Presented in poster form at the 2009 Annual Meeting of the American Society of Clinical Oncology; May 29-June 2, 2009; Orlando, Florida.
The New York City (NYC) public hospital system includes subspecialty care for gynecologic cancers, providing care to patients regardless of insurance status. The authors sought to determine the surgical patterns of care for ovarian cancer patients in the NYC public hospital system.
Ovarian cancer cases were identified in the New York State Department of Health Statewide Planning and Research Cooperative System database for years 2001 to 2006. Cases from NYC hospitals were separated into 2 cohorts: public and other NYC hospitals. Surgeons associated with each case were identified using the database and were stratified by volume of cases and presence of subspecialty training.
A total of 12,202 admissions for ovarian cancer were identified. Of these, 3639 involved major surgery, and 187 were performed at public hospitals. There were more African American and Asian patients in the public cohort (P < .001). The primary insurer was more likely to be Medicaid or a self-payer in the public cohort (P < 0.001). Urgent or emergent admissions comprised 55% of all admissions in public hospitals, compared with 29% of admissions in other NYC hospitals (P < .001). Patients in public hospitals were less likely to have their surgery performed by a gynecologic oncologist (57% vs 74%, P < .001) and less likely to have their surgery performed by a high-volume surgeon (21% vs 47%; P < .001) compared with patients in other NYC hospitals.
Ovarian cancer is the leading cause of death from gynecologic malignancy and is the fifth most deadly cancer for women in the United States. The American Cancer Society predicted that in 2009, 21,550 women would contract the disease and 14,600 would die as a result.1 The high mortality rate is secondary to the extent of disease at the time of diagnosis, with approximately 75% of patients having advanced stage disease at diagnosis.
Ovarian cancer is surgically staged according to guidelines established by the International Federation of Gynecologic Oncologists. This staging requires removal of the ovaries, fallopian tubes, and uterus, as well as extraovarian sites (lymph nodes, peritoneum, and omentum) known to harbor neoplastic cells. Performance of a staging procedure allows for correct assignment of stage to the patient and carries both prognostic and therapeutic implications. In addition to staging, the goal of surgery is to remove as much visible disease as is technically feasible. These procedures are stratified by the amount of tumor remaining after surgery. A clear survival advantage has been demonstrated for patients with residual disease of <1 cm in largest diameter (optimal cytoreduction). As a result, procedures with increased morbidity, such as bowel resection, are readily performed when they will leave a patient optimally cytoreduced.
In 1995, the National Institutes of Health published guidelines for the treatment of ovarian cancer affirming that a complete staging procedure is the standard of care.2 Despite this, only 50% of patients diagnosed with ovarian cancer in the United States undergo the correct procedure.3 Patients who receive their care from a gynecologic oncologist are more likely to achieve optimal cytoreduction4 and have improved survival5 compared with those treated by gynecologists or general surgeons. However, only 40% of patients with ovarian cancer receive their staging surgery from a gynecologic oncologist.6 Surgical volume has also been investigated as an independent variable affecting surgical outcome in ovarian cancer patients. Several studies have shown improved outcomes for those patients for whom high volume has been documented for the surgeon as well as the hospital,3, 7 although volume was not an independent variable in academic hospitals.3
Race and socioeconomic status are associated with suboptimal ovarian cancer treatment. African American and Hispanic patients are less likely to receive standard of care surgery, including lymphadenectomy.3, 6, 8 One reason for these differences may be differential access to centers with subspecialty care,9-10 although some authors refute this suggestion.11 A recent single-institution review showed that white and African American patients treated in the same hospital had similar outcomes.12
New York City (NYC) hosts the nation's largest public hospital system, including 11 acute care hospitals who serve patients without regard for their ability to pay for care. These hospitals are heterogeneous. Some are organized around their surrounding community with a relatively low patient census; others act as tertiary referral centers with high volume in their inpatient and outpatient services (Table 1). The patient mix is diverse, with significant representation of both minority and recent immigrant groups. The majority of patients are either self-paying or insured through government programs (health maintenance organization [HMO] Medicaid/Medicare). Several of these hospitals contract with academic medical centers to provide coverage by gynecologic oncologists. These physicians may cover the public hospital as well as the primary university-based hospital of the medical school. These arrangements are varied, and no referrals between institutions are mandated for patients with known or suspected diagnoses of gynecologic cancer.
