Measuring Institutional Quality in Head and Neck Surgery Using Hospital-Level Data: Negative Margin Rates and Neck...

Negative margins and lymph node yields (LNY) of 18 or more from neck dissections in patients with head and neck squamous cell carcinomas (HNSCC) have been associated with improved patient survival. It is unclear whether these metrics can be used to identify hospitals with improved outcomes.

To determine whether 2 patient-level metrics would predict outcomes at the hospital level.

A retrospective review of records from the National Cancer Database (NCDB) was used to identify patients who underwent primary surgery and concurrent neck dissection for HNSCC between 2004 and 2013. The percentage of patients at each hospital with negative margins on primary resection and an LNY 18 or more from a neck dissection was quantified. Cox proportional hazard models were used to define the association between hospital performance on these metrics and overall survival.

Margin status and lymph node yield at hospital level. Overall survival (OS).

We identified 1008 hospitals in the NCDB where 64 738 patients met inclusion criteria. Of the 64 738 participants, 45 170 (69.8%) were men and 19 568 (30.2%) were women. The mean SD age of included patients was 60.5 (12.0) years. Patients treated at hospitals attaining the combined metric of a 90% or higher negative margin rate and 80% or more of cases with LNYs of 18 or more experienced a significant reduction in mortality (hazard ratio [HR] 0.93; 95% CI, 0.89-0.98). This benefit in survival was independent of the patient-level improvement associated with negative margins (HR, 0.73; 95% CI, 0.71-0.76) and LNY of 18 or more (HR, 0.85; 95% CI, 0.83-0.88). Including these metrics in the model neutralized the association of traditional measures of hospital quality (volume and teaching status).

Treatment at hospitals that attain a high rate of negative margins and LNY of 18 or more is associated with improved survival in patients undergoing surgery for HNSCC. These surgical outcome measures predicted outcomes independent of traditional, but generally nonmodifiable characteristics. Tracking of these metrics may help identify high-quality centers and provide guidance for institution-level quality improvement.