Impact of race, socioeconomic status, and the health care system on the treatment of advanced-stage ovarian cancer in California

OBJECTIVE: We sought to investigate the impact of race, socioeconomic status (SES), and health care system characteristics on receipt of speciﬁc components of National Comprehensive Cancer Network guideline care for stage IIIC/IV ovarian cancer. STUDY DESIGN: Patients diagnosed with stage IIIC/IV epithelial ovarian cancer between Jan. 1, 1996, through Dec. 31, 2006, were identiﬁed from the California Cancer Registry. Multivariate logistic regression analyses evaluated differences in surgery, chemotherapy, and treatment sequence according to race, increasing SES (SES-1 to SES-5), and provider annual case volume. RESULTS: A total of 11,865 patients were identiﬁed. Median age at diagnosis was 65.0 years. The overall median cancer-speciﬁc survival was 28.2 months. African American race (odds ratio [OR], 2.04; 95% conﬁdence interval[CI], 1.45 e 2.87)andcarebya low-volumephysician (OR, 19.72; 95% CI, 11.87 e 32.77) predicted an increased risk of not undergoing surgery. Patients with SES-1 (OR, 0.71; 95% CI, 0.60 e 0.85) and those treated at low-volume hospitals (OR, 0.88; 95% CI, 0.77 e 0.99) or by low-volume physicians (OR, 0.80; 95% CI, 0.70 e 0.92) were less likely to undergo debulking surgery. African American race (OR, 1.55; 95% CI, 1.24 e 1.93) and SES-1 (OR, 1.80; 95% CI, 1.35 e 2.39) were both signiﬁcant predictors of not receiving chemotherapy. African American patients were also more likely than whites to receive no treatment (OR, 2.08; 95% CI, 1.45 e 2.99) or only chemotherapy (OR, 1.55; 95% CI, 1.10 e 2.18). Patients with low SES were more likely to receive no treatment (OR, 1.95; 95% CI, 1.44 e 2.64) or surgery without chemotherapy (OR, 1.67; 95% CI, 1.38 e 2.03). CONCLUSION: Among patients with advanced-stage ovarian cancer, African American race, low SES, and treatment by low-volume providers are signiﬁcant and independent predictors of receiving no surgery, no debulking surgery, no chemotherapy, and nonstandard treatment sequences.

O varian cancer is the second most common gynecologic cancer in the United States, with >22,000 cases diagnosed each year. 1 Because most patients present with advanced disease, >14,000 deaths are attributed to ovarian cancer annually. Significant survival gains have followed the widespread adoption of cytoreductive surgery and combined chemotherapy regimens, but improvements have not been distributed equally among races or socioeconomic categories. While 5-year survival in white women with ovarian cancer increased from 37-45% from 1975 through 2006, 5-year survival among African American patients decreased from 43-37% over the same time period. 2 Although biologic, socioeconomic, and cultural differences have been cited as reasons for this disparity, the widening survival gap suggests that African American patients have not benefited from recent improvements in ovarian cancer care, and it highlights treatment factors as important contributors to the survival disparity. Previous studies have found that African American patients with ovarian cancer are less likely to receive primary cytoreductive surgery, appropriate chemotherapy, and National Comprehensive Cancer Network (NCCN) guidelineeadherent care. 3,4 However, the specific deviations from recommended treatment programs have not been well defined. The objective of this study was to examine disparities in the quality of ovarian cancer care across a large, statewide population, as well as to identify specific treatment components that contribute to the receipt of nonstandard therapy in patients with advanced-stage ovarian cancer.

