|Home | About | Journals | Submit | Contact Us | Français|
The number of adult survivors of childhood cancer in the U.S. is increasing due to effective treatments and improved survival. The purpose of this study was to use a national, population-based sample to estimate the burden of illness in adult survivors of childhood cancer.
A total of 410 adult survivors of childhood cancer and 294,641 individuals without cancer were identified from multiple years of the National Health Interview Survey. Multiple measures of burden, general health, and lost productivity were compared using multivariate regression analyses including: logistic, polytomous logit, proportional odds, and linear models.
Controlling for the effects of age, gender, race/ethnicity, and survey year, adult survivors of childhood cancer reported poorer outcomes across the majority of general health measures and productivity measures than individuals without cancer. Survivors were more likely to report their health status as fair or poor (24.3% vs. 10.9%; p<0.001); having any health limitation in any way (12.9% vs. 3.4%; p<0.001); being unable to work due to health problems (20.9% vs. 6.3%; p<0.001); and being limited in the amount/kind of work due to health problems (30.9% vs. 10.6%; p<0.001). When categorized by time since diagnosis, cancer survivors had poor health outcomes in every time interval, with the greatest limitations in the initial 4 years following diagnosis and 30 or more years after diagnosis.
Adult survivors of childhood cancers have poorer health outcomes and more health limitations than similar individuals without cancer across multiple measures.
As a result of effective treatments and improved survival for childhood cancer patients 1, the number of survivors of childhood cancers has increased 2, 3 to an estimated 329,000 survivors in the U.S. in 2005 4. Numerous studies have documented the late effects of treatment, including second cancer diagnoses, psychosocial and cognitive problems, and other physiological problems 2, 5, 6. Among the studies examining these effects 2, 5, 6, most focus on physiological, cognitive, or psychosocial effects rather than other limitations in functioning and productivity beyond employment status. Furthermore, adult survivors of childhood cancers face increased risks of medical or psychological effects of childhood cancers which may impair functioning or the ability to work 3, 7.
The Childhood Cancer Survivor Study (CCSS) is a well-documented cohort of over 10,000 individuals from 26 clinical centers who were diagnosed with cancer as children and their siblings 8. While the CCSS is large and geographically diverse, with a long follow-up, the study is not nationally representative as childhood cancer patients were identified and treated at academic centers and care received at these institutions may not be representative of care received in a community setting. Geographic and socioeconomic variation in care have been observed for many health outcomes 9, 10,11, particularly between academic centers and community-based institutions 12,13,14,15,16,17, 18. Additionally, participation in the CCSS was based on only specific diagnoses8, and does not include the entire spectrum of childhood cancers. Other studies of adult survivors of childhood cancers have few survivors more than 30 years from their cancer diagnosis 6, 19, 20, and cannot examine changes in health by time since cancer diagnosis 2, or cannot examine differences between survivors and similar individuals without cancer 6. In a recent report on cancer survivorship, the Institute of Medicine identified the lack of nationally representative studies of economic factors such as employment as an important research gap 21. This study extends a previous study of the burden of illness in cancer survivors 22, and uses multiple years of data from a large nationally representative survey to focus on burden of illness in adult survivors of childhood cancers.
The sample was selected from the most recent years of the National Health Interview Survey (NHIS), an annual multistage survey conducted in the civilian non-institutionalized population of the U.S. The NHIS collects data on a broad range of health topics through personal household interviews with 35,000–40,000 households across the U.S. collecting information on about 75,000–100,000 individuals. During the study period, 1997–2006, approximately 31,400 adults on average completed the Sample Adult component of the survey each year, with response rates ranging from 69.0% to 80.4%. More information on survey design and content is available from http://www.cdc.gov/nchs/nhis.htm.
In each year, the survey contained a question “Have you ever been told by a doctor or other health professional that you had cancer or a malignancy of any kind?” Individuals who responded affirmatively were then asked about the type of cancer and the age at diagnosis for each cancer diagnosis. We identified 803 individuals ages 18 and above who reported they had been diagnosed with cancer before they were age 20. These data were not edited for reasonableness, and some survey respondents appeared to have misinterpreted the question about age at diagnosis and may have reported the number of years since their cancer diagnosis 23. For example, several respondents reported that they had been diagnosed with lung cancer before they were 10 years old, but lung cancer incidence in children is exceedingly rare and is too small to calculate a rate 24. As done in previous studies with these data 22, cancer survivors with responses to age at cancer diagnosis in age ranges where site-specific incidence rates were too small to estimate rates from SEER were excluded from the sample (N=393). Our final sample consisted of 410 childhood cancer survivors NHIS respondents who reported never being diagnosed with cancer or who reported only non-melanoma skin cancer were the comparison group.
