中国科学院心理研究所机构知识库

目的 基于国际量表分析并验证烧伤在院患者自杀意念预测模型的影响因素。方法 回顾性分析住院进行治疗的194例烧伤患者的临床资料，采用一般资料问卷、失眠严重指数量表(ISI)、汉密尔顿抑郁量表(HAMD)、汉密尔顿焦虑量表(HAMA)、急性应激障碍量表(ASDS)、精简烧伤健康量表(BSHS-B)对患者进行自杀意念影响因素的评估，根据有无自杀意念分为自杀意念组和无自杀意念组，分别比较组间基线资料，单因素筛选有意义的变量并进一步进行多因素Logistic回归建模，ROC分析评估模型区分度，并进行内部验证。结果 基线资料分析结果显示，与无自杀意念组(173/194)相比，自杀意念组(21/194)中年龄、体质量指数(BMI)、教育年限、吸烟史、烧伤面积估算百分比、有无头面颈烧伤、有无臀部及会阴烧伤、疼痛数字评分差异无统计学意义；性别(P=0.047)、有无躯干烧伤(P=0.022)、烧伤严重程度(中度烧伤：P=0.002;重度烧伤：P=0.458;特重度烧伤：P=0.169)、ISI评分(P=0.001)、HAMD评分(P=0.001)、HAMA评分(P<0.001)、ASDS评分(P=0.003)、BSHS-B评分(P=0.011)差异均有统计学意义。多因素Logistic回归分析的结果显示，烧伤严重程度(中度烧伤：OR=0.103,P=0.009;重度烧伤：OR=0.351,P=0.223;特重度烧伤：OR=0.103,P=0.095)及HAMA评分(OR=1.136,P=0.007)是烧伤患者伴发自杀意念的独立影响因素。通过两个独立影响因素建立Logistic回归预测模型，ROC分析结果显示模型的区分度较好(AUC=0.880,95%CI:0.808～0.952,P<0.001)且内部验证准确率为79.38%。结论 基于烧伤严重程度、HAMA评分两个独立影响因素构建的预测模型，具有良好的预测精度，有助于临床医生对烧伤在院患者有自杀意念的尽早进行干预措施，以期降低发生率并丰富临床心理相关的研究。

Objective To analyze and verify the factors influencing the prediction model of suicidal ideation of burn patients in hospital based on international scale． Methods The clinical data of 194 burn patients treated in hospital were retrospectively analyzed． General data questionnaire，ISI，HAMD，HAMA，ASDS and BSHS-B were used to evaluate the influencing factors of suicidal ideation． According to the presence or absence of suicidal ideation， the patients were divided into the suicidal ideation group and the non-suicidal ideation group． The baseline data between the groups were compared，univariate screening of meaningful variables was conducted，and multivariate Logistic regression modeling was further conducted． ROC analysis evaluated model differentiation，and internal verification was conducted． Results According to the baseline data analysis results，there were no statistically significant differences in age，BMI，years of education，smoking history，estimated percentage of burned area，head and neck burns，hip and perineal burns，and pain scores in the suicidal ideation group ( 21 /194) compared with the non-suicidal ideation group ( 173 /194) ． Gender ( P = 0. 047) ，presence or absence of trunk burn ( P = 0. 022) ， severity of burn ( moderate burn: P = 0. 002; severe burn: P = 0. 458; extremely severe burn: P = 0. 169) ，ISI score ( P = 0. 001) ，HAMD score ( P = 0. 001) ，HAMA score ( P ＜ 0. 001) ，ASDS score ( P = 0. 003) ，BSHS-B score ( P = 0. 011) had statistical significance． Multivariate Logistic regression analysis showed that the severity of burn ( moderate burn: OＲ = 0. 103，P = 0. 009; severe burn: OＲ = 0. 351，P = 0. 223; extremely severe burn: OＲ = 0. 103，P = 0. 095) and HAMA score ( OＲ = 1. 136，P = 0. 007) were independent influencing factors for burn patients with suicidal ideation。The Logistic regression prediction model was established by two independent influencing factors． ＲOC analysis results showed that the model had good differentiation ( AUC = 0. 880，95% CI: 0. 808 － 0. 952，P ＜ 0. 001) and the internal verification accuracy was 79. 38% ． Conclusion The prediction model built on the basis of two independent influencing factors，burn severity and HAMA score，has a good prediction accuracy，which is helpful for clinicians to intervene as soon as possible for burn patients with suicidal ideation in hospital，in order to reduce the incidence and enrich clinical psychological research．

国家自然科学基金(编号：81272092);; 中国科学院科研项目(编号：GJHZ2020)