Hospitalization and Mortality Outcomes Among Childhood Cancer Survivors by Race, Ethnicity, and Time Since Diagnosis.

Importance: Cancer outcomes are relatively poor in adults who belong to minoritized racial and ethnic groups. Survival and long-term outcomes by race and ethnicity in individuals with childhood cancers are less studied. Objective: To evaluate survival and hospitalization among American Indian and Alaska Native, Asian, Black, and Hispanic children compared with non-Hispanic White children with cancer. Design, Setting, and Participants: This cohort study evaluated all individuals born in Washington State who were younger than 20 years (hereafter referred to as children) and had been diagnosed with cancer during 1987 to 2012, with follow-up ranging from 1 to 27 years. The data subset was built in 2019, and statistical analyses were completed in January 2022. Exposures: Race and ethnicity. Main Outcomes and Measures: Mortality and hospitalization events for all other racial and ethnic groups relative to non-Hispanic White children estimated by Cox proportional hazards regressions for the first 5 years after diagnosis and among cancer survivors 5 or more years after diagnosis.
Results: A total of 4222 children (mean [SD] age, 8.4 [6.4] years; 2199 [52.1%] male; 113 American Indian and Alaska Native [2.7%], 311 Asian [7.4%], 196 Black [4.6%], 387 Hispanic [9.2%], and 3215 non-Hispanic White [76.1%]) with cancer diagnosed at younger than 20 years during 1987 to 2012 were included. Mortality was similar across all groups. Compared with non-Hispanic White survivors at less than 5 years after diagnosis, there were no greatly increased hazard ratios (HRs) for hospitalization. Among survivors at 5 or more years after diagnosis, hospitalization HRs were 1.7 (95% CI, 1.0-3.0) for American Indian and Alaska Native survivors and 1.5 (95% CI, 0.9-2.4) for Black survivors. Significantly increased HRs among Hispanic children were observed for infection-related (HR, 1.4; 95% CI, 1.2-1.6), endocrine-related (HR, 1.3; 95% CI, 1.1-1.6), hematologic-related (HR, 1.3; 95% CI, 1.1-1.5), respiratory-related (HR, 1.3; 95% CI, 1.0-1.5), and digestive-related (HR, 1.2; 95% CI, 1.0-1.5) conditions. American Indian and Alaskan Native children had increased HRs for infection-related (HR, 2.3; 95% CI, 1.2-4.5), hematologic-related (HR, 3.0; 95% CI, 1.4-6.5), and digestive-related (HR, 2.6; 95% CI, 1.3-5.4) conditions. Both American Indian and Alaska Native (HR, 3.6; 95% CI, 1.4-9.0) and Black (HR, 2.5; 95% CI, 1.2-5.5) children had increased mental health-related hospitalizations and death. Conclusions and Relevance: In this cohort study, disproportionately increased long-term risks of hospitalization for physical and mental conditions may have contributed to worse outcomes by race. A key component to bridging the morbidity gap by race is improved understanding of reasons for greater cause-specific hospitalizations in some groups, with development of culturally appropriate intervention strategies.

Introduction

In 2022, more than 15 900 children (age <20 years) are estimated to be diagnosed with cancer and 1600 will die of cancer,[1] making it the leading cause of disease-related death among children in the US.[2] In addition, there are approximately 483 000 survivors of childhood cancer,[3] with survivorship likely to increase because of advances in therapies for the most common childhood cancers.[4,5] However, this increase in survivorship has not equally benefited all racial and ethnic groups; Black and Hispanic children have a lower survival rate than White children.[6,7,8,9] Less is known about cancer outcomes among Asian and American Indian and Alaskan Native children. Factors associated with survival disparity may include socioeconomic status,[10,11] access to treatment,[12] and co-occurrence of other health-related conditions, such as cardiovascular health, injury, and mental health factors.[13,14,15]

Survivors of childhood cancer experience more hospitalizations than those without cancer,[16,17,18] with some hospitalizations related to cancer treatment[19] and others related to posttreatment factors.[18,20,21] Increased short- and long-term hospitalization and mortality among childhood cancer survivors relative to children without cancer have been reported.[18,21] However, an important gap remains in our understanding of the role of race and ethnicity in outcomes among long-term survivors. This gap exists largely because hospitalization and mortality data for children with cancer are largely informed only by clinical trials. However, enrollment and participation rates for racial and ethnic minority children are substantially lower than for White children,[22,23,24] and clinical trials are often limited to hospitalization information within 5 years of diagnosis. To address this gap, we used population-based data to estimate survival and hospitalizations among American Indian and Alaska Native, Asian, Black, and Hispanic compared with non-Hispanic White childhood cancer patients in the state of Washington.

