Trends in Childhood and Adolescent Cancer Incidence Rates in the United States between 2001 and 2022

Abstract

We estimated trends in age-standardized childhood and adolescent cancer rates between 2001-2022 using national data from the United States Cancer Statistics database. The incidence of cancer in 0-19-year-olds was 18.2 per 100,000 with rates increasing 0.94%/year between 2001-2016 and then decreasing 0.96%/year during 2016-2022. Lymphoma rates increased 0.49%/year during 2001-2022, while trajectories of other cancers varied over time. Leukemia rates increased by 1.03%/year during 2001-2010 and then plateaued. Rates of central nervous system tumors increased by 0.81%/year during 2001-2014 and then declined 2.10%/year during 2014-2022. Rates of other epithelial neoplasms were stable from 2001-2013, increased in 2013-2016, and were stable during 2016-2022. There were an estimated 1,040 additional childhood cancer diagnoses in 2022 compared with what would have been expected based on 2001 rates. Modifications in cancer classifications, screening practices, and diagnostic technology likely contributed to the observed changes, in addition to the potential contributions of putative risk factors.

INTRODUCTION

Major advances in cancer diagnosis and treatment over the last several decades have led to significant reductions in cancer-related mortality in children and adolescents (1). However, cancer remains the leading cause of death due to disease in these age groups in the United States (US) (2).

In contrast to clear long-term trends in decreasing childhood cancer mortality, studies using cancer registry data from the US and Europe (German Childhood Cancer Registry, Nordic Cancer Registries, and Cancer Incidence in Five Continents) have suggested an increase in cancer incidence of 0.5-1.0% per year in children aged 0-14 years and adolescents aged 15-19 years (3-11). Direct comparisons of data from these studies are challenging because of the use of different datasets, inclusion criteria, and time periods. In the US, two recent studies estimated trends in age-standardized childhood cancer rates through 2019. One study estimated an average annual increase of 0.7% per year in cancer incidence in 0-19-year-olds during 1975-2019 based on an analysis of 8 Surveillance Epidemiology and End Results (SEER) cancer registries, representing 8.3% of the US population (12). Another study used national data from the US Cancer Statistics Database (USCS) and reported a 1.1% per year increase during 2003-2016 and then a 2.1% decrease during 2016-2019 (13). In 2022, childhood and adolescent cancer incidence rates in North America were 19.4 per 100,000 compared with the global average of 10.7 (14).

In light of variability in the reported trends in childhood cancer incidence and use of different datasets by other studies, we conducted a comprehensive assessment using national data representing 99.0% of the US population to better understand cancer incidence trends between 2001 and 2022 in children and adolescents. This study investigated overall trends as well as trends in the International Classification of Childhood Cancer (ICCC)-defined major subcategories of childhood cancer: leukemia, lymphoma, central nervous system (CNS), and other epithelial neoplasms and melanoma. It additionally includes estimates of childhood cancer mortality trends and absolute changes in the number of new cancer diagnoses.

RESULTS

Overall Cancer Rates

During 2001-2022, 328,045 cancers occurred among 0-19-year-olds with 1.80 billion person-years of follow-up in the USCS database (age-standardized incidence rate: 18.2 per 100,000). Rates were highest among 0-4-year-olds (incidence rate [IR]=23.2 per 100,000) and 15-19-year-olds (IR=22.9), followed by 10-14-year-olds (IR=14.2) and 5-9-year-olds (IR=12.7). Compared to southern states (IR=17.8 per 100,000), childhood cancer rates were higher in western (IR=18.0; p=0.04), northeastern (IR=19.4; p<0.01), and midwestern states (IR=18.0; p=0.2).

Childhood cancer rates increased significantly 0.94% per year (95%CI 0.82, 1.09) during 2001-2016 and then decreased significantly 0.96% per year (95%CI −1.59, −0.51) during 2016-2022 (Table 1; Figure 1A). Age-specific and sex-specific trends followed a similar pattern. Rates increased significantly and then declined significantly for 0-4-year-olds (joinpoint=2015) and 5-9-year-olds (2015), increased significantly and declined non-significantly among 10-14-year-olds (2018), and increased 0.68% per year (95%CI 0.51, 0.85) across the entire time period among 15-19-year-olds (Figure 1B; Table 2). Among both boys and girls, rates increased during 2001-2016 and then declined significantly during 2016-2022 (APC_boys: −0.98; 95%CI −2.06, −0.36; APC_girls: −0.95; 95%CI −2.01, −0.31; Supplemental Tables 1 and 2). For comparison, childhood cancer rates in SEER-8 increased significantly 0.76% per year during 1975-2016 (95%CI 0.67, 1.14), before plateauing during 2016-2022 (APC=−0.76% per year; 95%CI −3.61, 0.61; Supplemental Figure 1).

