Using breast cancer quality indicators in a vulnerable population

Authors

  • Formosa Chen MD,

    1. Department of Surgery, Olive View-UCLA Medical Center, Sylmar, California
    2. Department of Surgery, University of California, Los Angeles, California
    3. Department of Surgery, Greater West Los Angeles, Veterans Administration, Los Angeles, California
    Search for more papers by this author
  • Melissa Puig MD,

    1. Department of Family Medicine, Riverside County Hospital, Moreno Valley, California
    Search for more papers by this author
  • Irina Yermilov MD,

    1. Department of Surgery, Olive View-UCLA Medical Center, Sylmar, California
    Search for more papers by this author
  • Jennifer Malin MD,

    1. Department of Hematology/Oncology, Greater West Los Angeles, Veterans Administration, Los Angeles, California
    Search for more papers by this author
  • Eric C. Schneider MD,

    1. Department of Health Policy and Management, Harvard School of Public Health, Boston, Massachusetts
    2. Brigham and Women's Hospital, Division of General Medicine, Boston, Massachusetts
    3. RAND Boston, Boston, Massachusetts
    Search for more papers by this author
  • Arnold M. Epstein MD, MA,

    1. Department of Health Policy and Management, Harvard School of Public Health, Boston, Massachusetts
    2. Brigham and Women's Hospital, Division of General Medicine, Boston, Massachusetts
    Search for more papers by this author
  • Katherine L. Kahn MD,

    1. David Geffen School of Medicine at UCLA, Los Angeles, California
    2. Jonsson Comprehensive Cancer Center, UCLA, Los Angeles, California
    3. RAND Health, Santa Monica, California
    Search for more papers by this author
  • Patricia A. Ganz MD,

    1. David Geffen School of Medicine at UCLA, Los Angeles, California
    2. Jonsson Comprehensive Cancer Center, UCLA, Los Angeles, California
    Search for more papers by this author
  • Melinda Maggard Gibbons MD

    Corresponding author
    1. Department of Surgery, Olive View-UCLA Medical Center, Sylmar, California
    2. Department of Surgery, University of California, Los Angeles, California
    3. Department of Surgery, Greater West Los Angeles, Veterans Administration, Los Angeles, California
    • UCLA Medical Center, Department of Surgery, CHS 72-215, 10833 LeConte Ave, Los Angeles, CA 90095
    Search for more papers by this author
    • Fax: (310) 268-4601


Abstract

BACKGROUND:

Adherence to quality indicators may be especially important to disease-specific outcomes for uninsured, vulnerable patients. The objective of this study was to measure adherence to National Initiative for Cancer Care Quality (NICCQ) breast cancer quality indicators in a public hospital and compare performance to published rates in a previously collected 5-city cohort.

METHODS:

One hundred five consecutive, newly diagnosed, stage I-III, breast cancer patients at a public hospital (from 2005 to 2007) were identified. Adherence rates to 31 quality indicators were measured by using medical record abstraction. Rates were calculated for individual indicators, aggregated domains, and components of care and were compared with the 5-city cohort results by using a 2-sided test of proportions.

RESULTS:

Overall adherence to the NICCQ indicators at the public hospital was 82%, versus 86% in the 5-city cohort. Public hospital adherence was better in 3 domains and components (Management of Treatment Toxicity 95% vs 73%, Referrals 76% vs 15%, and Documentation of Key Clinical Factors 72% vs 64%, P < .05 for all), but it was lower in others (Testing 82% vs 96%, Adjuvant Therapy 76% vs 83%, Surgery 72% vs 86%, Surveillance 63% vs 94%, and Respect for Patient Preferences 52% vs 72%, P < .001 for all).

CONCLUSIONS:

The results showed that it is possible to deliver breast cancer care to vulnerable patients comparable in quality to the care received by the broader population. Further study should identify the factors that lead to variation in adherence across domains of quality. Cancer 2011. © 2011 American Cancer Society.

