Skip to main content
NCBI home page
JNCI Journal of the National Cancer Institute logoLink to JNCI Journal of the National Cancer Institute
. 2021 Apr 23;113(10):1343–1351. doi: 10.1093/jnci/djab062

Association of Health-Care System With Prostate Cancer-Specific Mortality in African American and Non-Hispanic White Men

Daniella Klebaner 1,#, P Travis Courtney 1,2,#, Isla P Garraway 3, John Einck 1, Abhishek Kumar 1, Maria Elena Martinez 4,5, Rana McKay 6, James D Murphy 1,2, Humberto Parada 7, Ajay Sandhu 1, Tyler Stewart 6, Kosj Yamoah 8, Brent S Rose 1,2,✉
PMCID: PMC8486332  PMID: 33892497

Abstract

Background

Disparities in prostate cancer-specific mortality (PCSM) between African American and non-Hispanic White (White) patients have been attributed to biological and systemic factors. We evaluated drivers of these disparities in the Surveillance, Epidemiology, and End Results (SEER) national registry and an equal-access system, the Veterans Health Administration (VHA).

Methods

We identified African American and White patients diagnosed with prostate cancer between 2004 and 2015 in SEER (n = 311 691) and the VHA (n = 90 749). We analyzed the association between race and metastatic disease at presentation using multivariable logistic regression adjusting for sociodemographic factors and PCSM using sequential competing-risks regression adjusting for disease and sociodemographic factors.

Results

The median follow-up was 5.3 years in SEER and 4.7 years in the VHA. African American men were more likely than White men to present with metastatic disease in SEER (adjusted odds ratio = 1.23, 95% confidence interval [CI] = 1.17 to 1.30) but not in the VHA (adjusted odds ratio = 1.07, 95% CI = 0.98 to 1.17). African American vs White race was associated with an increased risk of PCSM in SEER (subdistribution hazard ratio [SHR] = 1.32, 95% CI = 1.10 to 1.60) but not in the VHA (SHR = 1.00, 95% CI = 0.93 to 1.08). Adjusting for disease extent, prostate-specific antigen, and Gleason score eliminated the association between race and PCSM in SEER (aSHR = 1.04, 95% CI = 0.93 to 1.16).

Conclusions

Racial disparities in PCSM were present in a nationally representative registry but not in an equal-access health-care system, because of differences in advanced disease at presentation. Strategies to increase health-care access may bridge the racial disparity in outcomes. Longer follow-up is needed to fully assess mortality outcomes.

Disparities between African American and non-Hispanic White (White) patients in cancer-specific mortality have been described across numerous cancer types and health-care systems (1-5). The survival gap between African American and White patients with prostate cancer has been well characterized, with 2-fold higher prostate cancer-specific mortality (PCSM) rates among African American patients depending on the setting (1,6–10). This disparity has been attributed to differences in prostate cancer biology in African American men, in addition to systemic factors in mediating this disparity, such as differential access to health care, prostate-specific antigen (PSA) screening, and distrust in the health-care system (1,11-16).

The Veterans Health Administration (VHA) is a relatively equal-access health-care system that treats a large, ethnically diverse population of veterans. The Surveillance, Epidemiology, and End Results (SEER) program is a national cancer registry program that collects data from the general US population. The goals of the present investigation were to: 1) compare PCSM between African American and White men within SEER and the VHA and 2) identify modifiable system-level contributors to these disparities. We hypothesized that PCSM would be comparable among African American and White men in an equal-access setting (the VHA) but not in a national registry (SEER) and that this disparity in SEER would be in part driven by more advanced disease at presentation.

Methods

Setting

For the SEER cohort, we obtained patient data from the SEER Incidence Database, which collects cancer incidence data from population-based cancer registries covering approximately 35% of the US population (17). Although SEER neither represents one specific health-care system nor is fully representative of the entire US population (18), it is reflective of a population that generally receives health care in fee-for-service, nonintegrated settings. For the VHA cohort, we obtained patient data from the VHA’s Corporate Data Warehouse (19), which contains the health records of more than 9 million veterans who received care at 170 VHA medical centers and more than 1000 outpatient sites, as well as a cancer registry of patients diagnosed with and/or treated for cancer at a VHA site (20). The current study was reviewed and approved by the VHA San Diego Healthcare System. Waivers of consent and authorization were granted by the institutional review board and the research and development committee of the VHA San Diego Healthcare System (institutional review board protocol number 150169).

Study Population

In both the SEER and VHA cohorts, we included male patients diagnosed with prostate adenocarcinoma between 2004 and 2015 who were without prior malignancy. The study population was further limited to African American and White patients without missing information on disease and sociodemographic characteristics, described below. Race and ethnicity were self-reported based on fixed categories. Cohort selection is further delineated in Supplementary Figure 1 (available online).

Measures

Covariates. For disease characteristics at presentation, we collected information on PSA, Gleason score, and extent of disease at diagnosis. Disease extent was classified as local if there was no disease outside the prostate (American Joint Committee on Cancer [AJCC] clinical nodal and metastasis stage 0) or extraprostatic extension (AJCC clinical tumor stage 1-2); regional if there was extraprostatic extension (clinical tumor stage 3-4) or regional nodal metastases (clinical nodal stage 1); and metastatic if there were organ or distant nodal metastases (clinical metastasis stage 1). Stage and grade were at diagnosis. Metastatic disease at diagnosis was determined in both SEER and the VHA using their respective registry “Clinical M” variables (21,22), which document clinical metastasis stage at diagnosis. For sociodemographic factors, we collected age at diagnosis, US geographic region, marital status at diagnosis, year of diagnosis, and neighborhood socioeconomic status (SES). In the SEER cohort, we included census-tract level SES index, a time-dependent composite score provided by SEER (23). In the VHA cohort, we included zip code–level median household income and education level, based on American Community Survey census data (24), as proxy measures for SES.

