Survivors of childhood cancer are at increased risk for osteoporosis, cardiovascular disease, and second malignancies—conditions for which modifiable risk factors are recognized and lifestyle interventions have shown benefit. Although some data regarding health behaviors of this population exist, receptivity to health promotion is largely unknown.
A survey was mailed to 380 survivors (age range, 11–33 years) of childhood leukemia, lymphoma, or central nervous system carcinomas (and guardians of survivors < 18 years old) to elicit data on exercise, dietary intake of calcium, fat, and fruits and vegetables, smoking status, readiness to pursue lifestyle change, quality of life, and interest in various health interventions.
Responses from 209 survivors (a 55% response rate) suggested that most did not meet guidelines for fruit and vegetable consumption (79%), calcium intake (68%), or exercise (52%), 42% were overweight/obese, and 84% consumed > 30% of calories from fat. Older (> 18 years) compared with younger (< 18 years) survivors were more likely to smoke (17% vs. 1%), to be obese (21.6% vs. 14.6%), and to have suboptimal calcium intakes (75.6% vs. 57.6%). No differences in lifestyle behaviors were observed between cancer groups. Compared with interventions aimed at weight control, improving self-esteem, or smoking cessation, the highest levels of interest were found consistently for interventions aimed at getting in shape and eating healthy. Survivors preferred mailed interventions to those delivered in-person, by telephone counselors, or via computers.
Currently in the United States, 1 in 900 young adults between the ages of 16 and 44 years is a survivor of childhood cancer.1 Encouragingly, early detection and therapeutic advancements have resulted in vastly improved 5-year relative survival rates, increasing from 30% in the 1960s to 77% at the turn of the new millennium.2 However, despite this victory, a childhood cancer diagnosis and associated therapies increase survivors' risk for a diverse range of functional deficits and comorbidities including osteoporosis, diabetes, cardiovascular disease, second malignancies, and premature death.1–9 As demonstrated in adult populations and adolescents without cancer, lifestyle interventions that promote a healthful diet, increased exercise, and abstinence from smoking lead to improvements in a wide range of chronic disease surrogate end points. However, no studies to date have examined the effectiveness of lifestyle interventions on chronic disease risk factors among childhood cancer survivors. Before such interventions are implemented, it is important to first assess the prevalence, receptivity and preferences of this population toward lifestyle and health promotion initiatives.
With the exception of smoking,10 little is known about the lifestyle behaviors of survivors of childhood cancer. In a recent study to assess multiple risk behaviors within this population, Butterfield et al.11 found that 68% consumed red meat > 3–4 times per week and 71% were sedentary. However, that study was conducted solely among adult survivors of childhood cancer who were enrolled in a smoking cessation intervention and therefore leaves a gap in knowledge regarding the health behavior practices of younger survivors and the majority of survivors who do not smoke. In addition, little is known about the readiness to undertake lifestyle change, as well as preferences for interventions of specific content (i.e., exercise and diet) and format (i.e., group classes and Internet-based programs). Such information is crucial if effective interventions are to be developed to respond to the tertiary and/or quaternary prevention needs of this high-risk population. A survey study was undertaken among three large groups of survivors of childhood cancer who received their care at Duke University Medical Center (DUMC; Durham, NC), the end result being that the data could be used to inform future prevention efforts. The specific aims of this exploratory study were 1) to determine current health and health behavior practices among survivors of childhood cancer of leukemia, lymphoma, and central nervous system (CNS) carcinomas; 2) to determine awareness and receptivity for various lifestyle interventions; and 3) to determine group differences in variables of interest across survivors of leukemia, lymphoma, or CNS carcinomas, as well as differences between younger (< 18 years) and older (≥ 18 years) survivors of childhood cancer.
MATERIALS AND METHODS
Survivors of childhood cancer (lymphoma, leukemia, or CNS carcinoma) between the ages of 11 and 33 years with ≥ 1 year after completion of treatment with no evidence of disease or with stable disease were identified through various registries at DUMC (i.e., the Central Cancer Registry, the Brain Tumor Center, the Young Cancer Survivors Program, and the Children's Cancer Center). Potential participants were approved for contact by their oncology care physician.
