Kim Tae Youn, Hardiker Nicholas, Coenen Amy
Betty Irene Moore School Nursing, University of California Davis, 4610 X Street, Sacramento, CA 95817, USA.
School of Nursing, Midwifery & Social Work, University of Salford, Mary Seacole Building, Greater Manchester M5 4WT, UK.
J Biomed Inform. 2014 Jun;49:213-20. doi: 10.1016/j.jbi.2014.03.001. Epub 2014 Mar 13.
The purpose of this study was to determine the degree of overlap between the International Classification for Nursing Practice (ICNP®) and the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT), with a specific focus on nursing problems, as a first step towards harmonization of content between the two terminologies.
Work within this study was divided across two ICNP subsets. The first subset (n=238) was made up of ICNP diagnosis/outcome concepts that had been included in previous experimental mapping activities with Clinical Care Classification (CCC) and NANDA-International (NANDA-I). These ICNP concepts and their equivalent concepts within CCC and NANDA-I were used within the Unified Medical Language System (UMLS) framework to derive automatically candidate mappings to SNOMED-CT for validation by two reviewers. The second subset (n=565) included all other ICNP diagnosis/outcome concepts plus those concepts from the first subset where the candidate mappings were rejected. Mappings from the second subset to SNOMED-CT were manually identified independently by the same two reviewers. Differences between the reviewers were resolved through discussion. The observed agreement between the two reviewers was calculated along with the inter-rater reliability using Cohen's Kappa (κ).
For the first semi-automated mapping, according to the two reviewers the great majority of ICNP concepts (91.6%) correctly mapped to SNOMED-CT in UMLS. There was a good level of agreement between the reviewers in this part of the exercise (κ=0.7). For the second manual mapping, nearly two-thirds of ICNP concepts (61.4%) could not be mapped to any SNOMED-CT concept. There was only a moderate level of agreement between the reviewers (κ=0.45). While most of the mappings were one-to-one mappings, there were ambiguities in both terminologies which led to difficulties. The absence of mappings was due to a large extent to differences in content coverage, although lexical variations and semantic differences also played a part.
This study demonstrated a degree of overlap between ICNP and SNOMED-CT; it also identified significant differences in content coverage. The results from the semi-automated mapping were encouraging, particularly for 'older' ICNP content. The results from the manual mapping were less favorable suggesting a need for further enhancement of both terminologies, content development within SNOMED-CT and further research on mechanisms for harmonization.
本研究旨在确定国际护理实践分类法（ICNP®）与医学系统命名法-临床术语（SNOMED-CT）之间的重叠程度，特别关注护理问题，作为实现这两种术语内容协调的第一步。
本研究的工作分为两个ICNP子集。第一个子集（n = 238）由ICNP诊断/结果概念组成，这些概念已包含在先前与临床护理分类法（CCC）和国际护理诊断协会（NANDA-I）的实验性映射活动中。这些ICNP概念及其在CCC和NANDA-I中的等效概念在统一医学语言系统（UMLS）框架内用于自动生成到SNOMED-CT的候选映射，以供两名评审员验证。第二个子集（n = 565）包括所有其他ICNP诊断/结果概念以及来自第一个子集但候选映射被拒绝的那些概念。由同两名评审员独立手动确定从第二个子集到SNOMED-CT的映射。评审员之间的差异通过讨论解决。使用科恩kappa系数（κ）计算两名评审员之间的观察一致性以及评分者间信度。
对于第一次半自动映射，根据两名评审员的意见，绝大多数ICNP概念（91.6%）在UMLS中正确映射到SNOMED-CT。在这项工作的这一部分，评审员之间有良好的一致性水平（κ = 0.7）。对于第二次手动映射，近三分之二的ICNP概念（61.4%）无法映射到任何SNOMED-CT概念。评审员之间只有中等程度的一致性（κ = 0.45）。虽然大多数映射是一对一映射，但两种术语中都存在歧义，这导致了困难。映射的缺失在很大程度上归因于内容覆盖范围的差异，尽管词汇变化和语义差异也起到了一定作用。
本研究证明了ICNP与SNOMED-CT之间存在一定程度的重叠；它还确定了内容覆盖范围的显著差异。半自动映射的结果令人鼓舞，特别是对于“旧的”ICNP内容。手动映射的结果不太理想，这表明需要进一步增强这两种术语，在SNOMED-CT内进行内容开发，并进一步研究协调机制。
