以「多屬性效用」模式探討減痛分娩的影響因素

Abstract

前言：俗稱「減痛分娩」的「硬脊膜外止痛」(Epidural analgesia, EA)是一種有效且安全的產痛緩減方法。其在大多數國家均有逐漸普遍之趨勢，唯台灣產婦的接受度仍遠較西方國家低。相關文獻報告多集中在其安全性與效果的研究，鮮少探究產婦接受減痛分娩與否的原因。為推廣減痛分娩以讓更多的產婦受益，闡明影響國內產婦接受減痛分娩決定的因素實屬必要。 目的：本研究以行為理論中的「多屬性效用」模式(Multi-attribute Utility, MAU)來進行產婦接受減痛分娩與否的決策分析，評估影響其決定的因素與相對重要性，並利用此一模式預測產婦是否會接受減痛分娩。 方法：本研究屬斷代性調查(cross-sectional survey)，收案日期自民國九十五年一月十八日起至四月八日止，對象為台北榮民總醫院行自然產與非計畫性(non-elective)剖腹產之產婦，由專家依MAU理論針對產婦的個別屬性(individual attribute)，對減痛分娩的認知與態度(knowledge & attitude)及行動線索(cue factor)，設計一多階層架構之問卷，並由專家依MAU理論加權計算出不同階層的分數差異，並藉所得的MAU總分預測決策者產前做減痛分娩的意願及生產時實際接受減痛分娩與否的行為，並評估不同切點(cut-off point)的敏感度(sensitivity)、特異度(specificity)與ROC (receiver operating characteristics)曲線下的面積以驗證模式的預測能力。 結果：本研究共納入205位產婦，其中167人為我們填寫問卷，其中151位產婦確實完成問卷。依據接受減痛分娩與否將所有產婦分成兩組。減痛分娩組(EA組)共有84人，而未接受減痛分娩組(Non-EA)則有83人。兩組的基本資料除了胎次與教育程度外並無顯著差異。EA組擁有大學以上學歷者較另一組多(85.7% vs. 66.3%, p=0.003)。此外，其第一胎的比例亦較非EA組高(76.2% vs. 47.0%, p<0.001)。以MAU模式分析結果發現，在20個問題中，有12個項目在兩組間呈現顯著差異，其中以「怕副作用」、「怕後遺症」、「怕打針」等三項在兩組間差異最大。十個次概念中，則有七個有顯著差異，其中屬「障礙」與「好處」的全部次概念均呈顯著差異，但屬「提示」的四個次概念中，只有“親友”一項有顯著差別。而第三層的三個概念「障礙」、「好處」與「提示」在兩組間均有明顯差異(所有p均<0.001)，至於第四層的最後決定在兩組間也呈現顯著差異，且兩組最後決定的MAU總分平均相差0.38 (p<0.001)。評估後發現最佳的切點值約為0.2，其對產前決定的預測敏感度與特異度分別為0.84與0.79，對最後決定的預測敏感度與特異度分別為0.84與0.72。其產前與最後決定的ROC曲線下面積分別為0.91 (0.86-0.96)與0.83 (0.76-0.89)。 結論：胎次與教育程度是影響產婦決定接受減痛分娩與否的最重要個別屬性，而認知與態度及行動線索則可利用MAU模式來評估其相對重要性，我們以MAU理論所建構的決策模式可準確預測產婦產前做減痛分娩的意願與其待產時接受減痛分娩的行為。

Background: Epidural analgesia (EA) is an effective and safe method to relieve labor pain. Although the prevalence of epidural labor analgesia has been increasing in most countries, it is still infrequently in Taiwan compared with most west countries. Most studies in relation to labor EA focused on its benefits and drawbacks. Little is known about the factors associated with decision on EA. In order to propagate the labor EA, the elucidation of decisive factors related to labor EA is of paramount importance. Aims: We applied multi-attribute utility (MAU) theory to ascertain possible factors on which we based to predict whether parturients would receive EA or not during their labor. Materials and methods: A cross-sectional survey was therefore conducted by enrolling all parturients with spontaneous delivery or non-elective cesarean section in Taipei Veterans General Hospital from Jan 18 2006 to Apr 8 2006. A hierarchical questionnaire based on MAU theory was designed by experts to include individual attributes, knowledge and attitude toward EA and cue factors. The experts also performed the weighting process and calculated the scores in different levels according to the MAU theory. The behaviors of parturients were also predicted by the application of MAU model. Sensitivity, specificity, and receiver operating characteristics (ROC) curve by different cut-off points were also evaluated to assess predictive validity of the MAU model. Result: There were 205 parturients recruited in our study. Among them, 167 parturients, consisting of 84 and 83 parturients in the EA and non-EA group, returned the questionnaire, respectively. Of 167, 151 subjects completed all questions. The only differences of individual attributes between the two groups are the parity and education level. More parturients have bachelor degree or above in the EA group compared with the non-EA group (85.7% vs. 66.3%, p=0.003). Besides, there are also more primipara in the EA group (76.2% vs. 47.0%, p<0.001). For items in MAU model, 12 of 20 items revealed significant differences between the two groups. Among these items, “Fear to side effects”, “Fear to severe complications” and “Fear to needle” had the most significant differences. Among the 10 sub-concepts, only 7 of them had significant differences. All sub-concepts under two concepts of “benefit” and “barrier” showed significant differences. In contrast, only 1 sub-concept in “cue to action” concept had significant difference. All 3rd level concepts showed significant differences between the two groups (all p<0.001) and the final decision also showed significant differences. The mean difference in final MAU score between the two groups was 0.38 (p<0.001). The best cut-off point of our model was 0.2 and the sensitivity and the specificity of pre-labor decision were 0.84 and 0.79, respectively. The sensitivity and specificity of final decision were 0.84 and 0.72. The area under ROC was 0.91 (0.86-0.96) for pre-labor decision and 0.83 (0.76-0.89) for final decision. Conclusion: The most influential individual attributes related to labor EA were parity and education level. MAU model can be used to evaluate the relative importance of items associated with knowledge and attitude toward EA and cue factors. Our MAU model can predict pre-labor decision and final decision.