精神分裂症住院病患攻擊行為之異質性與危險因子研究

Heterogeneity and Risk factors of Aggressive Acts among Schizophrenic Inpatients

陳杏佳(Shing-Chia Chen)

指導教授 ： 胡海國 ； 熊秉荃

國立臺灣大學/醫學院/護理學研究所/博士(2005年)

https://doi.org/10.6342/NTU.2005.00060

摘要

背景：精神疾病住院病患的攻擊行為在精神科單位是常見且重要的問題，目前對其臨床的認知與照護尚待充實與加強。因此針對攻擊行為，建構客觀的評量工具，探究其異質性結構和相關危險因子乃重要的課題，藉此改善臨床攻擊行為的處置與預防。
目的：本研究旨在探究精神分裂症住院病患攻擊行為的數個面向：描述與表徵其攻擊行為，並確認與指陳其異質性結構，及探尋與建立其預測的危險因子。主要探討的問題涵蓋下列四項：(1)建構攻擊行為的客觀評量工具，(2) 瞭解精神分裂症住院病患攻擊行為的發生特色，(3)探索精神分裂症住院病患攻擊行為的異質性結構，以及(4)確認精神分裂症住院病患攻擊行為的危險因子。
方法：本研究的概念架構主要參酌自Robert, Herman, 與 Hope的「自殺和暴力的兩階段模式」和MacArthur研究的「暴力風險評估模式」，藉以發展本研究對精神分裂症病患攻擊行為的異質性和危險因子之探討方向。首先，經由翻譯和回覆翻譯的步驟，發展以次數尺度來測量攻擊行為的中文版暴力量表，檢視其內容和建構效度以及內在一致性。該中文板量表涵蓋三種攻擊行為類型：(1)朝向物品(AggP)，(2)朝向他人(AggO)，及(3)朝向自己(AggS)。資料收集採用前瞻性個體時間序列的研究設計，針對某大學教學醫學中心的精神科急性病房，在徵得病患簽字同意後，收集民國90年4月至91年3月一年間內陸續入院接受治療，且出院診斷符合美國精神醫學會所出版「診斷與統計手冊第四版」的精神分裂症診斷標準之病患研究樣本共107位。本研究資料蒐集方法乃採用中文版暴力量表，參與觀察他們在住院期間至出院為止，每天各種攻擊行為的發生次數，累計其原始次數成為以週為單位的重複測量資料，但每位樣本病患依其實際住院期間而有不同的週數長度資料；並使用已發展的研究工具評量危險因子，包括「過去一個月的攻擊史」、「狀態特質憤怒表達量表」與「病患基本人口學與臨床資料」，「病房限制量表」，以及「精神症狀量表」。再者，精神分裂症住院病患攻擊行為的異質性結構可由下列三項指標評價之：(1) 病患住院後一週，其不同類別或類型攻擊行為同時發生之關係，(2) 病患攻擊性嚴重度在不同類別或類型的住院前一個月攻擊行為與住院一週後攻擊行為的相關性，以及(3) 以病患不同類別的住院前一個月攻擊行為確認其住院一週後攻擊行為之一致性。統計分析方法主要是運用SPSS12.0，S-PLUS6.0，以及LISREL8.43軟體，採用描述性統計、卡方檢定、皮爾森相關係數、驗證性因素分析、ROC曲線分析、複迴歸和交互作用分析。
結果：本研究發現暴力行為具有卜瓦松分配(Poisson distribution)和過度分散(over-dispersion)的情形，中文版暴力量表(VS-C) 共有16題，其信度在可接受的範圍，但不具較高值的內在一致性（Cronbach’s α = .67)。精簡版僅擷取全版量表項目中具核心意義的潛在變項6題，稱為「常見朝外的攻擊行為」。本研究的樣本病患平均年齡為33.4歲 (標準差=11.9歲)，其中多半為單身 (68.2%)、女性(69.2%)、無業 (73.8%)， 且大多信仰東方型態的宗教 (57.9%)。平均住院期間為4週(標準差=2.4週，範圍=2至13週)，約四分之ㄧ為非自願性住院 (26.2%).