Table 1. Characteristics of the 11 Acute Care Hospitals in the New York City Municipal Hospital Systema
The appropriate treatment of a patient with ovarian cancer often encompasses a variety of complex medical and surgical decisions. As a result, it is a good proxy for evaluating the appropriate referral system for specialist gynecologic oncology care within the municipal hospital system. We sought to determine the surgical patterns of care among patients with ovarian cancer in the NYC municipal hospital system.
MATERIALS AND METHODS
Institutional review board approval was obtained from the New York University School of Medicine. Cases were identified using the New York State Department of Health Statewide Planning and Research Cooperative System (SPARCS). This comprehensive data reporting system, established in 1979, collects patient and treatment information for every hospital discharge, ambulatory surgery, and emergency department admission. New York State mandates reporting by every hospital.
In the SPARCS database, each hospital discharge contains demographic data including age, sex, race/ethnicity, type of insurance (private, Medicare, Medicaid, self-pay/uninsured), zip code of residence, reason for admission by diagnosis-related group codes, primary and secondary (≤24) diagnoses, primary and secondary (≤20) procedures, dates of procedure, and length of stay.
Cases were identified for the years 2001 to 2006 inclusive, using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9) code 183.x: malignant neoplasm, ovary and uterine adnexa. Our study included patients with fallopian tube cancer (code 183.2) given the similarity in diagnosis and treatment. Each admission that resulted in a major gynecologic surgical procedure and carried 1 of the above diagnoses was included for study. Major surgery was defined as any case with an exploratory laparotomy (code 54.x) or any code including hysterectomy (code 68.x). Index cases were restricted to those procedures performed in NYC hospitals and were divided into 2 cohorts: cases in municipal (public) hospitals and cases from all other NYC hospitals. A comorbidity index, designed for use with large databases, was applied using a published protocol.13
Evaluation of Public Hospitals
Each acute care hospital is identified by a separate number within the SPARCS database; the codes for each public hospital were obtained, and the number of ovarian cancer surgeries were generated for each public hospital. Inquiries were made regarding the availability of subspecialist gynecologic cancer care at each site over the investigational period. Each site was stratified into 3 groups based on the availability of specialty care: full-time gynecologic oncology coverage; consistent, part-time gynecologic oncology coverage; and rare or no gynecologic oncology coverage. Full-time gynecologic oncology coverage was defined as the presence of a board-certified or board-eligible gynecologic oncologist on site at least 2 days a week, with around-the-clock availability for emergent cases. Part-time coverage was defined as having routine coverage at the facility, although less than 2 days a week, with or without around-the-clock availability for emergent cases. Rare or no gynecologic oncology coverage was defined as having no board-eligible or certified practitioner with routine hours, and no availability of on-site emergent coverage.
Determination of Surgeons' Training and Volume
The SPARCS database provides license numbers for the primary and secondary surgeons associated with each surgical procedure. We used these numbers to find all surgeons associated with major surgical procedures coded with the diagnosis “malignant ovarian neoplasm, primary.” Each surgeon was then stratified by presence or absence of subspecialty training using public physician databases. For each surgeon attributed to an ovarian cancer surgery, the SPARCS database was used to obtain the volume of ovarian cancer cases performed by that surgeon within New York State. An average number of cases per year was determined for each surgeon. This included only years in which at least 1 major ovarian surgery was attributed to that surgeon, in an attempt to account for physicians moving into or out of the treatment area. All cases performed within New York State were counted toward the surgeons' surgical volume. Based on previously reported standards,7, 14-16 high-volume surgeons were defined as those performing 10 or more ovarian cancer surgeries per year. All others were defined as low-volume surgeons.
