Statins’ safety and impact on the clinical outcomes in COVID-19 critically ill patients: A Multicenter, Cohort Study

Purpose: The complications of Severe Corona Virus Disease 2019 (COVID-19) are attributed to the overproduction of early response proinflammatory cytokines, causing a systemic hyperinflammatory state. Statins are potentially a potent adjuvant therapy in COVID-19 infection due to their pleiotropic and anti-inflammatory effects, which are independent of their cholesterol-lowering activity. This study investigates the impact of statin use on the outcome of critically ill patients with COVID-19. Methods A multicenter, retrospective cohort study of all adult critically ill patients with confirmed COVID-19 admitted to Intensive Care Units (ICUs) between March 1, 2020, and March 31, 2021. Eligible patients were classified into two groups based on statin use during ICU stay and were matched with a propensity score which was based on patient’s age and admission APACHE II and SOFA scores. The primary endpoint was in-hospital mortality. Other outcomes were considered secondary... Results A total of 1049 patients were eligible;502 patients were included after propensity score matching (1:1 ratio). The 30-day (hazard ratio 0.75 (95% CI 0.58, 0.98), P  = 0.03) and in-hospital mortality (hazard ratio 0.69 (95% CI 0.54, 0.89), P  = 0.004) were significantly lower in patients who received statin therapy on multivariable cox proportional hazards regression analysis. Moreover, patients who received statin have a lower risk of hospital-acquired pneumonia (OR 0.48(95% CI 0.32, 0.69), P  = < 0.001), lower levels of markers of inflammation on follow up and no increased risk of liver injury. Conclusion The use of statin during ICU stay in COVID-19 critically ill patients may have a beneficial role and survival benefits with a good safety profile.


Introduction
Since the spread of the Severe Acute Respiratory Syndrome Corona Virus 2 (SARS-CoV-2), leading to Coronavirus Infectious Disease 2019  in 2019, over four million people have died worldwide [1]. Although the mortality of patients with COVID-19 is usually due to respiratory-related complications [2], growing evidence shows increased morbidity and mortality related to multiorgan failure, including heart and kidney failure [3][4][5]. These complications are attributed to the overproduction of early response proin ammatory cytokines, causing a systemic hyperin ammatory state in patients with severe COVID-19 [6,7].
The pre-existence of cardiovascular disease (CVD) or cardiovascular (CV) risk factors has been associated with poor prognosis and increased mortality of hospitalized patients with COVID-19 [8,9]. In critically ill patients with COVID-19, the ICU and in-hospital mortality risks in patients with underlying heart failure were signi cantly higher, reaching 44.6% and 39%, respectively. [10] The use of statin therapy has been associated with reduced CV risk and CV death in non-COVID-19 patients [11]. Therefore, the international guidelines recommend using statin therapy in patients with atherosclerotic vascular disease (ASCVD) risk factors or existing ASCVD/CVD as primary or secondary prevention or cholesterol-lowering [12][13][14].
Part of the anti-in ammatory is ultimately to reduce in ammatory markers such as CRP, as shown in a published CARE study [34]. As a result, the suggested clinical bene t of statin therapy could emerge as early as 30 days after starting treatment, and it is consistent over time [35]. Moreover, statin therapy has been reported to have antiviral activity through immunomodulation and viral replication suppression [18]. These effects of statins suggest that they might have a promising role in indirectly improving the clinical outcomes in patients with COVID-19 [23].
Multiple studies reported a decrease in mortality and in ammatory response in patients with COVID-19 using statin therapy [20][21]. In contrast, a systematic review and another national observational study showed no improvement in patients' hospital outcomes and rather an increase in mortality, respectively [19,22]. However, most of the previous studies were general to all hospitalized patients with mild to severe COVID-19, while only limited studies investigated their bene t in critically ill patients with severe COVID-19 [23]. Hence, the bene t of statin therapy use in this population remains questionable. Thus, this study aimed to investigate the association of statin use on the outcome of critically ill patients with COVID-19.

