Variation in quality of care among older men with localized prostate cancer

Authors

  • Ravishankar Jayadevappa PhD,

    Corresponding author
    1. Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
    2. Leonard Davis Institute of Health Economics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
    • Department of Medicine, University of Pennsylvania, 224, Ralston-Penn Center, 3615 Chestnut Street, Philadelphia, PA 19104-2676
    Search for more papers by this author
    • Fax: (215) 573-8684

  • Sumedha Chhatre PhD,

    1. Department of Psychiatry, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Jerry C. Johnson MD,

    1. Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
    Search for more papers by this author
  • S. Bruce Malkowicz MD

    1. Division of Urology, Department of Surgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
    Search for more papers by this author

Abstract

BACKGROUND:

The objective of this study was to assess the racial and ethnic disparities in outcomes and their association with process-of-care measures for elderly Medicare recipients with localized prostate cancer.

METHODS:

The Surveillance, Epidemiology, and End Results-Medicare databases for the period from 1995 to 2003 were used to identify African-American men, non-Hispanic white men, and Hispanic men with localized prostate cancer, and data were obtained for the 1-year period before the diagnosis of prostate cancer and up to 8 years postdiagnosis. The short-term outcomes of interest were complications, emergency room visits, readmissions, and mortality; the long-term outcomes of interest were prostate cancer-specific mortality and all-cause mortality; and process-of-care measures of interest were treatment and time to treatment. Cox proportional hazards regression, logistic regression, and Poisson regression were used to study the racial and ethnic disparities in outcomes and their association with process-of-care measures.

RESULTS:

Compared with non-Hispanic white patients, African-American patients (Hazard ration [HR], 1.43; 95% confidence interval [CE], 1.19-1.86) and Hispanic patients (HR=1.39; 95% CI, 1.03-1.84) had greater hazard of long term prostate specific mortality. African-American patients also had greater odds of emergency room visits (odds ratio, 1.4; 95% CI, 1.2-1.7) and greater all-cause mortality (HR, 1.39; 95% CI, 1.3-1.5) compared with white patients. The time to treatment was longer for African-American patients and was indicative of a greater hazard of all-cause, long-term mortality. Hispanic patients who underwent surgery or received radiation had a greater hazard of long-term prostate-specific mortality compared with white patients who received hormone therapy.

CONCLUSIONS:

Racial and ethnic disparities in outcomes were associated with process-of-care measures (the type and time to treatment). The current results indicated that there is an opportunity to reduce these disparities by addressing these process-of-care measures. Cancer 2011. © 2010 American Cancer Society.

Prostate cancer is the most common malignancy among men in the United States. The burden of prostate cancer on the healthcare system is immense in both human and economic terms.1 Added to this, substantial disparities exist in the pattern of prostate cancer care and outcomes across institutions, regions, age groups, and ethnic groups.1-8 Racial and ethnic disparities also have been documented in various phases of prostate cancer care. Improving the quality of healthcare across racial and ethnic groups requires attention to the process and outcomes of health services rendered to individuals with adequate adjustments for various risk factors and personal characteristics.2, 8, 9 Although it is a priority in the health policy arena, quality of care continues to be uneven.1, 2, 8 Addressing the variation in quality of care is widely regarded as an important healthcare policy objective, and racial ethnic disparities in quality of care are recognized as a major quality problem. Increased variations in care patterns and outcomes are linked to a lack of agreement on the identification of quality-of-care measures.2-11

