Abinet M. Aklilu, MD, MPH1,2; Sanchit Kumar, MD, MBBS1; James Nugent, MD, MPH1,3; et al
JAMA Intern Med. 2024;184(4):414-423. doi:10.1001/jamainternmed.2023.8225
Question Are hospitalized patients who develop acute kidney injury (AKI) associated with COVID-19 at risk of worse kidney function trajectory and higher mortality than those with AKI associated with other illnesses?
Findings This multicenter cohort study using electronic health records of 9624 hospitalized patients with AKI found that patients with COVID-19−associated AKI had a 33% lower risk of major adverse kidney events with 22% lower risk of worsened kidney function and 69% lower risk of mortality compared with those with AKI due to other illnesses.
Meaning These findings indicate that survivors of hospitalization with COVID-19−associated AKI have a significantly lower risk of long-term kidney function decline and all-cause mortality than those with AKI associated with other illnesses.
Abstract
Importance COVID-19 infection is associated with a high incidence of acute kidney injury (AKI). Although rapid kidney function decline has been reported in the first few months after COVID-19−associated AKI (COVID-AKI), the longer-term association of COVID-AKI with kidney function remains unknown.
Objective To assess long-term kidney outcomes of patients who had COVID-19−associated AKI.
Design, Setting, and Participants This was a retrospective longitudinal multicenter cohort study conducted in a large hospital system using electronic health records data on adult hospitalized patients with AKI and COVID-19 or other illnesses. Included patients were hospitalized during the COVID-19 pandemic (March 2020-June 2022), were screened for SARS-CoV-2, had AKI, and survived to discharge, or had been hospitalized during the 5 years before the pandemic (October 2016-January 2020), had a positive influenza A or B test result, had AKI, and survived to discharge. Patients were followed up for a maximum of 2 years after hospital discharge. Data analyses were performed from December 2022 to November 2023.
Exposure COVID-19 and influenza.
Main Outcomes and Measures The primary outcome was major adverse kidney events (MAKE), defined as a composite of mortality and worsened kidney function (estimated glomerular filtration rate [eGFR] decline by ≥25% from discharge eGFR or kidney failure requiring dialysis). Multivariable time-to-event analyses were performed to compare MAKE between individuals with COVID-AKI and those who had AKI associated with other illnesses hospitalized during the same period. For further comparison, this outcome was assessed for a historic cohort of patients with influenza-associated AKI.
Results The study cohort included 9624 hospitalized patients (mean [SD] age, 69.0 [15.7] years; 4955 [51.5%] females) with AKI, including 987 patients with COVID-AKI, 276 with influenza-associated AKI, and 8361 with AKI associated with other illnesses (other-AKI). Compared with the other 2 groups, patients with COVID-19−associated AKI were slightly younger in age, had a higher baseline eGFR, worse baseline comorbidity scores, higher markers of illness severity, and longer hospital stay. Compared with the other-AKI group, the COVID-AKI group had lower MAKE (adjusted hazard ratio [aHR], 0.67; 95% CI, 0.59-0.75) due to lower all-cause mortality (aHR, 0.31; 95% CI, 0.24-0.39) and lower rates of worsened kidney function (aHR, 0.78; 95% CI, 0.69-0.88).
Conclusions and Relevance The findings of this multicenter cohort study indicate that survivors of hospitalization with COVID-AKI experience lower rates of MAKE, long-term kidney function decline, and mortality compared with patients with AKI associated with other illnesses.