Survival. SEER incidence data contains date and cause of death (cancer or other) for each patient, which we used to measure survival in this cohort. In the VHA, we used National Death Index data linked to the VHA to identify the date and cause of death. Patients who were alive at the date of last follow-up, defined as December 31, 2015, were censored on that date, as National Death Index data in the VHA database extended until December 31, 2015.

Statistical Analysis

Baseline disease-specific and sociodemographic characteristics of African American and White patients were compared using χ2 and Wilcoxon rank-sum for categorical and continuous variables, respectively. We performed logistic regression in both cohorts assessing the odds of metastatic vs nonmetastatic disease at diagnosis among African American men compared with White men, adjusting for age, region, marital status, year of diagnosis, and SES.

The primary endpoint of this study was PCSM, and the secondary endpoint was all-cause mortality. We measured the cumulative incidences of PCSM and all-cause mortality in African American and White men and used Gray test (25) to evaluate differences. We assessed the association between race and PCSM with Fine-Gray (26) competing-risks regression, considering death from nonprostate cancer causes as a competing event. We assessed the association of race with all-cause mortality with Cox proportional hazards regression after confirming the proportional hazards assumption was met through examination of Schoenfeld residuals. Additionally, as a sensitivity analysis, we combined both the SEER and VHA cohorts and tested for an interaction between race and health-care system.

Separate Fine-Gray models were performed in the SEER and VHA cohorts to evaluate for differences in the association of race with PCSM between the 2 registries. We first performed univariate analysis including only race. Covariates were generally categorized into 2 groups: 1) disease-specific characteristics at presentation, which included PSA, Gleason score, and extent of disease at diagnosis, and 2) sociodemographic characteristics at presentation, which included age at diagnosis, US geographic region, year of diagnosis, marital status, and SES. The coding of each covariate included in the multivariable models is provided in the Supplementary Methods (available online). Covariates were added individually and sequentially to the univariate model to assess the relative contribution of each variable to the disparity in PCSM between African American and White men. Finally, we ran additional multivariable Fine-Gray regression models adjusting for all disease-specific characteristics and/or sociodemographic characteristics. Given the exclusion of a substantial number of patients during the cohort selection process (Supplementary Figure 1, available online), we performed a sensitivity analysis on a cohort with imputed information for missing covariates. We performed the same Fine-Gray regressions of PCSM and Cox proportional hazards regressions of all-cause mortality with the imputed data to assess whether our results remained consistent.

To assess the contribution of covariates to the disparity in PCSM between African American and White men, we computed the relative attenuation of the subdistribution hazard ratio (SHR) for race (27). This was defined as [SHRRace-SHRRace+X] ÷ [SHRRace-1], where X is any number of covariates added to the Fine-Gray model. However, the univariate SHRRace in the VHA cohort is close to 1 (see the “Results” section), and thus this calculation could not be performed for the VHA cohort and is reported for only the SEER cohort.

We used SAS V9.2 (SAS Institute Inc, Cary, NC) and R 3.4.1 (R Foundation for Statistical Computing, Vienna, Austria) for all statistical analysis and figure design. All tests were 2-sided and a P value of less than .05 was considered statistically significant.

Results

Baseline Characteristics

The SEER cohort included 311 691 men (58 553 [18.9%] African American and 253 138 [81.1%] White), and the VHA cohort included 90 749 men (27 412 [30.2%] African American and 63 337 [69.8%] White) (Table 1). The median follow-up for the SEER cohort was 5.3  (interquartile range [IQR] = 2.6-8.1) years, and the VHA cohort was 4.7  (IQR = 2.4-7.6) years. In SEER, African American men were more likely than White men to present with metastatic disease (4.5% vs 3.3%), whereas no difference was seen in the VHA (3.2% vs 3.3%). Median baseline PSA was higher among African American men compared with White men in both cohorts, with a larger disparity in SEER (7.0 ng/mL vs 6.2 ng/mL) compared with the VHA (6.9 ng/mL vs 6.3 ng/mL). In SEER, African American men were slightly more likely than White men to present with a Gleason score of 8 or higher (17.6% vs 16.6%); however, in the VHA, African American patients were as likely as White patients to present with a Gleason score of 8 or higher compared with White patients (18.6% vs 19.0%).

Table 1.