Surveillance, Epidemiology, and End Results (SEER)-Medicare–linked data have been used to document the variation in health resource utilization and outcomes for prostate cancer.4-7 Reports indicated that African-American men were less likely to receive aggressive therapy (odds ratio [OR], 0.74; 95% confidence interval [CI], 0.70-0.79) compared with non-Hispanic white men.6, 12 In addition, African-American men who received curative treatment reported a differential recovery pattern compared with non-Hispanic white men who had prostate cancer.13 Because of higher use of prostate-specific antigen screening coupled with uncertainty in screening and treatment effects, debate continues about the effectiveness of various treatments. In another study that used SEER-Medicare data by Godley et al, racial and ethnic disparities were evident both in overall survival and prostate cancer-specific survival.14 African-American patients had poorer overall survival among those who underwent surgery.14, 15 Studies indicate that treatments for a given stage of prostate cancer vary by geographic region, age, and race, and ethnicity.4-7, 12-24 In men who undergo prostatectomy, the rates of postoperative and late urinary complications are significantly reduced if the procedure is performed in a high-volume hospital and by a high-volume surgeon.16

MATERIALS AND METHODS

Conceptual Model of Quality of Care

Quality of healthcare is defined as the degree to which healthcare services for individual patients and populations increase the probability of desired health outcomes and are consistent with current professional knowledge.3, 10 Figure 1 illustrates our conceptual model of quality of care, which consists of 3 main components: structure, process, and outcomes.9, 25 The structural component consists of characteristics of hospitals, physicians, and other healthcare workers and is defined as those resources used by providers or organizations that support the delivery of care to patients.8-11 Process of care includes the way physicians and patients interact, the appropriateness and timeliness of treatment, and other clinical and nonclinical factors associated with care. The outcome is derivative of the structure and process and includes change in patient health status, satisfaction with care, health-related quality of life, and functional status.8-11

Figure 1.

This chart illustrates the quality-of-care conceptual model.

Efforts to reduce racial and ethnic disparities must acknowledge the multidimensional nature of quality of care. There is paucity of knowledge regarding the process-of-care factors that contribute to racial and ethnic disparity in prostate cancer outcomes among elderly Medicare recipients. Hence, the objective of this study was to analyze the interplay of process-of-care measures (treatment type and time to treatment) and patient characteristics and their relation with short-term outcomes (mortality, complications, readmissions, and emergency room [ER] visits) and long-term outcomes (prostate-specific and all-cause mortality) among elderly patients with prostate cancer. We used SEER-Medicare–linked data to examine the racial and ethnic disparities in outcomes and their association with quality-of-care measures for elderly African-American, Hispanic, and non-Hispanic white men who were diagnosed with localized prostate cancer. We hypothesized that some of the racial and ethnic disparity in prostate cancer outcomes can be explained by process-of-care measures.

Data Sources and Study Sample

We adopted a retrospective design using the linked SEER-Medicare database. All African-American, Hispanic, and non-Hispanic white men aged ≥66 years who were diagnosed with prostate cancer (International Classification of Diseases codes: 185, 233.4, and 236.5) between 1995 and 1998 (n = 50,147) were identified. From this cohort, we retained those men who had localized stage (n = 42,522). Of these, 8476 men were excluded because information was not available regarding their treatment and treatment date in the SEER and Medicare claims data. Thus, the final cohort included 34,046 patients. For this cohort, data were obtained for the 1-year period before the diagnosis of prostate cancer and up to 8 years postdiagnosis. The lists of procedure codes, revenue center codes, and service codes were reviewed to ensure that appropriate codes were used for each year, because Healthcare Common Procedure Coding System codes change over time.