Variations in the quality of breast cancer treatment exist throughout the continuum of care, from screening, diagnosis and active treatment to surveillance.1 Receipt of recommended care varies by geographic region, institution, provider type, and patient characteristics.2-3 Depending on the type of care, 10% to 70% of women with breast cancer do not receive the evidence-based treatments shown to improve rates of cure and survival.4

Vulnerable populations, such as the uninsured and underinsured, racial/ethnic minorities, and those with lower socioeconomic status are at increased risk for receiving suboptimal quality care.5-6 These patients have lower rates of screening mammography,7-8 are more likely to experience delays between abnormal screening mammogram and diagnostic resolution,9 more likely to be diagnosed with advanced stage disease,10-11 less likely to receive breast-conservation surgery and appropriate adjuvant therapy,12-13 and to have worse outcomes such as higher rates of recurrence and shorter survival.7 Patients who live in rural areas or areas with lower socioeconomic status are also at risk for inadequate care and poorer outcomes.2, 12-15 Public hospitals care for a disproportionate number of uninsured and underinsured patients, racial/ethnic minorities, and those with lower socioeconomic status; thus, public hospitals are an ideal setting in which to evaluate and improve the quality of care for these vulnerable populations.

Quality of care indicators can be used to evaluate whether patients receive recommended care. Unlike clinical practice guidelines, quality indicators usually define highly specific inclusion and exclusion criteria as well as measurement algorithms, and when well designed, they can be used to compare quality of care across providers.16 A comprehensive set of 36 quality indicators for breast cancer was developed as part of the American Society of Clinical Oncology's (ASCO) National Initiative for Cancer Care Quality (NICCQ).17-18 These indicators measure specific processes of care, spanning the treatment continuum from diagnostic staging, surgical treatment, and adjuvant therapy to post-treatment surveillance. By using these indicators in a study cohort from 5 metropolitan areas in the United States (referred to as the 5-city cohort in the current study), Malin et al found that there was 86% adherence to the breast cancer indicators.

Before the ASCO study of the 5-city cohort, most evaluations of variation and disparities in breast cancer care focused on singular aspects of care or a small number of clinical processes, such as screening mammography. Although efforts are now being made to evaluate the quality of cancer care in community settings through programs such as the Quality Oncology Practice Initiative sponsored by ASCO, monitoring of the quality of care across the entire spectrum of breast cancer treatment is not routinely performed outside of large academic studies.19 Comprehensive assessment of breast cancer treatment in vulnerable and at-risk populations has also been lacking.20

The goals of the current project were to measure adherence to the NICCQ breast cancer quality indicators in a public hospital setting and to identify which areas of care could be prioritized for quality improvement. As a reference standard, we compared adherence rates between patients receiving care at the public hospital and patients who were included in the 5-city, population-sampled, NICCQ cohort.

MATERIALS AND METHODS

Patient Cohorts

The study was based at a Los Angeles public hospital that serves 2 million individuals. All newly diagnosed, consecutive, breast cancer cases were identified through a systematic search of the pathology database or samples submitted between August 15, 2005 and September 15, 2007 and included biopsies and surgical specimens. Terms to capture cases included breast, cancer, and carcinoma. Patients with newly diagnosed stage I to III cancers who were aged 21 years and older were included. Patients who transferred their care to other institutions (without undergoing surgery or chemotherapy at the public hospital), patients with in situ or metastatic disease, and men were excluded. Adherence to the NICCQ quality indicators was assessed for these patients through chart abstraction.

Findings from the public hospital were compared with the previously published ASCO study that sampled patients from 5 metropolitan areas (Atlanta, Georgia; Cleveland, Ohio; Houston, Texas; Kansas City, Kansas; Los Angeles, California) who were newly diagnosed in 1998 and registered in an American College of Surgeons-approved hospital cancer registry (5-city cohort). Eligible breast cancer participants had stage I to III cancers, were aged 21 to 80 years, were English-speaking, and were female. Adherence to the NICCQ quality indicators for the 5-city cohort was assessed through telephone survey of the patients and a medical record review conducted 4 years after diagnosis. In the current study, data were collected through medical record abstraction with 1-year follow-up; patient interviews and surveys were not used in this study.

