Disparities in Diagnosis, Treatment Access, and Time to Treatment Among Hispanic Men With Metastatic Prostate Cancer

Abstract

PURPOSE

Reporting racial/ethnic disparities in aggregate obscures within-group heterogeneity. We sought to identify disparities in diagnosis and treatment in Hispanic subpopulations with metastatic prostate cancer (mPCa).

METHODS

We disaggregated men with prostate adenocarcinoma from the National Cancer Database from 2004 to 2017 by racial subgroup and Hispanic background. We assessed (1) presenting with mPCa, (2) receiving any treatment, and (3) receiving delayed treatment beyond 90 days. Logistic regression and adjusted odds ratios (aOR) were reported.

RESULTS

Hispanic men had greater odds of presenting with mPCa (aOR, 1.54; 95% CI, 1.50 to 1.58; P < .001) compared with non-Hispanic White (NHW) men. All Hispanic racial subgroups were more likely to present with mPCa, with the highest risk in Hispanic Black (HB) men (aOR, 1.68; 95% CI, 1.46 to 1.93; P < .01). Men from all Hispanic backgrounds had higher odds of presenting with mPCa, especially Mexican men (aOR, 1.99; 95% CI, 1.86 to 2.12; P < .01). Hispanic men were less likely to receive any treatment (aOR, 0.60; 95% CI, 0.53 to 0.67; P < .001), and this effect was particularly strong for Hispanic White patients (aOR, 0.58; 95% CI, 0.52 to 0.66; P < .001) and Dominican men (aOR, 0.52; 95% CI, 0.28 to 0.98; P = .044). Hispanic men were more likely to experience treatment delays compared with NHW men (aOR, 1.38; 95% CI, 1.26 to 1.52; P < .001) and in particular HB (aOR, 1.83; 95% CI, 1.22 to 2.75; P = .002) and South/Central American men (aOR, 1.48; 95% CI, 1.07 to 2.04; P = .018).

CONCLUSION

Differences exist in stage at presentation, treatment receipt, and delays in treatment on disaggregation by racial subgroup and Hispanic heritage. We need to study the potential mechanisms of the observed variations to help develop targeted interventions.

Disparities in metastatic #PCa diagnosis and treatment among Hispanic subgroups: disaggregation is key!

INTRODUCTION

Hispanic people represent the largest ethnic minority in the United States, and in 2030, it is expected that one of five people living in the United States will self-identify as being Hispanic.¹ Among this group, prostate cancer is the most common nonskin cancer, accounting for 10% of cancer deaths.² Hispanic men face screening and treatment disparities across the prostate cancer disease continuum, from decreased likelihood of receiving treatment in the localized stage to higher rates of presenting with advanced disease.^3,4

CONTEXT

• Key Objective

• To identify disparities in metastatic prostate cancer (mPCa) diagnosis and treatment in Hispanic subpopulations using the National Cancer Database.

• Knowledge Generated

• Assessing key outcomes of presenting with metastatic disease, treatment receipt, and delay in treatment in mPCa subpopulations reveals disparities within the highly heterogenous Hispanic population. We find that overall, all Hispanic racial subgroups and countries of origins experience disparities, but even after taking into account key socioeconomic factors, certain subgroups experience persistent disparities: White and Hispanic Other racial subgroups and Mexican and Puerto Rican men were more likely to present with metastatic disease, and Mexican and Central/South American men were more likely to experience treatment-related delays.

• Relevance

• Disaggregating diverse populations to assess disparities allows for future targeted interventional approaches to mitigate inequities among specific populations while respecting their cultural influences.

Although Hispanic patients are often studied in aggregate,^5,6 significant within-group heterogeneity exists, particularly in cancer care.⁷ For instance, studies among Hispanic patients with localized prostate and breast cancers show observed differences in cancer stage at the time of diagnosis between racial and country-of-origin subgroups within the broader ethnicity.^8,9 Although Hispanic patients with metastatic prostate cancer (mPCa) as a whole have the second highest overall survival,¹⁰ this aggregated approach in Hispanic populations may mask different outcomes across Hispanic subgroups. Indeed, disaggregating oncologic outcomes in other large racial populations such as Asians also reveals important within-group differences that are often otherwise overlooked.^11,12 In the Hispanic population, oncologic outcomes of patients with mPCa have not been studied in a disaggregated fashion.

