Introduction

Acute cholecystitis (AC) is a common cause of emergency hospital admission that should be managed according to international guidelines [1, 2]. It can be classified into 3 grades of severity (mild, moderate, and severe). These grades affect the length of hospital stay, conversion to open surgery, medical costs, and prognosis [1]. Gangrenous cholecystitis (GC) is a severe form of AC. It occurs in approximately 15% of the patients (range 2–30%) and is associated with an increased risk of postoperative morbidity and mortality [3, 4]. During the COVID-19 pandemic, we observed an increased number of AC patients who presented with gangrenous acute cholecystitis. An early case series showed that COVID-19 infection and pneumonia were associated with GC with increased morbidity and mortality, mainly in elderly and frail patients [5,6,7,8,9].

GC requires prompt surgical management to reduce hospital stay and improve the clinical outcome. Several retrospective studies focused on the management of AC patients in the first period of COVID-19 pandemic. They reported increased non-operative management (NOM) in those patients. This was associated with increased conservative management failure, morbidity, and length of hospital stay (LOS). This was attributed to the limited access to the operating theaters in attempt to reduce the in-hospital spreading of the virus. Age, COVID-19 infection, AC severity, and NOM failure contributed to the increased death rate [10]. The aim of the ChoCO-W global prospective study is to compare the clinical course, biological and radiological findings, and clinical outcome of AC in patients who have COVID-19 disease with those who do not have it.

Patients and methods

Ethical considerations

Ethical committee approval was obtained from the CPP Sud-Méditerranée 3, University Hospital of Nîmes-France (2021.03.05 ter _ 21.01.16.09406). The ChoCO-W prospective study met and followed the standards outlined in the World Medical Association Declaration of Helsinki [11]. It did not change or modify the usual clinical practices of the participating acute care surgeons.

Study protocol

The ChoCO-W study was registered in the ClinicalTrials.gov (ID: NCT04542312). The details of the protocol were published [12]. This study was conceived and designed to run over 12 months (October 2020–October 2021). It is a global collaborative, prospective cohort study, including consecutive adult patients admitted to emergency departments with AC who were screened for SARS-CoV-2 using quantitative reverse transcription polymerase chain reaction (RT-PCR) swab test. The recruitment period was for 6 months (October 1, 2020, to April 30, 2021) with 1 month of postoperative follow-up. Two hundred and eighteen ChoCO collaborating centers joined the project and participated in the study. Each international center constituted a ChoCO team (1 local investigator and 2 collaborators) which was linked to an ID number for entering data anonymously in a secured web database. All local investigators were responsible of patients recruitment, data collection, and research ethical issues according to their local standards. All ChoCO collaborators who collected and entered the data were included in the ChoCO-collaborative authorship. The prospectively collected data were reported according to the STROBE guidelines [13].

Patients

A total of 2893, with a mean age of 61.3 years (SD 17.3), were prospectively included in the study. A total of 1481 (51%) patients were male. Three hundred and one patients did not have RT-PCR swab test for COVID-19 infection, or their results were non-conclusive, and they were excluded from the analysis. Out of the remaining 2592 patients with known PCR test result, 180 (6.9%) were proven to be COVID-19 positive and 2412 (93.1%) were COVID-19 negative. These two groups were compared. Concerning SARS-CoV-2 type, multiple variants emerged in the fall of 2020 and the most circulating in the recruitment period of the ChoCO-W study was the Delta variant (B.1.617.2), isolated firstly in India in October 2020. This variant showed higher virulence compared with wild-type SARS-CoV-2 [https://www.who.int/activities/tracking-SARS-CoV-2-variants#cms].

Study variables

Demography, clinical, laboratory, radiological, surgical, microbiological, and histopathological data were prospectively collected. These included gender, age, details of clinical presentation, preoperative diagnosis, radiological workup, markers of inflammation, surgical procedures, critical care support, complications, need for surgery, histopathological findings, hospital stay, and clinical outcomes. Clinical severity of the disease was assessed with the qSOFA score [14], PIPAS severity score [15], WSES sepsis severity score [16], while the severity of AC was assessed with the Tokyo severity classification [1]. Postoperative complications were reported according to the Clavien-Dindo classification [17].

