Patel Shivani A, Ali Mohammed K, Narayan K M Venkat, Mehta Neil K
Am J Epidemiol. 2016 Dec 15;184(12):933-942. doi: 10.1093/aje/kww081. Epub 2016 Nov 17.
We examined factors responsible for variation in cardiovascular disease (CVD) mortality across US counties in 2009-2013. We linked county-level census, survey, administrative, and vital statistics data to examine 4 sets of features: demographic factors, social and economic factors, health-care utilization and features of the environment, and population health indicators. County-level associations of these features (standardized to a mean of 0 with a standard deviation of 1) with cardiovascular deaths per 100,000 person-years among adults aged 45-74 years was modeled using 2-level hierarchical linear regression with random intercept for state. The percentage of CVD mortality variation (intercounty disparity) modeled by each set of features was quantified. Demographic composition accounted for 36% of county CVD mortality variation, and another 32% was explained after inclusion of economic/social conditions. Health-care utilization, features of the environment, and health indicators explained an additional 6% of CVD mortality variation. The largest contributors to CVD mortality levels were median income (-23.61 deaths/100,000 person-years, 95% CI: -26.95, -20.26) and percentage without a high school education (20.71 deaths/100,000 person-years, 95% CI: 16.48, 24.94). In comparison, the largest health-related contributors were health-care utilization (19.35 deaths/100,000 person-years, 95% CI: 16.36, 22.34) and CVD risk factors (4.80 deaths/100,000 person-years, 95% CI: 2.14, 7.46). Improving health-care access and decreasing the prevalence of traditional CVD risk factors may reduce county CVD mortality levels, but improving socioeconomic circumstances of disadvantaged counties will be required in order to reduce CVD mortality disparities across counties.
我们研究了2009 - 2013年美国各县心血管疾病（CVD）死亡率差异的相关因素。我们将县级人口普查、调查、行政和生命统计数据相联系，以考察4组特征：人口统计学因素、社会和经济因素、医疗保健利用及环境特征，以及人口健康指标。使用具有州随机截距的二级分层线性回归模型，对这些特征（标准化为均值为0、标准差为1）与45 - 74岁成年人中每10万人年心血管死亡人数的县级关联进行建模。对每组特征所建模的CVD死亡率差异（县间差异）百分比进行了量化。人口构成占县CVD死亡率差异的36%，纳入经济/社会状况后又解释了另外的32%。医疗保健利用、环境特征和健康指标又解释了CVD死亡率差异的6%。对CVD死亡率水平贡献最大的因素是收入中位数（-23.61例死亡/10万人年，95%置信区间：-26.95，-20.26）和未接受高中教育的人口百分比（20.71例死亡/10万人年，95%置信区间：16.48，24.94）。相比之下，与健康相关的最大贡献因素是医疗保健利用（19.35例死亡/10万人年，95%置信区间：16.36，22.34）和CVD危险因素（4.80例死亡/10万人年，95%置信区间：2.14，7.46）。改善医疗保健可及性并降低传统CVD危险因素的患病率可能会降低县CVD死亡率水平，但为了减少各县之间的CVD死亡率差异，需要改善弱势县的社会经济状况。