初次非特定症狀發生的平均年紀為24.1歲，107位病患中，91位(85.0%) 有在住院前一個月發生攻擊行為的紀錄，63位病患(58.9%)在住院首週發生攻擊行為(平均值=10, SD=9.8, 範圍= 1 to 46)，住院前一個月攻擊行為發生率依次為 AggO (70.1%)，AggP (44.9%)，與AggS (32.7%)，住院後首週攻擊行為發生率依次為AggO (47.7%)，AggP (27.1%)，與AggS (14.0%)，AggO類別發生的頻率較其他兩類高。所有攻擊行為在病患住院治療後皆降低發生的次數，依週次計算，攻擊行為發生率降低最多的為“與他人大聲口頭爭辯”, “低層次的敵意”, 以及 “以威脅態度運用物品但並未損壞”。住院期間攻擊行為發生率逐週降低至第二週33.3%、第三週26.8%及第四週16.9%，大多數病患的攻擊行為是在住院兩週即快速消失。
研究結果顯示樣本病患的攻擊行為呈現下列三點異質性結構的證據：(1) 住院後首週的AggO 與 AggP，特別是肢體行為的AggP (r= .51)，具高度同時發生率(r = .52)；而住院後首週只有肢體行為的AggO與AggS具同時發生性(r = .24)。(2) 病患攻擊性嚴重度在朝向物品、他人、或自我的不同類別之住院前一個月與住院後首週攻擊行為各自具有顯著的高度相關性 (配對的朝向物品, r = .46，朝向他人, r = .50，朝向自己, r = .41；以及朝向物品史和朝向他人, r = .34，朝向他人史和朝向物品, r= .43)；然而相較於朝向自我的攻擊行為與其他兩類皆具低度相關性。(3) 病患之住院前一個月與住院後首週攻擊行為在相對應的類別具高度一致性(AggP, 72.0%； AggO, 72.6%；AggS, 70.7%)；此外住院前一個月的AggP 與住院後首週的AggO(64.7%)，以及住院前一個月的AggO與住院後首週的AggP (64.6%)兩項皆具高度一致性；然而相較於朝向自我的攻擊行為與其他兩類則具低度一致性。
病患住院期間週平均次數的AggP之危險因子為住院前一個月週平均次數的AggP、活性與負性精神症狀三項。病患住院期間週平均次數的AggO之危險因子為住院前一個月週平均次數的AggO、情緒性精神症狀、以及住院前一個月週平均次數的AggO與性別之交互作用三項。
病患住院期間週平均次數的AggS之危險因子僅有住院前一個月的AggS乙項。顯而易見的，攻擊行為三種類型的危險因子皆不相同。整體而言，能夠顯著預測單一潛在因素“常見朝外的攻擊行為”之危險因子為住院前一個月的攻擊行為的相對應類型、精神症狀以及性別三項。
討論：本研究的強處乃在引用客觀測量工具與前瞻性觀察研究設計來探討精神分裂症住院病患的攻擊行為，過往的相關研究多半採用回溯式報告，通常可能低估病患攻擊行為的發生且僅能提供橫斷面的資料。本研究在探究精神病患在住院後每週的攻擊行為發生的相關訊息及其動態特質，確認攻擊行為的危險因子並考量其發展方向對形成攻擊行為的預測。攻擊行為的發生率在住院四週後顯現降低83.1%，此發現的訊息有助於強化精神分裂症病患住院後攻擊行為的處置與管理。病患在住院時首週特定類型的攻擊行為大約有70% 可以由住院前一個月的相對應類型預測得知，此類事證對臨床精神病患攻擊行為的預防極為重要。AggP可藉由高階的活性與低階的負性精神病症狀預測得知，AggO也可由高階的情緒性精神症狀預測之。而依據住院前一個月的AggO對住院後首週AggO的預測在男性病患的部分較女性病患準確度高，性別對AggP與AggS的預測則無影響。
精神分裂症病患住院發生攻擊行為的危險因子並不包括憤怒特質的個人變數，可能是因為生病的急性期可能會降低憤怒特質對攻擊行為的影響，憤怒特質也許在社區病患扮演生病穩定期發生的攻擊行為的重要角色，此項有待進一步研究。