Two methods of contact were employed and were dependent on the age of the survivor of childhood cancer. Survivors > 18 years were mailed a cover letter, an 11-page survey, a consent form, and a preaddressed, postage-paid envelope for return. Legal guardians were points of contact for younger survivors and received a cover letter, a 6-page survey, and a consent/assent form, along with an additional cover letter (addressed to the child) and the 11-page survey for the child to complete. Separate return envelopes were provided for both guardian and child. To enhance response rates, follow-up telephone calls were made to nonrespondents and $10 cash incentives were offered to survivors and respective guardians.12 Although 497 survivor surveys originally were posted, we subsequently learned that 10 addressees were deceased and 90 were unable to be contacted at the address provided by our databases. In verifying address information, we also became aware of miscodings related to diagnosis or age and subsequently deleted 17 subjects. Thus, our total potential sample included 380 survivors.
Surveys comprised validated and widely used scales or subscales, and were formatted using colorful headers and large font in an effort to engage participants and enhance response rates.12 The survey content was modified slightly to appropriately address adult-aged survivors, minor-aged survivors, and legal guardians of minor-aged survivors. Members of all three groups were questioned about their height, weight, and perceived health of the survivor,13 exercise,14 smoking,15 and dietary intakes of fat,16 fruits and vegetables,17 and calcium.18 In addition, members of all groups were asked to rate their level of interest on a 5-point scale (1 indicating the highest level of interest and 5 indicating the lowest level of interest) for programs that addressed specific content areas (i.e., weight control, eating healthy, getting in shape, feeling better about themselves, getting better grades or increasing performance at work, improving their social life, and smoking cessation) and that utilized specific formats or delivery channels (i.e., classroom, CD-ROM, Internet, telephone counseling, and mailed material-based interventions). Surveys targeting survivors (not guardians) also included the Pediatric Quality of Life Inventory (PedsQL),19 as well as items on self-efficacy with regard to smoking, exercise, and dietary change.20 Stage of readiness was assessed for smoking cessation and exercise.21 Survivors also were asked their preference regarding who, if anyone, they would choose to accompany them if they participated in a health promotion intervention. The name and address of this person (confidante) was requested. A brief 1-page survey (comprising the 5-point interest items, i.e., intervention content areas, delivery modes) was later mailed to all addressees listed. Adult survivors were asked about their employment status, whereas guardians of younger survivors were asked about their child's grade in school. Both sets of adults were asked their perception of risk for a variety of comorbid conditions (i.e., secondary cancers, overweight, diabetes, osteoporosis, cardiovascular disease, stomach ulcers, kidney stones, and arthritis).
Surveys were posted between July and November 2003, and completed surveys were received until February 2004. Completed surveys were reviewed and data were double-key entered. The sample was described with proportions, means (± standard deviations), and medians. Differences between cancer types (i.e., lymphoma, leukemia, or CNS malignancies) and age groups (i.e., < 18 years vs. ≥ 18 years) were tested using chi-square, Kruskal–Wallis, Wilcoxon, and Fisher exact tests. Agreement between responses of parents and minor-aged children was compared using kappa statistics (for categorical data) and interclass coefficients (for continuous data). Multiple regression analyses were used to explore associations between lifestyle behaviors and physical function status.
The potential evaluable sample comprised 205 survivors of cancer who were ≥ 18 years and 175 younger survivors (plus their guardians). A majority of this sample was white (83%) and male (54%), and the distribution among cancer types was approximately 41% CNS carcinomas, 21% lymphomas, and 38% leukemias. A 55% response rate was achieved overall, and no significant differences were observed between respondents and nonrespondents on the basis of age, gender, race, disease category, or time elapsed since diagnosis.