Sample characteristics included age at the time of the survey, gender, race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Hispanic, and Non-Hispanic Other), marital status, smoking status (current, former, never/missing), body mass index (underweight, normal weight, overweight, obese), and educational attainment (< high school, high school graduate, some college, college graduate or higher). Health insurance was categorized as private or military, public with no private or military, or none.
Co-morbid conditions were measured with a series of questions related to functional limitations. Individuals who identified any functional limitations were asked about the condition(s) or health problems underlying the limitation, therefore co-morbid conditions were reported only in relation to their impact on functional limitations. Co-morbid conditions were also categorized by absolute number of conditions (0, 1, 2, 3+).
Type of cancer was also abstracted for cancer survivors, who were grouped by leukemia/lymphoma, CNS/brain cancer, and solid tumors. Individuals who indicated they had been diagnosed with more than one cancer were categorized separately.
Time since diagnosis was calculated by subtracting age at diagnosis from age at the time of the survey (0–4 year, 5–9 year, 10–19 year, 20–29 year, and 30+ year intervals).
Health and functional limitation measures included self-rated health status (excellent, very good, good, fair, poor) and any problems with activities of daily living (ADL) including eating, bathing, dressing, getting in or out of bed or chairs, using the toilet, or getting around inside the home (yes, no) 25, 26. Other health limitations were measured with problems with instrumental activities of daily living (IADL)--needing help with routine needs, such as everyday household chores, or doing necessary business or shopping (yes, no); any limitation in any way (yes, no); and the number of days spent in bed in the past 12 months (0, 1–10, >10) 26, 27.
Lost productivity was measured by whether or not the individual had a job within the past 12 months (yes, no), limitations in ability to work due to health problems (yes, no), limitations in amount or kind of work because of health problems (yes, no), and number of days lost from work in the past year among individuals with jobs. To include lost productivity among 12,591 individuals (childhood cancer survivors and respondents without cancer) of working age but without jobs in the past week, and unable to work or limited in amount or type of work because of health and without data on the number of days lost from work in the past year, we used a measure of days lost from work variable by assigning a value of 260 days to represent a 2080-hour work year for these cases 22.
Descriptive statistics were calculated for sample characteristics. Frequencies of categorical variables were compared with chi-square statistics. We used multivariate logistic regression for binary response variables and linear regression for continuous responses, to assess the association between childhood cancer survivorship and health limitations and employment, controlling for the effects of age (<20, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, and ≥70 years), survey year, race/ethnicity, and gender. For nominal categorical response variables, a generalized logit link was used, and for ordinal categorical response variables, a cumulative logit link was used. Wald F p-values were used to compare childhood cancer survivors and similar individuals without cancer, and to compare time-since-diagnosis groups among childhood cancer survivors. An interaction term was used in some models to test for outcome differences in age groups among childhood cancer survivors and similar individuals without cancer. Results from the regression analyses are presented as predicted marginals. The predictive margins method 28 directly standardizes the outcome of each group to the covariate distribution of the population. Standardized results from logit models can be compared like percentages.
Because we excluded a relatively large number of cancer survivors who appeared to misinterpret the question about age at diagnosis, we also compared characteristics of cancer survivors included and excluded from our sample. All tests of statistical significance were two-sided and used SUDAAN 29 to incorporate the complex NHIS survey design and weighting for non-responses.
Adult survivors of childhood cancer were more likely to be younger, not married/separated, and non-Hispanic white than individuals without cancer (Table 1). Childhood cancer survivors were more likely to report having at least one co-morbid condition (42.8% vs. 29.4%; p<0.001). Of the cancer survivors, 23% reported previous leukemia/lymphoma diagnosis; 6%, CNS/brain cancer; 46%, solid tumors, (data not shown). Fewer than 1% reported a previous multiple cancer diagnoses, and 24% reported having other cancers.
