Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne, Victoria, Australia.
Diabet Med. 2011 Apr;28(4):414-23. doi: 10.1111/j.1464-5491.2010.03188.x.
To evaluate how to most efficiently screen populations to detect people at high risk of incident Type 2 diabetes and those with prevalent, but undiagnosed, Type 2 diabetes.
Data from 5814 adults in the Australian Diabetes, Obesity and Lifestyle study were used to examine four different types of screening strategies. The strategies incorporated various combinations of cut-points of fasting plasma glucose, the non-invasive Australian Type 2 Diabetes Risk Assessment Tool (AUSDRISK1) and a modified version of the tool incorporating fasting plasma glucose (AUSDRISK2). Sensitivity, specificity, positive predictive value, screening costs per case of incident or prevalent undiagnosed diabetes identified and intervention costs per case of diabetes prevented or reverted were compared.
Of the four strategies that maximized sensitivity and specificity, use of the non-invasive AUSDRISK1, followed by AUSDRISK2 in those found to be at increased risk on AUSDRISK1, had the highest sensitivity (80.3%; 95% confidence interval 76.6-84.1%), specificity (78.1%; 95% confidence interval 76.9-79.2%) and positive predictive value (22.3%; 95% confidence interval 20.2-24.4%) for identifying people with either prevalent undiagnosed diabetes or future incident diabetes. It required the fewest people (24.1%; 95% confidence interval 23.0-25.2%) to enter lifestyle modification programmes, and also had the lowest intervention costs and combined costs of running screening and intervention programmes per case of diabetes prevented or reverted.
Using a self-assessed diabetes risk score as an initial screening step, followed by a second risk score incorporating fasting plasma glucose, would maximize efficiency of identifying people with undiagnosed Type 2 diabetes and those at high risk of future diabetes.
评估如何最有效地对人群进行筛查，以发现有发生 2 型糖尿病风险的人群和有已确诊但未诊断的 2 型糖尿病的人群。
使用来自澳大利亚糖尿病、肥胖和生活方式研究的 5814 名成年人的数据，研究了四种不同的筛查策略。这些策略结合了空腹血浆葡萄糖的不同切点、非侵入性澳大利亚 2 型糖尿病风险评估工具（AUSDRISK1）和纳入空腹血浆葡萄糖的改良版工具（AUSDRISK2）的各种组合。比较了每种策略识别新发病例或已确诊但未诊断的糖尿病病例的灵敏度、特异性、阳性预测值、每例筛查成本以及预防或逆转每例糖尿病的干预成本。
在四种能够最大程度提高灵敏度和特异性的策略中，使用非侵入性的 AUSDRISK1，然后对 AUSDRISK1 结果提示风险增加的人群使用 AUSDRISK2，具有最高的灵敏度（80.3%；95%置信区间 76.6-84.1%）、特异性（78.1%；95%置信区间 76.9-79.2%）和阳性预测值（22.3%；95%置信区间 20.2-24.4%），可用于识别有已确诊但未诊断的糖尿病或未来新发糖尿病的人群。它需要最少的人（24.1%；95%置信区间 23.0-25.2%）进入生活方式改变计划，并且具有最低的干预成本和每例预防或逆转的糖尿病的筛查和干预计划总成本。
使用自我评估的糖尿病风险评分作为初始筛查步骤，然后使用纳入空腹血浆葡萄糖的第二个风险评分，可以最大限度地提高识别未确诊 2 型糖尿病患者和未来糖尿病高危人群的效率。