The SEER-Medicare–linked database brings together Medicare administrative claims data and clinical tumor registry data for Medicare recipients.26 The SEER program collects data on cancer incidence, treatment, and mortality in a representative sample of the US population. The data used in this analysis included 13 SEER sites (San Francisco, Connecticut, Detroit, Hawaii, Iowa, New Mexico, Seattle, Utah, Atlanta, San Jose, Los Angeles, Rural Georgia, and Alaska), encompassing 14% of the US population. With the exception of individuals who are enrolled in health maintenance organizations (HMOs) or who do not have Part B coverage, Medicare data provide information about all inpatient and outpatient use of medical care for residents of the United States aged ≥65 years. The SEER-Medicare file integrates the individual's SEER and Medicare records into a single data file. Among individuals in the SEER registry who were diagnosed with cancer at age ≥65 years, 93% have been matched with their Medicare enrollment records in a linked, customized file—the Patient Entitlement and Diagnosis Summary File. In addition to SEER diagnostic information, this file provides Medicare entitlement, utilization, and census tract and zip code-based socioeconomic data. SEER database provides characteristics of the tumor that are crucial to adequately adjust for disease severity, including histology, stage, and grade. SEER also provides information on the extent of disease that may have prognostic significance, such as the size of the primary tumor and the extent and location of lymph node involvement. The SEER-Medicare–linked record includes service codes in 3 Medicare files: 1) the inpatient file, 2) the hospital outpatient standard analytical file (claims for outpatient facility services), and 3) the physician part B file (claims for physician and other medical provider services).

Measurement Strategy

Process-of-care measures

As depicted in the conceptual model (Fig. 1), process of care measures are: 1) treatment type and 2) time to treatment. Treatment was categorized into 3 groups. The surgery group included mono therapy (surgery alone) or multimodal therapy (surgery with external-beam radiation therapy or brachytherapy or surgery with a combination of radiation and hormone therapy). The radiation therapy group included radiation therapy alone (external-beam radiation or brachytherapy) or radiation therapy with hormone therapy (multimodal). The third treatment group consisted of hormone therapy alone. The time to treatment (number of days) was defined as the time between diagnosis and treatment.

Outcome measures

The short-term outcome measures (within 30 days from treatment date) were: 1) mortality, 2) complications, 3) the number of readmissions, and 4) the number of ER admissions. The long-term outcome measures (up to 8 years) were prostate cancer-specific mortality and overall mortality. Survival was determined by Medicare vital statistics as well as SEER linkage to death certificates (the National Death Index). Consistent with Alibhai et al, and Beard et al, using Medicare inpatient and outpatient claims data, we identified the following groups of complications that occurred within 30 days from treatment: cardiac, respiratory, vascular, wound/bleeding, genitourinary, bowel, miscellaneous medical, and miscellaneous surgical.27-31

Disease severity and demographic covariates

Disease severity was adjusted by using data on prostate cancer stage, grade, and histology from the SEER database. The Charlson comorbidity index was used to assess medical comorbidity using inpatient and outpatient Medicare claims.32, 33 We used diagnostic information from all encounters 1 year before prostate cancer diagnosis to determine comorbidity, as outlined by Klabunde et al.32, 33 Age, race and ethnicity, marital status, geographic area, and income data were obtained from the Patient Entitlement and Diagnostic Summary File.

Statistical Analysis

We tested for the underlying difference in demographic and clinical attributes between African-American, Hispanic, and non-Hispanic white patients using t tests and chi-square tests. Our analyses, as indicated in our conceptual model (Fig. 1), consisted of 2 sets of models. In the first set (Models 1-6), we analyzed the association between race and ethnicity and outcomes (short-term mortality, complications, the number of readmissions and ER visits, prostate cancer-specific mortality, and all-cause mortality). A Cox proportional-hazards model was used for mortality outcomes, logistic regression was used for modeling any complications, and Poisson models (with zero inflation corrections) were used to model the number of ER visits and readmissions after adjusting for age, marital status, income, TNM stage, grade, comorbidity, and SEER region. The primary variable of interest in all models was race/ethnicity.34, 35 In the second set of models, we analyzed the modifying effects of process-of-care measures on outcomes. Separate models were developed for treatment (Model 7) and time to treatment (Model 8). We also tested for the effect of the interaction of race/ethnicity and process-of-care measures on outcomes. Analyses were conducted using Statistical Analysis System (SAS) version 9.2 (SAS Institute Inc., Cary, NC).