Baseline characteristics of the Surveillance, Epidemiology, and End Results (SEER) and Veterans Health Administration (VHA) cohorts

Variable SEER (n = 311 691)
 VHA (n = 90 749)
African American, No. (%) NHW, No. (%) P a African American, No. (%) NHW, No. (%) P a
All 58 553 (18.9) 253 138 (81.1) 27 412 (30.2) 63 337 (69.8)
Disease characteristics
 Disease extent at diagnosis <.001 .02
 Local 50 536 (86.3) 214 863 (84.9) 25 970 (94.7) 59 746 (94.3)
 Regional 5388 (9.2) 29 993 (11.8) 578 (2.1) 1503 (2.4)
 Metastatic 2629 (4.5) 8282 (3.3) 864 (3.2) 2088 (3.3)
 PSA, median (IQR) 7.0 (5.0-12.3) 6.2 (4.6-9.4) <.001 6.9 (5.0-11.7) 6.3 (4.8-9.8) <.001
 Gleason score <.001 <.001
 ≤6 23 426 (40.0) 110 538 (43.7) 10 614 (38.7) 26 870 (42.4)
 7 24 778 (42.3) 100 680 (39.8) 11 712 (42.7) 24 452 (38.6)
 8 5878 (10.0) 22 286 (8.8) 2847 (10.4) 6262 (9.9)
 ≥9 4471 (7.6) 19 634 (7.8) 2239 (8.2) 5753 (9.1)
Sociodemographic characteristics
 Age group, y <.001 <.001
 <60 20 794 (35.5) 62 835 (24.8) 8962 (32.7) 9957 (15.7)
 60 to <70 24 478 (41.8) 107 852 (42.6) 12 890 (47.0) 33 741 (53.3)
 70 to <80 11 257 (19.2) 66 702 (26.4) 4673 (17.1) 16 153 (25.5)
 ≥80 2024 (3.5) 15 749 (6.2) 887 (3.2) 3486 (5.5)
 SES index quintilesb <.001
 1st 20 354 (34.8) 22 220 (8.8) — —
 2nd 13 993 (23.9) 39 632 (15.7) — —
 3rd 10 301 (17.6) 51 573 (20.4) — —
 4th 7 954 (13.6) 62 521 (24.7) — —
 5th 5951 (10.2) 77 192 (30.5) — —
 Income, median (IQR), $c — — 40 861 (32 069-53 464) 48 731 (40 437-61 296) <.001
 Education, %d median (IQR), — — 12.9 (9.3-18.6) 14.3 (10.2-20.5) <.001
 Diagnosis year group <.001 <.001
 2004-2007 18 598 (31.8) 88 819 (35.1) 8 035 (29.3) 20 840 (32.9)
 2008-2011 20 844 (35.6) 90 894 (35.9) 10 366 (37.8) 23 964 (37.8)
 2012-2015 19 111 (32.6) 73 425 (29.0) 9011 (32.9) 18 533 (29.3)
 US geographic region <.001 <.001
 Midwest 7022 (12.0) 24 988 (9.9) 4516 (16.5) 15 866 (25.1)
 Northeast 8896 (15.2) 45 550 (18.0) 5721 (20.9) 10 884 (17.2)
 South 26 266 (44.9) 57 831 (22.8) 14 236 (51.9) 22 688 (35.8)
 West 16 369 (28.0) 124 769 (49.3) 2939 (10.7) 13 899 (21.9)
 Marital status <.001 <.001
 Unmarried 2024 (39.3) 53 206 (21.0) 16 218 (59.2) 28 235 (44.6)
 Follow-up time, median years (IQR) 4.8 (2.3-7.8) 5.3 (2.8-8.3) <.001 4.6 (2.4-7.2) 4.8 (2.4-7.5) <.001
Open in a new tab
a

Statistical significance was assessed with χ2 tests for categorical variables, and Wilcoxon rank sum test for continuous variables. All P values calculated were 2-sided. — = not applicable to cohort; IQR = interquartile range; NHW = non-Hispanic White; SES = socioeconomic status.

b

SEER-defined time-dependent composite score constructed from 7 variables measuring different aspects of a census tract, including median household income, median house value, median rent, percent below 150% of poverty line, education index, percent working class, and percent unemployed. 1st = lowest, 5th = highest.

c

Zip code–level median household income.

d

Percentage of zip code with attainment of at least a bachelor’s degree.

In both cohorts, African American men were statistically significantly more likely to be diagnosed with prostate cancer at a younger age compared with White men, and SES measures were statistically significantly lower among African American men. African American men were more likely to be diagnosed in the South and less likely to be diagnosed in the West.

Metastatic Disease at Diagnosis

Among men with a diagnosis of prostate cancer in SEER, African American men had statistically significantly higher odds of presenting with metastatic disease than White men (adjusted odds ratio [aOR]Race = 1.23, 95% confidence interval [CI] = 1.17 to 1.30, P < .001) (Table 2). SES was also statistically significantly associated with metastatic disease at diagnosis, with lower SES quintiles associated with progressively higher odds of presenting with metastatic disease compared with the highest SES quintile. Patients in the first SES quintile had a 67% increase in the odds of presenting with metastatic disease compared with those in the fifth SES quintile (P < .001) (Table 2).

Table 2.

Association between sociodemographic factors and metastatic disease at presentation in the Surveillance, Epidemiology, and End Results (SEER) and Veterans Health Administration (VHA) cohorts

