Context
COVID-19 has caused millions of deaths worldwide since it was reported in Wuhan, China, in late 2019. Nephrologists have been at the forefront of clinical care of patients with COVID-19 as kidney-related complications, including acute kidney injury (AKI), glomerular disease, and severe electrolyte derangements, are frequently encountered. In addition to the deleterious effects of COVID-19 upon the kidneys, patients with chronic kidney disease (CKD) have been disproportionately impacted, especially those with kidney failure receiving chronic dialysis and kidney transplant recipients (KTR).
CKD is the most common risk factor for death in patients with COVID-19 worldwide, and the risk increases with higher CKD stage, with the highest risk occurring in those with kidney failure receiving replacement therapy (KFRT) and KTR. Furthermore, patients with KFRT receiving in-center hemodialysis are at markedly higher risk of COVID-19 than those receiving home dialysis therapies who received care via telehealth. This is likely explained by increased risk of exposure of patients receiving in-center dialysis to SARS-CoV-2 during thrice weekly travel via group or public transportation to/from dialysis and exposure to patients and staff at the dialysis unit.
Single-center studies reporting clinical outcomes in patients with CKD and COVID-19 have reported widely variable results, and this variability is due in part to differences in patient characteristics such as age, sex, race, and comorbidities that are associated with worse outcomes. However, other factors, including differences in strain upon medical centers due to large numbers of patients with COVID-19, heterogeneity in data collection methods, methods used to ascertain and categorize CKD stage and clinical outcomes, and study time points, affect reported outcomes. Large multicenter studies that include a diverse array of medical centers across large regions and standardized and validated methods to reduce sources of variability are therefore needed.
Subject of Review
The Study of the Treatment and Outcomes in Critically Ill Patients With COVID-19 (STOP-COVID) is a multicenter cohort study involving 68 medical centers across the USA and was established in April 2020, during the peak of the COVID-19 surge in the northeastern USA. STOP-COVID enrolled consecutive adults (≥18 years) admitted to intensive care units (ICU) with laboratory-confirmed COVID-19. The overall goal of STOP-COVID is to identify independent risk factors for hospital mortality and acute organ injury and to identify treatment strategies associated with improved survival (https://clinicaltrials.gov/ct2/show/NCT04343898).
In a recently published ancillary study of STOP-CO­VID, Flythe et al. focused upon clinical outcomes in adults with pre-existing nondialysis-dependent CKD (ND-CKD) or with KFRT admitted to ICU with COVID-19 between March 4 and May 10, 2020. Patients were followed forward in historical time from ICU admission to in-hospital death, hospital discharge, or June 6, 2020, when the database was locked. There were 4,264 individuals included in the study, 143 (3%) of whom had pre-existing KFRT, 521 (12%) with ND-CKD, and 3,600 (85%) without CKD. Unsurprisingly, patients with ND-CKD and KFRT had more comorbidities, including hypertension, diabetes mellitus, and heart disease, than those without CKD. Of those with KFRT, 90% received in-center hemodialysis (HD), 6% peritoneal dialysis (PD), 1% home HD, and 3% had unknown dialysis modality.
Patients with KFRT were admitted to the ICU sooner after hospital admission (median 4 days) than patients with ND-CKD or those without CKD (7 days). Importantly, KFRT patients were less likely to have had documented COVID-19-related symptoms prior to ICU admission than those without CKD except for altered mental status, which was more common in those with KFRT (25% vs. 12%). Patients with KFRT also were less likely than those without CKD to require mechanical ventilation at the time of ICU admission (56% vs. 63%). However, patients with KFRT were more likely to require pressors at ICU admission (50%) than those with ND-CKD (42%) or without CKD (41%), though the presence of shock, defined as requirement for 2 or more pressors, was similar between groups. Baseline biomarkers also varied between groups with patients with KFRT having lower white blood counts, platelet counts, and fibrinogen levels but higher C-reactive protein (CRP), IL-6, ferritin, and troponin levels than those without CKD.
Many patients in this study received interventions intended to suppress SARS-CoV-2 and/or inflammation, despite the fact that patients in this study were enrolled before data were available from large COVID-19 clinical trials, and patients with CKD were excluded from most of those trials. Most patients received hydroxychloroquine or chloroquine and fewer received azithromycin or other antiviral medications. Seven percent of patients without CKD received remdesivir compared to 2% of those with ND-CKD and 0% of those with KFRT. Although corticosteroid use was similar across groups, those without CKD were more likely to receive the IL-6 antagonist tocilizumab (19%) compared to those with ND-CKD (14%) or KFRT (9%).
In addition to being less likely to require mechanical ventilation at the time of ICU admission, patients with KFRT were less likely than those without CKD to require mechanical ventilation at any time during hospital admission (74% vs. 80%). Patients with ND-CKD or KFRT were also less likely to require prone positioning than those without CKD. Laboratory parameters differed between groups during the first 14 days of hospitalization; patients with KFRT and ND-CKD had persistently lower lymphocyte and platelet counts and higher CRP levels. Patients with KFRT had higher lactate levels during the first week of ICU care compared to other groups.
