Structure and content of radiology reports, a quantitative and qualitative study in eight medical centers

https://doi.org/10.1016/j.ejrad.2008.06.023Get rights and content

Abstract

Rationale and objectives

In order to determine how the quality of the radiology report can be improved, the variation in content, form and length of the report needs to be examined. The purpose of the present study was to investigate and compare the length and structure of radiology reports (reports of abdominal CT examinations) in eight hospitals in the Netherlands and Flanders, the Dutch speaking part of Belgium.

Materials and methods

Eight hundred reports – 100 from each collaborating center – were collected. After exclusion of the reports that did not fulfill the criteria for a standard CT examination of the abdomen, 525 were retained for further study. The structure of each report was analyzed and the constituting parts (technical information, description of the findings, conclusion, advice, remark, note) were determined. The character and word count of the report and its constituting parts were measured and the results were grouped according to the country (Netherlands vs. Belgium/Flanders), the qualification of the radiologist (staff members vs. residents.) and the nature of the hospital (university medicals centers vs. community hospitals).

Results

Statistically significant differences were found between the Netherlands and Flanders, and between staff members and residents. The authors also found a wide variety in the lay-out and the overall content of the radiology report.

Conclusion

While most of the differences we found are probably rooted in local tradition, a standard model for the radiology report in centers sharing the same language may be a means to avoid interpretation mistakes by referring physicians, and to promote better care.

Introduction

Communication is the goal of radiological interpretation and reporting [1]. The radiology report – either in print or in digital form – is the most important communication tool by which radiologists transmit the result of their observations to referring physicians. For many of today's physicians, the radiology report has become the final product of the radiology department.

The structure and content of the radiology report has been a source of controversy for many years. The American College of Radiology has devised guidelines for the communication of diagnostic imaging findings, the latest edition having become effective on 1 October 2005 [2]. Although these guidelines have great value, they leave open for debate a large number of issues. Examples are: how long should a radiology report be? Should normal findings be mentioned, or even described extensively? Is it advisable to use standard reports or standard blocks of text to speed up reporting when findings are normal? Should the impression part of the report repeat all abnormal findings of the descriptive part or only those pertaining to the clinical requisitions?

While residents get extensive training in the execution of radiological examinations and the interpretation of their results, they hardly ever receive any training in authoring a good radiology report. In many training centers, residents are supposed to pick up the ‘noble art of reporting’ by imitating their trainers and peers. In a study by Steele et al. this ‘apprenticeship model’ did not give residents the confidence or competence a more structured program would provide [3]. Moreover, an apprenticeship model supposes that staff members are able to teach their apprentices to report, which implies that their reporting skills are better than those of the residents. Although this may seem self-evident, it has never been proven. In fact, Robert et al., using a grading system to evaluate the quality of reports of chest radiographs of patients admitted to an intensive care unit, found no significant quality difference between reports made by first-year residents and staff radiologists [4].

In a letter to the American Journal of Roentgenology in 1983, Revak presented his list of criteria to which a good radiology report should comply. Among other suggestions, he advises radiologists to practice brevity (avoid excessively long reports), clarity (avoid long sentences with a complicated internal structure) and clinical relevance (show that you have read the requisition) [5]. If staff members are indeed better at reporting than residents, they should be able to practice brevity and clarity; by consequence, their reports should be shorter. As to structure and form of the report, in a survey among radiologists and referring physicians, Naik et al. found that on the average 90% of the clinicians and 60% of the radiologists preferred a structured, itemized report to a report in prose [6].

In 2004, we started a research program to investigate the length, structure and content of radiology reports. In a small scale pilot study, the authors reviewed 94 radiology reports in the same way a college teacher would correct essays. Each report was given a score from 0 to 10 for each the following five items: clarity, clinical relevance, correct use of the language (i.e. Dutch), conciseness (brevity) and directness (precision). Using this scoring system, the authors calculated an average total score of 64.8/100 for all reports combined. The score of the best report in their sample was 76/100, while the score of the worst was 50/100. The authors concluded that the quality of the radiology reports in their sample was acceptable but not good [7]. Particularly worrisome, but in line with the results of Robert et al. [4] was the fact that reports made by staff members were found to be no better than residents’ reports.

Encouraged by the results of this small scale study, we decided to start a series of qualitative and quantitative investigations in academic and community medical centers in the Netherlands and Flanders, the Dutch speaking part of Belgium. We investigated the length of reports of abdominal CT examinations and the structure of these reports.

In the first part of this study, research questions were

  • Is there a difference in length between abdominal CT reports in academic hospitals and reports in community medical centers?

  • Is there a difference in length between abdominal CT reports from staff members and from residents?

  • Is there a difference in length between abdominal CT reports in two neighboring countries sharing one language?

In the second part of this study, we investigated, in the event of finding a difference in length, the reasons for this difference.

We obtained the collaboration of radiology departments in eight medical centers, four of which were situated in the Netherlands and four in Flanders. Four centers were part of an academic hospital, four of a community hospital. In the remainder of this article, we will refer to these hospitals as NLA1, FLC2, etc., the abbreviation NL pertaining to the Netherlands, FL to Flanders, A to academic hospitals and C to community medical centers.

The study was approved locally. In each center, 100 consecutive reports of abdominal CT examinations were collected. The electronic reports were provided in either Microsoft Word or plain text format. All reports were converted to Word 2003 format. Besides all patient specific information, all administrative and technical data were removed, except for a code allowing identification of the reporting radiologist(s).