Variable SEER VHA
Odds ratio (95% CI) Odds ratio (95% CI)
Race/Ethnicity
 Non-Hispanic White Referent Referent
 African American 1.23 (1.17 to 1.30) 1.07 (0.98 to 1.17)
Age at diagnosis, y
 <60 Referent Referent
 60 to <70 1.09 (1.03 to 1.15) 1.12 (0.99 to 1.26)
 70 to <80 1.72 (1.63 to 1.82) 1.94 (1.71 to 2.20)
 ≥80 5.76 (5.40 to 6.14) 8.49 (7.43 to 9.69)
US geographic region
 West Referent Referent
 Midwest 0.97 (0.91 to 1.04) 0.89 (0.80 to 1.00)
 Northeast 0.76 (0.72 to 0.81) 0.96 (0.86 to 1.08)
 South 0.88 (0.84 to 0.93) 0.81 (0.73 to 0.90)
Marital status
 Married Referent Referent
 Unmarried 1.76 (1.69 to 1.83) 1.22 (1.13 to 1.31)
Diagnosis year
 2004-2007 Referent Referent
 2008-2011 1.08 (1.02 to 1.13) 1.07 (0.98 to 1.18)
 2012-2015 1.73 (1.65 to 1.81) 1.40 (1.27 to 1.54)
SES quintilesa
 5th Referent —
 4th 1.18 (1.11 to 1.25) —
 3rd 1.29 (1.22 to 1.38) —
 2nd 1.42 (1.34 to 1.52) —
 1st 1.67 (1.57 to 1.83) —
Incomeb,c — 0.89 (0.76 to 1.03)
Educationd —  1.00 (0.99 to 1.01)
Open in a new tab
a

SEER-defined time-dependent composite score constructed from 7 variables measuring different aspects of a census tract, including median household income, median house value, median rent, percent below 150% of poverty line, education index, percent working class, and percent unemployed. 1st = lowest, 5th = highest. — = not applicable to cohort; CI = confidence interval; SES = neighborhood socioeconomic status.

b

Zip code–level median household income.

c

Log-transformed value used because of non-normal distribution of untransformed variable. This was treated as a continuous variable.

d

Percentage of zip code with attainment of at least a bachelor’s degree. This was treated as a continuous variable.

Among men with a diagnosis of prostate cancer in the VHA, African American men did not have statistically significantly higher odds of presenting with metastatic disease than White men (aORRace = 1.07, 95% CI = 0.98 to 1.17, P = .12) (Table 2). Unlike the SEER cohort, SES was not associated with metastatic disease presentation in the VHA (aORIncome = 0.89, 95% CI = 0.76 to 1.03, P = .11; aOREducation = 1.00, 95% CI = 0.99 to 1.01, P = .72).

Mortality

The 8-year cumulative incidence of PCSM in SEER was statistically significantly higher in African American (7.1%, 95% CI = 6.9% to 7.4%) compared with White men (5.5%, 95% CI = 5.4% to 5.6%) (P < .001) (Figure 1, A), whereas in the VHA cohort, the 8-year cumulative incidence of PCSM was not statistically significantly different between African American (5.5%, 95% CI = 5.2% to 5.9%) and White men (5.4%, 95% CI = 5.2% to 5.6%) (P = .93) (Figure 1, B). In univariate Fine-Gray regression in SEER, African American men had a 32% increased risk of PCSM compared with White men (SHRRace = 1.32, 95% CI = 1.10 to 1.60; P = .004). In SEER, the association of African American race with PCSM was attenuated most by disease extent at presentation (SHRRace = 1.16, 95% CI = 1.01 to 1.35; P = .04) and by PSA (SHRRace = 0.91, 95% CI = 0.87 to 0.96; P < .001), individually, and entirely when adjusted for all disease characteristics—disease extent, PSA, and Gleason score (SHRRace = 1.04, 95% CI = 0.93 to 1.16; P = .51) (Figure 2). Adding sociodemographic characteristics to this model did not result in further attenuation (SHRRace = 1.04, 95% CI = 0.96 to 1.12; P = .39). With adjustment for sociodemographic characteristics alone, African American men had a remaining 23% increased risk of PCSM (SHRRace = 1.23, 95% CI = 1.10 to 1.38; P < .001).

Figure 1.

Figure 1.

Open in a new tab

Cumulative incidence of prostate cancer-specific mortality (A, B) and all-cause mortality (C, D) among African American and non-Hispanic White men in the Surveillance, Epidemiology, and End Results (A, C) and Veterans Health Administration (B, D) cohorts. Two-sided P values calculated using Gray test.

Figure 2.

Figure 2.

Open in a new tab

Sequential Fine-Gray regression analysis assessing relative attenuation of the prostate cancer–specific mortality subdistribution hazard ratio (SHR) for African American race by selected covariates, with White race as reference group in (A) Surveillance, Epidemiology, and End Results and (B) the Veterans Health Administration (VHA). Relative attenuation (RA) could not be calculated for the VHA race SHR using the formula described in the Methods section as the unadjusted race SHR was 1.00—therefore there was no effect to be attenuated. Error bars represent 95% confidence interval. — = not applicable; PSA = prostate-specific antigen.

In the VHA, African American men did not have a statistically significantly increased risk of PCSM compared with White men in univariate models (SHRRace = 1.00, 95% CI = 0.93 to 1.08; P = .97) but had a slightly increased risk after adjustment for sociodemographic characteristics (SHRRace = 1.12, 95% CI = 1.04 to 1.21; P = .004) (Figure 2). When adjusted for disease characteristics in addition to sociodemographic characteristics in the VHA, the association of African American race with PCSM remained null (SHRRace = 0.93, 95% CI = 0.85 to 1.02; P = .10) (Figure 2). In our sensitivity analysis assessing interaction between race and health-care system, the interaction term was statistically significant (P < .001).

