DeFroda Steven F, Milner John D, Bokshan Steven L, Owens Brett D
a Department of Orthopaedic Surgery , Brown University, Warren Alpert School of Medicine , Providence , RI , USA.
Phys Sportsmed. 2018 Nov;46(4):499-502. doi: 10.1080/00913847.2018.1513755. Epub 2018 Aug 31.
The Internet is an easily accessible resource for both providers and patients. Despite this, the Internet is not peer reviewed, leaving Internet searches subject to inaccuracies, especially with regards to medical information. The purpose of this study was to review Internet images of meniscus tears using three popular search engines: Google, Bing, and Yahoo.
A search query was performed on the aforementioned search engines for the term: 'meniscus tear'. The first 100 images found for each individual search were analyzed by two independent reviewers with different levels of orthopedic training (orthopedic surgery resident and medical student). Inter-rate reliability and accuracy was determined for each of the search engines. The images were defined based on the source that published the image as either educational (published by hospital or medical association), commercial (published by a device company), or individual (published via a physician).
The inter-rater reliability was excellent (Cronbach's alpha = 0.91), (Cronbach's alpha = 0.94), (Cronbach's alpha >0.90) on Google, Bing, and Yahoo, respectively. When comparing the search engines for correctness, Google had 82% accuracy, compared to 81% for Bing and Yahoo. All three search engines have a similar mix of source material with educational images consisting of 86% of Google images, 84% for Yahoo, and 89% for Bing.
Our study revealed that the three search engines queried displayed meniscal tear images with >80% accuracy when evaluated by two independent reviewers. Despite this, many images may still be seen as highly technical, or esoteric to an untrained individual. Ultimately, physicians should take an active role in making high quality, easy to understand medical resources, and anatomic diagrams available to their patients to avoid confusion and enhance understanding.
互联网是供医疗服务提供者和患者使用的易于获取的资源。尽管如此,互联网未经同行评审,这使得互联网搜索容易出现不准确的情况,尤其是在医疗信息方面。本研究的目的是使用三个流行的搜索引擎(谷歌、必应和雅虎)来审查半月板撕裂的互联网图像。
在上述搜索引擎上对术语“半月板撕裂”进行搜索查询。由两名接受过不同程度骨科培训的独立评审员(骨科住院医师和医学生)对每次单独搜索找到的前100张图像进行分析。确定每个搜索引擎的评分者间信度和准确性。根据发布图像的来源将图像定义为教育类(由医院或医学协会发布)、商业类(由设备公司发布)或个人类(由医生发布)。
在谷歌、必应和雅虎上,评分者间信度分别为优秀(克朗巴哈系数=0.91)、(克朗巴哈系数=0.94)、(克朗巴哈系数>0.90)。在比较搜索引擎的正确性时,谷歌的准确率为82%,必应和雅虎为81%。所有三个搜索引擎的源材料组合相似,教育类图像在谷歌图像中占86%,雅虎为84%,必应为89%。
我们的研究表明,当由两名独立评审员评估时,所查询的三个搜索引擎显示半月板撕裂图像的准确率超过80%。尽管如此,但许多图像对于未经培训的人来说可能仍然显得技术含量过高或晦涩难懂。最终,医生应积极提供高质量、易于理解的医疗资源和解剖图给患者,以避免混淆并增进理解。