Original article
Use of Computer Databases to Reduce Radiograph Reading Errors

https://doi.org/10.1016/j.jacr.2006.08.016Get rights and content

Errors in reading radiologic images create health risks to patients and can adversely affect physicians through litigation and increased insurance costs. This study assessed the usefulness of a radiology teaching database for training physicians to improve their diagnostic accuracy. This tool is particularly important in pediatric radiology because children have unique anatomies and pathologies, and most physicians are generally more familiar with adults. The case collection consisted of normal cases and cases that had changes in the initial interpretations from an existing database of pediatric radiologic images at the Cincinnati Children’s Hospital Medical Center. Physicians in the radiology department were surveyed regarding their opinions on the case collections. The most frequent users of the collection were residents and fellows, who used the collection primarily for learning and reference. Attending radiologists used the collection for teaching and daily film reading. The most commonly used cases involved head computed tomography, brain magnetic resonance, and skull radiography. All respondents recognized the value of the database for the enhancement of residents’ and fellows’ training. Residents and fellows expressed the desire to add complete abnormal and frequently missed case collections to the normal collection. Physicians who used the collections did so very frequently, but many experienced physicians never accessed the collection. Inexperienced physicians’ reliance on reference books compared with Internet resources highlights a need for computer-based teaching and reference tools. This study illustrates that easy-to-use computer-based teaching files are useful for training physicians and as reference tools for experienced physicians.

Introduction

Errors in medical diagnosis and treatment are a serious problem that affects the quality of health care, with potentially significant impacts on patients’ well-being. Reducing errors should improve patient care and may reduce costs.

Espinosa and Nolan [1] reported that the rates of disagreement between emergency physicians and radiologists in the interpretation of radiographs range from 8% to 11%. Errors in interpreting radiographs can also have significant clinical and legal consequences. There are many reasons to place additional emphasis on training and on improving diagnostic quality in pediatric-related medical cases. Constant changes in anatomy and physiology during growth in children complicate the interpretation of pediatric radiologic images for both trainees and experienced pediatric radiologists. It is particularly important that university hospitals that educate trainees in different disciplines look for ways to detect and reduce commonly made medical errors.

Halsted et al [2] analyzed a total of 23,273 dictations of emergency radiographs for corrections after staff interpretation and for the types and incidence of missed abnormalities by radiology trainees. They determined that trainees were 9 times more likely to miss pathologic conditions than to wrongly identify one that did not exist and that most errors (69%) involved the diagnosis of fractures or dislocations.

To facilitate the training process and to be able to update the teaching database with new cases, many computer-based methods of collecting reference cases, including digital teaching files, have been developed [3, 4, 5, 6]. One such method developed by Weinberger et al [6] called MyPACS.net allows physicians to use a web-based service to create digital teaching files containing images and descriptive information that can then be accessed by authorized users. It is important to note that this teaching file is distinct from teaching files created on clinical picture archiving and communication systems (PACS) in that this teaching file is independent from a PACS. Halsted et al [7] developed a teaching tool based on PACS for use by radiology trainees in interpreting plain films from a pediatric emergency department.

The present research supplements the database developed by Halsted et al [7] and leverages the digital reference system they developed [3, 8] to investigate the most effective ways to make available a digital database for routine use by trainees and staff members, with a goal of improving training resources and reducing diagnostic errors in daily clinical use.

Section snippets

Materials and methods

The database used for this research included pediatric radiologic images that have been clinically reviewed by attending physicians, fellows, and residents at the Cincinnati Children’s Hospital Medical Center. Representative digital cases were selected from January 2000 through December 2004. Only images of 2 preselected categories were collected. One set of data, the reference set, included examples of normal radiologic pediatric images grouped by age. These were limited to the following

Results

The survey determined that most participants learned about the collection either through e-mail (48%) or through word of mouth from their colleagues (39%). A small percentage (13%) of users learned about the case collection through a formal user orientation.

The case collection was used over 5 months to varying degrees by the 3 different groups of respondents. The attending radiologists, as a group, accessed the collection the most, with 92 total accesses, followed by fellows with 36 uses and

Discussion

The normal case collection proved to be useful for all 3 groups of users: residents, fellows, and attending radiologists. Over the trial period, the collection was accessed more than 140 times. Most use involved imaging of the brain, including computed tomographic and MR images that were used for cases such as intracranial hemorrhage, cerebral edema, and abnormal myelination. Several factors could explain the particular popularity of the normal brain imaging cases: brain pathologies are

References (9)

  • M.J. Halsted et al.

    What is normal?A clinically useful reference collection of pediatric radiology cases created within PACS

    J Am Coll Radiol

    (2005)
  • J.A. Espinosa et al.

    Reducing errors made by emergency physicians in interpreting radiographs: longitudinal study—statistical data included

    Br Med J

    (2000)
  • M.J. Halsted et al.

    Diagnostic errors by radiology residents in interpreting pediatric radiographs in an emergency setting

    Pediatr Radiol

    (2004)
  • M.J. Halsted et al.

    A simple method of capturing PACS and other radiographic images for digital teaching files or other image repositories

    Am J Roentgenol

    (2002)
There are more references available in the full text version of this article.

Cited by (13)

  • Student-created independent learning modules: An easy high-value addition to radiology clerkships

    2014, Academic Radiology
    Citation Excerpt :

    The use of ILMs in radiology education has not been thoroughly evaluated in the medical education literature. Most studies have focused on digital resources designed to improve image interpretation and their use among radiology residents, not medical students (29–31). At least one study in the English literature has reported the value of using a digital module to teach medical students radiation safety (32).

  • Multimodal ranked search over integrated repository of radiology data sources

    2019, IC3K 2019 - Proceedings of the 11th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management
  • Big data integration case study for radiology data sources

    2018, 2018 IEEE Life Sciences Conference, LSC 2018
View all citing articles on Scopus
View full text