Data Abstraction

A medical record review was conducted on the patients in the public hospital sample to determine adherence to the NICCQ breast cancer quality indicators. A Microsoft Access computerized data abstraction tool was used (Microsoft, Redmond, Washington). Two physicians completed the abstractions, and the senior researcher reabstracted data that required further assessment (ie, to confirm dates or pathology reports in question). The senior researcher sampled 10% of the cases for review. In addition, tumor stage was reconfirmed independently for all cases by the senior researcher (each patient's tumor size, lymph node status, histology, and presence of metastases) from the electronic medical records. Study approval was obtained from both the University of California at Los Angeles (UCLA) and the Olive View-UCLA Medical Center Institutional Review Boards.

Analyses

Adherence to 31 of the 36 NICCQ breast cancer indicators was determined for the public hospital patients. The 5 indicators that were not measured pertained to the total duration of hormone therapy, whether the chemotherapy dosage was consistent with recommended guidelines, the amount of radiation to the chest wall after mastectomy, completion of planned radiation therapy among ambulatory patients, occurrence of grade 4 toxicity, and appropriate workup of vaginal bleeding when the patient was being given tamoxifen. Reasons for not assessing these 5 indicators included short follow-up time, data not reported, and insufficient data for accurate measurement of adherence. For examples, we were unable to determine whether the patient had been hospitalized at an outside institution, and we did not have consistent access to a data source for informing us whether the patient had a grade 4 toxicity.

The denominator for each indicator, termed eligible events, was defined by the specific inclusion and exclusion criteria for each indicator. For example, 1 of the indicators states: “If a patient has an axillary lymph node dissection, then the patient should have at least 6 lymph nodes removed.” The “if” portion of the indicator states inclusion and exclusion criteria that define the denominator for that indicator.

After the ASCO study of the 5-city cohort methodology,18 each indicator was also assigned to 1 of 5 Quality of Care Domains: 1) Diagnostic Evaluation, 2) Surgery, 3) Adjuvant Therapy, 4) Management of Treatment Toxicity, and 5) Post-treatment Surveillance as well as 1 of 8 Components of Care: 1) Testing, 2) Pathology, 3) Documentation of Key Clinical Factors, 4) Referrals, 5) Timing, 6) Receipt of Treatment, 7) Technical Quality of Treatment, and 8) Respect for Patient Preferences (Table 1). Percentage adherence was calculated for each domain and component. Adherence rates for individual indicators and aggregated domains and components were compared with the adherence rates reported in the 5-city cohort by a 2-sided test of proportions. Sensitivity analysis was performed to ensure that the comparisons across domains and components between the 5-city cohort and the public hospital data were not impacted by the 5 indicators that were not assessed in this study. This was performed by comparing the 5-city cohort's aggregated domain and component adherence rates both with and without the unmeasured indicators to the public hospital rates. Consistent with accepted quality improvement standards, a cutoff of <85% adherence rate was used to identify indicators, domains, and components in need of improvement.18 Analyses were performed using STATA statistical software version 9.0 (StataCorp, College Station, Texas).

Table 1. Assignment of the 31 Measured National Initiative on Cancer Care Quality Indicators into Quality of Care Domains and Components of Care
 Components of Care
 TestingPathologyDocumentation of Key Clinical FactorsReferralsTimingReceipt of TreatmentTechnical Quality of TreatmentRespect for Patient Preferences
Quality of Care Domains
 Diagnostic Evaluation3631
 Surgery112
 Adjuvant Therapy221421
 Management of Treatment Toxicity1
 Surveillance1

RESULTS

Cohort and Patient Characteristics

One hundred ninety-one patients with breast carcinoma were identified through a search of the public hospital pathology database. Eight patients had a breast cancer diagnosis reported but elected to have treatment at another hospital. One hundred five (55.0%) patients met study eligibility requirements and were included in the analysis (see Fig. 1 for reasons for exclusion). The 5-city ASCO study cohort included 1287 patients with breast cancer.

