以羅序分析檢驗糖尿病生活品質問卷與第二型糖尿病患者生活品質追蹤應用研究

Abstract

研究目的：目的在於使用現代測驗理論羅序模式檢驗糖尿病生活品質問卷，並且探討研究三年期間第二型糖尿病患者的生活品質可能的影響因子。 材料與方法：研究追蹤對象為2006年七月北台灣某醫學中心新陳代謝專科與一般科門診的糖尿病患共205位，追蹤時間從2009年五月至2010年一月。使用病人自填中文版糖尿病生活品質量表收集疾病相關生活品質。並且記錄基本人口學資料、糖尿病相關因子等資料。分析方法以羅序模式分別驗證與估計糖尿病生活品質問卷三個面向：滿意度、衝擊、憂慮的心理計量特徵、單向度與分數改變多寡。再以複迴歸方式討論生活品質分數變化與潛在預測因子的關係。 結果：排除第一型糖尿病患者後追蹤202位患者，共有149位完整參與此研究，平均年齡61±11歲。心理計量特徵部分，糖尿病生活品質問卷三個面向分別符合單向度建構，除了滿意度面向第八題(睡眠)、第十五題(糖尿病控制)與衝擊面向第十九題(不願回答)外。這些符合單向度假設的題目反應的滿意度、衝擊、憂慮程度，除滿意度的建構量尺均勻分布涵蓋滿意程度外，衝擊與憂慮程度的建構量尺顯示潛在性的天花板效應。經過三年時間，患者的生活品質在滿意度與憂慮面向達到統計顯著差異，滿意度上升(p<.0001) 與憂慮程度提升(p=0.01)。在控制其他潛在因子後，新發生微血管併發症(神經病變、腎病變)、曾經使用過胰島素治療與憂慮程度上升有關。開始運動或飲食控制分別對滿意度面向與憂慮面向有正向的影響。 結論：糖尿病生活品質問卷三個面向分別符合單向度假設。發生糖尿病相關併發症是生活品質下降的預測因子之一；開始運動或飲食控制有助於生活品質提升；此外糖化血色素對於生活品質改變的影響則沒有發現。

Objectives: The aim of this study is to use Rasch-based approach methods to assess the change of quality of life (QoL) scores and the effects of its predictors among patients with type 2 diabetes across three years’ follow-up period in Taiwan. Methods: We followed 205 patients recruited in July 2006 from one medical center in northern Taiwan during outpatient visits in the metabolism special clinic and general clinic of the study hospital in May 2009 to Jan 2010. The Diabetes Quality of Life (DQOL) questionnaire was used to assess the disease-specific QoL. Socio-demographic characteristics, diabetes-related factors and other medical factors were also collected as possible predictors. The psychometric properties of DQOL three domains (DQOL-S (satisfaction), I (impact) and W (worry)), and change of DQOL scores were examined by Rasch methods separately. Multiple linear regression models were used to test the association between possible predictors and the change of DQOL scores in the three years. Results: Of the 202 patients followed (after exclude 3 patient with type 1 diabetes), 149 patients completed the follow-up interview (mean age of 61±11 years). In psychometric properties, all items fitted Rasch method’s expectation except the item of “sleep” and “control of diabetes” in DQOL-S and “reluctant ” in DQOL-I. After removal of the misfit item, all of the remaining items all fitted the requirement of unidimensionality. In distribution of patients’ response and item difficulty, patients covered almost the full range on the construct of DQOL-S, while potential ceiling effects in structure of DQOL-I and DQOL-W. During the 3-year study period, patients showed a statistically significant increase on satisfaction level in satisfaction (p<.0001) but also a higher level of worry (p=0.01). After adjusting for all of potential predictors, presence of microvascular complications (neuropathy and nephropathy) and ever using insulin therapy were associated with a deterioration of worry scores. Starting exercise and / or diet control had positive effect on change in satisfaction and worry scores (p=0.01). Conclusion: Three domains in DQOL represent one common construct. In this cohort of patients with type 2 diabetes, the presence of complications is predictive of a lower QoL, while starting exercise and diet control predicting positive change of QoL. Effect of change of HbA1c values was not seen in this study.