Statistical analysis

Data were downloaded from the web database to Microsoft Excel (Microsoft Office 365, USA). Data were imported to an SPSS program, sorted, cleaned, and recoded as numbers. Missing data were not imputed, and the analysis was performed on all available data.

Patients were divided into 2 groups according to COVID-19 infection: non-COVID group and COVID group.

Data are presented as number (%) for categorical data, median (range) for ordinal data, and mean (SD) for continuous data. Data were presented as both median (range) and mean (SD) when there was statistically significant difference in the ranks which did not show in the median (range) numbers. This was meant for clarification as some may not appreciate the significant difference between the two groups despite having the same median (range). The reported valid percentages were calculated from the available data and not as percentage of the study population.

Nonparametric methods were used for the analysis as they are more protective and demanding than parametric methods; moreover, nonparametric methods can be used for small numbers and do not need a normal distribution. Fisher’s exact test was used to compare categorical data of independent groups, while Mann–Whitney U test was used to compare the ordinal or continuous data of two independent groups. A p value of less than 0.05 was accepted as significant.

Results

There were 180 patients in the COVID group and 2412 patients in the non-COVID group. Demography of the patients is shown in Table 1. There was no statistical difference of age and gender between the two groups. The rate of concomitant preexisting diseases including cardiovascular diseases (p < 0.0001), diabetes (p < 0.0001), and severe chronic obstructive airway disease (p = 0.005) was significantly higher in the COVID group. Markers of sepsis severity including ARDS (p < 0.0001), PIPAS score (p < 0.0001), WSES sepsis score (p < 0.0001), qSOFA (p < 0.0001), and Tokyo classification of severity of AC (p < 0.0001) were significantly higher in the COVID group (Table1 and Fig. 1).

Table 1 Epidemiological and clinical features of the ChoCO-w population study
Full size table
Fig. 1
figure 1

Box-and-whiskers plot of severity markers WSES score (A), PIPAS score (B), and qSOFA score (C), comparing the COVID and the non-COVID patients who were globally treated for acute cholecystitis in 42 countries from 234 centers over the period October 2020–April 2021. The box resembles the 25th percentile and the 75th percentile interquartile range (IQR), while the line within the box resembles the median. p value = Mann–Whitney U test

Full size image

Patients who had COVID-19 had significantly higher complications (32.2% compared with 11.7%, p < 0.0001), longer mean hospital stay (13.21 compared with 6.51 days, p < 0.0001), and higher mortality (13.4% compared with 1.7%, p < 0.0001) compared with non-COVID patients.

Table 2 shows the clinical presentation of the two groups. COVID patients had significantly more generalized abdominal pain compared with non-COVID patients (20.1% compared with 12.4%, p < 0.0001). The COVID group had also significantly higher mean (SD) core body temperature [(37.32 (0.92)°C compared with 36.87 °C (0.81) °C, p < 0.0001)], heart rate [(89.7 (14.8) bpm compared with 84.3 (16.6) bpm, p < 0.0001], lower systolic blood pressure [(124 (23.4) mmHg compared with 131.5 (23.4) mmHg, p < 0.0001], higher respiratory rate [(19.3 (3.73) breaths/min compared with 17.1 (3.25) breaths/min, p < 0.0001], lower SpO2 [(94% (80–100) compared with 97% (97–100), p < 0.0001), and higher incidence of shock (11.2% compared with 3.5%). There was no statistical difference in the modality of preoperative diagnosis between the two groups.

Table 2 Clinical findings in COVID and non-COVID patients
Full size table

Table 3 compares the laboratory tests results between the two groups. The mean white blood cell count and CRP were significantly higher in the COVID group [(8156 (8266)/mm3 compared with 7501 (18 690)/mm3 and 89.44 (98.3) mg/L compared with 80.15 (102.5); p = 0.04 and 0.002, respectively]. The most striking significant differences were in the total bilirubin and conjugated bilirubin which were almost doubled in the COVID group [9.07 (19.99) mg/dL compared with 5.38 (26.24) mg/dL and 5.38 (15.89) mg/dL compared with 2.31 (8.14), < 0.0001 in both]. Although there was statistical significance in the mean value of AST and ALT, the difference did not seem to impact on clinical features and outcomes. D-dimer was significantly higher, and arterial lactates were significantly lower in the COVID group [(858.5 (2382) nmol/L compared with 456.8 (1644); p = 0.02)] and [(3.52 (12.73) mmol/L compared with 16.96 (79), p = 0.03, respectively]. APTT time was significantly longer in the COVID patients [(31.52 (8.94) sec compared with 26.39 (11.54); p < 0.0001)].