Figure 1.

Inclusion and exclusion criteria from analytic public hospital cohort are shown.

The mean age of the public hospital patients was 54 years (range, 30-82 years), with 50% younger than 55 years (Table 2). Compared with the 5-city cohort, the public hospital patients were significantly younger. Fifty-two percent of patients at the public hospital were Hispanic compared with 4% in the 5-city cohort. In addition, 57% of the breast cancer population at the public hospital (including in situ and metastatic cases) preferred Spanish as their primary language compared with the 5-city cohort, which was entirely English-speaking. Nineteen percent of the public hospital patients presented with stage III disease compared with 5% in the 5-city cohort. The most notable difference between the study populations was that 100% of the public hospital patients were either uninsured or publically insured versus 3% of the 5-city cohort patients.

Table 2. Comparison of Public Hospital Versus 5-City Cohort Breast Cancer Patient Characteristics
 Public Hospital5-City CohortP
Total no. of patients1051287 
Age, y
 <5550%31%<.001
 ≥55 to <6534%42%.11
 ≥6516%27%<.001
Race/ethnicity
 Hispanic52%4%<.001
 White30%85%<.001
 Asian12%4%<.001
 African American5%7%.44
Insurance
 Uninsured/Public100%3%<.001
 Private71% 
 Medicare21% 
Pathological stage
 Stage I44%54%.05
 Stage II37%39%.69
 Stage III19%5%<.001

Overall and Individual Indicator Adherence Rates in the Public Hospital Cohort

Overall adherence to the 31 measured NICCQ breast cancer quality indicators in the public hospital cohort (n = 105) was 82%, with a total of 1720 eligible events (sum of the denominators of the 31 measured indicators).

Adherence to the individual indicators ranged from 0% to 100% in the public hospital cohort, with a median of 86% (Table 3). The denominators or eligible events for the individual indicators ranged from 6 to 104. In 14 of the 31 indicators, adherence was below 85%. The quality of pathology reporting was generally good with the exception of consistent inking of margins and reporting of tumor histologic grade (80% and 58%, respectively). Complete documentation of cancer stage in patients seen only by surgeons was poor (17%). For patients who did not receive surgical lymph node sampling (sentinel lymph node biopsy [SLNB] or axillary lymph node dissection [ALND]), breast conservation surgery (BCS), or reconstruction after mastectomy, documentation that these patients had been informed of possible treatment options was inadequate much of the time (0%, 27%, and 43% for lymph node sampling, BCS, and reconstruction, respectively).

