Racial differences in the surgical care of Medicare beneficiaries with localized prostate cancer

Abstract

INTRODUCTION

Prostate cancer (PCa) is the most frequently diagnosed non-cutaneous cancer in the male US population, with an estimated 233,000 new cases in 2014.¹ The management of PCa is driven by many factors including the severity of disease at presentation. Definitive therapy for localized PCa with curative intent is performed with radical prostatectomy (RP), radiotherapy (RT), or combinations thereof, and has shown to decrease PCa-specific mortality and improve overall survival, especially in patients with intermediate to high-risk disease.^2–6

Compelling data suggest that race and ethnicity strongly correlate with survival following a PCa diagnosis.^7–12 Godley et al found that Blacks suffer from higher PCa-specific mortality compared to Whites,¹¹ and this gap may be widening.¹³ The underlying reasons are unclear, but likely result from complex biological, cultural and sociodemographic differences. Nonetheless, there is evidence for a substantial disparity in the quality of received care. Some studies demonstrated substantial variability in treatment selection of racial and ethnic minorities, as well as inconsistency in outcomes of treatment.^11,14–16 Underwood et al showed that Black and Hispanics were less likely to receive definitive therapy than Whites,¹⁰ which has prompted investigators to hypothesize that a large part of care disparities stem from lower rates of definitive treatment for Blacks.

Based on these considerations, we assessed the effect of race on quality of care and survival of men receiving RP as definitive treatment of localized PCa. By restricting our cohort to a more homogeneous group of surgical candidates enrolled in Medicare, we sought to attenuate the effect of unmeasured confounders on the outcomes of treatment for PCa.

MATERIALS AND METHODS

Study cohort

The exclusion process for our cohort is illustrated in Figure 1. After selection, 26,482 men with localized PCa who underwent RP within the first year of PCa diagnosis remained for final analyses.

Open in a separate window

Figure 1

Flowchart of patient selection

Depiction of inclusion and exclusion criteria for the cohort of 26,482 men undergoing radical prostatectomy for localized prostate cancer. AJCC = American Joint Committee on Cancer, PCa = Prostate Cancer

Definitive treatment within 12 months of diagnosis was assessed by searching inpatient claims from the Medicare Provider Analysis and Review file, based on ICD 9^th Edition Clinical Modification (ICD-9-CM) codes, and physician claims in the Carrier Claims file, based on American Medical Association Current Procedural Terminology and ICD-9-CM codes for PCa diagnosis and procedure codes.

RESULTS

Treatment and quality of care

On average, Blacks experienced a longer treatment delay than nHW men (mean 79 vs. 71 days, p=.001; eTable1). In multivariate analyses, Blacks were less likely to receive RP within 3 months of diagnosis (Odds ratio [OR] (95% Confidence interval [CI]: 0.65 (0.59–0.71), p<.001; Table 2) and the top 50% had an absolute treatment delay of 7 days (95% Confidence Limit [CL] 3.64–10.13, eTable 2). This difference persisted at 6 and 9 months, where 18.0 and 12.3% of Blacks vs. 11.0 and 7.0% of nHWs have not had surgery, respectively (p<.001). Overall, 57.7% of Blacks underwent surgery without further adjuvant therapy as compared to 61.3% nHWs (p=.001). In trend analyses of all patients undergoing RP from 1992–2009, this gap has prevailed since the late 1990’s (eFigure 1). When surgery was performed, Blacks were less likely to undergo LND compared to nHWs (52.8 vs. 61.5%; OR: 0.76, 95% CI: 0.65–0.8, p<.001). This difference persisted in sensitivity analyses restricted to patients with intermediate and high-risk disease (OR: 0.96, 95% CI: 0.77–1.1, p=.33 and OR: 0.92, 95% CI: 0.67–1.05, p=.13, respectively; data not shown). In univariate analyses, Black race was correlated with more postoperative complications, ED visits, readmissions, mortality and less transfusions (all p<.05; eTable 1). In multivariable analyses, Blacks had increased odds of ED visits within 30 days and >30 days following RP (OR: 1.48, 95% CI: 1.18–1.86 and OR: 1.45, 95% CI: 1.19–1.8) and readmissions within 30 days and >30 days (OR: 1.28, 95% CI: 1.02–1.6 and OR: 1.27, 95% CI: 1.07–1.51) compared to nHWs (Table 2, Figure 2).

Open in a separate window

Figure 2

Forrest Plots of outcome measures and mortality

Odd’s ratios (OR) and Hazard ratios (HR) of outcome measures and mortality for Blacks (vs. non-Hispanic Whites) after radical prostatectomy (RP) for localized prostate cancer (PCa). ED = Emergency department

Table 2

Adjusted multivariable conditional logistic regression analyses testing quality of care measures in 26,482 men undergoing radical prostatectomy for localized prostate cancer within the first year after diagnosis stratified according to ethnicity, SEER-Medicare 1992–2009.