Study design
This was a multicenter, retrospective cohort study including adults critically ill patients con rmed COVID- 19

Study participants
We included adults (age ≥ 18 years) patients who were admitted to the ICU with con rmed COVID-19. Patients were diagnosed with COVID-19 using Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) nasopharyngeal and/ or throat swabs. Patients were excluded if they were < 18 years old, had ICU length of stay (LOS) ≤ one day, or died or were labeled as "Do-Not-Resuscitate" within 24 hours of ICU admission ( Figure 1). Eligible patients were then classi ed into two groups based on the statin therapy use during ICU stay. Statins were prescribed in the ICU as part of the medication reconciliation process if they had been prescribed in the pre-ICU period or were initiated in the ICU for various indications at the discretion of the ICU treating team and stopped based on physicians' clinical judgment.

Study setting
The study took place at four hospitals in Saudi Arabia; King Abdulaziz Medical City (Riyadh), King Abdulaziz University Hospital (Jeddah), King Abdullah bin Abdulaziz University Hospital (KAAUH) (Riyadh), and King Salman Specialist Hospital (Hail). We included both tertiary and secondary centers.
The selection of these centers were based on the geographic distribution, availability of electronic records and center's willing to participate in the national project. The primary site for this multicenter retrospective study was King Abdulaziz Medical City (Riyadh) which is considered a tertiary care center that include more than 1500 beds. There are more than nine ICU units. KAAUH is a 300-bed secondary teaching hospital that have one medical ICU unit with 28 beds.

Data collection
Each patients' data were collected and managed using Research Electronic Data Capture (REDCap®) software hosted by King Abdullah International Medical Research Center (KAIMRC). We collected patients' demographic data, comorbidities, vital signs and laboratory tests, Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA), and Nutrition Risk in Critically ill (NUTRIC) scores, Glasgow Coma Score (GCS), statin therapy use (type and dose), acute kidney injury, use of prone positioning, receipt of mechanical ventilation (MV) and MV parameters (e.g., PaO 2 /FiO 2 ratio, FiO 2 requirement) within 24 hours of ICU admission. Moreover, renal pro le, liver function tests (LFTs), coagulation pro le (i.e., INR, aPTT, brinogen, D-dimer), and in ammatory markers (Ferritin, procalcitonin and creatine phosphokinase (CPK)) within 24 hours of ICU admission were collected. Tocilizumab and corticosteroids use were recorded for the eligible patients. All patients were followed until they were discharged from the hospital or died during in-hospital stay, whichever occurred rst.

Outcomes
The primary endpoint was the in-hospital mortality. The secondary endpoints were 30-day mortality, hospital LOS, ICU LOS, receipt of MV, ventilator-free days (VFDs) at 30 days, and ICU-acquired complications (new onset atrial brillation, AKI, liver injury, hospital-acquired pneumonia, and secondary fungal infection).

Outcome de nition (s)
The 30-day mortality was de ned as a death occurring for any cause within 30 days of the admission date during hospital stay; patients who were discharged from the hospital alive were presumed to be survived. All patients were followed until they were discharged from the hospital or died during the in-hospital stay, whichever occurred rst.
Ventilator-free days (VFDs) at 30 days were calculated as the following: if the patients die within 30 days of MV, the VFDs = 0, VFDs = 30 − days after MV initiation (if patient survived and was successfully liberated from MV), and VFDs = 0 if the patient is on MV for >30 days.
Acute kidney injury (AKI) was de ned as a sudden decrease of renal function within 48 hours, de ned by an increase in absolute SCr of at least 26.5 μmol/L (0.3 mg/dL) or by a percentage increase in SCr ≥ 50% (1.5× baseline value) during ICU stay [24] .
Acute liver injury was de ned as alanine aminotransferase (ALT) exceeding three times the upper limit of normal or double in patients with elevated baseline ALT during the ICU stay.
Hospital-acquired pneumonia was de ned as pneumonia that occurs > 48 hours after admission and did not appear to be incubating at the time of admission [25].
Secondary fungal infection was identi ed through the blood, urine, wound, drainage, cerebrospinal uid, and/or respiratory cultures. The fungal growth was considered signi cant if the growth was ≥ 100,000 colony forming units (CFUs)/ml in sputum or endotracheal aspiration, ≥10,000 CFUs of single organism/ml in bronchoalveolar lavage or ≥ 1000 CFUs of single organism/ml in protected specimen brushes. Additionally, urinary cultures were considered signi cant if showing a growth ≥100,000 CFUs/ml of no more than two species of microorganisms. Cultures were excluded if the laboratory reported them as a "contaminant sample." [26,27] Respiratory failure was de ned as either low arterial carbon dioxide tension (PaCO 2 ) or hypoxemic respiratory failure (PaO 2 < 60 mm Hg with a normal or hypercapnic respiratory failure (PaCO 2 > 50 mm Hg) that requires mechanical ventilation.