Demographic and treatment characteristics for the 2 groups were compared using the Pearson chi-square test, Fisher exact test, and Student t test or their nonparametric equivalents where appropriate. In cases in which the results of parametric and nonparametric analyses were in concordance, the results of the parametric test are reported. Statistical analyses were performed using STATA version 9 software (StataCorp, College Station, TX).
A total of 12,202 admissions for ovarian cancer in NYC acute care facilities were identified in the study period (Figure 1). Of these admissions, 3643 involved major surgical procedures, 187 (5%) of which occurred within the public cohort.
The demographic information is listed in Table 2. Patients in the public cohort were younger than those in the other NYC group (54 years vs 59.7 years). There were more African American patients (36% vs 13%) and Asian patients (10% vs 5%) in the public cohort versus the other NYC cohort, respectively (P < .001). Unfortunately, ethnicity data was not well recorded by the database, particularly in the public cohort, in which a large percentage of cases (26%) were classified as unknown. A modified Charlson comorbidity index was applied to assess chronic health conditions present prior to surgery.13 The majority of patients had few or minor comorbidities, as evidenced by a score of 0, 1, or 2. There were no differences in the scores between the 2 cohorts.
Table 2. Demographic and Baseline Information by Cohorta
Public Hospital (n, 187)
Other NYC Hospital (n, 3456)
NYC indicates New York City.
Data are presented as n (%).
Average age, y
Fallopian tube cancer
1 to 2
3 to 4
Table 3 shows the primary insurance coverage that was designated for the index admission for each cohort. The payer mix was different between the 2 cohorts (P < .001). Medicaid was the primary insurer for 56% of patients in the public cohort compared with 8% of patients in the other NYC cohort. Private insurance was designated as the primary insurer for 65% of other NYC patients compared with 21% of public patients. Medicare was used less by public hospital patients compared with other NYC patients (12% vs 26%). The self-pay option was more common among public versus other NYC patients (11% vs 1%).
Hospital admission and surgical characteristics are described in Table 4. Urgent or emergent admissions were more likely in the public cohort, comprising 55% of the total for this group compared with 29% of the admissions in the other NYC cohort (P < .001). The types of procedures performed, as determined by ICD-9 codes, are also listed. There was a difference in the types of procedures performed between the 2 cohorts; however, without knowledge of the stage distribution in the 2 groups, clinical relevance of this difference cannot be determined. Patients in the public cohort were less likely to have their surgery performed by a gynecologic oncologist than patients in the other NYC cohort (57% vs 74%; P < .001). High-volume surgeons were defined as surgeons who had performed at least 10 major surgeries associated with code 183.x per year. Only those years in which at least 1 major surgery was performed were counted toward the average number. Patients in public hospitals were less likely to have their surgery performed by a high-volume surgeon compared with patients in other NYC hospitals (21% vs 47%, respectively; P < .001).
Using the database, the number of major surgical cases involving ovarian cancer was determined at each public hospital. The hospitals were grouped by the degree of gynecologic oncology presence (Figure 2). Overall, the case volume was low at all hospitals, with the majority of institutions performing fewer than 20 cases over the study period. Only 1 institution performed an average of 10 major surgical cases for ovarian cancer per year. Despite the availability of coverage, 2 of the 3 centers with full-time coverage had a low census of ovarian cancer patients. The remainder of municipal hospitals had either intermittent or rare gynecologic oncology coverage.
Each surgeon responsible for a major ovarian cancer surgery performed for a public hospital patient was identified over the study period (Figure 3). A total of 67 surgeons were identified. The majority of surgeons performed fewer than 5 major surgeries associated with ovarian cancer over the 6-year-period. Only 4 surgeons operating at public hospitals met the definition of a high-volume surgeon. For these surgeons, as well as most of the surgeons depicted in the graph, the majority of their ovarian cancer surgeries were performed at other NYC hospitals.
