Abstract
A minor data coding error in the individual patient database was identified regarding threshold growth: in a few studies, LI-RADS features requiring a prior imaging examination (threshold growth, subthreshold growth, size reduction, size stability, and US visibility) were documented as either present or absent, even if a prior imaging examination was not available. In the initial coding of threshold growth data for the evaluation of major features in the van der Pol 2022 study, these data were incorrectly coded as "not present" rather than "not evaluated" in the analysis. Based on follow-up discussions, the authors are not able to discern which observations labelled as threshold growth absent were truly absent versus "not evaluated" due to lack of prior imaging examinations. Repeat analysis of this data did not meaningfully change any of the data presented in the abstract or in the conclusion. Changes were made accordingly in Tables 3 and 4. As well, one of the P values in the Key Results section was changed to .19; in the Abstract, a CI was changed to 0.8, 3.4. In the Synthesis of Results section, a CI in the main text was changed to 0.8, 3.4; the number of observations was changed to 452; the P value was changed to .19; and an odds ratio was changed to 2.0 and its associated P value changed to .09.
Original language | English (US) |
---|---|
Article number | e239005 |
Journal | RADIOLOGY |
Volume | 307 |
Issue number | 1 |
DOIs |
|
State | Published - Apr 2023 |
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging