从烧伤识别疑似虐待:BuRN-Tool的推导与验证
研究基于1327例烧伤儿童的前瞻性数据,用逻辑回归推导出预测虐童调查转诊的BuRN-Tool,并在787例新病例中验证。工具以3分为转诊切点,对烫伤和非烫伤的灵敏度和特异度均较高,曲线下面积分别为0.87和0.85。结论认为该工具可作为急诊临床决策的辅助手段。
要点
- BuRN-Tool包含年龄<5岁、社会照护知晓、解释存疑、全层烧伤、非典型部位、双侧 pattern 和监管担忧等变量。
- 以3分为切点,烫伤灵敏度87.5%、特异度81.5%;非烫伤灵敏度82.4%、特异度78.7%。
- 该工具简单易用,可在急诊科辅助识别需虐待调查的烧伤儿童。
核心结论
- BuRN-Tool可预测哪些烧伤儿童需要接受儿童虐待调查,可作为临床决策的辅助工具。
研究设计与样本
- 前瞻性收集1327例烧伤儿童数据,用逻辑回归推导工具,并在787例新病例中验证。
主要发现
- 与转诊相关的变量包括年龄<5岁、社会照护知晓、解释存疑、全层烧伤、非典型部位、双侧 pattern 和监管担忧。
- 烫伤验证数据灵敏度87.5%(95%CI:61.7-98.4)、特异度81.5%(95%CI:77.1-85.4),曲线下面积0.87(95%CI:0.83-0.90)。
- 非烫伤验证数据灵敏度82.4%(95%CI:65.5-93.2)、特异度78.7%(95%CI:73.9-82.9),曲线下面积0.85(95%CI:0.81-0.88)。
关键数值与推荐
- BuRN-Tool以3分为转诊切点,由接收者操作特征曲线分析确定。
临床怎么用
- 在急诊科对烧伤儿童使用BuRN-Tool评分,≥3分时考虑转诊进行儿童虐待调查。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: 10-25% of childhood burns arise from maltreatment. AIM: To derive and validate a clinical prediction tool to assist the recognition of suspected maltreatment. METHODS: Prospectively collected data from 1327 children with burns were analyzed using logistic regression. Regression coefficients for variables associated with 'referral for child maltreatment investigation' (112 cases) in multivariable analyses were converted to integers to derive the BuRN-Tool, scoring each child on a continuous scale. A cut-off score for referral was established from receiver operating curve analysis and optimal sensitivity and specificity values. We validated the BuRN-Tool on 787 prospectively collected novel cases. RESULTS: Variables associated with referral were: age <5years, known to social care, concerning explanation, full thickness burn, uncommon body location, bilateral pattern and supervision concern. We established 3 as cut-off score, resulting in a sensitivity and specificity for scalds of 87.5% (95% CI:61.7-98.4) and 81.5% (95% CI:77.1-85.4) respectively and for non-scalds sensitivity was 82.4% (95%CI:65.5-93.2) and specificity 78.7% (95% CI:73.9-82.9) when applied to validation data. Area under the curve was 0.87 (95% CI:0.83-0.90) for scalds and 0.85 (95% CI:0.81-0.88) for non-scalds. CONCLUSION: The BuRN-Tool is a potential adjunct to clinical decision-making, predicting which children warrant investigation for child maltreatment. The score is simple and easy to complete in an emergency department setting.
引用
- Kemp AM, Hollén L, Emond AM, et al. Raising suspicion of maltreatment from burns: Derivation and validation of the BuRN-Tool. Burns. 2017. PMID: 28918905. doi:10.1016/j.burns.2017.08.018
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。