Our sensitivity analysis using imputed data found that in SEER (n = 463 402), African American race remained statistically significantly associated with an increased risk of PCSM on univariate Fine-Gray regression (SHRRace = 1.28, 95% CI = 1.24 to 1.32; P < .001), which again was entirely attenuated by the addition of disease characteristics to the model (SHRRace = 1.00, 95% CI = 0.97 to 1.03; P = .93), whereas sociodemographic characteristics alone resulted in minimal attenuation (SHRRace = 1.25, 95% CI = 1.20 to 1.29; P < .001) (Supplementary Table 2, available online). In the VHA (n = 127 788), African American race remained statistically unassociated with PCSM on univariate Fine-Gray regression (SHRRace = 1.04, 95% CI = 0.99 to 1.10).

The results of the all-cause mortality analyses are presented in Figure 1, C and D, and Supplementary Table 3 (available online) and overall followed a similar pattern as the PCSM analyses, in which African American race was associated with an increased risk for all-cause mortality in SEER but not in the VHA, including using the imputed datasets.

Discussion

The present analysis of 2 large national cancer registries demonstrates that race-based disparities in PCSM are evident in a general population cohort, SEER, but not in an equal-access health-care system, the VHA, supporting the role of health-care access in mitigating this disparity. Our results highlight the contribution of metastatic disease at diagnosis in the racial gap in PCSM, given that only in SEER were African American prostate cancer patients more likely than White patients to present with metastatic disease. Furthermore, on sequential regression, disease extent and PSA at diagnosis individually and cumulatively contributed to most of the risk of PCSM in African American men in SEER. Moreover, we demonstrate that metastatic disease at diagnosis is associated with SES within SEER, but not the VHA, suggesting that this is a marker of inequitable access to care.

Prior studies investigating cancer registry data have found that racial disparities in PCSM may be partially explained by modifiable factors such as barriers to health care, advanced disease at presentation, differential treatment, and SES (1,7,10,28-30). To our knowledge, ours is the first comprehensive report to examine this question between 2 different health system structures across all cancer stages, using a systematic approach to identify key drivers of the race-based survival gap. One study compared PCSM disparities among African American and White men in SEER, a VHA cohort, and randomized controlled trials (7), although it excluded patients with metastatic disease at presentation, which in our report was found to contribute to the PCSM gap between African American and White men. In addition, this study did not assess the role of individual disease characteristics in explaining this gap. Our results suggest that disease extent at presentation and PSA, but not Gleason score, are key components of the race-based disparity in PCSM, which is a novel finding in this multicohort context, although metastatic disease and high PSA are well-known poor prognostic factors (31). Of note, higher proportions of both African American and White men in the VHA presented with a Gleason score of 8 or higher compared with SEER. Whether this finding reflects differences in grading, patient characteristics, or other unmeasured factors between the 2 systems should be investigated in future studies.

Notably, we found that metastatic disease at presentation is associated with SES within a general population health system (SEER), but not an integrated health system (VHA), and therefore is likely influenced by racial gaps in screening and access to care. This suggests that it is not a universal truth that SES affects prognosis in prostate cancer but rather an association that is mediated by health-care access and health system structure. Indeed, prior studies have elucidated the interplay between race, SES, and prostate cancer stage at diagnosis (32,33). Altogether, this suggests that major determinants of disparities in metastatic disease at diagnosis, and subsequently PCSM, between African American and White men are systemic and modifiable and that delivering care in an integrated health system is one successful method of modifying these factors. However, one must consider the contribution of factors such as perceived racism in health-care experiences and physician mistrust among African American men, which have been associated with delayed prostate cancer screening (34-36). Subject matter experts have emphasized that structural racism remains at the root of differential access to care in the US health-care system (37,38). Furthermore, the lack of mortality disparities in the VHA does not imply a lack of racism within this system but rather a standardization of health-care access that serves to ameliorate disparities in outcomes (39). This suggests that changing policies and structures surrounding health-care access may be a temporizing solution in achieving equitable outcomes between African American and White patients as societies work more broadly to address systemic racism at the root of health inequities.

The contribution of diagnostic PSA to PCSM disparities seen in this study warrants further discussion. In both SEER and the VHA, PSA was higher at diagnosis for African Americans and attenuated the race-PCSM association such that African American race was associated with similar or improved PCSM compared with White race. This might reflect an independent relationship between PSA and PCSM disparities, or perhaps the association between PSA and disease extent (40), although the current study is unable to fully assess the mechanisms behind this finding. Furthermore, the influence of PSA screening on these disparities must be considered as well (41,42). Prior research has found a correlation between SES and prostate cancer incidence after the introduction of routine PSA screening, but not before (43), and that PSA screening rates were increased among patients of higher SES (44,45). Additional research is needed to fully evaluate the role of PSA and PSA screening in racial prostate cancer disparities.

Our study highlights the impact of nonbiological factors on race-based PCSM disparities and their potential interplay with existing research suggesting that these disparities arise from more biologically aggressive disease in African American men (46-50). There are indeed certain biological differences in prostate cancer presentation between African American and White men seen in the current and/or prior studies, such as higher diagnostic PSA, younger age of onset (51), and higher incidence (6) in African Americans; however, the current study demonstrates the contribution of systemic factors to race-based disparities after diagnosis of prostate cancer, suggesting that improved access to health care may diminish the differences in survival by race. Possible factors that lead to improved health equity in the VHA may include reduced financial barriers, integrated care delivery, and equal-access health-care coverage. It is known, for example, that the racial gap in US cancer mortality is decreased among patients older than 65 years of age, likely because of more universal Medicare coverage (52). Furthermore, additional factors specific to the VHA, such as the quality of care received or unique characteristics of the veteran population, may contribute to these findings as well.

