127 Cardiac surgery in Influenza-Like-Illness season 6 complications in noncardiac surgery shows conflicting results, with some studies reporting no seasonal differences15,16, a lower incidence of postoperative infections in winter compared with summer17, and higher mortality in winter compared with other seasons18,19. There are several strengths to this study. First, the large number of patients in the ILI season and baseline season. Second, seven consecutive ILI seasons were included, differing in length, starting time in the year, and severity. We found no clear relationship between the severity of the ILI season and the effect on IMV: 2014-2015 and 2017-2018 were relatively long and severe ILI seasons and 2015-2016 was short and relatively mild; however, in these three seasons, we found a significantly shorter duration of IMV in the ILI season compared with baseline season17. Also, all Dutch hospitals that perform cardiac surgery provide data to NICE. Furthermore, as followup in our data was continued until hospital discharge, and given the usually early development of pulmonary complications such as ARDS after cardiac surgery5, it was possible to compare the occurrence of pulmonary complications through the duration of mechanical ventilation, as it is unlikely severe complications will have occurred after hospital discharge. There are also some weaknesses that should be considered when interpreting our results. First, using ILI season as a proxy for the risk of respiratory viral infections was a practical solution but also has some downsides. It is an ecologic fallacy to assume that the only difference between ILI season and baseline season is the number of respiratory viral infections. There are environmental differences (average temperature, hours of sunlight), differences in behavior (hours spent outside, exercise), disease frequency (seasonal affective disorder), and differences in some possible risk factors (vitamin D and cholesterol levels)11,20. Ideally, the possible relationship between asymptomatic respiratory viral infection and pulmonary complications after cardiac surgery should be analyzed using proper viral diagnostics. Second, there may have been unmeasured confounding, for example, because influenza vaccination status was not available. Because influenza vaccination may also correlate with other factors, such as comorbidity, that could have influenced our results. Likewise, we did not have information on patients who were extubated before ICU admission, although this is very uncommon in the Netherlands. No further information was available on the “other cardiovascular surgery” group, but only a small number of patients were listed in this group. No data on the use of extracorporeal circulation during surgery were available, although the proportion of off-pump procedures is expected to be relatively low and thus have little influence on our results or conclusions21. The gradual increase in in-hospital mortality in our study from baseline season through intermediate season to the ILI season is striking and may indicate that current mortality prediction models can be improved. We found no evidence that the difference in in-hospital mortality is caused by direct postoperative pulmonary complications, given the lack of difference in the duration of mechanical ventilation (HR was larger than 1 in all but one ILI season during our study), the absence of a
RkJQdWJsaXNoZXIy MTk4NDMw