Table 3 Laboratory tests results in COVID and non-COVID patients
Full size table

The difference in mean value of INR in COVID and non-COVID groups [1.24 (SD 4.1) versus 1.4 (SD 0.71)] was not statistically significant (p = 0.017).

The management of patients admitted in ED with AC during the COVID-19 pandemic, without distinction of positivity to RT-PCR swab test for COVID infection, is shown in Table 4.

Table 4 Management of patients admitted with acute cholecystitis during the COVID-19 pandemic, without distinction of RT-PCR swab test for COVID infection result
Full size table

Table 5 compares the management between the COVID and non-COVID groups. There was highly significant difference in the surgical management between the two groups, p < 0.0001. Laparoscopic total cholecystectomy was performed less frequently in the COVID group (58.1% compared with 76.6%; p < 0.0001), while open total cholecystectomy was significantly higher in the COVID group (22.5% compared with 6.7%; p < 0.0001). Open total cholecystectomy after conversion was significantly decreased in the COVID group (0.7% compared with 5.4%; p < 0.0001). Reoperation was significantly higher in the COVID group (14.6% compared with 2.6%; p = 0.011).

Table 5 In-hospital management of ChoCO patients: comparison between COVID and non-COVID patients
Full size table

COVID patients needed significantly more mechanical ventilatory support (16.8% compared with 2.8%, p < 0.0001) and parenteral nutrition support (22.2% compared with 6.1%, p < 0.0001).

The COVID group had significantly higher postoperative complications compared with the non-COVID group (32% compared with 11%, respectively, p < 0.0001), including SSI, pulmonary infections, bleeding, and biliary generalized peritonitis (Tables 1, 2, 3, 4, 5 and 6). The Clavien-Dindo complication score was significantly higher in the COVID group [median (range) 2 (1–4) compared with 1 (1–4), p < 0.0001, Fig. 2]. The incidence of diffuse biliary peritonitis, biliary fistula, and common bile duct injury was 2.7% (5/180), 1.1% (2/180), and 0.6% (1/180), respectively, in the COVID group.

Table 6 Postoperative complications in the COVID and non-COVID-19 patients
Full size table
Fig. 2
figure 2

Box-and-whiskers plot of Clavien-Dindo postoperative complication classification comparing the COVID and the non-COVID patients. The box resembles the 25th percentile and the 75th percentile interquartile range (IQR), while the line within the box resembles the median. p value = Mann–Whitney U test

Full size image

Mortality rate was 13.4% (24/180) in the COVID group and 1.7% (40/2412) in non-COVID group (p < 0.0001).

The detailed postoperative complications of the two groups are shown in Table 6.

Table 7 shows the histopathological results in non-COVID and COVID groups. A statistical difference was shown between the two groups (p < 0.0001). The incidence of GC was doubled in the COVID group compared with the non-COVID group (40.7% compared with 22.3%). Gallbladder wall was significantly thicker in the COVID group [6.32 (2.44) mm compared with 5.4 (3.45) mm; p < 0.0001] (Fig. 3).

Table 7 Histopathologic findings in COVID and non-COVID patients
Full size table
Fig. 3
figure 3

Box-and-whiskers plot of gall bladder wall thickness (mm) in the COVID and the non-COVID patients who had total or partial cholecystectomy. The box resembles the 25th percentile and the 75th percentile interquartile range (IQR), while the line within the box resembles the median. p value = Mann–Whitney U test

Full size image

Discussion

To our knowledge, the ChoCO-W study is the largest global prospective study comparing COVID and non-COVID patients admitted with the diagnosis of AC. Recently, the CHOLECOVID study was published [18]. The methodology and aim of this study are different from ours. The CHOLECOVID study retrospectively compared the management of AC during the COVID pandemic with the pre-pandemic period. Instead we prospectively compared the characteristics and outcomes of patients who tested positive for SARS-CoV-2 during the episode of AC with those who did not.

