1. Introduction
The coronavirus Disease 2019 (COVID-19) pandemic put the global healthcare system to the test. Many countries, including Poland, were unable to provide patients with adequate care due to the insufficient number of hospital beds for patients requiring isolation during treatment of COVID-19. This situation pushed governments to create places such as temporary hospitals, mainly by rearranging existing wards but also by creating new hospitals. The latter was the case with a Temporary Hospital in Szczecin, Poland.
After two years of the pandemic, we already know that most COVID-19 cases are mild or asymptomatic, and many patients do not require hospital care. From the group of patients presenting with mild or moderate symptoms it was essential to identify those with a higher risk of rapid deterioration of respiratory function and probability of death and therefore who might require hospitalization and more intensive treatment. Being at the front line of the COVID-19 war, we wanted to find a simple tool to quickly identify patients with a poor clinical prognosis.
Numerous studies have suggested different risk factors of increased mortality and necessity for invasive ventilation in COVID-19 such as: older age, male sex, comorbidities or multimorbidity, high respiratory rate, radiographic severity score, elevated neutrophile count, higher C-reactive protein and creatinine concentration, low albumin, and blood oxygen saturation below 93% [
1,
2].
Frailty is a state of reduced ability to recover from a stressful event, which appears with older age, and is associated with increased risk of adverse outcomes and death. One of the most popular tools to grade frailty is the Clinical Frailty Scale (CFS), originally created in 2005 as a 7-point scale [
3] and later modified and expanded to a 9-point scale [
4]. CFS positions a patient’s frailty from very fit (CFS 1) to terminally ill (CFS 9), i.e., someone whose life expectancy is below six months.
The National Institute for Health and Care Excellence (NICE) provided guidelines, which recommend that patients admitted to hospital with COVID-19 should be assessed with the Clinical Frailty Scale (CFS) to evaluate baseline health and to consider treatment expectations [
5]. According to these guidelines, all patients admitted to the Temporary COVID-19 Hospital in Szczecin were assessed with CFS to predict health outcomes. Based on gathered information, our study aimed to confirm a high the CFS score as an independent risk factor of death for patients suffering from COVID-19, who required hospital treatment. Our secondary purpose was to analyze treatment strategies and possible complications according to CFS score.
3. Results
A total of 201 patients were included in the study: 129 patients (64.18%) without frailty (CFS 1–3), 46 patients (22.89%) with mild or moderate frailty (CFS 4–6) and 26 patients (12.94%) with severe frailty to terminally ill (CFS 7–9).
Table 1 presents demographic data, addictions, and co-morbidities. Patients from higher CFS group were older (
p < 0.001) and had more comorbidities (hypertension (
p < 0.001), chronic heart failure (
p < 0.001), atrial fibrillation (
p < 0.001), previous ischemic stroke (
p < 0.001), internal carotid artery stenosis (
p = 0.004), chronic peripheral ischemia (
p = 0.003), previous venous thrombosis (
p < 0.001) than patients with low CFS score. We decided to calculate the Charlson Comorbidity Index (CCI) to sum up comorbidities and predict 10-year survival. Patients with higher CFS also had higher CCI (
p < 0.001). Diabetes treated with insulin (
p < 0.001), chronic kidney diseases (CKD) (
p = 0.004) and chronic coronary syndrome (CCS) (
p < 0.001) with myocardial infarction (MI) (
p = 0.005) were most frequent in the CFS 4–6 group. Patients without frailty (CFS 1–3) most frequently smoked cigarettes (
p = 0.011).
Medications taken by the patients before admission are shown in
Table 2. Patients with CFS (7–9) more often were treated with acetylsalicylic acid (ASA) (
p < 0.016), new oral anticoagulants/oral anticoagulants (NOAC/OAC) (
p = 0.003), calcium channel blockers (CCBs) (
p = 0.049), statins/fibrates (
p = 0.006), nitrates (
p = 0.016), diuretics (
p = 0.022), opioids (
p = 0.024). Insulin (
p < 0.001) and immunosuppression (
p = 0.025) were more widespread in the CFS 4–6 group. Bronchodilators were taken most often by patients without frailty (
p = 0.05), less often in the mild to moderate frailty group and finally these drugs were not used before admission by the frailest patients.