	Blacks vs. Non-Hispanic Whites
	OR (95% CI)	p-value
Surgery within 3 month of PCa Diagnosis
Non-Hispanic Whites	1 (Ref.)
African Americans	0.65 (0.59–0.71)	<.001
Any radiotherapy
Non-Hispanic Whites	1 (Ref.)
African Americans	0.89 (0.80–1.0)	.05
Adjuvant Radiotherapy*
Non-Hispanic Whites	1 (Ref.)
African Americans	1.17 (1.00–1.38)	.05
Lymph node dissection
Non-Hispanic Whites	1 (Ref.)
African Americans	0.76 (0.656–0.80)	<.001
Transfusion at 30 days
Non-Hispanic Whites	1 (Ref.)
African Americans	0.84 (0.62–1.15)	.27
Any 30-day complications
Non-Hispanic Whites	1 (Ref.)
African Americans	1.01 (0.90–1.13)	.88
ED visits within 30 days of treatment
Non-Hispanic Whites	1 (Ref.)
African Americans	1.48 (1.18–1.86)	<.001
Readmission within 30 days
Non-Hispanic Whites	1 (Ref.)
African Americans	1.28 (1.02–1.61)	.04
Transfusion >30 days
Non-Hispanic Whites	1 (Ref.)
African Americans	0.98 (0.61–1.13)	.24
Any complications >30 days
Non-Hispanic Whites	1 (Ref.)
African Americans	1.0 (0.88–1.09)	.64
ED visits >30 days of treatment
Non-Hispanic Whites	1 (Ref.)
African Americans	1.45 (1.19–1.76)	<.001
Readmission >30 days
Non-Hispanic Whites	1 (Ref.)
African Americans	1.27 (1.07–1.51)	.006

Open in a separate window

Models adjusted for all Table 1 covariates and HSA

Abbreviations: Ref. = Referent (variable)

^*within 6 months after surgery

With regard to additional cancer therapies, we recorded a significantly shorter period from PCa diagnosis to RT after RP for Black men in the top 50% of our cohort. Specifically, Blacks experienced RT about 95 days earlier compared to nHWs (CL: −149 to −42 days, p=.001; eTable 2). They were also more likely to receive ADT (p=0.001). Additionally, once secondary cancer treatment (ADT, RT) was administered, time to treatment was significantly shorter than in nHW (eTable 1).

Median total calculated costs were 13,015 USD (IQR: 8,279–19,314) for Blacks compared with 15,758 USD (IQR: 8,439–17,080) for nHWs. The surgical management of Blacks was associated with higher incremental annual cost with the top 50% spending 1,186 USD more compared to nHWs (95% CI: 805–1,566; p<.001; eTable 2).

COMMENT

There is abundant controversy vis-à-vis the gap in PCa outcomes between Blacks and nHWs. Whilst population-based data suggest higher PCa-specific mortality across all tumor stages for Blacks,²⁸ reports originating from equal-access health care delivery systems have shown equivalent survival across all races after adjusting for stage at diagnosis and treatment.²⁹ Consequently, Underwood et al speculated that the survival disparities originate from the receipt (or lack) of definitive treatment in Blacks and Hispanics.¹⁰ Over the past decade, disparities in definitive treatment decreased significantly in Hispanics, whereas they have persisted in Blacks.³⁰ However, there is also biological evidence that PCa in Blacks is more aggressive than in nHWs, thus providing an alternate explanation for the differences. In the current article, we investigate disparities in elderly Black vs. nHW men who have chosen to undergo RP as definitive treatment for localized PCa.

Our study carries several major findings. First, we noticed a significant gap in of RP utilization between Blacks and nHWs. As postulated by Underwood and others^10,31, this shortcoming is likely responsible for an important share of disparities in PCa survival between Blacks and nHWs. Most importantly, as derived from trend analyses, our findings suggest that this gap has not significantly improved over time, which raises concerns that this problem is not being adequately addressed.

Second, we identified significant differences in the quality of care received by Blacks relative to nHWs. For example, we found that Blacks were less likely to undergo pelvic LND at the time of surgery. Although LND may be safely avoided in the context of low-risk disease, there are clear recommendations mandating a template lymphadenectomy for patients with a predicted probability of lymph node invasion above 2%,^32,33 which translates into the need for LND in most patients with intermediate and high-risk disease. In adjusted analyses of these subsets of patients, the disparities persisted. However, when we accounted for regional patterns of care by adjusting for HSA, Blacks with intermediate and high-risk disease were just as likely to undergo LND. Such findings emphasize that the geographic variation in quality of care is tightly linked to racial disparities, and may thus account for a significant proportion of the differences.³⁴ Moreover, we found a difference in the time from diagnosis to treatment between Blacks and nHWs. While the difference was clinically small (8 days, p=.001), this may be most significant in those with locally advanced disease. Indeed, O’Brien et al demonstrated, that treatment delay >6 months led to pathological upgrading, worse RP outcomes and higher rate of biochemical recurrence (BCR) in localized PCa.³⁵ Finally, we showed that racial disparities in pre- and perioperative care persist in surgical postoperative outcomes. In our study, Blacks had higher rates of ED visits and readmissions. The differences in comorbidities may ascertain for higher postoperative health care utilization rates among Blacks, as described previously.^20,31,36 However, our findings are significant as they account for measured confounders with multivariable adjustment, as well as other unmeasured confounders, given that individuals deemed ‘unfit’ for surgery were excluded. In sensitivity analyses, we could not find significant differences in surgeon caseload, experience or training between Blacks and nHWs. Therefore, the gap in perioperative outcomes could not be attributed to differences in provider characteristics.