Sociodemographic characteristics of the full respondent sample are provided in Table 1 and indicate that survivors across the three cancer groups were similar with respect to race, gender, and time elapsed since diagnosis. However, significant cancer group differences were observed with regard to current age and age at diagnosis, with survivors of lymphoma being older at both time points than survivors of leukemia or CNS carcinomas. In other words, there was a higher proportion of survivors of lymphoma among adult-aged respondents. By virtue of being older, adult-aged survivors also differed significantly from minor-aged survivors on age at diagnosis and time elapsed since diagnosis. Employment status did differ among cancer groups within adult-age survivors, with significantly more respondents with histories of CNS carcinomas reporting unemployment than survivors of leukemia or lymphoma.
Table 1. Sample Characteristics of Survivors of Childhood Cancer Participating in this Survey
All (N = 209) % (n)
CNS (n = 89) % (n)
Lymphoma (n = 43) % (n)
Leukemia (n = 77) % (n)
< 18 yrs (n = 87) % (n)
≥ 18 yrs (n = 122) % (n)
CNS: central nervous system; SD: standard deviation.
Data on respondents' perceived health, quality of life (QOL), and health behaviors are presented in Table 2. A majority of the sample reported good to excellent health, with no significant differences observed among the three cancer groups or between age groups. In addition, few differences were observed between younger and older survivors with regard to QOL, except for lower scores on the worry subscale among adults. In contrast, significant differences were observed between cancer groups on overall QOL, differences driven by the health and activity, getting along with others, and cognitive subscales, where scores were lower among survivors of CNS carcinomas. Exercise was significantly associated with QOL (health and activity subscale score) (ρ = 0.33, P < 0.01), suggesting a moderate link between survivors' physical activity and physical function.
Table 2. Health Indicators, Practices, and Intent among Survivors of Childhood Cancer (n = 209)
Overall (N = 209)
CNS (n = 89)
Lymphoma (n = 43)
Leukemia (n = 77)
< 18 yrs (n = 87)
≥ 18 yrs (n = 122)
BMI: body mass index; SD: standard deviation.
Fisher exact test.
Health and activities subscale: eight items cover ability to walk, exercise, and perform various tasks, as well as energy level.
Getting along with others subscale: five items relate to getting along with peers, forming friendships, and keeping up with peers.
Pain and hurt subscale: two items elicit information on pain and hurt.
Cognitive subscale: five items relate to solving problems, reading, writing, and memory.
Worry subscale: three items relate to worry about side effects and disease recurrence.
Perceived appearance subscale: three items elicit information on body image.
Unlike standards for adults in which a body mass index (BMI) threshold of 25 signifies the cutoff point for overweight, standards for children vary by age and gender, and tend to be more generous.22, 23 Despite this potentially conservative bias, 42% of child survivors in the current study either were overweight or obese. No differences in weight status across cancer groups were noted. Adult survivors were significantly more likely to be obese than younger survivors.
Overall, < 20% of survivors indicated that they ever smoked and approximately 10% reported that they currently smoked, with no differences observed among cancer groups for any tobacco-related measure. Older compared with younger survivors, however, were significantly more likely to report histories of tobacco use or current smoking behavior, with 17% of survivors ≥ 18 years classifying themselves as current smokers. Among current smokers, approximately one-half reported readiness to undertake smoking cessation, and a minority reported that they were sure or very sure that they could quit.
Overall, slightly less than one-half of survivors met guidelines for physical activity, a majority classified themselves as ready to undertake such change, and reported being sure to very sure that they could do so. The percentages of survivors adhering to low-fat diets and established guidelines for fruits and vegetables or calcium intake were far less.
Specifically, approximately 16% of survivors in the current study followed low-fat diets, approximately 21% practiced the 5-a-day guidelines for fruit and vegetable consumption, and less than one-third (32%) of the sample met dietary reference intakes for calcium. Although the practice of healthful dietary behaviors was far less prevalent than adherence to exercise, a majority of responders indicated that they were sure to very sure that they could make these dietary changes. No significant cancer group differences were observed in the practice of any lifestyle behavior measured, survivors' readiness to undertake healthful lifestyle change, or across levels of self-efficacy (i.e., confidence or sureness) for undertaking such change, although differences were observed between survivor age groups, with younger survivors reporting higher frequencies of strenuous exercise, readiness to exercise, calcium intake, and self-efficacy regarding consumption of five or more servings of fruits and vegetables per day compared with older survivors. Lower reported self-efficacy was observed among minors regarding perceived ability to adhere to a low-fat diet.
