Epidemiologic Assessment of Quality Gastric Cancer Surgery

INTRODUCTION: In May 2013, the American College of Surgeons Commission on Cancer (CoC) endorsed new gastric cancer quality measures recommending at least 15 lymph nodes (LN) in all resected non-metastatic non-proximal gastric cancers. We sought to determine whether hospital CoC-accreditation predicts compliance with this guideline. We also postulated that 3-month post-gastrectomy survival may differ for patients receiving cancer-directed surgeries in CoC vs non-CoC hospitals and that CoC LN count compliance (15+ vs 1-14) may predict gastric cancer survival, independent of CoC accreditation.

METHODS: We conducted a non-concurrent population-based cohort study in the statewide California Cancer Registry assessing survival through July 2012 among gastrectomy patients having AJCC stage I-III adenocarcinoma of the non-proximal stomach, 2004-2010. Logistic regression analysis assessed whether CoC accreditation (yes/no) predicted odds of compliance (yes/no) with the CoC minimum LN count guideline. Cox proportional mortality hazard ratios assessed 3-month post-gastrectomy survival for CoC-accredited hospitals (yes/no) and separately assessed mortality hazard ratios by compliance with the CoC LN guideline (yes/no) independent of CoC accreditation. Controlled covariates included age (continuous), race/ethnicity (4 categories), stage at diagnosis (6 categories) and diagnosis year (continuous).

RESULTS: Of 3,321 stage I-III gastric cancers, 50.5% received gastrectomies in CoC-accredited hospitals and 30.1% in non-CoC facilities. 44.7% of patients in CoC-accredited hospitals and 34.8% in non-CoC hospitals had 15+ LNs resected. Adjusting for covariates, current CoC-accreditation (yes/no) independently predicted removal of 15+ vs 1-14 LN (OR=1.61;95%CI=1.36-1.91). Adjusted Cox proportional hazards revealed that CoC-accreditation did not independently predict survival (HR CoC yes/no=0.98;95%CI=0.87-1.09), although compliance (yes/no) with the CoC 15+ LN count recommendation predicted lower mortality hazards (HR LN_{15+ vs. LN1-14}=0.69;95%CI=0.62-0.76), independent of CoC accreditation.

CONCLUSIONS: Odds of 15+ vs 1-14 LNs examined was higher in CoC vs non-CoC hospitals, while survival did not differ in CoC vs non-CoC hospitals. CoC LN count compliance predicted improved survival, regardless of CoC accreditation, suggesting that patients treated in CoC hospitals experienced unmeasured conditions, including comorbidities, adversly affecting survival.