Statistical analysis
We presented numerical variables (continuous variables) as mean and standard deviation (SD), or median and lower quartile (Q1) and upper quartile (Q3), as appropriate and categorical variables as number (percentage). The normality assumptions were assessed for all numerical variables using a statistical test (i.e., Shapiro-Wilk test) and graphical representation (i.e., histograms and Q-Q plots).
Model t assessed using the Hosmer-Lemeshow goodness-of-t test.
Baseline characteristics and outcome variables were compared between the two study groups. For categorical variables, we used the Chi-square or Fisher's exact test. We compared the normally distributed continuous variables using student t-test and other non-normally distributed continuous variables with the Mann-Whitney U test. Multivariable Cox proportional hazards regression analyses were performed for the 30-day and in-hospital mortality. Proportionality assumption was assessed before tting the cox model. Visual assessment was performed to assess the assumption by plotting log(-log) plot and also by testing the correlation of scaled Schoenfeld residuals with rank-ordered time. Additionally, Kaplan-Meier (KM) plots were generated for these outcomes. Multivariable regression analysis and negative binomial regression were used for the other outcomes considered in this study. Regression analysis was done by considering PS score as one of the covariates in the model. The odds ratios (OR), hazard ratio (HR), or estimates with the 95% con dence intervals (CI) were reported as appropriate. No imputation was made for missing data as the cohort of patients in our study was not derived from random selection. We considered a P value of < 0.05 statistically signi cant and used SAS version 9.4 for all statistical analyses.
Propensity score matching procedure (Proc PS match) (SAS, Cary, NC) was used to match patients who received statin therapy (active group) to patients who did not (control group) based on patient's age and baseline severity scores (APACHE II and SOFA scores) within 24 hours of ICU admission. A greedy nearest neighbor matching method was used in which one patient who received statin therapy (active) group matched with one patient who did not (control), which eventually produced the smallest within-pair difference among all available pairs with treated patients. Patients were matched only if the difference in the logits of the propensity scores for pairs of patients from the two groups was less than or equal to 0.5 times the pooled estimate of the standard deviation.

Results
A total of 1049 patients met our inclusion criteria. Of these patients, 279 (26.6 %) patients received statins during ICU stay. After propensity score matching, 502 patients were included (1:1 ratio) according to the selected criteria. The majority of the patients (86.4 %) received statin as a continuation due to underlying coexisting illness (e.g., DLP, IHD). Only twenty patients (8.6 %) required statin discontinuation during ICU stay, most often due to rhabdomyolysis (9 patients), followed by liver damage (8 patients).
Atorvastatin (81.3 % ) with a median dose of 20 mg was the most often used statin, followed by Rosuvastatin (14.1 % ) with a median dose of 20 mg.

