基于德尔菲法构建药物重整过程中药物差异分类工具

目的：构建药物重整过程中药物差异分类工具（Medication Discrepancy Taxonomy，MedTax），推动药物重整实施标准化。方法：翻译MedTax，运用德尔菲专家咨询法从分类的代表性、唯一性、分类名称的明确性和分类定义的清晰性四个方面对MedTax的每个二、三级分类指标进行评分。在分析专家协调系数、权威系数、内容效度和专家意见集中度的基础上，构建中文版药物差异分类工具，使用虚拟案例应用多自由边际Kappa检验可靠性。结果：经过11名专家两轮修改，构建了包含药物不匹配和药物部分匹配2个一级指标，遗漏药物、添加药物、药物重复等11个二级指标和23个三级指标的中文版药物差异分类工具。两轮专家咨询的专家权威系数均为0.85，专家协调系数分别为0.51和0.61，经χ²检验，P<0.01。第二轮专家咨询各指标均数为4.0~4.45分，变异系数为0.12~0.27，内容有效性指数为0.82~1.0，经可靠性测试，总体一致性评分为0.76。结论：该研究专家权威性可靠，专家意见集中度高，协调性好，结果可信。构建的中文版药物差异分类工具内部一致性高，可用于药物重整过程中药物差异的分类和测量。

Using the Delphi method to construct a classification taxonomy for medication discrepancies identified through medication reconciliation

OBJECTIVE To construct a medication discrepancy taxonomy (MedTax) for medication reconciliation to promote the standardization of medication reconciliation.METHODS The MedTax was translated. Two rounds of Delphi method were used to score each classification index at the second level and third level of the medication discrepancies taxonomy tool from four aspects of representativeness, uniqueness, clarity of classification name, and clarity of classification definition. Based on the analysis of expert coordination coefficient, authority coefficient, content validity and concentration of expert opinions, the Chinese medication discrepancies taxonomy (C-MedTax) was constructed. Virtual cases were used to test the reliability of the C-MedTax with multi-free marginal Kappa.RESULTS After two rounds of revisions by 11 experts, the C-MedTax was constructed including two first-level indicators of drug mismatched and drug partial matched, 11 second-level indicators such as omission drugs, commission drugs, and drug duplication, and 23 third-level indicators. The expert authority coefficients of the two rounds of consultation were both 0.85, and the expert coordination coefficients were 0.51 and 0.61, respectively. Chi-square test showed P<0.01. In the second round of consultation, the mean of each index was 4.0-4.45, the coefficient of variation was 0.12-0.27, and the content validity index was 0.82-1.0. After reliability testing, the overall consistency score was 0.76, and the scorers were compared in pairs, and 85% of the consistency scores were above 0.7.CONCLUSION The experts in this study were authoritative and reliable, expert opinions were highly concentrated and coordinated, and the results were credible. The constructed medication discrepancies taxonomy has high internal consistency and can be used to measure medication discrepancies in the process of medication reconciliation.