摘要
目的 分析健康体检表主检报告中的主检医师结论术语表达与单项检查结论的一致性.方法 对来自8家体检机构的11953份电子化体检表自然语言描述的血脂4项检查结论和主检医师结论,依据《中国成人血脂异常防治指南(2016年修订版)》的血脂异常临床分型、体检表血脂4项检查结论实际使用的血脂异常名词术语,构建了包含39个词条的"血脂异常"的词表库.对词表库的词条进行词频分析,计算血脂4项检查结论阳性率、主检医师结论阳性率,对主检医师结论与单项检查结论之间进行一致性检验.结果 词表库中39个词条,血脂4项检查结论不规范词条有18个,占总词条数的46%;词频分析能准确对应血脂异常临床分型的词条仅有1%;血脂4项检查结论血脂异常阳性率为47%,主检医师结论的阳性率为35%,主检医师结论与血脂4项检查结论之间的一致性检验,Kappa值=0.71,P〈0.01.结论 主检医师撰写的医师结论虽然没有严格使用标准术语,但绝大多数同义词条表达规范,便于建立"血脂异常"词表库进行文本挖掘分析.主检报告中主检医师结论与各单项检查结论是否一致,可作为主检报告质量的一个评价指标,如Kappa值〉0.75.
Objective To analyze the consistency of individual examination diagnosis with terminologies expressed in the conclusion report of physical examinations by the chief inspection physician. Methods Based on the clinical classifications used in the Guidelines on Prevention and Treatment of Blood Lipid Abnormality for Chinese Adults, and related terminology descriptions of Dyslipidemias that were actually used in four-item blood lipid examinations diagnosis, a lexicon database of Hyperlipidemias was constructed with 39 terms used in the four-item blood lipid examination diagnosis and chief physician conclusions. Totally 11953 electronic chief physician inspection reports from 8 health check-up institutions were included. We investigated the terms of lexicon database using word frequency analysis method, calculated the positive rate in the diagnosis of four single examinations of serum lipid and the positive rate in the chief physician's conclusion. Consistency of chief physician's conclusion with single examination diagnosis was analyzed by Kappa test. Results (1) Among the 39 terms of lexicon database, there are 18 nonstandard terms used in single examination diagnosis, accounted for 46% of all terms; (2) In word frequency analysis, there are only 1% of terms that corresponded to clinical classifications of Hyperlipidemias accurately;(3) The positive rate of Hyperlipidemias in serum lipid four single examinations diagnosis was 47%, the positive rate of physician diagnosis was 35%. The consistency analysis of chief physician conclusions with single examination diagnosis showed Kappa=0.71(P〈0.01). Conclusions Although the final conclusion written by the chief physician was not using standard terms strictly, most of synonymous terminology expressed in the physician conclusion is normative, which enabled the construction of lexicon database and text mining. Whether the single examination diagnosis are consistent with physician conclusions can be an evaluation indicator to assess the quality of chief physician conclusions for physical examinations. Kappa〉0.75 may be suggested as a favorable value.
出处
《中华健康管理学杂志》
CAS
2017年第4期344-348,共5页
Chinese Journal of Health Management
基金
国家自然科学基金资助项目(81573250)