This study has multiple limitations. The median follow-up times of 5.3 and 4.7 years for the SEER and VHA cohorts, respectively, are relatively short compared with the long natural history of prostate cancer, particularly in the context of prostate cancer mortality. We did not evaluate treatment gaps between African American and White men, as treatment data in SEER has been shown to have poor sensitivity and substantial variability by cancer site and is not recommended for estimations of proportion of individuals treated (53). Another limitation is possible data overlap between the SEER and VHA cohorts (54,55), but as the percentage of VHA patients included in SEER is approximately 3% (56), data duplication is unlikely to alter our results. Additionally, this study relies on cancer registry data to be accurately captured by hospital registrars. Although this information is encoded according to guidelines established by oversight bodies (22), it is subject to inaccuracies introduced during this process, though this limitation is common to all research using cancer registries (57,58). Last, only patients diagnosed with prostate cancer were included in this study, and although outcomes in the VHA are similar between African American and White patients diagnosed with prostate cancer, it is likely that the population mortality rate or rate of metastatic disease in African Americans at risk is higher than in Whites overall.

In conclusion, the present investigation demonstrates that disparities between African American and White men in PCSM persist in the United States on a nationwide basis but not in an integrated health-care setting. Our findings show that metastatic disease at diagnosis contributes to these disparities among African American men in SEER, suggesting that delayed diagnosis, possibly due to increased barriers to care, may in part contribute to the observed survival gap in SEER. Furthermore, metastatic disease at presentation is highly associated with SES in SEER but not in the integrated health-care setting of the VHA. As such, it is likely to be modifiable with interventions targeted at improved screening and more universal adoption of integrated models of health-care delivery. Going forward, it is vital to interrogate the role of PSA screening rates and other means of early diagnosis in reducing the disparity in PCSM between African American and White men and how these factors are modified in different health system structures. Longer-term follow-up is needed to fully assess these mortality outcomes.

Funding

This work was supported by the National Institutes of Health (TL1-TR001443 to PTC, CA132384 and CA132379 to MEM, JDM, and BSR) and the Department of Defense (W81XWH-17-PCRP-PRA to BSR).

Notes

Role of the funders: The funders listed above had no role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.

Disclosures: The authors have no disclosures.

Author contributions: DK: conceptualization, methodology, formal analysis, investigation, data curation, writing (original draft), writing (reviewing and editing), visualization. PTC: conceptualization, methodology, formal analysis, investigation, data curation, writing (original draft), writing (reviewing and editing), visualization, funding acquisition. IPG: methodology, writing (reviewing and editing). JE: writing (reviewing and editing). AK: data curation, writing (reviewing and editing). MEM: methodology, writing (reviewing and editing), funding acquisition. RM: methodology, writing (reviewing and editing). JDM: data curation, writing (reviewing and editing), funding acquisition. HP: methodology, writing (reviewing and editing). AJ: writing (reviewing and editing). TS: writing (reviewing and editing). KY: methodology, writing (reviewing and editing). BSR: conceptualization, methodology, writing (original draft), writing (reviewing and editing), supervision, funding acquisition.

Data Availability

Data available on request to bsrose@health.ucsd.edu.

Supplementary Material

djab062_Supplementary_Data
Click here for additional data file. (154.6KB, pdf)

