Abstract
Purpose:
Children with cancer are frequently hospitalized. However, hospitalization and death by disease category are not well defined <5 years from diagnosis.
Methods:
We conducted a retrospective cohort study using linked cancer registry-hospital discharge-vital records to identify cancer cases <20 years at diagnosis during 1987-2012 (n = 4,567) and comparison children without cancer, matched on birth year and sex (n = 45,582). Data linkage identified serious morbidities resulting in cancer- and non-cancer-related hospitalizations or deaths <5 years from diagnosis. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated to compare relative hospitalization and mortality by disease category and after excluding cancer-related outcomes. Among cancer cases, relative risks (RRs) of these outcomes for children with solid tumors compared with children with leukemia/lymphoma were also estimated.
Results:
Greater rates of all-cause hospitalization (281.5/1,000 vs. 6.2/1,000 person-years) and death (40.7/1,000 vs. 0.15/1,000 person-years) were observed in childhood cancer cases than comparators and across all diagnosis categories. Increased hospitalization (31.0/1,000 vs. 6.2/1,000 person-years; HR 5.0, 95% CI 4.5-5.5) and death (1.0/1,000 vs. 0.15/1,000 person-years; HR 10.4, 95% CI 5.6-19.1) rates remained when cancer-related outcomes were excluded. Although HRs for hospitalization and death did not differ greatly by treatment era, absolute rates of hospitalization were greater (1987-1999: 233.3/1,000; 2000-2012: 320.0/1,000 person-years) and death were lesser (1987-1999: 46.3/1,000; 2000-2012: 36.8/1,000 person-years) in the later treatment era among cases. Children with solid tumors were less likely to have a cancer-related hospitalization than were those with leukemia/lymphoma (RR 0.91, 95% CI 0.84-0.98).
Conclusion:
Even after excluding cancer-related diagnoses, children with cancer experience greater rates of hospitalization and death in all disease categories. Results may guide future toxicity mitigation initiatives and inform anticipatory guidance for families of children with cancer.
Keywords: Population-based study, cohort study, pediatric, cancer, hospitalization, mortality
Introduction
Cancer is the leading cause of disease-related death among children in the United States.[1] Children with cancer are frequently hospitalized for cancer-related treatment and management of treatment-related complications.[2] They may also experience health conditions (e.g., asthma, mental health disorders, injury, etc.) that cause hospitalizations among children without cancer.[3-5] Serious morbidities resulting in hospitalization, especially those that are not treatment-related, are not well described in children with cancer <5 years after diagnosis.
Hospitalization and mortality outcome data in children with cancer are largely informed by clinical trials.[2] However, <50% of children with cancer are treated on a clinical trial at time of diagnosis.[6, 7] Additionally, there are systematic gaps in inclusion on consortium-wide cancer clinical trials based on race, age, cancer diagnosis, and presence of comorbidities.[6-9] Among 5+ year childhood cancer survivors, population-based studies report greater rates of hospitalization and all-cause mortality compared to those without a history of cancer.[10-13] Population-based studies that include individuals <5 years from diagnosis are primarily limited to a single health outcome (i.e. infection-related mortality, early death), a specific malignancy, or a single-institution.[14-24]
Many hospitalizations following a childhood cancer diagnosis are treatment-related.[2] Improved knowledge of morbidity and mortality outcomes across cancer diagnoses may better guide clinicians in determining risk for adverse outcomes, treatment-related and otherwise, and provide insight regarding the overall burden experienced by families. We used population-based linked cancer-hospital discharge-vital registry data from Washington State to compare levels of hospitalization and mortality among childhood cancer cases <5 years from diagnosis to healthy children. We hypothesized that children <5 years from a cancer diagnosis have greater rates of hospitalization and mortality relative to similarly-aged children without cancer and that patterns in morbidity and mortality differ between cases versus comparators, even after excluding cancer treatment-related hospitalizations. We also explored variation in hospitalization indication among children with different types of cancer.
Methods
Subject Identification
Institutional Review Board approvals with waiver of consent were granted by the Washington State Department of Health and the Fred Hutchinson Cancer Research Center. This retrospective cohort study used population-based cancer registry data linked to birth (1974-2012), hospital discharge (1987-2013), and death (1987-2013) records from Washington State.
As part of a larger study of long term effects of cancer treatment,[10] all cancer cases <20 years old at diagnosis during 1974-2015 were identified from two Washington State population-based cancer registries: 1) National Cancer Institute’s Surveillance, Epidemiology and End Results Program-funded Cancer Surveillance System (13 counties surrounding Puget Sound), starting in 1974; and 2) Centers for Disease Control and Prevention’s National Program of Central Cancer Registries-funded Washington State Cancer Registry covering the entire state, starting in 1995. Cancer registry data include cancer type classified per the International Classification of Childhood Cancer (ICCC), third edition.[25] Cancer registry data additionally include International Classification for Oncology (ICD-O) morphology and topography codes; histology; stage (when applicable); diagnosis age, diagnosis date, initial therapy (chemotherapy, radiotherapy, surgery, etc.), and vital status at quarterly follow-up. Both cancer registries undergo comprehensive quality control assessment of completeness and accuracy.[26-28]
Cancer registry data were linked to Washington State birth records to identify all cancer cases <20 years at diagnosis between 1974-2015, and who were born in-state (N=6,320). This not only provided additional information about the cases but allowed identification of appropriate population-based comparison children for analyses. Birth records include date of birth, race/ethnicity, and sex as well as maternal (age, health conditions, education, prenatal smoking) and infant variables (birthweight, gestational age, malformations) that may be associated with childhood cancer risk and outcomes. Cases with unknown sex information were excluded (n = 16).
Cases were assigned a “reference date” defined as the date of cancer diagnosis. Cases with a non-malignant disease (primarily nonmalignant brain tumors, cervical carcinoma in situ, or skin cancers) were excluded (n = 605). For each case, 10 comparison subjects were selected from birth records, matched on birth year and sex. Comparators were assigned a reference date equal to their cases’ diagnosis date and those who were known to have died before that date (n = 88) were excluded, based on linkage to death certificates.
Since 1987, birth records have been routinely linked to hospital discharge data for the delivery hospitalizations of mother and infant, enriching subject records with information about possible co-morbidities and infant health conditions (i.e. congenital malformations were identified if indicated by birth record and/or presence of relevant ICD9 codes 740-759 in the hospital discharge record for the delivery hospitalization, as screening both sources has been shown to improve identification of many conditions).[29] We then linked the children’s records to subsequent hospital discharge records from 1987-2013 to identify all non-pregnancy-related outcomes (i.e. discharge diagnosis, hospitalization duration). Because hospital discharge data were only available from 1987-2013, cases diagnosed before 1987 were excluded (n = 416). To allow at least one year of follow-up, 716 cases diagnosed after December 2012 were excluded. Finally, data were linked to death records from 1987-2013 to identify deaths occurring in both groups. In total, 4,567 childhood cancer cases were identified and included in the analyses with 45,582 comparators.
Assessment of Outcomes
Our primary outcomes were hospitalization and death within 5 years after the diagnosis/reference date. Washington State hospital discharge data included all inpatient and observation discharges in non-Federal facilities. Multiple International Classification of Disease codes, version 9 (ICD9) were available for each discharge, using Medicare-Medicaid billing standards. Up to 25 discharge diagnoses were screened for each hospital discharge. Data quality activities by the Washington State Department of Health include tracking and verifying records monthly and follow-up on delinquent reports. Hospital discharge records for 5 years after the reference date were linked to subjects’ records to identify the occurrence of all subsequent hospitalizations. Hospitalizations for pregnancy-related conditions were excluded (ICD9 630-679, 760-779). Subjects’ records were also linked to State death records (maintained in partnership with the National Center for Health Statistics and the State Department of Health) to identify one underlying primary and up to 10 contributing causes of death, based on ICD9 and ICD10 codes. A diagnosis code within the applicable range for any field for any hospitalization indicated presence of the outcome. First hospitalization on record and death outcomes were evaluated overall and by ICD9/ICD10 diagnosis category, as previously described.[10, 30]
The Clinical Classification Software (CCS) program (adapted to incorporate ICD10 codes for deaths) was used to dichotomize outcomes as “cancer-related” (CCS diagnostic groups 11-43) or “non-cancer-related” (all other CCS diagnostic groups).[2, 31, 32] “Cancer-related” hospitalizations were further categorized into hierarchical mutually exclusive hospitalization indication groups: chemotherapy, procedure, infection, toxicity, or other (Table S1) using the algorithm developed by Russell et al.[2] Information regarding timing of chemotherapy and procedures within our dataset was not available: If the primary diagnosis and primary procedure fields were coded chemotherapy, then the hospitalization indication was coded “chemotherapy.” If both the primary diagnosis and procedure field did not indicate chemotherapy, and a cancer-related procedure code was used in the primary diagnosis or primary procedure fields, then the hospitalization indication was coded as a cancer-related “procedure.”
