SEARCH

SEARCH BY CITATION

Keywords:

  • childhood cancers;
  • survivorship research;
  • late morbidity;
  • late effects

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Our study examines inpatient, hospital-related morbidity in a geographically-defined cohort of long-term cancer survivors diagnosed before age 20 years in the province of British Columbia (BC), Canada. A total of 1374 survivors diagnosed from 1981 to 1995 surviving at least 5-years postdiagnosis, and a matched sample of 13,740 BC residents, were identified from population registers, and linked to provincial hospitalization records from 1986 to 2000. Logistic regression was used to assess relative risk and effect of sociodemographic, clinical, and temporal factors on risk. Approximately 41% of survivors vs. 17% of the population sample had at least one type of hospitalization-related late morbidity in the observation period (adjusted RR 4.1, 95% CI 3.7–4.5). Those at highest risk were survivors of leukemia (RR 4.8, 95% CI 4.0–5.8), central nervous system tumors (RR 4.8, 95% CI 4.0–5.8), bone and soft tissue sarcomas (RR 4.9, 95% CI 3.8–6.2), and kidney cancer (RR 4.9, 95% CI 3.4–7.0). Adjusted relative risk was elevated for all types of morbidity except pregnancy and birth complications, and highest for neoplasms (including second primary cancers) (RR 21.7, 95% CI 16.3–28.7). Morbidity was elevated for all combinations of primary treatment and highest for those with previous radiation, chemotherapy, and surgery (RR 7.1, 95% CI 5.5–9.0). Over time, morbidity for late effects other than neoplasms became more prevalent. These results suggest that survivors are at increased ongoing risk of many types of hospital-related late morbidity, implying that long-term monitoring for multiple health problems is warranted.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Improvements in treatment of childhood and adolescent cancer patients have resulted in dramatic increases in survival in the last three decades.1 However, survivors appear to have increased risks of long term problems, mainly treatment-related.2 Overall, it has been reported that up to 75% of childhood cancer survivors have at least one chronic therapy-related health problem, and between 30 and 50% of survivors have severe or life-threatening conditions.1, 3–11 Documented late effects include cardiopulmonary effects, endocrine effects, neurocognitive effects, psychosocial issues, and effects on other systems such as the genitourinary tract, the gastrointestinal tract and liver, ophthalmic structures, aural structures, musculoskeletal structures, and the immune system, as well as second malignancies.12, 13 Quantitative assessment of risks is affected, however, by the representativeness of the study group, and the sources and methods of identification of subjects, follow-up, and outcome data collection.

The childhood/adolescent/young adult cancer survivors (CAYACS) Research Program14 utilizes population registries, administrative datasets, and record linkage methodology to create a population-based longitudinal survivor research database with follow-up to December 31, 2000, to examine long-term health and educational effects of childhood cancer, and health care utilization, in the total population of 5-year cancer survivors diagnosed among residents of BC, Canada before age 25 years, from 1970.

Within the CAYACS program, the current study examines late hospital-related morbidity from 1986 to 2000 for childhood cancer survivors diagnosed under age 20 years between 1981 and 1995. Our objectives were to estimate the risk of developing late hospital-related morbidity as compared to the general population, and to assess the potential effects of sociodemographic, clinical and temporal factors on risk.

Material and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study group ascertainment and data linkage

Identification of the survivor cohort.

The survivor cohort was identified from the BC Cancer Registry. Individuals were eligible if they met the following criteria: diagnosed with a primary cancer included in the International Classification of Childhood Cancer (ICCC)15 before age 20 years between 1981 (5-year survivors from 1986, the start year for hospitalization records) and 1995; resident in BC at time of diagnosis; survived at least 5 years after diagnosis; and linked to registration files from the provincial health insurance plan using a unique Personal Health Number (PHN). A case that was registered in the plan on or after their 5-year survival date in at least 1 year between 1986 and 2000 was considered successfully linked. 1587 survivors were identified; 1374 (86.6%) of these were successfully linked to health insurance plan records and comprise the survivor study group.

Identification of comparison group.

A randomly selected frequency-matched sample of 13,740 BC (pop. 4 million) residents was obtained from health insurance plan registry records; frequency matching was done by gender and birth year to mirror the distribution of the study survivor group. In previous CAYACS work, 10:1 matching has been found to provide sufficient statistical power.

