Center for Statistical Consultation and Research, University of Michigan, 3555 Rackham, Ann Arbor, MI 48109-1070, USA.
J Clin Psychiatry. 2012 Oct;73(10):e1269-75. doi: 10.4088/JCP.12m07658.
To identify factors recorded in electronic medical chart progress notes associated with suicide among patients who had received treatment for depression.
The retrospective study sample consisted of 324 randomly selected US Veterans Health Administration (VHA) patients treated for depression who died by suicide from April 1, 1999, to September 30, 2004, stratified by geographic region, gender, and year of depression cohort entry and 312 control patients with depression who were alive on the date of suicide death (index date) and were from the same stratum as the matched suicide patient. In addition to constructing variables from administrative data, variables were abstracted from electronic medical chart notes in the year prior to the index date in 5 categories: clinical symptoms and diagnoses, substance use, life stressors, behavioral/ideation measures (eg, suicide attempts), and treatments received. Logistic regression was used to assess the associations.
Even after we adjusted for administratively available data, suicidal behaviors and substance-related variables were the strongest independent predictors of suicide. Prescription drug misuse had an odds ratio (OR) of 6.8 (95% CI, 2.5-18.5); history of suicide attempts, 6.6 (95% CI, 1.7-26.4); and alcohol abuse/dependence, 3.3 (95% CI, 1.9-5.7). Difficulty with access to health care was a predictor of suicide (OR = 2.9; 95% CI, 1.3-6.3). Receipt of VHA substance abuse treatment was protective (OR = 0.4; 95% CI, 0.1-0.9).
Prescription drug and alcohol misuse assessments should be prioritized in suicide assessments among depressed patients. Additionally, behavioral measures noted in electronic chart records may be useful in health system monitoring and surveillance and can potentially be accessed using word search or natural language processing approaches.
确定电子病历进展记录中与接受抑郁症治疗的患者自杀相关的记录因素。
本回顾性研究样本包括 324 名在美国退伍军人健康管理局（VHA）接受抑郁症治疗后自杀的患者，他们的自杀时间为 1999 年 4 月 1 日至 2004 年 9 月 30 日，按地理区域、性别和抑郁症队列进入年份分层，并选取 312 名在自杀死亡日期（索引日期）存活的抑郁症对照患者，且与匹配的自杀患者处于同一分层。除了从行政数据中构建变量外，还从索引日期前一年的电子病历记录中提取了 5 个类别中的变量：临床症状和诊断、物质使用、生活应激源、行为/观念措施（如自杀企图）和接受的治疗。使用逻辑回归评估关联。
即使在我们调整了行政上可用的数据后，自杀行为和与物质相关的变量仍然是自杀的最强独立预测因素。处方药物滥用的比值比（OR）为 6.8（95%CI，2.5-18.5）；自杀企图史为 6.6（95%CI，1.7-26.4）；酒精滥用/依赖为 3.3（95%CI，1.9-5.7）。获得医疗保健的困难是自杀的预测因素（OR=2.9；95%CI，1.3-6.3）。接受 VHA 物质滥用治疗具有保护作用（OR=0.4；95%CI，0.1-0.9）。
在对抑郁患者进行自杀评估时，应优先评估处方药物和酒精滥用评估。此外，电子病历记录中的行为措施可能有助于卫生系统监测和监测，并且可以使用词搜索或自然语言处理方法来获取。