關鍵字

攻擊行為 ； 發生率 ； 異質性 ； 住院病患 ； 精神分裂症 ； 危險因子

並列摘要

Background: Psychiatric inpatients’ aggressive acts can occur frequently and are recognized as a significant problem in psychiatric settings. Understanding of aggressive acts in clinical settings is still incomplete and the clinical care of aggressive acts remains unsatisfactory. It is therefore crucial to establish objective measures and to develop insights into the heterogeneity and related risk factors of aggressive acts, for improvement in the management and prevention of aggressive acts.
Objective: The purpose of this study has a many-fold intention depicted as follows: to describe and characterize schizophrenic inpatients’ aggressive acts, to identify and specify heterogeneity in their structure, and to look for and set up risk factors for prevention. This study investigates four major issues: (1) Establishing an objective behavior rating scale for measuring aggressive acts; (2) Understanding the characteristics and incidence of aggressive acts by inpatients with schizophrenia; (3) Exploration of the heterogeneous structure of aggressive acts by inpatients with schizophrenia; and (4) Identifying the risk factors of aggressive acts of inpatients with schizophrenia.
Method: The “Two-Stage Model of suicide and violence” coined by Plutchik, van Praag, and Conte and MacArthur’s “Violence Risk Assessment Model” have been incorporated in developing the conceptual framework of this study. At first, by back translation, an English version of the violence scale (VS) was translated into Chinese (VS-C) with counts scaling, including three categories of aggressive acts: (1) towards property (AggP), (2) towards others (AggO), and (3) towards self (AggS). The content validity, construct validity, and internal consistency of VS-C were examined. Prospective panel design was used for data collection. Patients fulfilling the DSM-IV criteria of schizophrenia in the acute psychiatric ward of a university teaching hospital were recruited over a one-year period (April 2001 to March 2002) as study subjects, after signing informed consent. The sample size was 107. Aggressive acts were rated daily using the VS-C by participant observation during the hospitalization period. The raw counts of aggressive acts were summed weekly as a data unit. The risk factors of schizophrenic inpatients’ aggressive acts were measured using “Past Month History of Aggressive Acts”, “State-Trait Anger Expression Inventory”, “Patients’ Basic Demographic and Clinical Data”, “Ward Restriction Scale”, and “Psychiatric Symptoms Checklist”. The heterogeneous structure of aggressive acts was assessed by three indicators: (1) the concurrence of specific categories and types of aggressive acts in the initial week after hospitalization, (2) the correlation of the severity of aggressiveness among specific categories and types between past-month and initial-week aggressive acts, and (3) the concordance of specific categories between past-month and initial-week aggressive acts in sensitivity, specificity, and positive and negative predictive value.
The software applications SPSS12.0, S-PLUS6.0, and LISREL8.43 were used for statistical analysis. The statistical methods included descriptive statistics, Chi-square test, Pearson’s correlation, Confirmatory Factor Analysis (CFA), Receiver Operating Characteristic (ROC) curves analysis, multiple regression model and interaction analysis.
Results: There was a Poisson distribution and over-dispersion on aggressive acts found in this study. The VS-C with 16 items showed an acceptable quality, but lacked a high internal consistency value (Cronbach’s α = .67). A shorter version of the VS-C could be drawn with one latent variable of 6 items sharing a core meaning labeled as “common outward aggressive acts.” In this study, participating subjects had a mean age of 33.4 (SD = 11.9). The majority of them was single (68.2%), female (69.2%), unemployed (73.8%), and had an Eastern religious affiliation (57.9%). The average duration of their hospitalization was 4 weeks (SD = 2.4, range = 2 to 13). About one quarter of them were involuntarily admitted (26.2%).
The mean age of the onset of initial nonspecific symptoms was 24.1 years old. Ninety-one out of 107 patients (85.0%) had a history of past-month aggressive acts, and 63 patients (58.9%) exhibited aggressive acts (mean = 10, SD = 9.8, range = 1 to 46) in the initial week after admission. The past-month incidence rate of aggressive acts was in the order of AggO (70.1%), AggP (44.9%), and AggS (32.7%). The initial-week incidence rate of aggressive acts was in the order of AggO (47.7%), AggP (27.1%), and AggS (14.0%). The category AggO occurred more frequently than the other two categories. All aggressive acts decreased during the treatment course of patients’ hospitalization. The incidence rate of aggressive acts that decreased the most, over weeks, were “loud verbal arguments with another person”, “low grade hostility”, and “used property in a threatening manner without damage”. The incidence rate of aggressive acts decreased to 33.3%, 26.8%, and 16.9% in the following second, third, and forth week respectively during hospitalization. Most patients’ aggressive acts were quickly decreased in two weeks.