Data collection

For survivors, demographic and disease information was obtained from the BC Cancer Registry, including birth date, diagnosis date, gender, and diagnosis coded to International Classification of Diseases for Oncology (ICD-O) Version 3.16 Treatment data were manually abstracted from health records at BC Cancer Agency and BC Children's Hospital. Date and cause of death (coded to the International Classification of Diseases) for deceased individuals was available from the Registry via routine linkage with BC Vital Statistics. Death information for the comparison group was obtained through linkage to Vital Statistics. For both study groups, yearly subject-specific information was made available from the health insurance plan registry, including current insurance status (indicating BC residence), and postal code of residence. In particular, hospitalization records, including deaths in hospital, of all inpatient (ie not emergency or outpatient) care in BC are available from 1986.

Only de-identified linked files were available for analysis.

Outcome measures

Late morbidity is defined as any chronic or late-occurring conditions, whether or not they are recognized late effects of the disease or treatment. For our study, morbidity leading to inpatient hospitalization was defined as the outcome. This level of severity generally corresponds to Grades 3, 4, and 5 in the Common Toxicity Criteria for Adverse Events version 3,17 which describes Grade 3 morbidity as that “severe undesirable events with multiple disruptive symptoms and serious interventions including hospitalization,” Grade 4 morbidity as “catastrophic, disabling, and potentially life-threatening,” and Grade 5 morbidity as that “leading to death.” Hospital-related morbidity was determined by the ICD Version 918 (ICD9)) code for the “diagnosis most responsible for hospitalization” on the provincial hospital discharge abstract records. These codes were grouped by ICD9 chapter, generally equivalent to organ systems, to categorize the type of late morbidity; for each subject, only the first hospitalization diagnosis in a particular ICD9 chapter was counted as the indicator of the presence of that type of late morbidity for each survivor (multiple diagnoses or hospitalizations with the same diagnosis were not counted again).

Potential modifying variables

Along with gender and birth year (used for frequency-matching), sociodemographic factors considered as potential risk modifiers in our study included, region of residence (categorized according to regional health authority), urban/rural status of residence, and socio-economic status (SES) quintiles, all determined at start of follow-up, using residential postal code data from the provincial health insurance plan annual registry records. Urban/rural residential status was determined from postal code linked to categories based on census information on population size and socioeconomic homogeneity provided by Statistics Canada19 and grouped as: metropolitan area (census metropolitan area), large community (tracted census aggregation), small community (untracted census aggregation) and rural (any smaller communities). Census-derived neighborhood-specific income quintiles, based on the average income per person equivalent in the area and determined by postal code,20 were used to determine SES of study subjects. Since hospitals are administered by the health authorities, this variable was included to assess possible regional differences in access to hospital care.

Risk was determined overall and for six diagnosis groups (leukemia, lymphoma, CNS, kidney tumor, bone tumors/soft tissue sarcoma, and carcinoma). As well, we examined the potential effect of primary treatment, age at diagnosis, calendar period of diagnosis and time since diagnosis, which was included as a time-dependent categorical factor.

Follow-up

Survivors were followed from the five-year anniversary of diagnosis, to end of follow-up period (end 2000), loss to follow-up (indicated by inactive status in health insurance plan, denoting outmigration from province or death), or date of death, if applicable. Comparison subjects were followed from the later of fifth birthday or entry date to the health insurance plan, to end of follow-up period (end 2000), date of death (if applicable), or inactive status in the health insurance plan.

Statistical analysis

The frequency and percentage of those having each type of late morbidity (as measured by hospitalization) were calculated for the survivor and comparison groups. Cox proportional hazard regression was used to estimate the risk of each category of late morbidity among cases compared to the population sample, adjusting for sociodemographic factors. Only the first morbidity of each type was counted; for each analysis, once an individual experienced an event, they were removed from the risk set. For each relative risk (RR) estimate, a 95% confidence interval was computed. All analyses were performed using R version 1.8.1.21