Methods

This retrospective cohort study used a subset of population-based cancer registry data linked to birth (1974-2012), hospital discharge (1987-2013), and death (1987-2013) records from Washington. The Washington State Department of Health and Fred Hutchinson Cancer Research Center institutional review boards approved this study with waiver of informed consent to use these existing data. This report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

As part of a larger study of childhood cancer among children born in Washington,[21] all individuals younger than 20 years with cancer during 1974 to 2015 who were born in state were identified from the National Cancer Institute’s Surveillance, Epidemiology and End Results program–funded cancer surveillance system (13 counties surrounding Puget Sound), starting in 1974, and the 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. Both registries undergo comprehensive assessments of completeness and accuracy.[25,26,27] Information used from the cancer registry included cancer type per the International Classification of Childhood Cancer (ICCC), Third Edition,[28] International Classification for Oncology (ICD-O) morphology and topography codes, histologic features, stage (when applicable), dates of and age at diagnosis, and vital status per quarterly follow-up.

In the parent study from which our data subset was derived, we identified via linkage to state birth records all individuals with cancer who were younger than 20 years at diagnosis (1974-2015) and who were born in state (N = 6320). Individuals with nonmalignant disease (primarily nonmalignant brain tumors, cervical carcinoma in situ, or skin cancers) were excluded (n = 715). Birth record information used in this analysis included sex assigned at birth, birth date, race, and ethnicity.

Outcomes were assessed by linkage of the children’s records to subsequent years of hospital discharge and death records (1987-2013) to identify all non–pregnancy-related hospitalizations and deaths (International Classification of Diseases, Ninth Revision [ICD-9] codes 630-679 and 760-779). We examined outcomes within 2 periods: less than 5 years from diagnosis (early outcomes, including all children; n = 4222 after excluding those with unknown sex or incomplete race and ethnicity data) and 5 or more years after diagnosis (≥5-year outcomes, including only children who survived at least 5 years; n = 2924 after excluding those with unknown sex or incomplete race and ethnicity data).

Hospitalizations and deaths were evaluated overall and categorized by ICD-9/ICD-10 diagnosis category, as previously described.[21,29] Hospital discharge records contain up to 25 fields of diagnosis codes; all were screened to identify first hospitalizations relevant to different diagnosis categories. State death records contain 1 primary and up to 10 contributing causes of death. Diagnosis in any field relevant to a diagnosis category indicated presence of the outcome. Separately, we also evaluated outcomes based on primary cause of hospitalization or death.

Race and ethnicity information from birth records was categorized as American Indian or Alaskan Native, Black, Asian, Hispanic, and non-Hispanic White based on screening both the mother’s and the father’s race and ethnicity field. This information is self-reported by the mother at the time of delivery. Children with disparate parental classifications have been included in more than 1 group, except for non-Hispanic White children, who included only those with both a non-Hispanic White mother and father. Children whose race and ethnicity variables indicated Pacific Islander ancestry (n = 10) were categorized as Asian because their numbers were too small to analyze separately. Three children with maternal or paternal race noted on the birth record as “other non-White” were excluded. Although maternal race in the birth record was missing for only 2% of records, our use of both maternal and paternal variables to identify the non-Hispanic White referent group resulted in race and ethnicity data being missing for 8% of study participants.

Statistical Analysis

Follow-up for early outcomes assessed deaths and hospitalizations at less than 5 years after diagnosis. Because hospital discharge data were only available to us for 1987 to 2013, cases diagnosed before 1987 were excluded, and, to allow at least 1 year of follow-up, those diagnosed after December 2012 were excluded, leaving 4567 cases. Follow-up accrued from diagnosis until whichever came first: December 31, 2013, death, or diagnosis date plus 5 years minus 1 day.[21,30] For the group of cancer survivors at 5 years or more, follow-up was accrued from the date 5 years after diagnosis until first hospitalization, death, or December 31, 2013, as relevant. To provide at least 5 years of follow-up in this group, only cases diagnosed from 1982 to 2008 were included, and follow-up accrued until December 31, 2013 (n = 3152 cases), with up to 27 years of follow-up. The data subset was built in 2019, and statistical analyses were completed in January 2022.

