Comparing Age at Cancer Diagnosis between Hispanics and Non-Hispanic Whites in the United States

In 2018, an estimated 149,000 new cancer cases were diagnosed and 42,000 cancer deaths occurred among Hispanics, the largest, youngest, and fastest growing ethnic minority group United States (1, 2). Among Hispanic men, the most frequently diagnosed cancers included cancers of the prostate, colon/rectum, and lung, which together accounted for 40% of cancers diagnosed in Hispanic men. Among Hispanic women, the most frequently diagnosed cancers included cancers of the breast, thyroid, and endometrium, which together accounted for 45% of cancers diagnosed in Hispanic women (3). Although the 2013 to 2017 age-adjusted incidence rates of cancer overall were lower among Hispanic men (370.4 per 100,000 men) and Hispanic women (339.5 per 100,000 women) than among non-Hispanic White (NHW) men (485.5 per 100,000 men) and NHW women (432.0 per 100,000 women; ref. 4), Hispanics are more often diagnosed at later stages and with larger tumors, and experience poorer quality of life after cancer (5).

Previous cancer registry- and hospital-based studies have reported younger ages at diagnosis of breast (6), gastric (7), renal (8), urinary bladder (9), and colorectal (10) cancers and non-Hodgkin's lymphoma (11) and multiple myeloma (12) among Hispanics as compared with NHWs. Although it is important to understand the difference in the crude ages at cancer diagnosis between Hispanics and NHWs as this has implications for cancer prevention and control, comparisons of mean age at cancer diagnosis across racial or ethnic groups may be confounded by differences in population age structure. This point was illustrated by Robbins and colleagues in their study comparing age at cancer diagnosis among U.S. Blacks and NHWs, which showed that most differences in the age at cancer diagnosis between Blacks and NHWs were smaller after adjusting for age structure (13). Similar to Blacks, U.S. Hispanics are also characterized by a younger age structure, which is attributed to high immigration of working-age persons and higher fertility rates among Hispanic women than among NHW women (14). In 2015, 11% of Hispanics were aged 65 years or older compared with 17% of NHWs (15), an important consideration given that the median age at cancer diagnosis in the United States is 66 years (4). Whether differences in age structure confound the comparisons of age at diagnosis among Hispanics and NHWs remains to be clarified.

Following the approach by Robbins and colleagues (13), we examined the differences in mean age at cancer diagnosis among U.S. Hispanics and NHWs with consideration of the role of population age structure on the estimated mean ages. We used data from the Surveillance, Epidemiology, and End Results (SEER) Program of the NCI, which collects and publishes cancer incidence and survival data from population-based cancer registries covering approximately 35% of the U.S. population, with oversampling of regions with higher minority populations (16). Given the younger age structure of Hispanics, we hypothesized that differences in the crude mean ages between Hispanics and NHWs would be largely attributed to the differences in the age structures, and that adjustment for the age structure of the source population would largely attenuate differences in ages at diagnosis between Hispanics and NHWs.

We used data from the following 18 SEER registries: Alaska Native Tumor Registry, Connecticut, Detroit, Atlanta, Greater Georgia, Rural Georgia, San Francisco-Oakland, San Jose-Monterey, Greater California, Hawaii, Iowa, Kentucky, Los Angeles, Louisiana, New Mexico, New Jersey, Seattle-Puget Sound, and Utah (17). The data were limited to U.S. NHWs and Hispanics who were diagnosed with invasive cancer in 2015. Our analysis was restricted to one year to avoid any secular trends by race or ethnicity. We classified cancer sites that were defined by the 2008 SEER recodes (18). Case counts and population estimates produced by the U.S. Census Bureau's Population Estimates Program from the year 2015 were extracted for each cancer type and stratified by race (NHWs vs. Hispanics), age (in single years, 0–84), and sex (male vs. female). Patients 85 years or older were excluded due to the unavailability of data on single year ages. We included only sites with at least 75 cases for both NHWs and Hispanics stratified by sex, to avoid sparse data problems.