The study subjects’ aggressive acts were heterogeneous as evidenced by: (1) a high concurrent rate of AggO with AggP (r = .52), especially with physical AggP (r = .51), whereas only physical AggO was concurrent with AggS (r = .24) in the initial week after hospitalization, (2) a high significant correlation of the severity of aggressive acts between corresponding categories of past-month and initial-week aggressive acts (toward property, r= .46 ; toward others, r = .50; toward self, r = .41; and Past-month AggP with AggO, r = .34, and Past-month AggO with AggP, r = .43), whereas the correlation between aggressive acts toward self and the other two categories were all low, and (3) a high concordance rate of corresponding categories of past-month and initial-week aggressive acts (toward property, 72.0%; toward person, 72.6%; toward self, 70.7%). The concordance between past-month AggP and initial-week AggO (64.7%), past-month AggO and initial-week AggP (64.6%) were also high, whereas the concordance between aggressive acts towards the self and other two categories were low.
The risk factors of weekly averages of AggP during hospitalization were weekly averages of past-month AggP, and positive and negative psychiatric symptoms. The risk factors of weekly averages of AggO during hospitalization were weekly averages of past-month AggO, affective psychiatric symptom, and the interaction of weekly averages of past-month AggO and gender factor. The risk factors of weekly averages of AggS during hospitalization were only the weekly averages of past-month AggS. Apparently, the risk factors of these three categories of aggressive acts were different. Globally, the risk factors that significantly predict the single latent factors of “common outward aggressive acts.” were the corresponding category of past-month aggressive acts, psychiatric symptoms, and gender factor.
Discussion: The strength of this study on aggressive acts of inpatients with schizophrenia was a prospective observation study design using an objective measure. Previous studies of patients’ aggressive acts mainly used retrospective reports, which might underestimate the occurrence of patients’ aggressive acts, and which provided a cross-sectional data only. This study revealed information on aggressive acts occurring weekly after admission into the psychiatric ward. In addition, to identify the risk factors of aggressive acts and consider their developmental directions to predict the formation of aggressive acts. This brings insight into the dynamic nature of aggressive acts of schizophrenia inpatients. Further, the incidence rate of aggressive acts decreased of 83.1% after 4 weeks of admission. This information is useful for emphasizing prevention and management of aggressive acts of inpatients with schizophrenia after admission. The specific category of initial-week aggressive acts was accurately predicted around seventy percent of the time by the corresponding category occurring in the past month prior to admission. This evidence is crucial for clinical psychiatric service in the prevention of aggressive acts. AggP was also predicted by high level of the positive and low level of the negative psychiatric symptoms; the AggO was also predicted by high level of the affective psychiatric symptoms. Male patients with a past-month history of AggO predict AggO more accurately than the same for female patients. Gender has no effect on the prediction of the other two categories of AggP and AggS.
The risk factors of aggressive acts among schizophrenic inpatients occurring during hospitalization did not include the personality variable of an anger trait. This might be due to the fact that the acute phase of illness might reduce the effect of the anger trait on aggressive acts. The anger trait may play a role on aggressive acts that occur in the stabilized state of illness. This needs to be studied further.

並列關鍵字

aggressive act ； incidence rate ； heterogeneity ； inpatient ； schizophrenia ； risk factor

參考文獻

Suicide and schizophrenia: A nationwide psychological

characteristics of 92 suicide victims with

polymorphism in schizophrenic patients: Further

behaviors of children with a schizophrenic patients--

Young, A., Taylor, S.G., & Mclaughlin-Renpenning, K.