Pattern
Setting up a companion code repository takes seven steps, and most research groups can do the first five in an afternoon once the analysis is finished. The steps go from storing the data in the right place to making the code rerun on new data, and the last one, automated checking, is optional.
Claim: recommendation
The author checklist in the program's companion turns these steps into a list you can tick off before you submit.
What would show this claim is wrong
Patient-level data goes to a governed repository such as Vivli or the YODA Project. Data with no patients in it, such as simulations, aggregates or bench measurements, can go in the companion repository itself, versioned with the code.
Use GitHub, GitLab or a similar service. Include every script, notebook and spreadsheet that made a result, the exact software versions, and one set of instructions from data to results.
Add a CITATION.cff file so people cite it correctly, and a codemeta.json file that describes the software. Both are existing standards; there's no need to invent a format.
Archive a release on Zenodo or a similar archive to get a DOI, a permanent address. Put that DOI in the paper, and put the paper's and the data's DOIs in the repository, so each points to the other two.
Some journals give badges or featured placement to papers whose code reruns, and some funders give credit for it in review. Whether these raise citations in medicine isn't known, but they do raise sharing.
When the method runs on new data, publish a new release with its own DOI. Never change the release the paper cites.
A continuous integration job, a program that reruns the code on every change, catches breakage early. Rules first, then a model only where rules can't answer, then a person.