Table 3 shows coronavirus-related symptoms on admission to the hospital. Reported symptoms differed among groups. Patients with the highest CFS more often suffered from nausea (
p = 0.003) and vomiting (
p = 0.010). Interestingly, these symptoms were not registered in the moderate frailty group. Patients with CFS 1–3 more often complained of cough (
p = 0.011), lack of taste (
p < 0.001), and lack of smell (
p = 0.042).
Laboratory results on admission are shown in
Table 4. Most results were worse in the severe frailty group (CFS 7–9). Only INR, CKMB and TnT were the highest in the mild to moderate frailty group (CFS 4–6). APRI (AST to platelet ratio index - determining liver function) was equally high in CFS 1–3 and 7–9 groups.
As presented in
Table 5, there were significant differences in respiratory parameters and type of oxygen therapy applied on admission between groups. Patients with CFS 4–6 and CFS 7–9 required more intense oxygen supplementation with non-rebreather masks (
p = 0.009) and HFNOT or NIV (
p = 0.018) with higher oxygen flows (
p = 0.009), and fraction of inspired oxygen (FiO
2) in the inhaled mixture of gases (
p = 0.006). Furthermore, the oxygenation ratio (partial pressure of oxygen/fraction of inspired oxygen (PaO
2/FiO
2) was higher in more frail patients (
p = 0.017). Acute respiratory distress syndrome (ARDS) occurred predominantly in patients without frailty, but the severity of ARDS increased along with CFS score.
Table 6 presents drugs applied during hospitalization in each CFS group. Patients with severe frailty more often needed therapeutic doses of low molecular weight heparin (LMWH) (
p = 0.011) and other than standard antibiotic therapy (
p < 0.001), most likely because of more frequent bacterial coinfection. Prednisone was most frequently given in the mild to moderate frailty group (
p = 0.020). We usually used dexamethasone to decrease immunological response to COVID-19 infection, but we continued with prednisone if it was taken beforehand due to other chronic disease. We also observed significant differences in remdesivir administration: 24.03% of patients in the CFS 1–3 group were treated with remdesivir compared to 3.85% in the CFS 7–9 group (
p = 0.007).
Table 7 compares registered complications. Frailty was associated with greater risk of cardiological complications (
p < 0.001) (heart failure (
p = 0.008), atrial fibrillation (
p < 0.001) renal complications (
p < 0.001) (acute kidney injury (AKI) or decompensation of CKD (
p < 0.001), urinary tract infection (
p < 0.001)), neurological complications (
p < 0.001) with impaired consciousness (
p < 0.001) and other complications (
p < 0.001) like pressure ulcers (
p < 0.001), gastrointestinal hemorrhage (
p = 0.003), or sepsis (
p < 0.001). Only respiratory failure (
p = 0.004) and mucosal bleeding (
p = 0.021) were more frequent in mildly frail patients. Finally, statistical analysis confirmed increased risk of death among frail patients, while most patients (83.72%) with CFS 1–3 were discharged home (
p < 0.001).
The ROC curves for age, continuous CCI, CFS, and PaO
2/FiO
2 are presented in
Figure 1. The ROC analysis showed that the CFS is better at predicting the mortality of patients with COVID-19. The data in
Table 8 show that CFS has the highest area under the ROC curve (AUC), 0.844, with
p < 0.001.
An assessment of survival probability in each CFS group was performed using the Kaplan–Meier curve (
Figure 2). It shows statistically significant differences in 30-day survival between all three groups of patients (
p < 0.001).
A detailed assessment of mortality was performed based on the CFS scale (score 1–9). Logistic regression was performed (
Figure 3). The analysis showed a relationship between the occurrence of death and an increase in the CFS score (OR = 1.978,
p < 0.001). After the results were corrected for data demographics and comorbidities, it was confirmed that mortality was associated with an increase in CFS (OR = 1.89,
p < 0.001).
