Characteristics and outcomes of patients with COVID‐19 admitted to ICU in a tertiary hospital in Stockholm, Sweden

Abstract Background Information on characteristics and outcomes of intensive care unit (ICU) patients with COVID‐19 remains limited. We examined characteristics, clinical course and early outcomes of patients with COVID‐19 admitted to ICU. Methods We included all 260 patients with COVID‐19 admitted to nine ICUs at the Karolinska University Hospital (Stockholm, Sweden) between 9 March and 20 April 2020. Primary outcome was in‐hospital mortality among patients with definite outcomes (discharged from ICU or death), as of 30 April 2020 (study end point). Secondary outcomes included ICU length of stay, the proportion of patients receiving mechanical ventilation and renal replacement therapy, and hospital discharge destination. Results Of 260 ICU patients with COVID‐19, 208 (80.0%) were men, the median age was 59 (IQR 51‐65) years, 154 (59.2%) had at least one comorbidity, and the median duration of symptoms preceding ICU admission was 11 (IQR 8‐14) days. Sixty‐two (23.8%) patients remained in ICU at study end point. Among the 198 patients with definite outcomes, ICU length of stay was 12 (IQR, 6‐18) days, 163 (82.3%) received mechanical ventilation, 28 (14.1%) received renal replacement therapy, 60 (30.3%) died, 62 (31.3%) were discharged home, 47 (23.7%) were discharged to ward, and 29 (14.6%) were discharged to another health care facility. On multivariable logistic regression analysis, older age and admission from the emergency department was associated with higher mortality. Conclusion This study presents detailed data on clinical characteristics and early outcomes of consecutive patients with COVID‐19 admitted to ICU in a large tertiary hospital in Sweden.


| INTRODUC TI ON
The first patient with confirmed coronavirus disease 2019 , caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), admitted to a Swedish intensive care unit (ICU) was reported on 6 March 2020. As of 20 April 2020, a total of 1100 patients with COVID-19 had been treated in Swedish ICUs, according to the Swedish Intensive Care Registry. 1 Approximately one quarter of these patients were admitted to the Karolinska University Hospital ICUs.
Data on baseline characteristics and outcomes of ICU patients with COVID-19 are essential for planning actions preceding local outbreaks and to assess the need of rehabilitation in ICU survivors.
Recent reports from China, Italy and the US indicate a mortality rate between 60% and 85% among ICU patients with data on definite outcomes (discharged alive from ICU or death). [2][3][4] Differences in patient characteristics and socioeconomic status, health care systems, ICU admission thresholds and availability of ICU beds between countries might explain such wide differences in mortality. [5][6][7] The aim of the current study was to examine clinical characteristics and outcomes of ICU-treated patients with COVID-19 in a large tertiary hospital in Stockholm, Sweden.

| ME THODS
This study was approved by the Swedish Ethical Review Authority (approval number 2020-01477) with a waiver of informed consent. The study has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

| Study population
The study was conducted at the Karolinska University Hospital, the largest (1600 beds) academic hospital in Stockholm, Sweden. As for other ICUs in Sweden, all intensive care beds are funded by the government. During the COVID-19 pandemic, the Karolinska University Hospital expanded intensive care from three to nine ICUs (from 38 to 182 ICU beds). We included all consecutive patients admitted to any of the ICUs at the Karolinska University Hospital in Stockholm, Sweden with confirmed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) by polymerase chain reaction, between 9 March 2020 and 20 April 2020.

| Data collection
We included routine clinical data prospectively recorded and validated in our local ICU COVID-19 quality registry. Registry data are retrieved from the Swedish Intensive Care Registry and from available electronic patient data management systems (Clinisoft [GE, Barringgton IL] and the Karolinska internal data warehouse, including data from Take Care [CompuGroup Medical, Koblenz, Germany]). In this study, we included all consecutive ICU patients with COVID-19 recorded in the local ICU COVID-19 quality registry from 9 March until 20 April 2020. Study end point was 30 April 2020.
Collected data included patient demographics, comorbidities, body mass index (BMI), duration of symptoms preceding ICU admission, medication associated with coronavirus disease administered before and/or during ICU admission, mechanical ventilation, renal replacement therapy, ICU length of stay, discharge destination and mortality. Information on comorbidities was obtained (from previous medical records, chronic medications, and/or from patients or relatives) and recorded by the treating clinician. There was no missing baseline admission data except for BMI (missing data for 30 patients). Clinical outcomes were available for all patients.