RESULTS

Sample Characteristics

The study sample consisted of 34,046 men who had localized prostate cancer. Table 1 indicates that the mean age of the non-Hispanic white group was slightly older than that of the other groups. The small difference in mean age was statistically significant, possibly because of the large sample size. Also, marital status and median income differed between the 3 groups. African-American men were less likely to be married and had lower income compared with their non-Hispanic white and Hispanic counterparts. The mean Charlson comorbidity score was higher for the African-American group compared with other 2 groups.

Table 1. Baseline Demographic and Clinical Characteristics (n = 34,046)
VariableNon-Hispanic whites, n = 28,200African Americans, n = 3642Hispanics, n = 2204P
  1. SD indicates standard deviation.

Age at diagnosis: Mean ± SD, y73.1 ± 5.472.4 ± 5.172.5 ± 5.4<.0001
Geographic area, %
 Metropolitan86.898.591.9<.0001
 Urban and rural13.21.58.1 
Marital status, %
 Married75.859.373.6<.0001
 Single19.234.522 
 Unknown56.24.4 
Charlson comorbidity score: Mean ± SD0.33 ± 0.860.60 ± 1.10.08 ± 0.44<.0001
Annual median income of census tract: Mean ± SD, $US42,330 ± 19,17625,979 ± 13,23733,408 ± 13,574<.0001
Tumor grade, %
 Well differentiated10.077.611.9<.0001
 Moderately differentiated65.6266.461.5 
 Poorly differentiated/undifferentiated24.312626.6 
Tumor classification, %
 ≤T2a48.146.248.5<.0001
 T2b and T2c41.640.739.8 
 ≥T3a10.313.111.7 

Variation in Process and Outcome Measures

Unadjusted comparison of process and quality-of-care measures revealed significant variation between racial and ethnic groups (Table 2). A higher proportion of Hispanic patients underwent radical prostatectomy (monomodal or multimodal), whereas a higher proportion of African-American patients received external-beam radiation therapy (monomodal or multimodal). African-American patients had a longer mean time to treatment from the date of diagnosis (95.8 days; P < .0001) and higher all-cause mortality. A greater proportion of African-American patients had ER visits within 30 days of treatment, whereas a greater proportion of white patients had inpatient visits. Overall, the proportion of patients with any complication was greater for the non-Hispanic white and Hispanic groups compared with the African-American group.

Table 2. Unadjusted Process and Quality-of-Care Measures Across Racial and Ethnic Groups (n = 34,046)
VariableNon-Hispanic whites, n = 28,200African Americans, n = 3642Hispanics, n = 2204P
  1. SD indicates standard deviation; ER, emergency room.

Treatment, %
 Surgery (monotherapy or multimodal therapy)42.335.147.9<.0001
 Radiation therapy (monotherapy or multimodal therapy)47.550.639.3 
 Hormone therapy alone12.214.312.8 
Time to treatment, d
 Mean ± SD64.9 ± 142.395.8 ± 231.960.6 ± 111.4<.0001
 Median31.059.031.0 
ER visits within 30 d of treatment, %3.414.782.99<.0001
Inpatient hospitalization within 30 d of treatment, %33.9428.4733.71<.0001
Complications at 30 d, %
 Any22.0620.8122.87.1345
 Cardiac2.762.92.36.4375
 Respiratory1.631.761.59.6793
 Vascular0.620.550.41.9345
 Wound9.598.19.07.0004
 Genitourinary10.210.8511.12.5183
 Miscellaneous4.474.395.22.0009
 Miscellaneous surgery0.520.550.59.7803
 Bowel1.962.442.51.0075
Short-term all-cause mortality at 30 d, %0.310.380.45.4419
All-cause mortality at 8-y follow-up, %19.9925.8918.10<.0001
Prostate cancer-specific mortality at 8-y follow-up, %1.592.032.04.0549