Ethics approvals

Ethics approval for our study was obtained from the BC Cancer Agency and BC Childrens Hospital Clinical Research Ethics Boards (both affiliated with the University of British Columbia). Approvals for access, use and linkage of data were obtained from the BC Cancer Registry and the BC Ministry of Health. The Ministry of Health required that table cell counts of less than 5 were to be suppressed.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Of 1374 survivors, 634 (46%) were male, 402 (29%) were born before 1975, and 383 (28%) were born after 1984 (Table 1). The comparison group in our study had a slightly higher proportion of individuals in the mid-range of SES compared to survivors, tended to live in larger communities, and differed somewhat from survivors in distribution across regions (Table 1). One-third (34.6%) of survivors were diagnosed under 5 years of age, one-quarter (24.5%) were diagnosed originally with leukemia, and 61% were diagnosed from 1981 to 1990 (Table 2). Almost 23% of survivors had surgery as their only treatment; 60% had chemotherapy as part of their treatment; and 35% had radiation therapy. Mean follow-up time for survivors was 7 years from 5 year-survival (maximum 15 years).

Table 1. Sociodemographic characteristics of survivors and comparison group
inline image
Table 2. Clinical characteristics of survivors and factors affecting hospital-related morbidity risk
inline image

Although risk did not differ significantly by type of original diagnosis (Table 2), survivors of a bone tumor or soft tissue sarcoma were at highest excess risk of developing any hospital-related late morbidity (RR 1.18, CI 0.87–1.61) compared to leukemia survivors, followed by brain tumor survivors (RR 1.08, CI 0.84–1.4).

The following factors resulted in significantly lower risk of developing hospital-related late morbidity: male gender (RR 0.7, CI 0.6–0.9) (not shown); longer time since diagnosis (RR 0.46, CI 0.3–0.7 for 10+ years since diagnosis compared to 5–7 years since diagnosis); and those diagnosed more recently (RR 0.59, CI 0.5–0.7 for those diagnosed 1991–1995, compared to survivors diagnosed 1981–1985) (Table 2). Survivors of a lymphoma, kidney cancer, or cancer type other than leukemia, CNS tumor, bone/soft tissue cancer, or carcinoma, had significantly lower risk of a subsequent neoplasm related to hospitalization (either a recurrent tumor or new primary diagnosis) (RR 0.3, CI 0.2–0.7 for lymphoma survivors; RR 0.3, CI 0.1–0.9 for survivors of kidney cancer; and RR 0.3, CI 0.2–0.7 for survivors of other diagnoses), compared to leukemia survivors (data not shown). Older age at diagnosis conferred a higher, but not statistically significant, risk of subsequent morbidity leading to hospitalization (RR 1.84, CI 0.83–4.08 for those diagnosed at age 15–19 years compared to those diagnosed before age 5), this risk was much higher and statistically significant for neoplasms requiring hospitalization (RR 3.75, CI 1.04–13.51 for those diagnosed age 10–14 compared to before age 5) (data not shown). All treatment modality combinations except chemotherapy only resulted in a higher risk of morbidity compared to surgery only, although only those survivors experiencing all three therapy modalities (surgery, chemotherapy, and radiation) had a statistically elevated risk relative to those who had surgery only (RR 1.96, CI 1.04–3.72).

Approximately 41.0% of the survivors had at least one type of morbidity leading to hospitalization, 19.1% had 2 or more types, 10.1% had 3 or more types, and 4.7% had 4 or more types (Table 3). In contrast, only 17% of the comparison group had any hospital-related morbidity, and only 6.3% had multiple types of hospital-related morbidity. CNS tumor survivors were at highest excess risk of developing multiple late conditions leading to hospitalization (22.8% had more than one type of condition).

Table 3. Hospital-related late morbidity in survivors and comparison group by number of types of admission diagnoses1
inline image

After adjusting for gender, SES, and region and urban/rural status of residence, the relative risk of hospital-related morbidity was significantly higher for survivors than for the comparison group (RR 4.1, CI 3.7–4.5); this was true also for all diagnostic groups of cancer survivors examined (Table 4; only overall, leukemia, lymphoma, CNS and bone/soft tissue sarcoma results shown). Those at highest increased risk were survivors of leukemia (RR 4.8, CI 4.0–5.8), CNS tumors (RR 4.8, CI 4.0–5.8), bone and soft tissue sarcomas (RR 4.9, CI 3.8–6.2), and kidney cancer (RR 4.9, CI 3.4–7.0).

