A novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was identified as the putative factor for the cluster of pneumonia cases in Wuhan, China, in December 2019. This outbreak has led to a global pandemic within a short period. As of October 1, 2021, there were more than 2.35 billion confirmed SARS-CoV-2 cases with 4.8 million deaths globally, according to the World Health Organization [1]. Although the vast majority of infected individuals developed mild or no respiratory symptoms, a significant proportion of patients progressed to severe lung damage characterized by acute hypoxic respiratory failure (AHRF), needing intensive care support [2]. Lung involvement and acute respiratory failure are the characteristics of coronavirus disease 2019 infection (COVID-19), but other organs, such as the heart, kidneys, liver, and central nervous system, are also involved.

Our hospital faced a massive burden of COVID-19 cases during the first and second pandemic waves. Till October 1, 2021, we had treated nearly 32,000 COVID-19 cases with either RTPCR positive or radiological evidence of COVID-19 or both. Since we are a tertiary care referral centre for nearly ten surrounding districts in southern Tamil Nādu in India, we received more and more critically ill patients. The incidence of renal failure among our COVID-19 patients was nearly 25%, and dialysis dependency was about 5% of total admission.

Acute kidney injury is one of the most critical complications of COVID-19, occurring in 0.5–7% of cases and 2.9–23% of intensive care unit (ICU) patients [3-5]. In previous reports by Cheng Y et al., acute kidney injury (AKI) develops in 5% to 15% cases and carries a high (60% - 90%) mortality rate in SARS and MERS-CoV infections [5]. A study by Fanelli V et al. involving > 2000 inpatients with COVID-19 reported an AKI incidence of 27.8% [6]. Renal failure is an independent risk factor linked to increased in-hospital mortality in patients with severe sepsis regardless of etiology [7, 8]. In a multicentre study involving 3099 COVID-19 ICU patients from the United States of America, around 20% had acute kidney injury (AKI) requiring renal replacement (RRT) [9]. Similarly, in the United Kingdom, during the first pandemic wave, a quarter of ICU COVID-19 patients needed renal replacement therapy [10]. Furthermore, an ongoing requirement for renal support and prolonged hospitalization poses a significant health care resource burden.

Patients with pre-existing ESKD appear to be at high risk for COVID-19 and its complications, as most of them are aged, with comorbidities, and some of them might be on immunosuppressant medication to treat an autoimmune disease or a graft rejection [11]. Also, those patients already having CKD are at a higher rate of becoming dialysis-dependent and critically ill. Hemodialysis patients have additional risk factors, including chronic immune dysregulation, the need to go to the hospital for in-centre hemodialysis (HD), require frequent visits to health care facilities, and sharing standard rooms with other patients [12]. Therefore, once infected, dialysis patients become sources of spreading the infection within these high-risk groups, and reports suggest a critical course in patients with ESKD on hemodialysis [13]. Even though earlier small case series have suggested a milder course [14], more and more studies have reported the significantly critical nature of the disease. A study conducted in Qatar by Hamid et al. found the incidence of COVID-19 to be 7.1% in a dialysis cohort versus a 4% national cohort [15]. They also suffered from more complications and mortality, given their age and comorbidities. The study population had approximately a hundred times higher risk of mortality than Qatar's general population (15% in the dialysis cohort versus 0.15% countrywide) [15].

The globally widespread nature of the pandemic has also placed enormous demands on the need for RRT across the world. The RRT requirements among patients with COVID-19 appear to be five-fold higher than those observed in historical populations (4.9% versus 0.9%) [16]. In datasets of community cohorts reported by Chan et al., 5%–15% of patients hospitalized with COVID-19 required dialytic support [17]. In specific United States centres, the requirement of CRRT increased 370% over typical levels [18]. Thus, the demand outweighed the supply worldwide during the initial six months of the COVID-19 pandemic [19]. Moreover, this extraordinary demand for RRT has strained the delivery of all dialysis therapies across the world. As a result, manufacturers of RRT have often faced unprecedented challenges in supply.

In view of the COVID-19 pandemic by April 2020, our hospital allotted two blocks with a bed strength of 1,500, with 1400 beds having oxygen facilities. We also created 400 ICU beds in three divided areas. In addition, we had allotted ten hemodialysis machines and one CRRT machine. We also dedicated a team of four nephrology consultants, four intensivists, eight junior residents in nephrology, ten staff nurses, and eight well-trained certified technicians to our dialysis unit.

