COVID-19 and Acute Ischemic Stroke Mortality and Clinical Outcomes among Hospitalized Patients in the United States: Insight from National Inpatient Sample

Coronavirus-19, primarily a respiratory virus, also affects the nervous system. Acute ischemic stroke (AIS) is a well-known complication among COVID-19 infections, but large-scale studies evaluating AIS outcomes related to COVID-19 infection remain limited. We used the National Inpatient Sample database to compare acute ischemic stroke patients with and without COVID-19. A total of 329,240 patients were included in the study: acute ischemic stroke with COVID-19 (n = 6665, 2.0%) and acute ischemic stroke without COVID-19 (n = 322,575, 98.0%). The primary outcome was in-hospital mortality. Secondary outcomes included mechanical ventilation, vasopressor use, mechanical thrombectomy, thrombolysis, seizure, acute venous thromboembolism, acute myocardial infarction, cardiac arrest, septic shock, acute kidney injury requiring hemodialysis, length of stay, mean total hospitalization charge, and disposition. Acute ischemic stroke patients who were COVID-19-positive had significantly increased in-hospital mortality compared to acute ischemic stroke patients without COVID-19 (16.9% vs. 4.1%, aOR: 2.5 [95% CI 1.7–3.6], p < 0.001). This cohort also had significantly increased mechanical ventilation use, acute venous thromboembolism, acute myocardial infarction, cardiac arrest, septic shock, acute kidney injury, length of stay, and mean total hospitalization charge. Further research regarding vaccination and therapies will be vital in reducing worse outcomes in patients with acute ischemic stroke and COVID-19.


Introduction
The Severe Acute Respiratory Virus Syndrome Coronavirus 2 (SARS-CoV-2), first detected in Wuhan, China, in December 2019, has caused a worldwide pandemic known as   [1,2]. Despite primarily being a respiratory virus, it also affects multiple organ systems, including the nervous system [3]. Neurological complications have been documented in approximately 36.4% of COVID-19 patients, and roughly 28.2% had severe central nervous system (CNS) injuries, contributing to worse morbidity and mortality [4]. Known neurological complications include acute ischemic stroke (AIS), neuropsychiatric symptoms, encephalopathy, seizures, myopathies, and polyradiculoneuropathies [5]. COVID-19 infection is associated with worse outcomes in AIS patients [6,7]; however, large-scale studies evaluating AIS outcomes with COVID-19 infection are limited.
The association of AIS with COVID-19 infection is well-established by multiple studies due to the increased reports of this condition during the pandemic [3,[8][9][10]. Ischemic stroke refers to a reduction in blood flow to part or all of the brain resulting in tissue damage and can be characterized as an acute, subacute, or chronic process [11]. Ischemic stroke subtypes can also be categorized using the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification based on the underlying etiology (i.e., large artery atherosclerosis or large vessel occlusion [LVO]; cardioembolism; small vessel occlusion; stroke of other, unusual, determined etiology; and stroke of undetermined etiology) [11]. LVOs constitute the greatest proportion (37.3%) of AIS, with the middle cerebral artery as the most common (49.6%) culprit [11]. Globally, there is a slight difference between sex and incidence of AIS, and this slight predominance in females between countries is due to AIS being more common among older ages and women dominating in senescence [5,12]. Its signs and symptoms are highly variable and include sudden unilateral face and extremity weakness or focal deficits, slurred speech, confusion, vision changes, and death [12]. AIS is associated with elevated biomarkers such as prothrombin time (PT), activated partial thromboplastin time (aPTT), c-reactive protein (CRP), D-dimer, fibrinogen, lactate dehydrogenase (LDH), and ferritin [3]. Clinical presentation and neuroimaging (computed tomography angiography [CTA] or CT perfusion studies) remain the mainstays for the evaluation and diagnosis [10].
The pathophysiology of AIS in COVID-19 infection remains unclear and debatable [3,8]. It is hypothesized to involve a prothrombotic state induced by the hyperinflammatory response from the underlying COVID-19 infection, which leads to elevated antiphospholipid antibodies, von Willebrand factor, factor VIII, fibrinogen, and D-dimer markers [3]. Another mechanism is believed to be due to the binding of the angiotensin-converting enzyme (ACE) 2 receptor and the resultant downregulation of blood pressure [4]. Additionally, cryptogenic causes are thought to be very common in this population, including endothelial dysfunction and cardiomyopathy after a viral or respiratory illness, resulting in a heightened risk of stroke [5,12,13].
Our study's objective is to assess outcomes between AIS patients with and without COVID-19 infection, by utilizing data from the National Inpatient Sample (NIS). The primary outcome was in-hospital mortality. Secondary outcomes included mechanical ventilation, vasopressor use, mechanical thrombectomy, thrombolysis, seizure, acute venous thromboembolism (VTE), acute myocardial infarction (MI), cardiac arrest, septic shock, acute kidney injury (AKI), AKI requiring hemodialysis (HD), length of stay (LOS), mean total hospitalization charge, and disposition.

