Post-acute and Chronic Kidney Function Outcomes of COVID-19 in Children and Adolescents: An EHR Cohort Study from the RECOVER Initiative

Lu Li, Ting Zhou, Yiwen Lu, Jiajie Chen, Yuqing Lei, Qiong Wu, Jonathan Arnold, Michael J. Becich, Yuriy Bisyuk, Saul Blecker, Elizabeth A. Chrischilles, Dimitri Christakis, Carol Geary, Ravi Jhaveri, Leslie Lenert, Mei Liu, Parsa Mirhaji, Hiroki Morizono, Abu Saleh Mohammad Mosa, Ali Mirza Onder, Ruby Patel, William E. Smoyer, Bradley Taylor, David A. Williams, Bradley P. Dixon, Joseph T. Flynn, Caroline Gluck, Lyndsay A. Harshman, Mark Mitsnefes, Zubin J. Modi, Cynthia G. Pan, Hiren P. Patel, Priya S. Verghese, Christopher B. Forrest, Michelle Denburg, Yong ChenmedRxiv 2024 · DOI

This study, part of the NIH’s RECOVER initiative, leverages electronic health record (EHR) data from over 1.8 million children to demonstrate that SARS-CoV-2 infection significantly increases the risk of long-term kidney dysfunction. Even in children with no prior history of renal issues, COVID-19 was associated with a higher incidence of chronic kidney disease (CKD) and sustained declines in kidney function for up to two years post-infection.

TL;DR

COVID-19 isn't just an acute respiratory threat for children; it carries a significant "long tail" of renal risk. This massive EHR study reveals that pediatric patients face a 20% to 35% increased risk of developing chronic kidney disease markers following infection, regardless of their prior health status.

What Problem Does This Paper Address?

While the acute phase of COVID-19 is generally milder in children than in adults, the "post-acute sequelae of SARS-CoV-2" (PASC), or Long COVID, remains a major concern. Most early research focused on adult kidney outcomes, leaving a gap in our understanding of how the virus affects the developing kidneys of children and adolescents.

The authors set out to determine if COVID-19 acts as a trigger for new-onset kidney disease or an accelerant for existing conditions. Specifically, they looked at the transition from acute infection to chronic kidney outcomes—a critical distinction for pediatricians managing long-term recovery.

Methods

The researchers conducted a retrospective cohort study using a "target trial" emulation framework. Key technical aspects include:

  • Propensity Score Stratification: To account for the "real-world" nature of EHR data, the team used propensity scores to balance the COVID-positive and COVID-negative groups across dozens of covariates, including age, sex, race, obesity, and clinical site.
  • Stratified Analysis: Patients were divided into three distinct "strata": those with no prior kidney issues, those with pre-existing CKD, and those who suffered Acute Kidney Injury (AKI) during the first 28 days of infection.
  • Outcome Modeling: They used Cox proportional hazards models to calculate Hazard Ratios (HR) for various kidney milestones, such as eGFR (estimated Glomerular Filtration Rate) declines and progression to End-Stage Kidney Disease (ESKD).

Datasets

The study utilized the RECOVER Database (Version s9), a massive repository of EHR data from the NIH Researching COVID to Enhance Recovery initiative.

  • Scale: 1,864,637 children and adolescents under age 21.
  • Scope: Data drawn from 19 major pediatric hospitals and health institutions across the United States.
  • Timeline: Follow-up data spanned from March 2020 to May 2023, providing one of the longest observation windows (up to 729 days) for pediatric COVID research to date.

Key Insights and Ideas

  • The "Healthy" Patient Risk: Perhaps the most striking finding is that children with no prior kidney disease or acute injury are still at risk. They showed a significantly higher risk of developing CKD Stage 2+ (HR 1.20) and Stage 3+ (HR 1.35).
  • COVID as an Accelerant: For children who already had CKD, COVID-19 increased the risk of further function decline (eGFR decline ≥30%). For those who had AKI during the acute phase, the virus significantly increased the risk of progressing to a composite of severe outcomes like dialysis or transplant.
  • Sustained Vulnerability: The risk doesn't disappear after the initial recovery. The study found elevated risks in both the "post-acute" phase (28–179 days) and the "chronic" phase (180–729 days), suggesting that the renal impact of the virus is persistent.