However, even with the above facilities, we could not provide extracorporeal dialysis methods to all needed patients. The factors that made hemodialysis difficult were, first, many patients were too critically ill to tolerate intermittent hemodialysis (IHD); second, there was an immediate need for RRT at the bedside; third, sometimes patients could not be shifted to hemodialysis unit; and last, there was a limited number of portable HD units. To avoid viral transmission and increased propensity to clot membranes, bloodline, and circuits in COVID-19 infection, we adapted the policy of single-use dialyzer, which needed more supply of consumables. In addition, the need for more CRRT machines and CVVH facilities was severely limited due to the lockdown. Nevertheless, so far, we have done RRT for 750 COVID-19 infected patients. But providing dialysis support to critically ill hemodynamically unstable patients was a big challenge for us.

Acute intermittent peritoneal dialysis (PD) can be performed at the bedside with minimal technical support and infrastructure. Even though the argument leans towards PD as an inferior procedure, more studies have shown PD to be effective in critically ill patients where continuous venovenous hemodiafiltration (CVVH) facilities are limited. The efficacy of PD is at least as greater as HD and it is possibly as efficient as hemodiafiltration [20, 21]. A meta-analysis by Liu et al. (analysis of six studies with 484 patients) compared PD therapy and extracorporeal RRT for management of AKI; PD therapy showed little or no difference from other extracorporeal therapies concerning all-cause mortality and recovery of kidney function [22]. Other clinical trials that have compared extracorporeal therapies versus PD indicate that the correction of uremia, acidosis, fluid overload, hyperkalemia, and incidence of infectious complications are not statistically different between the groups. Moreover, the cost-benefit advantage is more favourable for the PD group, especially in resource-limited settings [20]. The efficacy of correcting acidosis or Kt/V dosage may be higher in extracorporeal therapies; however, the evidence for this argument is meagre [22]. However, one limitation is that PD may be slightly less effective in removing ultrafiltration volume.

It is believed that the elevation of intraperitoneal pressure (IPP) generated by PD may alter diaphragmatic movement, thereby decreasing pulmonary compliance, which could hinder mechanical ventilation and worsen respiratory failure [22-24]. However, a recent prospective cohort study by Almeida et al. evaluated the respiratory mechanics of 154 patients with AKI and on PD versus HD; they showed that resistance in the respiratory system remained stable among the patients with PD and decreased among the HD patients, and oxygenation index increased in both the groups. Thus, the authors concluded an improvement in ventilatory mechanics in both the groups, without significant differences [25].

The clearance of inflammatory cytokines in PD has been well demonstrated in experimental models. The study by Altmann et al. [26] also proved interleukin 6 (IL-6) to be eliminated through PD effluent. Zhao et al. showed that inflammatory factors, including TNF- alpha, are removed by PD. The clearance of inflammatory cytokine developed during COVID-19 disease is also a possible advantage of PD. TNF-α, IL-6, and PCT are removed effectively in peritoneal dialysis [27].

To overcome the crisis during the pandemic, our hospital started an acute intermittent PD program in critically ill COVID-19 patients with renal failure, including both AKI and ESKD. Initial planning took place in the weeks after the surge and took about two months to implement the peritoneal dialysis program. From June 2020 to the end of September 2021, a total of 627 COVID-19 patients received intermittent hemodialysis, and 91 COVID-19 patients received acute intermittent peritoneal dialysis in our hospital (Fig. 1: x-axis represents months and the y-axis represents the total number of cases undergoing procedure).

Though the vaccination and COVID-prevention strategies are being implemented throughout, there is still a global concern about the third wave of COVID-19. So, the use of acute peritoneal dialysis as an option for RRT in critically ill patients with COVID-19 in resource-limited settings and with mass needs is still an area to explore.

The study involved a retrospective, single-centre observational design.

The study population included 91 patients diagnosed with hemodynamically unstable critically ill SARS-CoV-2 infection (Table 1) with renal failure, both AKI and CKD, and admitted to the intensive care unit of our hospital from July 2020 to September 2021.

Fig. (1). Monthly trends of haemodialysis and peritoneal dialysis in COVID-19 patients;x-axis: months from June 2020 to October 2021, y-axis: total number of cases (blue line indicates hemodialysis; brown line indicates peritoneal dialysis).

1. Primary outcome: This outcome involves recovery from critical illness or death of patients.
2. Secondary outcome: This involves efficacy (urea reduction ratio calculated at 24 and 48 cycles) of peritoneal dialysis and complications of peritoneal dialysis.