To the best of our knowledge, this is the first attempt to describe the surgical management of ovarian cancer patients in a large municipal hospital system. We chose ovarian cancer as the index disease because it is associated with a need for aggressive surgical management encompassing skills acquired through comprehensive training. As such, standardized treatment provided by a gynecologic oncologist has been advocated by both the American College of Obstetricians and Gynecologists17 and the National Institutes of Health.2
Ovarian cancer patients in the municipal hospital system are less likely to have a gynecologic oncologist as their surgeon than in other NYC hospitals. They are also less likely to have a high-volume surgeon associated with their care. This disparity may reflect the lack of availability of subspecialists in several public hospitals. However, the availability issue could be countered by strict referral practices to those centers with available subspecialists. Given the overall low census of ovarian cancer patients in the public system, a centralized care model could be instituted for patients who are at high risk for ovarian cancer. Referral systems have been advocated both in the United States18 and abroad14, 19 to ensure that patients with ovarian cancer have the benefit of high-volume surgeons and centers with appropriate ancillary services.
As it stands currently, surgeons at municipal hospitals qualify as high-volume surgeons if they perform ovarian cancer surgeries at facilities outside the public hospital system. Given the overall low census of ovarian cancer cases at municipal hospitals, this dual coverage allows patients at municipal hospitals access to high-volume cancer surgeons. Even with 100% centralization within the municipal hospital system, the patient census may be insufficient if these subspecialists were solely dedicated to municipal hospital patients. There is literature to support the sharing of physicians between public and private entities.20 A study evaluating the National Health System in the United Kingdom showed the most public service productivity among surgeons who used their maximum allowable private sector time.21 In our study, there was a clear benefit to using a shared physician practice model, and this may be reproducible in other specialties in which surgical volume is related to outcomes.
Patients from the public cohort were more likely to have their admission classified as urgent or emergent. In comparison, over 75% of patients in the other NYC cohort had elective admissions. Additional study is needed to determine the cause for this discrepancy, although this finding is not unique to patients with ovarian cancer, because underrepresented minorities are more likely to use emergency rooms for primary care needs.10
The use of a large, statewide database enabled us to evaluate a large pool of patients in diverse hospital systems. However, there are several important limitations. Although compliance with the database is mandatory for all acute care hospitals in New York State, the accuracy of the reported data is not validated. Specifically, the cases representing ovarian cancer require accurate coding to be captured for this project. Furthermore, stage of disease and degree of resection are not available for evaluation, limiting our ability to compare the overall quality of cancer care received by the 2 patient cohorts. Despite these limitations, the current study provides information regarding the short-term surgical patterns of care for patients in the municipal hospital system. We are undertaking a large chart review study to obtain more cancer-specific data on these patients, which will allow us to compare outcomes.
Other potential limitations of this study include the assumptions regarding surgical training. Of course, surgeons without gynecologic oncology training may be highly skilled. However, there are ample data in the ovarian cancer literature to support improved outcomes among patients whose surgeries were performed by gynecologic oncologists compared with gynecologists or general surgeons.4, 5, 22 There are also several studies showing a clear benefit in patients' outcomes when high-volume surgeons are involved in their care.3, 7 Inaccurate coding of physicians' license numbers could lead to incorrect categorization of physicians' status. Furthermore, if a surgeon is operating outside of New York State, we could be undercounting the number of procedures for that practitioner.
In conclusion, patients with ovarian cancer treated at municipal hospitals are less likely to undergo surgery performed by gynecologic oncologists or high-volume surgeons than patients at other NYC hospitals. While we as a nation continue to grapple with how best to provide care to the large percentage of underinsured and uninsured patients, it is critical that we assess the solutions that have been created to ensure that we are providing care in an equitable fashion. These findings, although preliminary, warrant further investigation.
CONFLICT OF INTEREST DISCLOSURES
This study was supported by an American Society of Clinical Oncology Foundation Young Investigator Award Grant (to L. R. B.).