Furthermore, in the ChoCO-W study recruitment period, Delta SARS-CoV-2 variant (B.1.617.2) was the most circulating virus and it was associated with higher transmissibility compared with wild-type SARS-CoV-2 and decreased vaccine effectiveness with higher incidence of secondary attack than the Alpha variant (B.1.1.7) [ https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/992983/21_May_2021_Risk_assessment_for_SARS-CoV-2_variant_VOC-21APR-02__B.1.617.2_.pdf].

During this first part of COVID-19 pandemic, health facilities were collapsing and people was recommended to stay home to limit human contact and the spreading of the virus.

The access to emergency departments was limited to patients with respiratory failure and acute abdomen with sepsis and septic shock.

Operating theaters were converted in ICUs and healthcare staff reallocated to manage patients with ventilatory support; consequently, access to OR was restricted to surgical patients non-eligible for NOM or after medical treatment failure in keeping the adequate personal protective equipment availability and decreasing the in-hospital circulation of the virus.

RT-PCR swab test result was mandatory to be admitted in OR.

The reported mortality of patients having GC is high mortality rate, and it increases in elderly and diabetic patients [19, 20].

Our study showed that COVID-19 patients with AC have an increased risk of presenting GC with higher postoperative complications and mortality rate.

This can be attributed to the associated comorbidity and frailty of COVID-19 patients, needing more frequently ventilatory mechanical support and parenteral nutrition and presenting with higher sepsis scores.

However, the environment may have contributed to enroll the most comorbid and severe patients in our study and probably to increase delay in surgical management (delay to ED admission + delay to OR admission) with negative outcomes and longer hospital stay.

Our data did not confirm an higher delay to surgical management; in fact, the mean (hours) delay from admission to surgical management was 63.44 (SD 201.4) and 45.9 (SD 110.1), respectively, for COVID and non-COVID groups (p = 0.89).

COVID patients had lower arterial lactate values compared to non-COVID patients [(3.52 (12.73) mmol/L compared with 16.96 (79), P = 0.03, respectively].

This is an unexpected result, since COVID patients had higher sepsis scores and signs of shock compared with non-COVID patients.

Carpenè et al. [20] reviewed 19 studies about hyperlactatemia and severe COVID disease, with 6459 patients included. They reported that COVID-19 patients with worse outcome have usually higher lactate values than those with better outcome, but most COVID-19 patients did not show hyperlactatemia, even if critically ill.

The association between blood lactate values and clinical outcome remains unclear in patients with SARS-CoV-2 infection. COVID-19 pathogenesis is multifactorial, in some way independent from severe ischemia and hyperlactatemia; in fact, patients with COVID-19 pneumonia or ARDS are reported with lower blood lactate values compared to those with non-COVID-19 pneumonia or ARDS of different etiologies [21].

Moreover, hyperlactatemia in COVID patients could be induced by medications such as metformin, propofol, acetaminophen [22,23,24], and catecholamines.

Iepsen et al. [25] reviewed the literature to assess if pathophysiology of lactate metabolism in sepsis and COVID patients is different from non-COVID septic patients. Evidence supports that elevated blood lactate value is strongly associated with mortality in septic patients. Lactatemia value seems unrelated to tissue hypoxia but likely reflects mitochondrial dysfunction and high adrenergic stimulation. Patients with severe COVID-19 exhibit near-normal blood lactate, indicating preserved mitochondrial function, despite a systemic hyperinflammatory state similar to sepsis.[25].

There is a need for further studies to assess this outcome. Nevertheless, serum lactate values monitoring in COVID patients may be useful for early identification of higher risk COVID-19 illness progression, but hyperlactatemia in severe COVID patients may not be present [22].

Our COVID-19 patients had higher total serum bilirubin, mostly conjugated, supporting the hypothesis that SARS-CoV-2 has a tropism for hepatic cells [26,27,28]. Several mechanisms were proposed to explain SARS-CoV-2 hepatic injury in critically ill patients including hypoxic hepatitis due to shock, high levels of positive end-expiratory pressure leading to hepatic congestion, and medications such as lopinavir/ritonavir. Most of our patients were not supported by mechanical ventilation. Despite that, they had abnormal liver functions most likely because of the hepatic ACE2 receptors which interact with SARS-CoV-2 causing direct cytopathic effects [26]. Patients with abnormal liver functions have at higher risk of progressing to severe COVID disease [28].

