207 Summary and general discussion 9 different in the other seasons (2013-2014: HR 1.14; 95% CI, 0.96-1.34, 2016-2017: HR 0.99; 95% CI, 0.92-1.07 and 2018-2019: HR 1.01; 95% CI, 0.93-1.10). The P/F ratio on postoperative day 1, day 3 and day 7 did not change significantly between patients who underwent cardiac surgery in ILI season and baseline season. However, undergoing cardiac surgery in ILI season did increase the odds of in-hospital mortality (OR 1.67, 95% CI 1.14-2.46). Based on these results, patients who underwent elective cardiac surgery during a season with a high prevalence of ILI were at increased risk of in-hospital mortality compared to patients in a season with low prevalence of ILI. No evidence was found that this difference is caused by direct postoperative pulmonary complications. In the second part of this thesis, different observational data sources were used to answer specific clinical questions regarding risk factors and treatment of viral respiratory infections. Using three different (epidemiological) approaches, clinical risk factors and treatment effects were studied that would have been difficult to study in a randomised controlled trial setting. In the first chapter of the second part of this thesis Chapter 4, an observational data set of over 200.000 patients was used to study the effect of viral co-infections on the need for IMV and mortality. Setting up a randomised controlled trial to study this (i.e., infecting patients with one or more viruses) would not be ethically acceptable, and a prospective study testing all consecutive SARS-CoV-2 positive patients for other respiratory viral infections is possible in theory, but difficult to organise in practice. Moreover, because viral co-infections are relatively rare, a prospective study would be expensive and time-consuming. In the second chapter of the second part of this thesis, Chapter 5, data from a natural experiment was used to compare survival between different treatment groups of IL-6 inhibitors in hospitalised COVID-19 patients. IL-6 inhibitors were shown to reduce mortality in severe COVID-19 patients early in the pandemic, but the surge in demand quickly led to severe drug shortages. These shortages resulted in changes in the recommended treatment and dose, with four different treatment regimens in the Netherlands. Since the treatment received by a patient was determined by the time at which they were hospitalised, this led to a natural experiment, in which the time period was used as an instrumental variable. Comparing the survival between the different treatment groups would also have been possible in an RCT, but this would have been more expensive and time-consuming than using routinely collected data. Moreover, randomising patients to a lower dose could be considered unethical if a higher dose in known to be effective, by denying patients the best available treatment. In the final chapter of the second part of this thesis, Chapter 6, data from the NICE database, containing data from 16 Dutch ICUs that receive patients after elective cardiac surgery, was used. ILI season was utilised as a proxy for respiratory viral
RkJQdWJsaXNoZXIy MTk4NDMw