Guidelines for reviewers and authors
Authoring or reviewing a MIMIC paper can be challenging due to the complex nature of the data. Our recommendations for authors when writing papers, or for reviewers when reviewing papers, are:
- The version of MIMIC should be specified. For example: MIMIC-III v1.4, or MIMIC-IV v1.0. If researchers are using an older version, inquire as to why.
- It is easy to conflate an ICU admission with a hospital admission. Ensure that it is clear.
- Mortality is a common outcome, but can be defined in a number of ways. Some researchers define 30-day mortality from admission, whereas others define it from discharge. Ensure the manuscript is clear. We recommend using 30-days from admission, as this better reflects the severity of illness of the patient on admision to the ICU.
- Avoid vague criteria. For example, “removed patients missing data” is unclear, whereas “removed patients with no heart rate measurements in the first 24 hours of their stay” is much more interpretable.
- Verify that the MIMIC data citation is referenced, and for MIMIC-III that the paper is also cited. See the acknowledgement page for details.
Most of important of all, code should be included with the paper. The MIMIC data use agreement requires researchers to publish code with their paper, and there is no substitute for code in explicitly describing the methodology. e.g. guidelines for reviewers.
Feedback
Was this page helpful?
Glad to hear it! Please raise an issue here to tell us how we can improve.
Sorry to hear that. Please raise an issue here to tell us how we can improve.
Last modified
June 25, 2021
: Update website links so they work with the new structure of the site (da7cf7c)