Adult survivors of childhood cancer reported poorer health across general health measures, functional limitations, and productivity compared to similar individuals without cancer, controlling for the effects of age, race/ethnicity, gender, and survey year in multivariate analyses (Table 2). Cancer survivors were more likely than individuals without cancer to report their health status was poor or fair, needing assistance with ADLs or IADLs, having any limitation in any way, and spending more days in bed in the past 12 months.
Fewer cancer survivors than similar individuals without cancer reported having a job in the past year. More cancer survivors were unable to work because of health problems (20.9% vs. 6.3%; p<0.001), or be limited in the amount or kind of work due to health problems (30.9% vs. 10.6%; p<0.001) (Table 2). Cancer survivors were more likely to have more days lost from work (40.2 days vs. 15.1 days; p<0.001).
When examining time since diagnosis among childhood cancer survivors at the time of the survey, 45 (11.0%) were diagnosed within 4 years, 42 (10.2%) within 5–9 years, 92 (22.4%) within 10–19 years, 77 (18.8%) were diagnosed within 20–29 years, and 154 (37.6%) were diagnosed more than thirty years before the survey. Cancer survivors had poor health outcomes in multiple intervals, with the greatest limitations in the initial 4 years after diagnosis and 30 or more years since diagnosis. For example, 21.1% of cancer survivors reported fair or poor health status in the initial 4 years after diagnosis and 39.8% fair or poor health in 30 or more years after diagnosis (Figure 1a). Approximately 20% of survivors reported more than 10 days spent in bed in the past 12 months in the initial 4 years and 30 or more years after diagnosis (Figure 1b).
Among measures of lost productivity (Figure 1c), the highest proportion of survivors limited in the amount or kind of work because of health problems was in the group 30 or more years following diagnosis. Similarly, survivors lost an average of more than 20 days from work in the 0–4 year, 10–19 year, and 20–29 year intervals, but in the 30 or more year interval, survivors lost approximately 69 days from work (Figure 1d).
We also compared health limitations and lost productivity in cancer survivors and similar individuals without cancer by age group in adjusted analysis (Figure 2). A higher percentage of cancer survivors than similar individuals without cancer reported fair or poor health (Figure 2a; p<0.001), spending more than 10 days in bed (Figure 2b; p<0.001), and being limited in the amount or kind of work (Figure 2c; p<0.001). Cancer survivors were more likely to have more mean work loss days, and differences were higher in older age groups, with a dose-response association(Figure 2d; p-value for interaction of age and survivor status p = 0.032).
Compared to the adult survivors of childhood cancer included in our sample, cancer survivors excluded from our sample were more likely to be female (76.2% vs. 53.1%) with lower levels of educational attainment (22.4% vs. 10.4% had not graduated from high school). Approximately 40% of individuals excluded reported a cervical cancer diagnosis, contributing to a higher proportion of females excluded. The age distribution and proportion with fair or poor health in the two groups were similar. Notably, there were no differences between the two groups by other measures of burden and lost productivity including health limitations, bed days, and days lost from work due to health problems.
In this study, we used a population-based, nationally representative sample of adult survivors of childhood cancer and similar individuals without cancer to assess the burden of illness. Adult survivors of childhood cancer were more likely to have diminished health and greater productivity losses compared to similar individuals without cancer, adjusted for the effects of age, gender, race/ethnicity and survey year in multivariate analyses. These findings were consistent across multiple measures of health and productivity. We also found that adult survivors of childhood cancer reported health limitations and productivity loss at multiple times since diagnosis, with the greatest limitations in the initial 4 years and 30 or more years after diagnosis. The prevalence of health limitations and productivity loss in adult survivors over 30 years from diagnosis might reflect earlier childhood cancer treatment patterns that are associated with more morbidities compared to more recent treatments 5 as particularly survivors who were treated in the 1970s and 1980s may have been exposed to a specific treatment or cumulative dose that puts them at particularly high risk for other chronic disease conditions 30. This population will be an important target for development of interventions to improve functionality and health outcomes.