Table 4. Relative risk of hospital-related late morbidity–survivors vs. comparison group
inline image

The most common types of late morbidity among survivors (according to ICD categories) were diseases of the digestive system [152 (11.1%) of all survivors], neoplasms including relapse or second cancer [137 (10.0%) of total survivors], and complications of pregnancy or birth [97 (15%) of total female survivors]. The relative risk of having any specific type of late morbidity was consistently higher among survivors than in the comparison group (RRs ranged from 2.2 to 21.7, p < 0.05), except for pregnancy/birth complications (RR 1.1, CI 0.9,1.3). The types of late morbidity with highest excess risk were neoplasms (RR 21.7, CI 16.3, 28.7), blood disorders (RR 10.5, 4.9, 22.2), diseases of the nervous system (RR 8.6, 6.4, 11.6), endocrine, nutritional and metabolic disorders (RR 7.9, 4.2, 14.7), and infective and parasitic diseases (RR 7.1, 4.8, 10.6). Leukemia survivors had particularly escalated relative risk for infective and parasitic diseases, neoplasms, and disease of the blood (RRs 11.4, 36.8, and 19.6, respectively); survivors of CNS tumors had high excess risks for neoplasms, endocrine, nutritional and metabolic disorders, diseases of the nervous system, and diseases of the circulatory system (RRs 33.6, 10.6, 17.8, and 8.7, respectively); and survivors of bone tumors and soft tissue sarcomas had high risks for subsequent neoplasms (RR 36.1, CI 21.9, 59.3), although numbers were small.

Survivors with any combination of treatment types had higher risks of overall hospital-related late morbidity than the population sample (Table 5), both overall and for survivors of leukemia, lymphoma, CNS and bone/soft tissue sarcoma tumors. The highest excess risk was found among survivors who had undergone all three modalities of treatment (chemotherapy, radiation, and surgery) for their original cancer (RR 7.1, CI 5.5, 9.0). Relative risk of hospitalization for infective and parasitic diseases, diseases of the digestive system, and musculoskeletal conditions, were significantly elevated for all treatment groups examined (range RR 4.2–12.9 for infections; range 2.6–7.9 for digestive system disorders; range 2.2–8.3 for musculoskeletal system problems) (data not shown).

Table 5. Relative risk of hospital-related late morbidity–survivors vs. comparison group by treatment
inline image

As shown in Figure 1, cumulative incidence of late hospital-related morbidity was significantly higher for survivors than the population sample, and increased over time (overall cumulative incidence for survivors 65%, vs. 26% for the population sample, 20 years postdiagnosis), although the rate of increase appears to decline with time. This pattern was also observed for both late-occurring neoplasms admitted to hospital (Fig. 1) (cumulative incidence for survivors 17%, vs. 1% for the population sample, 20 years postdiagnosis) and late hospital-related morbidity other than neoplasms (Fig. 1) (cumulative incidence for survivors 62%, vs. 26% for the population sample, 20 years postdiagnosis). Notably, late effects other than neoplasms accounted for most (95%, i.e., 62%/65%) of the morbidity leading to hospitalization at 20 years postdiagnosis.

thumbnail image

Figure 1. Hospital-related morbidity for childhood cancer survivors vs comparison group.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Our study examines inpatient, hospital-related late morbidity among a geographically-defined cohort of survivors of a cancer or tumor diagnosed before age 20 years. The results indicate that survivors of childhood cancer are burdened with an excessive risk of morbidity leading to hospitalization after 5 years of survival since diagnosis, as compared to the general population; that survivors have more types of hospital-related morbidity than the general population; and that excess risk persists over time, in particular hospital-related morbidity other than subsequent neoplasms. All combinations of treatment modalities conferred an excess risk, and increased risk was seen for all disease categories except pregnancy and birth complications, although not all results were statistically significant.

Our study differed from previous assessments of late effects in limiting outcomes to those who were admitted to hospital as inpatients. However, for conditions that would require hospitalization, sociodemographic factors, and health system factors such as urban/rural and regional residential characteristics, that might affect access to care, did not affect risk of late morbidity in our study, implying that access to hospital care within the Canadian health care system, which is mandated to provide universal access to medically necessary care, was not a factor in ascertainment of morbidity.