In this study, we assessed disparities in mPCa in Hispanic men disaggregated by race and countries of origin. Specifically, we examined risk of presenting with metastatic disease, access to treatment, and time to treatment. We hypothesized that Hispanic patients are likely to present with advanced disease at a higher rate compared with non-Hispanic White (NHW) patients, but significant differences will emerge across Hispanic subgroups when analyzed by disaggregated racial groups and Hispanic heritage.

METHODS

We define Hispanic people as individuals who identify as Hispanic, Latino, or Latinx and had a heritage background from a Spanish-speaking country. Consistent with federal guidelines, Brazilians are excluded from the US definition of Hispanic people. This approach was chosen to match the data source's disaggregation of Hispanic populations. Although Hispanic populations are often mischaracterized as a racial group rather than an ethnicity, we also used race to identify racial subgroups (ie, Hispanic White [HW], non-Hispanic Black [NHB]) within the Hispanic population.⁸

We included patients with prostate cancer diagnosed between 2004 and 2017 from the National Cancer Database (NCDB). Patients missing data on race, Hispanic ethnicity, or stage information were excluded. All patients were categorized by race and ethnicity subgroups on the basis of the NCDB prostate cancer Participant User Files into NHW, NHB, non-Hispanic Other (NHO), HW, Hispanic Black (HB), and Hispanic Other (HO). We also disaggregated patients by Hispanic countries of origin including Mexico, Puerto Rico, Cuba, South/Central America, and Dominican Republic. These disaggregated race, Hispanic ethnicity, and Hispanic countries of origins were our primary independent variables of interest. From these populations, we generated a subcohort of patients with metastatic disease.

We assessed our three main outcomes between groups by race and ethnicity and by country of origin. First, we compared the rate of metastatic disease at the time of diagnosis between groups. Second, for those with metastatic disease, we compared the rate of treatment, defined as the receipt of systemic treatment. Third, among patients who received treatment, we defined treatment delay as the receipt of treatment 90 days after diagnosis and compared the rate of treatment delay between groups.

For the three outcomes, we used the chi-squared test for univariable comparisons. We also incorporated logistic regression with two models—an adjusted model and a mediator model. The purpose of incorporating two models is to first look for disparities within subgroups and then assess how socioeconomic variables may mediate our outcomes. Our first adjusted model included age (continuous), facility type (Community Cancer Program, Comprehensive Community Cancer Program, Academic/Research Program, and Integrated Network Cancer Program), diagnosis year, and Charlson-Deyo score (0, 1, 2, or ≥3). Our second mediator analysis model incorporated both individual and neighborhood-level socioeconomic factors with neighborhood median household income (<$38,000 in US dollars [USD], $38,000 USD-$47,999 USD, $48,000 USD-$62,999 USD, or ≥$63,000 USD), percentage of adults in each patient's ZIP code of residence without high school education (≥21.0%, 13.0%-20.9%, 7.0%-12.9%, or <8.0%), and insurance status (uninsured, private insurance/managed care, Medicaid, Medicare, or other government). These three variables were chosen to reflect a more thoughtful approach in covariate adjustments when examining race and ethnicity as primary variables, assessing the potential influence of intervenable mediators.^13-15 Analyses were performed using StataMP 17.0 (StataCorp, College Station, TX).

RESULTS

Our cohort included 1,305,785 men with prostate cancer (Table 1). Hispanic men accounted for 4.5% (58,848) of the whole study cohort, and the majority of them (89.7%, 52,795) were White. Of the 103,187 (7.9% of whole cohort) men with metastatic disease, 73.2% (75,512) were NHW and 6.1% (6,254) were Hispanic. Stratifying the Hispanic population with metastatic cancer by country of origin, 17.19% (1,075) were Mexican, 5.04% (315) were Puerto Rican, 3.04% (190) were Cuban, 6.51% (407) were South or Central American, and 2.57% (161) were Dominican.