Statistical Analyses
Follow-up for both cancer cases and comparators accrued from diagnosis/reference date through whichever of the following came first: December 2013, death, or reference date plus five years minus one day.[10, 33] Race/ethnicity data were missing for 2% of subjects in each group. Levels of missing data for other variables were similar in cases and comparators. Incidence rates for outcomes were estimated per 1,000 person-years. Hazard ratios (HR) and 95% confidence intervals (CI) were estimated using Cox regression overall and by ICD9/ICD10 diagnosis category. HRs accounted for matching variables (birth year, sex) via baseline hazard stratification. Results were examined separately by gestational length (<37/37+ weeks), since prematurity is associated with childhood mortality,[34] by diagnosis age (<1 year/1+ year), to compare with prior studies,[17, 35], and time since reference date (<1 year, 1 - <3 years, 3 - <5 years), to review patterns over time. We also stratified by treatment era (1987-1999/2000-2012), based on mortality trends,[36] and by race (white/non-white) based on racial health outcome disparities in pediatric oncology;[37-39] further refinement was not possible due to small numbers. Important differences in risk estimates between strata were determined using the likelihood ratio test. Sensitivity analyses were conducted using only comparators with a history of hospitalization, due to possible differential loss to follow-up between cases and comparators. Because the Russell algorithm classified injuries as non-cancer-related, ICD9/10 injury/poisoning codes were manually reviewed and further dichotomized: probably/possibly cancer-related (e.g. complication of bone marrow transplant; transfusion reaction) vs. unlikely cancer-related. (Table S1). Relative risk for “treatment-related injuries/poisonings” and “non-treatment-related injuries/poisonings” was estimated in cases versus comparators. Results based on cell sizes <5 were suppressed and cells with <5 subjects masked. Analyses were conducted overall and repeated after exclusion of cancer-related outcomes as detailed above.
In case-only analyses, we described malignancy-specific morbidity and mortality outcomes, based on ICCC group: leukemia/lymphoma (ICCC I/II) and CNS/solid tumor (all others). Non-cancer-related diagnoses were then excluded and hospitalizations were categorized by indication (chemotherapy, procedure, infection, toxicity, other).[2] Number of hospitalizations, elapsed time to first hospitalization, and length of first hospitalization by indication were summarized. Poisson regression with robust standard errors was used to estimate relative risk (RR) and 95% CIs for each outcome (reference = leukemia/lymphoma). Median time to hospitalization and length of hospital stay (and interquartile ranges [IQRs]) were calculated among hospitalized cases. Analyses were conducted using Stata version 16 (StataCorp., College Station, TX).
Results
Cases and comparison children
A slight majority (52%) of cases were male and 80% of cases and comparators were White (Table 1). Leukemia (26%), CNS tumor (21%), and lymphoma (13%) were the most common malignancies. Most cases were treated initially with chemotherapy (69%), 52% underwent surgery, and 27% received radiation therapy. Median age at end of follow-up for cases was 11.0 years (range: 0 days – 24.0 years) and for comparators was 12.0 years (range: 3 days – 24.0 years). During the follow-up period, 55% of cases and 3% of comparators were hospitalized; 16% of cases and 0.1% of comparators died. Maternal age, gravidity, parity, and education were similar between cases and comparators (Table 2). A greater proportion of cases weighed >4000 grams at birth (16% versus 13%) and had congenital malformations (8% versus 5%). Premature birth was similar between cases and comparators (8% versus 7%).
Table 1.
Characteristics of childhood cancer patients diagnosed in 1987-2012 and comparison children.
| Characteristic | Cases n = 4,567 | Comparison children n = 45,582 | ||
|---|---|---|---|---|
| n | % | n | % | |
| Sex | ||||
| Female | 2,198 | 48.1 | 21,973 | 48.2 |
| Male | 2,369 | 51.9 | 23,609 | 51.8 |
| Race/ethnicity | ||||
| White | 3,720 | 82.7 | 35,848 | 80.1 |
| Black | 137 | 3.0 | 1,666 | 3.7 |
| Hispanic | 304 | 6.7 | 3,378 | 7.6 |
| Asian | 254 | 5.6 | 2,656 | 5.9 |
| Native American | 72 | 1.6 | 975 | 2.2 |
| Pacific Islander/Other | 10 | 0.22 | 224 | 0.5 |
| Age (years) at diagnosis/reference date | ||||
| 0-3 | 1,526 | 33.4 | 15,247 | 33.4 |
| 4-9 | 1,075 | 23.5 | 10,734 | 23.5 |
| 10-14 | 790 | 17.3 | 7,885 | 17.3 |
| 15-19 | 1,176 | 25.8 | 11,716 | 25.7 |
| Age (years) at end of follow-up | ||||
| 0-4 | 363 | 7.9 | 1,577 | 3.5 |
| 5-9 | 1,606 | 35.2 | 17,178 | 37.7 |
| 10-14 | 754 | 16.5 | 7,687 | 16.9 |
| 15-19 | 902 | 19.7 | 8,485 | 18.6 |
| 20-24 | 944 | 20.6 | 10,665 | 23.4 |
| Hospitalized during follow-up | ||||
| Yes | 2,503 | 54.8 | 1,282 | 2.8 |
| No | 2,064 | 45.2 | 44,300 | 97.2 |
| Died | ||||
| Yes | 754 | 16.5 | 32 | 0.1 |
| No | 3,812 | 83.5 | 45,550 | 99.9 |
| Cancer type | ||||
| Leukemia | 1,208 | 26.4 | ||
| Lymphoma | 600 | 13.1 | ||
| CNS | 961 | 21.0 | ||
| Neuroblastoma | 312 | 6.8 | ||
| Retinoblastoma | 101 | 2.2 | ||
| Renal tumor | 213 | 6.7 | ||
| Hepatic tumor | 62 | 1.4 | ||
| Bone sarcoma | 194 | 4.2 | ||
| Soft tissue sarcoma | 316 | 6.9 | ||
| Germ cell tumors | 207 | 4.5 | ||
| Other malignant epithelial | 372 | 8.1 | ||
| Other unspecified | 21 | 0.5 | ||
| Initial course of therapya | ||||
| Chemotherapy | 2,667 | 68.7 | ||
| Radiation | 1,228 | 27.1 | ||
| Surgery | 2,092 | 52.3 | ||
| None indicated | 950 | 24.6 | ||
| Treatment era | ||||
| 1987-1999 | 1,778 | 38.9 | ||
| 2000-2012 | 2,789 | 61.1 | ||
As indicated in cancer registry. Children may have received > 1 therapy
Table 2.
Parental, delivery, and infant characteristics of childhood cancer patients diagnosed in 1987-2012 and comparison children.