The results of the Childhood Cancer Survivorship Study (CCSS),1 Dutch3 and our study are generally similar. The size of the risk of late morbidity leading to hospitalization observed in our study (41% after 15 years postsurvivorship) was slightly higher than for severe conditions reported from the CCSS (27.5% after 30 years) and the Dutch study (37%), and patterns of risks were somewhat different, with different types of problems identified. In contrast to the CCSS, which shows generally linear increases in cumulative incidence of late effects for many cancers, in our study, the rate of increase of cumulative incidence of hospital-related morbidity appears to decrease in later years. These differences may be related to study differences such as the diagnostic groups included in the study cohorts, differences in follow-up periods, types of late effects identified through different sources, relative accuracy and reliability of the different methods of identification of outcomes, and effects of methodologic differences in recruitment, participation, and retention of subjects.

Strengths of our study include ascertainment from a population registry rather than pediatric treatment centres, thereby identifying a total survivor population; inclusion of all diagnoses, thereby providing an assessment of late morbidity in a representative survivor group; record linkage to population-based administrative databases, thereby minimizing selection and participation bias and loss to follow-up; objective and comprehensive criteria (hospitalization) for identification of conditions; reporting from administrative data (rather than self-report) for improved accuracy and consistency of reporting; and availability of a comparison group, allowing for calculation of relative risks. However, subjects were followed for a shorter time than the CCSS1 or Dutch3, 22 studies, and included fewer survivors than the CCSS study. Our study also included cases who died in hospital.

Limitations of the study include exclusion of conditions not captured in inpatient hospitalization data, lack of specificity and additional information on diagnoses (although audits of Canadian hospital diagnosis data show that, at the reported level of ICD code, diagnoses are sufficiently accurate for population study23). Only survivors and comparison subjects active in the health insurance plan were included in the analysis. Approximately 13.4% (213/1587) of survivors did not link to the provincial health insurance plan registry. Likely explanations for nonlinkage include mobility out of province before the 5-year anniversary of diagnosis, inaccurate identifying data for linkage, or nonregistration (younger males and recent immigrants in particular may not have registered either through lack of interest or knowledge of the process). Differential mobility out-of-province may have affected estimates of morbidity. We found that cumulative loss to follow-up was low in both groups, (4% for survivors diagnosed from 1970 vs. 6% for the comparison subjects); the slightly higher loss to follow-up among the general population sample may lead to a slight overestimate of morbidity among survivors.