Table 1.

Age-standardized cancer incidence (2001-2022) estimates and trends in the United States among 0–19-year-olds, by cancer. Data from the United States Cancer Statistics Database.

Cancer	2001 Rate	2022 Rate	Trend	Years	APC (95% CI)
Total Cancers	17.0	18.3	Trend 1	2001-2016	0.94 (0.82, 1.09)
			Trend 2	2016-2022	−0.96 (−1.59, −0.51)
I. Leukemias, myeloproliferative and myelodysplastic diseases	4.56	4.70	Trend 1	2001-2010	1.03 (0.42, 4.99)
			Trend 2	2010-2022	−0.18 (−2.12, 0.27)
II. Lymphomas	2.38	2.61	Trend 1	2001-2022	0.49 (0.28, 0.69)
III. CNS and miscellaneous intracranial and intraspinal neoplasms	3.01	2.85	Trend 1	2001-2014	0.81 (0.40, 1.39)
			Trend 2	2014-2022	−2.10 (−3.59, −1.26)
IV. Neuroblastoma and other peripheral nervous cell tumors	0.89	0.87	Trend 1	2001-2022	0.39 (−0.11, 0.87)
V. Retinoblastoma	0.33	0.29	Trend 1	2001-2015	0.72 (0.10, 3.12)
			Trend 2	2015-2022	−2.62 (−7.32, −0.66)
VI. Renal tumors	0.68	0.69	Trend 1	2001-2015	0.75 (−0.04, 9.00)
			Trend 2	2015-2022	−1.09 (−5.61, 0.43)
VII. Hepatic tumors	0.20	0.33	Trend 1	2001-2022	1.99 (1.19, 2.82)
VIII. Malignant bone tumors	0.96	0.99	Trend 1	2001-2010	−0.51 (−3.28, 0.16)
			Trend 2	2010-2022	0.88 (0.40, 2.42)
IX. Soft tissue and other extraosseous sarcomas	1.18	1.19	Trend 1	2001-2022	0.09 (−0.18, 0.34)
X. Germ cell tumors, trophoblastic tumors and neoplasms of gonads	1.03	1.15	Trend 1	2001-2022	0.20 (−0.27, 0.65)
XI. Other malignant epithelial neoplasms and malignant melanomas	1.62	2.29	Trend 1	2001-2013	1.37 (−0.16, 2.02)
			Trend 2	2013-2016	8.11 (3.74, 10.1)
			Trend 3	2016-2022	−0.96 (−3.18, 0.14)
XII. Other and unspecified malignant neoplasms	0.09	0.11	Trend 1	2001-2022	0.70 (−0.44, 1.82)

Figure 1.

Age-standardized cancer incidence rates among ≤19-year-olds in the US, 2001-2022 A) overall, and B) by age group. Data from the United States Cancer Statistics Database.

Table 2.

Age-standardized cancer incidence (2001-2022) estimates and trends in the United States by cancer and 5-year age groups. Data from the United States Cancer Statistics Database.