TABLE 1.

Cohort Characteristics

Metastatic Disease on Presentation

On univariable analysis, Hispanic patients in aggregate fared worse in all three outcome measures compared with NHW patients: 10.62% (6,254 of 58,848) presented with metastatic disease compared with 7.37% (75,512 of 1,024,561) in NHW (P < .001). Within the racial subgroups, 10.55% (5,570 of 52,795), 11.20% (227 of 2,026), and 11.35% (457of 4,027) of HW, HB, and HO presented with metastatic disease, respectively. When disaggregated by country of origin, patients of Mexican heritage presented with a high rate of metastatic disease at 13.27% (1,075 of 8,098; Fig 1).

FIG 1.

Likelihood of presenting with stage IV prostate cancer versus stage 0-III, disaggregated by race/ethnicity and country of origin. The adjusted model included for age, year of diagnosis, facility type, and Charlson-Deyo score. Mediator analysis included additional variables of income, education, and insurance. All subgroup percentages are column percentages, and stage IV presentations are row percentages. aOR, adjusted odds ratios; HB, Hispanic Black; HO, Hispanic Other; HW, Hispanic White; mOR, mediated odds ratio; NHB, non-Hispanic Black; NHO, non-Hispanic Other; NHW, non-Hispanic White; OR, odds ratio; PR, Puerto Rican; ref, reference.

In our logistic regression models, Hispanic patients as a group were more likely to present with metastatic disease in our adjusted model (adjusted odds ratios [aOR], 1.54; 95% CI, 1.50 to 1.58; P < .001) and this effect persisted after further accounting for socioeconomic factors (mediated odds ratio (mOR), 1.19; 95% CI, 1.15 to 1.23; P < .001). Disparities existed across all Hispanic racial subgroups in the adjusted model, but the effect seen in HB men became insignificant after adjusting for socioeconomic factors in the mediator analysis. When disaggregating by country of origin, all subgroups had higher risk of being diagnosed with metastatic disease, with the strongest effect among men of Mexican background (aOR, 1.99; 95% CI, 1.85 to 2.12; P < .02). After adjusting for socioeconomic status, Mexican (mOR, 1.47; 95% CI, 1.37 to 1.57; P < .01) and Puerto Rican heritage men (mOR, 1.23; 95% CI, 1.09 to 1.39; P < .01) still had higher odds of presenting with mPCa. Dominican men, conversely, after adjusting for socioeconomic status (SES) were less likely to present with metastatic disease (Fig 1; Appendix Table A1, online only).

Treatment for Metastatic Disease

Among men with metastatic disease, Hispanic patients received treatment at a slightly lower rate of 91.4% (3,974 of 4,350) compared with NHW patients (93.1%; 48,726 of 52,315; P < .001). Within the Hispanic population among racial subgroups, 91.0% (3,477 of 3,819) of HW and 91.9% (159 of 173) of HB received treatment. With country of origin–based disaggregation, Mexican and Dominican men had the lowest rates of treatment at 90.9% (655 of 721) and 91.2% (114 of 125), respectively compared with NHW men (93.1%; P < .001; Fig 2).

FIG 2.

Likelihood of treatment receipt if metastatic. The adjusted model included for age, year of diagnosis, facility type, and Charlson-Deyo score. Mediator analysis included for income, education, and insurance. All subgroup percentages are column percentages, and treated categories are row percentages. aOR, adjusted odds ratios; HB, Hispanic Black; HO, Hispanic Other; HW, Hispanic White; mOR, mediated odds ratio; NHB, non-Hispanic Black; NHO, non-Hispanic Other; NHW, non-Hispanic White; OR, odds ratio; PR, Puerto Rican; ref, reference.