Materials and Methods
This retrospective study utilized the NIS Healthcare Cost Utilization Project [HCUP], sponsored by the Agency for Healthcare Research and Quality [AHRQ] database, which is an all-payer database that approximates a 20% stratified sample of discharges from US community hospitals [14]. In this analysis, we used the 2020 National Inpatient Sample (NIS) data set, which included hospitalizations from 1 January 2020 to 31 December 2020, and was made available to the public in October 2022. The NIS database contains a deidentified collection of billing and diagnostic codes from participating hospitals. The NIS dataset does not directly involve 'human subjects' (consistent with federal regulations and guidance [15]) and is exempt from institutional review board approval.

Inclusion and Exclusion Criteria
All patients of age ≥ 18 years, admitted to the hospital with acute ischemic stroke and COVID-19 infection, were included in the study. We used ICD-10 clinical modification (CM) codes to retrieve patient samples and comorbid conditions, and ICD-10 procedure codes were used to identify in-hospital procedures. A detailed code summary is provided in Supplementary Table S1. Patient's age < 18 years, inter-hospital transfers, missing NIH stroke scale (NIHSS) on admission, contrast allergy, and hemorrhagic stroke were all excluded from the study (Supplemental Figure S1).

Covariates
The NIS data sample contains data regarding in-hospital outcomes, procedures, and other discharge-related information. Variables were divided into patient level, hospital level, and illness severity. a.
Patient-level: age, race, sex, comorbidities, insurance status, income in the patient's zip code, disposition; b.
Hospital level: location, teaching status, bed size, region; c.

Study Outcomes
The primary outcome assessed was in-hospital mortality. Secondary outcomes were (a) intubation rate and vasopressor use; (b) acute MI, AKI and AKI requiring HD, andVTE; (c) tPA and mechanical thrombectomy use; (d) septic shock and cardiac arrest; (e) length of stay; (f) financial burden on healthcare; and (g) resource utilization.

Statistical Methods
STATA 17 (StataCorp LLC, College Station, TX, USA) was utilized for statistical analysis [16]. The unweighted sample was 6.47 million observations, and the weighted sample was around 32.3 million discharges for the year 2020. Patients who were admitted with acute ischemic stroke were retrieved with ICD-10 CM codes, and this group was further divided based on COVID status. The chi-square test was used to compare categorical variables, and linear regression was used for continuous variables. For the primary outcome, univariate logistic regression was used to calculate the unadjusted odds ratio for variables of interest, and p values of <0.2 on univariate logistic regression were used to build a multivariate logistic regression model to adjust for potential confounders. A multivariate linear regression model was used for continuous variables (length of stay and total hospital charge). A two-tailed p-value of 0.05 was considered significant. We conducted a secondary analysis with propensity matching to confirm the results obtained by traditional multivariate analysis. Baseline demographics (Age, race, sex, income status, and insurance status) were matched using a 1:1 nearest neighbor propensity score with 0.05 caliper width n matched cohort, and a secondary multivariate regression model was built as described above (Supplementary Figure S2). All analysis was performed using Stata software version 17.0 (Stata Corporation, College Station, TX, USA). p values of less than 0.05 were considered statistically significant.
There was some variability in the geographic distribution of patients. Most patients in both cohorts were seen at urban teaching hospitals (83.9% vs. 79.6%, respectively) and had higher proportions of Medicare beneficiaries (60.5% vs. 64.0%, respectively, p = 0.02). The baseline characteristics of the study cohorts are outlined in Table 1. The baseline characteristics of the matched cohort after the propensity matching are provided in Supplementary Table S2.