Cancer	Age Group, years	2001 Rate	2022 Rate	Trend	Years	APC (95% CI)
Total Cancers	0-4	22.3	21.8	Trend 1	2001-2015	0.83 (0.53, 1.27)
				Trend 2	2015-2022	−1.69 (−3.22, −0.90)
	5-9	11.6	12.2	Trend 1	2001-2015	1.29 (1.00, 1.66)
				Trend 2	2015-2022	−1.65 (−2.74, −0.89)
	10-14	13.3	15.0	Trend 1	2001-2018	1.23 (1.04, 1.63)
				Trend 2	2018-2022	−1.27 (−3.44, 0.35)
	15-19	21.3	24.3	Trend 1	2001-2022	0.68 (0.51, 0.85)
I. Leukemias, myeloproliferative and myelodysplastic diseases	0-4	8.28	7.85	Trend 1	2001-2008	1.09 (0.03, 6.78)
				Trend 2	2008-2022	−0.57 (−2.96, −0.13)
	5-9	3.90	3.98	Trend 1	2001-2011	1.28 (0.71, 2.65)
				Trend 2	2011-2022	−0.55 (−1.54, −0.01)
	10-14	3.10	3.49	Trend 1	2001-2022	0.85 (0.46, 1.26)
	15-19	3.13	3.64	Trend 1	2001-2022	0.79 (0.51, 1.07)
II. Lymphomas	0-4	0.59	0.90	Trend 1	2001-2022	1.78 (1.10, 2.51)
	5-9	1.37	1.49	Trend 1	2001-2015	1.24 (0.38, 12.5)
				Trend 2	2015-2022	−0.97 (−6.56, 0.86)
	10-14	2.49	2.80	Trend 1	2001-2022	0.64 (0.19, 1.10)
	15-19	5.00	5.16	Trend 1	2001-2022	0.17 (−0.13, 0.46)
III. CNS and miscellaneous intracranial and intraspinal neoplasms	0-4	3.87	3.80	Trend 1	2001-2012	1.22 (0.39, 2.87)
				Trend 2	2012-2022	−2.07 (−4.09, −1.06)
	5-9	3.38	3.10	Trend 1	2001-2015	0.65 (0.22, 1.29)
				Trend 2	2015-2022	−2.66 (−5.00, −1.51)
	10-14	2.63	2.48	Trend 1	2001-2014	1.14 (0.53, 2.89)
				Trend 2	2014-2022	−1.67 (−4.87, −0.29)
	15-19	2.19	2.06	Trend 1	2001-2014	0.52 (−0.08, 5.41)
				Trend 2	2014-2022	−1.60 (−5.53, −0.30)
XI. Other malignant epithelial neoplasms and malignant melanomas	0-4	0.15	0.29	Trend 1	2001-2022	1.48 (0.09, 2.87)
	5-9	0.39	0.53	Trend 1	2001-2022	1.47 (0.43, 2.55)
	10-14	1.42	1.87	Trend 1	2001-2010	1.39 (−4.18, 2.88)
				Trend 2	2010-2019	5.13 (3.99, 10.6)
				Trend 3	2019-2022	−7.23 (−11.7, −2.12)
	15-19	4.49	6.41	Trend 1	2001-2013	1.08 (−0.13, 1.72)
				Trend 2	2013-2016	8.45 (3.99, 10.6)
				Trend 3	2016-2022	−1.02 (−3.23, 0.18)

Trends in total age-standardized childhood cancer rates in USCS resulted in an absolute increase of 2.29 cases per 100,000 from 2001 to 2016 and then a decrease of 1.06 cases per 100,000 during 2016 to 2022. Other epithelial neoplasms and melanoma contributed the most to the increase from 2001 to 2016 (+0.84 per 100,000), followed by leukemias (+0.41), lymphomas (+0.20), and CNS tumors (+0.19) (Figure 2A). Within the other epithelial neoplasms and melanoma site group, thyroid carcinomas (+0.54 per 100,000) and carcinomas of the appendix (+0.44) drove the observed increases through 2016. CNS tumors contributed the most to the decrease from 2016 to 2022 (−0.34 per 100,000), followed by leukemias (−0.27), and other malignant epithelial tumors and melanoma (−0.17; Figure 2B). The most common cancers remained the same throughout the time period. In 2001 and 2022, the most common cancer types were leukemias (26.6% and 25.4%), CNS tumors (17.6% and 15.4%) and lymphomas (14.5% and 15.3%).

Figure 2.

Absolute change in age-standardized incidence rates between A) 2001 and 2016, and B) 2016 and 2022 among ≤19-year-olds in the US. Data from the United States Cancer Statistics Database.