We calculated unadjusted mean ages at diagnosis for all cancer sites combined and for 34 cancer sites among NHWs and Hispanics separately. We also calculated a weighted mean age to determine the impact of age structure by race/ethnicity. The weighted mean ages were calculated with statistical weights that were specified for each stratum in single year ages “x” and by sex “y.” Equations used for the statistical weights for NHWs and US Hispanics were (P_NHW,x,y+P_Hispanic,x,y)/P_NHW,x,y and (P_NHW,x,y+P_Hispanic,x,y)/P_Hispanic,x,y, respectively, where P represents the underlying population size in each stratum. The adjusted mean age at diagnosis thus portrays the mean age if both NHWs and Hispanics had the same population age structure. Unadjusted and age-adjusted mean age differences (δ) were calculated by subtracting the mean ages in NHWs from the mean ages in Hispanics. Negative differences therefore represent younger ages at diagnosis in Hispanics compared with NHWs, and positive differences represent older ages at diagnosis in Hispanics. To test for statistically significant differences, we conducted independent samples t-tests comparing the unadjusted and adjusted mean ages at diagnosis by cancer site, race and sex. In sensitivity analyses, we examined crude and adjusted mean ages at cancer diagnosis among adults aged 20 years or older given that some cancers, especially leukemias, occur frequently in children which could skew the age distribution, and for consistency with other cancer surveillance research. Age-specific incidence rates were calculated through SEER*Stat for the four cancer sites in which we observed the largest negative and the largest positive mean age differences comparing Hispanics to NHWs among men and women separately.

A total of 318,218 incident cancers were included in this study: 138,982 NHW males (mean age = 64.9 years; SD = 12.5 years); 133,295 NHW females (mean age = 62.8 years; SD = 13.7 years); 20,913 Hispanic males (mean age = 59.0 years; SD = 16.5 years); and 25,028 Hispanic females (mean age = 56.4 years; SD = 16.1 years). Crude mean ages at diagnosis were lower among Hispanic males and females for all sites combined and for all cancer sites except for gallbladder cancer which was, on average, 1 year higher among Hispanic males than among NHW males (Table 1; Fig. 1). Among males, chronic myeloid leukemia had the largest unadjusted age difference, with U.S. Hispanic males diagnosed on average 15.0 years younger than NHW males. Acute myeloid leukemia among females had the largest unadjusted difference, with U.S. Hispanic females diagnosed on average 11.5 years younger than NHW females.

After adjustment for age structure, Hispanic males remained younger on average than NHW males at diagnosis of chronic myeloid leukemia [δ = −6.1; 95% confidence intervals (CI), −8.1 to −4.1 years], testicular cancer (δ = −4.7; 95% CI, −5.4 to −4.0), Kaposi sarcoma (δ = −3.6; 95% CI, −6.3 to −0.8), mesothelioma (δ = −3.0; 95% CI, −4.3 to −1.7), anal cancer (δ = −2.4; 95% CI, −3.9 to −0.8), and acute myeloid leukemia (δ = −2.2; 95% CI, −3.5 to −0.9), but were older on average at diagnosis of laryngeal cancer (δ = +1.8; 95% CI, 1.1 to 2.6), gallbladder cancer (δ = +3.8; 95% CI, 1.8 to 5.7) and Hodgkin's lymphoma (δ = +7.5; 95% CI, 5.7 to 9.4). After adjustment for age structure, Hispanic females remained younger on average than NHW females at diagnosis mesothelioma (δ = −3.7; 95% CI, −6.7 to −0.7), gallbladder cancer (δ = −3.0; 95% CI, −4.3 to −1.7), and ovarian cancer (δ = −1.6; 95% CI, −2.2 to −1.0), but were older on average at diagnosis of skin cancer (excluding basal and squamous cell carcinomas; δ = +3.8; 95% CI, 3.1 to 4.5), cervical cancer (δ = +4.1; 95% CI, 3.3 to 4.8), and Hodgkin's lymphoma (δ = +7.0; 95% CI, 5.0 to 9.1). Adjustment for age structure resulted in notable changes in mean differences in age at diagnosis of brain cancers in males (crude δ = −11.3 to adjusted δ = 0.4 years) and females (crude δ = −10.1 to adjusted δ = 3.1) and cancers of the bones and joints in males (crude δ = −13.4 to adjusted δ = −1.3) and females (crude δ = −10.1 to adjusted δ = 2.9). In sensitivity analyses in which we examined age structure-adjusted mean ages at cancer diagnosis among adults aged 20 years or older (Supplementary Table S1), the differences in age at diagnosis were largely similar among males and females as compared with when we included children.