In-Hospital Quality Measures and Disposition
Patients with COVID-19-associated AIS had an increased mean length of stay (10.2 days vs. 5.4 days, adjusted length of stay 3.2 days higher, p < 0.001) in comparison to AIS patients without COVID-19 infection. Patients with AIS and COVID-19 had a higher mean total hospitalization charge (145,497 USD vs. 88,320 USD, adjusted total charge 25,814 USD higher, p < 0.001). Of those who survived, patients with COVID-19-related AIS had more discharges to skilled nursing facilities (SNF), long-term acute care (LTAC) facilities, and nursing homes (48.3% vs. 40.8%, p < 0.001), as well as more discharges for home health (20.3% vs. 19.7%, p < 0.001) compared to those with AIS alone ( Table 2).
After PSM, AIS patients who were COVID-19-positive continued to have an increased mean length of stay (10.2 days vs. 5.7 days, adjusted length of stay 2.8 days higher, p < 0.001) and a higher mean total hospitalization charge (145,287 USD vs. 84,419 USD, adjusted total charge 31,400 USD higher, p < 0.001) than COVID-19-negative AIS patients (Table 4).

Discussion
In this retrospective analysis, we identified 329,240 hospitalized patients diagnosed with acute ischemic stroke between 1 January 2020 and 31 December 2020, out of which 6665 (0.4%) were diagnosed with AIS and COVID-19 infection, with a total of 1,659,040 COVID-19 hospitalizations. Our study included 322,575 AIS patients without COVID-19 during the same duration. Major findings of our study include: (1) AIS patients with COVID-19 had significantly increased in-hospital mortality compared to AIS patients who were COVID-19-negative. (2) Patients in the AIS with COVID-19 cohort required more mechanical ventilation and had significantly higher rates of acute VTE, acute MI, cardiac arrest, septic shock, and AKI. (3) AIS and COVID-19 patients had significantly higher NIHSS scores. (4) In-hospital mortality was higher among males, African Americans, Hispanics, Asians, and Native Americans with COVID-19-positive AIS than among females and Caucasians with COVID-19-negative AIS.
The prevalence of AIS with COVID-19 was reported as 0.2-2.7% in previous publications, which is consistent with our prevalence data of 0.4% [3,8,9,17]. A recent systematic review and meta-analysis by Nagraj et al. reviewed the incidence of stroke in randomized trials for COVID-19 therapeutics and included studies until 30 July 2021 [18]. This systematic review included 38,732 patients from inpatient and outpatient settings (65 patients with stroke) from 77 randomized studies, and all reported stroke events happened in the inpatient setting with an overall incidence of 0.18%. In contrast, our study included 329,240 inpatients (6665 patients with stroke) and found worse outcomes in patients with AIS and COVID-19 infection, with an overall incidence of 0.4%. They also concluded that there is a possibility of an association between the severity of the COVID-19 infection with the occurrence of stroke. Compared to prior studies, which have included smaller sample sizes [6,18], our study compares AIS outcomes in COVID-19 patients based on one of the largest cohorts of COVID-19 patients hospitalized in the United States in 2020. To our knowledge, our study is the largest study looking at the inpatient outcomes of patients hospitalized with AIS and COVID-19.
In the literature, patients with AIS who were COVID-19-positive were predominantly males, ranging from 58% to 62.3% [3,8]. Our study's results suggested that women were the most affected gender; however, when a mortality breakdown was performed, males were more likely than females to die from concomitant AIS and COVID-19. In AIS patients who were COVID-19-negative, females had the greatest predisposition toward mortality, correlating with the literature [19]. Our study contains a larger sample size, likely impacting the difference regarding which gender is most affected. We did not find any large studies investigating gender as a risk factor for mortality in COVID-19-positive AIS. Women have a slight predominance for ischemic stroke among non-COVID-19 cases, due to having a longer lifespan than males and an increased risk of AIS in senescence; however, men have become more afflicted in the setting of COVID-19 [8,19]. This increased occurrence of COVID-19 in men is consistent with the gender distribution observed in COVID-19 complications [20]. Men are more likely than women to succumb to physical stress (i.e., inflammation and infection) [8] due to the pro-inflammatory effects of testosterone, contrary to estrogen, which is protective and reduces inflammation [21].
In AIS alone, less than half of the data from the United States and the Canadian AIS randomized controlled trials include details about ethnicity and race [22]. In AIS with COVID-19 infection, racial disparities are underreported as well; however, one retrospective study found that African American patients with COVID-19 had a significantly higher likelihood of stroke (50%, p = 0.03) than their non-African American counterparts [7,23]. Another study found that Native Americans had a higher prevalence of conditions associated with COVID-19, such as cerebrovascular accidents, due to a higher predilection toward vascular risk factors (i.e., hypertension, atrial fibrillation, CAD, obesity, smoking, and alcohol abuse [24]. Our study suggests African Americans, Hispanics, Asians, and Native Americans, in descending order, are more predisposed to concurrent AIS and COVID-19. Furthermore, when mortality was divided, our results showed that these same races were also risk factors for mortality in this cohort, whereas Caucasians were more susceptible to mortality in COVID-19-negative AIS patients. We did not find any large studies investigating race as a specific risk factor for mortality, and these studies would be beneficial. These racial disparities can best be explained by socio-economic factors; however, definitive conclusions on causality for these findings cannot be drawn [25][26][27][28].