Cause-specific Cancer Rates

Leukemias

Age-standardized rates of childhood leukemia increased significantly 1.03% per year during 2001-2010 (95%CI 0.42, 4.99) and then plateaued during 2010-2022 (APC=−0.18; 95%CI −2.12, 0.27; IR₂₀₂₂=4.70 per 100,000; Table 1, Figure 3A). Leukemia rates among 0-4-year-olds increased significantly 1.09% per year (95%CI 0.03, 6.78) during 2001-2008 but declined 0.57% per year (95%CI −2.96, −0.13) during 2008-2022 (Table 2; Supplemental Figure 2). Among 5-9-year-olds, leukemia rates increased 1.28% per year (95%CI 0.71, 2.65) during 2001-2011 and then decreased 0.55% (95%CI −1.54, −0.01). Leukemia rates among both 10-14-year-olds (APC=0.85; 95%CI 0.46, 1.26) and 15-19-year-olds (APC=0.79; 95%CI 0.51, 1.07) increased significantly across the time period. Lymphoid leukemias and acute myeloid leukemias were the most common subtypes (IR₂₀₂₂=3.54 and 0.78 per 100,000, respectively); however, while rates of lymphoid leukemia continued to increase throughout the time period (APC_2001-2022: 0.47; 95%CI 0.11, 0.82), rates of acute myeloid leukemia increased and then declined significantly (APC_2001-2015: 0.48; 95%CI 0.11, 2.34; APC_2015-2022: −1.32; 95%CI −3.98, −0.23) (Supplemental Table 3).

Figure 3.

Age-standardized rates of cancer among ≤19-year-olds in the US by cancer type, 2001-2022: A) leukemias, B) lymphomas, C) central nervous system tumors and D) other epithelial neoplasms and melanoma. Data from the United States Cancer Statistics Database.

Lymphomas

Age-standardized rates of childhood lymphoma increased 0.49% per year during 2001-2022 (95%CI 0.28, 0.69; IR₂₀₂₂=2.61 per 100,000; Table 1, Figure 3B). Lymphoma rates increased 1.78% per year (95%CI 1.10, 2.51) among 0-4-year-olds and 0.64% per year (95%CI 0.19, 1.10) among 10-14-year-olds (Table 2; Supplemental Figure 2). Among 5-9-year-olds, lymphoma rates increased 1.24% per year (95%CI 0.38, 12.5) through 2015 and then stabilized. Rates did not change among 15-19-year-olds. Hodgkin lymphoma (IR₂₀₂₂=1.23 per 100,000) and non-Hodgkin lymphoma (NHL) (IR₂₀₂₂=1.04) rates were similar overall but the incidence patterns differed: Hodgkin lymphoma rates were stable over the time period evaluated, while rates of NHL were stable until 2011, increased 5.75% per year during 2011-14 (95%CI 1.89, 7.91) and then plateaued during 2014-2022 (Supplemental Table 4).

CNS tumors

Age-standardized rates of childhood CNS tumors increased significantly 0.81% per year during 2001-2014 (95%CI 0.40, 1.39) and then declined significantly during 2014-2022 (APC=−2.10; 95%CI −3.59, −1.26; IR₂₀₂₂=2.85 per 100,000; Table 1, Figure 3C). This pattern was consistent across each age group with annual declines of 2.07% during 2012-2022 (95%CI −4.09, −1.06) among 0-4-year-olds, 2.66% during 2015-2022 (95%CI −5.00, −1.51) among 5-9-year-olds, 1.67% during 2014-2022 (95%CI −4.87, −0.29) among 10-14-year-olds, and 1.60% during 2014-2022 (95%CI −5.53, −0.30) among 15-19-year-olds (Table 2; Supplemental Figure 2). Astrocytomas were the most common type of CNS tumor (IR₂₀₂₂=1.13 per 100,000), with age-standardized rates that did not change significantly from 2001-2017 and then declined 7.10% per year (95%CI −10.1, −4.98) during 2017-2022 (Supplemental Table 5).