Sampling procedure: Individuals who had positive nasopharyngeal swab for SARS-Cov-2 by reverse transcriptase method or a CT chest suggestive of COVID-19 (CORADS 4-6) or both were considered to have COVID-19 disease. Patients who were admitted with COVID-19 disease were classified as having mild, moderate, severe, and critically ill disease (Table 1). Patients who were diagnosed to be critically ill were admitted to the intensive care unit (Fig. 2).

1. Inclusion criteria: Those patients who were diagnosed with severe critically ill COVID-19 infection along with with acute kidney injury or chronic kidney disease and undergoing acute peritoneal dialysis were included in the study.
2. Exclusion criteria: Those patients who had developed poor PD flow, major leaks, bleeding complications, prone ventilated patients, sudden worsening of hypoxia on a ventilator soon after initiation of PD were taken for IHD/SLED/CVVH and excluded from the study.

After obtaining an informed written consent from the patient or first-degree relative (if the patient is in an altered sensorium), peritoneal dialysis was initiated. Bedside percutaneous catheterization was done as per our department protocol. After aseptic abdomen preparation and filling the peritoneal cavity with 20ml/kg of peritoneal dialysis fluid as priming (Table 2), a non-cuffed rigid plastic catheter was inserted just below the umbilicus in the midline by blind Seldinger procedure under local anaesthesia. Each acute peritoneal dialysis cycle (30 ml per kg) was prescribed at a frequency of one hour thirty minutes with an intraperitoneal dwell time of one hour and inflow time, and outflow (drain) time being 15 minutes each. The peritoneal dialysis cycles were repeated and continued until the patients were hemodynamically stable or recovered from renal failure or death. Patients with hypervolemia, CT chest suggestive of a mixed pattern of ground glass opacities, and pulmonary edema were prescribed hypertonic PD fluids (addition of 100 ml of 25% dextrose along with existing fluid and shorter dwell time). For those patients with pre-existing diabetes mellitus, eight units of short-acting insulin were added to each litre of PD fluid and subcutaneous or intravenous insulin for control of blood sugar as per patient glycaemic status and as per the protocol.

Fig. (2). Institutional protocol of bedside acute peritoneal dialysis.

All patients were treated with medications as per covid pandemic guidelines (Annexure 1). All patients were either on non-invasive ventilation (high flow nasal oxygen /HFNO) or invasive ventilation. The patients were managed in the ICU until hemodynamic stability was attained or weaned off the ventilator. Patients who recovered from hemodynamic instability were converted to regular hemodialysis or followed conservatively according to need. Furthermore, those patients who recovered were discharged after the isolation period.

All patients admitted in our covid block underwent routine biochemistry investigations, complete blood count, renal function test, serum electrolytes, serum ferritin, CRP, and radiological investigations immediately at the time of admission. Moreover, for patients who underwent peritoneal dialysis, RFT and electrolytes assessment was repeated at the end of 24 and 48 cycles to calculate the urea reduction ratio in order to assess the recovery and decide whether to continue or stop peritoneal dialysis.

From our hospital, COVID-19 database history, clinical findings, routine laboratory parameters, including a complete blood count and serum biochemistry (blood urea, serum creatinine, electrolytes, C-reactive protein, and ferritin) and radiological findings, were taken. The lung involvement in CT chest was divided into 4 grades (grade 1 = < 25% involvement, grade 2 = 25% to 49% involvement, grade 3 = 50% to 74% involvement, and grade 4 = > 74% involvement). The data related to the total number of PD cycles, complications like poor flow, haemorrhage, and peritonitis, were extracted from our department's COVID-19 databases. After 24 cycles and 48 cycles of peritoneal dialysis, the urea reduction ratio (URR 1 and URR 2) was assessed.

The study was approved by the Institutional Ethical Committee, Madurai medical college, Madurai. The confidentiality of research subject's personal information was ensured in accordance with the ethical principle of the Declaration of Helsinki.

SPSS 22.0 version and R-programming were used to carry out the statistical analysis. Demographic and comorbidity statuses of the patients were compared among the patients who recovered and the patients who expired using the chi-square test/Fisher's exact test. Shapiro-Wilk normality tests were used to find out the normality of the continuous data.

Box transformation was used to transform the non-normally distributed data into normally distributed data. Independent sample t-test was used to compare the biochemical parameters (continuous data) between the two group patients. The multivariate statistical model was used to find out the key factors associated with mortality. A five percent level of significance was considered statistically significant.