In addition to poorer quality of care and postoperative outcomes, our study finds that the cost of care for Blacks treated with RP is higher than in nHWs. As modest projections suggest that cost for PCa care will reach 18 billion USD by the end of this decade,³⁷ important work needs to be done to optimize and improve the value of PCa treatment paradigms. Much of the burden in increased costs has been attributed to the adoption of new technologies like robotic surgery or proton-beam therapy.³⁸ Although access to these technologies might be improving, the data is adamant that Blacks are in fact still discriminated from such new technologies.³⁹ Therefore, it is unlikely that such phenomenon is responsible for the increased financial burden of RP in Blacks. A careful examination of the data would suggest that the more prevalent use of RT and ADT in Blacks may be partly responsible.⁴⁰ Moreover, indicators of poorer quality of care like increased rates of ED visits and readmissions in Blacks may also ascertain for the increased expenditures.

The next major finding of our study shows that, despite important constellations of poor quality of care for Blacks undergoing RP, we did not detect significant differences in overall and cancer-specific survival. This is a remarkable shift from the generally accepted paradigm of worse PCa-survival in Blacks. However, most studies supporting these claims were unable to adjust for significant predictors of survival, and were subject to many unmeasured confounders which may have affected oncological outcomes.^41,42 Indeed, by excluding men who refused/were not offered surgical treatment for PCa, we are selecting a cohort of individuals of Black men who may be more directly comparable to nHWs than in previous studies, as they were deemed ‘fit’ for surgery. In those who make it to the operating table, despite poorer surgical quality of care as elucidated above, their survival rates were equivalent. To account for the relatively short follow-up (7.6 years), we performed sensitivity analyses restricting the cohort to patients diagnosed between 1992–1999 (mean follow-up 12.5 years); our findings were similar with regards to overall (p=.78) and PCa-specific mortality (p=.77), which is quite significant: even in earlier years, no significant disparity in long-term oncological outcomes were detected in this subset of patients. Finally, we examined the regional variation in mortality within our cohort. Although worse overall survival for Blacks was recorded in the South (OR 1.26, 95% CI 1.06–1.5, data not shown), no difference was found between regions with regard to PCa-specific mortality. The difference in overall mortality shown here is consistent with previous evidence,⁴³ however the lack of regional variation with regard to PCa-specific mortality further reinforces our findings. A possible interpretation of our findings is that the biological differences in tumor aggressiveness among Blacks may have been exaggerated, and that the perceived gap in survival is a result of lack of access or cultural perceptions with regard to surgical care for PCa or other factors that differentiate who makes it to the operating table.

Despite its strengths, our study has limitations, which are inherent to retrospective, observational studies relying on SEER-Medicare. Several key unmeasured confounders are not captured in administrative claims and may cause an underestimation of the severity of comorbidities.^44,45 While Blacks had less severe disease characteristics than nHWs, the exclusion of patients who refused or were not offered treatment may have introduced a bias by selecting only the healthiest Blacks for surgery. Although evidence for provider-specific screening and treatment recommendation disparities exist in PCa^46,47, this issue was outside the scope of this study. However, the contingent provider-specific selection bias would lead to selection of surgical patients with favorable unmeasured disease characteristics such as tumor volume, PSA, and number of positive cores. The lack of such preoperative characteristics may impede primary and secondary treatment choices after BCR. While our analyses adjusted for all available socioeconomic confounders, unmeasured sociodemographic variables other than race may explain the observed differences between Blacks and nHWs.^48,49 It is important to consider that our study comprises only traditional Medicare enrollees >65 years, thus limiting the general application of our findings. Specifically, we acknowledge that most men undergoing RP are younger and hold private insurance. Thus, our findings may not be applicable to the general population of men undergoing RP

To summarize, we provide robust evidence for the existence of a substantial difference in the quality of surgical care of PCa in Blacks. As the unfavorable quality of care did not translate into worse overall and cancer-specific survival in our sample, the commonly perceived detrimental survival in Black PCa patients may be the sequelae of barriers and selection bias in definitive treatment. The public and professional awareness needs to be raised to address these concerning issues and identify their underlying causes.

Supplementary Material

Acknowledgments

Footnotes

REFERENCES