| Characteristic | Cases n = 4,567 | Comparison children n = 45,582 | ||
|---|---|---|---|---|
| n | % | n | % | |
| Maternal (years) age at birth | ||||
| 12-19 | 381 | 8.4 | 4,771 | 10.5 |
| 20-24 | 1,163 | 25.5 | 12,166 | 26.8 |
| 25-29 | 1,392 | 30.5 | 13,717 | 30.2 |
| 30-34 | 1,034 | 22.7 | 9,970 | 21.9 |
| 35-39 | 503 | 11.0 | 4,031 | 8.9 |
| 40+ | 85 | 1.9 | 789 | 1.7 |
| Maternal educationa | ||||
| Less than High School | 379 | 16.5 | 4,036 | 17.7 |
| High School graduate | 674 | 29.4 | 6,912 | 30.3 |
| At least some college | 1,239 | 54.0 | 11,884 | 52.0 |
| Marital status | ||||
| Married | 3,528 | 77.8 | 34,404 | 76.0 |
| Single | 1,008 | 22.2 | 10,882 | 24.0 |
| Paternal (years) age at birth | ||||
| 12-19 | 150 | 3.5 | 1,451 | 3.4 |
| 20-24 | 670 | 15.7 | 8,145 | 19.3 |
| 25-29 | 1,253 | 29.3 | 12,458 | 29.5 |
| 30-34 | 1,218 | 28.5 | 10,907 | 25.9 |
| 35-39 | 636 | 14.9 | 6,101 | 14.5 |
| 40+ | 346 | 8.1 | 3,091 | 7.3 |
| Type of delivery | ||||
| Vaginal | 3,269 | 77.6 | 33,157 | 78.9 |
| Cesarean section | 946 | 22.4 | 8,887 | 21.1 |
| Number of prior pregnancies | ||||
| 0 | 1,354 | 32.6 | 13,139 | 31.7 |
| 1 | 1,180 | 28.4 | 11,949 | 28.8 |
| 2 | 780 | 18.8 | 7,736 | 18.6 |
| 3+ | 836 | 20.1 | 8,644 | 20.8 |
| Number of prior births | ||||
| 0 | 1,936 | 42.4 | 18,993 | 42.2 |
| 1 | 1,468 | 32.1 | 14,525 | 32.3 |
| 2 | 709 | 15.5 | 6,889 | 15.3 |
| 3+ | 392 | 8.7 | 4,546 | 10.1 |
| Prenatal smokingb | ||||
| No | 3,109 | 84.6 | 30,484 | 83.1 |
| Yes | 564 | 15.4 | 6,183 | 16.9 |
| Child birthweight (grams) | ||||
| 250-2499 | 243 | 5.3 | 2,377 | 5.2 |
| 2500-3999 | 3,584 | 78.7 | 37,000 | 81.5 |
| 4000+ | 728 | 16.0 | 6,000 | 13.2 |
| Gestational length (weeks) | ||||
| 20-36 | 366 | 8.3 | 3,275 | 7.5 |
| 37-41 | 3,767 | 85.2 | 37,658 | 85.8 |
| 42-45 | 290 | 6.6 | 2,978 | 6.8 |
| Child congenital malformationb | ||||
| No | 3,464 | 92.4 | 35,475 | 94.7 |
| Yes | 283 | 7.6 | 1,979 | 5.3 |
Data only available for children born 1992 and later
Data only available for children born 1984 and later
Hospitalization and death
Cases had a hospitalization rate of 281.5/1,000 person-years (comparators: 6.2/1,000 person-years) and a death rate of 40.7/1,000 person-years (comparators: 0.15/1,000 person-years) (Table 3). Relative to comparators, cases were nearly 50 times more likely to be hospitalized (HR 49.5, 95% CI: 45.0-54.5) and >300 times more likely to die (HR 372.9, 95% CI: 239.2-581.3). Cases had increased HRs across all cause-specific outcome categories. Cases were nearly 10 times more likely to have a mental health disorder-related outcome than comparators (HR 9.7, 95% CI: 7.1-13.1) and >40 times more likely to have an injury/poisoning-related outcome (HR 46.1, 95% CI: 39.8-53.5).
Table 3.
Hospitalization, mortality, and hospitalization/death for selected conditions in childhood cancer cases and comparison children.
| Person-years at risk | n | Rate per 1,000 | HR (95% CI)a | |
|---|---|---|---|---|
| Any Hospitalization | ||||
| Comparison | 205,189 | 1,282 | 6.2 | 1.0 (ref) |
| Case | 8,892 | 2,503 | 281.5 | 49.5 (45.0-54.5) |
| Death (any cause) | ||||
| Comparison | 208,850 | 32 | 0.15 | 1.0 (ref) |
| Case | 18,525 | 754 | 40.7 | 372.9 (239.2-581.3) |
| Cause-specific hospitalization/death | ||||
| Infectious | ||||
| Comparison | 208,141 | 230 | 1.1 | 1.0 (ref) |
| Case | 14,414 | 1,241 | 86.1 | 77.8 (65.9-91.7) |
| Cancer | ||||
| Comparison | 208,842 | 7 | 0.03 | 1.0 (ref) |
| Case | 18,388 | 97 | 5.3 | 159.6 (70.0-364.0) |
| Endocrine/metabolic | ||||
| Comparison | 207,955 | 313 | 1.5 | 1.0 (ref) |
| Case | 15,600 | 969 | 62.1 | 40.7 (35.3-46.9) |
| Hematologica | ||||
| Comparison | 208,589 | 96 | 0.5 | 1.0 (ref) |
| Case | 13,212 | 1,496 | 113.2 | 244.8 (189.5-316.2) |
| Mental health disorder | ||||
| Comparison | 208,661 | 98 | 0.5 | 1.0 (ref) |
| Case | 18,298 | 81 | 4.4 | 9.7 (7.1-13.1) |
| Nervous system | ||||
| Comparison | 208,236 | 217 | 1.0 | 1.0 (ref) |
| Case | 15,365 | 989 | 64.4 | 63.2 (53.4-74.8) |
| Circulatory system | ||||
| Comparison | 208,759 | 47 | 0.2 | 1.0 (ref) |
| Case | 16,692 | 692 | 41.4 | 186.4 (133.9-259.5) |
| Respiratory system | ||||
| Comparison | 207,547 | 427 | 2.0 | 1.0 (ref) |
| Case | 15,148 | 1,127 | 74.4 | 36.7 (32.3-41.7) |
| Digestive system | ||||
| Comparison | 208,066 | 293 | 1.4 | 1.0 (ref) |
| Case | 14,627 | 1,172 | 80.1 | 54.8 (47.4-63.4) |
| Genitourinary system | ||||
| Comparison | 208,503 | 126 | 0.6 | 1.0 (ref) |
| Case | 17,106 | 482 | 28.2 | 47.5 (38.3-58.8) |
| Congenital | ||||
| Comparison | 208,560 | 99 | 0.5 | 1.0 (ref) |
| Case | 17,879 | 194 | 10.8 | 21.7 (16.9-28.0) |
| Skin | ||||
| Comparison | 208,589 | 96 | 0.5 | 1.0 (ref) |
| Case | 16,807 | 539 | 32.1 | 74.6 (58.3-95.4) |
| Musculoskeletal | ||||
| Comparison | 208,497 | 191 | 0.9 | 1.0 (ref) |
| Case | 17,184 | 447 | 26.0 | 29.5 (24.5-35.5) |
| Injuriesb | ||||
| Comparison | 208,117 | 297 | 1.4 | 1.0 (ref) |
| Case | 15,399 | 987 | 64.1 | 46.1 (39.8-53.5) |
| Fracture | ||||
| Comparison | 208,591 | 104 | 0.5 | 1.0 (ref) |
| Case | 18,425 | 40 | 2.2 | 4.6 (3.1-6.6) |
Accounting for sex and birth year.
Includes fractures.
When cancer-related hospitalizations/deaths were excluded, cases were still 5 times more likely to have any hospitalizations (HR 5.0, 95% CI: 4.5-5.5) and 10 times more likely to die (HR 10.4, 95% CI: 5.6-19.1) (Table 4), and had increased HRs across all cause-specific outcomes except fracture. The RRs for cancer treatment-related and non-treatment-related injuries were 13.5 (95% CI 11.9-15.4) and 1.19 (95% CI 1.05-1.35) respectively (Table S2). When non-cancer-related hospitalizations/deaths were examined, cases had increased HRs across all cause-specific outcomes (Table 5). Differences in hospitalization and death outcomes were greatest in the year after the reference date and decreased over time. However, by 3 - <5 years from the reference date, cases remained more likely to be hospitalized or die with infection, endocrine, hematologic, nervous, circulatory, or injury diagnoses.
Table 4.
Hospitalization, mortality, and hospitalization/death for selected conditions in childhood cancer cases, among non-cancer-related diagnoses.