4. Discussion
Since the emergence of SARS-CoV-2 infection and the declaration of COVID-19 pandemic many patients requiring hospitalization overwhelmed medical systems across the world and therefore it soon became crucial to find a simple tool to identify patients with increased risk of death. The purpose of this study was to evaluate the relationship between frailty on admission to a hospital dedicated to treating patients infected with SARS-CoV-2 with certain outcomes and mortality. The results show that frailty is associated with increased risk of death during hospitalization.
We identified mild and moderate frailty in 22.8% of patients and severe frailty in 13% of patients. Prevalence of frailty in our cohort was lower than reported by other authors, who estimate the prevalence of frailty among hospitalized COVID-19 patients to be between 51% [
9] and 54% [
10] of patients.
There may be several reasons why our number of frail patients was lower. First, our ward was a temporary solution to the increased needs of hospitalization of infectious patients, with temporary staff, both nurses and doctors, who work permanently on several other units. Therefore, we hospitalized mostly patients, whose major problem was COVID-19 infection. Those with other severe diseases were often admitted to other hospitals in the area that were also taking care of COVID-19 patients. Second, doctors working in our units do not routinely assess CFS in their clinical practice. A recent study showed that the correlation between CFS scoring by untrained raters and geriatricians may be improved by using a new tool called classification tree. In that study, inexperienced clerks or residents tended to underestimate CFS score compared to geriatricians [
11]. Classification trees may guide a rater through the scoring process. This study was published after data collection in our research. Despite the lack of experience and the possibility of underestimating CFS in our study, it still predicted a poor outcome for COVID-19, which proves the utility of this scale. Third, we included all hospitalized patients, not only those aged 65 and more, for whom CSF was created. The use of CFS in younger patients with COVID-19 has already been discussed in several studies [
12]. It is possible to underrate young patients suffering from severe diseases or overrate patients with movement disabilities in this group. A high CFS should not be the only admission criterion, but our study showed it is an independent risk factor for death, even when corrected for age, confirming its universality.
Increased risk of complications and death during COVID-19 infection among frail patients has already been reported in several studies. The largest study included 2434 patients from 63 hospitals in 11 countries in Europe [
12]. In that study, the risk of in-hospital mortality was significantly higher both in the pre-frail group (defined as CFS 4–5) and the frail group (CFS 6–9) compared the fit group (CFS 1–3) with OR (odds ratio) of 1.54 (95% CI (confidence interval) of 1.16–2.06) and 2.71 (95% CI 2.04–3.6) respectively. In another large analysis of 1376 patients (age 70+) from 15 hospitals in the Netherlands, frailty (CFS 4–5) was associated with a 2-times higher risk of in-hospital mortality (OR 2.0) and severe frailty (CFS 6–9) was associated with almost a 3-times higher risk of death (OR 2.8) [
13]. A large meta-analysis of fifteen studies comprising of 23,944 COVID-19 positive patients confirmed that frailty is an independent predictor of mortality among COVID-19 patients [
9]. In our study, we also showed a relationship between death and an increase in CFS score (OR 1.98).
The CFS is a tool designed to evaluate physical impairments based on limitations patients experience in their everyday life. It does not consider comorbidities, which may influence survival rates. In our study, frail patients were more likely to have cardiological disorders, diabetes, and CKD, which are known to increase COVID-19 mortality. CCS, HT, and CHF are responsible for approximately a 2-times higher risk of death (RR 2.25 (95% CI 1.6–3.17); RR 1.81 (95% CI 1.43–2.32) and RR 2.03 (95% CI 1.28–3.29) respectively) [
14] and the history of CVD is associated with 3-fold increased risk of severe COVID-19 infection and over 11-fold higher risk of all-cause mortality [
15]. Mortality among diabetic patients is 2.39-times higher and the risk of severe course of COVID-19 is 1.45-times higher [
16]. Mortality among patients with CKD is 3.25 times higher [
14].