Introduction
Acute kidney injury (AKI) is frequently experienced by patients hospitalized for COVID-19.1 AKI has been reported in more than one-fifth of COVID-19 hospitalizations, twice as often in patients requiring treatment in the intensive care unit, and has been associated with approximately 10% dialysis requirement and high mortality.1–4 Although similar incidence of AKI is reported for patients hospitalized with COVID-19 and other respiratory viral illnesses, such as influenza,5 COVID-19 has been associated with more severe AKI, higher incident chronic kidney disease (CKD), and higher mortality.6–8 COVID-19 has also been associated with a range of functionally limiting long-term sequelae, even among individuals not requiring hospitalization.9 Therefore, in addition to the acute adverse effects of COVID-19 on kidney function, there is a growing concern for prolonged adverse effects.9–11 We have previously shown that compared with patients who develop AKI associated with other illnesses, patients with AKI during COVID-19 experience a greater decline in kidney function in the 6 months after hospital discharge.12 Other studies have similarly shown continued kidney function decline at a median of 4 months following COVID-19 regardless of illness severity.13,14
Although there is a breadth of evidence on acute and intermediate kidney outcomes after COVID-19 infection,9,10,14 there are limited data on the longer-term effects. Given the high prevalence of COVID-19 survivors, characterizing longitudinal kidney effects of COVID-19 infection has important public health implications for post−COVID-19 care.14 The objective of this study was to assess longer-term kidney outcomes in patients who experience COVID-19−associated AKI (COVID-AKI). We compared longitudinal kidney function progression, all-cause mortality, and their composite between patients hospitalized with and without COVID-19 who developed AKI and survived to discharge. We also compared these outcomes for influenza-associated AKI (flu-AKI) to assess whether kidney function trajectory after COVID-AKI was different from another viral respiratory infection. We hypothesized that patients with COVID-AKI would have a faster kidney function decline compared with those with AKI associated with other illnesses (other-AKI), including influenza.
Methods
This study was reviewed and approved by the Yale Institutional Review Board, and informed consent was waived because the research was deemed to be of minimal risk to patients. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Study Design and Setting
This retrospective longitudinal multicenter cohort study used data from the electronic health records (EHRs) of adults hospitalized in a large New England health system of 5 hospitals between March 10, 2020, and June 1, 2022, who underwent SARS-CoV-2 screening via reverse transcription−polymerase chain reaction, and a historic cohort of patients hospitalized during the 5 years before the COVID-19 pandemic (October 6, 2016-January 15, 2020) who had a positive influenza A or B test result. All had AKI during the index admission and survived to hospital discharge.
Eligibility Criteria
Patients included were adults (age ≥ 18 years) with AKI during hospital admission, results of 2 or more creatinine tests during hospitalization, survived hospitalization, and results for 1 or more creatinine level tests measured as an outpatient during the subsequent 2 years. The COVID-AKI and other-AKI groups were negative for other respiratory viral illnesses. Patients excluded were younger than 18 years; had a baseline estimated glomerular filtration rate (eGFR) of less than 15 mL/min/1.73m2; had been diagnosed with end-stage kidney disease (ESKD); had a solid organ transplant preindex hospital encounter; or had a serum creatinine level of 4 mg/dL or higher at admission. Also excluded were patients who had dialysis-requiring AKI and were discharged while receiving dialysis (defined as those who underwent dialysis within 3 days preceding their discharge date). Only outpatient creatinine values were considered in the analysis of kidney outcomes postdischarge. Patients who had opted out of research on their EHR (<1% of patients hospitalized within this hospital system) were also excluded.
AKI was defined according to the KDIGO (Kidney Disease: Improving Global Outcomes) consensus guidelines15 as a rise in serum creatinine concentration from baseline by 0.3 mg/dL in 48 hours or a 50% relative increase in 7 days. The baseline creatinine level was the lowest result during the preceding 7 days, or when available, the median of serum creatinine values in the 7 to 365 days16 before admission or imputed from an eGFR of 75 mL/min/1.73 m2 according to the Acute Dialysis Quality Initiative consensus guidelines17 using the 2021-CKD-EPI equation.18 Imputation was performed only if the patient did not have a history of CKD so as not to exclude those without pre-admission creatinine results who presented with AKI on admission. Stages of AKI were defined according to KDIGO criteria,15 where doubling of serum creatinine levels from baseline represents stage 2, and tripling of baseline creatinine or requirement of dialysis is stage 3. GFR was estimated using the 2021 race-free CKD-EPI eGFR-creatinine equation.18
Covariates
Independent variables included baseline demographic characteristics at index admission; comorbidities at index admission, extracted using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes; Elixhauser comorbidity score19; body mass index calculated as weight in kilograms divided by height in meters squared; baseline kidney function using serum creatinine concentration; weight change during hospitalization; markers of AKI severity, such as peak creatinine and dialysis receipt; and markers of illness severity, such as peak modified Sequential Organ Failure Assessment (mSOFA, scored using respiratory rate, platelet count, total bilirubin, mean arterial pressure, creatinine) and receipt of vasopressors, narcotics, any oxygen support, mechanical ventilation, and/or paralytics. Race and ethnicity information was included to identify any potentially significant differences in illness severity and/or outcomes. These data were obtained using self-identified race and ethnicity information from the EHR.