In order to obtain a set of comparable reports, all reports of ‘non-standard’ CT examinations of the abdomen were excluded. Examples of these are: combined CT thorax/CT abdomen examinations, CT examinations limited to one region or one organ system (CT of the liver, the kidneys, the aorta, the upper abdomen, the pelvis, etc.), CT procedures for interventional purposes (puncture).

The ACR Practice Guideline for Communication of Diagnostic Imaging Findings (2005) sums up the components of a report. These include demographics, relevant clinical information, the body of the report (procedure and materials, findings, potential limitations, clinical issues, comparison studies and reports) and an impression or conclusion [2].

In our reports, we identified a slightly modified list of parts we examined and measured separately. We emphasize that these parts were not necessarily explicitly stated as such by the author of the report, but were interpreted by us according to their implicit function in the report. In a few cases the whole content of the report was presented under the heading ‘conclusion’; in these cases we chose to ignore this heading.

The parts of the CT report we identified for examination were

  • a technical introduction explaining which examination was performed and under which conditions;

  • a descriptive section, representing in most cases the main body of the report;

  • a conclusion (diagnosis or impression);

  • an advice or recommendation for diagnostic or therapeutic follow-up;

  • a practical remark (e.g. “Please provide requirements with next examination request”, “Results have been discussed with you by phone”, etc.);

  • a note or addendum, pertaining to what was seen but not mentioned in any of the other parts of the report.

The last section was included because in FLA2 such an addendum was found in 41/79 CT reports. We did not find a note or addendum in any of the reports from the other centers.

We did not include the requisition, i.e. clinical information on the patient. We had two reasons for doing so. First, it was impossible to obtain this information from the reports of the one hospital that provided us with print-outs of the reports, the requisitions being hand-written and most often illegible. Instrumental in our decision however was the fact that the clinical information on the examination request is provided by the referring physician, and is generally literally repeated by the radiologist. Therefore, strictly speaking, it does not belong to the radiologist's lexicon but to that of the referring physician.

The number of characters and of words of the reports was determined with the word count function of Microsoft Word 2003 (not including spaces). We systematically measured both the number of characters and the number of words in the report, and calculated the proportion of the number of characters to the number of words in order to obtain mean word length. The results were exported to an Excel 2003 spreadsheet. Mean values were calculated for each hospital, per country, per type of hospital and per physician.

The distribution of the number of characters and words was tested for normality using the Kolmogorov–Smirnov test and we considered logarithmic or square root transformations in order to obtain normality where necessary. Groups were compared by independent t-test when normally distributed or by Mann–Whitney U-test if otherwise. Multiple linear regression analysis was performed in order to assess the independent contribution of each of the covariates (Netherlands vs. Flanders, university vs. community hospitals, staff members vs. residents). For the comparison of categorical data (frequencies) a chi-square test statistic was used.

Section snippets

Results

We received a total of 780 abdominal CT reports. After exclusion of ‘non-standard’ abdominal CT examinations, 255 were rejected and we retained 525 for further study. Results according to the covariates (type of hospital, country, and type of radiologist) are presented in Table 1

The longest report in our sample, a report from FLA2, had a character count of 2450 and a word count of 366. That is roughly equal to 40% of a page in a tightly printed scientific magazine. It was a report of a CT

Discussion

We investigated and compared the length and structure of consecutive reports of abdominal CT examinations in eight hospitals in the Netherlands and Flanders. The use of consecutive reports can be criticized. Indeed, in an ideal situation, identical abdominal CT investigations could have been read and reported independently by radiologists from the eight collaborating centers, so their reports would match perfectly. However, this would have been a costly and time-consuming experiment, and all

Acknowledgements

The authors wish to thank Johannes L. Bloem, M.D., Ph.D., Paul J. Bode, M.D., Hendrik R.M. Degryse, M.D., Roelof E. Fokkema, M.D., Ph.D., Peter A. Kint, M.D., Ph.D., Gabriel P. Krestin, M.D., Ph.D., Guy J. Marchal, M.D., Ph.D., Herman Pieterman, M.D. and Marijke Roosendaal, M.D.

References (7)

  • P.J. Friedman

    Radiologic reporting: structure

    Am J Roentgenol

    (1983)
  • Anon. ACR practice guideline for communication of diagnostic imaging findings (URL:...
  • J.L. Steele et al.

    Learning to report

    Acad Radiol

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

Cited by (45)

  • Dear Medical Students - Artificial Intelligence is Not Taking Away a Radiologist's Job

    2023, Current Problems in Diagnostic Radiology
    Citation Excerpt :

    However, studies have shown that using non-structured reports can be an obstacle to optimal patient care.22,23 Particularly, excessive variability in language, length, and style can diminish report clarity, making it difficult for referring providers to identify key information necessary for patient care.22-24 It has been suggested that structured reporting could be used to raise the caliber of radiology reports.

  • Variation in the length and structure of reports written by reporting radiographers: A retrospective study

    2018, Radiography
    Citation Excerpt :

    It would be beneficial to see if these referrers also consider both short and long reports to be clear in the context of the wide variation in report length and structure seen here, as UK A&E departments typically have a skills mix which include non-medical referrers. For CT reporting of the abdomen,5 no statistically significant differences were found in mean word length when comparing university versus community hospitals, residents versus staff members, or by country of report origin. The study presented here did demonstrate statistically significant differences in the mean word length in all but two reporting radiographer comparisons.

  • Structured Reporting in Radiology

    2018, Academic Radiology
View all citing articles on Scopus
View full text