Chanani Sheila, Wacksman Jeremy, Deshmukh Devika, Pantvaidya Shanti, Fernandez Armida, Jayaraman Anuja
Society for Nutrition, Education and Health Action, Mumbai, India
Dimagi, Inc., New Delhi, India.
Food Nutr Bull. 2016 Dec;37(4):504-516. doi: 10.1177/0379572116657241. Epub 2016 Jul 1.
Acute malnutrition is linked to child mortality and morbidity. Community-Based Management of Acute Malnutrition (CMAM) programs can be instrumental in large-scale detection and treatment of undernutrition. The World Health Organization (WHO) 2006 weight-for-height/length tables are diagnostic tools available to screen for acute malnutrition. Frontline workers (FWs) in a CMAM program in Dharavi, Mumbai, were using CommCare, a mobile application, for monitoring and case management of children in combination with the paper-based WHO simplified tables. A strategy was undertaken to digitize the WHO tables into the CommCare application.
To measure differences in diagnostic accuracy in community-based screening for acute malnutrition, by FWs, using a mobile-based solution.
Twenty-seven FWs initially used the paper-based tables and then switched to an updated mobile application that included a nutritional grade calculator. Human error rates specifically associated with grade classification were calculated by comparison of the grade assigned by the FW to the grade each child should have received based on the same WHO tables. Cohen kappa coefficient, sensitivity and specificity rates were also calculated and compared for paper-based grade assignments and calculator grade assignments.
Comparing FWs (N = 14) who completed at least 40 screenings without and 40 with the calculator, the error rates were 5.5% and 0.7%, respectively (p < .0001). Interrater reliability (κ) increased to an almost perfect level (>.90), from .79 to .97, after switching to the mobile calculator. Sensitivity and specificity also improved significantly.
The mobile calculator significantly reduces an important component of human error in using the WHO tables to assess acute malnutrition at the community level.
急性营养不良与儿童死亡率和发病率相关。基于社区的急性营养不良管理（CMAM）项目有助于大规模检测和治疗营养不良。世界卫生组织（WHO）2006年身高别体重/身长表格是用于筛查急性营养不良的诊断工具。孟买达拉维一个CMAM项目的一线工作人员（FWs）使用移动应用程序CommCare结合纸质版WHO简化表格对儿童进行监测和病例管理。开展了一项将WHO表格数字化并录入CommCare应用程序的策略。
通过FWs使用基于移动设备的解决方案，测量社区筛查急性营养不良时诊断准确性的差异。
27名FWs最初使用纸质表格，然后改用包含营养分级计算器的更新版移动应用程序。通过比较FWs给出的分级与每个儿童根据相同WHO表格应得的分级，计算与分级分类具体相关的人为错误率。还计算并比较了纸质版分级和计算器分级的Cohen kappa系数、敏感度和特异度。
比较至少完成40次无计算器筛查和40次有计算器筛查的FWs（N = 14），错误率分别为5.5%和0.7%（p < .0001）。改用移动计算器后，评分者间信度（κ）从0.79提高到0.97，几乎达到完美水平（>.90）。敏感度和特异度也显著提高。
移动计算器显著降低了在社区层面使用WHO表格评估急性营养不良时人为错误的一个重要组成部分。
