Variation in the Quality of Head and Neck Cancer Care in the United States

Abstract

This evaluation of quality metrics uses data from the National Cancer Database to summarize hospital-level overall quality of care for patients with head and neck cancer in the United States.

Variation in clinical care is one of the biggest obstacles facing health care organizations that prevents them from improving outcomes.¹ In previous work, Cramer et al² validated 5 quality metrics for head and neck cancer that meet validity criteria. High adherence to these metrics was associated with a 19% reduced hazard of mortality, suggesting strong suitability for more widespread adoption.² However, although variation of quality of care has been studied elsewhere, variation of these quality metrics has not been examined in depth. Commission on Cancer hospitals, which have been accredited based on several benchmarks for comprehensive, high-quality cancer care, treat approximately 70% of patients with head and neck cancer in the United States. Therefore, variation among these centers may identify opportunities for improvement that could be addressed through existing quality improvement programs.

Methods

From the National Cancer Database, we identified adult patients with surgically treated invasive, nonmetastatic squamous cell carcinoma of the oral cavity, oropharynx, larynx, and hypopharynx who were treated from January 1, 2004, through December 31, 2014. Using previously described methods, we calculated 5 quality metrics, including negative surgical margins, neck dissection lymph node yield of 18 or more, appropriate adjuvant radiotherapy for T3-T4 or N2-N3 disease (excluding patients with human papillomavirus), appropriate adjuvant chemoradiotherapy for positive margins or extranodal extension (excluding patients with human papillomavirus), and adjuvant therapy within 6 weeks.² We then calculated a hospital-level summary measure of overall quality that included in the denominator all the metrics for which a participant was at risk and in the numerator all the metrics for which a participant received recommended care, resulting in a proportion ranging from 0% to 100%.^2,3 We compared overall quality based on hospital volume and safety-net burden. Hospital volume was categorized into terciles. Hospital safety-net burden was calculated based on the percentage of uninsured and Medicaid-insured patients as described previously and categorized into quartiles.⁴ We excluded hospitals with a total volume of fewer than 20 patients. Data were analyzed from November 1, 2017, through May 31, 2018. All statistical analyses were performed using SPSS software (version 24.0; IBM). The institutional review board of University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, determined this study to be exempt from review and informed consent.

Results

A total of 72 322 patients from 770 hospitals met our inclusion and exclusion criteria (69.8% male and 30.2% female; mean [SD] age, 60.9 [12.5] years). The mean (SD) overall quality score at all hospitals was 69.2% (30.4%). However, overall mean quality across hospitals varied widely from 45% to 90%. Similar wide variation in overall quality existed across low- (range, 48%-89%), intermediate- (range, 45%-90%), and high-volume (range, 52%-82%) hospitals (Figure 1). This variation was also present among academic (range, 48%-82%), integrated (range, 54%-80%), community (range, 48%-89%), or comprehensive (range, 45%-90%) cancer centers. Correspondingly wide variation existed across the highest (range, 45%-84%), second highest (range, 44%-82%), third highest (range, 45%-85%), and lowest (range, 51%-90%) safety-net burden quartiles for hospitals (Figure 2).

Figure 1. Association of Hospital Volume and Quality.

Variation in quality is divided by volume terciles. Hospitals are ordered based on quality scores. Error bars indicate 95% CI. Each point represents a hospital-level summary measure of overall quality of the 5 individual quality metrics, including negative surgical margins, neck dissection lymph node yield of 18 or more, appropriate adjuvant radiotherapy for T3-T4 or N2-N3 disease, appropriate adjuvant chemoradiotherapy for positive margins or extranodal extension, and adjuvant therapy within 6 weeks. During the study period, high-volume centers treated 73 to 1520 cases; medium-volume centers, 38 to 72 cases; and low-volume centers, 20 to 37 cases.

Figure 2. Association of Safety-Net Burden With Quality.

Variation in quality is divided by safety-net burden quartiles as determined by the percentage of patients who are uninsured or insured under Medicaid. Each point represents a hospital-level summary measure of overall quality of the 5 individual quality metrics, including negative surgical margins, neck dissection lymph node yield of 18 or more, appropriate adjuvant radiotherapy for T3-T4 or N2-N3 disease, appropriate adjuvant chemoradiotherapy for positive margins or extranodal extension, and adjuvant therapy within 6 weeks. Quartile 1 represents a safety-net burden of less than 5.5%; quartile 2, 5.5% to 10.3%; quartile 3, 10.3% to 16.7; and quartile 4, greater than 16.7%.

Discussion

We identified widespread variation in the quality of care for head and neck cancer across Commission on Cancer hospitals. This variation is present across high- and low-volume hospitals as well as safety-net institutions, which suggests that participation in the Commission on Cancer or high-volume status alone does not ensure uniform high-quality care. Given challenges with travel and insurance that many patients face, centralizing all care of head and neck cancer at high-volume facilities with a low safety-net burden is not feasible.⁵ Instead, we should identify centers of excellence that consistently achieve high-quality care so that these lessons can be shared. In particular, understanding how some low-volume and safety-net hospitals reliably achieve high-quality care is important. Opportunities to spread this knowledge include regional collaboration and quality support initiatives through the Commission on Cancer, such as the Rapid Quality Reporting Tool, that should be encouraged to improve the care of patients with head and neck cancer.⁶ Such information would be valuable for physicians to identify quality improvement targets and for administrators to optimize network service lines.