Individuals with missing values for race and ethnicity were excluded. There were no missing values for birth year or age at diagnosis, and less than 1% of data were missing for sex. Outcome incidence rates for first events were estimated per 1000 person-years. Cox proportional hazards regression was used to estimate hazard ratios (HRs) and 95% CIs for hospitalizations and deaths overall and separately for outcome occurrences in each race and ethnicity group compared with non-Hispanic White individuals. The HRs for each diagnosis category were then recalculated for hospitalizations and deaths combined because of small numbers, accounting for birth year and sex in all analyses. The proportional hazards assumption was assessed by performing a χ2 test on the slope of the Schoenfeld residuals; none showed evidence of deviation from proportionality at a 2-sided P < .05. We examined possible differences by sex and age at diagnosis (<10 vs ≥10 years of age) as numbers permitted and by broad categories of cancer type (liquid vs solid tumor); because our results did not materially change by cancer type, results accounting for birth year and sex only are presented. Results of analyses with cell size less than 5 were suppressed. Analyses were conducted using Stata software, version 17 (StataCorp LLC).

Results

A total of 4222 children (mean [SD] age, 8.4 [6.4] years; 2199 [52%] male; 113 American Indian and Alaska Native [2.7%], 311 Asian [7.4%], 196 Black [4.6%], 387 Hispanic [9.2%], and 3215 non-Hispanic White [76.1%]) with cancer diagnosed at younger than 20 years during 1987 to 2012 were included, with a total overall follow-up of 41 132 person-years and 785 deaths (1021 person-years and 680 deaths in the first 5 years and 24 111 person-years and 105 deaths 5 years or more after diagnosis). Leukemia was the most frequent type of cancer for all groups, being most common among Hispanic children (153 [39.5%]), followed by Asian (93 [29.9%]) and American Indian and Alaska Native (34 [30.1%]) children, then Black (49 [25.0%]) and non-Hispanic White (805 [25.0%]) children (Table 1).

Table 1.

Characteristics of Children Younger Than 20 Years With Cancer in Washington State by Race and Ethnicity, 1987-2012

Characteristic	No. (%) of children
	Non-Hispanic White (n = 3215)	Hispanic (n = 387)	Asian (n = 311)	Black (n = 196)	American Indian and Alaska Native (n = 113)
Sex
Male	1696 (52.8)	185 (47.8)	163 (52.4)	100 (51.0)	55 (48.7)
Female	1519 (47.2)	202 (52.2)	148 (47.6)	96 (49.0)	58 (51.3)
Age at diagnosis, y
<5	1172 (36.5)	196 (50.6)	144 (46.3)	80 (40.8)	47 (41.6)
5-9	564 (17.5)	69 (17.8)	62 (19.9)	43 (21.9)	21 (18.6)
10-14	575 (17.9)	66 (17.1)	48 (15.4)	36 (18.4)	22 (19.5)
15-19	904 (28.1)	56 (14.5)	57 (18.3)	37 (18.9)	23 (20.4)
Type of cancer
Leukemias	805 (25.0)	153 (39.5)	93 (29.9)	49 (25.0)	34 (30.1)
Lymphomas	428 (13.3)	41 (10.6)	42 (13.5)	32 (16.3)	10 (8.8)
Central nervous system	683 (20.9)	63 (16.3)	62 (19.9)	43 (21.9)	30 (26.5)
Neuroblastoma	221 (6.9)	28 (7.2)	22 (7.1)	13 (6.6)	7 (6.2)
Retinoblastoma	59 (1.8)	13 (3.4)	8 (2.6)	7 (3.6)	3 (2.7)
Renal	143 (4.4)	22 (5.7)	8 (2.6)	16 (8.2)	7 (6.2)
Hepatic	37 (1.2)	2 (0.5)	11 (3.5)	3 (1.5)	1 (0.9)
Bone	141 (4.4)	15 (3.9)	8 (2.6)	11 (5.6)	5 (4.4)
Soft tissue	232 (7.2)	22 (5.7)	25 (8.0)	14 (7.1)	5 (4.4)
Germ cell	158 (4.9)	11 (2.8)	14 (4.5)	2 (1.0)	6 (5.3)
Other malignant neoplasm	303 (9.4)	15 (3.9)	16 (5.1)	6 (3.1)	5 (4.4)
Other unspecified	15 (0.5)	2 (0.5)	2 (0.6)	0	0

Overall Hospitalization and Mortality

Among all minoritized racial and ethnic groups in the first 5 years after diagnosis, the HRs for hospitalization among Hispanic and Asian children were both 1.1 (95% CI, 1.0-1.3) (Figure; eTable 1 in the Supplement). Mortality HRs for children of all races and ethnicities compared with non-Hispanic White children during the first 5 years were close to the null.

