Incidence of Long COVID Following Reinfection with COVID-19
M Daniel Brannock; Emily Hadley; Alexander Preiss; Megan L Fitzgerald; Nita Jain; Emily Taylor; Andrew Wylam; Yun J Yoo; Elaine Hill; Richard A Moffitt; N3C Consortium; RECOVER Consortium
BACKGROUND
COVID-19 reinfections have emerged as a critical concern, particularly in relation to post-acute sequelae of SARS-CoV-2 infection, commonly known as long COVID. Long COVID is known to manifest diverse, debilitating symptoms across all demographics. Limited studies have investigated the causal relationship of COVID-19 reinfections and long COVID.
METHODS
We leveraged demographically diverse electronic health records from the COVID-19 enclave of the National Clinical Cohort Collaborative, part of the RECOVER initiative, to create a matched cohort of reinfected and control adults. All participants had at least one documented COVID-19 infection. We used one-to-one coarsened exact matching on sex, race/ethnicity, age, healthcare utilization, existing comorbidities, site of care, and the timing and severity of first infection. Index dates were assigned to each matched pair as the date of reinfection for the reinfected case. Long COVID was defined using a machine learning computable phenotype trained on clinically diagnosed long COVID cases. Cumulative incidence one year after index was calculated using an Aalen-Johansen estimator. Risk ratios were calculated by taking the ratio of long COVID incidence among reinfected and control cases.
RESULTS
We found that reinfection resulted in a significantly higher risk of long COVID compared to not being reinfected (risk ratio, 1.35, 95% CI, 1.32-1.39; risk difference, 0.029, 95% CI, 0.027-0.031). This effect was consistent across most stratifications.
CONCLUSIONS
We found that COVID-19 reinfection resulted in a roughly 35% increase in the incidence of long COVID in a matched cohort using observational electronic health records.
Link | PDF | Preprint: MedRxiv | Open Access
M Daniel Brannock; Emily Hadley; Alexander Preiss; Megan L Fitzgerald; Nita Jain; Emily Taylor; Andrew Wylam; Yun J Yoo; Elaine Hill; Richard A Moffitt; N3C Consortium; RECOVER Consortium
BACKGROUND
COVID-19 reinfections have emerged as a critical concern, particularly in relation to post-acute sequelae of SARS-CoV-2 infection, commonly known as long COVID. Long COVID is known to manifest diverse, debilitating symptoms across all demographics. Limited studies have investigated the causal relationship of COVID-19 reinfections and long COVID.
METHODS
We leveraged demographically diverse electronic health records from the COVID-19 enclave of the National Clinical Cohort Collaborative, part of the RECOVER initiative, to create a matched cohort of reinfected and control adults. All participants had at least one documented COVID-19 infection. We used one-to-one coarsened exact matching on sex, race/ethnicity, age, healthcare utilization, existing comorbidities, site of care, and the timing and severity of first infection. Index dates were assigned to each matched pair as the date of reinfection for the reinfected case. Long COVID was defined using a machine learning computable phenotype trained on clinically diagnosed long COVID cases. Cumulative incidence one year after index was calculated using an Aalen-Johansen estimator. Risk ratios were calculated by taking the ratio of long COVID incidence among reinfected and control cases.
RESULTS
We found that reinfection resulted in a significantly higher risk of long COVID compared to not being reinfected (risk ratio, 1.35, 95% CI, 1.32-1.39; risk difference, 0.029, 95% CI, 0.027-0.031). This effect was consistent across most stratifications.
CONCLUSIONS
We found that COVID-19 reinfection resulted in a roughly 35% increase in the incidence of long COVID in a matched cohort using observational electronic health records.
Link | PDF | Preprint: MedRxiv | Open Access