References

  • 1. Ellis L, Canchola AJ, Spiegel D, et al. Racial and ethnic disparities in cancer survival: the contribution of tumor, sociodemographic, institutional, and neighborhood characteristics. J Clin Oncol. 2018;36(1):25–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Garner EI. Cervical cancer: disparities in screening, treatment, and survival. Cancer Epidemiol Biomarkers Prev. 2003;12(3):242s–247s. [PubMed] [Google Scholar]
  • 3. Hunt BR, Hurlbert MS.. Black:White disparities in breast cancer mortality in the 50 largest cities in the United States, 2005-2014. Cancer Epidemiol. 2016;45:169–173. [DOI] [PubMed] [Google Scholar]
  • 4. O’Keefe EB, Meltzer JP, Bethea TN.. Health disparities and cancer: racial disparities in cancer mortality in the United States, 2000-2010. Front Public Health. 2015;3:51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Ward E, Jemal A, Cokkinides V, et al. Cancer disparities by race/ethnicity and socioeconomic status. CA Cancer J Clin. 2004;54(2):78–93. [DOI] [PubMed] [Google Scholar]
  • 6. DeSantis CE, Siegel RL, Sauer AG, et al. Cancer statistics for African Americans, 2016: progress and opportunities in reducing racial disparities. CA Cancer J Clin. 2016;66(4):290–308. [DOI] [PubMed] [Google Scholar]
  • 7. Dess RT, Hartman HE, Mahal BA, et al. Association of Black race with prostate cancer-specific and other-cause mortality. JAMA Oncol. 2019;5(7):975–983. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Hoffman RM, Gilliland FD, Eley JW, et al. Racial and ethnic differences in advanced-stage prostate cancer: the Prostate Cancer Outcomes Study. J Natl Cancer Inst. 2001;93(5):388–395. [DOI] [PubMed] [Google Scholar]
  • 9. Merrill RM, Brawley OW.. Prostate cancer incidence and mortality rates among White and Black men. Epidemiology. 1997;8(2):126–131. [DOI] [PubMed] [Google Scholar]
  • 10. Riviere P, Luterstein E, Kumar A, et al. Survival of African American and non-Hispanic White men with prostate cancer in an equal-access health care system. Cancer. 2020;126(8):1683–1690. [DOI] [PubMed] [Google Scholar]
  • 11. Barocas DA, Penson DF.. Racial variation in the pattern and quality of care for prostate cancer in the USA: mind the gap. BJU Int. 2010;106(3):322–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Freedland SJ, Isaacs WB.. Explaining racial differences in prostate cancer in the United States: sociology or biology? Prostate. 2005;62(3):243–252. [DOI] [PubMed] [Google Scholar]
  • 13. Mahal BA, Aizer AA, Ziehr DR, et al. Trends in disparate treatment of African American men with localized prostate cancer across National Comprehensive Cancer Network risk groups. Urology. 2014;84(2):386–392. [DOI] [PubMed] [Google Scholar]
  • 14. Powell IJ, Dyson G, Land S, et al. Genes associated with prostate cancer are differentially expressed in African American and European American men. Cancer Epidemiol Biomarkers Prev. 2013;22(5):891–897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Shibata A, Whittemore AS.. Genetic predisposition to prostate cancer: possible explanations for ethnic differences in risk. Prostate. 1997;32(1):65–72. [DOI] [PubMed] [Google Scholar]
  • 16. Taksler GB, Keating NL, Cutler DM.. Explaining racial differences in prostate cancer mortality. Cancer. 2012;118(17):4280–4289. [DOI] [PubMed] [Google Scholar]
  • 17.National Cancer Institute Surveillance, Epidemiology, and End Results Program SEER 9 Regs Research Data, Nov 2018 Sub (1975-2016). National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch. http://www.seer.cancer.gov . Published 2018. Accessed January 2021.
  • 18. Merrill RM, Dearden KA.. How representative are the surveillance, epidemiology, and end results (SEER) program cancer data of the United States? Cancer Causes Control. 2004;15(10):1027–1034. [DOI] [PubMed] [Google Scholar]
  • 19.US Department of Veterans Affairs. Corporate Data Warehouse (CDW). https://www.hsrd.research.va.gov/for_researchers/vinci/cdw.cfm. Accessed January 2021.
  • 20.US Department of Veterans Affairs. Veterans Health Administration: about VHA. https://www.va.gov/health/aboutvha.asp. Accessed January 2021.
  • 21.Institute NC. SEER Registrar Staging Assistant TNM Data v1.9 - Prostate. https://staging.seer.cancer.gov/tnm/schema/1.9/prostate/?breadcrumbs=(∼schema_list∼). Accessed January 2021.
  • 22.US Department of Veterans Affairs. Veterans Affairs Central Cancer Registry (VACCR). https://www.data.va.gov/dataset/Veterans-Affairs-Central-Cancer-Registry-VACCR-/jvmd-8fgj. Accessed December 1, 2020.
  • 23. Yu M, Tatalovich Z, Gibson JT, et al. Using a composite index of socioeconomic status to investigate health disparities while protecting the confidentiality of cancer registry data. Cancer Causes Control. 2014;25(1):81–92. [DOI] [PubMed] [Google Scholar]
  • 24. Coogan PF, Castro-Webb N, Yu J, et al. Neighborhood and individual socioeconomic status and asthma incidence in African American women. Ethn Dis. 2016;26(1):113–122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Statist. 1988;16(3):1141–1154. [Google Scholar]
  • 26. Fine JP, Gray RJ.. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496–509. [Google Scholar]
  • 27. Birkmeyer JD, Stukel TA, Siewers AE, et al. Surgeon volume and operative mortality in the United States. N Engl J Med. 2003;349(22):2117–2127. [DOI] [PubMed] [Google Scholar]
  • 28. Jones BA, Liu WL, Araujo AB, et al. Explaining the race difference in prostate cancer stage at diagnosis. Cancer Epidemiol Biomarkers Prev. 2008;17(10):2825–2834. [DOI] [PubMed] [Google Scholar]
  • 29. Merrill RM, Lyon JL.. Explaining the difference in prostate cancer mortality rates between White and Black men in the United States. Urology. 2000;55(5):730–735. [DOI] [PubMed] [Google Scholar]
  • 30. Robbins AS, Yin D, Parikh-Patel A.. Differences in prognostic factors and survival among White men and Black men with prostate cancer, California, 1995-2004. Am J Epidemiol. 2007;166(1):71–78. [DOI] [PubMed] [Google Scholar]
  • 31. Chowdhury S, Robinson D, Cahill D, et al. Causes of death in men with prostate cancer: an analysis of 50,000 men from the Thames Cancer Registry. BJU Int. 2013;112(2):182–189. [DOI] [PubMed] [Google Scholar]