Figure.

Hazard Ratios (HRs) of Overall Hospitalization and Mortality Among Childhood Cancer Survivors by Race and Ethnicity Compared With Non-Hispanic White Children, Accounting for Birth Year and Sex

The HRs are estimated from Cox proportional hazards regression, adjusted for sex and year of delivery.

aP < .05.

Among cancer survivors at 5 years or more, an increased HR for hospitalization was noted for American Indian and Alaska Native (HR, 1.7; 95% CI, 1.0-3.0) compared with non-Hispanic White children. The HR for Black children was 1.5 (95% CI, 0.9-2.4). Hospitalization HRs for the other groups were close to the null. Mortality HRs for Hispanic and Asian survivors at 5 years or more were close to the null; there were too few deaths among Black and American Indian and Alaska Native children to assess.

Cause-Specific Hospitalization and Mortality in the First 5 Years

Significantly increased HRs among Hispanic children were observed for infection-related (HR, 1.4; 95% CI, 1.2-1.6), endocrine-related (HR, 1.3; 95% CI, 1.1-1.6), hematologic-related (HR, 1.3; 95% CI, 1.1-1.5), respiratory-related (HR, 1.3; 95% CI, 1.0-1.5), and digestive-related (HR, 1.2; 95% CI, 1.0-1.5) conditions (Table 2). Increased infection-related hospitalization or death (HR, 1.2; 95% CI, 1.0-1.5) and decreased congenital defect-related hospitalization or death (HR, 0.5; 95% CI, 0.2-0.9) were noted for Asian children. Although an increased HR for musculoskeletal-related conditions during the first 5 years was observed for American Indian and Alaska Native children (HR, 1.5; 95% CI, 0.9-2.5), the CI included 1. In analyses with sufficient numbers, these results were generally similar when primary cause of hospitalization or death was examined, except for attenuated HRs for infection-related outcomes in Asian children and hematologic-related outcomes in Hispanic children and an increased HR for musculoskeletal outcomes in Hispanic children (eTable 2 in the Supplement).

Table 2.

Cause-Specific Hospitalization and Death at Less Than 5 Years After Diagnosis Among Children Younger Than 20 Years With Cancer in Washington State by Race and Ethnicity