Age-specific incidence rates for cancer sites with the four largest negative (chronic myeloid leukemia and testicular among males, and mesothelioma and gallbladder among females) and positive (gallbladder and Hodgkin's lymphoma among males, and cervical cancer and Hodgkin's lymphoma among females) differences are shown in Fig. 2. For chronic myeloid leukemia, incidence rates were similar between Hispanic and NHW males until age 50, after which incidence rates were higher among NHW males than among Hispanic males, and for testicular cancer, incidence rates were higher among Hispanics compared with NHWs before the age of 30, but lower among Hispanics than NHWs after the age of 30. For mesothelioma, incidence rates were similar among women of both ethnic groups and were slightly higher among NHW women over the age of 75, and for cervical cancer, incidence rates were similar among women of both ethnic groups between the ages of 20 and 44, but higher among NHW women than among Hispanic women after the age of 44. Among males, gallbladder cancer incidence rates were similar for both ethnic groups before the age of 60, but were higher among Hispanics than among NHWs after the age of 60, and among females, gallbladder cancer incidence rates were higher among Hispanics than among NHWs after the age of 40. Finally, for Hodgkin's lymphoma, incidence rates were higher among NHW males and females between the ages of 10 and 49 than among males and females of the same ages, and among males, rates were also higher between the ages of 75 and 79, and slightly higher among Hispanic women between the ages of 65 and 79.

In 2015, U.S. Hispanics were diagnosed with cancers at younger ages on average than NHWs; however, these differences appear to be largely a result of the younger age structure in Hispanics compared with NHWs. Adjustment for population age structure, which simulates a scenario in which both populations have the same underlying age structure, is therefore of critical importance when making comparisons in average age at diagnosis across racial or ethnic groups. Of note, younger ages at diagnosis among Hispanics as compared with NHWs persisted after adjustment for age structure for several cancers including chronic myeloid leukemia, testicular cancer, Kaposi sarcoma, mesothelioma, and anal cancer among males and mesothelioma, gallbladder cancer, and ovarian cancer among females. Examination of the incidence rates for several of these cancer sites indicated higher incidence rates among Hispanics than among NHWs at younger ages for testicular cancer and Kaposi sarcoma among males, higher incidence rates among Hispanics than among NHWs at older ages at diagnosis of gallbladder cancer (both sexes), and for cervical cancer. Below we describe possible explanations and implications for cancers in which there was at least a two-year difference in the adjusted mean age at cancer diagnosis between Hispanics and NHWs.

Hispanic men were on average 6.1 years younger at diagnosis of chronic myeloid leukemia, 4.7 years younger at diagnosis of testicular cancer, 3.6 years younger at diagnosis of Kaposi sarcoma, 3.0 years younger at diagnosis of mesothelioma, and 2.4 years younger at diagnosis of anal cancer, but were on average 7.5 years older at diagnosis of Hodgkin's lymphoma and 3.8 years older at diagnosis of gallbladder cancer compared with NHW males after adjustment for population structure. Differences in the age at diagnosis of testicular cancer may be attributed to the population-specific rates of the two main subtypes, seminomas and nonseminomatous germ cell tumors. The incidence of seminomas, which largely affects men in their 30s and 40s, is 60% higher among NHWs than Hispanics whereas the incidence of nonseminomatous germ cell tumors, which largely affects men in their 20s, are identical between both groups which would explain the lower adjusted age at diagnosis for Hispanics (19). Moreover, differences in age at diagnosis of testicular cancer but also for other cancers such as mesothelioma, and leukemias (for which the etiology remains largely unknown; ref. 20) may be due to differences in environmental or occupational exposures. Several studies have suggested a relationship between agricultural employment and pesticide exposures and leukemias (21) and testicular cancer (22, 23), though more definitive studies are needed. Mesothelioma is a well-known occupational disease occurring as a result of asbestos exposure in the workplace (24).Younger ages at first exposure, greater willingness to perform tasks with higher risk, and greater hesitancy in requesting personal protection equipment may further contribute to younger ages at diagnosis of both testicular cancer and mesothelioma in U.S. Hispanic men compared with NHW men (25, 26). Younger ages at diagnosis of Kaposi sarcoma and anal cancers among Hispanic men may also be attributed to a higher prevalence of sexually transmitted infections among Hispanics. Human immunodeficiency virus (HIV) infection, together with the causal agent human herpesvirus-8 (HHV-8) (27), the main risk factor for Kaposi sarcoma in the United States (28), and Human Papilloma virus (HPV) infection, the most important risk factor for anal cancer (29), are more prevalent in Hispanic adolescent males than in NHW adolescent males (30, 31).