Table 3. National Initiative on Cancer Care Quality Breast Cancer Quality Indicators by Components of Care: Comparison Between Public Hospital and 5-City Cohort Adherence Rates
IndicatorPublic Hospital Cohort 2005-20075-City Cohort 1998 
No. EligibleNo. Meeting CriteriaAdherence RateNo. EligibleNo. Meeting CriteriaAdherence RateP
Testing
 IF a patient with a stage I-III breast cancer meets all of the following criteria: age <70 y, tumor size >1 cm, no evidence of metastatic disease within 3 months of diagnosis, and no documentation in the medical record that axillary lymph node sampling would not change treatment, THEN the patient should have axillary lymph node sampling (sentinel lymph node biopsy or axillary dissection).827996%78277799%.02
 IF a patient has a sentinel node biopsy that is positive, THEN the patient should undergo axillary lymph node dissection.302893%4747100%.07
 IF a patient has an axillary lymph node dissection, THEN the patientshould have at least 6 lymph nodes removed.595186%91987695%<.001
 IF a patient with stage I-III breast cancer has not had bilateral mastectomies, THEN the patient should have a mammogram in the last 12 mo.905763%1195112194%<.001
Pathology
 IF a patient has stage I-III breast cancer and had a breast tumorremoved, THEN the pathology report should state the tumor size.102102100%1166104389%<.001
 IF a patient has stage I-III breast cancer and had a breast tumor removed, THEN the pathology report should state that the margins were inked.1028280%116685173%.12
 IF a patient has stage I-III breast cancer and had a breast tumorremoved, THEN the pathology report should state the status of the margins.10210199%1166104890%<.001
 IF a patient has stage I-III breast cancer and had a breast tumorremoved, THEN a pathology report should state the histologic grade of the tumor.1025958%118096982%<.001
 IF a patient has stage I-III breast cancer and had a breast tumorremoved, THEN the hormone-receptor status of the tumor should be checked and reported.1029997%1280121395%.37
 IF a patient with stage I-III breast cancer undergoes an axillary lymphnode dissection, THEN the pathology report should state the number of lymph nodes evaluated and the number of positive lymph nodes.9595100%1087105297%.09
Documentation of Key Clinical Factors
 IF a patient with stage I-III breast cancer is not referred to another oncology specialist (eg, medical oncologist or radiation oncologist), THEN the medical record of the surgeon performing the last cancer surgery should document all of the following: tumor size and lymph node status (or AJCC stage or TNM category) and hormone-receptor status.12217%681725%.53
 IF a patient with stage I-III breast cancer is seen in consultation by amedical oncologist after last breast surgery, THEN the medical record of the medical oncologist should document all of the following: tumor size and lymph node status (or AJCC stage or TNM category) and hormone receptor status.514690%70457482%.15
 IF a patient with stage I-III breast cancer is seen in consultation by a radiation oncologist after completion of the staging evaluation, THEN the medical record of the radiation oncologist should document at least one of the following: tumor size and lymph node status (or AJCC stage or TNM category).474391%28220272%.01
 IF a patient is treated with chemotherapy, THEN body-surface area should be documented.504794%3138828%<.001
 IF a patient with stage I-III breast cancer is treated with radiation therapy THEN the radiation therapy medical record should document all of the following: total radiation dose, dose per fraction or number of fractions given, and the site.461124%34121362%<.001
Referrals
 IF a patient with stage I-III breast cancer undergoes breast-conservingsurgery and did not receive radiation therapy, THEN the patient should have a consultation with a radiation oncologist.13754%621118%.01
 IF a patient with invasive breast cancer undergoes a mastectomy andhas any of the following: positive margins, tumor size >5 cm, 4 or more involved lymph nodes or a T4 lesion, THEN the patient should have a consultation with a radiation oncologist.211990%2015%<.001
Timing
 IF a patient newly diagnosed with stage II-III breast cancer is aged <50 y and the tumor is >2 cm or the tumor involves the lymph nodes, THEN the patient should start adjuvant chemotherapy within 8 wk of the last therapeutic surgery.201575%17313679%.68
Receipt of Treatment
 IF a patient has stage I-III breast cancer, THEN the patient shouldreceive 1 of the following surgical treatments: breast-conserving surgery or mastectomy.104104100%12871286100%1.00
 IF a patient with stage I-III breast cancer meets all of the followingcriteria: ER+ or PR+ breast cancer, tumor size 1 cm or involved axillary lymph nodes, and not taking tamoxifen within 6 mo before diagnosis, THEN the patient should be started on tamoxifen.615285%77269991%.12
 IF a patient with stage II-III breast cancer is aged <50 y and the tumor is>2 cm or involves the lymph nodes, THEN the patient should receive chemotherapy.242083%17316394%.06
 IF a patient with stage I-III breast cancer has breast-conserving surgery,THEN the patient should receive local radiation therapy.272696%68366998%.54
 IF a patient with invasive breast cancer who undergoes a mastectomyhas any of the following: positive margins, tumor size >5 cm, 4 or more involved lymph nodes or a T4 lesion, THEN the patient should receive radiotherapy.161594%1219881%.21
Technical Quality Of Treatment
 IF a patient with stage I-III breast cancer undergoes surgical treatmentwith breast-conserving surgery, THEN the pathology report for the last cancer surgery should state that the surgical margins are clear or only focally involved with breast cancer.484288%68167499%<.001
 IF a patient is treated with chemotherapy, THEN the planned dose (doseper cycle x number of cycles) should be documented in the medical record.503774%29618864%.17
 IF a patient with stage I-III breast cancer has breast-conserving surgeryand the patient received radiation therapy and did not receive brachytherapy, THEN the patient should receive local radiation therapy 45Gy to 50.4 Gy to the whole breast.221777%46544696%<.001
 IF a patient ever receives highly emetogenic chemotherapy, THEN thepatient should receive potent antiemetic therapy.565395%26717867%<.001
Respect for Patient Preferences
 IF a patient with stage I-III breast cancer has a 1 cm tumor or lymphnode involvement, THEN a physician should have a discussion with the patient on possible treatment with chemotherapy.786077%104493990%<.001
 IF a patient has stage I-III breast cancer and does not undergo axillarylymph node sampling, THEN the patient should be informed about the option of a surgical check or removal of lymph nodes under the arm (ie, sentinel-node biopsy or axillary dissection).600%402255%.01
 IF a patient with stage I-III breast cancer undergoes mastectomy as firsttherapeutic procedure, THEN before undergoing mastectomy, the patient should be informed about the option to have either breast- conserving surgery followed by radiation therapy or mastectomy.481327%1073129%.80
 IF a patient with stage I-III breast cancer undergoes mastectomy, THENbefore undergoing mastectomy, the patient should be informed about the option of breast reconstruction after mastectomy.542343%59829850%.32