Hospitalization or deathb	Non-Hispanic White (n = 3215)	Hispanic (n = 387)	Asian (n = 311)	Black (n = 196)	American Indian and Alaska Native (n = 113)
Infection (all)
HR (95% CI)	1 [Reference]	1.4 (1.2-1.6)	1.2 (1.0-1.5)	0.9 (0.7-1.2)	1.0 (0.7-1.4)
Person-years at risk/No. of events	9772/1022	906/168	835/123	577/59	334/38
Rate per 1000 person-years	103.67	185.40	147.34	102.23	113.68
P value	NA	<.001	.04	.36	.92
Cancerc
HR (95% CI)	1 [Reference]	1.1 (1.0-1.3)	1.1 (1.0-1.3)	1.0 (0.8-1.2)	1.0 (0.8-1.2)
Person-years at risk/No. of events	6719/1882	588/259	518/202	378/116	237/67
Rate per 1000 person-years	275.19	440.21	384.07	301.63	282.85
P value	NA	.09	.16	.62	.74
Endocrine
HR (95% CI)	1 [Reference]	1.3 (1.1-1.6)	1.2 (1.0-1.6)	1.0 (0.7-1.3)	0.8 (0.5-1.3)
Person years at risk/No. of events	11 227/654	1161/112	983/83	639/40	402/20
Rate per 1000	58.08	96.48	83.38	62.64	49.77
P value	NA	.01	.08	.76	.40
Hematologic
HR (95% CI)	1 [Reference]	1.3 (1.1-1.5)	1.0 (0.8-1.3)	1.0 (0.8-1.3)	0.9 (0.7-1.3)
Person years at risk/No. of events	9587/1029	915/166	850/110	517/67	345/35
Rate per 1000	105.87	180.41	129.40	125.80	101.40
P value	NA	.01	.79	.99	.74
Mentalc
HR (95% CI)	1 [Reference]	1.5 (0.6-3.9)	NA	NA	NA
Person-years at risk/No. of events	13 046/54	1490/5	NA	NA	NA
Rate per 1000 person-years	4.14	3.35	NA	NA	NA
P value	NA	.419	NA	NA	NA
Nervous system
HR (95% CI)	1 [Reference]	1.2 (1.0-1.4)	0.9 (0.7-1.2)	0.9 (0.7-1.3)	1.2 (0.8-1.8)
Person-years at risk/No. of events	11 059/670	1141/110	1012/72	641/42	370/29
Rate per 1000 person-years	59.86	96.43	68.20	65.52	78.28
P value	NA	.12	.67	.67	.33
Circulatory
HR (95% CI)	1 [Reference]	1.3 (1.0-1.6)	1.3 (1.0-1.7)	1.0 (0.7-1.4)	1.1 (0.7-1.7)
Person-years at risk/No. of events	12 022/462	1278/76	1068/61	669/29	421/18
Rate per 1000 person-years	38.26	59.47	57.13	41.87	42.76
P value	NA	.08	.07	.81	.76
Respiratory
HR (95% CI)	1 [Reference]	1.3 (1.0-1.5)	1.0 (0.8-1.3)	1.0 (0.8-1.4)	1.0 (0.7-1.5)
Person years at risk/No. of events	10 999/756	1088/124	988/83	608/50	378/28
Rate per 1000	68.55	113.95	82.96	82.20	74.13
P value	NA	.02	.78	.77	>.99
Digestive
HR (95% CI)	1 [Reference]	1.2 (1.0-1.5)	1.0 (0.8-1.2)	0.9 (0.7-1.2)	1.0 (0.7-1.5)
Person-years at risk/No. of events	10 529/804	1065/132	958/85	602/50	376/30
Rate per 1000 person-years	75.51	123.97	87.72	83.06	79.85
P value	NA	.02	.72	.65	.92
Genitourinary
HR (95% CI)	1 [Reference]	1.0 (0.7-1.4)	1.2 (0.9-1.7)	1.1 (0.7-1.7)	1.1 (0.6-1.9)
Person-years at risk/No. of events	12 235/334	1385/43	1106/40	710/23	431/13
Rate per 1000 person-years	27.14	31.05	36.16	32.40	30.14
P value	NA	.96	.31	.71	.83
Congenital
HR (95% CI)	1 [Reference]	1.0 (0.6-1.6)	0.5 (0.2-0.9)	1.0 (0.5-1.9)	1.4 (0.7-3.1)
Person-years at risk/No. of events	12 761/134	1436/22	1200/8	744/10	438/7
Rate per 1000 person-years	10.27	15.32	6.66	13.44	15.99
P value	NA	.93	.02	.94	.36
Skin
HR (95% CI)	1 [Reference]	1.0 (0.7-1.3)	1.1 (0.8-1.5)	0.8 (0.5-1.3)	0.8 (0.4-1.4)
Person-years at risk/No. of events	12 036/366	1340/50	1107/44	693/20	434/11
Rate per 1000 person-years	30.16	37.32	39.74	28.84	25.32
P value	NA	.78	.57	.34	.42
Musculoskeletal
HR (95% CI)	1 [Reference]	1.3 (1.0-1.8)	1.1 (0.8-1.6)	1.1 (0.7-1.7)	1.5 (0.9-2.5)
Person-years at risk/No. of events	12 286/299	1366/52	1122/36	716/21	425/16
Rate per 1000 person-years	24.17	38.07	31.20	29.32	37.66
P value	NA	.07	.48	.67	.13
Injury
HR (95% CI)	1 [Reference]	1.0 (0.8-1.3)	1.2 (1.0-1.5)	1.0 (0.8-1.4)	1.2 (0.8-1.7)
Person-years at risk/No. of events	11 068/678	1211/89	954/83	628/43	381/29
Rate per 1000 person-years	60.99	73.47	85.96	68.46	76.19
P value	NA	.85	.10	.89	.37
Fracturec
HR (95% CI)	1 [Reference]	2.0 (0.8-5.0)	NA	NA	NA
Person-years at risk/No. of events	13 135/26	1496/6	NA	NA	NA
Rate per 1000 person-years	1.98	4.01	NA	NA	NA
P value	NA	.16	NA	NA	NA

Abbreviations: HR, hazard ratio; NA, not applicable.

The HRs are estimated from Cox proportional hazards regression, adjusted for sex and year of delivery.

Diagnosis years 1987 to 2012, with follow-up through 2013.

Results based on cell sizes less than 5 were suppressed.

