Article Text


Improving the care of sleep apnoea
P125 Digital Measurement Strategy is Robust in Non-Expert Hands in Measuring Clinically Relevant Effusions on Chest Radiograph
  1. M Nagendran,
  2. RJ Hallifax,
  3. M Maruthappu,
  4. A Manuel,
  5. NM Rahman
  1. Oxford Centre for Respiratory Medicine, Oxford, UK


Introduction The primary outcome of most studies investigating pleural infection treatment is the pleural effusion volume change between two defined time points. This may be performed using thoracic computed tomography (CT) scans, but this method is costly and involves increased radiation to the patient. Data from the MIST-2 trial demonstrates that a digital chest radiograph measurement strategy was effective and correlated well with CT scanning.

Aim To assess deviation of novice scorers from an expert scorer when measuring pleural effusion size on chest radiographs using the digital scoring system.

Methods 132 chest radiograph digital images (JPEG format) were obtained from the 3 month follow-up of the MIST-2 trial. The optimal method was observer-estimated position of diaphragm and mediastinum with mediastinum discounted from the measurement. For each image, the area occupied by the pleural collection and hemithorax was manually drawn as a polyhedron, permitting calculation of percentage area occupied by effusion. Medical students acted as novice scorers and a respiratory specialist was the expert scorer and both independently scored the images.

Results Clinically significant effusions were defined as those over 2% of the hemithorax area. 73 of 132 were rated by the expert as not clinically significant. Of the remaining 59, there was >2% difference between novice and expert scores in 38 cases. Novice and expert then reassessed these radiographs together to agree a consensus. Within the subgroup of 59 clinically significant effusions, the mean difference between the novice and expert scores was –0.10 (95% CI –3.2 to 3.0, p=0.95). Comparison of novice and expert respectively to the final consensus scores revealed a significant mean difference in the novice scores of 2.0 (95% CI 0.02 to 4.0, p=0.048, expert: –0.11, 95% CI –1.85 to 1.63, p=0.90).

Conclusions Novice scorers have a tendency to overestimate effusion size. However, where there is a clinically significant (>2% of hemithorax) effusion, scores between novice and expert correlate well. This implies that the digital measurement strategy is robust in non-expert hands in measuring clinically relevant effusions. This digital chest radiograph assessment tool is therefore a potential simple outcome for use in future studies assessing pleural drainage.

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.