Factors related to COVID‐19 mortality among three Swedish intensive care units—A retrospective study

Mortality due to acute hypoxemic respiratory failure (AHRF) in patients with coronavirus disease‐19 (COVID‐19) differs across units, regions, and countries. These variations may be attributed to several factors, including comorbidities, acute physiological derangement, disease severity, treatment, ethnicity, healthcare system strain, and socioeconomic status. This study aimed to explore the features of patient characteristics, clinical management, and staffing that may be related to mortality among three intensive care units (ICUs) within the same hospital system in South Sweden.


| INTRODUCTION
Worldwide, the coronavirus disease-19  pandemic has exceptionally strained healthcare resources. There have been improvements in the mortality rates of patients with  admitted to the intensive care unit (ICU), which could be attributed to increased knowledge, improved therapeutic options, and system adaptation to the pandemic. However, the Swedish Intensive Care Register demonstrated substantial differences in the unadjusted 30-day mortality rate related to COVID-19, which ranged from 5.9% to 67%, across Swedish ICUs between 2020 and 2021. 1 There are variations in patient outcomes following intensive care admission for COVID-19. [2][3][4][5] These variations may be attributed to several factors, including comorbidities, acute physiological derangement, disease severity, treatment, ethnicity, healthcare system strain, and socioeconomic status. 5 It might be difficult to compare outcomes among countries due to differences in population characteristics, admission criteria, and healthcare facilities. 6 Scandinavia is a unique region since its health-
Regarding acute physiological derangement measured on admission, there were between unit differences in the SOFA scores ( p = .0009), but not the SAPS 3 (p = .1).
Patients in the unit with the lowest mortality rate showed the highest ROX index and P/F ratios immediately before intubation.
Moreover, this unit showed the highest proportion of patients treated using IMV with subsequent prone positioning (PP Normal ( .04 1.6 (0.4-6.9) .09 Hypertensive disease 3.0 (1.6-6.0 Our study confirms previous findings that age and ARDS severity are established predictors of COVID-19-related death in the ICU. 5,12 In contrast, we did not find that comorbidities as assessed by the CCI or treatment received in ICU were independently associated with mortality. Importantly, this study identified the importance of physician-to-patient and nurse-to-patient ratios for mortality. Yet, unexpectedly, nursing workload and occupancy rates did not make a difference. Although we could not determine any causative relationships, our findings identify a potentially modifiable factor for outcome, and sheds light on variations in practice in a system that is widely assumed to be homogeneous in the delivery of health care.

| Staffing
During the pandemic, there were challenges in achieving the target staffing levels suggested by The Swedish Association for Anesthesia and Intensive Care Medicine. 13  Which in turn, could be a marker of high quality care. 26 However, the present article did not specifically aim to assess guideline adherence.
The ROX index predicts NIRS failure, which consequently predicts worse outcomes. 27 We observed significant between unit differences in the timing of intubation. Prolonged NIRS may cause patient self-inflicted injury (P-SILI), which may worsen the clinical outcomes of patients requiring IMV after NIRS. 28 Regarding NIRS, a recent study suggested that the optimal strategy should seek to limit IMV-related risks without increasing the risk of P-SILI. 29 Notably, patients with AHRF and COVID- Given the exploratory aim of our study, we sought to involve all potentially relevant factors to the univariable analysis. Although we were able to identify variation in practice between units, these factors could not predict 90-day mortality in the adjusted analysis. In terms of organizational factors the staffing, admission rate, and workload metrics were functions of the three different ICUs. Unsurprisingly, more nursing-staff was associated with better outcomes. Given the rapid influx of patients in initial phases of the pandemic waves, the staffing requirements have likely been challenging to meet in in several health-care systems. Our data suggest that staffing could have an impact on outcomes, which could be of interest for clinicians when planning ahead for future pandemics.
In conclusion, we identified variations in care related processes, which may be a modifiable risk factor, possibly due to interpretation of relevant guidelines, or guideline adherence. Also, age, disease severity and nurse staffing, but not treatment or socioeconomic status, were independently associated with 90-day mortality among critically ill patients with AHRF due to COVID-19.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.