[ad_1]
Uematsu H, et al. Development of a scoring system to quantify errors from semantic characteristics in incident reports. (BMJ Health & Care Informatics 2024;31(1):e100935). The authors have discovered an error in the text and one of the figures. This error is in the calculation of the group error score and may affect the interpretation and reproducibility of the model.
Description of the error: In this study, the group error score was determined by averaging the total report error scores based on the number of reports for each group, not the number of workers for each group. Therefore, there is a need to change ‘workers’ to ‘reports’ in the body text and in figure 1. The source code submitted during the initial submission clearly documented that all group error scores in this study were calculated using the number of reports from the outset. This is incorrect, although the data in the results were derived using the correct definition (ie, divided by the number of reports). As a result, no other parts of the manuscript need to be changed.
Process for calculating error scores from original text in incident reports.
Original Text: Page 3 of 8: ‘Then, the scores for a certain group (clinical or non-clinical departments/wards) were calculated by averaging the scores for individual reports in that group, adjusted by its number of workers.’
Correction: Page 3 of 8: ‘Then, the scores for a certain group (clinical or non-clinical departments/wards) were calculated by averaging the scores for individual reports in that group, adjusted by its number of reports.’
Original Text: Page 5 of 8: ‘The median group error score, which is the total report error score of a group divided by the number of workers therein, was 0.06 (IQR: −0.46 to 0.61).’
Correction: Page 5 of 8: ‘The median group error score, which is the total report error score of a group divided by the number of reports therein, was 0.06 (IQR: −0.46 to 0.61).’
[ad_2]
Source link