Levels of agreement between responses of younger survivors and their guardians are provided in Table 3. Highly significant associations were observed for all responses, with perfect agreement found for smoking status, and notably high associations were observed between survivor self-reports and guardian ratings for BMI, strenuous exercise, and calcium intake. Significant, albeit more moderate, associations were noted between survivors and guardians on moderate exercise, frequency of working up a sweat, and dietary intake of fat and fruits and vegetables.
Table 3. Agreement between Parents and Children on Health-Related Measures
Interclass correlation coefficient for continuous data.
Frequency of working up a sweat
Dietary fat intake
Fruit and vegetable intake
Data regarding level of interest in health promotion interventions of given content and delivery channel are depicted in Figures 1 and 2, respectively. Preference for interventions that focus on getting in shape garnered significantly higher scores among the sample as a whole (P < 0.01), and responses track similarly for all cancer types and age groups. It is noteworthy that although only 10% of the overall sample indicated extremely or very high levels of interest for smoking cessation interventions, these responses were reported by 70% of current smokers (n = 21). A majority of the parents of minor-aged children expressed extremely or very high levels of interest for all programs, except smoking where only 8.2% indicated this response. In terms of program delivery and compared with interventions that utilize computer-based, classroom, or telephone counseling approaches, a significantly larger proportion of survivors of childhood cancer expressed high levels of interest (P < 0.01) for mailed interventions. No significant between-group differences in preference were observed among survivors of cancer, and responses from guardians and confidantes paralleled those reported by survivors. Most survivors (approximately 58%) indicated a preference for participating in interventions with others (ranging from approximately 52% of adult-aged survivors to 63% of minor-aged survivors), with the preferred partner identified most often as a parent. Data obtained from confidantes (n = 99) also suggested a strong preference for mail-based interventions, with 81% indicating extremely or very high levels of interest for such programs. By contrast, far fewer reported extremely or very high levels of interest for alternative delivery formats, i.e., CD-ROM programs (67%), Internet-based programs (64%), telephone counseling (59%), or classroom interventions (51%).
To gauge responders' understanding of the relation between childhood cancer and other conditions, adult-aged survivors and guardians of minor-aged survivors were asked to endorse any of a number of conditions for which they believed survivors of cancer were at increased risk. A few respondents correctly identified such conditions as secondary cancers (∼ 44%), obesity (∼ 35%), osteoporosis (∼ 28%), diabetes (∼ 18%), and cardiovascular disease (∼ 15%).
Identical to a previous report by Meadows et al.,24 our survey results suggest that 89% of survivors of childhood cancer perceive their health to be good to excellent, with no differences detected in perceived health between the various groups of survivors. Data obtained on QOL also were remarkably similar to the normative data published by Varni et al.19 on both patients undergoing treatment and those receiving follow-up care for a variety of childhood cancers (N = 219: 50% leukemia, 7% CNS carcinomas, 9% lymphoma, and 33% other cancers). Data by Varni et al.19 establish QOL scores among a pediatric oncology sample which are a full 10 points lower than a healthy pediatric population, with an overall mean score of 72.2 ± 16.4 compared with our results, which reflect a mean score of 72.5 ± 16.6. The higher proportion of survivors of CNS carcinomas within our sample allowed us to detect differences in QOL between this group of survivors and others. Our data suggest that survivors of CNS carcinomas have significantly lower overall scores and lower scores for most subscales (except perceived physical appearance and pain and hurt), and thus experience relatively lower QOL. The compromised scores for function among survivors of CNS carcinomas may, in part, explain the lower rates of employment among adults within this population.