Key Results

  • New-Onset CKD: In previously healthy children, the hazard ratio for CKD Stage 3+ (eGFR < 60 mL/min/1.73m²) was 1.35 (95% CI: 1.15-1.59).
  • Severe Progression: In the AKI group, the risk for a composite outcome (50% eGFR decline, ESKD, dialysis, or transplant) remained high well into the second year, with an HR of 1.33 (95% CI: 1.21-1.47).
  • Function Decline: COVID-positive patients across all strata were consistently more likely to show eGFR declines of 30%, 40%, and 50% compared to their non-infected peers.

Limitations and Future Work

  • The "Home Test" Bias: EHR data only captures clinical visits. Children who tested positive at home but had mild symptoms might be misclassified into the "negative" control group, which likely biases the results toward the null (meaning the actual risk might be even higher).
  • Mechanistic Uncertainty: The study identifies a correlation but cannot confirm the why. Is it direct viral invasion of kidney tissue, the result of systemic inflammation (cytokine storm), or a side effect of intensive treatments used during hospitalization?
  • Future Directions: The authors call for further investigation into the biological pathways of these injuries and emphasize the need for long-term renal monitoring in pediatric patients who have had COVID-19, especially those who were hospitalized or had prior kidney vulnerabilities.

Summary updated

Abstract

We investigated the risks of post-acute and chronic adverse kidney outcomes of SARS-CoV-2 infection in the pediatric population via a retrospective cohort study using data from the RECOVER program. We included 1,864,637 children and adolescents under 21 from 19 children's hospitals and health institutions in the US with at least six months of follow-up time between March 2020 and May 2023. We divided the patients into three strata: patients with pre-existing chronic kidney disease (CKD), patients with acute kidney injury (AKI) during the acute phase (within 28 days) of SARS-CoV-2 infection, and patients without pre-existing CKD or AKI. We defined a set of adverse kidney outcomes for each stratum and examined the outcomes within the post-acute and chronic phases after SARS-CoV-2 infection. In each stratum, compared with the non-infected group, patients with COVID-19 had a higher risk of adverse kidney outcomes. For patients without pre-existing CKD, there were increased risks of CKD stage 2+ (HR 1.20; 95% CI: 1.13-1.28) and CKD stage 3+ (HR 1.35; 95% CI: 1.15-1.59) during the post-acute phase (28 days to 365 days) after SARS-CoV-2 infection. Within the post-acute phase of SARS-CoV-2 infection, children and adolescents with pre-existing CKD and those who experienced AKI were at increased risk of progression to a composite outcome defined by at least 50% decline in estimated glomerular filtration rate (eGFR), eGFR <15 mL/min/1.73m<sup>2</sup>, End Stage Kidney Disease diagnosis, dialysis, or transplant.
BibTeX
@misc{li2024post,
  author = {Lu Li and Ting Zhou and Yiwen Lu and Jiajie Chen and Yuqing Lei and Qiong Wu and Jonathan Arnold and Michael J. Becich and Yuriy Bisyuk and Saul Blecker and Elizabeth A. Chrischilles and Dimitri Christakis and Carol Geary and Ravi Jhaveri and Leslie Lenert and Mei Liu and Parsa Mirhaji and Hiroki Morizono and Abu Saleh Mohammad Mosa and Ali Mirza Onder and Ruby Patel and William E. Smoyer and Bradley Taylor and David A. Williams and Bradley P. Dixon and Joseph T. Flynn and Caroline Gluck and Lyndsay A. Harshman and Mark Mitsnefes and Zubin J. Modi and Cynthia G. Pan and Hiren P. Patel and Priya S. Verghese and Christopher B. Forrest and Michelle Denburg and Yong Chen},
  title = {{Post-acute and Chronic Kidney Function Outcomes of COVID-19 in Children and Adolescents: An EHR Cohort Study from the RECOVER Initiative}},
  year = {2024},
  doi = {10.1101/2024.06.25.24309488},
  howpublished = {medRxiv},
}
Figure