On multivariable analysis, Hispanic patients with metastatic disease overall were less likely to receive treatment (aOR, 0.60; 95% CI, 0.53 to 0.67; P < .001) and this effect persisted after adjusting for socioeconomic factors (mOR, 0.76; 95% CI, 0.67 to 0.86; P < .001). After disaggregation by racial subgroups, both NHB and HW were less likely to receive treatment despite adjusting for socioeconomic variables (NHB: mOR, 0.69; 95% CI, 0.63 to 0.74; P < .01; HW: mOR, 0.74; 95% CI, 0.65 to 0.84; P < .01). In the adjusted model, Mexican and Dominican heritage men were less likely to receive treatment (Mexican: aOR, 0.58; 95% CI, 0.45 to 0.75; P < .01; Dominican: aOR, 0.52; 95% CI, 0.28 to 0.98; P = .04), but odds were no longer significantly different when compared with NHW men after adjusting for SES (Fig 2; Appendix Table A1).

Treatment Delays

Hispanic men with mPCa appear to have a higher rate of treatment delays (10.9%; 562 of 5,141) when compared with NHW populations (7.2%; 4,746 of 61,504; P < .001). Within the Hispanic population when separated out by racial subgroups, 10.6% (484 of 4,071) of HW men, 14.8% (28 of 189) HB men, and 12.6% (50 of 397) of HO men experienced treatment delays. Men from Mexico, Puerto Rico, and South/Central America had the highest rate of delays at 10.6% (90 of 851), 10.2% (28 of 274), and 12.6% (43 of 340), respectively (Fig 3).

FIG 3.

Likelihood of delay to treatment by >90 days. The adjusted model included for age, year of diagnosis, facility type, and Charlson-Deyo score. Mediator analysis included for income, education, and insurance. All subgroup percentages are column percentages, and treated categories are row percentages. aOR, adjusted odds ratios; HB, Hispanic Black; HO, Hispanic Other; HW, Hispanic White; mOR, mediated odds ratio; NHB, non-Hispanic Black; NHO, non-Hispanic Other; NHW, non-Hispanic White; OR, odds ratio; PR, Puerto Rican; ref, reference.

In our logistic regression analysis, Hispanic men in aggregate experienced higher odds of treatment delay (aOR, 1.38; 95% CI, 1.26 to 1.52; P < .001; mOR, 1.40; 95% CI, 1.27 to 1.55; P < .001). With racial disaggregation, all subgroups except for NHO men experienced treatment delay relative to NHW men and this effect persisted in our mediator model. When disaggregating by country, Mexican (mOR, 1.42; 95% CI, 1.13 to 1.78; P < .01) and South/Central American heritage men (mOR, 1.52; 95% CI, 1.10 to 2.11; P = .012) were more likely to experience treatment delay (Fig 3; Appendix Table A1).

Sensitivity analysis defines treatment delay as >180 days to treatment and did not identify significant differences from using the 90-day definition (data not shown).

DISCUSSION

To our knowledge, our study is one of the first to assess mPCa outcomes among Hispanic populations in a disaggregated fashion, acknowledging the diversity inherent in this ethnic group. We found significant disparities in the Hispanic population with a higher proportion of Hispanic patients in aggregate presenting with metastatic disease and experiencing lower rates of treatment and higher rates of treatment delays. When the Hispanic population was disaggregated by race, all Hispanic racial subgroups were more likely to present with mPCa; HW men were less likely to receive cancer treatment. Furthermore, HW, HB, and HO men were more likely to experience treatment delays. When disaggregating by Hispanic heritage, men from all Hispanic countries were more likely to present with metastatic disease; Mexican and Dominican heritage men were less likely to receive treatment, and Mexican and South/Central American heritage men were more likely to experience treatment delays. After considering key socioeconomic factors, disparities persisted for several groups: HW, HO, Mexican, and Puerto Rican heritage men still were more likely to present with metastatic disease; Mexican and South/Central American heritage men were more likely to experience treatment-related delays.