With respect to age, our patients with COVID-19-related AIS had increasing susceptibilities to develop this condition when aged between 18 and 69 years, with mean ages of 69.5 ± 15.3 and 66.2 ± 13.0 years for females and males, respectively. Other published data were similar to these findings, with a documented age range from 48.1 to 75.7 years and a mean age of 63.4 ± 13.1 years [3,8]. Moreover, AIS in the setting of COVID-19 affects more young patients than during pre-pandemic times [29]. Although our study did not find that age was a predictor of mortality, the literature shows that poor functionality and mortality are higher with increasing age in this population [29]. Older patients, 70 years and above, with only COVID-19 infection, have been shown to have mortality between 10.9% and 26.6%, possibly due to many reasons, including aging immunity, numerous comorbidities, and congregate housing [30]. Age in AIS with COVID-19 infection could benefit from further studies as the pandemic continues.
Per our data, patients with AIS and COVID-19 were significantly more obese and more likely to have diabetes mellitus, which was supported by Luo et al., among other studies, in which the prevalence of diabetes in this group ranged from 20% to 65% with a pooled prevalence of 40% [3,8]. Our cohort with AIS alone was significantly susceptible to smoking, alcohol abuse, and CAD; however, it is unclear why these risk factors would not contribute to a predisposition to COVID-19 in our study. Prospective studies would be helpful for clarifying this discussion. Although we did not have data pertaining to hyperlipidemia, and there was no significant difference found for hypertension between our cohorts, common cerebrovascular comorbidities such as hypertension and hyperlipidemia, along with diabetes, were most frequently observed throughout the literature in patients with concomitant AIS and COVID-19 infection [3,8]. Other reports indicate that 22% to 24% of the COVID-19-associated AIS population did not have any known medical history or substance use, and approximately 44% of these patients presented with at least four vascular risk factors, which likely contributed to case severity [31].
In our study, patients with both AIS and COVID-19 had significantly higher in-hospital mortality at 16.9%, which was lower than the mortality of 20% to 63.6% documented in the literature yet higher than those with COVID-19 without AIS at 13.1% [5,9,10]. When further mortality analysis was performed, which evaluated gender, race, and age, it was found that gender and race have a significant role in mortality outcomes in COVID-19positive AIS patients, as discussed above. This high mortality can be explained by the additive effect of the COVID-19 infection. COVID-19 infection leading to a cytokine storm, excessive systemic inflammatory response, coagulation dysfunction, oxidative stress, direct neurological injury, and hypoxia-induced multiorgan failure all contributed to an increase in mortality, along with the lack of effective and available therapies and vaccinations for COVID-19 during the early phase of the pandemic and the characteristics and comorbidities discussed above. Additionally, social consequences of the pandemic (i.e., the overwhelmed healthcare system or fear of contracting COVID-19 if one sought healthcare) led patients to delay care, which may have caused the progression of a mild stroke into the moderate to severe presentations observed on admission [32,33]. It is important to recognize and treat AIS early and aggressively, in order to improve morbidity and mortality [33].
Our cohort of COVID-19-associated AIS patients had significantly increased mechanical ventilation requirements, acute VTE, acute MI, cardiac arrest, septic shock, and AKI compared to the non-COVID-19 AIS cohort. These results were consistent with the literature [5,8,34]. Ganesh et al. noted that up to 38% of AIS patients with high NIHSS scores undergo intubation, and this value is presumed to be even higher when combined with the complexity of COVID-19 [5]. Patients become additionally vulnerable to AKI, and thus, mortality increases when contrasted neuroimaging is used as part of the AIS workup [5]. Regarding elevated prothrombotic markers, studies show approximately 41.7% of COVID-19-associated AIS patients test positive for antiphospholipid antibodies, and 72.6% have elevated D-dimer levels [3]. That study, among others, suggested the severity of stroke presentation and worsened outcomes correlated with prothrombotic and inflammatory markers [3,8]. Moreover, worse outcomes can be explained by the reliance on biomarkers and clinical suspicion (regardless of neurological history) when imaging was not readily available for confirmatory diagnosis or when the use of standard modalities was delayed due to the early pandemic isolation effect [32,33]. Our study is limited due to the exclusion of lab values and imaging, as NIS diagnoses are based on ICD-10 codes. Despite higher NIHSS scores and therefore increased stroke severity in this population, our COVID-19-related AIS cohort did not require significant use of mechanical thrombectomy or thrombolysis. Irrespective of our results, there is accumulating evidence of negative outcomes (over a 9-fold increase in mortality) in AIS patients with COVID-19 who receive mechanical thrombectomy regardless of the timing and success of the intervention, necessitating risk-benefit discussions with these patients [35].