| Person-years at risk | n | Rate per 1,000 | HR (95% CI)a | |
|---|---|---|---|---|
| Any Hospitalization | ||||
| Comparison | 205,197 | 1,277 | 6.2 | 1.0 (ref) |
| Case | 16,715 | 519 | 31.0 | 5.0 (4.5-5.5) |
| Death (any cause) | ||||
| Comparison | 208,450 | 32 | 0.15 | 1.0 (ref) |
| Case | 18,525 | 21 | 1.0 | 10.4 (5.6-19.1) |
| Cause-specific hospitalization/death | ||||
| Infectious | ||||
| Comparison | 208,141 | 230 | 1.1 | 1.0 (ref) |
| Case | 18,263 | 86 | 4.7 | 4.2 (3.3-5.4) |
| Endocrine/metabolic | ||||
| Comparison | 207,958 | 311 | 1.5 | 1.0 (ref) |
| Case | 18,228 | 113 | 6.2 | 4.1 (3.3-5.1) |
| Hematological | ||||
| Comparison | 208,590 | 95 | 0.5 | 1.0 (ref) |
| Case | 18,117 | 123 | 6.8 | 14.1 (10.7-18.6) |
| Mental health disorder | ||||
| Comparison | 208,662 | 97 | 0.5 | 1.0 (ref) |
| Case | 18,489 | 17 | 0.9 | 2.0 (1.2-3.4) |
| Nervous system | ||||
| Comparison | 208,236 | 216 | 1.0 | 1.0 (ref) |
| Case | 18,044 | 148 | 8.2 | 8.0 (6.4-9.9) |
| Circulatory system | ||||
| Comparison | 208,760 | 47 | 0.2 | 1.0 (ref) |
| Case | 18,358 | 62 | 3.4 | 15.1 (10.2-22.4) |
| Respiratory system | ||||
| Comparison | 207,547 | 426 | 2.0 | 1.0 (ref) |
| Case | 18,250 | 96 | 5.2 | 2.5 (2.0-3.1) |
| Digestive system | ||||
| Comparison | 208,069 | 292 | 1.4 | 1.0 (ref) |
| Case | 18,226 | 93 | 5.1 | 3.6 (2.8-4.5) |
| Genitourinary system | ||||
| Comparison | 208,503 | 125 | 0.6 | 1.0 (ref) |
| Case | 18,390 | 46 | 2.5 | 4.3 (3.1-6.1) |
| Congenital | ||||
| Comparison | 208,560 | 99 | 0.5 | 1.0 (ref) |
| Case | 18,378 | 46 | 2.5 | 5.1 (3.6-7.3) |
| Skin | ||||
| Comparison | 208,589 | 96 | 0.5 | 1.0 (ref) |
| Case | 18,462 | 27 | 1.5 | 3.4 (2.2-5.2) |
| Musculoskeletal | ||||
| Comparison | 208,397 | 191 | 0.9 | 1.0 (ref) |
| Case | 18,353 | 55 | 3.0 | 3.4 (2.5-4.6) |
| Injuriesb | ||||
| Comparison | 208,117 | 297 | 1.3 | 1.0 (ref) |
| Case | 18,254 | 95 | 5.2 | 3.7 (3.0-4.8) |
| Fracture | ||||
| Comparison | 208,591 | 104 | 0.5 | 1.0 (ref) |
| Case | 18,499 | 8 | 0.43 | 0.9 (0.4-1.9) |
Accounting for sex and birth year.
Includes fractures.
Table 5.
Hazard ratios for non-pregnancy hospitalizations/death in childhood cancer cases relative to children without cancer, by post-diagnosis period, among non-cancer-related diagnoses.
| First year | 1 to < 3 years | 3 to < 5 years | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Person-years at risk | N | Rate/1,000 | HR (95% CI)a | Person-years at risk | N | Rate/1,000 | HR (95% CI)a | Person-years at risk | N | Rate/1,000 | HR (95% CI)a | |
| Any hospitalization | ||||||||||||
| Comparison | 45206 | 453 | 9.98 | 1.0 (ref) | 85681 | 522 | 6.08 | 1.0 (ref) | 76386 | 406 | 5.29 | 1.0 (ref) |
| Case | 4033 | 451 | 109.59 | 10.7 (9.4-12.2) | 7578 | 75 | 9.90 | 1.6 (1.3-2.1) | 6376 | 44 | 6.90 | 1.3 (1.0-1.8) |
| Death (any cause) | ||||||||||||
| Comparison | 45447 | 19 | 0.42 | 1.0 (ref) | 86259 | >5 | . | . | 76810 | <5 | . | . |
| Case | 4410 | 9 | 2.04 | 4.8 (2.2-10.7) | 7663 | <5 | . | . | 6427 | >5 | . | . |
| Cause-specific hospitalization/death | ||||||||||||
| Infection | ||||||||||||
| Comparison | 45389 | 102 | 2.23 | 1.0 (ref) | 86180 | 78 | 0.89 | 1.0 (ref) | 76749 | 57 | 0.74 | 1.0 (ref) |
| Case | 4367 | 64 | 14.66 | 6.5 (4.8-8.9) | 7643 | 18 | 2.36 | 2.6 (1.6-4.4) | 6415 | 17 | 2.65 | 3.6 (2.1-6.1) |
| Endocrine/metabolic | ||||||||||||
| Comparison | 45388 | 110 | 2.42 | 1.0 (ref) | 86126 | 127 | 1.46 | 1.0 (ref) | 76709 | 91 | 1.17 | 1.0 (ref) |
| Case | 4346 | 89 | 20.48 | 8.4 (6.3-11.1) | 7646 | 18 | 2.35 | 1.6 (1.0-2.6) | 6407 | 17 | 2.65 | 2.3 (1.3-3.8) |
| Hematological | ||||||||||||
| Comparison | 45429 | 40 | 0.86 | 1.0 (ref) | 86222 | 33 | 0.38 | 1.0 (ref) | 76783 | 29 | 0.38 | 1.0 (ref) |
| Case | 4322 | 114 | 25.22 | 28.8 (20.0-41.6) | 7654 | 12 | 1.57 | 4.1 (2.1-7.9) | 6422 | 7 | 1.09 | 2.9 (1.3-6.6) |
| Mental health | ||||||||||||
| Comparison | 45440 | 15 | 0.31 | 1.0 (ref) | 86222 | 35 | 0.41 | 1.0 (ref) | 76753 | 60 | 0.78 | 1.0 (ref) |
| Case | 4405 | 6 | 1.36 | 4.4 (1.7-11.5) | 7660 | 6 | 0.78 | 1.9 (0.8-4.6) | 6423 | 5 | 0.78 | 1.0 (0.4-2.5) |
| Nervous system | ||||||||||||
| Comparison | 45403 | 82 | 1.81 | 1.0 (ref) | 86171 | 84 | 0.97 | 1.0 (ref) | 76744 | 65 | 0.82 | 1.0 (ref) |
| Case | 4306 | 126 | 29.03 | 15.7 (11.9-20.8) | 7642 | 22 | 2.88 | 2.9 (1.8-4.7) | 6411 | 15 | 2.34 | 2.9 (1.6-5.0) |
| Circulatory system | ||||||||||||
| Comparison | 45442 | 15 | 0.33 | 1.0 (ref) | 86240 | 17 | 0.20 | 1.0 (ref) | 76795 | 18 | 0.23 | 1.0 (ref) |
| Case | 4378 | 46 | 10.28 | 30.7 (17.1-55.1) | 7651 | 13 | 1.70 | 8.5 (4.1-17.6) | 6422 | 7 | 1.09 | 4.7 (2.0-11.2) |
| Respiratory system | ||||||||||||
| Comparison | 45355 | 175 | 3.84 | 1.0 (ref) | 86068 | 166 | 1.93 | 1.0 (ref) | 76696 | 105 | 1.34 | 1.0 (ref) |
| Case | 4354 | 73 | 16.76 | 4.3 (3.3-5.7) | 7649 | 17 | 2.22 | 1.1 (0.7-1.9) | 6416 | 13 | 2.03 | 1.5 (0.8-2.7) |
| Digestive system | ||||||||||||
| Comparison | 45400 | 84 | 1.85 | 1.0 (ref) | 86125 | 118 | 1.37 | 1.0 (ref) | 76715 | 96 | 1.25 | 1.0 (ref) |
| Case | 4353 | 74 | 16.31 | 8.7 (6.3-11.9) | 7642 | 16 | 2.09 | 1.5 (0.9-2.6) | 6419 | 9 | 1.40 | 1.1 (0.6-2.2) |
| Genitourinary system | ||||||||||||
| Comparison | 45425 | 41 | 0.88 | 1.0 (ref) | 86203 | 54 | 0.63 | 1.0 (ref) | 76769 | 37 | 0.48 | 1.0 (ref) |
| Case | 4384 | 34 | 7.76 | 8.7 (5.5-13.8) | 7656 | 8 | 1.04 | 1.7 (0.8-3.5) | 6422 | 5 | 0.78 | 1.6 (0.6-4.1) |
| Congenital | ||||||||||||
| Comparison | 45422 | 45 | 0.99 | 1.0 (ref) | 86212 | 42 | 0.49 | 1.0 (ref) | 76780 | 28 | 0.36 | 1.0 (ref) |
| Case | 4381 | 39 | 8.67 | 8.7 (5.6-13.3) | 7652 | 8 | 1.05 | 2.1 (1.0-4.5) | 6422 | 5 | 0.78 | 2.1 (0.8-5.5) |
| Skin | ||||||||||||
| Comparison | 45432 | 33 | 0.73 | 1.0 (ref) | 86223 | 36 | 0.42 | 1.0 (ref) | 76779 | 29 | 0.36 | 1.0 (ref) |
| Case | 4398 | 17 | 3.87 | 5.3 (3.0-9.5) | 7658 | 5 | 0.65 | 1.6 (0.6-4.0) | 6425 | 5 | 0.78 | 2.1 (0.8-5.5) |
| Musculoskeletal | ||||||||||||
| Comparison | 45428 | 38 | 0.84 | 1.0 (ref) | 86172 | 87 | 1.01 | 1.0 (ref) | 76736 | 70 | 0.91 | 1.0 (ref) |
| Case | 4373 | 44 | 10.06 | 11.9 (7.7-18.3) | 7660 | 7 | 0.91 | 0.9 (0.4-2.0) | 6421 | 7 | 1.09 | 1.2 (0.6-2.6) |
| Injuriesb | ||||||||||||
| Comparison | 45410 | 77 | 1.70 | 1.0 (ref) | 86123 | 131 | 1.52 | 1.0 (ref) | 76718 | 93 | 1.21 | 1.0 (ref) |
| Case | 4369 | 56 | 12.82 | 7.5 (5.3-10.6) | 7642 | 26 | 3.40 | 2.2 (1.5-3.4) | 6407 | 20 | 3.12 | 2.6 (1.6-4.2) |
| Fracture | ||||||||||||
| Comparison | 45436 | 21 | 0.46 | 1.0 (ref) | 86209 | >25 | . | . | 76773 | >25 | . | . |
| Case | 4406 | 6 | 1.36 | 2.9 (1.2-7.3) | 7663 | <5 | . | . | 6427 | <5 | . | . |
Accounting for sex and birth year.