HB populations experienced the worst disparities in stage at presentation and delayed treatment. Few prostate cancer studies have disaggregated Hispanic populations by race, perhaps in part, due to the controversial nature of racial identity in the Hispanic/Latino/Latinx community. Racial identifiers used in the United States, such as White and Black, do not directly correspond to perceptions of race in many Latin American countries. Conversely, racial identities like mestizo play a much more significant role in Latin America with no similarly available identifier in American categorizations.¹⁶ Although health data for indigenous and Afro-descendant populations in Latin American countries are limited, early studies suggest widespread disparities across a wide spectrum of outcomes.^17,18 When these populations migrate to the United States, multiple factors influence racial self-identification, including nativity status, length of stay in the United States, spoken language, and societal perceptions, with many rejecting US Census Bureau categorizations altogether and categorizing themselves as other race.^19-23 Despite these complexities, Afro-Latino people, sometimes also referred to as Black Latino or Black Hispanic people, have become increasingly recognized as a unique population with distinct inequities in health when compared with other Hispanic/Latino/Latinx communities.²⁴ Our findings appear to corroborate these observations, supporting the need for a focused study in these populations.

While few studies have examined disparities among disaggregated Hispanic populations by country of origin in mPCa, our findings corroborate localized prostate cancer disparities among Hispanic communities.²⁵ Within localized prostate cancer, Mexican, Puerto Rican, and South or Central American populations present with higher-risk disease when compared with NHW populations: SEER-based studies have documented worse prostate cancer–specific survival in Mexican men with locoregional prostate cancer and higher risk of harboring aggressive disease with active surveillance.^7,26 In addition, Puerto Rican men tend to have a higher incidence of prostate cancer relative to other Hispanic men and worse prostate cancer–specific mortality after radical prostatectomy.^2,8,27 In our study, men of Dominican heritage were more likely to present with metastatic disease and less likely to receive treatment in our adjusted model, but the effect became reversed and insignificant in the mediator analyses, respectively. This suggests that insurance status and neighborhood-level socioeconomic factors may play a greater role in explaining disparate outcomes in Dominican populations. Populations of Mexican descent experienced the greatest disparities even after adjusting for social determinants of health in both our study and studies examining disparities in localized cancer. Given the worsening disparities in the metastatic setting when compared with the localized prostate cancer setting, our findings suggest that lack of intervention at early higher-risk cancer stages could play a key role in worsening rates of metastasis in these populations.

Among men with metastatic cancer, NHB and HW men were less likely to receive treatment. For those receiving treatment, almost all Hispanic racial subgroups and Mexican and South/Central American men were less likely to receive timely treatment compared with NHW men in our adjusted model. As a whole, ethnic minorities face higher barriers to care compared with NHW populations such as transportation and cost,²⁸ especially in the metastatic disease state, which requires consistent and often intensive follow-up.²⁹ Cultural differences between the patient and the physician may also increase barriers to care. Moreover, our treatment variable is a binary variable, which does not consider whether a patient receives ADT monotherapy or appropriate treatment intensification and adjuncts. As treatment intensification becomes the standard of care,^30-32 we need more granular details to capture the quality of care and how that affects oncologic outcomes. As treatment intensification requires more social support and financial resources, it is likely that minorities including Hispanic patients will face higher barriers.

What explains these various country and racial differences in outcomes? Social determinants of health, rather than genetics, are most likely given the lack of genetic distinctiveness in either racial or country of origin identities. As noted by Stern, Hispanic people descend from European, Amerindian, and African ancestral populations, with varying degrees of admixture.³³ When Hispanic and NHW populations were analyzed within equal-access systems, prostate cancer oncologic outcomes were similar.^6,34,35 Among Hispanic Americans, however, large variations in access to care, language, insurance status, cultural practices, areas of settlement within the United States, and citizenship status have all been noted when disaggregated by both race and country of origin.^9,36-39 Causal analyses examining the varying degrees of influence of these factors remain unexplored in prostate cancer. Our mediation analysis suggested that factors like insurance status, neighborhood income, and neighborhood education level could at least partially explain the differences we observed. Differences in stage at presentation appeared to be mitigated for Cuban and Central and South American populations when compared with NHW populations after adjusting for these additional factors. By contrast, disparities persisted in patients of Mexican, Puerto Rican, and Dominican populations, further supporting that distinct factors affect each of these populations that warrant targeted evidence-based interventions.

