Department of Surgery, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina.
Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina.
J Surg Res. 2022 Dec;280:304-311. doi: 10.1016/j.jss.2022.07.036. Epub 2022 Aug 26.
There are multiple measures of area socioeconomic status (SES) and there is little evidence on the comparative performance of these measures. We hypothesized adding area SES measures improves model ability to predict guideline concordant care and overall survival compared to models with standard clinical and demographic data alone.
We included patients with colorectal cancer from 2006 to 2015 from the North Carolina Cancer Registry merged with insurance claims data. The primary area SES study variables were the Social Deprivation Index, Distressed Communities Index, Area Deprivation Index, and Social Vulnerability Index. We used multivariable logistic modeling and Cox proportional hazards modeling to assess the adjusted association of each indicator, with guideline concordant care and overall survival, respectively. Model performance of the SES measures was compared to a base model using likelihood ratio testing and area under the curve (AUC) assessments to compare SES indicator models with each other.
We found that the Area Deprivation Index, Social Vulnerability Index and Social Deprivation Index, but not Distressed Communities Index, were significantly associated with receiving guideline concordant care and significantly improved model fit over the base model on likelihood ratio testing. All models had similar AUCs. With respect to overall survival, we found that all indices were independently and significantly associated with survival and had significantly improved model fit over the base model on likelihood ratio testing. AUC analysis again showed all area SES measures had comparable performance for overall survival at 5 y.
This analysis demonstrates the importance of including these measures in risk adjustment models. However, of the commonly available measures, no one measure stood out as superior to others.
有多种衡量区域社会经济地位（SES）的方法，但关于这些衡量方法的比较性能的证据很少。我们假设，与仅使用标准临床和人口统计学数据的模型相比，加入区域 SES 衡量标准可以提高模型预测指南一致护理和总体生存率的能力。
我们纳入了 2006 年至 2015 年来自北卡罗来纳癌症登记处与保险索赔数据合并的结直肠癌患者。主要的区域 SES 研究变量是社会剥夺指数、困境社区指数、区域剥夺指数和社会脆弱性指数。我们使用多变量逻辑建模和 Cox 比例风险建模来评估每个指标与指南一致的护理和总体生存率的调整关联。使用似然比检验和曲线下面积（AUC）评估来比较 SES 指标模型彼此之间的关系，比较 SES 措施模型的性能与基本模型。
我们发现，区域剥夺指数、社会脆弱性指数和社会剥夺指数与接受指南一致的护理显著相关，并且在似然比检验中，与基本模型相比，模型拟合度显著提高。所有模型的 AUC 相似。关于总体生存率，我们发现所有指数均与生存率独立且显著相关，并且在似然比检验中，与基本模型相比，模型拟合度显著提高。AUC 分析再次表明，所有区域 SES 衡量标准在 5 年时对总体生存率的表现均相当。
这项分析表明了在风险调整模型中纳入这些衡量标准的重要性。然而，在常用的衡量标准中，没有一个衡量标准明显优于其他衡量标准。