With respect to NIHSS scores, our cohort of AIS patients with concurrent COVID-19 infection had significantly higher scores. This correlated with the literature in that patients with AIS with COVID-19 had moderate-to-severe stroke severity on admission compared to non-COVID-19 cohorts, which may explain the poor outcomes, as analyzed above [6,36]. In a systematic review and meta-summary by Tan et al., the mean NIHSS score was 19 ± 8, consistent with LVOs observed in 40.9% of AIS with COVID-19 patients [3], and was also proportional to increased mortality [9]. A limitation of our study was the exclusion of stroke subtypes within our cohort data.
In our study, the co-diagnosis of AIS and COVID-19 resulted in a significantly increased length of stay, an increased mean total hospitalization charge, and more discharges to a facility or home services for continued care than non-COVID-19 AIS patients. These results were predictable given that patients with AIS and COVID-19 had significant complications and required higher levels of support, as discussed above.
AIS treatment in the general, non-COVID-19 population primarily includes blood pressure control, anticoagulation, long-term dual antiplatelet agents and statin therapy, potential thrombolysis using recombinant tissue plasminogen activator (rt-PA; i.e., alteplase) and mechanical thrombectomy depending on the case and if the patient is within the therapeutic window of 3 to 4.5 h or within 6 h, respectively [3,12,37,38]. The mainstay of management for COVID-19-associated AIS is the aforementioned stroke treatment in combination with supportive care and treatment of the underlying COVID-19 infection [10]. COVID-19 patients typically receive steroids, antivirals, and comfort medications, all of which have since shown benefit throughout the pandemic [39]. However, it remains uncertain if these interventions directly benefit AIS. Further evaluation is necessary to identify possible benefits and effects of antivirals (i.e., remdesivir) on mitigating adverse vascular events [40], given that many articles did not investigate stroke specifically [41]. Acute ischemic stroke recurrence, defined as new ischemic lesions within one week to 90 days after a clinically symptomatic stroke, has a 2% recurrence rate, with LVOs being the most common recurrence subtype [42][43][44]. It is controversial if any of the aforementioned interventions help prevent or reduce recurrence, especially when the original presentation occurs during the same admission as the patient's COVID-19 illness, and our study lacks data pertaining to the intensity and efficacy of certain treatment modalities (i.e., antiplatelet medications) that could play a role in prognosis [45]. Further research is necessary to identify if long COVID factors affect the incidence of recurrent AIS given the reports of persistent neurological symptoms, such as headaches and fatigue, months after COVID-19 infection [5,46]. Our study is limited in that we could not study long COVID as the 2020 NIS data does not include the 2021 ICD-10 code for this diagnosis.
There have been concerns surrounding COVID-19 vaccination with DNA and mRNA vaccines and the potential development of complications, such as acute ischemic stroke [47,48]. In Kolahchi et al., reportedly 22 cases (51.1% of that study's cohort) developed AIS associated with the vaccine-induced immune thrombotic thrombocytopenia (VITT; thrombocytopenia and antibodies against platelet factor 4 leading to thrombosis) four weeks after all, except one patient, received a viral vector vaccine out of > 1.3 billion (< 0.000%) COVID-19 administered vaccinations from December, 2020, to December, 2021 [47,49]. In contrast, Stamenkovic et al. found no connection between vaccination and stroke [48]. Based on these publications, the benefits outweigh the risks of vaccination (i.e., decreased severity of COVID-19 infection, hospitalization, mortality, etc.) [50]; however, supplementary literature to determine if vaccination reduces COVID-19-related AIS specifically is welcomed.
The 2020 NIS database has provided new data that can be used to further investigate COVID-19-related AIS. Our study supports the important learning lesson that AIS with COVID leads to worse clinical outcomes and that COVID-19 infection itself can increase the risk of AIS. This information emphasizes the importance of closer inpatient monitoring and outpatient follow-up with neurology for those who develop neurologic complications due to COVID-19 illness. With this, there can be further research into the management of COVID-19-related AIS in order to improve mortality and other outcomes. This evidence also promotes and encourages vaccination.