The results of our study support the need to continue to assess morbidity risks as survivors age, as evidence of long term health risks can inform the development of strategies for surveillance and treatment of long term survivors, and future demands on the health care system. We also need to investigate the effects of coexisting and future comorbidity on health risks as this population ages, and further explore changes in risk over time, in particular risks of late effects other than cancer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors gratefully acknowledge the BC Cancer Registry, BC Cancer Agency, BC Children's Hospital, the BC Ministry of Health Services, 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.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Oeffinger KC, Mertens AC, Sklar CA, Kawashima T, Hudson MM, Meadows AT, Friedman DL, Marina N, Hobbie W, Kadan-Lottick NS, Schwartz CL, Leisenring W, Robison LL. Chronic health conditions in adult survivors of childhood cancer. N Engl J Med 2006; 355: 157282.
  • 2
    Hewitt ME, Weiner SL, Simone JV; National Cancer Policy Board (US). Childhood cancer survivorship: improving care and quality of lifeed. Washington, DC: National Academies Press, 2003.
  • 3
    Geenen MM, Cardous-Ubbink MC, Kremer LC, van den Bos C, van der Pal HJ, Heinen RC, Jaspers MW, Koning CC, Oldenburger F, Langeveld NE, Hart AA, Bakker PJ, Caron HN, van Leeuwen FE. Medical assessment of adverse health outcomes in long-term survivors of childhood cancer. JAMA 2007; 297: 270515.
  • 4
    Curry HL, Parkes SE, Powell JE, Mann JR. Caring for survivors of childhood cancers: the size of the problem. Eur J Cancer 2006; 42: 5018.
  • 5
    Garrè M, Gandus S, Cesana B, Haupt R, De Bernardi B, Comelli A, Ferrando A, Stella G, Vitali M, Picco P. Health status of long-term survivors after cancer in childhood; a mulit-institutional study in Italy. Am J Pediatr Hematol Oncol 1994; 16: 14352.
  • 6
    Von der Weid N, Beck D, Caflisch U, Feldges A, Wyss M, Wagner H. Standardized assessment of late effects in long-term survivors of childhood cancer in Switzerland: results of a Swiss Pediatric Oncology Group (SPOG) pilot study. Int J Pediatr Hematol Oncol 1996; 3: 48390.
  • 7
    Stevens MC, Mahler H, Parkes S. The health status of adult survivors of cancer in childhood. Eur J Cancer 1998; 34: 6948.
  • 8
    Sklar CA. Overview of the effects of cancer therapies: the nature, scale and breadth of the problem. Acta Paediatrica Suppl 1999; 88: 14.
  • 9
    Oeffinger KC, Eshelman DA, Tomlinson GE, Buchanan GR, Foster BM. Grading of late effects in young adult survivors of childhood cancer followed in an ambulatory adult setting. Cancer 2000; 88: 168795.
  • 10
    Humpl T, Fritsche M, Bartels U, Gutjahr P. Survivors of childhood cancer for more than twenty years. Acta Oncol 2001; 40: 449.
  • 11
    Hudson MM, Mertens AC, Yasui Y, Hobbie W, Chen H, Gurney JG, Yeazel M, Recklitis CJ, Marina N, Robison LR, Oeffinger KC. Health status of adult long-term survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. JAMA 2003; 290: 158392.
  • 12
    Monteleone PM. Late effects of childhood cancer and treatment. Available at http://www.emedicine.com/ped/topic2591.htm, Accessed on June 2006.
  • 13
    Oeffinger KC, Hudson MM. Long-term complications following childhood and adolescent cancer: foundations for providing risk-based health care for survivors. CA Cancer J Clin 2004; 54: 20836.
  • 14
    McBride ML, Rogers PC, Sheps SB, Glickman V, Broemeling AM, Goddard KJ, Hu J, Lorenzi M, Peacock S, Pritchard S, Rassekh SR, Siegel L, Spinelli JJ, Teckle P, Xie L. Childhood, adolescent, and young adult cancer survivor research program of british columbia: objectives, study design, and cohort characteristics. Pediatr Blood Cancer 2010; 55: 32430.
  • 15
    Steliarova-Foucher E, Stiller C, Lacour B, Kaatsch P. International classification of childhood cancer. Cancer 2005; 103: 145767.
  • 16
    Kramarova E, Stiller CA. The international classification of childhood cancer. Int J Cancer 1996; 68: 75965.
  • 17
    National Cancer Institute (US), Common Terminology Criteria for Adverse Events, v3.0 (CTCAE). National Cancer Institute, 2003.
  • 18
    World Health Organization. International statistical classification of diseases and related health problems, Geneva, 1992.
  • 19
    Rambeau S, Todd K, Census metropolitan area and census agglomeration influenced zones (MIZ) with census data. Geography Division, Statistics Canada 2000.
  • 20
    Wilkins R, PCCF+ Version 3J User's Guide (Geocodes/PCCF). Automated geographic coding based on the statistics canada postal code conversion files, including postal codes to May 2002. Health Analysis and Measurement Group, Statistics Canada, Ottawa, 2002.
  • 21
    R Development Core Team. R: a language and environment for statistical computing., ed. 1.8.1 Vienna, Austria: R Foundation for Statistical Computing, 2005.
  • 22
    Van Leeuwen FE, Cardous-Ubbink MC, van den Bos C, van der Pal HJ, Kremer LC, Koning CC, Oldenburger F, Bakker PJ, Caron HN. Overall assessment of late effects in a Dutch cohort of long-term survivors of childhood cancer. In: Proceedings of the 9th International Conference: Long-term Complications of Treatment of Children and Adolescents for Cancer. 2006. Niagara-on-the-Lake, ON, Canada: Roswell Park Cancer Institute.
  • 23
    Williams J, Young W. Inventory of studies on the accuracy of Canadian health administrative databases. Institute for Clinical Evaluative Sciences, 1996.