Cause-Specific Hospitalization and Mortality Among Survivors of 5 Years or More

Although some increased cause-specific hospitalization and death HRs were noted among Asian and Hispanic children, none were statistically significant at P < .05 (Table 3). Among Black children, an increased HR was observed for mental health–related conditions (HR, 2.5; 95% CI, 1.2-5.5). American Indian and Alaskan Native children had increased HRs for infection-related (HR, 2.3; 95% CI, 1.2-4.5), hematologic-related (HR, 3.0; 95% CI, 1.4-6.5), mental health–related (HR, 3.6; 95% CI, 1.4-9.0), and digestive-related (HR, 2.6; 95% CI, 1.3-5.4) conditions. In analyses with sufficient numbers, these results were generally similar when primary cause of hospitalization or death was examined (eTable 3 in the Supplement).

Table 3.

Cause-Specific Hospitalization and Death at 5 Years or More After Diagnosis Among Children Younger Than 20 Years With Cancer in Washington State by Race and Ethnicity

Hospitalization or deathb	Non-Hispanic White (n = 2300)	Hispanic (n = 228)	Asian (n = 200)	Black (n = 125)	American Indian and Alaska Native (n = 71)
Infection (all)
HR (95% CI)	1 [Reference]	1.1 (0.6-2.1)	0.9 (0.4-1.7)	0.9 (0.4-2.1)	2.3 (1.2-4.5)
Person-years at risk/No. of events	22 395/135	1538/11	1734/9	1052/6	629/9
Rate per 1000 person-years	6.03	7.15	5.19	5.70	14.30
P value	NA	.79	.66	.82	.03
Cancerc
HR (95% CI)	1 [Reference]	1.0 (0.5-1.7)	1.1 (0.6-1.8)	1.1 (0.6-2.1)	1.3 (0.6-2.8)
Person-years at risk/No. of events	22 392/188	155/13	1752/15	1067/10	638/7
Rate per 1000 person-years	8.31	8.42	8.56	9.37	10.98
P value	NA	.91	.85	.75	.48
Endocrine
HR (95% CI)	1 [Reference]	0.9 (0.4-1.9)	0.8 (0.4-1.7)	1.0 (0.4-2.2)	1.7 (0.7-3.8)
Person-years at risk/No. of events	22 531/129	1550/8	1787/8	1078/6	637/6
Rate per 1000 person-years	5.73	5.16	4.48	5.57	9.41
P value	NA	.84	.55	.90	.26
Hematologicc
HR (95% CI)	1 [Reference]	1.3 (0.6-2.8)	NA	1.2 (0.5-2.9)	3.0 (1.4-6.5)
Person-years at risk/No. of events	22 836/92	1557/8	NA	1078/5	631/7
Rate per 1000 person-years	3.94	5.14	NA	4.64	11.10
P value	NA	.47	NA	.74	.02
Mentalc
HR (95% CI)	1 [Reference]	2.1 (0.8-5.6)	NA	2.5 (1.2-5.5)	3.6 (1.4-9.0)
Person-years at risk/No. of events	22 888/67	1565/5	NA	1064/7	633/5
Rate per 1000 person-years	2.93	3.19	NA	6.58	7.89
P value	NA	.15	NA	.04	.02
Nervous systemc
HR (95% CI)	1 [Reference]	1.3 (0.7-2.4)	1.1 (0.6-2.1)	NA	NA
Person-years at risk/No. of events	22 665/118	1546/12	1731/10	NA	NA
Rate per 1000 person-years	5.21	7.76	5.78	NA	NA
P value	NA	.42	.83	NA	NA
Circulatoryc
HR (95% CI)	1 [Reference]	NA	NA	NA	NA
Person-years at risk/No. of events	22 909/86	NA	NA	NA	NA
Rate per 1000 person-years	3.75	NA	NA	NA	NA
P value
Respiratoryc
HR (95% CI)	1 [Reference]	1.3 (0.6-2.7)	1.4 (0.7-2.7)	NA	1.9 (0.8-4.7)
Person-years at risk/No. of events	22 697/98	1557/8	1773/10	NA	638/5
Rate per 1000 person-years	4.32	5.14	5.64	NA	7.84
P value	NA	.54	.35	NA	.20
Digestive
HR (95% CI)	1 [Reference]	0.9 (0.4-1.9)	0.9 (0.5-1.9)	1.0 (0.4-2.4)	2.6 (1.3-5.4)
Person-years at risk/No. of events	22 660/105	1553/7	1780/8	1078/5	629/8
Rate per 1000 person-years	4.63	4.51	4.49	4.64	12.72
P value	NA	.71	.85	.94	.02
Genitourinaryc
HR (95% CI)	1 [Reference]	1.8 (0.7-4.8)	NA	NA	NA
Person-years at risk/No. of events	22 932/59	1577/5	NA	NA	NA
Rate per 1000 person-years	2.57	3.17	NA	NA	NA
P value	NA	.25	NA	NA	NA
Congenitalc
HR (95% CI)	1 [Reference]	NA	NA	NA	NA
Person-years at risk/No. of events	23 175/12	NA	NA	NA	NA
Rate per 1000 person-years	0.52	NA	NA	NA	NA
P value	NA	NA	NA	NA	NA
Skinc
HR (95% CI)	1 [Reference]	NA	NA	NA	NA
Person-years at risk/No. of events	23 015/41	NA	NA	NA	NA
Rate per 1000 person-years	1.78	NA	NA	NA	NA
P value	NA	NA	NA	NA	NA
Musculoskeletalc
HR (95% CI)	1 [Reference]	0.7 (0.3-1.8)	0.8 (0.4-1.9)	1.5 (0.7-3.3)	NA
Person-years at risk/No. of events	22 639/97	1563/5	1743/6	1074/7	NA
Rate per 1000 person-years	4.28	3.20	3.44	6.51	NA
P value	NA	.48	.62	.32	NA
Injury
HR (95% CI)	1 [Reference]	1.3 (0.7-2.5)	0.7 (0.3-1.5)	1.1 (0.5-2.3)	1.4 (0.6-3.4)
Person-years at risk/No. of events	22 368/137	1557/11	1751/7	1073/7	633/5
Rate per 1000 person-years	6.12	7.06	4.00	6.52	7.90
P value	NA	.413	.363	.818	.494
Fracturec
HR (95% CI)	1 [Reference]	NA	NA	NA	NA
Person-years at risk/No. of events	23 013/26	NA	NA	NA	NA
Rate per 1000 person-years	1.13	NA	NA	NA	NA
P value	NA	NA	NA	NA	NA

