论文部分内容阅读
目的本文通过检验45岁及以上的人群体重指数(BMI)与跌倒性伤害发生的联系,详尽地阐述了复杂抽样Poisson回归分析方法及应用的必要性。方法本报告应用2010年美国德克萨斯州BRFSS数据,分析了跌倒性伤害与体重指数的联系强度。结果普通Poisson回归分析高估了肥胖、患有心血管疾病、小于25000美元的年收入与跌倒性伤害的联系;低估了中等和差的健康状况、未被雇用、未婚和受教育程度低与跌倒性伤害的联系,但假设检验结果没有变;与复杂抽样负二项回归(NB)分析结果相比,年龄与跌倒性伤害的联系强度没有改变,其统计学检验也保持未变。数据拟合复杂抽样NB回归,离散参数α=8.15,标准误1.68,α的95%CI=5.44-12.22,表明跌倒性伤害数据的方差是其均数的8.15倍,因此复杂抽样NB回归模型更适合BRFSS数据。结论普通Poisson回归分析低估了参数估计的方差和标准误,造成了肥胖、患有心血管疾病、小于25000美元的年收入与跌倒性伤害的假关联关系;复杂抽样NB回归调整了模型的参数估计,适合多阶段抽样设计的数据分析。
Objective This article elaborates the necessity of using complex sampling Poisson regression analysis method and its application by examining the association between body mass index (BMI) and falls injury in people aged 45 years and above. Methods This report uses 2010 Texas BRFSS data to analyze the association between falls injury and body mass index. General Poisson regression analysis overestimated the association between obesity and cardiovascular disease, annual income of less than $ 25,000 and fall injuries; underestimated moderate and poor health status, unexpired, unmarried and low education and falls However, the test results were not changed. Compared with the negative binomial regression (NB) analysis of complex samples, the correlation between age and fall injuries did not change, and the statistical test remained unchanged. Data fitting complex sampling NB regression, the discrete parameter α = 8.15, standard error 1.68, α 95% CI = 5.44-12.22, indicating that the variance of fall injury data is 8.15 times the average, so complex sampling NB regression model more Suitable for BRFSS data. Conclusion Ordinary Poisson regression analysis underestimates the variance and standard error of the parameter estimation, resulting in a fake association between obesity and cardiovascular diseases, with an annual income of less than 25000 USD and falling injuries. The complex sampling NB regulates the parameter estimation of the model, Data Analysis for Multistage Sampling Design.