Limitations
Our study data were obtained from the NIS, which may result in selection bias. This is a collection from a United States in-hospital database, and it lacks data in the out-patient, nursing home, and global settings. The diagnosis of acute ischemic stroke and COVID-19 were based on ICD-10 codes because NIS data lacked lab values and imaging, which may make diagnoses prone to error. Our larger sample size helps reduce the effects of these errors. The study cohorts included non-vaccinated individuals, as the FDA first approved vaccination against COVID under the EUA on 11 December 2020. It is probable that COVID-19 vaccination may alter the outcomes of acute ischemic stroke between vaccinated and unvaccinated patients.

Future Directions
The NIS 2020 database primarily includes non-vaccinated patients, as the first vaccine against COVID-19 was made available in December 2020. A recent study by Kim et al. noted that vaccination against COVID-19 is associated with reduced stroke risk in patients with COVID-19 infection [51]. Given the high mortality reported in AIS with COVID-19 infection in our study, it may serve as a valuable reference to address vaccine hesitancy in unvaccinated individuals t. In contrast, some studies have also suggested an increased risk of stroke after COVID-19 vaccination [52,53]. We could not study the effect of vaccination on AIS outcomes with COVID-19 infection. Long COVID may be associated with an increased risk of AIS stroke [54]. Conversely, patients with the COVID-19 infection and AIS may have an altered risk of different long-term outcomes from the stroke, e.g., recurrent strokes and bleeding. More studies are needed to improve understanding of stroke outcomes after a COVID-19 infection and long COVID.
Our study supports the important learning lesson that AIS with COVID leads to worse clinical outcomes and that COVID-19 infection itself can increase the risk of AIS. These results are similar to other diseases showing worse outcomes with COVID-19 infection, e.g., STEMI [51]. Successful stroke care requires complex, timely collaboration between multiple clinical teams, and factors like delay in access during the early COVID-19 pandemic may have played a role in worse outcomes [55]. A better understanding of these factors can guide the development of tools and policies to better prepare for future pandemics [55].

Conclusions
Our study found worse outcomes among patients with acute ischemic stroke with COVID-19 infection in terms of mortality and morbidity, including an increase in NIHSS scores, acute venous thromboembolism, acute myocardial infarction, cardiac arrest, septic shock, acute kidney injury, and mechanical ventilation compared with patients without COVID-19 infection. Our study primarily included non-vaccinated patients who may be at risk of worse outcomes with COVID-19 infection. These results highlight the need for more research on the role of vaccination and treatment options to improve acute ischemic stroke-related outcomes in COVID-19-positive patients.

Institutional Review Board Statement:
The NIS database used in this study contains de-identified information from billing and diagnostic coding from the participating hospitals, does not involve 'human subjects' directly, and is exempt from Institutional Review Board approval.

Informed Consent Statement: Not applicable.
Data Availability Statement: Restrictions apply to the availability of these data. Data were obtained from the National Inpatient Sample database, US.

Conflicts of Interest:
The authors declare no conflict of interest.