Davis W K, Hess G E, Hiss R G
Department of Postgraduate Medicine/Health Professions Education, University of Michigan Medical School, Ann Arbor 48109.
Diabetes Care. 1988 Jul-Aug;11(7):538-45. doi: 10.2337/diacare.11.7.538.
The goal of this research was to quantify the relationships between patient survival and a set of explanatory variables in a randomly selected sample of community-based patients with non-insulin-dependent diabetes mellitus (NIDDM). The sample included 343 patients with NIDDM initially entered into the study in 1981-1982 and reexamined in 1985-1986. Mortality data were collected on reexamination in 1985 and updated from death-certificate data through 1 January 1986. The data collected from the patients included demographic and clinical variables, psychosocial variables related to diabetes, measures of physiologic control, hospitalization, and mortality. The Cox proportional-hazards model was used to compute a hazard rate for each individual and to determine risk covariates. The results indicated that the variables most associated with the risk of mortality were patient age, social impact of diabetes, renal function, complexity of diet regimen, and history of smoking. Two of these variables (social impact and complexity of diet regimen) were obtained from the Diabetes Educational Profile completed by all patients on entry to the study. The five predictor variables were more closely related to mortality than diabetes control as measured by HbA1, previous hospital admissions, previous heart attacks, and other physiologic measures frequently used as outcome measures. The only physiologic predictor was renal function.
本研究的目的是在一组随机选取的社区非胰岛素依赖型糖尿病（NIDDM）患者样本中，量化患者生存率与一组解释变量之间的关系。该样本包括343例NIDDM患者，他们于1981 - 1982年首次进入研究，并于1985 - 1986年进行复查。1985年复查时收集了死亡率数据，并通过死亡证明数据更新至1986年1月1日。从患者收集的数据包括人口统计学和临床变量、与糖尿病相关的心理社会变量、生理控制指标、住院情况和死亡率。采用Cox比例风险模型计算每个个体的风险率并确定风险协变量。结果表明，与死亡风险最相关的变量是患者年龄、糖尿病的社会影响、肾功能、饮食方案的复杂性和吸烟史。其中两个变量（社会影响和饮食方案的复杂性）是通过所有患者在进入研究时填写的糖尿病教育概况表获得的。这五个预测变量与死亡率的关系比通过糖化血红蛋白（HbA1）、既往住院次数、既往心脏病发作以及其他常用作结局指标的生理指标所衡量的糖尿病控制更为密切。唯一的生理预测指标是肾功能。
