心理援助热线来电者自杀危险预测模型的建立和信效度检验

Valid prediction for suicidal behavior is the key to suicide prevention. In this project, suicidal risk evaluation is applied to psychological aids hotline, to establish a suicide risk assessment model, to develop a suicidal risk assessment scale for the callers, and test the reliability and validity of the scale. All the effective calls to Beijing psychological aid hotline would be recruited. The first effective call of each callers is treated as baseline call. The suicidal risk assessment scale is administered to each callers during their baseline calls. The callers will be divided into three groups, there are high risk, moderate risk, or low risk to suicide, based on the assessment. In the next 12 months after the baseline call, all the high suicidal risk callers would be followed up in the next day, 1 week, 1 month, 3 month and 12 month; and 20% random selected callers who were assessed as moderate suicidal risk would be followed up in the 3 and 12 month; and 10% random selected callers who were assessed as low suicidal risk would be followed up in the 12 month. Whether the callers commit suicidal behavior or not in the follow up period would be identified. A suicidal behavior is defined as "a self-injurious behavior with or without a fatal outcome, and there is evidence that the callers intended at some (nonzero) level to kill him/herself". The predictive validity of the scale would be assessed by the results of the follow up study. The tape records of the baseline calls of randomly selected 10 high suicidal risk callers, 10 moderate suicidal risk callers, and 20 low suicidal risk callers would be recruited. To assess the inter-rater reliability of the scale, every crisis hotline operaters will listen to these 40 tape records and complete the scale independently. Completing the scale and the following up will not interfere with crisis treatments flow of the hotline.

有效预测自杀行为是自杀干预的关键问题。本项目将自杀危险评估与心理援助热线电话结合，依据定性分析研究建立热线来电者自杀危险的预测模型，并编制适合热线使用的标准化评估工具。在研究入组期间被北京市心理援助热线有效接听的来电者首次来电作为基线来电接受自杀危险程度评估。评估结果分为自杀高危、中危和低危。被评为高危的所有来电者将于基线来电后12个月内接受5次随访。随机抽取20%中危和10%低危来电者于基线来电后12个月内分别进行2次和1次随访。随访内容为从基线来电或上次随访至今有无发生自杀行为。凡基线来电后12个月内出现来电者自己实施的有结束自己生命意图的伤害行为定义为有自杀行为。根据随访结果评估危机来电者自杀危险程度评估工具的预测效度。随机抽取10个高危来电、10个中危来电和20个低危来电评估录音，接线员听录音独立评分以检验本工具评定员之间一致性。本项目的评估与随访过程不干扰现有的危机来电干预流程。

心理热线干预是自杀预防的重要措施之一，但需要事先对来电者的自杀行为风险进行有效预测。本项目在既往研究结果基础上建立热线来电者自杀危险预测模型并编制了自杀行为危险程度评估量表。该量表包括抑郁、生活事件、严重躯体疾病、目前自杀意念、既往自杀史、物质滥用等12个项目，采用加权相加的办法计算来电者自杀行为危险程度得分，总分范围0-16分。来电者可被划分为高危（8-16分）、中危（4-7分）和低危（0-3分）。该量表被做成软件化纳入热线呼叫系统。热线接线员使用呼叫系统对来电者进行自杀危险程度评估以及心理干预。所有接线员经过量表使用的培训，其中6名接线员通过听40个来电录音的方式分别再次独立评定来电者的自杀行为程度，检验量表的评定者之间一致性。量表总分的一致性（ICC）系数为0.79，各项目的一致性为0.56-0.95。从2015年1月到2016年9月，共有9982个来电者接受了该量表评定，评为自杀低危、中危和高危的分别为2112、5450、2420例。我们对所有高危，50%的中危和50%的低危来电者预约了12个月的随访，最终有4482例来电者完成了随访。随访期间，共有305例来电者自杀未遂、11例自杀死亡。以随访期间自杀行为（未遂或死亡）的发生为阳性结局，以≥4分为临界值，中危+高危的灵敏度为97.75%，特异度为16.38%，阳性预测值为7.85%，阴性预测值98.98%。以≥8分为临界值的高危灵敏度为74.28%，特异度为59.99%，阳性预测值为11.95%，阴性预测值97.16%。该量表的ROC曲线下的面积为0.739，说明有较好的预测能力。基线来电时被评估为高危、中危和低危的来电者，他们的自杀风险有明显的差异，高危来电者在早期即出现高比例的自杀未遂行为，而低危来电者持续处于较低的风险水平，中危来电者的自杀风险缓慢上升。Cox比例风险模型分析结果显示，与低危来电者相比，高危来电者的自杀行为危险上升到约12倍（HR= 12.50, 95% CI：7.42-21.06），中危来电者的自杀行为危险上升为3倍（HR=3.55, 95% CI：2.02-6.26）。本研究结果的科学意义在于，我们可以较好地预测热线来电者此后1年内的自杀行为，并据此将有限的自杀干预资源放在自杀危险最高的来电者身上，提高预防效果。本研究结果还可以进一步推广至全国范围内的心理援助热线。