Other epithelial neoplasms and melanoma

Age-standardized rates of the broad group of malignancies classified as “other epithelial neoplasms and melanoma” among children were stable during 2001-2013, increased significantly 8.11% per year (95%CI 3.74, 10.1) during 2013-2016, and then did not change significantly during 2016-2022 (IR₂₀₂₂=2.29; Table 1, Figure 3D). Rates increased significantly 1.48% per year among 0-4-year-olds and 1.47% per year among 5-9-year-olds (Table 2; Supplemental Figure 2). In contrast, rates among 10-14 and 15-19-year-olds were stable, increased sharply (APC_2010-2019: 5.13; 95%CI 3.99, 10.6) and (APC_2013-2016: 8.45, 95%CI 3.99, 10.6), respectively, and then declined significantly among 10-14-year-olds and became stable among 15-19-year-olds through 2022. Thyroid carcinomas (IR₂₀₂₂=0.99 per 100,000), appendix carcinomas (IR₂₀₂₂=0.50) and melanomas (IR₂₀₂₂=0.30) were the most common specified cancers in this diverse group of malignancies. Rates of thyroid carcinoma increased annually 4.44% (95%CI 3.93, 5.14) during 2001-2018 and then decreased 5.32% per year (95%CI −10.5, −1.96). This pattern was observed after restricting to differentiated thyroid carcinomas (papillary and follicular) both for males (APC_2018-2022: −6.45; 95%CI −16.3, 1.19) and females (APC_2018-2022: −5.54; 95%CI −11.2, −1.54). Rates of appendix cancer did not change during 2001-2010 but then increased significantly 61.3% per year (95%CI 48.5, 81.2) during 2010-2015, followed by a 3.10% per year increase (95%CI 0.48, 5.57) during 2015-2022. Melanoma rates decreased significantly 4.01% per year (95%CI −4.67, −3.45) from 2001-2022. (Supplemental Table 6).

Other cancer types

The next most common groups of cancers were “soft tissue and other extraosseous sarcomas” (IR₂₀₂₂=1.19 per 100,000), “germ cell tumors, trophoblastic tumors and neoplasms of gonads” (IR₂₀₂₂=1.15), and “malignant bone tumors” (IR₂₀₂₂=0.99; Table 1). Rates of soft tissue and other extraosseous sarcomas and germ cell tumors, trophoblastic tumors and neoplasms of gonads were stable during 2001-2022, while rates of malignant bone tumors increased 0.88% per year (95%CI 0.40, 2.42) during 2010-2022.

Absolute increases in cancer diagnoses relative to 2001

We estimated the number of additional childhood cancer diagnoses that occurred relative to 2001 in each subsequent calendar year (Figure 4). Compared with what would have been expected based on 2001 rates, the absolute number of additional cancer diagnoses peaked in 2015 (n=2,097; 16,136 observed and 14,039 expected cases) and then declined through 2022 (n=1,040; 14,871 observed and 13,831 expected cases; Figure 4A). In 2022, 62% of the additional cancers occurred among 15-19-year-olds (n=645), with 131 additional cancers in 5-9-year-olds, 358 additional cancers among 10-14-year-olds and 94 fewer cancers among 0-4-year-olds. Estimates and patterns differed across cancer types. In 2022, there were an estimated 133 additional leukemias (including 215 additional lymphoid leukemias and only 4 additional acute myeloid leukemias), 182 additional lymphomas (including 146 additional NHLs and 8 fewer Hodgkin lymphomas), 567 additional other epithelial neoplasms and melanomas (including 342 additional thyroid cancers and 247 fewer melanomas) and 126 fewer CNS tumors than expected (including 301 fewer astrocytomas; Figures 4B-E).

Figure 4.

Additional childhood cancers diagnosed in the US during 2002-2022, relative to 2001: A) total childhood cancers, B) leukemias, C) lymphomas, D) central nervous system tumors and E) other epithelial neoplasms and melanoma. Data from the United States Cancer Statistics Database.

Cancer Death Rates

There were 44,977 cancer deaths among 0-19-year-olds in the US during 2001-2023 (age-standardized mortality rate=2.36 per 100,000), declining 21.4% from 2001 (mortality rate=2.75) to 2023 (mortality rate=2.22) – an average annual decline of 1.11% per year (95%CI −1.46, −0.87) over the time period. Cancer mortality rates declined non-significantly during 2001-2009 and 2009-2021, before increasing non-significantly during 2021-2023. Cancer death rates declined significantly among 0-4 (APC_2001-2023: −1.66; 95%CI −2.03, 1.31) and 5-9-year-olds (APC_2001-2023: −1.70; 95%CI −2.07, −1.34), were stable among 10-14-year-olds and increased non-significantly among 15-19-year-olds (Supplemental Figure 3). During 2001-2023, age-standardized CNS tumor mortality rates decreased non-significantly 0.40% per year (95%CI −0.83, 0.03), and lymphoma mortality rates decreased 4.55% per year (95%CI −5.22, −3.97). Leukemia mortality rates decreased 2.97% per year (95%CI −3.39, −2.70) through 2021, but did not change significantly during 2021-2023 (APC=8.16; 95%CI −0.70, 12.6)