The Charlson Comorbidity Index (CCI) summarizes comorbidities and estimates 10-year survival. The CCI includes several diseases, such as CVD, COPD, diabetes, liver disease, tumors, and others. In our study, frail patients had significantly higher CCI, which is consistent with higher burden of comorbidities found in these groups. The calculated CCI was nearly 6 in severely frail patients, in the moderately frail nearly 5 and in the fit nearly 3 points, which makes an estimated 10-year survival of 2%, 21%, and 77% respectively. The usefulness of CCI in predicting poor outcome of COVID-19 has been evaluated in several studies. The risk of mortality increases by 16% for each increase in CCI [
17] and even CCI > 0 is associated with increased risk of severe COVID-19 and death [
17,
18].
We corrected the risk of death for comorbidities, age, sex, and BMI and found that a high CFS score is still significant, independent risk factor of death in patients hospitalized with COVID-19 with OR 1.89 (CI 1.35–2.63, p < 0.001)
The negative effect of renin-angiotensin-aldosterone drugs (RAAS) on COVID-19 outcome reported in several studies at the beginning of the pandemic brough attention to a possible relationship between pretreatment with some drug groups and COVID-19 outcome. Therefore, we analyzed what drugs were used by our patients before admission to the hospital (
Table 2). Frailer patients had more comorbidities; they used more drugs and were more prone to possible side-effects. Studies on this subject confirmed safety of chronic treatment, which should be continued whenever possible during hospitalization. Large meta-analyses have shown that neither RAAS, nor other antihypertensive drugs, nor NSAIDs are associated with increased risk of death. The reason for previous observations was not the influence of drugs on COVID-19, but it was the effect of underlying comorbidities alone [
19,
20]. Moreover, statins, due to their anti-inflammatory effect, are associated with lower risk of fatal COVID-19 [
21,
22,
23] among non-ICU patients and lower risk of mechanical ventilation treatment [
22].
The primary symptoms of COVID-19 are well known and include fever, cough, and dyspnea. On the other hand, there are some atypical manifestations, which occur mainly in elderly, frail patients, such as hypotension, low body temperature, functional decline, and acute mental change [
24]. In our study, patients with CFS 1–3 presented mainly olfactory and taste disorders as well as cough. We observed that severely frail patients (CFS 7–9) more frequently complained of gastrointestinal symptoms, such as nausea and vomiting, which may be associated with longer hospitalization and severity of infection [
25,
26]. This phenomenon can be explained by the viral load and replication within the digestive tract [
25]. SARS-CoV2 can invade the human cells by connecting to the angiotensin converting enzyme 2 (ACE-2) receptors, which are localized not only in the alveolar cells in the lung but also in the gastrointestinal tract [
26]. Doctors must be aware of atypical symptoms of COVID-19 among older and frail patients.
The usefulness of some biochemical markers in staging risk in COVID-19 has been studied in many recent analyses. Many investigations focused on LDH and lymphocyte count changes as an independent risk factor of mortality [
27,
28,
29]. Lower lymphocyte counts, higher leukocyte counts, and higher neutrophil-lymphocyte ratio (NLR) were observed in severe cases [
28,
30,
31]. In our study, we did not find a statistically important difference between frailty and LDH level and lymphocyte count, but severely frail patients (CFS 7–9) had higher: WBC, neutrophil count, NLR (neutrophil—leukocyte ratio), D-dimer, and APRI (AST platelet ratio index) and lower HGB and HCT, which is consistent with previous studies [
24]. CFS 7–9 was also associated with elevated renal biomarkers (creatinine, urea) and lower GFR.
Patients hospitalized in the Temporary COVID-19 Hospital in Szczecin were treated according to the Polish Society of Epidemiologists and Infectious Disease Doctors guidelines. Treatment strategy includes corticosteroids (patients requiring oxygen therapy due to respiratory failure, most frequently dexamethasone), LMWH (in prophylactic, intermediate, or therapeutic doses, depending on patient’s clinical condition), antibiotics (in case of suspicion of bacterial superinfection; most frequently ceftriaxone, levofloxacin or azithromycin), and remdesivir [
32]. Tocilizumab and convalescent plasma were not available in our department. We found that patients with higher frailty stages were treated differently than those less frail. To the best of our knowledge, this is the first analysis that compares COVID-19 treatment depending on CFS score.