Table 3 presents the results from Cox models for the association between race and ethnicity and outcomes with adjustments for age, marital status, TNM stage, comorbidity, geographic area, and SEER region. Race/ethnicity was not associated with short-term all-cause mortality. Compared with non-Hispanic white patients, Hispanic patients had a greater odds of developing any complications (odds ratio [OR], 1.12; 95% confidence interval [CI], 1.02-1.22). African-American patients had more ER visits and fewer inpatient visits compared with their non-Hispanic white counterparts (OR, 1.64 [95% CI, 1.42-1.89] and 0.93 [95% CI, 0.89-0.99], respectively). African-American patients (hazard ratio [HR], 1.43; 95% CI, 1.19-1.86) and Hispanic patients (HR, 1.39; 95% CI, 1.03-1.84) had a greater hazard of prostate cancer-specific mortality compared with non-Hispanic white patients. African-American patients also had a greater hazard of overall long-term mortality (HR, 1.39; 95% CI, 1.30-1.50) compared with non-Hispanic white patients.

Table 3. Association Between Race and Ethnicity and Outcomes (n = 34,046)
VariableShort Term 
 HR (95% CI)OR (95% CI)Long Term Up to 8 Years: HR (95% CI)
 Model 1: MortalityModel 2: Any ComplicationsModel 3: No. of ER VisitsModel 4: No. of Inpatient VisitsModel 5: Prostate Cancer-Specific MortalityModel 6: All-Cause Mortality
  1. HR indicates hazard ratio; CI, confidence interval; OR, odds ratio; ER, emergency room; TNM, tumor, lymph node, metastasis classification system.

Race
 African American1.52 (0.92-2.6)1.06 (0.92-1.15)1.64 (1.42-1.89)0.93 (0.89-0.99)1.43 (1.19-1.86)1.39 (1.3-1.5)
 Hispanic1.7 (0.89-3.3)1.12 (1.06-1.25)1.04 (0.83-1.29)1.02 (0.94-1.1)1.39 (1.03-1.84)0.93 (0.84-1.01)
 Non-Hispanic white, reference
Age1.2 (1.14-1.39)1.03 (1.01-1.15)1.1 (1.09-1.10)1.01 (0.99-1.02)1.13 (1.10-1.42)1.1 (1.1-1.13)
Married0.67 (0.45-0.98)0.99 (0.93-1.14)0.76 (0.68-0.85)1.0 (0.98-1.11)0.79 (0.66-0.94)0.81 (0.77-0.85)
TNM classification
 T2b andT2c0.49 (0.38-0.79)0.69 (0.66-0.74)0.72 (0.64-0.82)0.73 (0.69-0.76)1.5 (1.3-1.91)1.13 (1.1-1.2)
 ≥T3a0.93 (0.54-1.6)0.73 (0.67-0.79)1.10 (0.89-1.3)0.76 (0.72-0.81)3.6 (2.8-4.5)2.74 (2.39-2.69)
 ≤T2a, reference
Geographic area
 Metropolitan0.41 (0.27-2.7)0.55 (0.42-0.67)0.69 (0.49-1.05)0.75 (0.66-0.86)0.73 (0.39-1.3)0.78 (0.65-0.92)
 Urban0.82 (0.28-2.3)0.90 (0.74-1.11)0.99 (0.65-1.53)0.98 (0.86-1.12)1.0 (0.55-1.9)0.94 (0.78-1.12)
 Rural, reference
Charlson comorbidity score1.04 (0.73-1.33)1.07 (1.04-1.09)1.15 (1.10-1.22)0.95 (0.93-0.97)1.07 (1.04-1.10)1.07 (1.04-1.5)

Modifying Effect of Process-of-Care Measures on Racial and Ethnic Disparity in Outcomes

