Comparison of renal replacement therapy and renal recovery before and during the COVID-19 pandemic- A single centre observational study

Rececca Ryan Intensive care unit, University College London Hospital, UK Isabel Taylor Intensive care unit, University College London Hospital, UK Chris Laing Department of Renal Medicine, University College London, Royal Free Hospital, London, UK Mervyn Singer Intensive care unit, University College London Hospital, Bloomsbury Institute of Intensive Care Medicine, University College London, UK; Bloomsbury Institute of Intensive Care Medicine, University College London, UK Niall MacCallum Intensive care unit, University College London Hospital, UK Nish Arulkumaran (  nisharulkumaran@doctors.org.uk ) Intensive care unit, University College London Hospital, UK; Bloomsbury Institute of Intensive Care Medicine, University College London, UK

Comparison of renal replacement therapy and renal recovery before and during the COVID-19 pandemic-A single centre observational study Introduction Provision of renal replacement therapy (RRT) for patients with COVID-19 was challenging for both logistical and disease-related factors [1]. A signi cant proportion of COVID-19 patients developed acute kidney injury (AKI), many of whom require RRT [2]. In our institution, as with many others, shortages of continuous venovenous hemodia ltration (CVVHDF) machines and replacement uids necessitated reduced intensity of hemo ltration (exchange rates) to facilitate the greatly increased demand for RRT capacity. Patients with COVID-19 are at increased risk of thrombotic complications [3], creating challenges around optimizing the anticoagulation to preserve lter life. This was further compounded by the lack of citrate regional anticoagulation.
To assess the impact of these modi cations, and any differences in thresholds for the initiation of RRT, and we drew comparison against consecutive non-COVID-19 critically ill patients in our intensive care unit requiring RRT over the previous year. We speci cally addressed differences in initiation criteria for RRT, uid balance and azotaemia with different clearance rates, incidence of hemo lter clotting, duration of RRT prior to renal recovery, hospital survival, and renal recovery.

Materials And Methods
We conducted a single-centre retrospective case control study of patients with COVID-19 requiring acute RRT on the intensive care unit at University College London Hospital between 1st March and 30th June 2020. Comparison was made against those receiving RRT in the pre-COVID period from January 2019 to February 2020. Our hospital lacks a renal dialysis unit so ICU admission is required for all patients with acute kidney injury needing RRT.
Data were extracted from electronic healthcare records on patient demographics, RRT parameters, anticoagulation, biochemistry and uid balance on the day of RRT initiation, creatinine clearance rates, the incidence of lter clotting, duration of RRT, renal function on discharge from hospital, and hospital survival. Patients receiving chronic dialysis were excluded as our study sought to assess criteria for RRT initiation and renal recovery rates.
Details of the UCLH pre-COVID-19 protocol and shortage-driven modi cations during COVID-19 are detailed in the Supplementary Appendix. Brie y, prior to COVID-19, the standard of care was citrate-based regional anticoagulation and initiation of CVVHDF at exchange rates of 1000 ml/hr for both hemo ltration and dialysis. During shortage periods, COVID-19 patients received continuous venovenous hemo ltration (CVVHF) initiated at an exchange rate of 1000 ml/hr. In cases where augmented small molecule clearance was required, 1000 ml/hr dialysis was subsequently added until desired biochemistry was achieved. Prophylactic anticoagulation using low molecular weight heparin (enoxaparin) was routinely increased in all COVID-19 patients from 40 mg once daily to 40 mg twice daily. Therapeutic doses were given to those with diagnosed venous, pulmonary or systemic arterial thromboembolism. During periods of citrate unavailability, therapeutic dose low molecular weight heparin (LMWH) was used to maintain lter patency.
As this was a retrospective observational study, we did not de ne any sample size. Anonymised data were used for analysis. No patients were missing the primary outcome or key confounders (details in Supplementary data). Continuous and categorical variables are reported as median (interquartile range) and n (%), respectively. For continuous variables, the Mann Whitney U test was used for comparison between groups. Categorical data were compared using the chi-squared test. Changes in biochemistry or uid balance over time was assessed using 2-way ANOVA. Statistical analysis was performed, and . As data on routine RRT management are collected as part of routine service evaluation within our institution, speci c permission was not required for data obtained from pre-COVID patients.  Table 1).

Demographics
There were no signi cant between-group differences in gender, body mass index (BMI), time from hospital admission to ICU admission, and proportions of patients with hypertension and diabetes mellitus ( Table 1). Non-COVID-19, patients were older, more likely to have chronic kidney disease (CKD), and had a higher serum creatinine on hospital admission. Censored for patients without CKD, the between-group difference in admission serum creatinine remained signi cant (Table 1).