Includes fractures.
Sensitivity analyses
Results were similar when we restricted analyses to children one year or older at diagnosis/reference date (Table S3). When comparators were restricted to children with at least one hospitalization, cancer cases remained more likely to be hospitalized (HR 2.8, 95% CI: 2.5-3.2) or to die (HR 128.3, 95% CI: 47.8-344.4) (Table S4). Regarding cause-specific hospitalization and death, cases had increased HRs across all diagnosis categories except mental health and fracture and no difference in risk for congenitally-related outcomes. HRs remained greatly increased for all outcomes regardless of gestational age (Table S5) or age at diagnosis (Table S6). We observed no large differences in HRs by treatment era (Table S7) or race (data not shown). However, the absolute hospitalization rate was greater for cases in the later treatment era (1987-1999: 233.3/1,000 person-years; 2000-2012: 320.0/1,000 person-years), whereas the death rate was lesser for cases in the later treatment era (1987-1999: 46.3/1,000 person-years; 2000-2012: 36.8/1,000 person-years). Absolute rates of many cause-specific outcomes were greater in the later treatment era, including infectious, endocrine, circulatory, nervous, and musculoskeletal outcomes.
Cause-specific hospitalizations among cancer cases
A slight majority of cases were hospitalized for a cancer-related diagnosis (51%); 25% of cases had at least one hospitalization with a primary indication for chemotherapy, 47% for procedure, 23% for infection, 26% for toxicity, and 18% for other cancer-related diagnoses (Table 6). For the vast majority of hospitalized cases, the elapsed time to first hospitalization was <1 year from diagnosis for infection, toxicity, and other cancer-related diagnoses; most hospitalizations were <1-week duration. Hospitalizations for other cancer-related diagnoses were relatively longer than hospitalizations for chemotherapy. Relative to children with solid tumors, patients with leukemia/lymphoma were more likely to have hospitalizations lasting ≥1 week, slightly more likely to be hospitalized (RR 0.94, 95% CI: 0.87-1.0), and slightly less likely to die (RR 1.15, 95% CI: 1.1-1.2). These results were similar when restricted to cancer-related diagnoses. Cases with leukemia/lymphoma were more likely to be hospitalized for procedure, infection, or toxicity indications and were more likely to have a later first hospitalization (3 - <5 years after diagnosis).
Table 6.
Cause-specific hospitalizations overall and by indication among children with solid and liquid tumors.
| Outcome | All cases N= 4,567 | Leukemia/Lymphoma N= 1,808 | CNS/Solid tumor N= 2,759 | RRa (95% CI) | |||
|---|---|---|---|---|---|---|---|
| n | % | n | % | n | % | ||
| Any hospitalization | |||||||
| Yes | 2,503 | 54.8 | 1,035 | 57.3 | 1,468 | 53.2 | 0.94 (0.87-1.01) |
| No | 2,064 | 45.2 | 773 | 42.7 | 1,291 | 46.8 | 1.0 (ref) |
| Death (any cause) | |||||||
| Yes | 754 | 16.5 | 250 | 13.8 | 504 | 18.3 | 1.15 (1.09-1.21) |
| No | 3,813 | 83.5 | 1,558 | 86.2 | 2,255 | 81.7 | 1.0 (ref) |
| Cancer-related hospitalization | |||||||
| Yes | 2,346 | 51.4 | 994 | 55.0 | 1,352 | 49.9 | 0.91 (0.84-0.98) |
| No | 2,221 | 48.6 | 814 | 45.0 | 1,407 | 50.1 | 1.0 (ref) |
| Cancer-related death | |||||||
| Yes | 733 | 16.1 | 237 | 13.1 | 496 | 18.0 | 1.16 (1.10-1.22) |
| No | 3,834 | 83.9 | 1,571 | 86.9 | 2,263 | 82.0 | 1.0 (ref) |
| Number of hospitalizations by intent | |||||||
| Chemotherapy | |||||||
| 0 | 3,411 | 74.7 | 1,203 | 66.5 | 2,208 | 80.0 | 1.0 (ref) |
| 1-4 | 721 | 15.8 | 466 | 25.8 | 255 | 9.2 | 0.35 (0.31-0.41) |
| 5+ | 435 | 9.5 | 139 | 7.7 | 296 | 10.7 | 1.39 (1.15-1.69) |
| Procedure | |||||||
| 0 | 2,421 | 53.0 | 905 | 50.1 | 1,516 | 54.9 | 1.0 (ref) |
| 1-4 | 1,969 | 43.1 | 822 | 45.4 | 1,147 | 41.6 | 0.91 (0.85-0.98) |
| 5+ | 177 | 3.9 | 81 | 5.5 | 96 | 3.5 | 0.78 (0.58-1.04) |
| Infection | |||||||
| 0 | 3,520 | 77.1 | 1,260 | 69.7 | 2,260 | 81.9 | 1.0 (ref) |
| 1-4 | 954 | 20.9 | 498 | 27.5 | 456 | 16.5 | 0.60 (0.53-0.67) |
| 5+ | 93 | 2.0 | 50 | 2.8 | 43 | 1.6 | 0.56 (0.38-0.84) |
| Toxicity | |||||||
| 0 | 3,381 | 74.0 | 1,231 | 68.1 | 2,150 | 77.9 | 1.0 (ref) |
| 1-4 | 1,052 | 23.0 | 522 | 28.9 | 530 | 19.2 | 0.66 (0.60-0.74) |
| 5+ | 134 | 3.0 | 55 | 3.0 | 79 | 2.9 | 0.94 (0.67-1.32) |
| Other | |||||||
| 0 | 3,747 | 82.0 | 1,483 | 82.0 | 2,264 | 82.1 | 1.0 (ref) |
| 1-4 | 756 | 16.6 | 309 | 17.1 | 447 | 16.2 | 0.95 (0.83-1.08) |
| 5+ | 64 | 1.4 | 16 | 0.9 | 48 | 1.7 | 1.96 (1.11-3.45) |
| Elapsed time to first hospitalization (years) | |||||||
| Infection | |||||||
| No hospitalization | 3,520 | 77.1 | 1,260 | 69.7 | 2,260 | 81.9 | 1.0 (ref) |
| < 1 | 888 | 19.4 | 445 | 24.6 | 443 | 16.0 | 0.65 (0.58-0.73) |
| 1 - <3 | 129 | 2.8 | 82 | 4.5 | 47 | 1.7 | 0.37 (0.26-0.53) |
| 3 - <5 | 30 | 0.7 | 21 | 1.2 | 9 | 0.4 | 0.28 (0.13-0.61) |
| Toxicity | |||||||
| No hospitalization | 3,381 | 74.0 | 1,231 | 68.1 | 2,150 | 77.9 | 1.0 (ref) |
| < 1 | 1,008 | 22.0 | 481 | 26.6 | 527 | 19.1 | 0.72 (0.64-0.80) |
| 1 - <3 | 138 | 3.0 | 73 | 4.0 | 65 | 2.3 | 0.58 (0.42-0.81) |
| 3 - <5 | 40 | 1.0 | 23 | 1.3 | 17 | 0.7 | 0.48 (0.26-0.90) |
| Other | |||||||
| No hospitalization | 3,747 | 82.0 | 1,483 | 82.0 | 2,264 | 82.1 | 1.0 (ref) |
| < 1 | 681 | 15.0 | 263 | 14.5 | 418 | 15.1 | 1.04 (0.90-1.20) |
| 1 - <3 | 109 | 2.4 | 43 | 2.4 | 66 | 2.4 | 1.00 (0.69-1.47) |
| 3 - <5 | 30 | 0.6 | 19 | 1.1 | 11 | 0.4 | 0.38 (0.18-0.79) |
| Length of first hospitalization (days) by intent | |||||||
| Chemotherapy | |||||||
| No hospitalization | 3,411 | 74.7 | 1,203 | 66.5 | 2,208 | 80.0 | 1.0 (ref) |
| < 7 | 867 | 19.0 | 415 | 23.0 | 452 | 16.4 | 0.71 (0.63-0.80) |
| 7+ | 289 | 6.3 | 190 | 10.5 | 99 | 3.6 | 0.34 (0.27-0.43) |
| Procedure | |||||||
| No hospitalization | 2,421 | 53.0 | 905 | 50.1 | 1,516 | 54.9 | 1.0 (ref) |
| < 7 | 1,493 | 32.7 | 570 | 31.5 | 923 | 33.5 | 1.06 (0.97-1.16) |
| 7+ | 653 | 14.3 | 333 | 18.4 | 320 | 11.6 | 0.63 (0.55-0.72) |
| Infection | |||||||
| No hospitalization | 3,520 | 77.1 | 1,260 | 69.7 | 2,260 | 81.9 | 1.0 (ref) |
| < 7 | 764 | 16.7 | 363 | 20.0 | 401 | 14.5 | 0.72 (0.63-0.82) |
| 7+ | 283 | 6.2 | 185 | 10.3 | 98 | 3.6 | 0.35 (0.27-0.44) |
| Toxicity | |||||||
| No hospitalization | 3,381 | 74.0 | 1,231 | 68.1 | 2,150 | 77.9 | 1.0 (ref) |
| < 7 | 886 | 19.4 | 402 | 22.2 | 484 | 17.5 | 0.79 (0.70-0.89) |
| 7+ | 300 | 6.6 | 175 | 9.7 | 125 | 4.6 | 0.47 (0.37-0.58) |
| Other | |||||||
| No hospitalization | 3,747 | 82.0 | 1,483 | 82.0 | 2,264 | 82.1 | 1.0 (ref) |
| < 7 | 548 | 12.0 | 178 | 9.8 | 370 | 13.4 | 1.36 (1.15-1.61) |
| 7+ | 272 | 6.0 | 147 | 8.2 | 125 | 4.6 | 0.56 (0.44-0.70) |
| Median (IQR) time to hospitalization (days) by intent among those who were hospitalized | |||||||
| Infection | 128 (64-259) | 156 (73-308) | 105 (58-223) | - | |||
| Toxicity | 100 (48-247) | 120 (55-251) | 88 (44-234) | - | |||
| Other | 133 (54-331) | 146 (62-425) | 120 (46-278) | - | |||
| Median (IQR) length of hospital stay (days) | |||||||
| Chemotherapy | 4.7 (3.3-7.0) | 4.9 (3.4-8.2) | 4.3 (3.3-6.1) | - | |||
| Procedure | 5.0 (3.3-7.8) | 5.2 (3.5-9.1) | 4.6 (3.1-7.0) | - | |||
| Infection | 5.0 (3.5-7.2) | 5.1 (3.6-8.3) | 4.7 (3.5-6.3) | - | |||