Models 7 and 8 present the effects of the interactions of process-of-care measures and race/ethnicity on short-term outcomes (Table 4). Model 7 indicated that the interaction between race/ethnicity and treatment type was not associated with 30-day all-cause mortality. However, Hispanic men who received radiation therapy were less likely to have complications (OR, 0.56; 95% CI, 0.33-0.95) compared with their non-Hispanic white counterparts who received hormone therapy (Table 4). Similarly, African-American patients who underwent surgery or received radiation were less likely to have more ER visits or inpatient visits compared with their non-Hispanic white counterparts who received hormone therapy. Model 8 indicated that African-American patients who had a longer time to treatment had a greater odds of more ER visits (OR, 1.4; 95% CI, 1.2-1.7). The interaction between race/ethnicity and treatment type was not associated with all-cause mortality or prostate cancer-specific mortality (Table 5). However, African-American patients who had a longer time to treatment had a greater hazard of all-cause mortality (HR, 1.1; 95% CI, 1.05-1.2) compared with non-Hispanic white patients. Hispanic patients who underwent surgery (HR, 5.4; 95% CI, 1.6-18.1) or received radiation therapy (HR, 5.4; 95% CI, 1.5-19.7) had a greater hazard of long-term prostate cancer-specific mortality compared with non-Hispanic whites who received hormone therapy.

Table 4. Association Between Race and Ethnicity, Process Variables, and Short-Term (30-Day) Outcomes (n = 34,046)
  OR (95% CI)
ModelMortality: HR (95% CI)Any ComplicationsNo. of ER VisitsNo. of Inpatient Visits
  • OR indicates odds ratio; CI, confidence interval; HR, hazard ratio; ER, emergency room.

  • a

    All models are adjusted for age, marital status, disease stage, comorbidity, and geographic area.

Model 7: Interaction between race and treatmenta
 Race
  African American1.9 (0.65-5.7)1.41 (0.97-1.8)2.7 (1.9-3.61)1.5 (1.2-1.8)
  Hispanic2.6 (0.77-8.9)1.04 (0.71-1.54)1.3 (0.71-2.2)1.19 (0.82-1.5)
  Non-Hispanic white, reference
Treatment
  Surgery1.7 (1.0-2.94)6.8 (5.71-7.7)1.9 (1.6-2.39)5.9 (4.69-6.4)
  Radiation therapy0.14 (0.05-0.39)0.61 (0.54-0.69)0.46 (0.37-0.57)0.42 (0.38-0.48)
  Hormone therapy alone, reference
 Interaction
  African American×surgery0.64 (0.17-2.43)0.77 (0.57-1.14)0.51 (0.36-0.73)0.64 (0.52-0.79)
  African American×radiation therapy1.7 (0.23-12.1)0.78 (0.55-1.12)0.52 (0.33-0.82)0.71 (0.54-0.95)
  Hispanic×surgery0.50 (0.12-2.2)0.99 (0.66-1.5)0.72 (0.38-1.4)0.79 (0.58-1.1)
  Hispanic×radiation therapy0.05 (0.10-1.3)0.56 (0.33-0.95)0.64 (0.27-1.5)0.88 (0.58-1.3)
Model 8: Interaction between race and time to treatmenta
 Race
  African American0.70 (0.05-11.1)1.17 (0.83-1.50)0.56 (0.43-0.85)1.16 (0.86-1.39)
  Hispanic0.35 (0.02-5.3)1.56 (1.08-2.52)1.3 (0.57-2.83)0.99 (0.77-1.31)
  Non-Hispanic white, reference
 Time to treatment0.19 (0.12-0.30)0.75 (0.73-0.77)0.72 (0.68-0.77)0.79 (0.78-0.81)
  African American×time to treatment1.30 (0.46-3.93)0.99 (0.89-1.14)1.40 (1.16-1.67)0.96 (0.91-1.01)
  Hispanic×time to treatment1.9 (0.67-5.12)0.91 (0.82-1.01)0.93 (0.76-1.21)0.99 (0.94-1.14)
Table 5. Association Between Race and Ethnicity, Process Variables, and Long-Term (Up to 8-Year) Outcomes (n = 34,046)
ModelHR (95% CI)
All-Cause MortalityProstate Cancer-Specific Mortality
  • HR indicates hazard ratio; CI, confidence interval.

