本觀察性研究利用廣義估計方程式分析方法，對台灣三軍總醫院失智症合併精神行為症狀患者臨床資料進行分析與說明。該資料包括病患個案的特徵、症狀描述，及接受治療的結果。每位個案的總觀察時間為12週。研究人員評估個案的臨床整體評估分數、神經精神量表與簡式智能測驗，並記錄結果在個案報告表。在所有資料收集完成後，提供統計數據之總結和相應的數據信息，且將結果以描述性方式呈現。其後，進行臨床整體評估分數之相關性分析。推論統計的進行採用廣義估計方程式來估算重複測量和藥物使用與否對神經精神量表評估總分及其照護者困擾總分，與簡式智能測驗分數的影響。 經由分析結果得到，只有使用情緒穩定劑與治療時間在個案神經精神量表評估總分與照護者困擾總分的交互作用上達顯著差異(p=0.022，p=0.012)。在有使用情緒穩定劑的狀況下，第4週神經精神量表評估總分的平均分數與第12週的平均分數分別比基準期減少13.90與14.26且達顯著差異(p<0.001)；第4週神經精神量表照護者困擾總分的平均分數與第12週的平均分數分別比基準期減少7.66與7.88且達顯著差異(p<0.001)。 In this observational study, we use generalized estimating equations model to analyze the behavioral and psychological symptoms of dementia clinical data of the Tri-Service General Hospital in Taiwan. The data includes patient’s characteristics, symptoms descriptions, and treatment results. The total observational duration of each subject is 12 weeks. Investigators assess each patient and record their clinical global impression scores, neuropsychiatric inventory and mini-mental state examination scores in case report forms. Summary statistics and appropriate data listings are provided. The correlation analysis of clinical global impression score is presented. The statistical inference for neuropsychiatric inventory with their caregiver distress scores, and mini-mental state examination results are conducted using generalized estimating equations to study the repeated measures data and explore the impact of drug use. It is found that for neuropsychiatric inventory score and caregiver distress score the interactions of treatment effect of mood stabilizers and duration of treatment are statistically significant (p = 0.022, p = 0.012). For patients treated with mood stabilizers, the average neuropsychiatric inventory scores at Week 4 and Week 12 are 13.90 and 14.26 less than that in the baseline, and the differences are statistically significant (p <0.001); the average neuropsychiatric inventory caregiver distress scores at Week 4 and at Week 12 decrease 7.66 and 7.88 from the baseline and the differences are significant (p <0.001).