The demographic and clinical characteristics of the cohort are shown in Table 3a. 91 patients were included in the study. Of the 91 patients in our study population of renal failure with critically ill COVID-19, 60 (65.9%) patients were ESKD on regular hemodialysis before admission, 7 (7.7%) patients were known CKD on conservative management prior to COVID-19 infection, and 24 (26.3%) patients were newly diagnosed AKI by appropriate investigations. Males constituted 63.7% of our cohort, but neither sex was significantly associated with mortality. The average age in the survivor group was around 49.88 years, while in non-survivors, it was 59.07 years. Patients belonging to the age group more than 75 years and 50 to 74 years were found to have significantly higher mortality (p-value = 0.04). The common comorbidities were diabetes mellitus (53.8%), coronary artery disease (29.6%), and pre-existing pulmonary disease (16.4%). When considering the comorbidity status of the patients' coronary artery disease, Type 2 diabetes mellitus and pre-existing lung disease were significantly associated with mortality (P-value <0.05). The most common symptoms at initial presentation were dyspnoea (98.9%), fever (91.2%), cough (86.8%), and diarrhoea (29.6%).

Among the patients, 26.3% (n=24) were diagnosed with AKI and 73.7% as CKD. Patient admission urea (173.85 mg/dl vs. 217.76 mg/dl) and creatinine (10.4 vs. 14.08) values exhibited a significant correlation with mortality (value < 0.05) . Hypertonic PD was done in 47.2% (n=43) of patients, and it was significantly associated with decreased mortality (Table 3b). The urea reduction ratio was calculated after 24 cycles (URR1) and 48 cycles (URR 2) of PD. Both URR1 and URR 2 were significantly higher in the survivor group (Fig. 3). The mean number of PD cycles was significantly higher in the survivor group than in the non-survivor group (66.52 vs. 44.26, p-value < 0.05). Complications, like haemorrhage and peritonitis, were found in 13.1% (n=12) of patients. However, none of the complications were significantly associated with mortality. The mean dialysis vintage in the non-survivor group was 7.38 months and was significantly higher than in the survivor group (2.57 months). Mean blood pressure at the time of presentation in survivor vs. non-survivors was systolic (84.3 mm of Hg vs. 77.5 mm of Hg) and diastolic BP (59.09 mm of Hg vs. 42.93 mm of Hg), and it was found to be statistically significant with the outcome (p-value < 0.05). In the non-survivor group, around 83% of patients were in double ionotropic support and 17% in single ionotropic support. Patients with double ionotropic supports were found to have significantly increased mortality (p-value < 0.05). Mechanical ventilation was used in 50.5% (n= 46) of patients, and its use was significantly associated with mortality. The mean duration of ICU stay (duration from entry to ICU to an outcome) among survivors was 127.28±88.69 hours (6-9days on average), whereas, in non-survivors, it was 65.32±30.79 hours (3-4 days on average) and found to be statistically significant.

Patients in the non-survivor group had significantly lower hemoglobin when compared to the survivor group (p-value=0.02). Biochemical parameters (Table 3c), including total leukocyte count, neutrophil-lymphocyte ratio, blood urea and creatinine at presentation, were significantly increased in the non-survivors group. Patients who expired had significant hyperglycemia (p-value <0.05). Despite higher blood glucose levels in the non-survivor group, only four patients in the expired group had diabetic ketoacidosis (DKA) with urine acetone positive (p-value =0.701). The lung involvement in the CT chest was divided into four grades. Patients having grade 3 and grade 4 were found to have significantly higher mortality. 47.2% (n=43) patients were detected to have features of both COVID-19 pneumonia and pulmonary edema in the CT chest, but the combination was not found to be a significant risk factor for mortality. Significant elevation of serum bilirubin, SGOT, and SGPT was observed in the non-survivors group. Serum sodium was found to be significantly lower in the non-survivor group (p-value < 0.05). Inflammatory markers, CRP, and ferritin were significantly increased in the non-survivors group (p-value< 0.01). The mortality in our cohort was 63.7% (n=58) (Table 3d).

Multivariate analysis was conducted to find out the crucial predictors of mortality. The higher percentage of lung involvement in CT chest (Fig. 3a) and increased neutrophil-lymphocyte ratio were found to be independently associated with increased mortality, whereas a higher number of PD cycles (Fig. 3b) and higher urea reduction ratio at 24 cycles (URR1) (Fig. 3c) were found to be independently associated with decreased mortality (Fig. 3d).

Fig. (3a). Box plot analysis of CT chest with the outcomes (R-recovered; E-expired).

Fig. (3b). Box plot analysis of PD cycle with the outcomes (R-recovered; E-expired).

Fig. (3c). Box plot analysis of URR1 with the outcomes (R-recovered; E-expired).

Fig. (3d). Box plot analysis of URR2 with the outcomes (R-recovered; E-expired).