Abbreviations: HR, hazard ratio; NA, not applicable.

The HRs were estimated from Cox proportional hazards regression, adjusted for sex and year of delivery.

Diagnosis years 1982 to 2008, with follow-up through 2013.

Results based on cell sizes less than 5 were suppressed.

Discussion

There are few population-based studies evaluating hospitalizations and mortality among short- and long-term childhood cancer survivors by race and ethnicity. In this cohort study, we observed no increased or decreased mortality by race during the early (<5 years after diagnosis) or later (≥5 years after diagnosis) periods. Hospitalization patterns, however, varied by race, being 70% increased among survivors at 5 or more years for American Indian and Alaska Native compared with non-Hispanic White children. We also observed increased hospitalization and death related to specific conditions in the first 5 years after cancer diagnosis among Hispanic children (infection-, hematologic-, endocrine-, respiratory-, and digestive-related disorders).

Among survivors at 5 or more years, we observed 2.5- to 3.6-fold increased HRs for mental health–related hospitalization or death for Black and American Indian and Alaska Native children. American Indian and Alaskan Native children also had increased HRs for infection-, hematologic-, and digestive-related disorders.

Increased rates of hospitalization at less than 5 years and 5 years or more after diagnosis among childhood cancer survivors compared with children without cancer have been reported previously using Washington State population-based data[18,21] and other registry and self-reported data.[17,31] Previous studies of morbidities in children with cancer found that infection,[31,32] cardiovascular,[33,34,35] endocrine,[31,32,36,37] respiratory,[31,32,37] digestive,[31,32,37] circulatory,[31,32] and mental health[31,32,38] conditions are increased among childhood cancer survivors compared with children without cancer. In addition, nervous system conditions,[31,32,33,37] blood conditions,[31] pregnancy complications,[32] nutritional or metabolic conditions,[32] hearing or eyesight loss,[35] kidney failure,[35] and other self-reported chronic health conditions[35] have been reported as reasons for hospitalization among childhood cancer survivors. However, none of these studies made comparisons by race or ethnicity, so the increased risks we observed in minoritized racial and ethnic groups (already at increased risk for these outcomes) compared with non-Hispanic White survivors emphasizes the even greater magnitude of risk among racial and ethnic minority children relative to children without cancer. Previous studies[18,21,31,32] also identified increased injury among childhood cancer survivors relative to children without cancer. Injury outcomes are difficult to understand in these studies because this outcome category is usually heterogeneous, including trauma, poisoning, and burns as well as fractures potentially associated with cancer treatment-related bone density reduction. Small numbers precluded estimation of the relative occurrence of injury in most racial and ethnic minority children compared with non-Hispanic White children as an outcome in the current analysis. Studies with larger numbers and the ability to examine nontreatment-related injuries are warranted.