DISCUSSION

The analysis of temporal changes in cancer rates provides a broad view of childhood cancer in the U.S. to inform research and public health priorities. In this study, we assessed trends in childhood cancer incidence by analyzing 328,045 incident cancers that occurred between 2001 and 2022 among 0-19-year-olds in the US population. Age-standardized childhood cancer rates increased 0.94% per year during 2001-2016 and then decreased by a similar degree (0.96% per year) from 2016-2022. This equated to an absolute increase of 2.28 cases per 100,000 (2001 compared with 2016) followed by an absolute decrease of 1.06 cases per 100,000 (2016 compared with 2022). There were an estimated 1,040 additional childhood cancer diagnoses (7.5% increase) in 2022 compared with what would have been expected based on 2001 rates. However, overall patterns mask substantial variability by cancer subtype and age that could provide clues regarding the potential drivers of the observed trends.

We focused our cancer type-specific analysis on the most common pediatric cancer categories defined in ICCC: leukemia, lymphoma, CNS tumors, and the group designated as other epithelial neoplasms and melanoma. We found that childhood leukemia rates increased significantly until 2010 and then did not change significantly through 2022, driven mostly by declines among children aged 0-9 years. Steady rate increases in acute lymphoid leukemia (ALL) and rates of acute myeloid leukemia (AML) that peaked in 2015 and then declined were the largest contributors to the overall trend. Certain genetic predispositions (e.g., trisomy 21 or DNA repair disorders) are associated with high risks of leukemia but the etiology of the majority of childhood leukemia cases is unknown (15). Some studies have found increased risk of childhood leukemia associated with exposures to pesticides or ionizing radiation, and reduced risks related to history of allergies or infections (16-19). Per- and polyfluoroalkyl substances (PFAS), persistent environmental contaminants used in industrial processes for decades, have more recently been identified as a potential risk factor for childhood leukemia (20-22). However, the current data suggest that the contribution of environmental exposures to childhood leukemia etiology is small. Large, well-powered longitudinal studies are required to understand if changes in environmental exposures contributed to changes in childhood leukemia rates.

Childhood lymphoma rates increased slowly over the time period with different patterns across lymphoma types and age groups. Rapid increases in rates of Burkitt lymphoma during 2019-2022 reflect coding changes implemented in 2021 that recoded Burkitt leukemias as Burkitt lymphomas (23). Elevated risk of lymphoma occurs in immunodeficiency disorders and has been associated with certain infections (e.g., Epstein Barr virus, EBV) (24). It is unlikely that changes in EBV infections are directly responsible for changes in incidence rates, as EBV prevalence in children is high and does not appear to have increased over time (25).

The overall incidence of CNS tumors increased between 2001 and 2014 and sharply declined beginning in 2014. These changes appear to be primarily driven by changes in astrocytoma rates, the most common CNS tumor in 0-19-year-olds, and are likely due to the 2016 changes in CNS tumor classification with the publication of the updated World Health Organization (WHO) guidelines (26). This coding change is also likely reflected by increases in the “mixed and unspecified gliomas” category noted from 2019-2022. The etiology of childhood CNS tumors is complex, including cancer predisposition syndromes (e.g., choroid plexus carcinoma in Li-Fraumeni syndrome) and ionizing radiation exposure, most commonly associated with diagnostic medical exams (e.g., computed tomography [CT] scans) (27-29). CNS malignancies are especially challenging to categorize and differences in their incidence could be reflected by changes in diagnostic criteria over the last several decades (26). Notably, our data do not indicate an increase in malignant brain tumors in the last decade, consistent with other analyses of malignant and non-malignant brain tumors (30,31). This is supportive of prior studies that have shown no association between cell phone use and CNS tumors in children, as the prevalence of cell phone use has increased while CNS tumor rates have declined (32,33).

