109 IL-6 inhibitors in hospitalised COVID-19 patients 5 Supplementary material Methods Since it was not possible to follow patients if they were transferred to another hospital, the complete hospital and ICU LOS was not always clear. For example, if a patient would be admitted at the ICU and then transferred to another hospital, the registration would stop, as the original hospital would no longer supply data. The following assumptions were made. If registration ended at an ICU day, hospital and ICU admission were assumed to be incomplete, as patients are always discharged home from the ward and not from the ICU. However, if registration ended at an ICU and restarted again at an ICU day, it was assumed that the patient was transferred to another ICU, and the total ICU LOS was the first ICU stay + period of missing data + second ICU stay. Patients who were admitted via transfer from another hospital were also considered to have an incomplete LOS, as no data from the previous hospital was available. A sensitivity analysis excluding patients with an incomplete LOS was performed. Some patients had two separate COVID-19 hospital admission registered. It was not possible to differentiate between a temporary transfer to another hospital or discharge home and re-admission. If the period between the two hospital admission was less than 90 days, we included those days and added them to stay one and stay two to calculate the total hospital LOS, assuming that they were transferred to another hospital. However, since it is likely that some of these patients were not in another hospital receiving continuous care for COVID-19 but were discharged home, we performed a sensitivity analysis including only the days of their first admission as their hospital LOS. We randomly selected 25% of patients with <90 days between the two hospital admission, and only used the first admission as their hospital LOS. We repeated this process with 50% and 75% randomly selected patients. For patients with more than 90 days between two COVID-19 admission, we used the LOS from the first admission (in which the IL-6 inhibitor was given) for the hospital LOS. While in-hospital death was registered in the database, discharge home was not. If patients had not died during admission, they were either discharged home or transferred to another hospital, from which they could have been discharged home or died. If patients had any activity registered after the end of the hospital admission, such as an outpatient clinical visit or physiotherapy appointment, they were assumed to have been discharged alive. Results Survival analysis In an unadjusted Kaplan-Meier survival curve, we found that survival was best in the
RkJQdWJsaXNoZXIy MTk4NDMw