In extending previous analyses of short- and long-term hospitalization and mortality outcomes among childhood cancer survivors,[18,21,39] we expand knowledge of these outcomes by race and ethnicity, in particular for American Indian and Alaska Native children, a group rarely examined. The Childhood Cancer Survivor Study[40] reported an age- and sex-adjusted risk of all-cause mortality among Black children compared with non-Hispanic White children of 1.2 (95% CI, 1.0-1.4), which was attenuated to 1.0 (95% CI, 0.8-1.0) after adjustment for socioeconomic characteristics, health behaviors, and health care use.

We confirmed racial disparity in rates of several adverse outcomes among racial and ethnic minority children relative to non-Hispanic White children with cancer and elucidated some conditions that may be relevant in the short term (potentially treatment related) and longer term (less likely to be treatment related) by race and ethnicity. Our data suggest that Hispanic children may experience greater hospitalization in the first 5 years, whereas American Indian and Alaska Native and Black children experience greater hospitalization later. Race and ethnicity can serve as a proxy for the associations among socioeconomic, sociocultural, and environmental factors. Differences by race may reflect variations in patterns of socioeconomic position and health care access (urban or rural, cultural knowledge, medical mistrust, etc).[41,42] For example, previous studies have established that American Indian and Alaska Native adults are less likely to have private health insurance[43,44] and experience increased mental health–related hospitalizations,[45,46,47] emergency department use,[48] and increased travel time and distance to access medical care.[49,50] These factors may play a larger role among cancer patients 5 years or more from diagnosis, when active treatment is usually complete and patients are no longer within the support system of a cancer care facility.

Our results on mortality, but not hospitalizations, are consistent with those of the Childhood Cancer Survivor Study,[10] which compared health status and mortality among Black, Hispanic, and White adult survivors of childhood cancer and did not find greater mortality or other adverse outcomes by race or ethnicity after adjustment for income, educational level, and health insurance status. However, a review by Bhatia[12] described increased adverse outcomes among minoritized racial and ethnic groups examined in several studies and identified possible factors for racial differences in disease-free survival among long-term childhood cancer survivors, including lack of knowledge about potential late toxic effects and screening, health behaviors such as smoking that vary by race and ethnicity, and possible differences in underlying susceptibility and pharmacological responses.

Strengths and Limitations

Our study had key strengths, including our longer follow-up time, which allowed assessment of cause-specific conditions for hospitalization and mortality beyond the early effects of cancer treatments, and evaluation of outcomes among groups that are rarely examined, including American Indian and Alaska Native children and Asian children. Our study also has several limitations. Small numbers limited our ability to assess some comparisons, and we were unable to evaluate outcomes by cancer type. We may not have linked all state-born cases in the cancer registry to the birth records. However, birth records located more than 70% of cases identified in the registry, and the age and sex distributions of linked cases were similar to all registry cases. ICD-9 and ICD-10 diagnosis codes used to assess causes of death and hospitalization have the potential for misclassification, and we were unable to include outcomes in outpatient settings, focusing analyses on serious morbidities and mortality only. We were unable to assess the impact of factors such as health insurance and educational level, which may have resulted in uncontrolled confounding. A recent study[51] indicated adjustment for health insurance type and social deprivation index in comparisons of childhood cancer survival attenuated the differences, although adjustment for these factors still did not remove significant leukemia, lymphoma, and brain tumor survival disparities for Black and Hispanic children compared with non-Hispanic White children. There is also the possibility of loss to follow-up for hospitalization because of emigration, although US Census data indicate that most people who move stay within state.[52] Additionally, results must be interpreted with caution in light of multiple comparisons.

Conclusions

Racial disparities in short- and long-term hospitalizations and mortality exist and may be patterned and compounded by racial differences in socioeconomic factors and access to care. Results from this population-based cohort study with longer-term follow-up help elucidate these disparities. By evaluating specific conditions for hospitalizations, we add results for cause-specific outcomes to help further identify high-risk groups at the system level and during long- and short-term periods since diagnosis. Racial and ethnic minority survivors of childhood cancer may have increased long-term disease burden and a range of conditions that require hospitalizations and compound medical costs, shorten life expectancy, and decrease quality of life. These findings underscore the long-term needs of racial and ethnic minority survivors and need for attention from health care practitioners to monitor survivors’ status and inform appropriate interventions. Pooling similar linked registry data from several states would increase the precision of these estimates and expand generalizability to other regions. Future studies are needed to assess short- and long-term health outcomes and patterns of health care use and co-occurrence of other health-related conditions by race and ethnicity among childhood cancer survivors.