Division of Research and Optimal Patient Care, American College of Surgeons, Chicago, IL, USA.
Ann Surg Oncol. 2013 Jun;20(6):1766-73. doi: 10.1245/s10434-013-2867-z. Epub 2013 Jan 26.
Quality initiatives are increasingly focusing on the quality of oncologic surgery. However, there is concern that a lack of cancer-specific variables may make risk-adjusted hospital quality comparisons inadequate. Our objective was to assess whether hospital quality rankings for cancer surgery are influenced by the addition of cancer-specific variables to the risk-adjusted models.
Patients from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) and National Cancer Data Base (NCDB) who underwent colon or rectal resection for cancer were linked (2006-2008). Hierarchical models were developed predicting ACS NSQIP outcomes based on ACS NSQIP only vs a model using NSQIP and NCDB-derived cancer variables (e.g., stage and neoadjuvant therapy). Changes in hospital quality rankings were compared.
A total of 11,405 patients underwent colon (n = 9,678, 146 hospitals) or rectal (n = 1,727, 135 hospitals) resection for cancer (2006-2008). Hospital-level complication rates (and standard deviation) after colon surgery were 2.2 % (±2.7 %) for mortality and 17.2 % (±8.7 %) for serious morbidity. After rectal cancer resection, complication rates were 0.9 % (±3.8 %) for mortality and 22.3 % (±20.4 %) for serious morbidity. When cancer-specific variables were included in risk-adjustment, outlier agreement was very good (kappa >0.85), and hospital odds ratio correlations were nearly identical (R > 0.98) for all outcomes assessed. Median changes in hospital rankings with the addition of the cancer-specific variables ranged from 1 to 2 after colon resection to 2-4 after rectal resection.
Addition of the available cancer-specific variables to risk-adjustment models did not affect hospital quality rankings for cancer surgery. Existing ACS NSQIP risk-adjustment variables appears to be sufficient for accurate comparisons of hospital quality.
质量改进措施越来越关注肿瘤外科的质量。然而，人们担心缺乏癌症特异性变量可能会使风险调整后的医院质量比较不充分。我们的目的是评估将癌症特异性变量添加到风险调整模型中是否会影响癌症手术的医院质量排名。
从美国外科医师学院国家外科质量改进计划（ACS NSQIP）和国家癌症数据库（NCDB）中选取 2006 年至 2008 年间接受结肠癌或直肠癌切除术治疗癌症的患者进行链接。基于仅使用 ACS NSQIP 或使用 NSQIP 和 NCDB 衍生的癌症变量（例如，分期和新辅助治疗）的模型，开发了预测 ACS NSQIP 结果的分层模型。比较了医院质量排名的变化。
共有 11405 例患者接受结肠癌（n=9678，146 家医院）或直肠癌（n=1727，135 家医院）切除术治疗癌症（2006-2008 年）。结肠癌手术后的医院级并发症发生率（标准差）为死亡率 2.2%（±2.7%）和严重发病率 17.2%（±8.7%）。直肠癌手术后的并发症发生率为死亡率 0.9%（±3.8%）和严重发病率 22.3%（±20.4%）。当将癌症特异性变量纳入风险调整时，异常值一致性非常好（kappa>0.85），并且评估的所有结果的医院优势比相关性几乎相同（R>0.98）。添加癌症特异性变量后，结肠癌切除术后医院排名中位数的变化范围为 1-2，直肠癌切除术后为 2-4。
将可用的癌症特异性变量添加到风险调整模型中不会影响癌症手术的医院质量排名。现有的 ACS NSQIP 风险调整变量似乎足以进行医院质量的准确比较。