MATERIALS AND METHODS
This was a retrospective populationbased case study of primary invasive epithelial ovarian cancers reported to the California Cancer Registry from Jan. 1, 19961, , through Dec. 31, 2006. The study received exempt status from the Institutional Review Board of the University of California, Irvine (Human Sub-jects#2011-8317). The California Cancer Registry is a standardized, qualitycontrolled population-based cancer surveillance registry that has collected information about tumor characteristics, patient characteristics, diagnosis, and treatment for all cancers diagnosed in California since 1988. Case reporting in the state is estimated to be 99%, and follow-up completion rates are >95%. 2,4-7 International Classification of Diseases for Oncology, Second Edition was used to identify tumor location and histology. Cases were identified using the ovarian Surveillance, Epidemiology, and End Results Program (SEER) primary site code (C569). 8 The study population consisted of women at least 18 years of age who were diagnosed with primary advanced-stage epithelial ovarian cancer from Jan. 1, 19961, , through Dec. 31, 2006. There were 21,044 incident ovarian cancer cases identified during the time period with follow-up continuing through January 2008. After sequentially excluding borderline tumors; germ-cell tumors; sex cord tumors; cases with missing International Classification of Diseases for Oncology, Second Edition morphology codes; cases prepared solely from autopsy or death certificates; and cases with unknown or incomplete surgery, chemotherapy, or hospital information, 18,327 cases of all stages remained. As this study included only patients with stage IIIC or IV ovarian cancer, a total of 11,865 cases were finally analyzed.
Explanatory variables included patient, tumor, and health care provider characteristics. Race/ethnicity was categorized into 4 groups: white, African American, Hispanic, and Asian/Pacific Island. Patient insurance type was grouped into 4 categories: private insurance (managed care, health maintenance organization, preferred provider organization, or other private insurance), Medicaid, Medicare, or other insurance type. Socioeconomic status (SES) was classified into 5 categories: lowest, lower-middle, middle, higher-middle, and highest SES based on quintiles of Yost's index of socioeconomic status (YOSTSCL) score. 9 Age at diagnosis was used as either a continuous variable or categorical variable with groups including those age <45, 45-54, 55-69, and !70 years.
Hospital volume was derived based on the average number of ovarian cancer cases treated at each hospital annually. Hospitals with !20 cases per year were classified as high-volume hospitals; hospitals with <20 cases per year were low volume. Physician volume was derived from the average number of cases treated annually by each physician (surgeon, medical oncologist, or attending physician). Physicians involved in !10 cases per year were considered high volume.
Outcome variables included the concordance of surgery type, chemotherapy type, and treatment sequence with NCCN treatment guidelines. [10][11][12][13][14] Surgery type was classified as follows: no surgery, oophorectomy with or without hysterectomy, oophorectomy with omentectomy, and/or debulking surgery. Chemotherapy type was categorized into 4 groups including multiple-agent chemotherapy, singleagent chemotherapy, no chemotherapy despite recommendation, and no chemotherapy for other reason. Treatment sequence had 6 categories: surgery and adjuvant chemotherapy, neoadjuvant chemotherapy and surgery, surgery and chemotherapy in unknown sequence, surgery only, chemotherapy only, and no surgery or chemotherapy.
Differences among treatment groups (surgery, chemotherapy, and treatment sequence) were analyzed with c 2 or Fisher exact test. A multinomial logistic regression model was used to perform multivariate analysis for outcomes with >2 categories. The guideline-adherent treatment category was used as the referent for each outcome variable, and binary logistic regression was performed for this outcome. Race and SES were interpreted as independent variables, as interaction terms for these variables were not significant.

Population characteristics
Patient, tumor, and provider characteristics are shown in Table 1 19.4% for patients in the highest SES category (SES-5), although these values were not statistically significant. SES-1 patients were significantly less likely than SES-5 patients to receive debulking surgery (OR, 0.71; 95% CI, 0.60e0.85).

