Cancer Control Research Program, British Columbia Cancer Research Centre, British Columbia Cancer Agency, Vancouver, BC, Canada
School of Population and Public Health, University of British Columbia, Vancouver, BC, Canada
Correspondence to: Mary L. McBride, Cancer Control Research Program, British Columbia Cancer Agency, 2-107, 675 West 10th Avenue, Vancouver, British Columbia, V5Z 1L3, Canada, Tel.: 604-675-8059, Fax: 604-675-8180, E-mail: firstname.lastname@example.org
To estimate the risk of late morbidity leading to hospitalization among young adult cancer 5-year survivors compared to the general population and to examine the long-term effects of demographic and disease-related factors on late morbidity, a retrospective cohort of 902 five-year survivors of young adult cancer diagnosed between 1981 and 1999 was identified from British Columbia (BC) Cancer Registry. A matched comparison group (N = 9020) was randomly selected from the provincial health insurance plan. All hospitalizations until the end of 2006 were determined from the BC health insurance plan hospitalization records. The Poisson regression model was used to estimate the rate ratios for late morbidity leading to hospitalization except pregnancy after adjusting for sociodemographic and clinical risk factors. Overall, 455 (50.4%) survivors and 3,419 (37.9%) individuals in the comparison group had at least one type of late morbidity leading to hospitalization. The adjusted risk of this morbidity for survivors was 1.4 times higher than for the comparison group (95% CI = 1.22–1.54). The highest risks were found for hospitalization due to blood disease (RR = 4.2; 95% CI = 1.98–8.78) and neoplasm (RR = 4.3; 95% CI = 3.41–5.33). Survivors with three treatment modalities had three-fold higher risk of having any type of late morbidity (RR = 3.22; 95% CI = 2.09–4.94) than the comparators. These findings emphasize that young adult cancer survivors still have high risks of a wide range of late morbidities.
the Classification of Cancers in Adolescents and Young Adults
BC Cancer Agency
British Columbia Cancer Registry
Childhood, Adolescent, and Young Adult Cancer Survivors Research Program
childhood cancer survivors
central nervous system
the International Classification of Diseases, version 9
International Classification of Disease for Oncology
Medical Services Plan
Personal Health Number
Vital Statistics Agency
Young adult cancer survivors
Although improvements in treatment have dramatically increased the survival rate of cancer diagnosed in young people since the 1970s, cancer survivors may suffer from a variety of significant adverse late and chronic complications. These include serious life-threatening health problems which can affect multiple organ systems. These complications are associated with the cancer itself and previous therapy which generally includes chemotherapy, radiation therapy (RT) and surgery. Late effects have been shown to become increasingly clinically significant with aging. Life-time follow-up with regular monitoring is important for survivors of childhood and young adult cancer and is likely to lead to prevention and earlier diagnosis of long term health complications. Research on these complications among long-term survivors is likely to improve our understanding of the etiology of these late effects.
Most late morbidity studies involving survivors of cancer in young people focus on childhood cancer survivors (CCS). The risk of late effects in this group is very significant. Oeffinger et al. using self-reported questionnaire data, found that 62.3% of CCS diagnosed before 21 years of age between 1970 and 1986 had at least one chronic condition and 27.5% had severe or life-threatening conditions. Geenen et al. reported that 74.5% of CCS diagnosed with cancer prior to 17 years of age between 1966 and 1996 had one or more adverse events, and 36.8% had at least one severe or life-threatening or disabling adverse event, using medical assessment of adverse events from a single institute in the Netherlands. In British Columbia, over 40% of CCS had at least one type of late morbidity leading to hospitalization, after up to 25 years of follow-up. Young adult cancer survivors (YACS) have a unique cancer spectrum as carcinomas are much commoner in this age group. Although certain cancer types can occur in all age groups from childhood to adulthood, such as lymphoma and acute leukemia, cancers in the young adult cohort tend to have different biologic and immunophenotypic characteristics. Therefore, risks and type of late effects in CCS and YACS are likely to be different. Little is known about the overall burden of late morbidity affecting YACS. The magnitude of risks of late morbidity, and disease-related risk factors, have not been fully explored among YACS.