Quality indicators assessing adjuvant chemotherapy identified the following processes as needing improvement: discussion regarding possible treatment with chemotherapy (77%), receipt of chemotherapy (83%), starting chemotherapy within 8 weeks of last therapeutic surgery (75%), and documentation of planned chemotherapy dose (74%). Documentation of the radiation treatment plan (24%) and the total amount of radiation after BCS (77%) were also suboptimal. Only 54% of patients who did not receive radiation therapy after BCS had a consultation with a radiation oncologist. And, timely surveillance mammography was performed in only 63% of patients (of note, surveillance mammography is performed on site at the public hospital, and patients typically do not have another source of care option). These indicators identified specific clinical processes in need of improvement.

In 17 quality indicators, performance was 85% or greater (Table 3). These indicators addressed aspects of pathology reporting, diagnostic evaluation, quality of surgical therapy, documentation, receipt of hormone and radiation therapy, management of treatment toxicity, and appropriate referrals to specialists. For example, adherence rate for diagnostic evaluation was 96% for performing surgical axillary lymph node sampling, 96% of patients received radiation after BCS, and 90% received consultation with a radiation oncologist after mastectomy when appropriate.

Quality of Care Domains and Components of Care Adherence Rates in the Public Hospital Cohort

In addition to evaluating specific processes, each NICCQ indicator was classified into a Quality of Care Domain and a Component of Care (Table 1). Examining performance in these domains and components can help identify more global or systems-level deficiencies in quality of care within an organization or healthcare system.

For the public hospital, adherence was ≥85% in Diagnostic Evaluation and Management of Treatment Toxicity domains (Fig. 2, Table 4). Conversely, less than 85% adherence in the domains of Surgery, Adjuvant Therapy, and Surveillance identified systems-level elements in need of quality improvement.

Figure 2.

Illustrated are (A) comparison of public hospital to 5-city cohort adherence rates across aggregated quality of care domains and (B) comparison of public hospital to 5-city cohort adherence rates across aggregated components of care.

Table 4. Comparison of Public Hospital Versus 5-City Cohort Adherence to NICCQ Breast Cancer Quality Indicators by Aggregated Quality of Care Domains and Components of Care
 No. of MeasuresEligible EventsAdhered Events% AdherencePa
 Public Hospital Cohort5-City CohortPublic Hospital Cohort5-City CohortPublic Hospital5-City CohortPublic Hospital5-City Cohort 
  • a

    P for comparison of percentage adherence to domains and components of care scores for patients receiving care in public hospital compared with the 5-city cohort using a 2-sided test of proportions.