| Toxicity | 4.8 (3.4-7.0) | 5.0 (3.5-8.0) | 4.6 (3.4-6.4) | - | |||
| Other | 5.3 (3.3-8.4) | 6.3 (3.9-10.8) | 4.8 (3.0-7.0) | - | |||
Leukemia/lymphoma = reference group
Discussion
We observed increased rates of hospitalization and death across all diagnosis categories during the 5 years after diagnosis among childhood cancer cases compared with children without cancer, even after excluding cancer-related outcomes. The risk was greatest in the year following diagnosis. Although the overall hospitalization rate for cases approached that of comparators by 3-5 years post-diagnosis, it was still increased and cases remained more likely to be hospitalized for specific diagnoses, including infections, circulatory diseases, endocrine disorders, nervous system diseases, and injuries.
Hospitalization is not unexpected after a childhood cancer diagnosis. Certain outcomes may be under-reported in many studies. For example, one population-based study described greater early mortality rates than those reported in cooperative clinical trials.[16] Population-based studies reflect the risk of morbidity and mortality more completely than single-institution or trial-based studies by better representing all socio-demographic groups and cancer types. Diagnosis age <1 year reportedly increases risk of hospitalization and death;[16, 40, 41] we observed greatly increased HRs regardless of age at diagnosis. Although a decrease in cancer-related mortality has been described from the 1950s-1990s,[17, 36, 42, 43] mortality outcomes have not substantially changed in recent decades.[42, 44] Our data suggest absolute mortality is decreased in the more recent treatment era. Hospitalization rates among children with cancer appear to be increasing. This may be due to increased treatment intensity for some malignancies (e.g. autologous transplant and chemoimmunotherapy for high risk neuroblastoma), as well as more conservative supportive care practices for selected treatment regimens (e.g. prolonged hospitalization following myelosuppressive regimens for acute myelogenous leukemias).[2, 31, 45] Similarly, we observed greater absolute rates of hospitalization for all children in the more recent treatment era. This causes greater resource utilization when cost-effectiveness in health care is increasingly important.[31] Although resource utilization in childhood cancer survivors is well documented,[11, 46] the greatest hospitalization burden occurs within 5 years after diagnosis.[14, 18] Our results suggest an additional risk beyond what is expected from a child’s cancer diagnosis, which persists beyond the first year following diagnosis for many diagnoses. This may help identify patterns in potentially avoidable outcomes.
Chronic disease in long-term survivors of childhood cancer is well documented.[10, 46-49] However, many chronic diseases likely develop earlier than five years post-therapy. Higher rates of respiratory illnesses have been reported in childhood cancer cases <5 years from diagnosis compared with children without cancer.[22] Similarly, childhood cancer cases have greater proportions of admissions for a range of chronic conditions, including asthma and mental health disorders, as early as 2-years post-therapy.[50, 51] A greater risk of hospitalization for various organ system diseases compared with the general population is also reported among 1-year childhood cancer survivors.[52-57] Our results support the development of chronic disease in children <5 years post-therapy, even after excluding cancer-related diagnoses. Altogether, this highlights the need to identify the timing of chronic disease development, facilitating early intervention.
Skeletal morbidity is also a well-known consequence of selected malignancies and chemotherapy regimens in pediatrics.[19, 58-60] Children with acute lymphoblastic leukemia are reported to be twice as likely to experience fracture compared to the general population.[20] Decreased bone mineral density post-therapy for bone sarcomas has also been described.[61] We observed a >4-fold increased risk of fracture; however no increased risk remained when cancer-related diagnoses were excluded. Small numbers precluded risk estimation beyond the first year. We found a >45-fold increased risk of injury-related outcomes overall. This may be due to the increased risk of complications from injury in childhood cancer cases or because cases are more likely to be closely monitored following an injury. When we further refined injuries to non-treatment-related diagnoses, a 20% increased risk remained, suggesting a child’s cancer diagnosis produces some difference in risk of injury that persists after the first year. Our findings are supported by a recent report examining mortality due to unintentional injury in cancer patients of all ages.[62] Notably, chronic illness is a well-defined risk factor for non-accidental trauma.[63] Moreover, adolescents with cancer continue to participate in risky-behaviors.[64, 65] This comes at a time when many children are not routinely seen by a primary care provider and highlights the importance of ongoing health care maintenance screening during and after cancer treatment.[66, 67]
This study has several limitations. ICD9/10 diagnosis codes have the potential for misclassification. Because all discharge diagnoses were evaluated with similar weight (except the hierarchical assessment of cancer related outcomes, as described), cases may be more likely to have a diagnosis captured because they are under closer surveillance, which would bias estimates away from the null. It is difficult to assess the possible bias due to exclusion of hospitalizations in federal facilities, which were not included in the state hospital discharge database, although it is likely to be quite small as children are much less likely than adults to be hospitalized at a federal facility, which includes only military and veteran’s institutions. Small sub-sample sizes limited our ability to detect differences in some analyses. It is likely that we were unable to link all state-born cases to their birth records. Although we have no way to assess the linkage accuracy, birth records were located for a majority (>70%) of cases identified in the registry and the age- and sex-distributions of linked cases is similar to that of all registry cases. There is a possibility for loss to follow-up for hospitalization due to emigration out of state. Childhood cancer cases may emigrate at lower rates compared with healthy children, over-estimating the number of comparators without hospitalization. If this is true, our estimates would be biased away from the null. However, significantly increased HRs were observed even when restricted to comparators with at least one prior hospitalization. Similar bias for mortality measures may be less likely due to cross-jurisdictional sharing of death record information for state residents, although it is likely some deaths were missed. Hospital discharge records have limited treatment information and cancer registry data only include initial treatment details. Although we may speculate based on cancer type, we were unable to draw conclusions on treatment-specific associations and to analyze outcomes with relapsed disease.