This study was conducted using linked registry, clinical and long-term data for survivors of cancer diagnosed between 20 and 24 years of age from the Childhood, Adolescent, and Young Adult Cancer Survivors (CAYACS) Research Program. The CAYACS Program is an ongoing, population-based research resource for survivors of cancer diagnosed before 25 years old in the province of British Columbia (BC), Canada. It consists of a retrospective survivor cohort and a randomly selected comparison group, frequency-matched by sex and birth year, from population-based registries, linked with medical records and provincial government health administrative databases. Since all “medically necessary” healthcare services are government-funded, this linkage allows us to obtain information on all “medically necessary” services for the study groups. We estimated the risk of late morbidity leading to hospitalization among YACS, who had survived at least 5 years, compared to the general population; and we examined the effects of demographic and disease-related factors on late morbidity.
Material and Methods
The survivor cohort was identified from the population-based British Columbia Cancer Registry (BCCR). BCCR ascertains all newly diagnosed cancer cases among residents of BC. The ascertainment of cancer incident cases by BCCR is estimated to be 95% or better. The survivors in this study were restricted to individuals diagnosed between 20 and 24 years old from 1981 to 1999 with a first primary cancer included in the Classification of Cancers in Adolescents and Young Adults (AYA), resident in BC at the time of diagnosis, who had survived at least 5 years after diagnosis, and who linked to the Client registry of the Medical Services Plan (MSP), the provincial health insurance plan (using a person-specific Personal Health Number (PHN), available since 1986).
We defined valid linkage for survivors as a minimum of one year of MSP registry linkage during the follow-up period between January 1, 1986 and December 31, 2006 from 5 years post-diagnosis (start of follow-up). In total, 1,252 young adult survivors were identified and 902 were successfully linked to the MSP (including 448 men and 454 women) yielding a linkage rate of 72.0%.
A comparison group 10 times the size of the survivor group (9,020 individuals resident in BC 1986 or later) was randomly selected from the MSP Client registry records. The population sample was frequency-matched by sex and birth year to resemble the survivor cohort.
Late morbidity outcomes
Late morbidity was defined as any chronic or late-occurring health condition, whether or not it was a recognized late effect of the disease or treatment. In this study, morbidity leading to hospital admission was measured. Person-specific hospitalization records containing morbidity information was linked to records of survivors and comparison subjects using PHN. These records capture data on hospital admissions of all inpatient care from all health care facilities in BC, including hospitalization dates and discharge diagnosis. Emergency visits and outpatient care visits were unavailable. Each discharge record has a recorded diagnosis “most responsible” for the hospitalization, and up to 15 additional diagnoses, coded according to the International Classification of Diseases version 9 (ICD-9, Ref. ). In this study, only the “most responsible” hospital discharge diagnosis was identified, and classified according to ICD-9 chapter (chapters correspond generally to organ systems). The first hospitalization attributable to a chapter-specific type of problem was counted as the indicator of that type of late morbidity. Consequently, multiple hospitalizations with the same type of condition for each individual were counted only once in the analysis.
Data collection and follow-up
The BCCR provided demographic and diagnostic information for YACSs, including date of birth, sex, date of primary cancer diagnosis and morphological and histological codes of this diagnosis based on International Classification of Disease for Oncology (ICDO), 3rd edition. All the primary cancer diagnoses were further converted into a 10-category AYA classification based on morphological and histological diagnosis codes. Cancer-related treatment and relapse information was manually abstracted by research staff from medical records, and included chemotherapy, RT and surgery. Second cancer information was ascertained from BCCR. The MSP Client Registry provided the demographic information for the comparison group.
