疑有精神病發病前狀態者之精神病發作預測：神經心理測驗與精神分裂性人格問卷的比較

Abstract

目標：本研究主要針對疑有精神病發病前狀態的青少年或年輕成人，早期偵測他們如何轉變成精神病發作，並比較神經心理學測驗和精神分裂性人格問卷在轉變為精神病和未轉變為精神病兩組間的差別。此外，本研究試圖了解這些有精神病發病前狀態的族群中，神經心理學測驗和精神分裂性人格問卷對於日後精神分裂症是否有預測能力，並比較其預測力，期望能提昇早期篩檢之精確度以利早期介入並降低精神分裂症所造成的影響。 方法：本研究為前瞻性追蹤研究，經由臨床評估符合精神病極高風險標準的59人進行追蹤。所有參加研究的個案均先接受臨床訪談，利用潛伏性精神分裂症量表(LSS)、遺傳問卷診斷問卷(DIGS)、前驅症狀量表(SOPS)所結合的思覺量表來評估精神病風險。接著在神經心理學方面，利用電腦版的威斯康辛卡片分類測驗(WCST)測量執行能力和連續表現測驗（CPT）測量持續注意力缺失。在精神分裂性人格特質方面，利用知覺異常量表(PAS)、準精神分裂人格量表(SPQ)、中國人健康量表(CHQ)，三種自陳量表進行測量，並且使用青少年自陳量表測量(YSR)測行為問題。最後利用追蹤期間發病與未發病作為分組，以Cox比例風險回歸分析探討精神病發病與神經心理學和精神分裂性人格特質的關係並建立預測模型，因為個案資料具有左截斷資料的特性，因此利用左截斷右設限的統計方法進行調整，接著比較模型間的預測力。 結果：平均追蹤兩年，總共有21位個案最後轉變為精神病。臨床特徵方面，前軀症狀量表中的正性症狀在發病族群中有顯著較高的分數。神經心理學測驗方面，威斯康辛卡片分類測驗的卡片分類成功次數以及非固執反應錯誤次數在發病族群中有較差的表現。精神分裂性人格特質和行為問題方面，社交缺損、中國人健康量樣表分數以及退縮問題在發病族群較未發病族群低。在Cox比例風險回歸分析上分為三種模型作比較，只考慮神經心理學的預測模型、只考慮精神分裂性人格特質的預測模型以及全部考慮的預測模型，其中全部納入考慮的預測模型有較高的ROC。 討論：部分神經心理學測驗指標以及精神分裂性人格特質與極高風險族群最後轉變為精神病有關連性。結合神經心理學以及精神分裂性人格特質在預測發病上有很好的預測能力。這差異可能在精神病發病過程中扮演著重要的角色，因此需要的進一步了解，可以幫助提昇早期篩檢的精確度以利於早期介入。

Aim: This study aimed to 1) assess whether the scores on neuropsychological performance or schizotypy questionnaires were associated with psychosis initiation among subjects with suspected pre-psychotic state; and 2) compare the predictive accuracy of transition to psychosis by different models incorporating neuropsychological tests, schizotypy questionnaires, or both. Method: Fifty-nine subjects at ultra-high-risk (UHR) of psychosis were followed for a mean of 2 years. Subjects were interviewed using the Diagnostic Interview for Genetic Studies (DIGS) and Scale of Prodromal Symptoms (SOPS). Schizotypy was measured using the Perceptual Aberration Scale (PAS), Schizotypal Personality Questionnaire (SPQ), Chinese Health Questionnaire (CHQ) and Youth Self Report (YSR). Neuropsychological functioning was evaluated using Continuous Performance Test (CPT) and Wisconsin Card Sort Test (WCST). Outcome comparisons were conducted using either chi-square test or t test, and correlates of baseline measures were examined using cox regression. Model building used Cox regression and found the minimum of AIC. The discriminatory ability of prediction models was determined using ROC curve. Results: At 2-years follow-up, 21 UHR subjects converted to full-blown psychosis. In neuropsychological functioning, WCST Categories Achieved and Nonperseverative Errors (both Ps < 0.05) were associated with psychosis conversion. In interview schedules and self-administered questionnaires, YSR Withdrawn (P < 0.05) and SPQ Excessive Social Anxiety, Interpersonal Deficits (both Ps < 0.05) and SOPS Positive Subscales score (P < 0.05) were associated with psychosis conversion. The predictors for neuropsychological model and questionnaire model, respectively, were selected by stepwise regression. The estimated AUC for the model combined neuropsychological variables and questionnaire variables had the best AUC. Conclusion: Neuropsychological functioning and schizotypal personality questionnaires were practically association with conversion. The prediction model combining neuropsychological assessment and psychosocial assessment identified an increased risk of psychosis with appropriate prognostic accuracy in our sample. This prediction model could be future replicated in independent samples