In a previous study, Emmons et al.10 reported that, compared with healthy young adults, the prevalence of smoking among survivors of childhood cancer is lower, i.e., 17%. Our results among survivors ≥ 18 years are identical. However, unlike Emmons et al.,10 we did not observe increased resistance or low self-efficacy for quitting. Differences between these findings may be explained by potential bias and by the finding that our data were obtained from respondents to a health behavior survey.
The most recent National Health and Nutrition Examination Survey indicates that approximately 15% of U.S. children and adolescents are overweight.25 As in previous reports on survivors of acute lymphoblastic leukemia,26–31 and especially on subsets receiving cranial radiotherapy,29, 30 we observe within our sample an increased prevalence of overweight (41.6%). Furthermore, of those who are overweight, approximately one-half are obese, and no apparent differences are noted between the prevalence of overweight and/or obesity with regard to cancer type. Just as in the general population, the prevalence of obesity increases with age, but is more troubling among survivors of childhood cancer because of higher overall percentages and the increased vulnerability of this population to weight-related illnesses, such as cardiovascular disease. Therefore, these data strongly suggest a need for weight control interventions within the broad spectrum of survivors of childhood cancer.
Findings of the Youth Risk Behavior Surveillance Survey (YRBSS) indicate that the majority (64.6%) of U.S. youth report participating in exercise that caused them to “work up a sweat” within the past 7 days.32 This statistic contrasts sharply with our data on survivors of cancer, which show that less than one-third (28%) of the sample are physically active. Further, only 48% of our sample met national guidelines established for exercise (i.e., ≥ 1 hour a day on most days of the week for adolescents and ≥ 30 minutes a day on most days of the week for adults).33 These findings underscore the need for increased exercise among survivors of cancer—a need that appears to exist within a context of readiness for behavior change because a majority (72%) report that they are at least thinking about getting more exercise. Moreover, interventions aimed at getting in shape garnered the highest level of interest (59% responded that were extremely or very interested), although the timing of these interventions may be important because readiness to exercise appears to decrease with age. Thus, interventions aimed at survivors < 18 years may be easier to implement and elicit greater adherence than interventions targeted to survivors ≥ 18 years. Survivors of CNS carcinomas had borderline trends to even lower levels of physical activity, as well as readiness for and self-efficacy to incorporate exercise into their lives. Therefore, survivors of CNS carcinomas may represent a population in which the need for physical activity interventions is relatively great, but for whom distinct barriers exist. Importantly, given the significant association identified between exercise and physical functioning, interventions aimed at getting in shape have the potential to not only affect health, but functional outcomes as well.
Compared with interventions aimed at getting in shape, survivors' interest level for initiatives targeting weight control per se were decidedly lower (40.3% responded extremely or very interested). This finding suggests that efforts aimed at reducing the prevalence of obesity within this population are likely to gain better acceptance if the focus of energy balance is on increasing energy expenditure rather than reducing energy intake.
It is noteworthy that participants' interest levels in interventions aimed at eating healthy obtained the second highest ranked score (48%) and their dietary intake data clearly indicate a need for increased consumption of fruits and vegetables and calcium.
The prevalence of consuming ≥ 5 servings of fruits and vegetables per day within our sample (21%) is identical to that reported by the YRBSS,32 and similar to the 26% found by Pesa and Turner in a survey of 16,200 healthy high school students.34 Still, the finding that only a minority of these survivors of cancer consume adequate amounts of fruits and vegetables raises concern, especially given established guidelines for survivors that endorse increased fruit and vegetable consumption as a probable means of improving overall health.35
An additional concern is our finding that a minority (32%) of these survivors of childhood cancer met guidelines established for calcium intake—a much lower percentage than reported in the general population where 88.9% of males and 54.1% of females (age range, 14–18 years) meet reference levels.36 Granted, osteoporosis and osteopenia are fairly prevalent in the general population. however, conditions of progressive bone loss are even more common among survivors of childhood cancer, and argue for increased rather than decreased intakes. Recent conjecture that high calcium diets not only protect against osteoporosis but obesity as well37 provides additional reason for survivors of childhood cancer to increase their intakes and also supports the need for interventions in this area.
