Thank you for this interesting study
What has been shown is an association rather than causation.: A multivariate regression meta -analysis may help clarify.
The study cannot be appropriately standardized for confounders (there was a reason why the prescribing doctor thought steroids were clearly required at that time in the patient) and thus outcomes are not unexpected in the cohort of users vs non-users. At worst this study could possibly be completely misleading.
Hypothetically: if the group prescribed were instead randomized to half prescribed and half not - then we could tell whether steroids in the short term are harmful or beneficial on the outcomes recorded. It is possible steroids could actually shorten length of illness (and immobility) and actually reduced DVTs etc OR perhaps it is indeed harmful.
The study however is valuable in generating hypotheses and also cautioning in the potential indiscriminate use of steroids. Yet it may cause harm if it influences colleagues to refrain from using steroids in short courses for appropriate reasons.
Competing interests: No competing interests