The COVID group showed a longer aPTT time and lower INR value compared with the non-COVID group in our study, and this would suggest intrinsic clotting factor deficiency.

This evidence supports published data about coagulability disorders of COVID-19 patients, characterized by significantly elevated D-dimer and fibrinogen (hyper-coagulability), mild thrombocytopenia and a mildly prolonged PT/aPTT (hypo-coagulability), based mainly on immunothrombosis mechanism which is triggered by hyperinflammatory response and diffuse endotheliopathy. This endothelial derangement most often manifests as an early hypercoagulable state with high risk of venous and arterial thromboembolic events and then results in a hemostatic derangement known as fibrinolytic shutdown [29, 30].

Elevated D-dimer levels in COVID patients are consistently reported, whereas their gradual increase during disease course is particularly associated with disease progression. PT and aPTT prolongation and fibrin degradation products’ increase with severe thrombocytopenia are correlated with life-threatening disseminated intravascular coagulation (DIC) [31,32,33].

Tang et al. [34] reported early that high D-dimer and fibrin degradation product (FDP) levels are risk factors for DIC and death in severe COVID-19 patients. Their study showed a significantly higher D-dimer and FDP levels and longer PT and aPTT in non-survivors compared to survivors on admission (p < 0.05) [34].

Venous or arterial thrombotic complications are reported in one-third of ICU COVID-19 patients despite pharmacological thrombo-prophylaxis [29, 35].

COVID-19 disease is associated with hypo-fibrinolysis as shown by thromboelastogram assays, but due to the costs of this laboratory exam, we did not collected sufficient data for analysis. Elevated D-dimer suggests hyper-fibrinolysis. This increases the risk of thrombotic events and renal failure which increases mortality rate [29]. SARS-CoV-2 may lead to direct endothelial injury and increased levels of pro-inflammatory cytokines (such as tumor necrosis factor-α, interleukin-1, and interleukin-6 leading to a cytokine storm). This has been associated with micro- and macrovascular thrombosis and organ failure [31]. The WSES was the first society to recommend early administration of prophylactic anticoagulation with LMWH in COVID-19 surgical patients to reduce the risk of thromboembolism [36]. The CORIST (Italian retrospective multicentric observational) study [37], which enrolled 2574 patients, showed that in-hospital heparin treatment was associated with a lower mortality, particularly in severely ill COVID-19 patients and in those with strong coagulation activation.

The International Society of Thrombosis and Haemostasis recommended measuring D-dimers, prothrombin time, and platelet count in all patients who present with COVID-19 infection in stratifying patients who may need admission and close monitoring or not [38].

The COVID-induced micro-angiopathy and hyper-coagulability could be correlated with the high incidence of GC in COVID-19 patients, but the ChoCO-W study cannot confirm this. Nevertheless, our study showed that the incidence of GC was doubled in COVID patients group compared with non-COVID (40.7% compared with 22.3%; p > 0.0001) and gallbladder wall was significantly thicker in COVID patients.

This was previously considered as a risk factor for “difficult gallbladder” surgery associated with higher conversion rate. In contrast, our data have shown that laparoscopic cholecystectomy, performed in 58% (75/180) of COVID-19 patients, is a safe and reproducible procedure in expert hands with a conversion rate of only 0.7% (compared with 5.4% in non-COVID group; p < 0.0001), that is, lower than the reported conversion rates for GC (ranging from 18 to 25%) [39, 40].

Open total cholecystectomy in our study was performed in 22.5% of the COVID-19 patients compared with 6.7% of the non-COVID patients. This is probably due to the hemodynamic instability and respiratory failure of COVID patients enrolled in our study: Nobody will perform a laparoscopic approach in hemodynamic unstable patients and in surgical patients presenting hypoxic respiratory failure.

Furthermore, several international surgical societies recommended against performing laparoscopic cholecystectomy because of the potential risk of SARS-CoV-2 transmission correlated with surgical smoke and artificial pneumoperitoneum: This may have leaded surgeons to reduce the use of laparoscopy in COVID patients.

To our knowledge, there are no data confirming increased risk of contamination among healthcare providers during laparoscopy and laparoscopic cholecystectomy is the golden standard treatment for cholecystitis in all patients [2].