| Outcomes
The primary outcome was in-hospital mortality, at study end point (30 April 2020). Secondary outcomes included ICU length of stay, the proportion of patients receiving mechanical ventilation and renal replacement therapy, and hospital discharge destination. During the review process, an extended follow-up time of mortality was requested for all patients. Therefore, we included 120-day mortality as a secondary outcome measure.

| Statistics
Categorical variables are presented as number (with percentages).

Editorial Comment
In this important, timely, and well-conducted study, clinical characteristics and outcomes of 260 patients with COVID-19 admitted to the intensive care units at the Karolinska University Hospital, Stockholm, are presented.
Time to death until 30 April 2020 was presented using the Kaplan-Meier methodology. Assessment of the following potential risk factors for in-hospital mortality was performed using multivariable logistic regression analysis: age, sex, symptom duration (log-transformed to achieve normality distribution), number of comorbidities (none vs. ≥1), location before ICU admission (ward or other ICU vs emergency department), and chloroquine phosphate use.
Model discrimination was determined by the area under the curve.
Additionally, time to death within 120 days was assessed using multivariable Cox regression analysis and presented using the Kaplan-Meier methodology. Proportional hazard assumption was assessed using the Shoenfeld Residuals test. A two-sided P Value < 0.05 was considered statistically significant.
Mortality and ICU support for patients with data on definite outcomes (discharged alive from ICU or died; n = 198) and for patients still in ICU (n = 62) are presented in Table 3 Figure S1 in Electronic Supplementary Material 1.
Among the 62 patients who were still in ICU at study end point, we observed a median ICU length of stay and duration of mechanical ventilation above 20 days and use of renal replacement in more than 40% across all age groups.  The proportion (95% CI) of patients who died in hospital by age interval, sex, BMI interval, location before ICU admission, symptom duration, number and type of comorbidity and chloroquine phosphate use are summarized in Figure 2. Additional outcomes by subgroup are presented in the Supplementary Appendix (Table S1 to

| D ISCUSS I ON
We studied the first 260 patients with COVID-19 admitted to 9 ICUs in a tertiary hospital in Stockholm, Sweden. In this cohort the majority was men, median age was 59 years and 60% had at least one comorbidity. Among 198 patients with data on definite outcomes (discharged alive from ICU or death), mortality was 30.3%, median ICU length of stay was 14 days and 31.3% were discharged home. On adjusted analysis, we found a significant association of older age and admission from the emergency department with higher mortality.
While most European countries share the same goal of ensuring and controlling health care and limit economic consequences of the COVID-19 pandemic, Sweden has taken a slightly different route to reach these common goals. In contrast to political and governmental reinforcement of country-or region-wide lockdowns, Sweden have built its strategy on a combination of binding regulations and We observed a lower mortality compared to ICU cohorts from China, Italy and the US. [2][3][4]11 It seems that illness severity from SARS-CoV-2 varies substantially in different patient groups, with the highest mortality reported in older men with multiple coexisting diseases. 11,12 Mortality differences in ICU-treated patients with should be interpreted with caution, as treatment was not given in a randomized fashion and no association between chloroquine phosphate use and mortality was observed in the multivariable analysis.
This study has limitations. First, the single-hospital design limits generalizability. However, as of 20 April 2020 approximately one quarter of all patients with COVID-19 admitted to Swedish ICUs had been treated at the Karolinska University Hospital. Second, at study end point, one quarter of the patients were still treated in ICU. These patients have an extended length of stay, a high proportion of renal replacement therapy, and potentially different outcomes than those included in our outcome analysis. Finally, information on severity of comorbidity and socioeconomic factors, which may impact outcome in critically ill patients, were not available in our data set. However, the study is based on validated data with available clinical outcome data for all patients.

| CON CLUS IONS
In this study of critically ill patients with COVID-19 admitted ICUs in a tertiary hospital in Stockholm, Sweden, presenting baseline characteristics and clinical course, mortality was 30.3%. Older age and admission from the emergency department were associated with higher mortality.