Kapil Ojha MBBS, Presenter: Nothing to Disclose

Angela McKinnie, Abstract Co-Author: Nothing to Disclose

Richard D. White MBChB, FRCR, Abstract Co-Author: Nothing to Disclose

Ian Alexander Zealley MD, Abstract Co-Author: Nothing to Disclose

In order to be clinically useful it is essential that the content of radiology reports can be readily assimilated by the clinician. Journalists use automated text analysis tools, for instance to ensure that a pre-determined “reading age” for a particular publication is met. In order to determine their potential utility in radiology practice we set out to assess the relationship between automated text analysis findings and subjective ease of readability of radiology reports as determined by clinicians.

A test set of 100 lung cancer staging CT scan reports was generated, compromising 10 randomly selected reports issued by 10 randomly selected radiologists from our centre. The text of each report was analysed using four automated tools: Flesch-Kincaid Score (FKS), Flesch Reading Ease Score (FRES), Simple Measure Of Gobbledygook score (SMOG) and Word Count (WC). Each report was also rated for readability by 3 chest physicians using a 100 mm visual-analogue scale (rating range 0-100) with the mean of these 3 scores providing the reference standard.

The correlation coefficients for the automated scores and the clinician readability ratings were -0.39 FKS, 0.36 FRES, -0.24 SMOG, and -0.76 WC. The mean report WC for the 10 reports issued by each of the 10 radiologists ranged from 96-360 words. The mean WC exhibited a strong negative correlation with the clinician readability rating.

The strong negative correlation between Word Count and clinician readability rating indicates that short reports are a more effective communication tool than long reports. More complex text analysis tools correlated poorly with readability.

An effective radiology report must strike a balance between comprehensiveness and readability: where possible, short is sweet.

Ojha, K, McKinnie, A, White, R, Zealley, I, Short is Sweet: Automated Text Analysis of Radiology Reports.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12025456.html