Demographic and Clinical Characteristics
The majority of the included patients in both arms were male (68.9 %) with a mean age of 61.6 (SD +14.79). The predominant underlying comorbidities were diabetes mellitus (59.0 %), followed by hypertension (55.2 %) and dyslipidemia (19.3 %). There were some notable differences between the two groups before propensity score matching Following the propensity score matching based on patient's age and baseline severity scores (APACHE II and SOFA scores), most of these baseline and demographic characteristics were similar between the two groups. Summary of the patients' baseline characteristics in available in the (Additional le-1)

30-day and in-hospital mortality
In a crude analysis, there was no signi cant difference in the 30-day mortality (44.2 % vs. 49.2 %, P = 0.27) or in-hospital mortality (46.8 % vs. 53.0 %, P = 0.57) between the statin and the control group, respectively (Table 1). However, using the cox proportional hazards regression analysis, patients who received statin had a lower 30-day mortality (HR 0.75 (95% CI 0.58, 0.98)), P =0.03) as well as in-hospital mortality (HR 0.69 (95% CI 0.54, 0.89, P =0.004)) compared with the control group ( Table 1). The overall survival probabilities were higher during hospital stay among patients who received statin therapy before and after propensity score-matched ( Figure 2).

Ventilator free days and Length of stay
During the ICU stay, the mean ventilator-free days (VFD) was 10.6 days (±12.2) and 10.2 days (±12.3) in patients who received statin therapy compared to the control group respectively; however, it was not signi cant between the two groups (beta coe cient : 0.03 (95% CI -0.38, 0.45), P= 0.87). On the ip side, among those who survived during ICU stay, we observed that critically ill patients who received statin had a longer ICU length of stay (LOS) and hospital LOS with a (beta coe cient 0.24 (95% CI 0.07, 0.42), P=0.007)), and (beta coe cient 0.39 (95% CI 0.21, 0.58), P <0.0001 respectively)) ( Table 1).

Complications during ICU stay
Complications during ICU stay were reported in (Table 2). During ICU stay, we observed that patients who received statin were shown to have statistically signi cantly lower odds of hospital-acquired pneumonia (bacterial or fungal) (OR 0.48(95% CI 0.32, 0.69), P= 0.001)). On the other hand, statin users have a higher odd for secondary fungal infection (OR 2.48(95% CI 1.44, 4.24), P= 0.001) compared with the control group ( Table 2).

Follow-up biomarkers during ICU stay
The follow-up biomarkers such as D-dimer (p-value <0.001), and in ammatory markers such as procalcitonin (p-value <0.001), and total WBC count (p-value 0.04) were signi cantly lower during ICU stay in patients who received statin compared with the control group. In contrast, creatine phosphokinase (CPK) and ferritin levels were similar among the groups, as described in Table 2.