Outcomes
The primary outcome was major adverse kidney events (MAKE) operationalized as a composite of worsened kidney function (WKF; defined as eGFR decline by ≥25% or ICD-10 code for ESKD) or death. Secondary outcomes included the individual components of MAKE and annualized eGFR changes from discharge through 2 years of follow-up. Death data were extracted from the EHR, which obtains this information from direct report as well as the US National Death Index.
Statistical Analysis
Continuous variables are presented as mean (SD) if normally distributed, and median (IQR) if skewed. Categorical variables are presented as counts (percentages). Descriptive statistics were compared using the χ2 test for categorical variables and the Kruskal-Wallis test for continuous variables with post hoc Wilcoxon rank-sum test for pairwise comparisons. MAKE and its individual components were compared using a time-to-event Cox proportional hazards regression analysis adjusted for multiple potential confounders including age, sex, race and/or ethnicity, baseline comorbidity (CKD, hypertension [HTN], diabetes [DM], coronary artery disease [CAD], heart failure [HF], Elixhauser comorbidity score), markers of AKI severity (baseline, peak, and discharge creatinine levels; dialysis receipt), and markers of illness severity during hospitalization (ICU admission, peak mSOFA, vasopressor, antibiotic, narcotic, paralytic, oxygen support, mechanical ventilator support, length of stay).
In all time-to-event analyses, discharge time was considered the start time. Only outpatient eGFR values were considered in this analysis. For the WKF analysis, patients were censored at their last eGFR value during the 2 years of follow-up or on September 20, 2022 (the time of data lock), whichever came first. For the mortality analysis, patients were considered to be living if no death date was available in their EHR and censored at 2 years postdischarge or on September 20, 2022, whichever came first. The slope of eGFR was estimated using linear mixed effects models with random intercepts and random slope with the exposures modeled as interaction with time from discharge, adjusting for baseline demographic characteristics and comorbidities and the aforementioned markers of AKI severity and illness severity. The model for goodness-of-fit tests used the Akaike Information Criteria.20
Sensitivity Analysis
As a sensitivity analysis, we performed linear mixed-effects models with the exposure modeled as an interaction between the exposure (infection status), time from discharge, and time intervals from discharge (ie, 0-6, 6-12, and 12-24 months) to explore differences in eGFR slope over time. Prespecified subgroup analyses of MAKE were performed similarly, including baseline and discharge eGFR range categories; age younger than 65 years compared with 65 years or older; female compared with male sex; non-Black race compared with Black patients (due to concerns for higher risk of eGFR progression in the latter group)21; CKD, HTN, DM, CHF, and AKI stages; and COVID-19 vaccination status (0, 1, or ≥2 doses). Subgroup analyses are only hypothesis-generating and should be interpreted with caution.
Furthermore, we performed propensity score−matched analysis of time-to-MAKE and time-to-WKF among patients with COVID-AKI compared with patients with other-AKI and flu-AKI compared with other-AKI using 1:1 matching of these pairs without replacement. We planned to adjust for covariates with percentage bias greater than 10%. Matching was performed using all covariates adjusted for in the multivariable analyses. We also performed multivariable Cox-regression analyses of MAKE and WKF using an eGFR decline cutoff of 40% or more from discharge eGFR, ESKD alone, and a composite of ESKD and mortality. Furthermore, to account for potential ascertainment bias in our WKF outcome from between-group differences in follow-up frequency, we performed an inverse probability of follow-up weighted analysis.
Statistical analyses were performed using Stata, version 18 (StataCorp), and R Studio, version 4.1.3 (The R Foundation for Statistical Computing), from December 2022 to November 2023. Statistical tests were 2-tailed and P values < .05 were considered statistically significant for all analyses, except for interactions where a P value < .10 was used. There was no adjustment for multiplicity in any analysis other than the primary analysis.