In this population-based analysis, we compared the overall hospitalization and death outcomes experienced by children within 5 years of a cancer diagnosis compared with children in the general population and found increased risks of hospitalization and death, even after excluding cancer-related diagnoses. This may provide insight into which types of complications and non-cancer-related outcomes childhood cancer cases have greatest risk of encountering early after diagnosis, inform anticipatory guidance, and help identify opportunities to prevent toxicity.
Supplementary Material
Acknowledgements:
The authors thank Mr. Bill O’Brien for programming and data management.
Funding:
This work was conducted with grant support from the Alex’s Lemonade Stand Foundation for Childhood Cancer awarded to B. Mueller and E. Chow, a T32 Training Grant (5T32CA009351-40) awarded to A. Steineck, and with cancer registry support from # HHSN261201300012I, N01-CN-005230, N01-CN-67009, N01-PC-35142, HHSN261201000029C from the Surveillance, Epidemiology and End Results (SEER) Program of the National Cancer Institute with additional support from the Fred Hutchinson Cancer Research Center, and Centers for Disease Control and Prevention #DP12-1205 DP003899-02.
Footnotes
Publisher's Disclaimer: This Author Accepted Manuscript is a PDF file of an unedited peer-reviewed manuscript that has been accepted for publication but has not been copyedited or corrected. The official version of record that is published in the journal is kept up to date and so may therefore differ from this version.
Disclosures/Conflicts of Interest: The authors declare that they have no conflicts of interest.
Ethics approval: Institutional Review Board approvals, including waiver of consent, were granted by the Washington State Department of Health and the Fred Hutchinson Cancer Research Center.
Consent to participate: Not applicable
Consent for publication: Not applicable
Availability of data and material: Not applicable
Code Availability: Not applicable
References
- 1.Cunningham RM, Walton MA, and Carter PM, The Major Causes of Death in Children and Adolescents in the United States. N Engl J Med, 2018. 379(25): p. 2468–2475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Russell HV, et al. , Algorithm for analysis of administrative pediatric cancer hospitalization data according to indication for admission. BMC Med Inform Decis Mak, 2014. 14: p. 88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Leyenaar JK, et al. , Epidemiology of pediatric hospitalizations at general hospitals and freestanding children’s hospitals in the United States. J Hosp Med, 2016. 11(11): p. 743–749. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Witt WP, W.A., Elixhauser A, Overview of Hospital Stays for Children in the United States, 2012, in HCUP Statistical Brief. 2014. [PubMed] [Google Scholar]
- 5.Keren R, et al. , Prioritization of comparative effectiveness research topics in hospital pediatrics. Arch Pediatr Adolesc Med, 2012. 166(12): p. 1155–64. [DOI] [PubMed] [Google Scholar]
- 6.Koschmann C, Thomson B, and Hawkins DS, No evidence of a trial effect in newly diagnosed pediatric acute lymphoblastic leukemia. Arch Pediatr Adolesc Med, 2010. 164(3): p. 214–7. [DOI] [PubMed] [Google Scholar]
- 7.Shaw PH and Ritchey AK, Different rates of clinical trial enrollment between adolescents and young adults aged 15 to 22 years old and children under 15 years old with cancer at a children’s hospital. J Pediatr Hematol Oncol, 2007. 29(12): p. 811–4. [DOI] [PubMed] [Google Scholar]
- 8.Liu L, et al. , Childhood cancer patients’ access to cooperative group cancer programs: a population-based study. Cancer, 2003. 97(5): p. 1339–45. [DOI] [PubMed] [Google Scholar]
- 9.Hunger SP, et al. , Improved survival for children and adolescents with acute lymphoblastic leukemia between 1990 and 2005: a report from the children’s oncology group. J Clin Oncol, 2012. 30(14): p. 1663–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mueller BA, et al. , Hospitalization and mortality among pediatric cancer survivors: a population-based study. Cancer Causes Control, 2018. 29(11): p. 1047–1057. [DOI] [PubMed] [Google Scholar]
- 11.Kirchhoff AC, et al. , Risk of hospitalization for survivors of childhood and adolescent cancer. Cancer Epidemiol Biomarkers Prev, 2014. 23(7): p. 1280–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lorenzi MF, et al. , Hospital-related morbidity among childhood cancer survivors in British Columbia, Canada: report of the childhood, adolescent, young adult cancer survivors (CAYACS) program. Int J Cancer, 2011. 128(7): p. 1624–31. [DOI] [PubMed] [Google Scholar]
- 13.de Fine Licht S, et al. , Long-term inpatient disease burden in the Adult Life after Childhood Cancer in Scandinavia (ALiCCS) study: A cohort study of 21,297 childhood cancer survivors. PLoS Med, 2017. 14(5): p. e1002296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kaul S, et al. , A retrospective analysis of treatment-related hospitalization costs of pediatric, adolescent, and young adult acute lymphoblastic leukemia. Cancer Med, 2016. 5(2): p. 221–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Smits-Seemann RR, et al. , Infection-related mortality in Hispanic and non-Hispanic children with cancer. Pediatr Blood Cancer, 2017. 64(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Green AL, et al. , Death Within 1 Month of Diagnosis in Childhood Cancer: An Analysis of Risk Factors and Scope of the Problem. J Clin Oncol, 2017. 35(12): p. 1320–1327. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hamre MR, et al. , Early deaths in childhood cancer. Med Pediatr Oncol, 2000. 34(5): p. 343–7. [DOI] [PubMed] [Google Scholar]
- 18.Rosenman MB, et al. , Hospital resource utilization in childhood cancer. J Pediatr Hematol Oncol, 2005. 27(6): p. 295–300. [DOI] [PubMed] [Google Scholar]
- 19.Elmantaser M, et al. , Skeletal morbidity in children receiving chemotherapy for acute lymphoblastic leukaemia. Arch Dis Child, 2010. 95(10): p. 805–9. [DOI] [PubMed] [Google Scholar]
- 20.Hogler W, et al. , Incidence of skeletal complications during treatment of childhood acute lymphoblastic leukemia: comparison of fracture risk with the General Practice Research Database. Pediatr Blood Cancer, 2007. 48(1): p. 21–7. [DOI] [PubMed] [Google Scholar]
- 21.Warner EL, et al. , Financial Burden of Pediatric Cancer for Patients and Their Families. J Oncol Pract, 2015. 11(1): p. 12–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ramsay JM, et al. , Respiratory emergency department use from diagnosis through survivorship in children, adolescents, and young adults with cancer. Cancer, 2018. 124(19): p. 3924–3933. [DOI] [PubMed] [Google Scholar]
- 23.Pole JD, et al. , Subsequent Malignant Neoplasms in a Population-Based Cohort of Pediatric Cancer Patients: A Focus on the First 5 Years. Cancer Epidemiol Biomarkers Prev, 2015. 24(10): p. 1585–92. [DOI] [PubMed] [Google Scholar]
- 24.Smitherman AB, et al. , Early Posttherapy Hospitalizations Among Survivors of Childhood Leukemia and Lymphoma. J Pediatr Hematol Oncol, 2016. 38(6): p. 423–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Steliarova-Foucher E, et al. , International Classification of Childhood Cancer, third edition. Cancer, 2005. 103(7): p. 1457–67. [DOI] [PubMed] [Google Scholar]
- 26.Warren JL and Harlan LC, Can cancer registry data be used to study cancer treatment? Med Care, 2003. 41(9): p. 1003–5. [DOI] [PubMed] [Google Scholar]
- 27.Zippin C, Lum D, and Hankey BF, Completeness of hospital cancer case reporting from the SEER Program of the National Cancer Institute. Cancer, 1995. 76(11): p. 2343–50. [DOI] [PubMed] [Google Scholar]
- 28.White MC, et al. , The history and use of cancer registry data by public health cancer control programs in the United States. Cancer, 2017. 123 Suppl 24: p. 4969–4976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lydon-Rochelle MT, et al. , The reporting of pre-existing maternal medical conditions and complications of pregnancy on birth certificates and in hospital discharge data. Am J Obstet Gynecol, 2005. 193(1): p. 125–34. [DOI] [PubMed] [Google Scholar]