  • 32. Chornokur G, Dalton K, Borysova ME, et al. Disparities at presentation, diagnosis, treatment, and survival in African American men, affected by prostate cancer. Prostate. 2011;71(9):985–997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Coughlin SS. A review of social determinants of prostate cancer risk, stage, and survival. Prostate Int. 2020;8(2):49–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Powell W, Richmond J, Mohottige D, et al. Medical mistrust, racism, and delays in preventive health screening among African-American men. Behav Med. 2019;45(2):102–117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Shariff-Marco S, Klassen AC, Bowie JV.. Racial/ethnic differences in self-reported racism and its association with cancer-related health behaviors. Am J Public Health. 2010;100(2):364–374. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Talcott JA, Spain P, Clark JA, et al. Hidden barriers between knowledge and behavior: the North Carolina prostate cancer screening and treatment experience. Cancer. 2007;109(8):1599–1606. [DOI] [PubMed] [Google Scholar]
  • 37. Ogedegbe G. Responsibility of medical journals in addressing racism in health care. JAMA Netw Open. 2020;3(8):e2016531. [DOI] [PubMed] [Google Scholar]
  • 38. Amutah C, Greenidge K, Mante A, et al. Misrepresenting race—the role of medical schools in propagating physician bias. N Engl J Med. 2021;384(9):872–1056. [DOI] [PubMed] [Google Scholar]
  • 39. Deville C Jr, Lee WR.. Reconciling outcomes for Black men with prostate cancer within and outside the Veterans Health Administration. Cancer. 2021;127(3):342–344. [DOI] [PubMed] [Google Scholar]
  • 40. Lojanapiwat B, Anutrakulchai W, Chongruksut W, et al. Correlation and diagnostic performance of the prostate-specific antigen level with the diagnosis, aggressiveness, and bone metastasis of prostate cancer in clinical practice. Prostate Int. 2014;2(3):133–139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Carpenter WR, Howard DL, Taylor YJ, et al. Racial differences in PSA screening interval and stage at diagnosis. Cancer Causes Control. 2010;21(7):1071–1080. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Powell IJ, Vigneau FD, Bock CH, et al. Reducing prostate cancer racial disparity: evidence for aggressive early prostate cancer PSA testing of African American men. Cancer Epidemiol Biomarkers Prev. 2014;23(8):1505–1511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Liu L, Cozen W, Bernstein L, et al. Changing relationship between socioeconomic status and prostate cancer incidence. J Natl Cancer Inst. 2001;93(9):705–709. [DOI] [PubMed] [Google Scholar]
  • 44. Guessous I, Cullati S, Fedewa SA, et al. Prostate cancer screening in Switzerland: 20-year trends and socioeconomic disparities. Prev Med. 2016;82:83–91. [DOI] [PubMed] [Google Scholar]
  • 45. Rundle A, Neckerman KM, Sheehan D, et al. A prospective study of socioeconomic status, prostate cancer screening and incidence among men at high risk for prostate cancer. Cancer Causes Control. 2013;24(2):297–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Huang FW, Mosquera JM, Garofalo A, et al. Exome sequencing of African-American prostate cancer reveals loss-of-function ERF mutations. Cancer Discov. 2017;7(9):973–983. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Mahal BA, Alshalalfa M, Spratt DE, et al. Prostate cancer genomic-risk differences between African-American and White men across Gleason scores. Eur Urol. 2019;75(6):1038–1040. [DOI] [PubMed] [Google Scholar]
  • 48. Mahal BA, Berman RA, Taplin ME, et al. Prostate cancer-specific mortality across Gleason scores in Black vs nonblack men. JAMA. 2018;320(23):2479–2481. [DOI] [PubMed] [Google Scholar]
  • 49. Martin DN, Starks AM, Ambs S.. Biological determinants of health disparities in prostate cancer. Curr Opin Oncol. 2013;25(3):235–241. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Smith ZL, Eggener SE, Murphy AB.. African-American prostate cancer disparities. Curr Urol Rep. 2017;18(10):81. [DOI] [PubMed] [Google Scholar]
  • 51. Tsodikov A, Gulati R, de Carvalho TM, et al. Is prostate cancer different in Black men? Answers from 3 natural history models. Cancer. 2017;123(12):2312–2319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52. Siegel RL, Miller KD, Jemal A.. Cancer statistics, 2020. CA A Cancer J Clin. 2020;70(1):7–30. [DOI] [PubMed] [Google Scholar]
  • 53. Noone AM, Lund JL, Mariotto A, et al. Comparison of SEER treatment data with Medicare claims. Med Care. 2016;54(9):e55–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54. Zullig LL, Smith VA, Jackson GL, et al. Colorectal cancer statistics from the Veterans Affairs Central Cancer Registry. Clin Colorectal Cancer. 2016;15(4):e199–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.VHA DIRECTIVE 1412: US Department of Veterans Affairs Cancer Registry System. Department of Veterans Affairs, Veterans Health Administration. https://www.va.gov/vhapublications/ViewPublication.asp?pub_ID=8368. Accessed December 1, 2020.
  • 56. Howlader N, Ries LA, Stinchcomb DG, et al. The impact of underreported Veterans Affairs data on national cancer statistics: analysis using population-based SEER registries. J Natl Cancer Inst. 2009;101(7):533–536. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57. Gliklich RE, Dreyer NA, et al. (eds). Registries for Evaluating Patient Outcomes: A User’s Guide. Rockville, MD: Agency for Healthcare Research and Quality U.S. Department of Health and Human Services; 2014. [PubMed] [Google Scholar]
  • 58. Sarkar RR, Courtney PT, Bachand K, et al. Quality of care at safety-net hospitals and the impact on pay-for-performance reimbursement. Cancer. 2020;126(20):4584–4592. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

djab062_Supplementary_Data
Click here for additional data file. (154.6KB, pdf)

Data Availability Statement

Data available on request to bsrose@health.ucsd.edu.

Articles from JNCI Journal of the National Cancer Institute are provided here courtesy of Oxford University Press

RESOURCES