  • a

    All models are adjusted for age, marital status, stage, comorbidity, and geographic area.

Model 7: Interaction between race and type of treatment, surgerya
 Race
  African American1.3 (1.10-2.61)1.3 (0.82-2.1)
  Hispanic0.79 (0.63-0.99)0.30 (0.09-0.95)
  Non-Hispanic white, reference
 Treatment
  Surgery0.83 (0.77-0.89)0.89 (0.71-1.1)
  Radiation therapy0.63 (0.58-0.67)0.35 (0.27-0.46)
  Hormone therapy, reference
 Interaction
  African American× surgery1.02 (0.87-1.2)1.2 (0.64-2.1)
  African American× radiation therapy1.10 (0.91-1.3)1.0 (0.54-2.1)
  Hispanic×surgery1.20 (0.88-1.5)5.4 (1.6-18.1)
  Hispanic×radiation therapy1.31 (0.96-1.7)5.4 (1.5-19.7)
Model 8: Interaction between race and time to treatmenta
 Race
  African American1.13 (0.86-1.4)1.40 (0.60-3.2)
  Hispanic0.79 (0.55-1.13)1.4 (0.49-4.2)
  Non-Hispanic white, reference
 Time to treatment0.80 (0.78-0.82)0.72 (0.65-0.79)
  African American×time to treatment1.1 (1.05-1.2)1.0 (0.82-1.3)
  Hispanic×time to treatment1.1 (0.951.2)0.98 (0.72-1.4)

DISCUSSION

Medicare-enrolled, elderly patients with prostate cancer have great variability in healthcare utilization, morbidity, and other quality outcomes.1-8 Quality of care in prostate cancer is a multidimensional construct that consists of structure, process, and outcomes. Racial and ethnic disparities in prostate cancer care are documented for each phase of care: prevention, treatment, follow-up, and terminal care.4-7, 12-24 In continuation of our earlier research on quality of prostate cancer care assessment, in the current study, we have documented the modifying effect of process-of-care measures on the racial and ethnic differences in prostate cancer outcomes using an administrative database. To begin with, we observed that, in concordance with earlier research, our results confirmed racial and ethnic disparities in prostate cancer outcomes.4-7 We observed racial and ethnic disparity in the process-of-care measures (treatment type and time to treatment). Both the mean and the median time to treatment were longer for African-American patients, who also were more likely to receive radiation therapy. The results lend substantial support to our hypothesis that some of the racial and ethnic disparity in prostate cancer outcomes can be explained by process-of-care measures. The longer time to treatment among African-American patients was indicative of a greater hazard of overall mortality as well as more ER visits. Also, the interaction between treatment and race/ethnicity was significant for the outcomes of complications, ER visits, inpatient visits, and prostate cancer-specific and overall mortality.

The journey to assessing quality-of-care measures in the arena of prostate cancer began with an exploration of the variations in outcomes across geographic regions, age groups, and racial and ethnic groups. The foundation for assessing quality of care was built by earlier RAND studies.36-43 In those studies, the Donabedian model of structure, process, and outcome was used to conceptualize the dimensions of quality of care. Quality-of-care assessment is essential to develop appropriate clinical and healthcare policy measures that address the burden exerted by prostate cancer. In a premier study, Spencer et al developed 63 potential indicators and covariates to measure quality of care in prostate cancer,40 which is a multidimensional construct that encompasses various levels of care and clinical and environmental domains.36 Given the multiple treatment options for localized prostate cancer, quality-of-care measures are useful for comparing treatment efficacy and variations across provider and geographic level attributes. Process-of-care quality indicators were identified as superior for those who received radiation therapy compared with those who underwent radical prostatectomy.38 An examination of compliance with 25 quality-of-care measures that were developed by RAND to assess structure and process indicated significant variation across hospital type and census-based geographic regions.43 A recent study demonstrated that prostate cancer treatments differed significantly between county hospitals and private providers.44 Although those studies successfully applied quality-of-care measures, clearly, there is room for further measurement and assessment of the quality of care for patients with prostate cancer.41