Older ages at diagnosis of Hodgkin's lymphoma between Hispanic and NHW men may also be attributed to differences in the incidence of subtypes for this disease, which are remarkably different between Hispanics and NHWs. The incidence rate of nodular sclerosis Hodgkin's lymphoma, for example, has been reported to be lower in younger Hispanic adults compared with NHWs while the incidence of mixed cellularity Hodgkin's lymphoma has been reported to be higher among older Hispanic adults compared with NHWs (32). Gallstones are a major risk factor for gallbladder cancer (33), although gallbladder cancer is rare and most people with gallstones never develop gallbladder cancer, and the prevalence of gallstones has been reported to be similar in Hispanics and NHWs (34). Additional research is therefore needed to explain the reasons for the older ages at diagnosis of gall bladder cancer among Hispanic men compared with NHW men.

Hispanic women were on average 3.7 years younger at diagnosis of mesothelioma and 3.0 years younger at diagnosis of gallbladder cancer, but were on average 7.0 years older at diagnosis of Hodgkin's lymphoma, 4.1 years older at diagnosis of cervical cancer, and between 2 and 4 years older at diagnosis of cancers of the skin, liver, brain, anus, bones and joints, small intestine, and thyroid, compared with NHW females, after adjustment for population structure. We hypothesize that younger ages at diagnosis of mesothelioma among Hispanic women compared with NHW women are likely attributed to exposure to asbestos prior to immigrating to the United States as asbestos continues to be used in most Latin American countries (35). For gallbladder cancer, younger ages at diagnosis among Hispanic women may be attributed to differences in the prevalence of type 2 diabetes, which increases the risk of gallbladder cancer (36–40) and is 1.7 times as prevalent in Hispanic women than in NHW women (41), or to differences in the risk of gallbladder disease, which is higher among Hispanic women than among NHW women (34).

As in men, reasons for older ages at diagnosis of Hodgkin's lymphoma among Hispanic women compared with NHW women may be due to differences in the incidence of Hodgkin's lymphoma subtypes (32). For anal and cervical cancers, U.S. Hispanics females have been reported to have an older median age of first sexual exposure compared with NHW females (42) potentially resulting in an older age at HPV infection and persistent infection of HPV into older ages and thus an older age at diagnosis of these cancers among Hispanic women. Hepatocellular carcinoma (HCC), the main form of liver cancer is primarily driven by hepatitis C virus infection in U.S. males whereas in women, and particularly Hispanic women, metabolic disorders, such as obesity, diabetes and metabolic syndrome and nonalcoholic fatty liver disease is the most common etiology of HCC and this disease is associated with older age (43, 44). Older ages at diagnosis of skin cancer among Hispanic women than among NHW women may be due ultraviolet radiation exposure through occupations such farm work (45), to lower skin screening rates due to less awareness of risks or symptoms, or to less access to health insurance and health care (46). Because cancers of the bones and joints and brain occur in relatively high proportions among children and because Hispanics are proportionally younger, the lower unadjusted ages at diagnosis among Hispanic women are not surprising; however, reasons for older adjusted ages at diagnosis between Hispanic and non-Hispanic women require further investigation. Finally, risk factors for cancers of the thyroid and small intestine are not well understood; however, younger ages at diagnosis among Hispanic women than among NHW women could be due to higher prevalence of obesity in Hispanic women (for thyroid cancer; ref. 47) or to diets higher in saturated fat (48) in Hispanic women (for small intestine cancer; ref. 49), though studies comparing diet quality among Hispanic and non-Hispanic women are needed.

