Department of Family Medicine, McMaster University, 1280 Main Street West, David Braley Health Sciences Centre 3rd Floor, Hamilton, Ontario, L8S 4K1, Canada.
Ateneo de Zamboanga University, Zamboanga City, Philippines.
BMC Public Health. 2019 Aug 27;19(1):1169. doi: 10.1186/s12889-019-7402-0.
As the prevalence of type 2 diabetes (T2DM) increases in low- to middle-income countries, the burden on individuals and health care systems also increases. The use of diabetes risk assessment tools could identify those at risk, leading to prevention or early detection of diabetes. The aim of this study was to evaluate the appropriateness of 6 existing T2DM risk screening tools in detecting dysglycemia in Zamboanga City, Philippines.
This study used a case-control design in an urban setting in the southern Philippines. There were 200 participants in two groups: 1) those diagnosed with diabetes (n = 50; recruited from diabetes clinics) and 2) those with no previous diagnosis of diabetes (n = 150; recruited from community locations). Participants completed six tools (the Finnish Diabetes Risk Score [FINDRISC], the Canadian Diabetes Risk Score [CANRISK], the Indian Diabetes Risk Score [IDRS], the American Diabetes Association [ADA] risk score, an Indonesian undiagnosed diabetes mellitus [UDDM] scoring system, and a Filipino tool). Scores were compared to fasting plasma glucose levels, which are recommended in Philippines clinical practice guidelines as a valid, available, and low cost option for T2DM diagnosis. Appropriateness of tools was determined through accuracy, sensitivity, specificity, positive/negative predictive value (PPV, NPV), and positive/negative likelihood ratios.
The Filipino tool had the highest specificity (0.73) and PPV (0.27), but lowest sensitivity (0.68). The IDRS and Indonesian UDDM tool had the highest NPV at 0.96, but were not amongst the highest in other scores. The CANRISK tied for highest area under the receiver operating characteristic (ROC) curve (AUC), AUC (0.80), but other scores were not noteworthy. Overall, the FINDRISC was the most effective with highest sensitivity (0.94), tied for highest AUC (0.80), and with middle scores in other variables (specificity: 0.45, PPV: 0.20, NPV: 0.95), when using the published cut-off score of 9. When increasing the cut-off score to 11, specificity increased (0.71) and sensitivity was not greatly affected (0.86).
Our results suggest that the FINDRISC is more suitable than other known diabetes risk assessment tools in an urban Filipino population; effectiveness increased with a higher cut-off score.
随着 2 型糖尿病（T2DM）在中低收入国家的流行率上升，个人和医疗保健系统的负担也随之增加。使用糖尿病风险评估工具可以识别出处于风险中的人群，从而预防或早期发现糖尿病。本研究旨在评估 6 种现有的 T2DM 风险筛查工具在菲律宾三宝颜市检测糖基化血红蛋白中的适用性。
本研究采用病例对照设计，在菲律宾南部的城市环境中进行。有 200 名参与者分为两组：1）确诊为糖尿病的患者（n=50；从糖尿病诊所招募）和 2）无糖尿病既往诊断的患者（n=150；从社区地点招募）。参与者完成了 6 种工具（芬兰糖尿病风险评分[FINDRISC]、加拿大糖尿病风险评分[CANRISK]、印度糖尿病风险评分[IDRS]、美国糖尿病协会[ADA]风险评分、印度尼西亚未诊断的糖尿病评分系统[UDDM]和菲律宾工具）。评分与空腹血浆葡萄糖水平进行比较，空腹血浆葡萄糖水平是菲律宾临床实践指南中推荐的用于 T2DM 诊断的有效、可用和低成本的选择。工具的适宜性通过准确性、敏感性、特异性、阳性/阴性预测值（PPV、NPV）和阳性/阴性似然比来确定。
菲律宾工具的特异性最高（0.73）和 PPV（0.27），但敏感性最低（0.68）。IDRS 和印度尼西亚 UDDM 工具的 NPV 最高，为 0.96，但在其他评分中并不高。CANRISK 的 ROC 曲线下面积（AUC）最高（0.80），但其他评分则不显著。总体而言，FINDRISC 是最有效的，敏感性最高（0.94），AUC 最高（0.80），其他变量的评分居中（特异性：0.45，PPV：0.20，NPV：0.95），当使用发表的 9 分切点时。当将切点提高到 11 分时，特异性增加（0.71），敏感性影响不大（0.86）。
我们的结果表明，在菲律宾城市人群中，FINDRISC 比其他已知的糖尿病风险评估工具更适合；随着切点的提高，其有效性也随之增加。