Patients with ND-CKD and KFRT had significantly higher mortality than those without CKD in unadjusted and adjusted analyses at 14 and 28 days after ICU admission. At 28 days after ICU admission, death occurred in 51% and 50% of patients with ND-CKD and KFRT but in only 35% of those without CKD. In the fully adjusted model (including age, sex, race, Hispanic ethnicity, diabetes, hypertension, coronary artery disease, heart failure, and atrial fibrillation or flutter), the hazard ratios for 28-day mortality were 1.41 and 1.25 for patients with KFRT and ND-CKD compared to those without CKD. Among those with KFRT, those dialyzing via a catheter had 1.94-fold higher risk of death than those dialyzing with an arteriovenous vascular access. Among those with ND-CKD, there was a nonsignificant trend toward higher mortality for those with baseline creatinine ≥2 mg/dL. The most common cause of death in all groups was respiratory failure. Patients with KFRT had a nonstatistically significant trend toward increased secondary outcomes, including shock, ventricular arrhythmia/cardiac arrest, and major bleeding events.
Second Opinion
Important strengths of this study include its large size, geographical diversity (though limited to the USA), standardized data collection methods, rigorous statistical analyses, and inclusion of consecutive ICU admissions, which should reduce selection bias. However, as acknowledged by the authors, there are limitations that may affect the interpretation of the results. Although patients were enrolled consecutively, data were collected retrospectively, and the study was observational. Therefore, though the investigators attempted to control for multiple factors, it remains possible that residual confounding may have affected the results. Baseline CKD status may have been misclassified in some patients, though the use of hospital laboratory data would be expected to yield more accurate assessment of kidney function than the use of diagnostic codes. Another limitation is the fact that biomarker data were missing for many patients, and it is likely that levels were checked preferentially in the more severely ill patients, which may have skewed results.
Despite the geographic diversity of study sites from all major regions of the USA, nearly all were large academic centers and 58% of patients were from the northeastern USA. Many medical centers in the northeastern USA experienced an overwhelming surge of COVID-19 during spring 2020 that necessitated a rapid expansion of ICU capacity, resulting in severely constrained nursing and other resources, including the capacity for renal replacement therapies. STOP-COVID investigators have demonstrated that medical center strain is associated with increased mortality in critically ill patients with COVID-19. Since this study was limited to patients after ICU admission, it is also possible that patients with severe ND-CKD and/or KFRT may have experienced differences in pre-ICU care that may have affected outcomes and may have died prior to ICU admission or been denied ICU care due to perceived medical futility. However, if only "healthier" patients with severe CKD were admitted to ICU, this would have biased outcomes to be better in those with ND-CKD and KFRT.
Clinical outcomes for patients with COVID-19 have improved in the USA since the time when these data were collected, likely because health care systems have been less severely strained and due to improvements in pharmacologic and nonpharmacologic treatments. It is therefore possible that patients admitted to the ICU, including those with CKD, may have better outcomes than that reported in this article. Although patients with KFRT had higher 28-day mortality than those without CKD (50% vs. 35%), mortality from single-center studies of KFRT patients with COVID-19 admitted to ICUs in the USA during the same time period was significantly higher. The reasons for this disparity are unclear but may be explained in part by the fact that this cohort comprised mostly large academic medical centers, and measures of hospital quality and number of ICU beds are predictors of mortality in ICU patients with COVID-19. It is also important to note that during the spring 2020 COVID-19 surge, there was a large increase in mortality among US patients with KFRT but a simultaneous large decrease in hospitalizations in these patients, strongly suggesting that a large proportion of patients with KFRT and COVID-19 died without being hospitalized.
It is unclear why patients with severe CKD were less likely to present with classic COVID-19 symptoms and more likely to present with atypical symptoms such as altered mental status, but this observation should alert clinicians to be especially vigilant for COVID-19 in patients with CKD and unexplained change in clinical status. It is also unclear why patients with KFRT were less likely to require mechanical ventilation in the ICU than patients without CKD. Since patients with KFRT had a higher prevalence of vasopressor use at the time of ICU admission, clinicians may have admitted them to the ICU over concerns of worsening hemodynamics during renal replacement therapy.
The reasons why CKD is an independent risk factor for death in patients with COVID-19 are poorly understood. The most common causes of death in STOP-COVID were respiratory failure, septic shock, and kidney failure. The presence of severe kidney disease can make assessment and optimization of volume status in patients with respiratory failure more difficult. Also, since CKD is a potent risk factor for AKI in patients with COVID-19 and AKI is strongly associated with increased mortality in patients with COVID-19, increased AKI may have contributed to excess mortality in ND-CKD. Moreover, persons with ND-CKD and KFRT are also at increased risk of bacterial infections and poor outcomes after infection than those without CKD. Although patients with CKD did not have markedly increased prevalence of bacteremia at the time of ICU admission in STOP-COVID, it is unclear whether there were more episodes of bacteremia during the subsequent ICU course in patients with CKD. This is particularly relevant for patients with KFRT since HD catheter use compared to arteriovenous access was associated with nearly 2-fold increased mortality in this study.
Although the increased mortality attributable to CKD in patients with COVID-19 likely reflects differences in underlying physiology, including dysregulated immune responses, since these patients were less likely to receive immunomodulatory and antiviral therapies, it is also possible that differences in clinical care may have contributed to worsened outcomes. One of the most important results from this study was the lower incidence of treatment with emerging therapies that clinicians thought might improve outcomes in COVID-19. Although some, such as remdesivir, are not recommended for patients with eGFR <30 mL/min/1.73 m2, others, such as tocilizumab, are not contraindicated and do not require dose adjustment for reduced GFR. Although the limited efficacy of these treatments would not fully explain the marked difference in mortality in patients with CKD, there may have been other unmeasured differences in care between those with/without CKD that may have influenced mortality. Unfortunately, patients with kidney disease are often denied potentially beneficial medical therapies due to inappropriate concerns over safety or futility.