Results
The study population included 9624 hospitalized patients (mean [SD] age, 69.0 [15.7] years; 4955 [51.5%] females and 4669 [48.5%] males; 1576 Black [16.4%], 1007 (10.5%) Hispanic, and 7098 [73.8%] White individuals), composed of 987 patients with a positive COVID-19 test result and AKI (COVID-AKI group); 276 from a historic cohort with a positive influenza test result and AKI (flu-AKI group); and 8361 with AKI associated with other illness (other-AKI group). Demographic characteristics for each of the 3 AKI study cohorts are detailed in Table 1. The flu-AKI group was older compared with the COVID-AKI and other-AKI groups. There was a higher proportion of self-identified Black and Hispanic individuals in the COVID-AKI group (288 [29.2%] and 175 [17.7%], respectively) compared with the other 2 groups (52 [18.8%] and 34 [12.3%] for flu-AKI and 1236 [14.8%] and 798 [9.5%] for other-AKI, respectively). The participant flow diagram is presented in eFigure 1 in Supplement 1.
Baseline Characteristics
In addition to demographic information, Table 1 presents baseline comorbidity and hospital admission-associated data. The COVID-AKI group had the highest prevalence of DM (467 [47.3%]). The flu-AKI group had the highest prevalence of CKD and CHF. The other-AKI group had the lowest prevalence of DM, HTN, CKD, CHF, and CAD.
Baseline serum creatinine levels were lowest in patients with COVID-AKI. Patients with COVID-AKI were more likely to be mechanically ventilated and require dialysis, but had similar vasopressor use as the other-AKI group during admission. The COVID-AKI group had a significantly higher use of glucocorticoids. The COVID-AKI group had the highest eGFR at discharge (median [IQR] eGFR, 64.0 mL/min/1.73 m2 [44.0-89.5] vs other-AKI, 51.6 [38.8-76.4] vs flu-AKI, 53.0 [36.4-77.7]; P < .001) despite having the highest peak mSOFA score and the longest hospital stay (COVID-AKI, 10 [6-19] days; flu-AKI, 7 [4-12] days; other-AKI, 6 [4-12] days; P < .001). The COVID-AKI group also had the largest weight loss at the time of discharge with a median (IQR) of −1.9 (−6.2 to 0.5) kg compared with −0.9 (−3.6 to 1.0) kg for the flu-AKI group and −0.4 (−3.9 to 1.7) kg for the other-AKI group. There was substantial missingness (overall approximately 20%) in baseline, admission, and follow-up proteinuria and weight data, less so among the COVID-AKI group (14%).
Primary Outcome
MAKE
MAKE (compared with other-AKI) was lower for the COVID-AKI group but no different for the flu-AKI group on unadjusted analysis. After adjustment, MAKE was lower for both the COVID-AKI (adjusted hazard ratio [aHR], 0.67; 95% CI, 0.59-0.75; P < .001) and flu-AKI (aHR, 0.81; 95% CI, 0.68-0.96; P = .02) (Table 2). The Kaplan-Meier curves for MAKE, along with WKF and mortality, are shown in Figure 1.
Subgroup Analyses—MAKE
Adjusted subgroup analyses of MAKE are presented in Figure 2. The COVID-AKI group had lower MAKE compared with other-AKI in all subgroups, and lower MAKE compared with flu-AKI in all except for patients without DM. Among patients with DM, surviving COVID-AKI was associated with a lower risk of MAKE compared with flu-AKI, adjusted for other potential confounders.
Secondary Outcomes
Descriptive secondary outcomes after hospital discharge are presented in eTable 1 in Supplement 1. The median (IQR) of durations of follow-up were: for COVID-AKI, 302 (100-496) days; other-AKI, 242 (79-450) days; and flu-AKI, 481 (123-638) days. The median (IQR) number of eGFR measurements were 3 (2-6) for COVID-AKI compared with 4 (2-7) for other-AKI and 4 (2-7) for flu-AKI. Other-AKI had the highest incidence rate of follow-up at a median (IQR) of 8.4 (8.3-8.5) per person-year compared with COVID-AKI at 5.2 (5.1-5.4) and flu-AKI at 5.9 (5.6-6.1).
