The purpose of this study was to identify factors associated with a high risk of involuntary psychiatric in-patient hospitalization both on the individual level and on the level of mental health services and the socioeconomic environment that patients live in. The present study expands on a previous analysis of the health records of 5764 cases admitted as in-patients in the four psychiatric hospitals of the Metropolitan City of Cologne, Germany, in the year 2011 (1773 cases treated under the Mental Health Act and 3991 cases treated voluntarily). Our previous analysis had included medical, sociodemographic and socioeconomic data of every case and used a machine learning-based prediction model employing chi-squared automatic interaction detection (CHAID). Our current analysis attempts to improve the previous one through (1) optimizing the machine learning procedures (use of a different type of decision-tree prediction model (Classification and Regression Trees (CART) and application of hyperparameter tuning (HT)), and (2) the addition of patients’ environmental socioeconomic data (ESED) to the data set. Compared to our previous analysis, model fit was improved. Main diagnoses of an organic mental or a psychotic disorder (ICD-10 groups F0 and F2), suicidal behavior upon admission, admission outside of regular service hours and absence of outpatient treatment prior to admission were confirmed as powerful predictors of detention. Particularly high risks were shown for (1) patients with an organic mental disorder, specifically if they were retired, admitted outside of regular service hours and lived in assisted housing, (2) patients with suicidal tendencies upon admission who did not suffer from an affective disorder, specifically if it was unclear whether there had been previous suicide attempts, or if the affected person lived in areas with high unemployment rates, and (3) patients with psychosis, specifically those who lived in densely built areas with a large proportion of small or one-person households. Certain psychiatric diagnoses and suicidal tendencies are major risk factors for involuntary psychiatric hospitalization. In addition, service-related and environmental socioeconomic factors contribute to the risk for detention. Identifying modifiable risk factors and particularly vulnerable risk groups should help to develop suitable preventive measures.

中文翻译：

---

识别非自愿精神病住院的风险因素：利用环境社会经济数据和机器学习方法来改进预测。


本研究的目的是确定与非自愿精神科住院患者住院高风险相关的因素，无论是在个人层面还是在心理健康服务水平以及患者居住的社会经济环境方面。本研究扩展了之前的一项研究对 2011 年德国科隆市四家精神病医院收治的 5764 例住院患者的健康记录进行了分析（1773 例根据《精神卫生法》接受治疗，3991 例自愿接受治疗）。我们之前的分析包括每个病例的医学、社会人口统计和社会经济数据，并使用基于机器学习的预测模型，该模型采用卡方自动交互检测（CHAID）。我们当前的分析试图通过（1）优化机器学习程序（使用不同类型的决策树预测模型（分类和回归树（CART）以及超参数调整（HT）的应用）来改进前一个分析，以及（ 2）将患者的环境社会经济数据（ESED）添加到数据集中，与我们之前的分析相比，模型拟合度有所提高，主要诊断为器质性精神障碍或精神障碍（ICD-10 组 F0 和 F2）、自杀倾向。入院时的行为、正常服务时间以外入院以及入院前未接受门诊治疗均被证实是拘留的有力预测因素。 对于 (1) 患有器质性精神障碍的患者，特别是退休、在正常服务时间以外入院并住在辅助住房中的患者，(2) 入院时有自杀倾向但未患有情感障碍的患者，风险特别高。障碍，特别是如果不清楚以前是否有过自杀企图，或者受影响的人是否居住在失业率高的地区，以及 (3) 精神病患者，特别是居住在人口稠密、人口比例较高的地区的精神病患者。或一个人的家庭。某些精神病诊断和自杀倾向是非自愿精神病住院的主要危险因素。此外，与服务相关的因素和环境社会经济因素也会增加拘留风险。识别可改变的风险因素和特别脆弱的风险群体应有助于制定适当的预防措施。