Our study is the first to consider the impact of population age structure on mean age at cancer diagnosis when comparing U.S. Hispanics to NHWs. To do so, we used data from the SEER 18 database, which oversamples the U.S. Hispanic population providing many cases for analysis. However, several limitations should be noted. First, SEER data do not to capture information about country of origin and so our study collapses U.S. Hispanics into one single ethnic group. Hispanics are highly heterogeneous and thus vary greatly with respect to culture, socioeconomic status, genetics, age structure extent of regional dispersion, and cancer burden (50, 51). Cubans, for example, have an older age structure and reside predominantly in the state of Florida, whereas Mexicans have a younger age structure and have migrated throughout the United States (14). The lack of consideration of the variability in cancer rates among these groups (52) may affect the generalizability of our findings to all Hispanics highlighting the importance of considering each heritage group when studying cancer. Second, our results are potentially subject to bias due to emigration at older ages. Less-healthy and older Hispanics have been noted to return to their country of birth (i.e., salmon bias hypothesis; ref. 53) when they retire, become severely ill, or after a temporary job. Differential under-ascertainment of cancer cases among Hispanics at older ages may bias estimates towards younger ages at diagnosis among Hispanics. However, emigration that is the direct consequence of a cancer diagnosis made in the United States would still be captured by the registries, thus having minimal impact in the estimates reported here. Third, SEER does not collect information about cancer screening, which may be important to consider as delays in diagnosis of colorectal, cervical, female breast, prostate and lung (among heavy smokers) cancers may result in older ages at diagnosis for Hispanics. Fourth, SEER registry data can be incomplete due to delays or errors in reporting (54); however, limiting our data to the year 2015 ensures more accurate and complete ascertainments of cancer cases. Fifth, although SEER includes 38% of the U.S. Hispanic population, it does not include the states of Florida or Texas, which have large numbers of diverse Hispanic populations. Finally, in this study we omitted cancer sites with very low incidence due to instability of calculations using sparse data, but we were able to make inferences about the most frequently diagnosed cancers as well as some less well-studied cancers.

Cancer disparities epidemiology is largely concerned with understanding differences in cancer risk factors across racial ethnic groups; when exposure to these factors also varies by age, they may result in Hispanics being diagnosed at younger ages than NHWs. However, age differences which are largely attenuated or reversed (as in our study with urinary bladder cancer) after age-adjustment also demonstrate the contribution of the population age structure to comparisons across groups that are made using means. Statistical adjustments create a hypothetical contrast, which can elucidate, but also sometimes inadvertently hide important disparities and thus should be undertaken with care (55). In our study, we simulated a scenario in which both U.S. Hispanics and NHWs have the same underlying age structure. Although this is currently not the case, as the U.S. Hispanic population continues to age, crude differences in the age of cancer diagnosis may converge over time. Departures from this convergence, however, may signal important areas of investigation.

Our comparison of unadjusted ages at cancer diagnosis between U.S. Hispanics and NHWs indicates that although Hispanics have lower rates of most cancers, they are diagnosed at younger ages than NHWs for most cancers examined. These reports align with the crude mean age differences reported in previous studies, which reflect “real-world” observations that cancer is diagnosed, on average, at earlier ages in Hispanics. Our study demonstrates that age at cancer diagnosis is strongly impacted by differences in age structure, and that adjustment may be important for various cancer sites when comparing populations with different underlying age structures. Future studies should aim to understand the underlying differences between these two groups that may contribute to younger or older ages at cancer diagnosis, and should further disaggregate Hispanic subgroups, which may mask disparities in age at diagnosis between Hispanics. This has important implications for cancer surveillance and cancer prevention given that the identification of many of the potential causal factors may be modifiable and the identification of particularly susceptible subgroups may help inform targeted prevention strategies. The findings of our study give us a deeper understanding behind the epidemiology of cancer among U.S. Hispanics and the use of crude and adjusted mean age at cancer diagnosis. As the U.S. population diversifies, more attention should be put to the earlier ages at which these cancers affect specific racial or ethnic groups.

This study was determined to be non-human subjects research by the San Diego State University Institutional Review Board.

The SEER data and the SEER*Stat software used in this study are freely available online: https://seer.cancer.gov/data-software/. The computer code for replicating these results would require a collaboration with the senior author, a formal data use agreement, and institutional review board approval from the participating institutions.

No disclosures were reported.

H. Parada: Conceptualization, resources, supervision, visualization, writing–original draft, writing–review and editing. A.H. Vu: Data curation, formal analysis, visualization, writing–original draft, writing–review and editing. P.S. Pinheiro: Writing–review and editing. C.A. Thompson: Data curation, formal analysis, supervision, writing–review and editing.

H. Parada Jr was supported by the NCI (K01 CA234317), the National Institute on Minority Health and Health Disparities (U54 MD012397), the SDSU/UCSD Cancer Center Comprehensive Partnership (U54 CA132384 and U54 CA132379), and the Alzheimer's Disease Resource Center for Advancing Minority Aging Research at the University of California San Diego (P30 AG059299). C.A. Thompson received support from the National Center for Advancing Translational Sciences (KL2 TR001444). The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the views of the NIH.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.