In unadjusted analyses, the COVID-AKI group had the lowest 1-year mortality at 5.4% compared with 9.8% (other-AKI) and 9.0% (flu-AKI). There was a similar trend at 2 years (7.4%, 13.9%, and 16.7%, respectively; P < .001). The incidence of WKF at 6 months was highest for flu-AKI (19.5%, 19.9%, and 22.4%, respectively), and there was no difference in WKF incidence between the groups at 1 year (each approximately 28%), although the 2-year incidence differed and was lowest among the COVID-AKI group (33% vs 42% for flu-AKI and 34% for other-AKI). The flu-AKI group had the highest 2-year ESKD diagnosis (5.1% vs 1.3% for other-AKI vs 1.6% for COVID-AKI).
Time-to-Event Analyses: WKF and Mortality
With other-AKI as reference, there was no difference in time-to-WKF between the groups on unadjusted analysis: for COVID-AKI, HR of 0.99 (95% CI, 0.89-1.11; P = .86) and for flu-AKI, HR of 1.01 (95% CI, 0.84-1.21; P = .94). On multivariable Cox proportional hazards analysis, WKF was lower for COVID-AKI (aHR, 0.78; 95% CI, 0.69-0.88; P < .001) compared with the other-AKI group but no different for flu-AKI (aHR, 0.88; 95% CI, 0.72-1.06; P = .17).
The COVID-AKI group had markedly lower unadjusted all-cause mortality compared with the other-AKI group (HR, 0.51; 95% CI, 0.40-0.65; P < .001), which became even lower after multivariable adjustment (aHR, 0.31; 95% CI, 0.24-0.39; P < .001). The flu-AKI group, on the other hand, had no different all-cause mortality risk compared with other-AKI on unadjusted (HR, 1.19; 95% CI, 0.89-1.59; P = .25) or adjusted analysis (aHR, 0.79; 95% CI, 0.59-1.08; P = .14).
Kidney Function Trajectories
Kidney function trajectories of the 3 groups are shown in eFigure 2 in Supplement 1 as plots of median and bootstrap 95% CIs of creatinine concentration and eGFR at relevant instances during hospitalization, at discharge, and at selected time periods after discharge. All groups had similar peak creatinine levels. All groups experienced a decline in kidney function during the follow-up period. Compared with the other 2 groups, the COVID-AKI group had higher kidney function at baseline and discharge.
Annualized Slope of eGFR
eGFR slopes over 2 years of follow-up are shown in Table 3. In unadjusted mixed-effects models, the eGFR slope was −6.86 (95% CI, −8.00 to −5.73) mL/min/1.73 m2 per-year for patients with COVID-AKI and −5.24 (95% CI, −7.10 to −3.37) for flu-AKI and −2.50 (95% CI, −2.90 to −2.10) for other-AKI. This trend persisted after adjusting for baseline demographic and clinical characteristics and kidney function during admission. Adjusted-annualized rate of eGFR change following COVID-AKI, flu-AKI, and other-AKI was −6.79 (95% CI, −7.93 to −5.64), −5.55 (95% CI, −7.45 to −3.65), and −2.60 (95% CI, −3.01 to −2.20) mL/min/1.73 m2 per-year, respectively. The slopes of both COVID-AKI and flu-AKI were statistically different from other-AKI (P < .001 and .003 for interaction, respectively).
Sensitivity Analyses
The slopes of eGFR during 6, 6 to 12, and 12 to 24 months are shown in eTable 2 in Supplement 1. All 3 groups experienced the fastest eGFR decline during the first 6 months after hospital discharge. In the adjusted model, those with COVID-AKI had the fastest eGFR decline in the first 6 months with a mean change of −7.64 (95% CI, −8.73 to −6.56) mL/min/1.73 m2. However, this substantially slowed to 0.07 (95% CI, −1.32 to 1.45) mL/min/1.73 m2 during the subsequent 6 months, and −1.97 (95% CI, −4.54 to 0.60) mL/min/1.73 m2 per year during the year 2. The eGFR decline after the first 6 months was not significant and no different from that of the other-AKI group. Flu-AKI had a similar trend. eFigure 3 in Supplement 1 shows the COVID-AKI group starting out with a relatively higher eGFR at discharge, followed by a rapid initial drop, and then a much slower decline.