However, in our study (laparoscopic and open) cholecystectomy showed a slightly higher rate of biliary leakage in COVID patients (1.1%) compared with non-COVID patients (0.8%) although not statistically significant. These data are slightly higher than biliary leakage rates reported in the literature [41,42,43].

Subtotal cholecystectomy, which was reported to be useful in the management of difficult gallbladders [44], was performed laparoscopically in 1.1% of the non-COVID patients and 0.8% of the COVID patients in our study.

Open partial cholecystectomy after conversion was performed in 1% of the non-COVID patients and 0.8% of the COVID patients. A second surgical exploration was required for 5.5% of the COVID patients compared with 2.6% of the non-COVID patients. COVID-19 patients had statistically higher postoperative complications, higher mean hospital stay (13.21 days compared with 6.51 days), and higher mortality (13.4% compared to 5.4%), similar to other studies [45].

The COVID group had more SSI, pulmonary infections, postoperative bleeding, and diffuse biliary peritonitis, compared with the non-COVID group.

This evidence supports the recommendation to delay surgical management in COVID patients having AC, according to their comorbidities, frailty, severity of pneumonia, and surgical risk in order to decrease postoperative complications and mortality rate, when it is possible [36, 46].

Several early retrospective studies reported an increased use of NOM and percutaneous cholecystostomy (PC) in treating both COVID and non-COVID patients presenting with AC during the early phase of the pandemic because of concerns about the safety of laparoscopy, artificial pneumoperitoneum, and biological fluids in spreading the virus in the operating rooms, and because of limited access to the operating rooms. This approach was associated with increased hospital stay, NOM failure, and increased in-hospital COVID infection [10, 47, 48].

In our study, laparoscopic cholecystectomy was performed in 1474/2869 (51%); NOM including antibiotics alone was used in 14% (414/2869) of COVID and non-COVID patients. The overall open cholecystectomy rate was 8% (250/2869), and PC was performed for 7% of (COVID and non-COVID) patients (211/2869).

To our knowledge, this confirms that PC is not an alternative to laparoscopic cholecystectomy in stable, non-critically ill patients, when an early and safe laparoscopic cholecystectomy can be performed. PC can be considered as a bridge to surgery in unstable, high risk, and unfit patients for surgery [49].

Strengths and limitations of the study

We enrolled prospectively all the COVID and non-COVID patients admitted with acute cholecystitis in ED in a 6-month period from October 2020 to April 2021. In this first period of Delta variant (higher virulence compared with wild-type SARS-CoV-2) COVID pandemic, only comorbid patients with acute abdominal pain and signs of sepsis were addressed and admitted to ED, overcrowded by severe COVID patients requiring ventilatory support and admission in ICU, because of governments lockdown and limited resources (beds, personal protective equipment, ventilators, operating rooms, and healthcare personnel).

Furthermore, several emergency surgeons opted for open cholecystectomy, when a safe laparoscopy was not possible in limit the spreading of the virus in OR.

We have to acknowledge that the COVID cohort is small and sicker and that the follow-up period of 1 month is short.

The long-term follow-up especially in those who had COVID-19 would be of interest in a future study.

However, this study has a wholistic approach looking for the global outcome without having a specific management protocol despite the major variation between the different countries. This is useful for the generalizability of the study.

To our knowledge, the ChoCO-W study is the first global study about AC comparing COVID and non-COVID patients during the ongoing pandemic.

Conclusions

The incidence of gangrenous cholecystitis is higher in COVID patients, and it is associated with high-grade Clavien-Dindo postoperative complications, higher length of hospital stay and higher mortality.

When it is possible, it is recommended to delay the surgical treatment in COVID-19 patients to decrease morbidity and mortality rates. Laparoscopic cholecystectomy is the golden standard treatment for acute cholecystitis in all patients. In expert hands, laparoscopic cholecystectomy is a safe and reproducible surgical procedure for acute cholecystitis, without significant increase in biliary leakage rate in COVID and non-COVID patients.

The rate of open cholecystectomy is higher in COVID patients compared with non-COVID patients, without statistically significant difference. To our knowledge, the laparoscopic approach is not associated with an increased biological risk of SARS-CoV-2 transmission in operating room, in presence of adequate protective personal equipment, protocols and skilled staff to manage COVID patients. Gangrenous cholecystitis is not an absolute contraindication to the laparoscopic approach in COVID and non-COVID patients.