Both the survivor and the comparison groups were followed from January 1, 1986 to December 31, 2006, loss to follow-up, or date of death, depending on which occurred first. Loss to follow-up was identified by inactive status in the MSP. Possible reasons include moving out of BC or death out of country. Date and cause of death for survivors were obtained from the BCCR via routine linkage with the BC Vital Statistics Agency (VSA). Death information for the comparison group was obtained from a separate linkage to the VSA.
Socioeconomic status (SES), urban/rural residential classification, and region of residence at start of follow-up, were determined using residential postal code data. SES was classified according to the average neighborhood-adjusted income quintiles using census data linking residential postal codes to Statistics Canada geographic areas. Urban/rural residential classification was generated through conversion of postal code into six groups based on census data of population size and density provided by Statistics Canada. Region of residence was classified according to regional health authority.
The frequency and proportion of types of late morbidity leading to hospitalization, overall and by demographic and clinical characteristics (for survivors), were calculated for both the survivor and comparison groups, and compared using chi-square tests. In order to estimate the relative risk for the late morbidity leading to hospitalization overall and in each type of late morbidity, the Poisson regression model was applied, adjusting for demographic and clinical factors. Person-years at risk were calculated as the time from the later of 5 years after the original cancer diagnosis or the entry to the MSP to death, loss to follow-up, or the end of follow-up period (December 31, 2006). Person-time was divided into 5-year age categories to adjust for changes in the number of individuals contributing at each age group. Person-time was used as an offset in the Poisson regression. For individuals experiencing more than one type of late morbidity, only the first event in each ICD-9 chapter was counted in the analysis. The statistical software package SAS was used for data management and analysis (SAS Institute, Cary, NC).
The data file for analysis did not contain any personal identifying information. Ethical approval for this study was obtained from the BC Cancer Agency (BCCA)/University of British Columbia clinical Research Ethics Board. Data access approvals were obtained from the BCCR, BCCA Health Records for access to medical records, and the BC Ministry of Health for administrative health files.
Sociodemographic characteristics of the study groups are shown in Table 1. Among the survivors, mean age at the end of follow-up was 38.2 years (SD ± 5.8; range, 26.4–50.4), and mean time from diagnosis was 11.1 years (SD ± 5.3; range, 5.2–25.9). Since start of follow-up was at 5 years post-diagnosis, the mean follow-up time in this study was 6.1 years. Over 50% (50.3%) were female and 7.1% were deceased at the end of 2006. The mean age of the comparison group was 37.5 years (SD ± 6.3; range, 25.0–51.0) at the end of follow-up, mean follow-up time was 6.2 years (SD ± 5.2; range, 0.1–21.0), and 1.0% were deceased at the end of the study.
Table 1. Sociodemographic characteristics of survivors and comparison group
Among 902 eligible survivors, the most common primary cancer diagnoses were lymphoma (n = 209, 23.2%), carcinoma (n = 170, 18.8%), melanoma (n = 169, 18.7%) and germ cell tumor (n = 163, 18.1%) (Table 2). Almost half (n = 445, 49.3%) of the survivors were diagnosed with their primary cancer before 1990, and about one-third (n = 273, 30.2%) had surgery as their only treatment. Another third (n = 314, 34.8%) received chemotherapy as part of treatment, and 29.4% (n = 265) received RT. There were 55 survivors (6.1%) who relapsed and 10 survivors (1.1%) diagnosed with a second cancer in the first 5 years post-diagnosis.
