文献速递免疫规划卫生政策与健康促进

ICD与SNOMED搜索策略在COVID-19疫苗接种后不良事件中的准确性:医院行政数据分析

Accuracy of ICD and SNOMED search strategies for adverse events following COVID-19 vaccination: Analysis of hospital administrative data.
PubMed 文献 · Vaccine · 2026-01-31

专家小组识别了24个ICD-10-AM码和2个SNOMED码,随机抽取761份病历(主要2016-2019年入院)进行审查。结果显示,仅吉兰-巴雷综合征有单一特异码对应单一AESI,各码PPV范围为11%至100%,许多码准确性不足,需逐案评估。

要点

  • 仅吉兰-巴雷综合征有单一编码对应单一AESI,多数神经系统疾病编码重叠。
  • 各编码满足病例定义的PPV范围为11%至100%,需逐案评估。
  • 许多编码准确性不足,不宜单独用于识别目标AESI。

核心结论

  • 每个AESI编码的PPV需逐案评估,许多编码准确性不足以识别目标AESI。

主要发现

  • 仅吉兰-巴雷综合征有单一编码特异对应单一AESI,多种神经系统疾病存在编码、疾病和病例定义重叠。
  • 各编码满足病例定义的PPV范围为11%至100%,具体取决于编码和确信度水平。

关键数值与推荐

  • 专家小组识别了24个ICD-10-AM码和2个SNOMED码,预期识别9种AESI,优先选择特异性而非敏感性编码。
  • 编码的PPV范围为11%至100%,各编码的确信度和PPV有详细说明。

临床怎么用

  • 使用医院编码识别COVID-19疫苗接种后AESI时,应了解各编码的PPV和确信度,避免仅依赖编码进行判定。

局限性

  • 未评估编码组合的PPV,方法无法估计编码漏掉的病例数。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Hospital discharge codes can be used to identify possible Adverse Events of Special Interest (AESI) following COVID-19 vaccinations. AIM: We sought to estimate the positive predictive value (PPV) and level of certainty of ICD-10-AM and SNOMED coding for meeting Brighton Collaboration case definitions of AESIs in Aotearoa, New Zealand. METHODS: Our expert panel identified 24 ICD-10-AM codes and 2 SNOMED codes expected to identify 9 AESIs. We sought codes likely to be specific rather than sensitive. Medical record reviews were conducted by an experienced coder, adjudicated by medical specialists, for the level of certainty that each case met the target case definition. Admissions coded to an explicit alternative diagnosis were classified as Not a Case. RESULTS: Our data covered over 3 million people. Reviews were conducted on 761 medical records, randomly selected from admissions principally in calendar years 2016 to 2019. We report the PPV of each code with respect to its level of certainty of meeting the case definition. Only Guillain-Barré had a single code specific to a single AESI. Several neurological conditions had overlapping codes, conditions and case definitions. PPVs for individual codes to meet case definitions ranged from 11% to 100%. Level of certainty of each code and PPV are specified. CONCLUSION: PPVs of codes for each AESI need to be assessed on a case-by-case basis. Many codes are arguably insufficiently 'accurate' to identify a target AESI. We did not assess PPVs for combinations of codes. Our method did not allow us to estimate the number of cases missed by the coding.

来源:Vaccine,查看原文。版权归原作者所有。

引用

  1. Kenealy T, Aguirre-Duarte N, Roxburgh RH, et al. Accuracy of ICD and SNOMED search strategies for adverse events following COVID-19 vaccination: Analysis of hospital administrative data. Vaccine. 2026. PMID: 41621119. doi:10.1016/j.vaccine.2026.128275

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。