Although the NCDB collects data from over 1,500 Commission on Cancer–accredited facilities, selection and sampling bias may still occur. As noted above, disaggregation on the basis of country of origin or American definitions of race may obscure other cultural factors that may share equal importance in explaining the observed disparities within this population. The NCDB does not unfortunately provide robust data on other such factors, such as language barriers or level of acculturation. Sample size in smaller disaggregated populations may also be subject to type II error. Granular oncologic data such as burden of disease or type of treatment intensification were not available. Hispanic background data contain a high degree of missingness, similar to Hispanic data within the SEER database. These categorizations also have imperfect definitions and groupings, such as the South/Central American subgroup, which aggregates multiple uncharacterized countries of origins together. They inadequately capture the nuances of one's background, but at the same time, we cannot do away with them altogether.⁴⁰ Finally, residual effects of socioeconomic status may still be present in mediator analyses given the imperfect nature of both the individual and neighborhood-level SES variables provided in the NCDB data set.

Our findings support several areas of future exploration, particularly in the mechanisms of the observed differences. Although our study suggests that socioeconomic status may account for some of the observed disparities among Hispanic populations, further studies are needed to validate this observation. In addition, our finding suggesting that Dominican populations may benefit more from individual and neighborhood-level socioeconomic interventions should be further studied. Some groups, like Mexican populations, continued to experience disparities despite adjusting for socioeconomic status, suggesting that other mechanisms, such as language, social ties, and cultural factors, may account for some of the observed disparities. In addition, health outcomes in Afro-Latino/HB populations remain understudied and merit independent studies. In addition, given their aggregation in our study, populations like Central and South Americans should be better examined in future work. Qualitative studies are likely to shed further light on the reasons underlying the observed disparities and may help to clarify modifiable factors that could be leveraged to promote equity. Finally and most importantly, acting on disparities found in descriptive studies such as this is the important next step. So far, interventional studies targeting disparities found in patients with cancer remain rare but increasing.⁴¹ Among prostate cancer, most interventional studies are centered around prostate-specific antigen screening and shared decision making. Carlson et al demonstrated that education at the time of prostate cancer screening in the setting of outreach events increases patient knowledge and additional randomized trials are underway.^42-44 Targeted community outreach programs such as barbershop engagement programs are feasible among Black patients⁴⁵; similar creative interventions informed by cultural humility should be explored in other minority groups including Hispanic men. In the more advanced disease states, interventional studies are sorely needed.

In conclusion, our study examined disparities in mPCa, including stage at presentation, treatment status, and delayed treatment, using the National Cancer Database, focusing primarily on Hispanic populations stratified by race and country of origin. We observed wide variations across subgroups, with socioeconomic status potentially playing a large role, particularly in Dominican populations. These findings should be validated in other data sets, and other potential mechanisms, including language, acculturation status, and region of settlement within the United States, should also be explored. By taking a targeted interventional approach that respects these diverse influences, physicians and public health leaders can better mitigate the inequities that these populations continue to experience.

APPENDIX

TABLE A1.

Adjusted and Mediator ORs for Outcome Measures Disaggregated by Race/Ethnicity and Country of Origin

AUTHOR CONTRIBUTIONS

Conception and design: Helen Y. Hougen, Nishwant Swami, Edward Christopher Dee, Brandon A. Mahal

Administrative support: Brandon A. Mahal

Collection and assembly of data: Nishwant Swami, Edward Christopher Dee

Data analysis and interpretation: Nishwant Swami, Edward Christopher Dee, Helen Y. Hougen, Mohammed Alshalalfa

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Disparities in Diagnosis, Treatment Access, and Time to Treatment Among Hispanic Men With Metastatic Prostate Cancer

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).