Finally, although in recent years interest in low-fat diets has waned, it must be remembered that dietary fat is a concentrated energy source and is associated with increased BMI.38 Thus, the finding that 84% of this sample had fat intake > 30% is concerning in light of the high prevalence of overweight and obesity. Further, although dietary fat composition was not measured, it is highly likely that saturated fat intake tracked with total fat intake, and therefore was elevated. Data from the current study support nutritional guidance in survivor populations aimed at heart healthy diets.39
These findings suggest both a need for and interest among survivors of childhood cancer and their guardians in programs aimed at increasing exercise and improving dietary intake. Interest in interventions aimed at improving lifestyle behaviors likely will be fueled even more if efforts are undertaken to educate survivors and their guardians about the increased risk for comorbid conditions—conditions for which healthy lifestyle practices may indeed make a difference. To be sure, psychosocial interventions also have important roles and evoke moderate levels of interest. However, it is entirely possible that exercise and dietary interventions may act directly or indirectly to increase satisfaction and feelings of well-being within these areas as well.
Given the barriers associated with time and travel, as noted in a previous study of 978 survivors of breast and prostate carcinoma,40 we fully expected the survivors in the current sample to express a preference for home-based interventions. However, findings along these lines were surprising. First, lower preference scores were obtained for home-based telephone counseling (only 14.5% expressed extreme to high interest) and classroom-based formats (17.5% reported extreme or high interest). Second, despite the finding that this group of survivors of cancer was raised in the computer age, they reported significantly higher levels of interest in mail-based interventions, compared with those delivered via the Internet or CD-ROM (48.5% vs. 37.9% and 37.2%, respectively). Similar preferences also were expressed by guardians and confidantes. Strong bonds were noted between survivors and their guardians by data showing that a majority of survivors, both adults and minor-aged, named their guardian as the partner they would choose to accompany them when participating in an intervention. Thus, partner-based interventions that capitalize on social support may hold large promise in this population.
The primary limitation of the current study relates to response bias. In conducting our initial mailing, we underestimated the number of unusable addresses that we would encounter and have come to appreciate a methods paper by Mertens et al.41 that details the sizable effort necessary to identify and recruit living survivors of childhood cancer who have current and accurate contact information. That said, our survey response rate of 55% still is within the limits purported by Dillman as acceptable.42 Finally, the data that we collected on perceived health, as well as QOL, corresponded nearly exactly to results reported by Varni et al.19 Therefore, although we acknowledge the potential for response bias, our data suggest it to be of minimal threat.
In summary, except for lower scores on QOL reported by survivors of CNS carcinoma, we found few other differences in reported health and lifestyle behaviors among survivors of leukemia, lymphoma, and CNS carcinomas who participated in the survey. Overall, scores for perceived health and QOL were lower than norms published for healthy populations, yet similar to those reported previously among survivors of childhood cancer. Likewise, smoking rates were lower than those reported in the general population and identical to data previously reported among survivors of childhood cancer.10 Although a minority of the sample reported consuming five or more servings of fruits and vegetables per day, prevalence data for meeting guidelines were similar to the general population. In contrast, our findings among survivors of childhood cancer suggest higher prevalences of overweight and obesity, physical inactivity, and inadequate calcium intake compared with population norms. Given that survivors of childhood cancer have an increased risk for secondary cancers, cardiovascular disease, osteoporosis, and diabetes, these findings initially give pause and then provide evidence that lifestyle interventions may indeed hold particular promise for this vulnerable population. The need to promote and encourage healthy behaviors and lifestyles among survivors of childhood cancer was recently identified by a delphi panel of healthy policy experts,43 and also is documented in risk-based guidelines for survivors of pediatric cancer established by the Children's Oncology Group.44 The time has come to aggressively address this need with interventions that are acceptable and bear optimal potential for success.
The authors gratefully acknowledge the efforts of Denise Snyder, Shelley Rusincovitch, Amy Hensley, Meredith Wood, Shannon Eaton, Cheri Willard, Bercedis Peterson, Nicholas Wahnfried, Miriam Nelles, Tammy Manago, and Petra Wahnfried.