Propensity score−matched analysis of the primary outcome (ie, MAKE) and its components comparing COVID-AKI to other-AKI revealed findings consistent with the multivariable analyses showing lower risk of MAKE (HR, 0.60; 95% CI, 0.51-0.71; P < .001), WKF (HR, 0.69; 95% CI, 0.58-0.82) and mortality (HR, 0.31; 95% CI, 0.23-0.41; P < .001) (eFigure 4 and eTable 3 in Supplement 1). In propensity−score matched analysis, the flu-AKI group had a similar risk of MAKE, WKF, and mortality as the other-AKI group. Covariate balance is shown in eTable 4 in Supplement 1.
Inverse probability of follow-up-weighted multivariable Cox-regression analysis of WKF revealed findings consistent with the unweighted analyses, showing lower risk of WKF for COVID-AKI (aHR, 0.84; 95% CI, 0.76-0.92; P = .002) vs other-AKI and similar risk of WKF for flu-AKI vs other-AKI (eTable 5 in Supplement 1). Multivariable analyses for MAKE and WKF using 40% or greater eGFR decline from discharge rather than 25% or greater were consistent with the analyses that used the latter (eFigure 5 and eTable 6 in Supplement 1). ESKD risk alone was no worse for COVID-AKI survivors compared with other-AKI in adjusted analyses (aHR, 0.62; 95% CI, 0.34-1.15; P = .13) and the risk of a composite of ESKD or death was markedly lower in the COVID-AKI group (aHR, 0.33; 95% CI, 0.26-0.42; P < .001), whereas the risk of ESKD was substantially higher for the flu-AKI group compared with the other-AKI group (aHR, 2.24; 95% CI, 1.23-4.09; P = .009) (eFigure 6 and eTable 7 in Supplement 1).
Discussion
To our knowledge, this is the first longitudinal follow-up of COVID-AKI that includes 2 comparator groups—a control group of patients with negative COVID-19 test results who were admitted during the COVID-19 pandemic, and a historic cohort of patients who had been hospitalized with influenza during the 5 years before the pandemic. Contrary to existing concerns, survivors of COVID-AKI appear to be a resilient population with overall better long-term survival and kidney outcomes compared with AKI associated with influenza or other illnesses. Among hospitalized adults who had AKI associated with COVID-19, influenza, or other illnesses and were followed up for up to 2 years postdischarge, we found that individuals with COVID-AKI were relatively protected from MAKE, progressive kidney dysfunction, and mortality. These results were consistent in propensity−score matched and inverse−probability of follow-up−weighted analyses as well as multivariable analyses using more stringent kidney function decline definitions. Patients with COVID-AKI also had lower MAKE than those with flu-AKI, largely driven by the lower mortality rate.
