| 摘要: | 目的：
利用現代測量理論來評估世界衛生組織生活品質問卷（World Health Organization Quality of Life-Bref, WHOQOL-BREF）與聖喬治呼吸問卷（St. George''s Respiratory Questionnaire, SGRQ）兩份問卷應用在慢性阻塞性肺部疾病（chronic obstructive pulmonary disease, COPD）患者是否具有良好的心理計量特質，並評估不同疾病嚴重程度分期患者適用題目，作為發展短式問卷或電腦適性測驗之基礎。
材料與方法：
本研究共計收案240名男性COPD患者，利用項目反應理論（item response theory, IRT）中的Rasch模式來分析各題目的特質。分析項目包括：以適合度指標（infit statistic）來檢驗各範疇單一向度的假設，並檢驗題目難度、問卷的信度、題目門檻值範圍以及題目適中性。再以試題差別功能（differential item functioning, DIF）來檢驗各題目在不同疾病嚴重度族群與不同年齡族群之心理計量特質是否相同。最後針對題目具有難度順序（hierarchy）特質的SGRQ日常活動範疇，計算各題之題目訊息量（item information），挑出各疾病分期患者較適用的題目，用以探討縮減題目或發展適性測驗的可行性。
結果：
兩份問卷各範疇題目大多符合單一向度的假設。在題目難度分佈上，SGRQ各題目平均難度範圍較WHOQOL-BREF要廣。在信度的評估方面，WHOQOL-BREF各範疇的信度指標介於0.74 ~ 0.80，而SGRQ除症狀範疇信度較低（0.59）外，日常活動與疾病衝擊範疇皆在0.8以上。題目門檻值分析結果顯示WHOQOL-BREF題目門檻值所能涵蓋的患者能力範圍較SGRQ要廣。在題目適中性分析上，WHOQOL-BREF生理與心理範疇題目難度適中，但社會與環境範疇的題目則較為簡單；SGRQ症狀與日常活動範疇題目難度適中，但疾病衝擊範疇則較簡單。兩份問卷在疾病嚴重程度與年齡的DIF上，WHOQOL-BREF題目很少出現試題差別，但是在SGRQ的日常活動與疾病衝擊範疇則有許多的DIF存在。針對SGRQ的日常活動範疇分疾病嚴重度的挑題結果，狀況較佳的族群，則挑出較具挑戰性的題目，隨著疾病愈嚴重，所挑出的題目愈趨簡單。尤其是針對疾病較嚴重的族群，不論是在「某些活動使您喘不過氣來」題組或「呼吸問題如何影響您的活動」題組，挑選出來的少數題目即能夠達到與原有題目相當的準確性。
結論：
從分析結果可以看出兩份問卷用於COPD患者皆呈現不同的心理計量特質。由於在不同疾病嚴重度與年齡的比較上，SGRQ問卷有許多題目產生DIF，因此在比較不同疾病嚴重程度與年齡族群時，要注意結果的詮釋與其所代表的臨床意義。訊息分析顯示SGRQ的日常活動範疇具有合理的結構性，不同疾病分期可選擇不同難度的題目來回答，尤其對疾病較嚴重的COPD患者，可以不需要回答冗長的問卷。本研究結果可以作為縮短問卷或發展適性測驗的基礎。
Aims:
The aim of this study was to apply the modern measurement theory to evaluate the psychometric characteristics of the World Health Organization Quality of Life-Bref (WHOQOL-BREF) and the St. George’s Respiratory Questionnaire (SGRQ) in chronic obstructive pulmonary disease (COPD) patients. The items selected based on information function can provide a useful reference for item reduction and computerized adaptive testing applied in COPD.
Methods:
There were 240 male COPD patients included in this study. We analyzed the characteristics of each item by Rasch model. The infit statistics of Rasch model was used to examine the unidimensionality of each domain, and also evaluated the item difficulty, the reliability, the range of item threshold and item targeting of the two questionnaires. In addition, the differential item functioning (DIF) was used to examine the item psychometric properties for different groups stratified by age and disease severity. Finally, for the activity domain of the SGRQ, which is with hierarchy items, we employed the information function of the Rasch model to collect 3 items for each disease stages and compared how their performance of estimation compared with the original lengthy items.
Result:
Most of the items within each domain fitted the assumption of unidimensionality. In the item difficulty distribution, the range of mean difficulties of items with its assigned domain for the SGRQ was wider than for the WHOQOL-BREF. The reliability of each domain with WHOQOL-BREF is from 0.74 to 0.80. With SGRQ, the reliability is lower in symptoms domain, but the reliabilities in activity and impacts domains were above 0.8. The range of item thresholds for WHOQOL-BREF could cover wider range of the patient’s ability than that of the SGRQ. For the WHOQOL-BREF, the items of physical and psychological domain were targeting to the patients, but items of social and environmental domain were simpler for the patients. For the SGRQ, the items of symptoms and activity domain were targeting to the patients, but items of impacts domain were simpler for the patients. Among disease severity and age group, items of WHOQOL-BREF appeared almost no DIF, but lots of items of activity and impacts domain with SGRQ appeared DIF. There were comparable estimating power based on the shorten selected items with the original lenthy items, especially for the COPD patients with severe status.
Conclusion:
Our results show the WHOQOL-BREF and the SGRQ had different psychometric characteristics for COPD patients. Quite a few items presented item differenting among groups with different disease stages and age, which impled we need to pay more attention in interpreting the results or comparing the results among these groups. In addtion, more clinical information could be found from the items with severe DIF. Our result showed the feasibility of items shorting process, especially for the COPD patients with severe stages. The analysis based on the modern measurement theory can provide an useful structure in the development of computerized adaptive testing. |