RRT initiation, clearance, and discontinuation
No inter-group differences were seen in cumulative uid balance, urine output in the preceding 24 hours, pH, potassium, serum creatinine or urea on initiation of RRT (p>0.05) (Supplementary Table 2

Anticoagulation and thrombosis
The overall incidences of lter clotting and blood transfusion requirements were higher in COVID-19 patients (Supplementary Table 4). All non-COVID-19 patients received citrate regional anticoagulation.
Twenty-seven of the 47 (57%) COVID-19 patients requiring acute RRT were diagnosed with a venous thromboembolic event (VTE), all of whom received therapeutic low molecular weight heparin (LMWH). In addition, seven patients received regional citrate anticoagulation. Of the 20 patients without any evidence of VTE during admission, 17 were initiated on citrate anticoagulation with prophylactic LMWH while 3 patients were initiated on RRT with therapeutic anticoagulation with LMWH without citrate. Nineteen patients switched from regional citrate anticoagulation to systemic LMWH anticoagulation or vice versa depending on citrate availability. The 20 COVID-19 patients without diagnosed VTE had a similar incidence of lter clotting compared to the 27 patients with VTE. There was also no difference in the incidence of circuit clotting between days on citrate and days on LMWH.

Clinical outcomes
Hospital mortality was similar between patients with and without COVID-19 (60% vs. 68%; p=0.508) ( Table 2). Among survivors, the duration of RRT, IMV, ICU stay and time from RRT cessation to hospital discharge was greater among COVID-19 patients. On hospital discharge, serum creatinine was signi cantly lower among patients with COVID-19 (Figure 3), excluding one patient with COVID-19 who was referred for ongoing RRT. Censored for patients without CKD, hospital discharge serum creatinine was comparable to patients without COVID (Table 2).

Discussion
Data on acute RRT during the COVID-19 pandemic have not been described in detail, nor a comparison made against a non-COVID cohort. The cohort of COVID-19 patients in our center were comparable in age and gender to the published literature [4]. Compared to patients without COVID-19, patients with COVID-19 were younger, had a lower serum creatinine on hospital admission, and were less likely to have CKD. As respiratory failure is the primary manifestation of COVID-19, it is unsurprising that all COVID-19 requiring RRT also required IMV. All however required vasopressor support which, in view of the normal lactate level in the majority of patients, is more likely to re ect hypotension due to a combination of hemodynamic consequences related to sedation use and high airway pressures rather than the metabolic effects of sepsis.
The metabolic phenotype in COVID-19-associated AKI differs from that seen in non-COVID patients in other respects. Firstly, the much later requirement for RRT in COVID-19 patients was not related to any signi cant delay in initiation as urea, creatinine, uid balance and urine output values were similar to non-COVID-19 patients despite a far less severe metabolic acidosis. In the COVID-19 patients RRT was primarily initiated for ultra ltration to achieve a neutral uid balance with the aim of improving oxygenation [5]. However, our data show no effect on P:F ratio over the rst 48 hours, despite active uid removal. The much earlier use of RRT in non-COVID-19 patients who were more acidotic also implies a much higher degree of catabolism in these patients.
Severe COVID-19 disease is associated with endothelial activation and a hyperin ammatory, prothrombotic state [6,7]. Despite patients with COVID-19 receiving augmented prophylactic dose LMWH with regional citrate anticoagulation, the incidence of lter circuit clotting was higher than in patients without COVID-19 on regional citrate anticoagulation. On average, one in 11 CVVHDF circuits in COVID-19 patients clotted each day. There was an associated increase in RBC transfusion requirements among COVID-19 patients requiring one unit every eight days. No differences were found however in the rate of lter clotting between the use of regional citrate anticoagulation or therapeutic systemic anticoagulation with LMWH.
The duration of RRT was longer compared to non-COVID-19 AKI, and IMV was often required after renal function recovered. Despite a greater proportion of COVID-19 patients requiring three organ support (respiratory, renal and cardiovascular), survival rates were similar to non-COVID-19 patients. Rates of mortality of COVID-19 patients receiving both IMV and RRT are comparable to other published data [8].
The aetiology of AKI in COVID-19 is multifactorial including haemodynamic compromise, in ammatory mediators, and direct viral infection of glomerular and tubular cells [9,10]. However, renal function recovered in almost all survivors by hospital discharge.
As with all retrospective analyses, we acknowledge that ndings are associative. This is a single centre study with relatively limited numbers. Our clinical criteria for ICU admission and RRT may differ from other centres, potentially limiting generalisability. However, our patient demographics are consistent with published COVID-19 data. Baseline renal function was unavailable for some patients so the presence of CKD may be under-estimated. The PaO 2 : FiO 2 ratio does not re ect details on mechanical ventilator settings, although is unlikely to alter the conclusion given the large difference between non-COVID and COVID-19 patients. Furthermore, data on sequential uid balance and serum creatinine following initiation of RRT do not factor in survival bias, where some patients may not have survived 3 days following RRT initiation (details in supplementary data).
In summary, a third of critically ill patients with COVID-19 on the UCLH ICU required acute RRT. Compared to patients without COVID-19, they were younger, had a lower serum creatinine on hospital admission, and were less likely to have CKD. RRT was initiated much later following ICU admission in COVID-19 patients, primarily for uid balance rather than acidaemia or hyperkalaemia. COVID-19 patients were successfully managed with reduced CVVHDF exchange rates. Duration of RRT requirement is longer compared to non-COVID-19 patients despite lower serum creatinine on hospital admission and lower rates of CKD in COVID-19 patients. Requirement for ongoing IMV precludes intermittent RRT in a nephrology acute dialysis unit. Hospital mortality was equally high in both COVID-19 and non-COVID-19 patients, and renal recovery among survivors was comparable.