young children and adolescents for malignant disease. In terms of this investigation, the CCSS is unique in that it is able to describe the educational achievements of a much broader spectrum of childhood cancer diagnoses than any other publication to date. Although treatment factors, such as CRT and IT MTX clearly play a role in poor academic achievement, this study shows that, to some degree, other non-CNS treatment variables also can impact scholastic functioning.

Although the results from the current analysis of the CCSS cohort provide important information, there is a clear need for researchers to further investigate the utilization of SE and educational attainment among childhood cancer survivors. The CCSS is under-represented in terms of racial minority survivors; thus, the potential confounding effect of race and SES needs to be examined in other, more diverse survivor populations. Understanding the influence of race and SES may provide important insights into how to maximize educational opportunities within racial/ethnically defined subgroups. Moreover, with the increasing number of immigrant children of non-English speaking families, it also would be beneficial to focus research on this distinct population. Another area that the current analysis was unable to address adequately related to the reason surrounding the need for SE over time. It is likely that individuals who are newly diagnosed with a childhood cancer may be placed into SE due to an excessive number of school days missed. However, is it still true that children are in SE years later for the same reason, potentially due to somatic issues as a result of their treatment, or do low test scores and behavioral issues tend to develop at a later time? Our analysis showed that rates of SE differed significantly across the period of eligibility for CCSS. SE went through a considerable amount of reform during the 1970s, which may or may not have had a significant impact on our findings. Additional research may be limited to include more recently diagnosed and treated children to evaluate current SE standards and criteria more closely. Future research clearly is needed to address these issues and aid physicians, nurses, social workers, and educators to work together to develop individualized plans that help survivors accomplish educational goals.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors express their appreciation to Catherine Moen for her valuable contribution to this effort.


  1. Top of page
  2. Abstract
  6. Acknowledgements