Table 2. Factors affecting late morbidity leading to hospitalization risk among survivors
Half (n = 455, 50.4%) of the survivors had at least one type of late morbidity leading to hospitalization during follow-up (Table 2). The rates of having at least one type of hospital-related condition ranged from 67.9% for central nervous system (CNS) tumor survivors to 40.5% for germ cell survivors. One quarter of the survivors (n = 222, 24.6%) had two or more types of late morbidity, and 112 (12.4%) had three or more types of late morbidity (results not shown). Among CNS tumor survivors, 25 (44.4%) had at least two types of hospital-related morbidity. On the other hand, 37.9% (n = 3,419) of the comparators had at least one condition leading to hospitalization, and only 16% (n = 1,443) had two or more types of hospital-related morbidity. Female survivors had 50% more risk of having hospital-related conditions than males [rate ratio (RR) =1.50, 95% CI = 1.20–1.86]. Those who received all three treatment modalities had a higher risk of later problems (RR = 1.80, 95% CI = 1.06–3.05) than survivors with chemotherapy only. Survivors with a second cancer during follow-up had three times the increased risk of having at least one type of hospital-related morbidity (RR = 3.16, 95% CI = 1.36–7.32).
Figure 1 shows the overall incidence rates of late morbidity, excluding pregnancy-related hospitalizations, by sex, for both survivors and comparators during the follow-up period. For both sexes, younger survivors had a higher incidence of hospital-related morbidity compared to the population sample than older survivors. For those aged 25–29 years old, the incidence of hospital-related morbidity for survivors was 1.65 times higher than the corresponding rate for the comparison group (95% CI = 1.43–1.91, results not shown). The RR decreased to 1.5 (95% CI = 1.13–2.00) for the 35- to 39-year age group and 1.4 (95% CI = 0.93–2.10) for the 40- to 60-year age group. Meanwhile, both female survivors and comparators had a higher risk of a hospital-related condition than males before 30–34 years old, and a lower risk beyond this age.
After excluding pregnancy-related conditions, 326 instances of late morbidity were reported among survivors, representing an incidence of 65.7 per 1,000 py (95% CI = 58.7–73.2), and 3,097 were reported among the population sample with an incidence of 50.3 per 1,000 py (95% CI = 48.5–52.1) (Table 3). Neoplasms and digestive system diseases had the highest incidence (15.2 per 1,000 py and 14.8 per 1,000 py, respectively) among survivors. Of the 127 survivors who were subsequently hospitalized for cancer, 38 were hospitalized for the same cancer type; 88 had a different cancer, and in 1 case it was not clear from the coded information if the diagnosis at hospitalization was the same as the original diagnosis. After adjusting for attained age, sex, SES, region of residence, and urban/rural residential status, the risk of all hospital-related morbidity among survivors was 1.37 times higher than the comparison group (95% CI = 1.22–1.54). The highest risks for morbidity leading to hospitalization were found for neoplasms (RR = 4.33; 95% CI = 3.51–5.33) and blood diseases (RR = 4.17; 95% CI = 1.98–8.78).
Table 3. Rate ratios of late morbidity leading to hospitalization for YACSs vs. comparison groups
Compared to the population sample, the risk of having hospital-related morbidity was significantly higher for survivors who had received any combination of treatment modality (Table 4). Rate ratio was significantly higher among survivors with three treatment modalities than in the comparison group (RR = 3.22; 95% CI = 2.09–4.94). Rate ratios for survivors of lymphoma (RR = 2.74; 95% CI = 1.42–5.28) and carcinoma (RR = 3.85; 95% CI = 1.73–8.60) were also higher compared to the population sample.
Table 4. Rate ratio of late morbidity leading to hospitalization for YACSs vs. comparison by treatment
This study is the first to examine the overall burden of long-term late morbidity leading to hospitalization among YACS. We found that over half of 5-year YACS developed at least one type of late morbidity leading to hospitalization. Compared with the general population, YACS have increased risks of having hospital-related morbidity, particularly for blood diseases and neoplasms. The risks were higher among survivors receiving all three treatment modalities. These findings emphasize the importance of follow-up surveillance for late effects among YACSs.