Chemotherapy
When compared to whites, African American patients were significantly more likely to receive no chemotherapy (OR, 1.55; 95% CI, 1.24e1.93) or singleagent chemotherapy (OR, 1.42; 95% CI, 1.04e1.93). Of African American patients, 28% were found to have received no chemotherapy with "other" reason supplied, while only 4.6% of African American patients (compared to 6.7% of whites and 7.2% of Asian/Pacific Islanders) did not receive chemotherapy despite practitioner recommendation (Table 3).
There was a statistically significant inverse linear relationship between SES quintile and nonreceipt of chemotherapy. SES-1 had the highest risk of

Treatment sequence
Order of treatments also differed by race and SES (     Research Gynecology

ajog.org
It is clear that quality of care is an important factor in ovarian cancer outcomes. Several retrospective studies associate optimal cytoreduction and platinum/taxane chemotherapy regimens with improved survival, [16][17][18] and a prior analysis of California Cancer Registry data by Bristow et al 19 found decreased disease-specific survival (hazard ratio, 1.18; 95% CI, 1.07e1.32) in patients who received care not adherent to NCCN guidelines. Despite the importance of appropriate care, minority and low-SES patients are also less likely to receive care that conforms to NCCN guidelines. In a recent study by Bristow et al, 3 African American race, Medicaid or uninsured status, and median household income of <$35,000/y were independently associated with lower rates of NCCN guidelineeadherent care, and non-NCCN guidelineeadherent care was an independent predictor of shorter overall survival. Still, specific aspects of treatment deviations within the context of NCCN ovarian cancer treatment guidelines have not been well defined. The objective of this study was to investigate differences in ovarian cancererelated surgical procedures, chemotherapy regimens, and treatment sequence according to racial and socioeconomic classification in women with stage IIIC/IV disease and identify which of these factors contribute to the observed deviations from guideline care.
Strengths of the current study include the large study population, the reliability of the California Cancer Registry, and the inclusion of a recent time period during which there were no major changes to treatment guidelines. Treatment during this time period is presumed to be homogenous, as it occurred prior to the addition of intraperitoneal chemotherapy was widely adopted in response to GOG-172. There are also several limitations. First, this was a retrospective study using a population-based data set. This type of data carries an inherent risk of election and reporting bias. We were also unable to control for unreported variables that could influence the likelihood of receiving recommended care.
Such variables include the presence of medical comorbidities, the extent of initial disease, and cumulative chemotherapy dose and intensity. We were also unable to analyze the complexity of the surgical procedures, amounts of residual disease, or physician specialty, as this information is not recorded by the California Cancer Registry.
Despite these limitations, the current data offer several new observations that could account for the disparities in ovarian cancer survival. Race and SES are significantly and independently associated with specific treatment elements contributing to NCCN guideline care, and a linear association is noted between decreasing SES and an increasing risk of not receiving appropriate surgery or chemotherapy. Because debulking surgery is the cornerstone of modern ovarian cancer treatment, disparities in surgical care likely account for much of the survival gap in this disease. Our results show that African American patients are significantly less likely to undergo any surgery and more likely to undergo inappropriate surgery (ie, removal of an ovarian mass without staging or debulking). Lower SES was not associated with a statistically significant increase in the risk of not undergoing surgery for ovarian cancer, although patients classified as SES-1 were less likely to receive debulking surgery. While higher rates of comorbid conditions in African American and low-income patients could contribute to the decreased rate of surgical intervention in these groups, such conditions are unlikely to completely account for the disparities seen in this study. Both low SES and African American race were associated with nonreceipt of chemotherapy, although the strongest association was seen with decreasing SES. Decreased rates of chemotherapy seen in low-SES patients may be explained by lack of insurance funding for chemotherapy, as "other" payer status was the only payer group statistically associated with no chemotherapy despite provider recommendations.
While additional research is needed to further characterize the survival gap in ovarian cancer, the current study highlights several areas where survival gains could be made. Interventions should focus on improving access to high-volume providers and hospitals that provide NCCN guidelineeadherent regimens. Further research should aim to define other reasons for deviation from guidelines, control for variation due to differences in medical comorbidities, and develop appropriate riskadjusted measurement models. -