COVID-AKI has been associated with high rates of short- and intermediate-term morbidity and mortality. Studies have reported faster eGFR loss, not only among those with severe disease requiring ICU admission and kidney replacement therapy,22–24 but also among nonhospitalized individuals.13 Studies assessing kidney function decline after COVID-19 have been limited by short duration of follow-up.12,23,25,26 Various types of kidney injury have been reported in the setting of COVID-19 infection, including cytokine release−associated tubular injury, pigment nephropathy, endothelial dysfunction, and podocytopathies. Although similar kidney manifestations and nonspecific postviral syndromes have been described with other viral illnesses,27–29 COVID-19 infection of any severity has been increasingly associated with a constellation of residual multiorgan dysfunctions10,30–32 termed long-COVID or post−COVID-19 condition, spawning a growing concern for continued kidney function decline.10
Secondary analyses evaluating eGFR slope revealed a seemingly faster overall annualized eGFR decline among survivors of COVID-AKI compared with those of other-AKI. This was due to an initial rapid eGFR drop post-hospital discharge, similar to that observed during a shorter duration by previous studies.8,12–14 This faster annual eGFR decline among survivors of COVID-AKI may be partly due to this group’s higher eGFR at discharge compared with the other 2 groups, and partly due to unmeasured confounders contributing to differences in follow-up. Despite higher baseline comorbidities (more than twice the baseline CKD and CHF prevalence of the other-AKI group) and worse measures of illness severity (mSOFA, vasopressor, narcotic, paralytic, and ventilator requirements), the COVID-AKI group had a higher discharge eGFR. In fact, approximately 57% had achieved more than 50% improvement from their eGFR nadir at the time of discharge and 31% had an improvement of 100% or more. Glucocorticoid-induced hyperfiltration and loss of muscle mass are possible contributors, the former via hemodynamic effect on eGFR and the latter due to reduced creatinine generation. The COVID-AKI group had longer hospital stay, which possibly contributed to loss of muscle mass during hospitalization with subsequent weight recovery in the months after discharge, which may lead to an increase in creatinine, and hence, a faster eGFR slope. Frequent kidney recovery has been reported among survivors even after severe COVID-19 associated AKI requiring dialysis. In a cohort of patients with severe COVID-AKI discharged alive from the hospital while receiving dialysis, nearly 92% no longer required dialysis within approximately 6 months and approximately one-third recovered back to their baseline.33
Surprisingly, the COVID-AKI group had lower risks of mortality and MAKE than the other-AKI group. This may indicate that the former had better long-term outcomes compared with the latter. Those who survive a COVID-19 infection complicated by AKI may have inherent unmeasured characteristics that are associated with favorable longer-term outcomes. Notably, the COVID-AKI group’s risk of MAKE became lower and WKF became statistically significant on adjusted analysis. Adjustment for illness severity markers likely contributed to these findings because survivors of intense treatment could be more likely to have better long-term outcomes. The mortality benefit may further be attributable to differential hospital course and treatment strategies. Those hospitalized for reasons other than COVID-19 during the early pandemic period may have been sicker than the usual hospitalized AKI population, with possibly new diagnoses of chronic illnesses.
Limitations
Our study had limitations that must be considered when interpreting the results. Although we adjusted for multiple potential confounders, there are likely residual unmeasured confounders contributing to longitudinal eGFR trajectory. Absence of standardized follow-up and accurate tracking of postdischarge exposures are limitations of retrospective studies. Ascertainment bias, whereby those followed more closely are more likely to meet the outcome, is certainly a concern for WKF, although it should affect all study groups equally. Those with multiple postdischarge measurements (implying more frequent follow-up) are likely different than those with fewer measurements. To mitigate ascertainment bias, we performed an inverse probability of follow-up−weighted analysis of WKF that showed consistent findings of lower risk of progression in those with COVID-AKI compared with other-AKI. Differential exposures during the follow-up period (eg, initiation of renoprotective medications) may affect the kidney function trajectory. Evolving trends in the care of patients with CKD, including the use of new therapeutics, may contribute to differing outcomes between the historic cohort and the COVID-19 era cohort. Due to lack of socioeconomic status−related data, we were not able to adjust for social determinants of health. However, given the disproportionate effect of COVID-19 among those of low socioeconomic position, the COVID-AKI group would be expected to do worse. Furthermore, GFR is estimated from creatinine, an endogenous biomarker influenced by muscle mass and nutritional status. Patients with critical illness and prolonged immobility experience muscle wasting.34,35 The role of this may be greater among the COVID-AKI group, which had a substantially longer hospital stay and greater weight loss compared with the other 2 groups. Postdischarge weight gain and eventual return to baseline may partly explain the COVID-AKI group’s rapid eGFR decline in the immediate postdischarge period.
Conclusions
The findings of this multicenter cohort study are reassuring because, despite their seemingly higher overall postdischarge eGFR decline compared with AKI associated with influenza or other illnesses, COVID-AKI survivors experienced a rapid attenuation of their kidney function decline rate and had overall lower rates of long-term kidney progression compared with the other 2 groups. Replication in broader cohorts as well as assessment of the effects of other kidney dysfunction markers (eg, proteinuria) and the association of COVID-19−specific therapeutics with kidney function trajectory are worth investigating in future studies.