Cox regression models for intermediate events, with discharge from hospital as an example

Area of Science:

• Surgical outcomes research
• Patient safety and healthcare quality

Background:

• Hospital discharge is an intermediate event impacting patient health outcomes.
• Post-hospitalization disease onset is common, particularly for surgical site infections.

Purpose of the Study:

• To explore the role of hospital discharge in the risk of surgical infections following hernia repair.
• To evaluate direct and interactive effects of discharge on infection risk, considering methodological challenges.

Main Methods:

• Utilized Cox models to analyze data from hernia repair operations with 1-month follow-up.
• Addressed methodological difficulties including waiting-time bias and selective discharge bias.

Main Results:

• Fifty percent of surgical infections occurred after patients were discharged home.
• Cox models could correct for waiting-time bias, but the protective effect of discharge remained challenging to interpret.

Conclusions:

• Hospital discharge timing significantly influences the risk of post-operative surgical infections.
• Further research is needed to fully interpret the protective effect of discharge and refine bias correction methods in surgical outcomes studies.