These high levels of burden, co-morbid conditions, and poorer health status are consistent with other reports of adult survivors of childhood cancer 2, 3, 5, 6, 20, 31. Studies from the CCSS reported survivors were twice as likely to have performance limitations than their siblings without cancer (19.6% vs. 11.8%) 3. Additionally, a population-based study in Canada found fewer survivors reported very good or excellent general health compared to controls (62.3% vs. 71.2%) and survivors were more likely to have one or more co-morbid conditions 19. Because of the potential of selection bias in a clinic-based population, estimates of prevalence of chronic disease may be overestimated 32 and may not reflect care in the community. Our study complements prior studies conducted in academic institutions or cancer centers by providing nationally-representative, population-based estimates of burden of illness in the U.S.
We found significant limitations in workforce participation and productivity in cancer survivors compared to similar individuals without cancer. These findings are consistent with other studies 20, 33 and a recent meta-analysis which reported that cancer survivors were less likely to be employed compared to healthy controls 7. We also found that adult survivors of childhood cancer were more likely to report having public health insurance without supplemental private or military coverage, which is also consistent with other studies 34. Having public health insurance can impact access and quality of care as some physicians do not accept patients with public health insurance, patients may not have access to specialty care, or patients may delay care due to cost 21, 35. Because health insurance is generally obtained through employment in the U.S. and adult survivors of childhood cancer are less likely to participate in the workforce, future work on the inter-relationship between work, health insurance, and health care utilization will be important 36.
Clinical guidelines recommend risk-based preventative services to cancer survivors 21, 37, 38. However, childhood cancer survivors do not always attend follow-up clinics 39, do not receive appropriate mammography screening 40, or do not report receiving medical care based on being a cancer survivor 41. Lack of risk-based care has been reported to be greatest among the oldest survivors 41, the population with the largest differences in burden of illness in our study. In addition to risk-based care, the development and dissemination of effective interventions specific to childhood cancer survivors and their unique needs as they age to improve quality of life are needed. For example, workplace reintegration programs or other primary prevention programs such as physical activity or tobacco cessation programs, is an important area of research for this population. The costs associated with medical care for adult survivors of childhood cancer are largely unknown and are also an important area of future research.
Despite the strengths of a nationally representative, population-based study with high response rates, there are several limitations to our study. While the weighted number of adult survivors of childhood cancer in the NHIS is a reflection of the size of this population in the U.S., our sample of survivors is relatively small. We excluded a large number of cancer survivors because of possible misinterpretation of the question about age at diagnosis. Approximately 40% of those excluded from the analysis were women who reported they had cervical cancer under the age of 20 years. Because invasive cervical cancer is extremely rare in this age group, it is more likely they were diagnosed with a pre-invasive lesion or had an abnormal Pap smear under the age of 20 or were referring to time since diagnosis. Furthermore, the majority of the cancer sites reported by those excluded were unlikely to be childhood cancer diagnoses (for example, breast and prostate cancer), making it unlikely we excluded any childhood cancer survivors who misinterpreted the question or had cognitive late effects hampering their ability to answer the survey. Although the excluded group was more likely to be female and have lower educational attainment than our sample of adult survivors of childhood cancer, their age distribution, self-reported health status, health limitations, bed days, and days lost from work were similar to the included population. This suggests that excluding this group was unlikely to have impacted our estimates of health limitations and lost productivity. Cancer patients have been found to under-report history of cancer compared to medical records or tumor registries 42, therefore the possible inclusion of some survivors in the comparison group may understate the differences between the two groups. Although there is also a possibility of recall bias as survivors may be less likely to remember the type of cancer they had as a child, this study focused on reporting of current health rather than prior behaviors. Due to the relatively small size of the general population of adult survivors of childhood cancer in the U.S., the survivors in some of the rare cancer diagnoses groups may not have been represented in this analysis. Finally, burden of illness and late effects in adult survivors of childhood cancer varies by cancer site and treatment 37, particularly as treatment patterns in this population have changed over time, but we were unable to examine differences due to the absence of treatment data in the NHIS.
In this study, we found adult survivors of childhood cancers had poorer health outcomes than did similar individuals without cancer across several burden and productivity measures. The prevalence of health limitations and lost productivity was highest in survivors in the initial 4 years and 30 or more years after diagnosis. Given the increasing number of adult survivors of childhood cancers, it is important for this population to receive risk-based healthcare and for physicians to be aware of the risks for this population for decades following diagnosis.
Funding: The National Cancer Institute provided support for this manuscript.
Presented in part at the 2009 American Society of Preventive Oncology (ASPO) in Tampa, Florida.