Incidence rates of the broad category of “other epithelial neoplasms and melanoma” increased rapidly during 2013-2016 with non-significant trends during 2001-2013 and 2016-2022. This pattern was also present among 15-19 and 10-14-year-olds, while rates increased steadily among younger age groups. Changes in rates in this heterogeneous group were driven by appendix cancer, melanoma, and thyroid carcinoma rates. Rapidly increasing appendix cancer rates were consistent with the previously described change in the reportability of carcinoid tumors to cancer registries. (34,35). Notably, rates of childhood melanoma declined significantly, a finding possibly consistent with increased use of sun protective measures in this time frame (36). Childhood thyroid carcinoma increased 4.44% annually through 2018 and then decreased by 5.32% per year. A similar pattern has been observed for adulthood thyroid carcinoma incidence in the U.S., albeit with a slightly earlier plateau (2014-2016) and subsequent downturn (37). The rise in adulthood thyroid carcinoma incidence has been largely attributed to overdiagnosis, owing to the rapid growth in diagnostic imaging and use of more sensitive diagnostic tools since the early-1980s, although a more modest influence of environmental or lifestyle factors has not been ruled out (38). The impact of overdiagnosis on U.S. childhood thyroid carcinoma incidence trends has been less clear (39,40); however, enhanced thyroid cancer surveillance has been shown in other settings to have a marked influence on thyroid cancer detection and diagnosis in children (41). The recent declines in thyroid carcinoma incidence, both in adults and children (as observed in the current study), coincided with major shifts in clinical practice recommendations for adult and pediatric thyroid carcinoma diagnosis and treatment over the last 10-15 years, including improvements in risk stratification and stronger recommendations against screening (37,38). Thus, our results suggest a strong influence of diagnostic and screening practices on thyroid carcinoma incidence trends not only in adults, but also in children.

The age-standardized cancer incidence rates in SEER-8 were consistent with our findings in USCS. Analyses in SEER-8 showed an overall slow increase in childhood cancer incidence from the start of reporting until an inflection point in 2016 when rates began to decline, though the downward trend was not statistically significant. This is consistent with prior studies utilizing SEER data to investigate different time periods (9-12), though older studies did not include a sufficient number of years of data after 2016 to detect the downward trend. While USCS data covers a shorter time period, the nationwide dataset allows for more robust findings, with greater statistical power than SEER (e.g., SEER-8 includes only 8.3% of the US population) to detect temporal trends in incidence rates across age groups and cancer types. Our results extend prior work in USCS that presented trends through 2019, showing that declining incidence rates continued through 2022 (13). The current study additionally includes estimates of childhood cancer mortality trends and absolute changes in the number of new cancer diagnoses.

We have outlined potential explanations for changing childhood cancer rates in the last two decades, including both etiologic drivers as well as cancer classification changes and changes in clinical practice. Changes in coding and ascertainment are likely responsible for some of the most rapid changes in rates over time. In addition, the growing understanding of the role germline genetic risk plays in cancer etiology of children and adults could contribute. Historically, a genetic cancer-prone syndrome was identifiable primarily by associated syndromic features (e.g., somatic overgrowth in Beckwith-Wiedemann syndrome or dysmorphic features in trisomy 21)(15,42). Germline testing of non-syndromic children with cancer suggests an underlying genetic etiology in at least 10% of cases (15,43). As more adults with cancer are undergoing genetic testing, cascade testing of their family members, including children, has identified more individuals with a cancer predisposing syndrome which may prompt more cancer screening (44). This, in turn, could lead to increased number of cancer diagnoses and earlier diagnoses, though it is unclear if this has had or will have a meaningful impact on trends.

Death due to childhood cancer declined non-significantly through 2021 with significant declines noted in children less than 10 years old. Continued monitoring of childhood cancer mortality rates is warranted, as death rates increased non-significantly during 2021-2023. However, given that cancer therapies continue to be refined and new efficacious treatment modalities are being discovered, improvements in long-term survival are expected to continue (1).

This nationwide surveillance study provided a detailed overview of childhood and adolescent cancer incidence and mortality rates over the last two decades. We found a slow increase in childhood cancer rates followed by a decrease occurring around 2016. Updates in disease classification likely impacted some of the observed changes in cancer rates along with changes to diagnostic technology and perhaps increased screening of high-risk groups. The prevalence of other putative risk factors may also influence cancer rates, including yet-to-be-defined environmental exposures. Robust monitoring of cancer rates is critical for the prioritization of public health efforts and future studies focused on etiology and prevention. A commitment to large, prospective, well-powered longitudinal studies should advance knowledge of specific factors that could influence childhood cancer rates and perhaps lead to new strategies for early interception or prevention.