Remdesivir is an antiviral agent which can inhibit SARS-CoV-2 replication in vitro [
33]. Beigel et al. proved in their randomized, double-blind study that patients who were treated with remdesivir had a shorter recovery and discharge time and had lower mortality in comparison to placebo group [
34]. Remdesivir was applied more frequently in the CFS 1–3 group compared to the CFS 7–9 group. The difference in results from elevated biochemical markers of liver and renal dysfunction (alanine aminotransferase > 5-times upper limit of normal and decreased GFR < 30 mL/min) [
35] that we observed among frail patients, are contraindications to remdesivir treatment.
The other medications which have proven activity against COVID-19 are glucocorticosteroids, especially dexamethasone. Many recent studies have shown that dexamethasone could reduce mortality [
36,
37]. In our department, all patients with respiratory failure (defined as SpO
2 < 94% without oxygen supplementation) received dexamethasone.
It has been proven that anticoagulant therapy with LMWH is associated with a better prognosis and lower mortality in COVID-19 patients [
38]. There are a few mechanisms that lead to prothrombotic state in COVID-19, for example: cytokine storm, pathological complement-activation, endothelial dysfunction caused by hypoxia and virus-related damage which leads to vasoconstriction, platelet activation and aggregation [
39]. In our cohort, patients with CFS 7–9 more frequently received LMWH in therapeutic doses due to previously diagnosed atrial fibrillation, deep vein thrombosis, or higher D-dimer levels. We did not observe any differences in occurrence of thromboembolic complications between groups.
We report differences between complications found in patients depending on CFS score. Patients with frailty had higher risk of mucosal bleeding, respiratory failure, and heart failure. Patients with severe frailty suffered from renal, neurological, and infectious complications as well as arrhythmias. These results are similar to those reported by other authors, who mention acute renal failure, delirium and ARDS [
40] as the most common. Some of the complications have prognostic significance. Hagg et al. noticed that AKI was an additional risk factor for mortality in the group of patients with CFS > 5 [
41].
The severity of COVID-19 may be assessed by the intensity of oxygen therapy applied on admission. In our study, patients with CFS 4–6 and CFS 7–9 more often required oxygen supplementation via non-rebreather mask and HFNOT with more intense oxygen flow and oxygen concentration. Those patients had a higher probability of developing severe ARDS. Although this was associated with higher mortality, the qualification and admission rate to ICU was similar in all groups. In a study by Andres-Esteban et al study patients with frailty, defined as CFS ≥ 5 was less often admitted to the ICU than non-frail (CFS 1–3) and pre-frail (CFS = 4) patients [
40]. In the COMET study the incidence of ICU admission was higher in the CFS 6–9 group, but lower in CFS 5–6 group compared to patients without frailty. The difference was even more visible in patients younger than 65 years (in the ≥65 years old subgroup there was no statistically significant difference) [
12]. Usually, severe frailty may be a contradiction to ICU admission, especially in case of limited resources in the COVID-19 pandemic. In our unit, all patients requiring intensive oxygen therapy were consulted by an ICU-doctor and the decision about admitting them to ICU in case of deterioration was made in advance. Older age, severe comorbidities and severe frailty were the most common reasons for disqualifying patients from ICU treatment.
Limitations
Our study comes with some limitations. First, the retrospective nature of the study should be noticed. The database was collected from 3 up to 6 months ago. During this time, some changes have been implemented to the general guidelines of COVID-19 treatment (i.e., corticosteroids or antibiotics dosing), thus its potential impact on the final prognosis in particular individuals might remain considerable. Second, we reported only short-term outcomes and did not do any long-term follow-up. Finally, doctors working in our unit were mostly internal medicine and cardiology residents, who do not routinely assess CFS in their clinical practice. As laboratory tests are irrefutable, the approach to assessing CFS may differ among different specialists. Additionally, CFS is only validated for the elderly patients (over 65), and it may not be suitable for assessment of younger people or patients with a learning or progressive disability who were included in the study.