- 30.Chow EJ, et al. , Morbidity and Mortality Differences Between Hematopoietic Cell Transplantation Survivors and Other Cancer Survivors. J Clin Oncol, 2017. 35(3): p. 306–313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Price RA, Stranges E, and Elixhauser A, Pediatric Cancer Hospitalizations, 2009: Statistical Brief #132, in Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. 2006: Rockville (MD). [Google Scholar]
- 32.Elixhauser AM, EM., Clinical classifications for health policy research, version 2: Hospital inpatient statistics. Healthcare Cost and Utilization Project (HCUP 3) Research Note 1. AHCPR Pub, 1996(98–0049). [Google Scholar]
- 33.Chow EJ, et al. , Cardiovascular hospitalizations and mortality among recipients of hematopoietic stem cell transplantation. Ann Intern Med, 2011. 155(1): p. 21–32. [DOI] [PubMed] [Google Scholar]
- 34.Crump C, et al. , Gestational age at birth and mortality from infancy into mid-adulthood: a national cohort study. Lancet Child Adolesc Health, 2019. 3(6): p. 408–417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Winther JF, et al. , Childhood cancer survivor cohorts in Europe. Acta Oncol, 2015. 54(5): p. 655–68. [DOI] [PubMed] [Google Scholar]
- 36.Armstrong GT, et al. , Reduction in Late Mortality among 5-Year Survivors of Childhood Cancer. N Engl J Med, 2016. 374(9): p. 833–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Bhatia S, et al. , Childhood cancer survivorship research in minority populations: A position paper from the Childhood Cancer Survivor Study. Cancer, 2016. 122(15): p. 2426–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Liu Q, et al. , Racial/Ethnic Differences in Adverse Outcomes Among Childhood Cancer Survivors: The Childhood Cancer Survivor Study. J Clin Oncol, 2016. 34(14): p. 1634–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Todd J, et al. , Increased rates of morbidity, mortality, and charges for hospitalized children with public or no health insurance as compared with children with private insurance in Colorado and the United States. Pediatrics, 2006. 118(2): p. 577–85. [DOI] [PubMed] [Google Scholar]
- 40.Burcham MD, et al. , Emergency Department Chief Complaints Among Children With Cancer. J Pediatr Hematol Oncol, 2018. 40(6): p. 445–449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Mueller EL, et al. , Why pediatric patients with cancer visit the emergency department: United States, 2006-2010. Pediatr Blood Cancer, 2015. 62(3): p. 490–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Smith MA, et al. , Declining childhood and adolescent cancer mortality. Cancer, 2014. 120(16): p. 2497–506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Miller RW and McKay FW, Decline in US childhood cancer mortality. 1950 through 1980. JAMA, 1984. 251(12): p. 1567–70. [PubMed] [Google Scholar]
- 44.Pritchard-Jones K and Hargrave D, Declining childhood and adolescent cancer mortality: great progress but still much to be done. Cancer, 2014. 120(16): p. 2388–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Aprile G, et al. , Unplanned presentations of cancer outpatients: a retrospective cohort study. Support Care Cancer, 2013. 21(2): p. 397–404. [DOI] [PubMed] [Google Scholar]
- 46.Mueller EL, et al. , Insurance, chronic health conditions, and utilization of primary and specialty outpatient services: a Childhood Cancer Survivor Study report. J Cancer Surviv, 2018. 12(5): p. 639–646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Hudson MM, et al. , Health status of adult long-term survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. JAMA, 2003. 290(12): p. 1583–92. [DOI] [PubMed] [Google Scholar]
- 48.Mertens AC, et al. , Cause-specific late mortality among 5-year survivors of childhood cancer: the Childhood Cancer Survivor Study. J Natl Cancer Inst, 2008. 100(19): p. 1368–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Oeffinger KC, et al. , Chronic health conditions in adult survivors of childhood cancer. N Engl J Med, 2006. 355(15): p. 1572–82. [DOI] [PubMed] [Google Scholar]
- 50.Harrington RL, et al. , Impact of multimorbidity subgroups on the health care use of early pediatric cancer survivors. Cancer, 2020. 126(3): p. 649–658. [DOI] [PubMed] [Google Scholar]
- 51.Harrington RL, et al. , Multimorbidity and healthcare utilization among early survivors of pediatric cancer. Pediatr Blood Cancer, 2019. 66(6): p. e27655. [DOI] [PubMed] [Google Scholar]
- 52.Asdahl PH, et al. , Gastrointestinal and liver disease in Adult Life After Childhood Cancer in Scandinavia: A population-based cohort study. Int J Cancer, 2016. 139(7): p. 1501–11. [DOI] [PubMed] [Google Scholar]
- 53.Bonnesen TG, et al. , Liver diseases in Adult Life after Childhood Cancer in Scandinavia (ALiCCS): A population-based cohort study of 32,839 one-year survivors. Int J Cancer, 2018. 142(4): p. 702–708. [DOI] [PubMed] [Google Scholar]
- 54.Bonnesen TG, et al. , Long-term risk of renal and urinary tract diseases in childhood cancer survivors: A population-based cohort study. Eur J Cancer, 2016. 64: p. 52–61. [DOI] [PubMed] [Google Scholar]
- 55.de Fine Licht S, et al. , Hospital contacts for endocrine disorders in Adult Life after Childhood Cancer in Scandinavia (ALiCCS): a population-based cohort study. Lancet, 2014. 383(9933): p. 1981–9. [DOI] [PubMed] [Google Scholar]
- 56.Gudmundsdottir T, et al. , Cardiovascular disease in Adult Life after Childhood Cancer in Scandinavia: A population-based cohort study of 32,308 one-year survivors. Int J Cancer, 2015. 137(5): p. 1176–86. [DOI] [PubMed] [Google Scholar]
- 57.Winther JF, et al. , Risk of cardiovascular disease among Nordic childhood cancer survivors with diabetes mellitus: A report from adult life after childhood cancer in Scandinavia. Cancer, 2018. 124(22): p. 4393–4400. [DOI] [PubMed] [Google Scholar]
- 58.Wilson CL, et al. , Fractures among long-term survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. Cancer, 2012. 118(23): p. 5920–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Blaes AH, et al. , Pathologic femur fractures after limb-sparing treatment of soft-tissue sarcomas. Journal of cancer survivorship : research and practice, 2010. 4(4): p. 399–404. [DOI] [PubMed] [Google Scholar]
- 60.Paulino AC, Late effects of radiotherapy for pediatric extremity sarcomas. International journal of radiation oncology, biology, physics, 2004. 60(1): p. 265–74. [DOI] [PubMed] [Google Scholar]
- 61.Muller C, et al. , Early decrements in bone density after completion of neoadjuvant chemotherapy in pediatric bone sarcoma patients. BMC musculoskeletal disorders, 2010. 11: p. 287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Yang K, et al. , Incidence of Death From Unintentional Injury Among Patients With Cancer in the United States. JAMA Netw Open, 2020. 3(2): p. e1921647. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Leventhal JM, et al. , Fractures in young children. Distinguishing child abuse from unintentional injuries. Am J Dis Child, 1993. 147(1): p. 87–92. [DOI] [PubMed] [Google Scholar]
- 64.Fladeboe KM, et al. , Sexual Activity and Substance Use Among Adolescents and Young Adults Receiving Cancer Treatment: A Report from the PRISM Randomized Controlled Trial. J Adolesc Young Adult Oncol, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Rosenberg AR, et al. , Intimacy, Substance Use, and Communication Needs During Cancer Therapy: A Report From the “Resilience in Adolescents and Young Adults” Study. J Adolesc Health, 2017. 60(1): p. 93–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Wiener L, et al. , Standards for the Psychosocial Care of Children With Cancer and Their Families: An Introduction to the Special Issue. Pediatr Blood Cancer, 2015. 62 Suppl 5: p. S419–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Kazak AE, et al. , Psychosocial Assessment as a Standard of Care in Pediatric Cancer. Pediatr Blood Cancer, 2015. 62 Suppl 5: p. S426–59. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.