METHODS

Data Sources

We utilized 2001-2022 (most recent year available) incidence data from the United States Cancer Statistics (USCS) Database (RRID:SCR_024896), which includes population-based cancer registries from all US states funded by both the SEER program and the National Program of Cancer Registries. Analyses were limited to 0-19-year-olds – this age range includes both childhood and adolescent cancers, henceforth referred to in aggregate as “childhood cancers.” Cancer types were defined with the ICCC recode extended 3^rd edition (45), restricted to cancer with malignant behavior. Multiple primary tumors in the same individuals were included. Analyses were restricted to states that met USCS publication criteria for the full time period (https://www.cdc.gov/national-program-cancer-registries/about/npcr-standards.html), which excludes data from Mississippi and includes 99.0% of the US population. We also estimated longer-term temporal trends (1975-2022) using data from SEER-8 (RRID: SCR_006902).

Statistical Analysis

Annual age-standardized incidence rates (standardized to the 2000 US population), and age-specific incidence rates (0-4, 5-9, 10-14 and 15-19-year-olds) were estimated for each cancer type using SEER*Stat software version 8.4.5 (RRID:SCR_025808). Joinpoint regression was used to estimate annual percentage changes (APCs) in rates from 2001 to 2022 using Joinpoint software version 5.4.0.0 (RRID:SCR_018129). Inflection points (or “joinpoints”) in the trajectory of cancer trends could occur in different calendar years for different combinations of cancer type and age group. 95% confidence intervals (CIs) and p-values were calculated in Joinpoint using the Empirical Quantile Method (https://surveillance.cancer.gov/help/joinpoint/setting-parameters/method-and-parameters-tab/apc-aapc-tau-confidence-intervals/empirical-quantile); p<0.05 was considered statistically significant. 2020 was excluded from trend analysis given the known impact of the COVID-19 pandemic on cancer diagnoses in that year (46). We have focused our main presentation on the four largest site groups of childhood cancers as classified by ICCC – “leukemias, myeloproliferative and myelodysplastic diseases” (i.e., leukemias), “lymphomas and reticuloendothelial neoplasms” (i.e., lymphomas), “central nervous system (CNS) and miscellaneous intracranial and intraspinal neoplasms” (i.e., CNS tumors), and “other malignant epithelial neoplasms and malignant melanomas” (i.e., other epithelial neoplasms and melanoma); however, the results for all cancer categories are presented in Table 1. We excluded miscellaneous lymphoreticular neoplasms from our lymphoma definition as ascertainment changes over time resulted in pronounced and artifactual temporal trends that strongly influenced the overall lymphoma trends (47,48). We also estimated rates stratified by sex and geographic region (i.e., northeast, south, midwest and west) (49).

In addition, we assessed the contribution of specified cancer site groups to the overall trend on an absolute scale by comparing the change in age-standardized incidence rate between 2001 and 2016 (time period during which the overall trend increased) and 2016 and 2022 (time period during which the overall trend decreased). To estimate the absolute change in the number of childhood cancers due to increasing incidence rates, we compared the observed number of diagnoses in 2002 through 2022 to the expected number based on 2001 rates among 0-19-year-olds. The expected number of diagnoses were estimated by multiplying age-specific 2001 rates by population estimates in each subsequent year (2002-2022).

Finally, we estimated trends in age-standardized cancer mortality rates using national death certificate data from the National Center for Health Statistics (RRID:SCR_025830) for 2001-2023 (most recent year of data available) with cancer deaths classified based on ICD-10 codes overall (C00-C97) and for lymphomas (C81-86, C96.3), leukemias (C91-C95) and brain and central nervous system tumors (C70-C72).

Significance.

Childhood and adolescent cancer rates increased during 2001-2016 and then declined significantly. This temporal pattern was largely driven by rates of leukemias, lymphomas, CNS tumors and other epithelial tumors. Changes in cancer classification, screening and diagnostic technology likely influenced cancer-specific patterns, in addition to potential changes in risk factor prevalence.

DATA AVAILABILITY

US Cancer Statistics data are publicly available from the Centers for Disease Control and Prevention (https://www.cdc.gov/united-states-cancer-statistics/index.html). The US SEER data are publicly available from the National Cancer Institute (https://seer.cancer.gov/)