Overall3136172020,281140517,45582%86%<.001
Quality of care domains
 Diagnostic evaluation13138929887787869188%88%1.00
 Surgery442542673182228972%86%<.001
 Adjuvant therapy12164286148326507776%83%<.001
 Management of treatment toxicity12563785327795%73%<.001
 Surveillance1190119557112163%94%<.001
Components of care
 Testing442612943215282182%96%<.001
 Pathology666057045538617689%88%.47
 Documentation key clinical factors552061708149109472%64%.02
 Referrals223482261276%15%<.001
 Timing11201731513675%79%.68
 Receipt of treatment552323036217291594%96%.14
 Technical quality of treatment491763505149301185%86%.71
 Respect for patient preferences44186178996129052%72%<.001

Among the Components of Care, adherence was ≥85% in Pathology, Receipt of Treatment, and Technical Quality of Treatment. The components of Testing, Documentation of Key Clinical Factors, Referrals, Timing, and Respect for Patient Preferences all had lower scores.

Comparison of Adherence Rates for the Public Hospital and 5-City ASCO Cohorts

Overall performance across all the NICCQ indicators was slightly lower at the public hospital than in the 5-city cohort (82% vs 86%, P < .001), but the discrepancy was small.

Adherence rates in the public hospital cohort and 5-city cohort differed in 16 of the 31quality indicators (Table 3). Performance at the public hospital was better for 7 of the 16, which included for example, pathology reporting of tumor size and margin status, documentation of staging, and appropriate referral to radiation oncologists. In 9 indicators, performance at the public hospital was lower compared with the 5-city cohort, for example, informing patients who did not receive lymph node sampling that it was an option, discussing the option for chemotherapy, and performing timely surveillance mammography.

Across the aggregated Quality of Care Domains and Components of Care, performance was better in the public hospital in the Management of Treatment Toxicity, Referrals, and Documentation of Key Clinical Factors (Fig. 2, Table 4). Domains and components in which the public hospital had lower performance included Adjuvant Therapy, Surgery, Surveillance, Testing, and Respect for Patient Preferences.

DISCUSSION

When we used the NICCQ breast cancer quality indicators, we found that the quality of breast cancer treatment at the public hospital was good with 82% overall adherence, and 17 of the 31 individual indicators had scores of ≥85%. From an institutional standpoint, analysis of these indicators identified areas of good and lower quality in the continuum of breast cancer care. Indicators with <85% performance identified specific clinical processes in most need for quality improvement. By analyzing performance in the aggregated Quality of Care Domains and Components of Care, we were able to identify more global and systems-level elements that can be addressed to achieve more widespread impact with our quality-improvement efforts.

Comparison of indicator adherence to the previously published 5-city NICCQ cohort allowed us to assess quality of care at the public hospital relative to a study cohort that was more likely to be representative of breast cancer patients in the United States. The public hospital and 5-city patient cohorts were significantly different in terms of age, race/ethnicity, language spoken, and insurance status. As expected, the public hospital cohort had sociodemographic characteristics that suggested this was a more vulnerable population and, therefore, at-risk for decreased access to care and/or suboptimal quality of care. By comparison, the 5-city cohort was more similar to the national breast cancer population than to the public hospital cohort. Although overall performance at the public hospital was lower compared with the ASCO study, the magnitude of the difference was small. Of course, public hospitals can vary significantly in terms of size, organizational structure, patient demographics, and services provided. Some features of the study site which may have led to higher performance include a multidisciplinary tumor board that is attended by physicians in all treating disciplines, including surgery, medicine, and radiation oncology. All patients receiving breast cancer treatment are discussed by the tumor board. Also, medical oncology treatment takes place on site at the hospital, which allows close communication between the surgical and oncologic services. Although radiation oncology is off-site, there is also close communication between the on-site and off-site providers. In comparison, patients treated in private hospitals often receive care in a more fragmented manner through different institutions and unaffiliated providers. Areas in which the public hospital performed better, such as in Referrals and Documentation of Key Clinical Factors, may have benefited from the limited options for uninsured patients to choose other providers and the paradoxically higher degree of continuity of care between surgical and oncologic disciplines that this may create. Although our study findings do not indicate which features of the public hospital resulted in higher quality care, they do suggest that such care is possible, at least in selected settings. Also, many other public hospitals have close affiliations with academic centers, and this may be of benefit on some of the indicators.