Discussion
In this retrospective, multicenter study, we investigated the impact of statin's use on the outcome of critically ill patients with COVID-19. Additionally, we evaluated the statins' possible effects on the ICUacquired complications, including safety outcomes. After using cox proportional hazards regression analysis, patients who received statins had a lower 30-day mortality P=0.03) as well as in-hospital mortality ( P=0.004). Also, patients who received statins were associated with lower hospital-acquired pneumonia and in ammatory markers with no statistically signi cant difference in liver injury (P=0.05).
This study demonstrated that statin therapy in ICU patients with COVID-19 was associated with reduced risks of 30-day and all-cause in-hospital mortality. This bene t of statin might be attributed to its pleiotropic anti-in ammatory properties reducing the CRP levels and interleukin 6 [28]. These in ammatory markers are known to increase during COVID-19 disease and increase the risk of mortality [28]. In addition, statins have an antioxidant effect and improves endothelial dysfunction, which might help decrease the CV events due to the hyper-coagulopathy status during the course of COVID-19 [28]. Theoretically, statins may help reduce the cytokine storm that is usually associated with the poor prognosis of patients infected with COVID-19 [17]. Similar to our ndings, a meta-analysis conducted by Kow et al. reported a signi cant reduction in fatal or severe COVID-19 disease with the use of statins [17]. However, not all the included studies in this meta-analysis had critically ill patients in their study population [21]. On the other hand, another meta-analysis by Scheen et al. and an observational study by Russo et al. showed that the use of statins was not associated with mortality bene ts in patients with COVID-19. [29,30] This variation from our ndings might be related to the differences in the population included in the previous studies, however our study only focuses on critically ill patients with severe COVID-19. [22,30].
In contrary to our ndings, a randomized controlled trial (INSPIRATION-S trial) presented in the American College of Cardiology (ACC 2021) comparing atorvastatin 20 mg daily vs. placebo showed that initiating statin in critically ill patients with COVID-19 was not associated with mortality bene ts. [31] However, they included only the new initiation of statin therapy compared to our study in which the majority of our population were long term statins users. The prolonged anti-in ammatory effect of statins might play a major effect on reducing the levels of in ammatory markers such as CRP only included new initiation of statin therapy and they excluded patients who use statins prior to the admission. As a result, the suggested clinical bene t of statin therapy could emerge as early as 30 days after starting treatment and it is consistent over time. [34] Thus, prolonged statin use prior to admission might lessen the severity of cytokines storm and its complications.
We found that the use of statins was not associated with any signi cant reduction in MVFD or in preventing respiratory failures that require MV eventually. Contrary to our ndings, a preliminary study in non-ICU patients demonstrated a signi cant reduction in the risk of MV in patients using statin among patients infected with COVID-19 [32].
Interestingly, patients who received statin have lower odds of pneumonia. This nding is consistent with previous data which could be related to a proposed antibacterial effect for statins [36,37,38]. On the other hand, we observed that critically ill patients who received statins had signi cantly longer ICU and hospital LOS. The higher survival rate could explain the prolonged ICU and hospital LOS in our cohort in the statins group. Another explanation, the more prolonged ICU and hospital LOS may be secondary to higher odds of secondary fungal infection in patients who received statin therapy.
Besides the observed survival bene ts in our study with statins therapy use in critically ill patients with COVID-19, we did not report any safety concerns related to statins' side effects or complications. Even though earlier reports from a cohort of 1099 patients with COVID-19 from China showed that up to 39.4% had AST >40 U/L and 28.1% had ALT >40 U/L, and most of these elevations occurred in critical COVID-19 cases [33]. We did not nd any signi cant differences in liver injury among the statins and non-statins critically ill groups with COVID-19. Statin showed a good safety pro le in our cohort. Still, its use might be hindered by clinicians' reluctance to utilize it in critically ill patients with COVID-19 due to (liver injury, myotoxicity, and rhabdomyolysis-related kidney injury). Therefore, continuing statin therapy as primary or secondary prevention is advisable prior to ICU admission and during ICU stay unless contraindicated.
We believe that our multicenter cohort study is one of few studies that evaluated statins' clinical and safety outcomes in COVID-19 critically ill patients using propensity score matching and multiple regression analysis to minimize the bias. Nevertheless, we also determined some limitations in our study. The retrospective nature of our study may have been affected by missing documentation that could be translated to unmeasured confounders. Also, the medication history use before admission might be affected by limited reconciliation during COVID-19 pandemics. Even though we observed mortality bene ts, we did not assess the long-term bene t of statin use after COVID-19 survival. Thus a large randomized controlled trial is needed to investigate the e cacy and safety of statin use in critically ill patients with COVID-19.

Conclusion
Statin use during ICU stay in COVID19 critically ill patients was associated with lower mortality risk with no safety concerns. The continuation or the initiation of statin therapy during the ICU stay in patients with COVID-19 might be a reasonable approach unless statins are contraindicated. The result of this multicenter retrospective study motivates prospective clinical studies to con rm our ndings.

Declarations
The datasets used and/or analyzed during the current study are available from corresponding author on reasonable request.

Ethics approval and consent to participate
The study was approved in January 2021 by King Abdullah International Medical Research Center Institutional Review Board, Riyadh, Saudi Arabia ((Ref.# NRC21R/015/R). Participants' con dentiality was strictly observed throughout the study by using anonymous unique serial number for each subject and restricting data only to the investigators. Informed consent was not required due to the research's method as per the policy of the governmental and local research center.

Consent for publication
Not applicable.   ∆^ Denominator of the percentage is non-atrial brillation as comorbidity.
∆ Denominator of the percentage is the total number of patientŝ^ Chi-square test is used to calculate the P-value/** Fisher's Exact teat is used to calculate P-value.
Wilcoxon rank sum test is used to calculate the P-value. $* Generalized linear model is used to calculate estimates and p-value. Figure 1 Critically ill patients with COVID-19 owchart