The incidence of late morbidity leading to hospitalization found in our study was higher than the risk of severe or life-threatening conditions (CTCAE Grade 3 or 4, Ref. ) reported from studies of CCS,[4, 5, 15-17] which examined risks among survivors diagnosed from birth to 21 years old. The survivors in these studies were also diagnosed in earlier eras (1970–1986),[15-17] when treatments may have been significantly different, although a recent study on childhood cancer included survivors diagnosed in a later time period (1966–1996). In two recent studies, the estimated risks of having chronic conditions or adverse events were higher than our findings (62.3% and 74.5%, respectively), whereas the risks of severe conditions were lower (27.5% and 36.8%, respectively).[4, 5] The differences among studies may be caused by different definitions of outcomes, the inclusion of different groups of diagnoses and the proportion of survivors with different diagnoses that would be expected to have different risks of late effects, as well as the differences in follow-up periods.
Previous studies of CCS have showed that female sex was associated with a higher risk of long-term adverse outcomes, including intellectual function, cardiac dysfunction, obesity, abnormal pubertal timing, steroid-induced osteonecrosis, and primary hypothyroidism.[4, 5, 18-20] Oeffinger et al. reported 1.5 times higher risk (95% CI = 1.3–1.6) for female survivors of Grade 3 or higher health conditions. Hudson et al. found that being female was associated with reporting at least one adverse general health and mental health (OR = 1.4, 95% CI = 1.3–1.6) for all level of impairment. In our study, we have reported a similar difference of 1.5 times increased risk (95% CI = 1.2–1.9) of late morbidity leading to hospitalization for female YACS. However, the difference we observed was mainly caused by the second cancer and diseases in genitourinary system among female survivors (results not shown), suggesting that cancer treatment received in young adult period may have a different impact to the survivors comparing with the treatment delivered at an earlier age.
Our findings are consistent with previous reports that combination therapy with multiple treatment modalities, including chemotherapy, RT and surgery, is associated with an increased risk of significant morbidity.[4, 5] In our study, we found three times higher risks of hospital-related morbidity associated with the use of chemotherapy, RT and surgery for all YACS, and two times higher risks for YACS who had received treatment with RT and surgery.
Some limitations of this study are related to the record linkage methodology and the administrative data sources. Our study focused on the late effects leading to hospitalization among YACSs; as a result, conditions not leading to hospitalization were not considered in the analysis, as well as the repeated events within the same type of late morbidity. Because only active subjects from MSP were included in the study, information about subjects who had moved out of the province, was not available which may lead to overestimates in morbidity for survivors. However, linkage rates (72% for this study; 79% for the total original cohort) were high compared to participation rates for studies using conventional identification, recruitment, tracking, and data collection methods (e.g., recruitment and data collection rate = 61.4% in the CCSS study). Furthermore, loss to follow-up (4% for survivors vs. 6% for the comparison group) was extremely low. It is possible that a small number of survivors may be included in the selection of the comparison group, however, because of the small prevalence of survivors in the population, this should not compromise our ability to estimate the risk of late morbidity in this study.
Advances in diagnosis, treatment, and supportive care have improved the 5-year survival of young adult cancer patients. However, these survivors still face the high risks for a wide range of late morbidities. In our study, more than half of the survivors developed problems leading to hospitalization over a maximum of 21 years follow-up. In addition, it is possible that some exposures may have cumulative effects that are expressed over long periods of time and other late-onset effects may occur in the future. It is also possible that changes in therapy over recent years may be associated with a different spectrum of late effects in YACSs, which have not been captured in our current study cohort. Therefore, it is imperative that survivors of young adult cancer have ongoing long term follow up to screen for potential late toxicity related to their previous disease and therapy. Future studies should be conducted on screening and early detection of these late health problems, in the hope that the overall late effect risk can be reduced. In this way effective clinical programs and guidelines can be developed to meet the needs of YACSs.
We acknowledge the BC Cancer Registry, BC Cancer Agency, the BC Ministry of Health, BC Vital Statistics, and the Centre for Health Services and Policy Research at the University of British Columbia for their Cooperation in allowing access, use and linkage of the data to support this program.