Tsimiklis James, Howson Sarah, Kovoor Joshua, Tan Sheryn, Stretton Brandon, Gupta Aashray, Evans Shaun, Booth Andrew, Satheakeerthy Shrirajh, Lim Lauren, Stranks Jessica, Gluck Samuel, Maddison John, Gilbert Toby, Bacchi Stephen
Lyell McEwin Hospital, Adelaide, South Australia, Australia.
University of Adelaide, Adelaide, South Australia, Australia.
Intern Med J. 2025 Jan;55(1):84-88. doi: 10.1111/imj.16590. Epub 2024 Nov 28.
Electronic medical records (EMRs) provide multiple efficiencies in communication to clinicians. The ability to copy and paste text in an EMR can be useful; however, it also conveys a risk of inaccurate documentation. Studies in international settings have described such overuse of copying to result in 'note bloat', with the dilution of relevant clinical information and potential clinical detriment.
To determine the frequency of erroneous copying, characterise the component of notes in which this occurs and determine the performance of similarity metrics in the prediction of notes likely to have erroneous copying.
A cross-sectional evaluation of all ward round notes over a 48-h period for all long-stay (>48 h) medical services, except the Acute Medical Unit, at the Lyell McEwin Hospital, a 257-bed tertiary hospital in South Australia. Four similarity metrics were evaluated: longest-sequential series of unchanged characters, similarity score (Difflib SequenceMatcher), Levenshtein distance and the Jaccard index.
One hundred twenty-eight patients were included. The number of patients who had a ward round note on two consecutive days was 97 out of 128 (75.8%). Erroneous copying was found in 8.3% of ward round notes. All (eight out of eight, 100%) of these instances of erroneous copying were in the 'issues list'. A threshold of >850 unchanged sequential characters, when compared with the ward round note the preceding day, demonstrated reasonable performance in the prediction of erroneous copying.
Erroneous copying may occur in up to 8.3% of ward round notes in a variety of medical services. Automated strategies to help address this issue should be explored.
电子病历（EMR）在与临床医生的沟通中提供了多种效率提升。在电子病历中复制和粘贴文本的功能可能很有用；然而，它也带来了记录不准确的风险。国际上的研究描述了这种过度复制的情况会导致“记录臃肿”，相关临床信息被稀释，并可能对临床造成损害。
确定错误复制的频率，描述发生错误复制的记录部分，并确定相似性度量在预测可能存在错误复制的记录方面的性能。
对南澳大利亚一家拥有257张床位的三级医院莱尔·麦克尤恩医院所有长期住院（>48小时）医疗服务（急性医疗单元除外）在48小时内的所有查房记录进行横断面评估。评估了四种相似性度量：最长连续不变字符序列、相似性得分（Difflib SequenceMatcher）、莱文斯坦距离和杰卡德指数。
纳入了128名患者。在128名患者中，有97名（75.8%）连续两天有查房记录。在8.3%的查房记录中发现了错误复制。所有这些错误复制的实例（8例中的8例，即100%）都出现在“问题列表”中。与前一天的查房记录相比，当连续不变字符数>850时，在预测错误复制方面表现出合理的性能。
在各种医疗服务中，高达8.3%的查房记录可能会出现错误复制。应探索有助于解决这一问题的自动化策略。