Nevertheless, our analysis identified areas of deficiencies in the quality of care provided at the public hospital that are consistent with known disparities in quality of care across socioeconomic strata. Overall the quality of care was good, with 82% adherence to the indicators. However, adherence was lower in the public hospital compared with the 5-city cohort in several indicators that could have a significant impact on patient outcomes in terms of recurrence and survival. For example, clear margins were achieved in 88% of patients who underwent BCS at the public hospital compared with 99% in the 5-city cohort, receipt of appropriate chemotherapy in patients with stage II-III cancer was 83% versus 94%, and receipt of appropriate hormone therapy was 85% versus 91%. These results were presented to our multidisciplinary staff, and targeted approaches to improve care were subsequently implemented, specifically, receipt of hormone treatment, surveillance mammography, and better patient education. A patient registry for tracking receipt of care was also initiated.

This study had several limitations. First, unlike the 5-city ASCO study, there was no patient survey conducted for the public hospital cohort because of resource constraints. Therefore, we had to rely on information limited to the medical records. Specifically, for the 3 indicators assessing Respect for Patient Preferences, we measured whether a relevant discussion was documented in the medical record instead of asking patients whether they were informed of their possible treatment options,. This might have introduced measurement bias into these indicators and resulted in lower adherence rates. For indicators that assessed receipt of treatment, it is possible that patients received some aspects of their treatment at other facilities. This, too, might have introduced measurement bias and resulted in lower adherence rates. However, this is not likely to be a significant factor in our study cohort given their limited options for access to care.

There were additional differences between the public hospital and the 5-city cohort in terms of patient selection and timeframe, and variations in performance were more challenging to interpret. For example, temporal changes in care between 1998 and 2005 might have contributed to some of the measured differences in performance. Documentation within the medical record was often incomplete. Finally, this was a single-institution study and it may lack generalizability. Our study, thus, suggests that high-quality care can be provided to traditionally underserved patients at a public hospital but cannot prove that it is routinely provided at all municipal hospitals.

On the basis of the findings of this study, several interventions have been implemented to improve performance and quality of care in areas identified as deficient. For example, post-treatment surveillance will be automatically initiated through the radiology department, which already had a system in place to follow abnormal screening mammograms; that system will be extended to ensure that patients receive timely surveillance mammograms after curative treatment. Importantly, many cultural and language barriers in this vulnerable patient population were likely responsible for some of the lower rates of receipt of care. Recently, a patient navigator program (bilingual and bicultural) has addressed some of these barriers, but future assessments are warranted. Furthermore, a prospective patient registry was instituted to track all components of patient treatment to ensure receipt of appropriate care. Our findings show that not only can high-quality care be provided to an underserved population but also that evaluations of performance can identify additional areas that are ripe for quality improvement.

CONFLICT OF INTEREST DISCLOSURES

Veterans Affairs (VA) Greater West Los Angeles Department of Surgery provided support for Dr. Chen and Dr. Maggard Gibbons. Dr. Maggard Gibbons' time was supported in part by a grant from the Robert Wood Johnson Physician Faculty Scholars program and by the NIH/NCI R25 CA 87,949 UCLA Cancer Education and Career Development Program. Funding was also provided through seed grants from the Jonsson Cancer Center Foundation and Olive View-UCLA Medical Center. We thank the AVON Cares for Life program, which supported the patient navigator program in our study.

Ancillary