Our findings must be considered in light of limitations that are intrinsic to the SEER-Medicare–linked dataset. The study sample included only African-American, Hispanic, and non-Hispanic white men aged ≥66 years who lived in a SEER area and were not enrolled in an HMO. Although SEER is designed to provide a representative sample of the United States, it includes only a relatively small Hispanic population. However, the age and sex distributions for individuals aged ≥66 years in the SEER areas are comparable to those of the US elderly population. Our findings may not be generalizable to men aged <66 years or men enrolled in HMOs. Medicare claims data provide an excellent opportunity to analyze prostate cancer care in a broad population, however, these data have certain limitations. Although our analysis controlled for treatment and other covariates, there is potential for some unaccounted bias. Finally, as depicted in the conceptual model, in this study, we did not address the effect of measures of structure (such as hospital and physician attributes) and other potential process-of-care measures on outcomes. Our objective is to address these measures in future studies.

In conclusion, healthcare quality, which is a key concept for medical practice and research, also is a widely used construct in healthcare administration and market research. A major concern of policymakers in the United States is the persistent presence of unacceptable variation in quality of care across racial and ethnic groups.1-12 The first step in eliminating these disparities is to identify their determinants and minimize the overuse, under use, and misuse of health resources while accounting for individual preferences. More important, policy measures are needed to identify and implement strategies at the geographic level, health system level, and individual level. A 2-dimensional approach can be used to address the issues of quality of care. One dimension is the internal or system/individual-level effort to improve the quality of care. The second dimension is the broader or external level, in which monitoring efforts can be incorporated through collaborations among organizations, such as the American Urological Association, the National Cancer Institute, the American Society of Clinical Oncology, and the Agency for Healthcare Research and Quality. Some of the measures that can be adopted at the external level are establishing guidelines and achievable performance targets, an infusion of technology and expertise, monitoring structure and process measures to ensure enhanced outcomes, and, finally, providing the necessary funding.

Consistent with earlier studies, our study revealed ethnic variations in quality of care and outcomes. However, our data suggest opportunities to reduce these disparities. For example, the large difference in time to treatment (approximately an entire month) suggests the need for an aggressive, patient-centered approach immediately after the diagnosis is established. Systems can be modified to ensure that the initial appointment with patient and family occurs as soon as possible. During the first and subsequent visits, practitioners can demonstrate sensitivity to the emotional impact of this diagnosis, explore attitudes of suspicion and distrust, and exhibit respect for the individual as a person. Health systems can implement proactive, aggressive monitoring systems from diagnosis through treatment and early recovery to ensure that minority patients do not delay care because of a lack of understanding or miscommunication. Finally, the current data speak to a need for health systems and health plans to identify minority patients. Without this identification, issues of health equity cannot be measured and, thus, cannot be improved.

In summary, significant opportunities exists to minimize ethnic variations in the process of care for patients with prostate cancer and to improve their overall quality of care. In the era of value-based healthcare and accountability in healthcare, it is vital to consider these factors while developing appropriate performance incentives to minimize ethnic disparity in prostate cancer care. We have developed a conceptual model of quality of care that consists of elements of structure, process, and outcome of care. This model was envisioned to facilitate linkage of the multiple dimensions of quality of care, and our study is a small step in this direction. Many studies have demonstrated that the structure of care, which includes the hospital and the physician, is associated with improved outcomes. Future studies are planned to identify additional measures of structure and process and their association with racial and ethnic disparity in outcomes.

CONFLICT OF INTEREST DISCLOSURES

Supported in part by the National Cancer Institute, National Institutes of Health (Grant 5RO3CA 121338-2) and by the University of Pennsylvania Resource Center for Minority Aging Research, National Institute on Aging, National Institutes of Health (Grant 5P30AG031043-03).

Ancillary